You are on page 1of 70

Review

pubs.acs.org/CR

Porous Anodic Aluminum Oxide: Anodization and Templated


Synthesis of Functional Nanostructures
Woo Lee*,, and Sang-Joon Park

Korea Research Institute of Standards and Science (KRISS), Yuseong, 305-340 Daejeon, Korea
Department of Nano Science, University of Science and Technology (UST), Yuseong, 305-333 Daejeon, Korea

6.2.3. Morphological Instability


6.3. Steady-State Pore Formation
6.3.1. Joules Heat-Induced Chemical Dissolution
6.3.2. Field-Assisted Oxide Dissolution
6.3.3. Average Field Model for Steady-State
Pore Structure
6.3.4. Direct Cation Ejection Mechanism
6.3.5. Flow Model for Steady-State Pore
Formation
7. Self-Ordered Porous Anodic Aluminum Oxide
(AAO)
7.1. Mild Anodization (MA)
7.2. Hard Anodization (HA)
7.3. Pulse Anodization (PA)
7.4. Cyclic Anodization (CA)
7.5. Anodization of Thin Aluminum Films Deposited on Substrates
8. Long-Range Ordered Porous AAO
9. AAO Template-Based Synthesis of Functional
Nanostructures
9.1. Electrochemical Deposition (ECD)
9.2. Electroless Deposition (ELD)
9.3. SolGel Deposition
9.4. Surface Modication
9.5. Template Wetting
9.6. Mask Techniques
9.7. Chemical Vapor Deposition (CVD)
9.8. Atomic Layer Deposition (ALD)
10. Closing Remarks and Outlook
Author Information
Corresponding Author
Notes
Biographies
Acknowledgments
Abbreviations
References

CONTENTS
1. Introduction
2. Types of Anodic Aluminum Oxide (AAO)
3. Ionic Conduction in Anodic Oxide Films
3.1. High-Field Conduction Theory
3.2. Elementary Interfacial Reactions
3.3. Transport Numbers
3.4. Stress-Driven Ionic Transport
4. Electrolytic Breakdown
4.1. Factors Inuencing Breakdown
4.1.1. The Nature of Anodized Metal
4.1.2. Electrolyte Conditions
4.1.3. Current Density (j)
4.1.4. Other Factors Inuencing Breakdown
4.2. Models for Breakdown
4.2.1. Electron Avalanche Multiplication
4.2.2. Stress-Driven Breakdown
5. Structure of Porous Anodic Aluminum Oxide
(AAO)
5.1. General Structure
5.1.1. Pore Diameter (Dp)
5.1.2. Interpore Distance (Dint)
5.1.3. Barrier Layer Thickness (tb)
5.2. Structure of Pore Wall (Anion Incorporation)
5.3. Eect of Heat Treatments
6. Growth of Porous Anodic Aluminum Oxide (AAO)
6.1. Stress Generation in Anodic Oxide Films
6.1.1. Volume Expansion
6.1.2. Stress Measurements
6.1.3. Eects of External Stresses on Pore
Growth
6.2. Initial-Stage Pore Formation
6.2.1. Qualitative Description on Pore Formation
6.2.2. Kinetics of Porosity Initiation
2014 American Chemical Society

7487
7488
7490
7490
7490
7491
7492
7493
7493
7494
7494
7494
7494
7495
7495
7496
7497
7497
7497
7499
7499
7499
7502
7502
7502
7502
7503

7507
7509
7509
7509
7510
7511
7512
7513
7514
7516
7518
7521
7521
7524
7528
7528
7531
7531
7533
7538
7539
7540
7541
7543
7544
7544
7544
7545
7545
7545
7547

1. INTRODUCTION
In ambient atmospheres, aluminum becomes rapidly coated
with a compact 23 nm thick oxide layer. This native oxide
layer prevents the metal surface from further oxidation. Because
of the surface native oxide, aluminum generally has good
corrosion resistance. However, local corrosion of metal can
occur in rather aggressive outdoor environments, containing

7505
7506
7506
7506

Received: January 2, 2014


Published: June 13, 2014
7487

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

discussed (section 5). Anodization of aluminum is a volume


expansion process, and thus is accompanied by mechanical
stresses. Recent studies have indicated that the stresses have
profound implications not only on the ionic transport, but also
on the self-ordering behavior of oxide nanopores. We will
discuss in detail the eect of stress on pore growth (section
6.1), the kinetics of pore initiation, and morphological
instability associated with the early stage of anodization
(section 6.2), and recent models describing steady-state pore
formation (section 6.3). After that, recent progress on
anodization of aluminum used in fabricating self-ordered
porous AAO and also for engineering internal pore structures
will be discussed (section 7). In addition, various approaches to
long-range order porous AAO will be reviewed (section 8). In
the last part of this Review (section 9), various chemical
approaches for the syntheses of low-dimensional functional
nanostructures and the fabrications of advanced nanodevices
will be discussed. These approaches include electrochemical
deposition (ECD), electroless deposition (ELD), solgel
deposition, surface modication, template wetting, shadow
mask techniques, chemical vapor deposition (CVD), and
atomic layer deposition (ALD). Chemistry issues encountered
in the template-based synthesis of functional nanostructures
will be discussed in detail. Finally, we will present the
challenges and future prospects of the eld (section 10).

corrosive chemicals (e.g., chlorides or sulfates). In 1857, Bu


rst found that aluminum can be electrochemically oxidized in
an aqueous solution to form an oxide layer that is thicker than
the native one.1 This phenomenon has been called anodization because the aluminum part to be processed constitutes
the anode in an electrolytic cell. In the early 1920s, the
phenomenon observed by Bu was exploited for industrial scale
applications, for example, protection of seaplane parts from
corrosive seawater.2 In general, the anodic aluminum oxide
(AAO) lms form with two dierent morphologies (i.e.,
nonporous barrier-type oxide lms and porous-type oxide
lms) depending mainly on the nature of the anodizing
electrolyte.3 Because the process was rst implemented for
protection purposes, the anodization of aluminum and its
alloys, particularly porous-type anodization, has received
considerable attention in the industry because of its extensive
practical applications. Many desirable engineering properties
such as excellent hardness, corrosion, and abrasion resistance
can be obtained by anodizing aluminum metals in acid
electrolytes.4 In addition, due to its high porosity, the porous
oxide lms formed on the metals serve as a good adhesion base
for electroplating, painting, and semi-permanent decorative
coloration. The anodized products can be easily found in
electronic gadgets, electrolytic capacitors, cookware, outdoor
products, plasma equipment, vehicles, architectural materials,
machine parts, etc. Recently, this nearly century-old industrial
process has been drawing increasing attention from scientists in
the eld of nanotechnology. This trend originated with the
seminal works of Masuda and co-workers, who reported on
self-ordered porous AAO in 19955 and the subsequent
development of the two-step anodization process in 1996.6
Porous AAO lm grown on aluminum is composed of a thin
barrier oxide layer in conformal contact with aluminum, and an
overlying, relatively thick, porous oxide lm containing
mutually parallel nanopores extending from the barrier oxide
layer to the lm surface.7 Each cylindrical nanopore and its
surrounding oxide region constitute a hexagonal cell aligned
normal to the metal surface. Under specic electrochemical
conditions, the oxide cells self-organize into hexagonal closepacked arrangement, forming a honeycomb-like structure.57
Pore diameter and density of self-ordered porous AAOs are
tunable in wide ranges by properly choosing anodization
conditions: pore diameter = 10400 nm and pore density =
1081010 pores cm2. The novel and tunable structural features
of porous AAOs have been intensively exploited for synthesizing a diverse range of nanostructured materials in the forms of
nanodots, nanowires, and nanotubes, and also for developing
functional nanodevices.
The objective of this Review is to provide a solid information
source for researchers entering this eld and to establish a
broad and deep knowledge base. This Review introduces the
fundamental electrochemical processes associated with anodic
oxidation of aluminum, and discusses the recent progress on
anodization of aluminum for the development of ordered
porous AAOs, and nanotechnology applications of porous
AAOs. We organize this Review as follows: after discussing the
growth characteristics of two dierent types of AAOs (section
2), we will describe the theory of ionic conductions and
elementary interfacial reactions (section 3), followed by
electrolytic breakdown (section 4) to understand the
fundamental electrochemistry associated with anodic oxidation
of aluminum. Next, the electrochemical factors dening the
geometric and chemical structures of porous AAOs will be

2. TYPES OF ANODIC ALUMINUM OXIDE (AAO)


Anodization of aluminum in aqueous electrolytes forms anodic
oxide lms with two dierent morphologies, that is, the
nonporous barrier-type oxide lms and the porous-type oxide
lms. The chemical nature of the electrolytes mainly
determines the morphology of AAOs.3,7,8 A compact nonporous barrier-type AAO lms can be formed in neutral
electrolytes (pH 57), such as borate, oxalate, citrate,
phosphate, adipate, tungstate solution, etc., in which the anodic
oxide is practically insoluble.9,10 Meanwhile, porous-type AAOs
are formed in acidic electrolytes, such as selenic,11 sulfuric,12
oxalic,12 phosphoric,7,12,13 chromic,12,14 malonic,12,1517 tartaric,12,18 citric,12,1720 malic acid,12,18 etc., in which anodic oxide
is slightly soluble. Early models describing anodic oxide growth
were developed on the basis of the barrier-type oxide.2124
Moreover, in the early stage of porous-type oxide growth, the
formation of the initial barrier oxide is followed by the
emergence of incipient pores. Therefore, in this Review, we will
mention the barrier-type oxide growth to the extent needed for
understanding porous-type oxide formation. Some excellent
review articles covering the barrier-type anodic oxide lms are
given in refs 3 and 25.
The two types of anodic oxides (i.e., barrier- vs porous-type
AAO) dier in their oxide growth kinetics. In the case of
barrier-type oxide formation under potentiostatic conditions
(i.e., U = constant), current density (j) decreases exponentially
with time (t). Correspondingly, the lm growth rate decreases
almost exponentially with time (t), which places a limit on the
maximum lm thickness obtainable for barrier-type AAO lms
(Figure 1). It has been experimentally veried that the
thickness of barrier-type lm is directly proportional to the
applied potential (U). On the other hand, current density (j) in
porous-type anodization under potentiostatic conditions
remains almost constant within a certain range of values during
the anodization process, due to the constant thickness of the
barrier layer at the pore bottom. The thickness of the resulting
porous oxide lm is linearly proportional to the total amount of
7488

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

aluminum, new oxide forms above and below the marker


layer (Figure 2b). The marker layer is located at a depth of 40%
of the lm thickness in a plane corresponding to that of the
original metal surface. On the other hand, when a barrier-type
lm is formed at 60% current eciency (j), the plane of the
marker layer is immobile and 40% of the Al3+ cations are shed
into the electrolyte via direct cation ejection mechanism
without contributing to the oxide formation (Figure 2c). In this
case, anodic oxide grows at the metal/oxide interface via the
inward migration of O2/OH ions. When porous-type anodic
oxide forms at 60% current eciency, the marker plane is
located above that of the original metal surface (Figure 2d). In
this case, the metal/oxide interface is also the oxide growth
front, and 40% of Al3+ ions are ejected into the solution.
Because cations are being shed into the electrolyte, the
current eciency (j) of porous-type oxide growth is typically
much lower than that of the barrier-type. Accordingly, the
PillingBedworth ratio (PBR = the ratio of molar volume of
the grown oxide to molar volume of the consumed metal; see
section 6.1.1) for the initial barrier oxide formation in poroustype oxide growth at the early stage of anodization is lower than
that for barrier-type oxide growth: PBR = 0.90 for porous-type
oxide growth at j = 53.5% in phosphoric acid solution and
PBR = 1.7 for barrier-type oxide growth at j = 100% in neutral
adipate solution.9,33 Shimizu et al.33 suggested that the initial
barrier oxide grows under increasing tensile stress (PBR < 1),
which causes local oxide cracking most probably at the
randomly present metal protrusions. The generated surface
cracks were considered to be local paths for electrolyte
penetration, causing non-uniform local thickening of the initial
barrier oxide. Non-uniform thickening of the initial oxide causes
concentration and redistribution of the current lines into the
relatively thin oxide regions between the protrusions (i.e., a
local increase in electric eld, E). Consequently, localized
scalloping of the metal/oxide interface takes place. Shimizu et
al.33 pointed out that the non-uniform thickening of anodic
oxide (i.e., morphological instability) in the initial barrier oxide
is one of the most distinctly dierent growth features between
porous- and barrier-type AAO lms. Unlike porous-type oxide
growth in acid electrolytes, anodic oxides in neutral electrolytes
grow highly uniformly on surface nished aluminum,
maintaining at metal/oxide and oxide/electrolyte interfaces.
Even the smoothing of initially rough aluminum surfaces during
the growth of barrier oxide lms has been experimentally
observed.34

Figure 1. Two dierent types of anodic aluminum oxide (AAO)


formed by (a) barrier-type and (b) porous-type anodizations, along
with the respective current (j)time (t) transients under potentiostatic
conditions.

charge (i.e., anodization time, t) involved in the electrochemical


reaction.
Radiotracer studies, employing an immobile marker (125Xe),
have indicated that, in the case of barrier-type oxide formation,
anodic alumina grows simultaneously at the oxide/electrolyte
interface and at the metal/oxide interface, through Al3+ egress
and O2/OH ingress, respectively, under a high electric eld
(E).2628 In the case of porous-type anodic alumina formation,
on the other hand, oxide grows at the metal/oxide interface via
the inward migration of O2/OH ions. 18O tracer studies have
shown that outwardly migrating Al3+ cations do not contribute
to the oxide growth at the oxide/electrolyte interface, but are all
shed into the anodizing electrolyte via direct ejection
mechanism (see section 6.3.4).10,2931 Otherwise, egressing
Al3+ ions would form anodic alumina at the oxide/electrolyte
interface to heal any developing or embryonic pores there.
Schematic diagrams illustrating the dimensional changes of
aluminum during the barrier-type and porous-type anodic oxide
formation are shown in Figure 2.32 An immobile marker layer is
implanted into the starting aluminum with a native oxide layer
(Figure 2a). When a barrier-type lm is formed at 100% current
eciency (j) by anodization of the marker-implanted

Figure 2. Schematic diagrams illustrating dimensional changes of an aluminum specimen following anodizing. (a) Initial aluminum with a thin airformed oxide lm. The red dashed line represents an immobile marker layer implanted into the initial aluminum with a thin air-formed oxide lm.
(b) Anodized at 100% eciency with formation of a barrier-type anodic lm. (c) Anodized at just above 60% eciency with formation of a barriertype anodic lm. (d) Anodized at 60% eciency with formation of a porous anodic lm. Reproduced with permission from ref 32. Copyright 2006
The Electrochemical Society.
7489

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

in the oxide is high enough (e.g., 106107 V cm1), the ionic


current density (j) can be expressed as25

3. IONIC CONDUCTION IN ANODIC OXIDE FILMS


3.1. High-Field Conduction Theory

W azFE

j = va exp
RT RT

When a valve-metal is anodized under either potentiostatic or


galvanostatic condition, anodic oxide lm forms on the metal.
For anodizing aluminum (Al) and tantalum (Ta), an empirical
exponential dependence of the ionic current density (j) on the
electric eld (E) is established. Ionic current density (j) under
high-eld conditions, which is the case for anodic oxide growth,
can be associated with the movement of charged ions in the
barrier oxide, and can be related to the potential drop (U)
across the barrier oxide through the exponential law of
Guntherschulze and Betz, as follows:21,35
j = j0 exp(E) = j0 exp( U /tb)

(2)

where v is the hopping attempt frequency of the ion, is the


density of concentration of mobile charge in C cm3, a is the
hopping inter-distance, W is the hopping activation energy at
zero eld, is a parameter describing the asymmetry of the
activation barrier at non-zero eld, z is the valence of the
mobile ions, and F is Faradays constant. From eqs 1 and 2, the
following relations can be obtained:
W

j0 = a exp
RT

(1)

where j0 and are material-dependent constants at a given


temperature, and U/tb is the eective electric eld (E,
typically 106107 V cm1) impressed on the barrier layer with
thickness tb. For anodic alumina, a large range of j0 and values
has been reported: j0 = 3 104 to 1 1018 A cm2 and = 0.1
106 to 5.1 106 cm V1.25
For anodic oxidation of metal in an electrolyte, three theories
based on the following possible rate-determining steps for oxide
formation have been developed:3 ion transfers (i) across the
metal/oxide interface (MottCabrera theory),23,24 (ii) through
the oxide bulk (Verwey theory),22 and (iii) across the oxide/
electrolyte interface (Dewald theory).36,37 In the point defect
model of Macdonald et al.,38 the oxide lm is assumed to
contain a high concentration of non-interacting positive and
negative point defects, and the rate-determining step for the
oxide growth is assumed to be the transport of metal and oxide
vacancies across the oxide lm. All of these theories can explain
the empirical exponential relationship proposed by Guntershultz and Betz. On the other hand, transient experiments
favorably indicate that the rate-determining step is the
movement of charged ions within the oxide.25
On the basis of the rate-determining movement of ions
within the oxide, the high-eld model relates the parameters j0
and in eq 1 to the nature of oxide materials. The high-eld
conduction model is based on a hopping mechanism, in which
the activation energy for hopping ions is dependent on electric
eld E (Figure 3).25 Ions at regular sites or interstitial positions
jump to vacancies or other interstitial positions in their
neighborhood. The model assumes that the oxide is defect-free
and of homogeneous composition. When the electric eld (E)

azF
RT

(3)
(4)

Because the parameter a can be related to the inter-atomic


distance in the oxide, one can expect that the electric eld
strength (E) increases when the oxygen ion density increases
(i.e., a decrease in parameter a) provided that the other
parameters are constant. Equation 1 can be modied to obtain
a Tafel equation:

ln j = ln j0 + E

(5)

For a constant oxide thickness tb, a constant Tafel slope is


obtained.
The electric eld (E) in the oxide can be related to the
applied (or measured, in the potentiostatic condition) electrode
potential (U). The measurable potential drop between the
metal and the electrolyte is equal to
U = U + m/o + o/e

(6)

where U is the potential drop in the oxide, and m/o and o/e
are the potential drops at the metal/oxide and oxide/electrolyte
interfaces, respectively.39 In a typical anodization, the potential
drops at the metal/oxide and oxide/electrolyte interfaces are
quite small, as compared to the several tens of volts of potential
drop in the oxide (i.e., U m/o + o/e). Therefore, the
following approximation for the electric eld (E) is possible for
the high-eld ionic transport:
E = U /tb U /tb

(7)

where tb is the thickness of oxide.


3.2. Elementary Interfacial Reactions

As was already discussed in section 2, it is now widely accepted


that for the anodic growth of alumina both Al3+ cations and
oxygen-containing anions (e.g., O2 or OH) are mobile within
the anodic oxide under high electric eld (E).10,2628,40 Al3+
ions migrate outwardly toward the oxide/electrolyte interface,
while O2 or OH anions move inwardly toward the metal/
oxide interface. Therefore, one can consider both (i) the metal/
oxide and (ii) the oxide/electrolyte interfaces as the growth
front of anodic oxide during anodization of a valve-metal. For
anodizing aluminum, the following elementary reactions are
considered to be possibly occurring at the interfaces (Figure 4).
(i) At the metal/oxide interface:
Figure 3. Inuence of the electric eld strength (E) on the activation
energy of hopping ions. Reproduced with permission from ref 25.
Copyright 1993 Elsevier.
7490

3+
Al Al(ox)
+ 3e

(8)

3+
2
2Al(ox)
+ 3O(ox)
Al 2O3

(9)

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 4. Schematic diagrams showing elementary interfacial reactions for (a) barrier-type and (b) porous-type anodic oxide.

and t = 1 t+ for anion. Transport numbers can be


determined by employing a marker layer, whose position in
the anodic oxide lm indicates the extent of oxide that was
formed at each interface. If the metal ions are the only mobile
species, new oxide should be formed at the oxide/electrolyte
interface on top of the marker layer. On the other hand, if
oxygen anions are the only mobile species, the new oxide
should be formed at the metal/oxide interface below the
marker layer. Davies et al.26 stated that the ideal marker atoms
for determination of transport numbers should fulll the
following requirements: The markers should be (i) uncharged,
so that they do not migrate in the oxide under the inuence of
the applied eld; (ii) large in size, so that they do not diuse
signicantly within the oxide lattice; (iii) present in trace
amount, so that the macroscopic properties of the tagged oxide
remain unaltered; and (iv) detectable, in order to assess their
depth in the oxide.
These conditions can be satised by implanting radioisotopes
125
Xe inert gas atoms or 222Rn, which are heavier than typical
valve-metals and oxygen, in a preformed thin oxide lm and
subsequently anodized.26,27,42 Radioactive tracers allow the
position of the buried marker to be assessed by monitoring the
energy of emitted - or -particles.26,27,42,43 Other techniques
to measure the buried marker position in oxide include
Rutherford backscattering spectrometry (RBS)40 or direct
observation of voids formed by implanted Xe by employing
cross-sectional transmission electron microscopy (TEM).28,44 A
representative cross-sectional TEM image showing an immobile
Xe marker layer is given in Figure 5. The sample in the gure
was formed in near-neutral potassium phosphate electrolyte at a
high current eciency (j 100%).44 The approximately 10nm-thick straight Xe marker layer is located at about the
midpoint of the lm. The anodic oxide above the marker layer
formed by the eld-driven outward migration of Al3+ ions and
that beneath the marker layer by the eld-driven inward
migration of oxygen carrying anions, O2/OH. Assuming that
all egressing Al3+ ions contribute to the oxide formation, the
cation transport number was directly estimated to be t+ =
0.49.44 For anodizing conditions under which oxide grows with
appreciable metal dissolution, however, TEM-based direct
measurement may underestimate the cation transport number.
In such cases, Al3+ ions dissolved in anodizing electrolyte
should be quantied to estimate the equivalent oxide thickness.
Davies et al.26 pointed out that the location of an immobile
marker in anodic oxide markedly depends on the anodization
conditions, such as current density (j) and the nature of

(ii) At the oxide/electrolyte interface:


3+
2
2Al(ox)
+ 3O(ox)
Al 2O3

(10)

+
3+
Al 2O3 + 6H(aq)
2Al(aq)
+ 3H 2O(1)

(11)

3+
3+
Al(ox)
Al(aq)

(12)

2
2O(ox)
+ O2(g) + 4e

(13)

+
2
2H 2O(1) + O(ox)
+ OH(ox)
+ 3H(aq)

(14)

Reactions 9 and 10 correspond to the formation of anodic


oxide at the metal/oxide and oxide/electrolyte interfaces,
respectively. Reaction 11 describes dissolution of anodic
alumina by Joules heat-induced oxide dissolution and/or
eld-induced oxide dissolution, which will be discussed in
section 6.3.1 and section 6.3.2, respectively. On the other hand,
reaction 12 occurs through eld-assisted direct ejection of Al3+
ions from the metal/oxide interface through oxide into the
electrolyte, which will be discussed in detail in section 6.3.4.
Reactions 1113 decrease the net current eciency (j)
associated with the anodic oxide formation. Reaction 14
describes the heterolytic dissociation of water molecules at the
oxide/electrolyte interface, which supplies oxygen anions to the
metal/oxide interface to form anodic oxide. By assuming that
all oxide anions from the dissolution of Al2O3 at the oxide/
electrolyte interface migrate to the metal/oxide interface to
reform Al2O3, and that all oxide anions from the dissociation of
water contribute to the oxide formation, Su et al. proposed the
following overall reaction at the oxide/electrolyte interface:41
3+
2
Al 2O3 + nH 2O(1) 2Al(aq)
+ (3 n x)O(ox)

+
+ xOH(ox)
+ (2n x)3H(aq)

(15)

where n denotes the amount of water dissociated per mole of


Al2O3 that is dissolved at the same time. Su et al. claimed the
eld-dependent nature of the heterolytic dissociation of water
in reaction 14 and related the dissociation rate of water to the
porosity (P) of AAO, which will be touched upon in section
7.1.
3.3. Transport Numbers

As mentioned in previous sections, anodic oxide formation can


occur at both the metal/oxide and the oxide/metal interfaces.
The relative amount of mobile ions transported to the oxide
forming interfaces is called the transport number: t+ for cation
7491

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 5. Cross-sectional transmission electron microscopy (TEM)


image showing immobile 125Xe marker layer. The sample (i.e., barriertype anodic oxide lm) was formed at a constant current density of 1
mA cm2 to 100 V in near-neutral potassium phosphate electrolyte.
Adapted from ref 44 with permission. Copyright 1987 Taylor &
Francis (www.tandfonline.com).

Figure 6. Schematics showing (a) the movement of Al3+ and O2 ions


during the re-anodizing process and (b) the corresponding cell
potential (U)time (t) curve. Reproduced with permission from ref
46. Copyright 1978 Elsevier.

electrolyte. The authors performed transport number experiments using radioisotope 125Xe marker atoms and also
quantitative analyses on dissolving Al3+ ions during anodization
in two dierent near-neutral electrolytes (i.e., sodium
tetraborate and ammonium citrate). According to their
experiments, both metal and oxygen ions are mobile during
oxide growth. In the borate solution, anodic alumina grew with
high current eciencies (i.e., negligible cation loss), and thus
the immobile 125Xe markers were completely buried in the
oxide. The mean cation transport number (t+) was estimated to
be t+ = 0.58 for anodic alumina formed at the current density
range of 0.110 mA cm2. On the other hand, in citrate
solution, the amount of aluminum passed into the solution was
as high as 40% of the total oxidized metal at low current
density, but decreased as the current density increased. The
125
Xe markers in anodic alumina remained very close to the
outer surface. The cation transport number varied with current
density, from about t+ = 0.37 at 0.1 mA cm2 to t+ = 0.72 at 10
mA cm2.
In the case of porous AAO, the transport numbers of mobile
ions can be estimated by the so-called pore-lling method,
which was originally used to determine the porosity (P) of
porous AAO by Dekker and Diddelhoek.45,46 In this method,
aluminum is rst anodized in an acid electrolyte to form
porous-type anodic oxide and subsequently re-anodized in a
neutral electrolyte electrolyte to form barrier-type oxide under a
galvanostatic condition. During anodizations, potential (U)
time (t) transients are monitored. During the barrier-type
anodizing (i.e., re-anodizing process), new oxide gradually
forms simultaneously within the pores and underneath the
barrier layer of the pre-formed porous anodic oxide, because
both Al3+ and O2 ions contribute to the oxide formation at the
metal/oxide and oxide/electrolyte interfaces, respectively. As a
result, the cell potential gradually increases with time during the
re-anodizing process. Figure 6 schematically shows (a) the
movement of Al3+ and O2 and (b) the cell potential (U)time
(t) prole during the re-anodizing process.46 The non-zero
value of U at t = 0 is due to the original barrier layer of preformed porous AAO. The complete lling of pores is
accompanied by the change of the slope in the Ut curve at
time tp due to the sudden increase of the oxide/electrolyte
interfacial area. For the time t < tp, the following relation can be
obtained:46

dh+
jM
dh
P
+
=
dt
dt nFk

(16)

where (=2.95 g cm ) is the density of oxide, P is the porosity


of porous AAO, dh+/dt and dh/dt are, respectively, the rates of
the increase of the barrier oxide thickness at the oxide/
electrolyte and metal/oxide interfaces, j is the current density,
M is the atomic weight of Al, n (=3) is the number of electrons
involved in oxidation reaction, F is Faradays constant, and k
(=0.505) is the weight fraction of aluminum in the oxide. The
cation transport number is given by the ratio of the weight of
new oxide formed within the pores per unit time to the total
weight of new oxide formed per unit time:46
dh+ dh+
dh
t + = P
+
/ P

dt dt
dt

(17)

The slopes m1 and m2 of the Ut transient in Figure 6b are


given by46
m1 =

1 dh+
dh
+

AR dt
dt

(18)

m2 =

1 dh+
dh
+
P

AR dt
dt

(19)

where AR is the anodizing ratio (=the ratio of the barrier layer


thickness to the cell potential, in nm V1), and assumed to be a
constant. From eqs 1619, the porosity (P) of porous AAO is
given by
P=

t +(m2 /m1)
1 (1 t +)(m2 /m1)

(20)

For porous AAO formed in 1.125 M oxalic acid at 30 V,


Takahashi and Nagayama reported that the transport numbers
of mobile Al3+ and O2 ions are t+ = 0.4 and t = 0.6,
respectively.46
3.4. Stress-Driven Ionic Transport

The high-eld conduction model describes the relation


between ionic current density (j) and the electric eld (E)
well. However, stress gradients in the oxide may possibly
contribute to the ionic transport. Hebert and Houser47,48 have
7492

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

to tensile transition at 0.51.0 mA cm2), while average tensile


stress of the order of 50 MPa was predicted above 1 mA cm2,
which is in good agreement with the experimental stress data of
Bradhurst and Leach.50 In addition, by taking into consideration the viscous ow of oxide material, the model predicted
the increases of the cation transport number (t+) as a function
of current density (j). On the basis of experimental evidence
that cation transport number (t+) is largely dependent on the
electrolyte condition, Hebert and Houser pointed out that the
oxide viscosity and conduction parameters may depend on the
solution composition as a result of electrolyte anion
incorporation into the anodic oxide lm. They suggested that
bulky electrolyte anions disrupt the local packing of oxygen
ions and inuence transport properties by the introduction of
additional free volume into the amorphous oxide.48

developed a model for ionic transport in growing amorphous


anodic alumina lms, in which ion migration in the oxide is
driven by gradients of mechanical stress as well as electric
potential. It also considers the viscoelastic creep of the oxide. In
other words, both stress gradient-driven ionic migration and
stress gradient-driven creep are considered in the model. It is
assumed that stress originates at the metal/oxide interface due
to the volume change upon oxidation. For stress gradientdriven ionic migration, the empirical high-eld conduction
relation is generalized by considering the dependence of the
ionic current density on the gradient of the ionic chemical
potential i:4749
Ji = 2

i
|i |

C iu i0 sinh
|i |
RT

(21)

where Ji, Ci, and are, respectively, the ux, the concentration,
and the pre-exponential velocity of ion i (i = M and O for
metal and oxygen, respectively), and a is the migration jump
distance in the oxide. The chemical potential is related to the
mean stress () and electrical potential () as follows:48
u0i

i = u i0 + z iF Vi

4. ELECTROLYTIC BREAKDOWN
When valve-metals (e.g., Al, Ta, Nb, Zr, etc.) are anodized
under galvanostatic conditions, the thickness of the oxide lms
increases linearly with time. Correspondingly, the applied
potential (U) increases linearly with time to keep the electric
eld (E) constant during the process. Under this condition, the
anodizing potential (U) nally reaches a value at which visible
sparking on the anode starts appearing, and local thickening,
cracking, blistering, or even burning of oxide lm commences.
This local event is called electrolytic breakdown, which not
only prevents the uniform growth of anodic lms over the
macroscopic metal surface, but also permanently degrades the
dielectric properties of the oxide. The anodizing potential at the
onset of this local event is called breakdown potential (UB).
Because the oxide thickness increases linearly with the
anodizing potential (U) in galvanostatic conditions, the
breakdown is dependent on the oxide thickness and occurs at
a critical oxide thickness. Breakdown during anodization can be
associated with a number of phenomena. These include the
appearance of visible sparking/luminescence,5159 the local
crystallization of oxide, 6066 oxygen evolution at the
anode,63,67,68 retardation of potential rise,69,70 occurrence of
audible cracking,71 and rapid voltage uctuations.69,70,72 In
porous AAO growth, breakdown can occur under high current
density anodizing conditions.4 If the reaction heat cannot be
adequately dissipated from the anode, electrolyte heating may
cause local increase in conductivity and a current run away
process. This results in local thickening or burning of anodic
oxide, terminating uniform growth of porous AAO. The anodic
oxide in the burnt area exhibits typically a dierent color from
the burnt-free areas. For a given anodizing electrolyte, on the
other hand, porous AAO formed at a potential just below
breakdown value (i.e., U < UB) exhibits the best self-ordering of
pores (section 7.1).18 Improving the breakdown characteristics
of anodic oxide lms through proper control of the electrolyte
composition, surface state of the starting aluminum, and
reaction heat can allow one not only to explore new anodizing
conditions for self-ordered pore growth, but also to engineer
internal pore structures (see sections 7.27.4). In this section,
we discuss some of the electrochemical factors inuencing
breakdown, and models that explain the breakdown phenomena.

(22)

where
zi, and V i are the standard chemical potential, the
charge number, and the molar volume of ion i, respectively. For
barrier-type anodic alumina lm, the mean normal stress is
dened according to = 1/3(xx + yy) = 2/3xx, where x- and
y-directions are parallel to the interface.47 For the stress
gradient-driven oxide creep, the model enforces the conservation of electrical charge and volume and the momentum
balance in a Newtonian uid. For galvanostatic anodization of
aluminum at the applied current density j, the constraint of
charge conservation can be written as follows:
u0i ,

j = 2FJO + 3FJM

(23)

On the other hand, the volume balance is


jM
= VO JO v
3F

(24)

where M is the molar volume of the Al atom in the metal and


v is the creep velocity in the oxide. By employing the Maxwell
viscoelastic model and also by assuming a large elastic modulus,
the momentum balance in a Newtonian uid is expressed as47
0=

1
1
+ 2 v + (v)

(25)

where is the viscosity.


For porous AAO lm growing under steady-state,47 the
model predicted that a large compressive interfacial stress
causes the lateral ow of oxide materials from the center of
pore base toward the cell boundaries and the upward ow in
the pore wall oxide, as in the ow pattern experimentally
observed from W tracer studies (see section 6.3.5). Simulation
results indicated that the stress eld driving the ow results
from the following three origins: the volume expansion
occurring at the metal/oxide interface, nonlinearity of the
equations governing conduction of mobile ions (i.e., Al3+ and
O2/OH), and incorporation of electrolyte-derived anionic
species within the anodic oxide near the oxide/electrolyte
interface.47
For barrier-type anodic alumina lm,48 the model predicted
the average stress in the oxide to be compressive when the
current density is smaller than 0.5 mA cm2 (i.e., compressive

4.1. Factors Inuencing Breakdown

In general, the breakdown potential (UB) is dependent on the


nature of the metal being anodized, the current density (j), and
the composition (or resistivity) of the electrolyte. Meanwhile,
7493

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

lm. For anodic oxide of aluminum, some authors have


reported that anion concentration (CA) inuences UB.76,77
Kato et al. showed that at a xed solution resistivity UB
decreases linearly with an increase in the logarithm of the
anion concentration, or more specically the anion charge with
the following relation:77

the electrolyte temperature, stirring rate, and history of anodic


oxide have no inuence on the breakdown potential.
4.1.1. The Nature of Anodized Metal. Wood and
Pearson investigated metals whose anodization in 3%
ammonium tartrate ended in sparking, and associated the
breakdown potential (UB) with the ionic bonding characteristics of the anodic oxides by employing the criteria of Pauling
and Wells. They established a descending order of UB according
to the melting point of the corresponding oxide: Zr (300 V) >
Al (245 V) > Ta (200 V) > Nb (190 V).72 However, Alwitt and
Vijh reported a dierent descending order of UB for
anodizations of the same metals in the same conditions: Al
(350 V) > Zr (315 V) > Ta (275) > Nb (190 V).73 They
correlated the increase in UB with the increasing heat of
formation per equivalent (Hf/equiv) of oxide, which is
approximately equal to one-half the value of the forbidden band
gap of the oxide. Further, they noted that the dependence of UB
on the band gap would reect the electronic nature of the
breakdown phenomena. As such, rather conicting reports have
been published for the dependence of UB on the intrinsic solidstate properties of anodic oxides. Iknopisov et al.69 pointed out
that the dependence of UB on the nature of the metal is
considerably smaller than the dependence on the electrolyte
resistivity (e).
4.1.2. Electrolyte Conditions. Early studies have reported
that the breakdown potential (UB) increases linearly with the
logarithm of the electrolyte resistivity (e) with the following
equation:

UB = A + B log e

UB = A B log C A

(27)

On the basis of anodization experiments with tantalum in


sulfuric, phosphoric, and hydrochloric acids, Arifuku et al.78
reported that UB is dependent upon the detailed distribution
proles of incorporated anions in the anodic oxide. Later, the
role of incorporated electrolyte species in the electrical
breakdown was emphasized by Albella et al., who have put
forward a theory of avalanche breakdown during anodic
oxidation.7981
4.1.3. Current Density (j). For tantalum anodization in
ammonium sulfate, Yahalom and co-workers76,82 reported that
the breakdown potential (UB) is almost independent of the
current density (j). For anodic lms on aluminum, Ikonopisov
et al.69 also reported that a 500-fold increase of current density
(j) only lowers the breakdown potential (UB) by 15%. On the
other hand, Di Quarto et al.74,75,83 pointed out the occurrence
of two dierent kinds of breakdown, that is, mechanical and
electrical breakdown. For anodic oxides of tungsten,74
zirconium,75 and titanium84,85 under limited conditions, they
noted that anodic oxides grew with an increasing number of
defects at a retarded rate (i.e., reduced slope in Ut curve)
during galvanostatic anodizations, until electrical breakdown
(EB) eventually occurs with visible sparks. They termed this
characteristic growth as mechanical breakdown (MB). For
electrical breakdown (EB), they reported that current density
(j) has little eect on the breakdown potential (UEB), which is
in line with the reports of Yahalom et al. and Ikonopisov et
al.69,76,82 In the case of mechanical breakdown (MB), however,
they observed that current density (j) has a signicant eect on
the breakdown potential (UMB) according to the following
equation:

(26)

where A and B are the constants depending on the electrolyte


composition and the anodized metal.69,71,74,75 Figure 7 shows
the dependence of UB on log e during anodization of Nb, Ta,
Al, and Zr.69 It appears from the gure that the dierent
inuences of e on UB defeat attempts to set the metals in series
with respect to the breakdown characteristics of their anodic

UMB = AMB + BMB log j

(28)

where AMB and BMB are constants, which depend mainly on the
kind of anion in the electrolyte and slightly upon pH and
concentration of electrolyte: BMB > 0 for anodic oxides of
zirconium and titanium75,85 and BMB < 0 for anodic oxide of
tungsten.74
4.1.4. Other Factors Inuencing Breakdown. The
surface state of the starting metal (i.e., the surface defects
(aws), purity, processing history, etc.) also strongly inuences
the breakdown potential (UB).61,86 In general, the surface
defects unavoidably cause a decrease of the breakdown
potential (UB) with the commencement of sparks.87 On the
other hand, post-breakdown anodization experiments have
shown that breakdown characteristics are independent of the
history of the anodic oxide lm.72,88,89 When a valve-metal was
anodized in electrolyte A until breakdown occurred at UB,A, and
then the resulting sample was re-anodized in electrolyte B with
a higher breakdown potential (UB,B), the lm formation during
the post-breakdown anodization continued at normal kinetics
until breakdown occurred at UB,B.88,89 Temperature (T) is one
of the easily controllable parameters of the electrolyte.
Ikonopisov formulated the temperature dependence of the
breakdown potential UB (section 4.2.1).90 However, a change in
the electrolyte temperature can alter both the electrolyte

Figure 7. Dependence of the breakdown potential (UB) on the


logarithm of electrolyte resistivity (e) for anodizations of Ta, Nb, Al,
and Zr in solutions of ammonium salicyalte in dimethylformamide.
Reproduced with permission from ref 69. Copyright 1979 Elsevier.
7494

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

resistivity (e) and the property of the growing anodic oxide.


When the dependence of electrolyte resistivity (e) is
considered, no clearly pronounced dependence of the breakdown potential (UB) on the temperature (T) was obtained.71,77,91
4.2. Models for Breakdown

4.2.1. Electron Avalanche Multiplication. The rst


attempt to develop a quantitative model of breakdown was
made by Iknopisov.90 He considered experimentally observed
breakdown characteristics, and noting that the breakdown
potential (UB) mainly depends on the nature of the anodized
metal and the electrolyte resistivity (e), he inferred that
breakdown is dependent upon the solid-state properties of the
anodic oxide and is controlled by electrochemical reactions at
the oxide/electrolyte interface. In his model, the initial
electrons are injected from the electrolyte into the oxide
conduction band (CB) by either a FowlerNordheim or a
Schottky mechanism (Figure 8). The injected electrons

je, x = 0 = 1 exp[2E1/2]

(31)

ln je, x = 0 = 1/T + 2

(32)

je, x = 0 = 1e2

(33)

From eqs 30 and 31, the dependence of breakdown potential


(UB) on the electric eld (E) is given by
UB = (i /rq)(ln je,B ln 1) (i2/rq) E

(34)

This equation explains a slight decrease of UB with increasing


current density (j). Regarding the relation between the
breakdown potential (UB) and temperature (T), the following
expression is obtained by combining eqs 30 and 32:
UB = (i /rq)(ln je,B 1/T 2)

(35)

Equation 35 predicts that UB is dependent on the temperature


(T), which conicts with experimental observations.71,77,91 For
this discrepancy, Ikonopisov pointed out the interplay between
temperature (T) and the solution resistivity (e). For the
dependence of breakdown potential (UB) on the electrolyte
resistivity (e), from eqs 30 and 33, one may obtain
UB = (i /rq)(ln je,B ln 1 + 2 ln e )
= (i /rq)(ln je,B ln 1) + (2.3i2/rq) log e

(36)

Equation 36 has exactly the same form as eq 26, which


describes an empirical relation between UB and e.
Although Ikonopisovs model explains some of the
experimental results, it has been criticized by many authors.
Shimizu pointed out the unrealistic value of the mean free path
(E) of ionized electrons, which from eqs 26, 29, and 36 is
given by
(E) = 1/(E) = i /rqE = (1/2.3E)(B /2)

By using the experimental values from Ikonopisov et al. for E


(=8.7 106 V cm1) and B/2 (=1000 V), Shimizu obtained
= 500 nm, which roughly corresponds to the thickness of oxide
lms formed up to the potential 400 V and indicates the
absence of the electron avalanche capable of causing the
breakdown.94 Albella et al. questioned the origin of electrons in
Ikonopisovs model.87 They pointed out Ikonopisovs model
lacked a reasonable explanation of the role of the electrolyte
and the absence of specic electrochemical reactions required
for the injection of the initial electrons.
Albella et al. explicitly considered the eect of the anodizing
electrolyte by posulating that the initial electrons for the
avalanche come from the electrolyte species incorporated into
anodic oxide.7981 The incorporated electrolyte species act as
impurity centers close to the oxide conduction band (CB),
releasing electrons to the conduction band via the eld-assisted
PooleFrenkel mechanism (Figure 9).80,87,95,96 In the model of
Albella et al., the total current density (jt) consists of three
components:

Figure 8. Schematic representation of the band structure and the


avalanche breakdown in Ikonopisovs model. Reproduced with
permission from ref 79. Copyright 1984 The Electrochemical Society.

accelerate and multiply in avalanche during their travel in the


oxide of thickness (tox) to the anode, until the avalanche current
reaches a critical value for breakdown. In this multiplication
process, the electronic current (je) depends on the travel
distance (x) with x = 0 being the oxide/electrolyte interface,
which can be expressed by
je, x = t = je, x = 0 exp[(E)tox ] = je, x = 0 exp[rqEtox /i]
ox

(29)

where (E) is the impact ionization coecient at the electric


eld E, r is a recombination constant (r < 1), q is the electron
charge, and i is the threshold energy for impact ionization.
Breakdown occurs if the electronic current (je) exceeds a critical
value je,B at a critical oxide thickness (tox,B). Because UB = Etox,B,
the breakdown potential (UB) is given by
UB = (i /rq)(ln je,B ln je, x = 0 )

(37)

jt = j1 + j2 + je

(38)

where j1 is the oxidation current density, j2 is the current


density consumed by the incorporated electrolyte species and is
assumed to be a constant faction of j1 (i.e., j2 = j1), and je is
the electronic current density. The electronic current density
(je) at the anode can be expressed as

(30)

The dependences of je,x=0 on the electric eld (E),69,92


temperature (T), and electrolyte resistivity (e)93 were
empirically determined to be, respectively:
7495

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

V (t ) =

1 +
Kj t
1+ t

V (t ) =

E
[exp(U /E) 1]
1+

(45)

(46)

The factor (1 + )/(1 + ) in eq 45 describes the correction


of the anodizing rate due to the incorporation of electrolyte
species. On the other hand, eq 46 enforces deviation of
potential from the linearity due to the avalanche eect.
Accordingly, eq 44 predicts a gradual decrease of slope (dU/
dt) of the potentialtime curve during galvanostatic anodization. Albella et al. conrmed the validity of eq 44 by tting it on
the experimental results of tantalum anodization (Figure 10).80

Figure 9. Band diagram showing the avalanche multiplication of


electrons in the model of Albella et al. The impurity level in
conduction band (CB) is denoted by A. Reproduced with
permission from ref 81. Copyright 1987 Elsevier.

je = je, x = 0 exp[tox ]=je, x = 0 exp[U ]

(39)

where is the impact ionization coecient, and is the ratio of


the oxide thickness to the anodization potential (U) and is
equal to the inverse of the electric eld E (i.e., = 1/E = AR,
the anodizing ratio). Because the initial electronic current
originates from incorporated species, je,x=0 should be a constant
fraction of oxyanion current (j2) and is je,x=0 = j2 = j1.
Under the assumption that the critical current density is a
fraction z of the oxidation current j1, the breakdown potential
(UB) should satisfy the following relation:

je, x = 0 exp[UB] = zj1

(40)

Figure 10. Experimental result for the evolution of the potential (U)
as a function of time (t) during tantalum anodization in 1.2 M H3PO4
at 1.78 mA cm2. The theoretical curve (solid line) has been tted
according to eqs 4446. Reproduced with permission from ref 80.
Copyright 1985 Elsevier.

Accordingly, the breakdown potential (UB) is given by


UB = (1/) ln(zj1 /j0 ) = (E /) ln(z /)

(41)

The time derivative of the potential is given by


1 E M
M2
dU
= 1 j1 +
j
dt
x 2y2 2
ox F x1y1

Further, by tting eq 44 on the experimental potential


evolutions in dierent electrolyte concentrations (C), they
obtained a relation between and C:95

(42)

where ox is the oxide density, F is the Faraday constant, and


M1 and M2 are the molecular weights of the oxide and the
incorporated species, respectively, whose corresponding anion
and cation valences are x1,y1 and x2,y2, respectively. Combining
eqs 38 and 42 yields the following dierential equation:
dU /dt = Kjt (1 + )[1 + + exp(U )]1

aC b

with a and b being electrolyte-dependent constants. From eqs


41 and 47, the concentration dependence of the breakdown
potential (UB) is given by
UB (E /)[ln(z /a) b ln C ]

(43)

(48)

which is in good agreement with the experiments.


4.2.2. Stress-Driven Breakdown. Sato97 distinguished ve
dierent possible contributions to the mechanical stresses in
anodic oxide: (a) electrostriction pressure, (b) interfacial
tension of the lm, (c) internal stress caused by the volume
expansion, (d) internal stress due to partial hydration/
dehydration of the anodic oxide, and (e) local stress caused
by impurities. By considering the rst two contributions as the
most general factors for breakdown, he mathematically derived
a thermodynamic model of stress-driven breakdown, and

where K is the unitary rate of anodization for oxide without


electrolyte incorporation and given by M1E/x1y1oxF, and is
the ratio of the equivalent weight of the incorporated species to
that of oxide, that is, = (M2/x2y2)/(M1/x1y1). Assuming a
constant eld in the oxide, the integration of eq 43 yields the
relation between the anodizing potential (U) and time (t):
U (t ) = V (t ) V (t )

(47)

(44)

with V(t) and V(t) given by


7496

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

lm pressure. The alignment of pores in a close-packed pattern


was considered to be a process of relieving the lm stress. As
the breakdown in the anodic oxidation of aluminum proceeds,
therefore, a porous oxide layer progressively forms and thickens
on the compact barrier oxide layer, of which thickness remains
constant with continuous plastic deformation. This electrostriction-stimulated breakdown model is somewhat in line with
the recent ow model accounting for the steady-state formation
of porous AAO lm by Skeldon et al. (see section 6.3.5).98

explained the breakdown potential (UB) and the eect of


adsorbed anionic species on it. According to the model,97 the
stress (P) accumulated in the oxide lm is equal to
P =

( 1)

8
tox

(49)

where is the oxide permittivity and is the surface tension. In


eq 49, the rst term represents the electrostriction eect and
the second the interfacial tension eect. According to Sato, for
an oxide dielectric with > 10, an electric eld (E) of 5 106 V
cm1 produces a compressive electrostriction pressure exceeding 1000 kg cm2, which is higher than the critical mechanical
strength of oxides and thus may be a cause of their mechanical
failure. Because the compressive stress within the anodic oxide
increases with thickness, there is a limiting oxide thickness
(tox,B) above which breakdown occurs. The incorporated
anionic species causes a decrease in the surface tension (),
and thus increases the stress by lowering the value of the
second term in eq 49. The dependence of the breakdown
potential (UB) on the anion concentration (CA) in the
electrolyte was also established, as follows:
dUB
8kT
=
A
( + 1)
d ln C A

5. STRUCTURE OF POROUS ANODIC ALUMINUM


OXIDE (AAO)
5.1. General Structure

Figure 12 shows schematically an idealized structure of porous


AAO, together with scanning electron microscopy (SEM)
images of each part of the porous AAO. Porous AAO has a
honeycomb-like structure. Porous oxide layer formed on
aluminum substrate contains a large number of mutually
parallel pores. Each cylindrical nanopore and its surrounding
oxide constitute a hexagonal cell aligned normal to the metal
surface. Each nanopore at the metal/oxide interface is closed by
a thin barrier oxide layer with an approximately hemispherical
morphology. Under proper anodization conditions, the oxide
cells are self-organized to form a hexagonally close-packed
structure.7 On the other hand, the surface of the aluminum
after complete removal of the porous oxide layer is textured
with arrays of concave features. The thickness of the porous
AAO layer on aluminum is proportional to the total charge
(Qc) involved in the electrochemical oxidation. Therefore, the
depth of oxide nanopores is easily tunable from a few tens of
nanometers up to hundreds of micrometers by controlling
anodization time (t). In general, the structure of self-ordered
porous AAO is often dened by several structural parameters,
such as interpore distance (Dint), pore diameter (Dp), barrier
layer thickness (tb), pore wall thickness (tw), pore density (p),
and porosity (P). For ideally ordered porous AAO, the
following relationships can be drawn by simple geometric
consideration:

(50)

where A is the anion density at the oxide surface at the


breakdown potential (UB). The model predicts a lower
breakdown potential for electrolytes having higher anion
concentration. Kato et al. have also used electrostriction to
explain the enhancement of breakdown by incorporated anionic
impurities (see eq 27).77 They suggested that the incorporated
anions lead to additional electrostrictive input into the
mechanical stress in oxide lms.
Sato noted three dierent forms of mechanical breakdown
depending on the mechanical property of the lms: brittle crack
for rigid anhydrous anodic oxides, and plastic deformation or
ow for visco-plastic hydrous anodic oxides (Figure 11).97 He
suggested that the formation of porous AAO lms on
aluminum is associated with continuous mechanical breakdown,
accompanied by a continuous plastic ow of oxide under high

Dint = Dp + 2tw

(in nm)

(51)

2
1014 cm2
P =
2
3 Dint

(52)

Dp
P(%) =

100
2 3 Dint

(53)

These structural parameters of porous AAO are known to be


dependent on the anodizing conditions: the type of electrolyte,
anodizing potential (U), current density (j), temperature (T),
etc. Among those, anodizing potential (U) and current density
(j) are the most important electrochemical parameters. A
review on this matter has recently been published by Sulka.99
Here, we briey discuss the major structural parameters of
porous AAO and the electrochemical factors inuencing them.
5.1.1. Pore Diameter (Dp). OSullivan and Wood used
electron microscopy to quantitatively study the morphology of
porous AAO potentiostatically formed in phosphoric acid
(H3PO4) electrolyte.100 The pore diameter (Dp), interpore
distance (Dint), and barrier layer thickness (tb) were observed to
be directly proportional to the anodizing potential (U). Their
microscopic analysis revealed that the pore diameter increases

Figure 11. Three modes of mechanical breakdown of surface lms.


Reproduced with permission from ref 97. Copyright 1971 Elsevier.
7497

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 12. Schematic structure of (a) porous anodic aluminum oxide (AAO) on Al foil and (b) cross-sectional view. (ce) SEM images of porous
AAO, showing top surface, barrier layer, and bottom surface, respectively. Scale bars are 1 m. Panels ce were reprinted with permission from ref
111. Copyright 2006 Macmillan Publishers Ltd.: Nature Materials.

at the rate (p) of 1.29 nm V1 with respect to the anodizing


potential (U):
DP = pU = 1.29U

(54)

They pointed out that the electrolyte concentration does not


signicantly inuence the pore diameter (Dp), while the
temperature of the electrolyte is positively correlated with
pore diameter.100 On the other hand, theoretical modeling of
porous AAO growth performed by Parkutik and Shershulsky
predicted a decrease in pore diameter with decreasing
electrolyte pH (i.e., increasing electrolyte concentration) due
to the enhanced dissolution velocity of anodic oxide at the pore
base.101 Moreover, a recent study by Sulka and Parkoa
indicated that the pore diameter decreases with decreasing
temperature.102 In general, pore diameter close to the surface of
a porous AAO lm is larger than that close to the pore bottoms
(i.e., truncated pore channels), especially when anodization is
conducted at an elevated temperature and/or for an extended
period of time. This can be attributed to the chemical
dissolution of the pore wall oxide by acid electrolyte.
Accordingly, pore diameter measured from the pore bottom
(not from the surface of AAO lm) is more relevant for
investigating the intrinsic eect of electrochemical parameters
on the structure of porous AAO.
Recently, Lee et al.103 reported that pore diameter (Dp)
increases with current density (j) under potentiostatic
anodization conditions (Figure 13). Under specic experimental conditions, they observed spontaneously oscillating current

Figure 13. Cross-section SEM micrographs of AAOs prepared from


two separate anodization experiments, whose reactions were
terminated near j = 86 mA cm2 and j = 881 mA cm2 in sinusoidally
oscillating currents under potentiostatic condition (U = 200 V). (c) A
schematic cross-section of AAO on Al. (d) The parameters dening
the geometry of the pore bottom. Scale bars = 250 nm. Reproduced
with permission from ref 103. Copyright 2010 Wiley-VCH Verlag
GmbH & Co. KGaA, Weinheim.

during potentiostatic hard anodization (HA) at the potential


range of 140200 V (see section 7.2). They suggested that at a
given potentiostatic condition (i.e., U = constant) the
7498

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

V1.7,100,114 For self-ordered porous AAOs formed by mild


anodization (MA, see section 7.1) in oxalic and phosphoric acid
at various anodizing potentials (U), Vrublevsky et al. have
reported empirical equations based on results from their reanodizing experiments.115117 They used the equations for
estimating the barrier layer thickness by assuming the anodizing
ratio, ARMA = 1.14 nm V1. On the other hand, reduced
anodizing ratios for sulfuric and oxalic acid have recently been
reported for hard anodization (HA); ARHA = 0.61.0 nm V1
(see Figure 13d and section 7.2).18,111,113,118,119 Chu et al.18
determined the anodizing ratio for less-popular anodizing
electrolytes (e.g., tartaric, citric, glycolic, and malic acids) by
performing what they termed critical-potential anodization.
The anodizing ratio for various electrolytes was determined to
be AR 1 nm V1 (Figure 14). It should be noted that this
value of anodizing ratio is the averaged proportionality constant
determined from mild and hard anodization experiments of
aluminum.

distribution of the current lines and electric eld (E) may be


sensitively varied by the geometric details of the barrier oxide
layer. That would inuence the movement rates of the
electrolyte/oxide and oxide/metal interfaces, hence the pore
diameter (Dp).103
5.1.2. Interpore Distance (Dint). It has long been
established that the interpore distance (Dint) is also linearly
proportional to the anodizing potential (U).7,100,104109 A
detailed study on this matter was performed for sulfuric and
oxalic acid by Ebihara et al.104,105 Their empirical expressions
on the relationship between the interpore distance (Dint) and
anodizing potential (U) are as follows:
Dint = 12.1 + 1.99 U

for sulfuric acid:

(U = 318 V)
(55)

for oxalic acid:

Dint = 14.5 + 2.00 U

(U 20 V)
(56)

=1.70 + 2.81U

(U 20 V)

(57)

For oxalic acid-based anodizations in the potential range of 20


60 V, Hwang et al. reported that interpore distance only
depends on anodizing potential (U), not on the temperature of
the electrolyte:107
for oxalic acid:

Dint = 5.2 + 2.75 U

(U = 2060 V)
(58)

This temperature independence of the interpore distance is in


line with the results of Keller et al.,7 but conicts with the
experimental results of Sulka and Parkoa,102 who observed that
interpore distance is positively correlated with temperature for
self-ordered porous AAOs formed by sulfuric acid-based
anodization; interpore distance at an elevated temperature
(10 C) is about 10% larger than that at a low temperature (i.e.,
8 to 1 C). OSullivan and Wood100 reported for phosphoric
acid-based anodization that increasing the temperature or the
electrolyte concentration decreases the interpore distance. For
self-ordered porous AAOs formed by mild anodization (MA)
conditions using sulfuric, oxalic, and phosphoric acid, it has
generally been accepted that the interpore distance (Dint) is
linearly proportional to the anodizing potential (U) with a
proportionality constant MA of 2.5 nm V1 (see section
7.1):109
Dint = MA U = 2.5U

Figure 14. Eect of anodizing potential (U) on the barrier layer


thickness (tb) for porous AAO formed in dierent acid electrolytes.
(Solid symbols, measured values; open symbols, calculated values from
the half-thickness of the pore walls). Reproduced with permission
from ref 18. Copyright 2006 The Electrochemical Society.

(59)

5.2. Structure of Pore Wall (Anion Incorporation)

However, this empirical formula is not valid for hard


anodization (HA), under conditions in which a high electric
eld (E) is exerted across the barrier layer due to high current
density (j) during anodization.16,110113 This will be discussed
in detail in section 7.2.
5.1.3. Barrier Layer Thickness (tb). The thickness of the
barrier layer (tb) is one of the most important structural
parameters of porous AAO for understanding the kinetics of
the electrochemical oxidation of aluminum. Like other
structural parameters, barrier layer thickness (tb) is also
dependent on the anodizing potential (U). The potential
dependence of the barrier layer thickness has also been known
as anodizing ratio (AR = tb/U), the inverse of which
corresponds to the electric eld (E) across the barrier layer, and
it determines the ionic current density (j) (see eq 1).
Accordingly, at a given anodizing potential (U), current density
(j) increases exponentially as a function of the inverse of the
anodizing ratio AR (i.e., the electric eld strength E). Earlier
studies have indicated that the anodizing ratio equals 1.2 nm

The incorporation of electrolyte-derived anions into anodic


alumina is considered a general phenomenon for both barrierand porous-type anodization, occurring least for the former and
greatest for the latter.3 For three major pore-forming acid
electrolytes (e.g., H2SO4, H2C2O4, and H3PO4), incorporation
of acid anions occurs via inward migrations under an electric
eld (E) during the anodization of aluminum. The incorporated acid anions inuence the chemical, optical, and
mechanical properties of the resulting porous AAO. For
example, incorporated oxalate (C2O42) anions together with
singly ionized oxygen vacancies (F+ center) have been known
to contribute to the blue photoluminescence (PL) of porous
AAO formed in oxalic acid solution.120122 The mechanical
properties (e.g., hardness, wear resistance, and elasticity) of
anodic alumina are also known to be aected by the
incorporated chemical species (e.g., water and acid
anions).123125 The amount of incorporated acid anions and
their distribution in anodic alumina depend on the anodization
potential (U), current density (j), and temperature (T), as well
7499

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 15. (a) Schematics illustrating the duplex structure of pore walls of porous AAO: vertical (left) and transverse (right) cross-sections. TEM
plane view (b) of H3PO4-AAO and the corresponding X-ray maps of the elements: (c) phosphorus, (d) oxygen, and (e) aluminum. (f) TEM plane
view of H3PO4-AAO, showing the dierent parts of the pore wall (i.e., the outer pore wall, cell-boundary band, and interstitial rod). Reproduced with
permission from ref 135. Copyright 2009 Elsevier.

as the type and concentration of electrolytes.44,117,126129


Accordingly, the chemical structure of the pore wall of AAOs
varies with the anodization conditions. Han et al.129 have
recently reported that, even at a steady-state growth condition
(i.e., xed U, j, and T), the content of anionic impurities and
their incorporation depth decrease as a function of anodization
time due to the progressive reduction of the electrolyte
concentration, which markedly aects pore widening as well as
the opening of the barrier oxide layer by wet-chemical etching.
On the basis of TEM investigations of disordered porous
AAOs formed in sulfuric, oxalic, phosphoric acid solutions,
Thompson and co-workers suggested that pore wall oxide has a
duplex structure in terms of chemical composition: an acidanion contaminated outer oxide layer next to the pores and a
relatively pure inner oxide layer, as schematically shown in
Figure 15a.130 TEM micrographs of porous AAO lms formed
in phosphoric and oxalic acids (i.e., H3PO4-AAO and H2C2O4AAO) showed a cell structure with cell-boundary bands,
although the cell-boundary band in H2C2O4-AAO did not
appear as well-dened as that observed in H3PO4-AAO. The
cell-boundary bands became markedly more apparent upon
continued exposure to the electron beam due to the preferential
crystallization of the cell boundary regions. Meanwhile, the
presence of a cell-boundary band was not apparent for porous
AAO lms formed in sulfuric acid (i.e., H2SO4-AAO). The
duplex structure of the pore walls was conrmed by Ono and
Masuko,131 who reported that the depth and the amount of
anion incorporation in H3PO4-AAO increased linearly with
anodizing potential (U), but the presence of a duplex structure
was not conrmed by TEM for samples formed at U < 10 V.
They also found that the crystallization rate of pore wall oxide
under a strong electron-beam irradiation decreases with the
increasing content of incorporated electrolyte species.132
Scanning transmission electron microscopy (STEM) and
energy dispersive X-ray (EDX) point analysis on H3PO4-AAO
by Thornton and Furneaux revealed that the cell-boundary

bands are composed of relatively pure alumina, whereas the


material adjacent to the pores contains incorporated phosphate
species from the electrolyte.133,134 Recent microscopic chemical
analyses of the pore wall material of highly ordered H3PO4AAO by Le Coz et al.135 clearly indicated the presence of
phosphorus-free cell-boundary bands (Figure 15be). The
dierent parts of the unit cell were found to have a
heterogeneous chemical composition of Al2O30.197AlPO4
0.034H2O, which supports the results of the previous works by
Thompson et al.10,130 The work also highlighted, as a new
nding, that there is an interstitial rod material with a
composition of Al2O30.018AlPO4xH2O at the triple junction
connecting three cells (Figure 15f).135
Thompson and Wood related the steady-state anodizing
behavior of porous AAO lms formed in the major anodizing
acids to the distribution of the acid anions within the barrier
layers and the true eld strengths across the relatively pure
alumina regions.14 Starting with the knowledge that the
thickness ratio of the inner to outer pore wall oxide layer
increases in the order sulfuric acid < oxalic acid < phosphoric
acid < chromic acid, they depicted the same order of thickness
ratio for the barrier oxide at the bottom of pores and correlated
it to the rates of oxide formation at the same anodizing
potential (U). The averaged eective electric eld (E = U/tb)
across the barrier layer is approximately constant for the barrier
layers of AAOs formed in dierent acid electrolytes, because
the measured anodizing ratios are similar (ARMA = tb/U 1.2
nm V1). On the other hand, the potential drop (U) is
greater and linear across the relatively pure alumina region and
smaller across the outer acid anion-contaminated region, where
the potential decreases progressively toward the oxide/
electrolyte interface (Figure 16). Therefore, the true electric
eld across the relatively pure alumina region of the barrier
layer is in the order:
7500

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

H 2SO4 AAO > H 2C2O4 AAO > H3PO4 AAO


> H 2CrO4 AAO

Figure 16. Distribution of the potential drop and electric eld, E


(slope of the voltagedistance plot), across barrier layers of porous
AAOs formed in (a) sulfuric, (b) oxalic, (c) phosphoric, and (d)
chromic acid. Reproduced with permission from ref 14. Copyright
1981 Macmillan Publishers Ltd.: Nature.

Figure 17. (a) The evolution of pore diameter (Dp) as a function of


time (tetch) upon wet-chemical etching of porous AAOs formed in 0.3
M oxalic acid at 40 V. Wet-chemical etchings of pore wall oxide were
performed in 5 wt % H3PO4 at 29 C. The numbers in the plot are the
slopes (in nm min1) of the corresponding linear ts. (b) SEM images
showing systematic increase in pore diameter (Dp) as a function of
time (tetch) upon wet-chemical etching in 5 wt % H3PO4 (29 C).
Reproduced with permission from ref 129. Copyright 2013 The
American Chemical Society.

Accordingly, the rate of formation of the anodic oxide lms


in the dierent acids varies with the same order given above if
solid-state ionic migration across the inner layer is the ratedetermining step.14 The current density (j) is related to the
potential drop (U) across the barrier oxide by the high-eld
conduction theory (eq 1). Among the acid electrolytes above,
sulfuric acid will give the highest current density (j), because of
the highest potential drop across the relatively pure alumina
region adjacent to the metal interface.
The duplex nature of the pore wall can be experimentally
evidenced by investigating the rate of pore widening. For a
given set of etching conditions (i.e., temperature and
concentration of an etchant solution, typically H3PO4), the
rate of etching is dependent on the chemical composition of the
pore wall oxide of AAO.129 Figure 17 shows (a) the evolution
of pore diameter (Dp) as a function of pore wall etching time
(tetch) for porous AAO formed in 0.3 M oxalic acid (H2C2O4),
together with (b) representative SEM micrographs. As
presented in Figure 17a, Dp versus tetch plot is characterized
by an inection point, at which the slope of the curve changes.
Pore wall oxide in the early stage is etched at a higher rate (1.04
nm min1) than that (0.36 nm min1) in the later stage. The
retarded rate of etching in the later stage can be attributed to
the relatively pure nature of the inner pore wall oxide, as
compared to the less dense outer pore wall oxide due to the
incorporation of anionic species. As shown in Figure 17b, the

ability to precisely control the pore diameter by the pore


widening process is one of the most attractive features of
porous AAO for template-based nanofabrication. This feature
allows one to systematically investigate the size dependence of
chemical or physical properties of ordered arrays of nanodots,
nanowires, or nanotube materials prepared from porous AAO
templates.
As-prepared porous AAOs are amorphous and contain
varying amounts of water depending on anodizing condition.3
The local coordination environments of aluminum in
amorphous AAOs have been extensively studied using X-ray
radial distribution analysis,136 electron-yield extended X-ray
absorption ne structure (EXAFS) spectroscopy,137 and magicangle spinning nuclear magnetic resonance (27Al MAS
NMR).138141 Amorphous AAOs have been considered to
have a close structural relation to spinel (fcc) -Al2O3 with
tetra- and hexa-coordinated aluminum cations in the mixing
ratio of 1:2. Farnan et al.138 have reported that the coordination
numbers of aluminum cations are dependent on the anodizing
electrolyte: hexa-coordination for porous AAO formed in
chromic acid (i.e., H2CrO4-AAOs), tetra-, penta-, and hexa7501

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

thermodynamically stable hcp -Al2O3 (corundum) is obtained


above 1150 C. Crystalline phase transformations of porous
AAOs formed in dierent electrolytes have also been
investigated by thermogravimetric (TG) and dierential
scanning calorimetry (DSC) analyses.139,140,144,147151 The
results have indicated that the type of anodizing electrolytes
(i.e., the nature of incorporated acid anions) strongly inuences
transformation temperatures during heat treatment. For
H2C2O4-AAOs, Mardilovich et al.144 reported that dehydration
occurs up to 400 C, dehydroxylation occurs at 400700 C,
and incorporated oxalates pyrolyze at higher temperatures
before the nal transformation into thermodynamically stable
-Al2O3. A comparative study of H2SO4-, H2C2O4-, and
H3PO4-AAOs by Mata-Zamora et al.150 has shown that
incorporated phosphates are stable up to 1400 C, while
sulfates and oxalates decompose into gaseous SO2 and CO2 at
900 and 870 C, respectively. Later, Kirchner et al.140
conrmed decomposition of sulfates in H2SO4-AAOs at around
900 C by mass spectrometry (MS).
In general, the initial crystallization and subsequent phase
transformation of porous AAO are accompanied by changes in
the morphology of pores and also in mechanical properties.139,144,151 MacQuaig et al. reported 6.7% loss of pore
circularity, 15% increase in pore diameter (Dp), 13%
decrease in pore density (), and about a 2-fold increase in
microhardness (from 2.5 to 4.7 GPa) upon heat-treatment of
as-prepared H2C2O4-AAO up to 1200 C.151 They attributed
the changes of the overall pore structure to the densication of
pore wall materials, which is associated with dehydration,
dehydroxylation, and the loss of incorporated acid anions in the
course of the phase transformations from amorphous to
crystalline -Al2O3. Mardilovich et al. observed sharp decreases
in the exibility of H2C2O4-AAO lms at the onset of
crystallization at 820840 C and on phase transformation
from metastable ccp-alumina to hcp-alumina (i.e., -Al2O3) at
11001150 C.144 They pointed out that the fragility of porous
-Al2O3 membranes is so high that they cannot be considered
for routine practical use as membranes. High temperature
treatment of initially planar porous AAO membranes often
leads to serious mechanical deformation (e.g., curling/rolling)
or even cracking.139,144,148 Therefore, careful heat-cycling
processing is required. Recently, Chang and co-workers
reported that severe deformation of porous AAO membranes
can be prevented by removing the acid-anion contaminated
outer pore wall oxide before annealing at high temperatures.152
A proper hydrothermal treatment of H3PO4-AAOs improved
crystallinity of the relatively pure inner oxide layer (i.e., the cellboundary band in Figure 15), which enhanced the etching
contrast between the inner and outer oxide layers, after which
the outer pore wall oxide was selectively removed from the unit
cells of the AAO membranes by wet-chemical etching.
Deformation-free porous -Al2O3 membranes could be
obtained by annealing of the resulting samples at 1300 C
(see Figure 19).

coordination for H2SO4- and H2C2O4-AAOs, and tetra- and


penta-coordination for H3PO4-AAOs. It is now generally
accepted that the aluminum exists in four-, ve-, and six-fold
coordination with oxygen, although the ratio of the
coordination numbers varies in dierent samples formed in
dierent anodizing conditions. Farnan et al.138 attributed such
variations of aluminum coordination to the presence of
hydroxyl groups (OH) within alumina, with an increase of
anodizing temperature favoring the hexa-coordination. Incorporated electrolyte species (e.g., acid anions, proton, and water)
may be more responsible for such variations of aluminum
coordination. Yet, a clear explanation for such variation of
coordination number has not been given yet.
5.3. Eect of Heat Treatments

The surfaces of pore walls are hydrophilic due to the surfacebound hydroxyl (OH) groups, and this feature allows easy
modications of the surface property via self-assembly of
various functional molecules. Extensive recent reviews on this
matter are given in refs 142 and 143. On the other hand, asprepared porous AAOs are highly labile to both acid and base
attack. Proper high temperature heat treatments of as-prepared
porous AAOs markedly improve their thermal stability and
resistance against acid, base, and other corrosive chemicals,
allowing the resulting AAOs to be useful as starting materials
for developing various devices, which will operate in high
temperatures or harsh environments. TEM and X-ray
diraction (XRD) studies have indicated that the porous
AAOs undergo a series of polymorph transformations upon
heat treatment up to 1500 C in air with the following route
(see Figure 18 for XRD): amorphous AAO -Al2O3 Al2O3 -Al2O3 -Al2O3.121,139,140,144147 Amorphous
AAOs crystallize into almost pure -Al2O3 at a temperature
range of 820900 C, and then undergo successive transformations through metastable ccp - and -Al2O3 until

6. GROWTH OF POROUS ANODIC ALUMINUM OXIDE


(AAO)
6.1. Stress Generation in Anodic Oxide Films

6.1.1. Volume Expansion. Oxidation of aluminum is a


volume expansion process. The volume expansion during
anodization can be quantitatively expressed by the Pilling
Bedworth ratio (PBR). The PBR is rigorously dened by the

Figure 18. XRD spectra of heated H3PO4-AAO (Co K radiation).


Key: AlP = AlPO4, A = -Al2O3, T = -Al2O3. Reproduced with
permission from ref 139. Copyright 2005 Elsevier.
7502

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 20. A schematic illustration of volume expansion during


anodization of aluminum.

For three major pore-forming acid electrolytes (i.e., H2SO4,


H2C2O4, and H3PO4), a wide range of volume expansion
factors (kv) between 0.86 and 1.90 have been reported (Table
1).106,116,162,163,165168 For anodizations of aluminum in
H2SO4, Jessensky et al.165 investigated the relation between
the volume expansion factor (kv) and the degree of selfordering of pores. They reported that optimal conditions for
the best ordered arrays of pores are accompanied by a moderate
volume expansion (i.e., kv = 1.22), while contraction or very
strong volume expansion can result in the disordering of pores.
Further, they proposed that the mechanical stress at the metal/
oxide interface, which is associated with volume expansion and
increases with the anodizing potential (U), is the driving force
for the formation of ordered hexagonal pore arrays, by causing
repulsive forces between the neighboring pores. Li et al.
reported a similar line of experimental results, noting that the
volume expansion factor (kv) for the optimum anodization
conditions required for ordered pore arrays should be close to
1.4, irrespective of the anodizing electrolytes (i.e., H2SO4,
H2C2O4, or H3PO4).106 Later, on the other hand, Nielsch et al.
proposed that the best self-ordering of pores requires a porosity
(P) of 10%, the condition that corresponds to a volume
expansion of porous AAO of about 1.23, independent of the
electrolyte.109
More systematic investigations on the eect of anodization
conditions on the volume expansion factor (kv) have been
performed by Vrublevsky and co-workers. For porous AAO
growths in H2SO4 and H2C2O4 electrolytes, they determined
that the volume expansion factor (kv) has a linear dependence
on the anodizing potential (U) and also noted a linear relation
between the logarithm of the current density (i.e., ln j) and the
inverse volume expansion factor (i.e., 1/kv).116,163,167 For
anodizations under galvanostatic conditions, an increase of
temperature led to a decrease of the volume expansion factor
(kv) due to the corresponding decrease of the formation
potential (U), in which the slopes of ln j versus 1/kv curves are
invariant with respect to the anodizing temperature.163,167
However, the slopes of curves were found to be dierent for
dierent electrolytes, which was explained by the inuence of
acid-anion incorporation on the volume expansion: the larger is
the amount of incorporated acid anions, the larger is the
volume expansion factor (kv).167 On the basis of the
experimental results of Vrublevsky et al., one may obtain the
following relation:

Figure 19. A photograph of the porous -Al2O3 membranes obtained


by annealing the porous AAOs with (a) and without (b) acid-anion
contaminated outer pore walls. (ce) SEM images of porous Al2O3
membranes obtained by annealing the porous AAOs after removal of
outer pore walls at (c) 1115 C, (d) 1250 C, and (e) 1300 C; the
scale bars are 1 m. Reproduced with permission from ref 152.
Copyright 2012 The Royal Society of Chemistry.

molar volume ratio of grown oxide (Vox) to the consumed


metal (Vm) as follows:
PBR =

Mox m
Vox
=
Vm
nM mox

(60)

where Mox is the molecular weight of oxide, Mm is the atomic


weight of metal, n is the number of atoms of metal per one
formula of the oxide, and m and ox are the densities of metal
and oxide, respectively. In science on the corrosion of metals,
PBR has been the basis for judging the protectiveness of a
passivating oxide: if PBR < 1, the passivating oxide is under
tensile stress and easily cracked; if 1 < PBR < 2, the oxide
covers the metal uniformly and is protective; if PBR > 2, the
passivating oxide is under too much compressive stress and
easily crumbles (e.g., iron oxide on iron).153,154 For anodic
alumina growth, PBR can be experimentally determined from
the current eciency (j) of oxide formation and the density
(AAO) of the resulting AAO, provided that the composition of
anodic oxide is well-dened. The densities of barrier- and
porous-type AAOs have been reported to be in the range of
AAO = 2.73.5 g cm3.104,111,155157 Assuming composition
stoichiometry of Al2O3 and AAO = 3.0 g cm3, PBR for AAO
growth is 1.70 at 100% current eciency (j). For porous-type
AAO growth, on the other hand, PBR can vary between 1.02
and 1.58 due to the lower current eciency (j = 60
93%).158161 The exact determination of PBR is rather
complicated by the composition of anodic oxide, its density,
and the current eciency (j) of oxide formation. Consequently, volume expansion in anodic oxide growth has been
considered instead, by employing a volume expansion factor
(kv), and a simple thicknesses ratio:116,162164

k v = hAAO/hAl

6.1.2. Stress Measurements. Anodic alumina is a


dielectric material. During anodization, a very large electric
eld (typically, 106107 V cm1) is impressed on the oxide
lm. The electric eld drives the inward movement of O2 ions
and the outward migration of Al3+ ions within the anodic oxide.
The resistance to these counter-migrations and the attraction of

(61)

where hAAO and hAl are the vertical heights of the AAO and the
consumed aluminum, respectively (Figure 20).
7503

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Table 1. Volume Expansion Factors (kv) Reported for Porous Anodic Aluminum Oxide (AAO) Growth in Three Major PoreForming Acid Electrolytes
electrolyte

concentration

H2SO4

20 wt %
20 wt %
1.7 wt %
1.7 wt %
10 wt % mixture solutiond

temp (C)
1
1
1
10
1822
0

H2C2O4

2.7 wt %
0.34 M
0.220.92 M
0.12.9 M

H3PO4

1
15
1618
18

Ua or jb

kvc

refs

1825 V
19 V
25 V
25 V
620 mA cm2
50 V
100 V
300 V
40 V
40 V
1.624.8 mA cm2
213 mA cm2

0.861.62
1.41
1.36
1.40
1.41.6
1.72
1.76
1.90
1.42
1.18
1.251.42
1.21.62

165
106
106
106
167
162
162
162
106
168
163,167
166

U = anodizing potential (in V). bj = current density (in mA cm2). ckv = volume expansion factor. dA mixture solution of 0.41 M H2SO4 and 0.16
M H3BO3.

and Leach50 conducted deection measurements on aluminum


and reported the eects of current density and oxide thickness
on the stress. During anodization of aluminum in ammonium
borate and ammonium citrate solutions, they observed that the
stress can be either tensile or compressive depending on the
current density. Compressive stress was observed below 1 mA
cm2, while tensile stress was observed at higher current
densities. A similar line of results was obtained by Nelson and
Oriani,169 who performed deection measurements during
anodization of aluminum and titanium in 0.1 M H2SO4. The
deection was related to the stress as follows:169

the charged species result in a compressive electrostatic stress


along the direction of the electric eld. The compressive stress
normal to the oxide surface is proportional to the dielectric
constant () and the square of the electric eld (E):169
2
=
E
(62)
8
This compressive stress occurs only under the electric eld.
Thus, the stress will relax when the anodization is stopped.
Proost et al. have determined electrostatic stress in anodic oxide
lms of thickness tb as follows:67,170
ES =

(1 + 2) E2
v
0
1v
2
tb

(63)

where v is the Poisson coecient of the oxide lm, is its


relative dielectric constant, o is the vacuum permittivity, and 1
and 2 are both characteristic electrostriction constants.
Porous AAO has been known to incorporate water as well as
varying amounts of electrolyte-derived anionic species (see
section 5.2). Compositional eects arising from anion
incorporation and oxide inhomogeneities may be sources of
stress in anodic oxides. The stress during anodization can be
measured by monitoring the changes of the substrate
deection. Stoney related the measured deection (i.e., the
radius of curvature) to the stress () as follows:171

Yt 2
6ktb

EMt 2k
2
3(1 vM
)L2tb

(65)

where EM and vM are the elastic modulus and Poissons ratio of


metal strip of length, L, and thickness, t; tb is the thickness of
the oxide layer where the stress is generated; k is the radius of
curvature of the metal strip. The compressive stress due to
electrostriction was found to increase linearly with anodizing
potential during oxide growth. Slowly grown oxides have
greater electrostrictive deections than more rapidly grown
oxides. For aluminum anodized in 0.1 M sulfuric acid, the
deections are compressive at low current densities and
become tensile above 0.6 mA cm2. The development of
tensile stress was attributed to the volume dierence between
the metal being oxidized and the oxide formed at the metal/
oxide interface.169 Moon and Pyun investigated the eect of
electrolyte concentration and current density on the deection
behavior of aluminum in sulfuric acid.173,174 Their experiments
found that the deections become more tensile as the current
density increases. Compressive stresses were always observed at
the relatively low current density (j = 2 mA cm2). On the
other hand, at larger concentrations of sulfuric acid, the rates of
deection with respect to the current density became nonlinear.
The observed compressive stress at low current densities and
tensile stress at higher current densities was explained in terms
of the annihilation of cation vacancies and the formation of
oxygen vacancies at the metal/oxide interface, respectively. The
authors suggested that the stresses that developed are not
distributed over the entire oxide lm, but are limited to a
narrow region of the metal/oxide interface below 1 nm.174 An
opposite evolution of internal stress has recently been reported.
Proost et al.175 measured the internal stress of porous AAO

(64)

where Y is Youngs modulus of the substrate metal, t is the


thickness of the metal foil, tb is the thickness of the oxide on the
metal, and k is the radius of curvature. Equation 64 has been
modied over the decades to account for lateral strain,
dierences in the elastic moduli of the oxide and its metal
substrate, and non-uniformity of the stress distribution in the
oxide. Vermilya172 applied the Stoney method for anodizing
dierent metals. Upon applying potential, the substrate
deected, indicating tensile stress. The stress observed with
the forming voltage applied was always more compressive than
at zero voltage except for tungsten. Higher stress was observed
when the rate of oxide formation was high. Vermilya attributed
the observed stress to a dynamic hydration process. As the
oxide lm is buried by newly generated oxide, it is dehydrated
by proton migration, producing tension in the lm. Bradhurst
7504

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

during the very initial stage of pore formation in 0.4 M


phosphoric acid by employing a high-resolution in situ
curvature measurement technique based on multiple-beam
deectometry. The internal stress was found to remain constant
during the initial barrier growth. However, the internal stress
value increases in the compressive direction with increasing
current density. When the current density is lower than 4 mA
cm2, the internal stress in the barrier oxide is tensile, while at
higher current densities, the internal stress becomes compressive. The dierent stress evolutions in dierent electrolyte
systems have not yet been fully understood. More systematic in
situ measurements of stress as a function of the anodizing
conditions at the very initial stage of pore formation are
required to resolve conicting stress evolutions reported in the
literature. A meaningful advance in this direction has recently
been made by Hebert and co-workers,176 who demonstrated
the eectiveness of phase-shifting curvature interferometry as a
new technique for high-resolution in situ monitoring of stress
evolution during anodization. The newly developed metrology
allows extremely stable and reliable measurement of curvature
changes at a curvature resolution of 103 km1, which is
comparable to or higher than that of the high-resolution
multiple-beam deectometry technique.175,177179 From stress
measurements during galvanostatic anodizing of aluminum in
0.4 M phosphoric acid, Hebert and co-workers176 found that
the apparent stress in the barrier oxide is tensile (+100 MPa) at
low current density but became increasingly compressive at
higher current densities. In addition, they observed that
transition from tensile to compressive stress occurs at current
density of 4 mA cm2, in good agreement with the report of
Proost and co-workers.175
6.1.3. Eects of External Stresses on Pore Growth. The
eect of an applied external tensile stress on the self-ordering of
pores was rst studied by Sulka and co-workers.180 It was found
that the magnitude of applied tensile stress inuences the
ordering degree of pores. At a relatively low external tensile
stress, regular hexagonal arrangement of pores was observed.
However, a high tensile stress completely destroyed the pore
arrangement of porous AAO. Large holes and pits, rather than
nanopores, appeared on a highly stressed surface. This
experimental result revealed that the mutually repulsive
mechanical force between neighboring pores, which is
associated with volume expansion due to oxidation of the
metal, may be the driving force for the self-organized formation
of hexagonally close-packed arrangement of nanopores, as
suggested by Jessensky et al.165
As to the eects of compressive stress on pore growth, Park
et al.181 have performed the anodization of aluminum conned
in micrometer-sized vertical trench patterns. The authors
observed that during the anodization of the conned aluminum,
the anodization rate is signicantly retarded at the vertical
sidewall of the trench. Because of the retarded anodization rate,
most of the aluminum at the edge part of the structure remains
and its thickness decreases gradually with increasing distance
from the vertical sidewall toward the central part of the trench
structure (Figure 21). The authors attributed this phenomenon
to the accumulation of compressive stress at the vertical
sidewall of the trench structure, where linear vertical volume
expansion is severely prohibited by additional stress. The
authors noted that because compressive stress is an additional
kinetic barrier to the electrochemical oxidation of aluminum,
the anodization kinetics of aluminum should be severely
retarded.

Figure 21. Cross-sectional SEM images of the AAO conned in a


micrometer-sized vertical trench pattern: (a) low and (b) high
magnication image. Scale bar = 1 m. Reproduced with permission
from ref 181. Copyright 2006 The Electrochemical Society.

Retardation of anodization rate has also been reported for


lateral anodization processes. In that process, an aluminum thin
lm deposited on a substrate is sandwiched by a rigid insulating
top layer, and then the side edges of the aluminum are anodized
to produce horizontal arrays of pores on the substrate.182186
Oh and Thompson186 reported abnormal behavior in the
anodic oxidation of aluminum in mechanically conned
structures used for the formation of horizontal nanoporous
AAO. The authors observed that dendrites, periodic internal
pore structures (see Figure 22a), formed with a 5% retarded
growth rate, as compared to its value during bulk anodization
under the same conditions. They attributed the observed
anomalies to the suppressed volume expansion and a plastic
ow of anodic oxide conned by an insulating top layer;
because volume expansion by plastic ow in the pore growth

Figure 22. (a) Cross-sectional SEM image of AAO nanopore formed


by horizontal anodization. Scale bar = 100 nm. (b) Schematic
illustration of the formation mechanism of horizontal AAO with
dendritic internal pore structure. Reproduced with permission from ref
186. Copyright 2011 The Electrochemical Society.
7505

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 23. Schematic diagram of the kinetics of porous AAO growth in (a) potentiostatic and (b) galvanostatic conditions: (a) Current (j)time (t)
curves for potentiostatic anodization (i.e., U = constant) and (b) potential (U)time (t) curve for galvanostatic anodization (i.e., j = constant).

maximum due to the ready diusion of electrolyte (stage III).


After that, current (j) reaches a steady value after passing
through an overshoot (stage IV). The appearance of current
overshoot has been related to the decrease of the initial pore
density with the steady-state growth of major pores:168 pores
increase in size by persistent merging with adjacent pores. For a
given set of anodization conditions, the rate of potential
increase at the very beginning of anodization, the value of the
minimum current, the time needed for anodizing current to
reach a steady value, and the appearance of the current
overshoot have been known to be directly dependent on the
anodizing potential (U), electrolyte pH and temperature, and
the initial surface state of the aluminum.168,187,188
For the case of galvanostatic anodization, a similar
progression can be observed for stages IIV, while the
potential (U) changes as a function of time (Figure 23b).
Under constant-current conditions, the oxide growth rate
should be proportional to the applied current density (j) and
constant according to the Faradays law. In addition, a constant
electric eld (E = U/tb) is required to sustain the applied
constant current (j).76 Accordingly, the potential (U) increases
with the thickness of the growing barrier oxide (tb), as shown in
the inset of Figure 23b. However, in practice, the evolution of
potential (U) deviates from a simple linear increase with time,
as shown in Figure 23b. For convenience, various mechanisms
governing such a deviation have been referred to as growth
instabilities, which include, for example, mechanical breakdown
during zirconium anodization, and surface undulation/pore
initiation during aluminum anodization.178 Figure 23b shows a
gradual retardation of potential (U) increase at stage II. Such a
potential evolution can be attributed to a morphological
instability, that is, transition from the stage of barrier oxide
growth to the stage of porous oxide growth.178 In the following
sections, we will discuss in detail the kinetics and morphological
instability involved in the early stage of anodization, which have
been systematically investigated by Proost and co-workers.160,175,178,189,190
6.2.2. Kinetics of Porosity Initiation. Recent publications
have shown that the growth of a porous AAO and its selforganization are most likely driven by the internal stresses
developing in the anodic oxide.47,190192 Proost and co-workers
have investigated the initiation of porosity during galvanostatic

direction is prohibited by traction at the insulating top layer,


the extra volume of newly formed anodic alumina is extruded
inside the primary pores, resulting in a dendritic structure, as
schematically illustrated in Figure 22b.
6.2. Initial-Stage Pore Formation

6.2.1. Qualitative Description on Pore Formation.


Porous AAO can be easily fabricated by anodization of
aluminum in acid electrolytes either under a constant potential
(i.e., potentiostatic) or a constant current (i.e., galvanostatic)
condition. In general, potentiostatic anodization is widely
employed for the fabrication of self-ordered porous AAO,
because of the linear relation between the applied potential (U)
and the structural parameters of the resulting AAO (i.e., pore
diameter Dp, interpore distance Dint, and barrier layer thickness
tb, section 5). Figure 23 shows (a) a typical current (j)time (t)
curve for potentiostatic anodization, (b) potential (U)time (t)
curve for galvanostatic anodization, together with (c) schematic
illustrations of the stages of porous structure development.101
When a constant anodic potential (U) is applied, a thin
compact barrier oxide starts to grow over the entire aluminum
surface (stage I). Thickening of the initial barrier oxide over
time (t) results in an increase of the series resistance (R) of the
anodization circuit. Current (j) is initially maintained at the
limiting current (jlimit) of the power supply, and correspondingly potential (U = jR) increases linearly with time (t) (see the
inset of Figure 23a). When the thickness (or the resistance, R)
of the compact barrier oxide layer reaches a certain value,
current (j) drops rapidly to hit the minimum value (stage II).
For this stage, OSullivan and Wood100 suggested that current
(i.e., electric eld) concentrates on local imperfections (e.g.,
defects, impurity, pits) existing on the initial barrier oxide,
resulting in non-uniform oxide thickening and pore initiation at
the thinner oxide areas. Thompson and co-workers10,33,44 have
proposed that local cracking of the initial barrier oxide due to
accumulated tensile stress (PBR < 1) may develop the paths for
electrolyte penetration. Local increase in eld strength at the
penetration paths eectively polarizes the AlO bonds,
facilitating eld-assisted oxide dissolution there (section
6.3.2),100 and eventually leads to development of individual
penetration paths into embryo pores.10,44 Accordingly, further
anodization leads to a gradual increase in current (j) to a local
7506

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 24. (a) Curvature change and cell voltage evolution as a function of anodization time at 5 mA cm2 in 1 M H2SO4 (22 C); (b) radial power
spectral density (PSD) of samples anodized for dierent times. Reprinted with permission from ref 190. Copyright 2009 AIP Publishing LLC.

anodization of aluminum thin lm on a silicon substrate in 1 M


sulfuric acid (22 C) at j = 5.0 mA cm2 by in situ monitoring
of the internal stress-induced curvature of the substrate.190
They observed that the magnitude of the substrate curvature
(K, in km1) increases at a constant rate in the compressive
direction up to a transition time (ca. 7.9 s) at which the rate of
curvature change suddenly increases (Figure 24a). The
observed curvature transition was attributed to the initiation
of porosity accompanied by an increase of the oxide growth
rate. This was conrmed by a quantitative analysis of the radial
power spectral density (PSD) distributions of the anodized
surfaces as presented in Figure 24b, in which the rst (1) and
second (2) correlation breaks are associated with the
aluminum grain morphology, and the appearance of porosity,
respectively. Further, the authors pointed out that the increase
in growth rate after the curvature transition would not be
expected if the porous layer grew by dissolution of the oxide at
the pore base.190
Hoar and Yahalom76,187 have proposed that the pore
initiation occurs at a suciently low electric eld due to
proton entry into the initial barrier oxide at preferred sites,
where concentrated electric eld (i.e., current) accelerates oxide
dissolution to develop pores. The authors187,193 suggested that
the Al3+ ions found in the solution originated from the oxide at
the pore base as a result of eld-assisted (rather than thermal)
oxide dissolution (see section 6.3.2). Yet later studies29,30,161
have revealed that Al3+ ions in the electrolyte are the result of
direct ejection of Al3+ ions from the metal/oxide interface
through the oxide into the solution (see section 6.3.4), rather
than coming from the eld-assisted oxide dissolution process.
Recently, Proost et al. have for the rst time correlated the
kinetics of Al3+ loss to the morphological changes occurring
during the very early stage of galvanostatic anodization of
aluminum in 1 M sulfuric acid160 and 0.4 M phosphoric acid189
by employing in situ inductively coupled plasma optical
emission spectrometry (ICP-OES). For both anodization
cases, the authors observed three distinct regimes of Al3+ loss.
However, the evolution of the Al3+ loss rate turned out to be
markedly dierent, as follows. For the same current density, the
rate of Al3+ loss is higher in phosphoric acid than in sulfuric acid
during the barrier layer growth stage (regime A, stage I in
Figure 23b). On the other hand, the Al3+ loss rate is lower
during the pore initiation stage (regime B, stage II in Figure
23b), as compared to the barrier oxide layer formation stage

(i.e., regime A). This implies that the oxide formation eciency
(j) during the initiation of porosity is higher than that during
the initial barrier oxide formation, which is in line with the
increase in the oxide growth rate upon pore initiation, as shown
by in situ monitoring of the internal stress-induced curvature of
the substrate (vide supra).190 On the other hand, the authors
attributed the lower Al3+ loss rate during the pore initiation
stage to the non-uniformity of current distribution upon
commencement of pore initiation.189 In sulfuric acid, the Al3+
loss rate during stages II and III was constant, and increased
slightly at the beginning of stage IV. In phosphoric acid,
however, the cation loss rate decreased again during stage II
and then remained more or less constant during stages III and
IV (regime C). The authors associated the dierence in the
observed evolution of Al3+ loss rate with the morphological
dierences of the growing anodic oxide (i.e., pore size and
spacing).189 In sulfuric acid, the rate of Al3+ loss during the
steady-state pore growth stage was similar to the level of the
barrier layer growth stage, whereas in phosphoric acid both
rates were markedly dierent; there was a higher rate during the
barrier layer growth stage. For both cases of anodization, the
rates of Al3+ loss were observed to linearly increase with the
current density (j). Proost et al.189 suggested that the direct
proportionality between the rate of Al3+ loss and the current
density (j) can be attributed to the direct ejection of Al3+ ions
into the electrolyte, because the eld-assisted contribution may
be relatively independent of the current density according to
the high eld conduction theory.25 For both cases of
anodizations, the oxide formation eciency (j) during the
porous oxide growth stage was found to increase with current
density (j). For phosphoric acid anodizing, however, the
eciency (j) during the barrier oxide growth stage did not
increase with current density (j), and remained constant in the
current range j = 2.010 mA cm2 (Figure 25), which is
dierent from the case of sulfuric acid anodizing.160 The
abnormal eciency values at 1.0 mA cm2 in Figure 25 were
suggested to occur when eld-assisted dissolution (see section
6.3.2) rather than direct cation ejection (see section 6.3.4)
becomes the predominant mechanism of Al3+ loss for current
densities lower than 2.0 mA cm2.164,189
6.2.3. Morphological Instability. As was briey mentioned at the end of section 6.2.1, at the very early stage of
anodization, pores initiate as a result of the morphological
instability of growing anodic oxide. This becomes evident at an
7507

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

(i.e., potential vs time) was characterized by three well-dened


regimes, as discussed in the previous section. By converting the
curvature versus time curve (see Figure 24a) into a stress
thickness versus potential curve, the authors obtained internal
stress data corresponding to the barrier layer growth regime
from the slope of the stressthickness versus potential curve
(Figure 27). On the basis of the anodizing ratio (AR) of 1.1 nm

Figure 25. Dependence of the oxide formation eciency (j) on


current density (j) for Al anodizing in 0.4 M H3PO4 during barrier
growth (regime A, i.e., stage I in Figure 23b, ) and steady-state
porous growth (regime C, i.e., stages IIIIV in Figure 23b, ).
Reproduced with permission from ref 189. Copyright 2011 The
Electrochemical Society.
Figure 27. (a) Stressthickness versus cell potential curves during
galvanostatic anodization of aluminum for 20 s in 1.0 M H2SO4 (22
C) at 5 mA cm2. The vertical dashed line indicates the moment of
pore initiation. (b) Oxide thickness at pore initiation versus
instantaneous internal stress in the barrier. The numbers near data
points indicate corresponding current density in mA cm2. Reprinted
with permission from ref 178. Copyright 2011 Elsevier.

oxide thickness of a few nanometers as shown in Figure


26.67,161,168,178,190,194 Previous research has suggested that

V1 for aluminum anodizing in 1.0 M H2SO4, they then could


plot the thickness at pore initiation (the rst vertical dashed
line in Figure 27a) as a function of stress for a set of anodizing
experiments performed at dierent current densities, j = 1.525
mA cm2 (Figure 27b). It turned out that oxide thickness at the
point of pore initiation increases with the compressive stress.
The authors claimed that such a dependency revealed that
pore initiation itself does not correspond to a stress-aected
instability, although further development of the instability into
an ordered steady-state pore morphology can be considered to
be stress-aected (i.e., porosity-induced stress relaxation).178
Instead, as an alternative instability criterion, they suggested
that electrostatic energy acts as a driving force not only for
porosity initiation, but also for selection of the interpore
distance (Dint) of anodic oxides.175
More recently, Hebert and co-workers194 performed a linear
stability analysis of an instability mechanism controlled by oxide
dissolution and ionic migration at the initial stage of pore
formation. The authors claimed that previous models101,192,196198 based on nonlinear interface kinetics may be
unrealistic, because the oxide formation eciency (j) is weakly
dependent on the current density (j).199,200 Their model
predicted that the range of oxide formation eciencies (j)
producing pattern selection depends on the cation charge (z)
and PBR; patterns with a minimum pore spacing occur within a
narrow range of the oxide formation eciency (j = 6570%
for porous anodic alumina and 5058% for anodic titania),
which occurs if z > 2. According to the model, the wavelength
for the maximum disturbance growth rate is proportional to the
thickness of anodic oxide, which quantitatively explains the
proportionality of interpore distance (Dint) to anodizing
potential (U). This holds for both disordered and self-ordered
porous AAOs, and also for diverse anodizing electrolytes.

Figure 26. Cross-sectional TEM micrographs of anodic oxide lm on


Al anodized in 0.4 M H3PO4 at 4.5 mA cm2 for (a) 17 s, (b) 34 s, and
(c) 55 s. Adapted with permission from ref 161. Copyright 2010 The
Electrochemical Society.

internal stress may play a role in the growth instability of


anodic oxides.64,65,75,192,195 Van Overmeer and Proost178 have
investigated the relation between the internal stress, the
morphological instability, and the pore initiation during the
growth of porous AAO. They employed a high-resolution in
situ curvature measurement technique to monitor the internal
stress during anodization of a thin aluminum lm on a silicon
substrate in 1.0 M H2SO4 at 5 mA cm2. The anodization curve
7508

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

anodization. Hoar et al.187,193 suggested that the local oxide


dissolution is assisted by the increased electric eld (E) due to
the geometry of the pore base. The proposed model suers
disadvantages, however, in that it does not specify the details
associated with eld-assisted oxide dissolution and it adopts the
idea that the outwardly migrating Al3+ ions contribute to the
oxide formation at the oxide/electrolyte interface, which
actually does not occur in the case of porous-type oxide
growth (see section 6.3.4).29,30
OSullivan and Wood100 proposed a detailed physical
mechanism for eld-assisted chemical dissolution of anodic
oxide, and qualitatively explained the dependence of the porous
morphology on anodizing conditions. They explained the eldassisted oxide dissolution in terms of the eective polarization
of AlO bonds in the lattice at the oxide electrolyte interface
under the eld (Figure 28).100 In this model, the electric eld

6.3. Steady-State Pore Formation

6.3.1. Joules Heat-Induced Chemical Dissolution. For


steady-state pore growth, the movement rates of the metal/
oxide interface and the oxide/electrolyte interface should be
balanced, keeping the barrier layer thickness (tb) constant. It
has long been believed that this balance is achieved by a
dynamic equilibrium between the oxide formation at the metal/
oxide interface and the removal of oxide at the oxide/
electrolyte interface.100,101,168,193 Joules heat-induced chemical
dissolution of the barrier oxide by acid electrolyte has been
suggested.7,127,168,187,201 However, this thermal mechanism
does not reasonably explain the dynamic balance of the
movement rates of the interfaces (i.e., the metal/oxide and
oxide/electrolyte interfaces), because the rate of chemical
dissolution is typically much lower than that of oxide formation
even at an elevated temperature.29 Generated heat has been
considered to play a role in enlarging pores by assisting
dissolution of the pore wall oxide, resulting in truncated pore
channels.202 From the experimentally determined oxide
formation rate (372.5 nm/min) in 1.5 M H2SO4 (21 C) at
20 mA cm2 and the chemical dissolution rate (0.084 nm/min)
in the same electrolyte, Hunter and Fowle201 inferred that the
electrolyte condition at the pore base must locally change to the
equivalent of boiling of 5.10 M H2SO4 at 124 C to satisfy the
required rate of chemical dissolution (i.e., 372.5 nm/min). On
the other hand, on the basis of their calculations of steady-state
temperature distribution, Nagayama and Tamura203 reported
that local temperature rise due to Joule heating under a
comparable anodization condition (i.e., 1.0 M H2SO4, 27 C,
and j = 9.4 mA cm2) is not more than T = 0.07 C. They
claimed that the high rate of local oxide dissolution should be
interpreted as a consequence of high electric eld (E)
impressed on the barrier layer, as had been suggested by
Hoar and Mott.193 Recent in situ measurements of anode
temperature during anodizations in 0.3 M H2C2O4 at 40 V have
shown that the maximum temperature change is T 1 C,
which contradicts Joules heat-induced chemical dissolution of
anodic oxide at the pore base.204
6.3.2. Field-Assisted Oxide Dissolution. Hoar and Mott
proposed193 that the thickness of the barrier oxide layer is
maintained by the dynamic rate balance between the following
two processes occurring at the oxide electrolyte interface: (1)
the oxide formation by the reaction between O2 ions and Al3+
ions migrated from the metal/oxide interface, as in the
formation of barrier-type oxide, and (2) the oxide dissolution.
The authors assumed that oxide formation takes place both at
the metal/oxide interface and at the oxide/electrolyte interface,
as in the barrier-type oxide formation. In their model, oxide at
the oxide/electrolyte interface is decomposed to Al3+ and O2
ions. The resulting Al3+ ions go into the electrolyte. Meanwhile,
O2 ions in contact with acid electrolyte become OH and
move through the oxide to form new oxide at the metal/oxide
interface. The proton (H+) released by the oxide formation
reaction would then diuse back to the electrolyte by proton
transfer between the lattice O2 ions. Since the oxide ions from
the oxide/electrolyte interface are spread over a larger area at
the metal/oxide interface, the oxide dissolution occurs at a
greater rate compared to oxide formation at the metal/oxide
interface. In other words, the net result of process (2) is the
progressive thinning of the barrier oxide layer due to the
requirement of oxygen volume conservation in the oxide, which
is compensated by the oxide formation through process (1) to
keep the thickness of the barrier layer constant during

Figure 28. Schematics of eld-assisted dissolution mechanism by


OSullivan and Wood: (a) before polarization, (b) after polarization,
(c) removal of Al3+ and O2 ions, and (d) remaining oxide. Adapted
with permission from ref 100. Copyright 1970 Royal Society
Publishing.

(E) across the barrier oxide can eectively polarize (i.e.,


stretch) the AlO bond along the applied eld direction,
lowering the eective activation energy for bond dissociation.
Solvation of Al3+ ions by water molecules via activated complex
(i.e., Al(H2O)63+) and the removal of O2 ions by H3O+ ions as
H 2O are facilitated. Because the electric eld (E) is
concentrated on the pore base, the oxide dissolution rate is
the greatest there, and a dynamic equilibrium between the
oxide formation and the oxide dissolution can be established.
This model has popularly been cited in the literature to explain
the growth and morphology of porous AAO.
Several models based on eld-assisted oxide dissolution have
been developed. In their theoretical modeling, Parkhutik and
Shershulsky101 considered the three-dimensional (3D) distribution of electric eld and current in the barrier oxide layer and
included the eld-assisted dissolution at the oxide/electrolyte
and oxide formation at the metal/oxide interface as boundary
conditions to predict the steady-state pore morphology. Their
model predicted that the movement rate of the oxide/
electrolyte interface by eld-assisted dissolution is pH-dependent and has an exponential dependence on the electric eld (E).
7509

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

The model also predicted a linear dependence of interpore


distance (Dint) on anodizing potential (U).
Thamida and Chang198 were also able to predict the linear
dependence of the interpore distance (Dint) on the anodizing
potential (U). They performed a perturbation analysis of the
equation describing the movements of the metal/oxide
interface by oxide formation, and oxide/electrolyte interface
by oxide dissolution, the rate of which was considered to be
governed by the local electric eld. The electric eld at both
interfaces was considered to be dependent on the shape or
topography of the interfaces. Their model predicted that the
ratio of pore diameter (Dp) to interpore distance (Dint) is a
factor independent of anodizing potential (U), but varies with
the electrolyte pH.198 Singh et al.192 have further put forward
previous models. The main element of their model is the
ButlerVolmer relation, which describes the exponential
dependence of the current on the overpotential and the
dependence of the activation energy of oxide dissolution
reaction on the Laplace pressure and the elastic stress within
anodic oxide.192,205 Their model predicted ordered hexagonal
pore arrays for a range of volume expansion factors close to the
one observed for ordered porous lms.
Recently, Friedman et al.206 performed a systematic
experimental investigation to study the stability phase diagram
as a function of pH and anodizing potential (U) in an attempt
to validate the above-mentioned theoretical models. By
considering the discrepancies between the previous theoretical
models and their experimental results, they concluded that the
previous models must include an appropriate weighting factor
to account for the oxide formation and dissolution mechanism
during the pore formation. The discrepancies may originate
from the fact that these models were based on pore formation
by eld-assisted oxide dissolution at the pore base. In fact, the
eld-assisted dissolution mechanism for steady-state pore
formation has recently been rejected by several authors (see
sections 6.3.4 and 6.3.5).
For the initial stage of anodization, on the other hand, Oh
and Thompson207 have recently reported direct experimental
evidence of the impact of the electric eld (E) on the oxide
dissolution rate and the existence of a threshold electric eld
(E*) for pore initiation. To assess the eect of an electric eld
(E) on the Al2O3 dissolution rate, the authors re-anodized a
planar pre-formed Al2O3 layer (thickness h0 = 160 nm) on
aluminum and investigated the evolution of thickness of the
preformed oxide layer as a function of time (Figure 29). The
thickness of the planar barrier oxide was found to have
decreased from 160 to 131 nm after re-anodization in 5 wt %
H3PO4 at 86 V for 49 min, without changing the thickness of
the aluminum (i.e., no anodic oxidation of aluminum). In
addition, the dissolution rate of the barrier oxide of a xed
initial thickness was found to increase with electric eld (E).
The results indicated that eld-assisted oxide dissolution is
operative at the initial stage of anodization. On the basis of the
invariance of the thickness of aluminum and the morphology of
the metal/oxide interface, Oh and Thompson suggested that
the formation of the incipient pores is associated with eldinduced instability at the oxide/electrolyte interface at
suciently high electric eld. In a separate work, Skeldon
and co-workers have reported conrming experimental results
of eld-assisted oxide dissolution at the initial stage of pore
formation during potentiostatic anodization of aluminum in
phosphoric acid.208 In their study of pore initiation, the authors
employed immobile arsenic species as tracers. 18O-labeled

Figure 29. (a,b) SEM images showing eld-assisted dissolution of


anodic oxide under an electric eld (E): (a) before and (b) after reanodization of a planar preformed Al2O3 layer in a 5 wt % H3PO4
solution at 86 V for 49 min. Scale bar = 200 nm. (c) Changes in the
thickness of a planar preformed Al2O3 layer due to electric eld- and
time-dependent dissolution behavior in a 5 wt % H3PO4 solution.
Reprinted with permission from ref 207. Copyright 2011 Elsevier.

barrier-type oxide lms were rst formed in sodium arsenate


(Na2HAsO47H2O) solution, and subsequently the resulting
samples were re-anodized in phosphoric acid under electric
elds (i) below the threshold electric eld (E*) reported by Oh
and Thompson for the formation of incipient pores, (ii) close
to the threshold eld to induce signicant anodic oxidation of
aluminum, and (iii) well above the threshold eld. From
microscopic analyses of the arsenic and 18O contents, and the
pore morphologies and arsenic distribution in the resulting
anodic oxide lms, it was found that the eld-assisted oxide
dissolution is mainly responsible for the formation of incipient
pores at oxide formation eciency (j) = 2030%, while
eld-assisted ow of oxide materials is operative for the growth
of major pores at oxide formation eciency (j) = 57
66%.208 The authors suggested that the preferential growth of
incipient pores locally increases the current density at the pore
bases, which may inuence the transport numbers of mobile
ions, insertion of electrolyte-derived anionic species into the
anodic oxide, and thus oxide viscosity.208
6.3.3. Average Field Model for Steady-State Pore
Structure. As discussed in section 6.2.1, after the formation of
incipient pores, some large incipient pores develop into major
pores by increasing their size through persistent merging with
neighboring smaller incipient pores. As the anodization
proceeds, the growing major pores readjust their sizes and
spatial arrangement to establish equilibrium morphology (i.e.,
hexagonally close-packed pore distribution). During this period,
merging, dying, or even branching of the pores may occur. For
a given set of anodization conditions, the barrier layer thickness
(tb), pore size (Dp), and interpore distance (Dint, or cell size) in
the equilibrium pore structure are mainly determined by the
anodizing potential (U). OSullivan and Wood suggested a so7510

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

called average eld model to explain the establishment of the


equilibrium structure of growing pores.100 The essence of the
model is that eld-assisted dissolution at the base of growing
pores should occur to some dierent extents until the electric
eld (E) across the barrier layers eventually becomes the same
for every pore, the condition under which pores of an
equilibrium size grow at a constant rate, being determined by
the average eld. The model can be qualitatively explained as
follows. Figure 30 schematically shows three adjacent pores

would be enlarged in the lateral direction. Accordingly, for all


three cases in Figure 30, rearrangement of pores in conjunction
with the self-adjustment of pore size (Dp) and barrier layer
thickness (tb) would take place until the electric eld (E) across
the barrier layers of every pore approaches the average eld.
6.3.4. Direct Cation Ejection Mechanism. The average
eld model discussed above is based on the eld-assisted oxide
dissolution.100 However, this dissolution mechanism alone does
not adequately explain some of the experimental results, in
particular, that the quantity of Al3+ ions in the electrolyte after
anodization exceeds that of Al3+ ions originating from the
formation of the pores. On the basis of pore-lling experiments,
Takahashi and Nagayama reported that transport numbers for
Al3+ and O2 ions moving across the barrier layer are t+ = 0.4
and t = 0.6, respectively.46 The result implies that both Al3+
and O2 ions migrate in anodic oxide, with 40% of the ionic
current carried by the Al3+ ions and the remainder by the O2
ions, and further that about 40% of the Al3+ ions are lost into
the electrolyte without contributing to the porous oxide
formation. Considering the porosity (P 10%) of porous
AAO formed in typical anodization conditions,109 30% of Al3+
ions should be lost through a mechanism dierent from eldassisted oxide dissolution, even if the latter is still operative.
Siejka and Ortega studied the pore formation mechanism by
employing 18O tracing techniques.30 They rst formed a
compact base oxide lm on aluminum in electrolyte enriched in
H218O and subsequently anodized the sample in H216Oenriched H2SO4 electrolyte for pore formation. From the
nuclear microanalyses for 18O-isotope content and depth
distribution, the 18O tracer was found to be located at the
lm surface conserving its initial isotopic concentration (Figure
31b), which was attributed to oxide decomposition inside the

Figure 30. Schematic representation of growing three adjacent pores


(a, b, and c) with dierent pore sizes (Dp1, Dp2, and Dp3) and barrier
layer thicknesses (tb1, tb2, and tb3). R1, R2, and R3 are the radii of
curvature of the metal/oxide interfaces of the respective pores.
(45.7) is the angle subtended from the center of curvature to the
pore bases. The red lines extending from the metal/oxide interface to
the oxide/electrolyte interface represent current lines. The regions
marked with A correspond to the area of unoxidized metal near the
concave ridges.

growing with dierent pore sizes (Dp1, Dp2, and Dp3) and
barrier layer thicknesses (tb1, tb2, and tb3). We assume that pore
(b) has the equilibrium dimension of a given set of anodization
conditions, in which tb2 and R2 are the equilibrium barrier layer
thickness and radius of curvature of the metal/oxide interface,
respectively. Although OSullivan and Wood assumed that the
barrier layer thickness (tb) and the angle (w = 45.7) in a
potentiostatic anodization are constant irrespective of the pore
sizes (Dp), we suppose here that only w remains constant while
tb changes with Dp, which represents the real cases better. For
the three cases given in Figure 30, the electric eld (E) across
the barrier layers is dierent, because at a given anodizing
potential (U) the eld (E) is inversely proportional to the
barrier layer thickness (i.e., E = U/tb). Accordingly, eldassisted oxide dissolution at the pore base would occur to a
greater extent for pore (a), as compared to pores (b) and (c),
while anodic oxidation of metal would take place to the same
extent at the metal/oxide interface due to higher current
density (j) according to eq 1, until Dp1, R1, and tb1 have reached
their equilibrium values (i.e., Dp2, R2, and tb2). For pore (c), on
the other hand, eld-assisted oxide dissolution and anodic
oxidation reactions would be progressively retarded because of
lower electric eld (E), until pore (c) has the equilibrium
dimension. Unoxidized metal near the hemispherical concave
ridges (i.e., the points marked with A in Figure 30) is acted
upon by two lateral eld components of dierent magnitudes
from two neighboring pores. The lateral eld component of
greater magnitude drives the eld-assisted dissolution and
anodic oxidation to a greater extent. The corresponding pore

Figure 31. Schematic diagrams illustrating ndings of oxygen tracer


experiments for the growth of (a) barrier and (b) porous anodic lms
on aluminum. In each case, the anodizing is carried out rst in
electrolyte enriched in H218O and second in electrolyte enriched in
H216O. Adapted with permission from ref 32. Copyright 2006 The
Electrochemical Society.

oxide and reincorporation of the released oxygen to form new


oxide at the metal/oxide interface. 18O-tracer studies along with
the analyses of the components of ionic current (jtot) have
further revealed that the number of oxygen in porous AAO
accounts for about 60% of the total ionic current (i.e., jox =
60%jtot), which is numerically equal to the current eciency
(j) of porous AAO formation.29,30 The remainder is associated
with the loss of Al3+ ions into the electrolyte (i.e., jloss = 40%
jtot). Siejka and Ortega claimed that pore formation in the
absence of oxygen losses should be associated with the direct
ejection (jAl = 30%jtot) of Al3+ ions from the metal/oxide
interface into the electrolyte across the barrier oxide and the
outward movement (jdec = 10%jtot) of Al3+ ions produced by
7511

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 32. Cross-section transmission electron micrographs (TEM) of (a) the sputter-deposited aluminum containing a tungsten tracer layer, and
following anodization for (b) 180 s, (c) 240 s, and (d) 350 s at 5 mA cm2 in 0.4 M H3PO4 at 293 K. Reprinted with permission from ref 98.
Copyright 2006 Elsevier. The distributions of tungsten tracer in anodic oxide are schematically illustrated above the respective TEM images. (e)
Simulated current lines and potential distribution for porous AAO growth in oxalic acid at 36 V (color scale is potential in volts). (f) Simulated ow
velocity vectors and mean stress (color scale is dimensionless stress). Panels (e) and (f) reproduced by permission from ref 200. Copyright
2009Macmillan Publishers Ltd.: Nature Materials.

immediately beneath the pores initially lies slightly below the


adjacent tracer near cell wall regions (Figure 32b). Upward
displacement of the tracer band at the cell wall regions became
pronounced upon further anodizations, as compared to the
tracer below the pores, where the band is getting fainter
because of a reduced tungsten concentration (Figure 32c,d). It
was suggested that the ne-tungsten lines along the cell
boundaries beneath the tracer layer are associated with the
tungsten enrichment in the aluminum adjacent to the
aluminum/lm interface, because formation of WO3 requires
higher Gibbs free energy as compared to Al2O3.98,212,213 The
authors pointed out that the observed behavior of the tracer
band is contrary to what one would expect from the
conventional dissolution-based model of porous AAO growth
as follows.32,98 According to the conventional model of pore
formation, tungsten tracer should be incorporated into the
anodic oxide rst at the regions immediately below the pores
and then near the cell boundary regions because of the
scalloped geometry of the barrier oxide layer. Thus, the tracer
band at the pore regions should lie ahead of the tracer at the
pore wall regions because of the outward migration of tungsten
species, contrary to the experimental observations. If eldassisted oxide dissolution occurred at the oxide/electrolyte
interface, to maintain a constant barrier layer thickness (tb), a
tungsten-rich layer with a sharp image contrast (instead of
diminished contrast, as in Figure 32c) would have been

oxide decomposition inside the barrier layer toward the


electrolyte: jloss = jAl + jdec = 40%jtot.30 Consistent results
have recently been reported by Skeldon and co-workers,161 who
observed that the amounts of 18O tracer species in lms formed
by sequential anodizing in phosphoric acid do not change
signicantly during the formation of anodic oxide between the
barrier and porous stages. The direct cation ejection
mechanism is also in line with the recent report by Wu et al.,
who have experimentally shown that the formation of pores
does not occur through oxide dissolution process at the oxide/
electrolyte interface, but through direct ejection of Al3+ ions
into the electrolyte.31 However, the process involved in oxide
decomposition inside the barrier layer and its role in the
dynamic balance between the movement rates of interfaces (i.e.,
the metal/oxide and the oxide/electrolyte interfaces) need to
be specied.
6.3.5. Flow Model for Steady-State Pore Formation.
Skeldon et al.32,98,209,210 have investigated the development of
pores in porous AAO by performing tracer studies. They
anodized Al/W-tracer/Al substrates (W-tracer layer; 35 nm
thick, Al30 at. % W, Figure 32a) in H3PO4 solution and
investigated the movement of the tracer band (WO3) by TEM
and RBS.32,98 During anodic oxidation, tungsten species
incorporated into the anodic oxide migrated toward the
oxide/electrolyte interfaces at about 0.38 of the rate of Al3+
ions.211 It was observed that the incorporated tracer band
7512

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Skeldon and co-workers213,218 have also investigated the


eect of the types of tracer elements on the distributions of
tracer layers in anodic alumina. Figure 33 schematically

observed by TEM at the pore bases due to preferential


dissolution of Al3+ ions of a higher mobility in anodic oxide.
On the basis of the results of tracer experiments, Skeldon and
co-workers32,98 proposed the ow mechanism of pore
generation, as an alternative process to conventional eldassisted oxide dissolution. According to the new model, the
constant thickness of the barrier layer (tb) during porous AAO
formation is maintained by viscous ow of oxide materials from
the pore base to the cell boundary. It was suggested by the
authors that the displacement of oxide materials is driven by
large compressive stresses (100 MPa) from electrostriction
due to a high electric eld (E 106 V cm1)97 and also by
volume expansion due to oxidation of aluminum,165 and that
can be facilitated by the involvement of most of the oxide
constituents in ionic transport (i.e., the plastic ow of oxide
materials).97,214,215
In most anodizations for porous AAO formation, acid anions
from electrolytes are incorporated into the barrier layer through
inward migration. They migrate slowly in the barrier oxide, as
compared to O2/OH ions, due to their relatively large size.
As a consequence, the barrier layer exhibits a duplex structure
in terms of chemical composition, with an acid anion
contaminated outer oxide layer and a relatively pure inner
oxide layer adjacent to the metal/oxide interface (see section
5.2).14 The relative thickness of the outer oxide layer to the
inner one depends mainly on the nature of the anodizing
electrolyte in a given set of anodizations and is invariant
throughout anodization. The ow model accounts for
incorporation of electrolyte anions into anodic oxide and also
their migration behavior in the barrier layer. Incorporated
anionic species are transported toward the cell walls in addition
to their inward migration in the oxide without dissolutionrelated loss; otherwise, the barrier oxide would eventually
contain no incorporated acid anions because of their slow
migration. The ow model considers Al3+ ions as the only ionic
species being lost into the electrolyte by eld-assisted direct
ejection.30 Garcia-Vergara et al. have pointed out that bulky
acid anions play a key role in the ow of oxide materials,
inuencing not only the pore and cell dimensions but also the
self-organization of pores through the redistribution of
stress.32,98,216 The isotopic order and the absence of tracer
loss in the earlier 18O tracer studies of Siejka and Ortega (see
section 6.3.4)30 have also been recently reinterpreted in terms
of the eld-induced plastic ow of oxide materials from the
pore base toward the cell boundary, during which most of the
initially incorporated 18O tracers are displaced to the cell
walls.161 The formation of dendritic (or sh-bone-like) pores in
a mechanically constrained environment has also been
explained in the framework of the ow mechanism.186,217
The ow mechanism of steady-state pore generation has
been supported by the theoretical modeling of Houser and
Hebert,47,199,200 which highlighted not only the ionic migration
under the gradients of mechanical stress and electric potential,
but also its implication on the Newtonian viscoelastic ow of
oxide materials from the pore base toward the pore bottoms
and further into the cell walls (see Figure 32e,f). Yet, the model
does not take into account the volume expansion stress at the
metal/oxide interface for the viscous ow of the oxide materials.
The authors concluded that the compressive stress at the pore
base drives the ow of oxide materials in association with the
competition of strong anion adsorption with deposition of
oxygen ions.200

Figure 33. Schematic diagram comparing (a) hafnium-, (b) neodymium-, and (c) tungsten-containing tracer layers in porous anodic
lms formed on aluminum in phosphoric acid. Reprinted with
permission from ref 213. Copyright 2009 Elsevier.

compares the distributions of hafnium-, neodymium-, and


tungsten-containing tracer layers in porous lms formed in
phosphoric acid. Contrary to what was observed from the tracer
studies using tungsten, the hafnium and neodymium exhibited
the types of movement behaviors that one might expect from
eld-assisted oxide dissolution. The authors attributed the
observed behaviors of tracer bands to the fast movement rates
of the tracer cations (i.e., Hf4+ and Nd3+) in anodic alumina as
compared to W6+ ions and also to the loss of fast moving tracer
cations into the electrolyte through direct ejection mechanism;
Hf4+ ions migrated outward in anodic alumina at about the
same rate as Al3+ ions,219 while Nd3+ ions migrated about twice
as fast as Al3+ ions (i.e., 6 times faster than W6+ ions).220
Although the authors explanation of the unexpected distortion
seems to be plausible within the present knowledge of
anodization, more systematic experimental investigations
using other electrolyte systems and theoretical study are
required to fully verify the oxide ow model.

7. SELF-ORDERED POROUS ANODIC ALUMINUM


OXIDE (AAO)
Studies on porous-type anodization have been mainly led by
the surface nishing industry, and hence focused on the
development of cost-eective anodizing processes and the
improvement of engineering properties of anodized products.
Although various anodizing processes have been intensively
explored by industry, the size uniformity and spatial ordering of
pores have not been considered a major concern. Typical
porous AAOs formed by industrial processes, represented by
hard anodization (HA), exhibit disordered pore structures with
numerous micrometer-sized cracks.221224 Therefore, classical
HA processes have not been implemented in the current
nanotechnology research, until the recent development of new
HA processes. On the other hand, the mild anodization (MA)
7513

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 34. (Left) A schematic showing a conventional two-step mild anodization (MA) process for self-ordered porous anodic aluminum oxide
(AAO): (i) the rst long-term anodization, (ii) removal of disordered porous AAO, and (iii) the second anodization at the identical condition to the
rst one. Representative surface SEM micrographs of the respective samples are shown in panels (a), (b), and (c), respectively. Reprinted with
permission from ref 228. Copyright 1998 Springer Science and Business Media. (d) A color-coded SEM image of AAO formed by two-step MA
using 0.3 M oxalic acid at 40 V, showing a poly-domain structure. An area with the same color consists of a domain. The pores are color-coded on
the basis of the average angle to the six nearest neighbors. Pores that have no apparent hexagonal coordination (i.e., defect pores) are marked with
white. Reprinted with permission from ref 226. Copyright 2008 The American Chemical Society.

aluminum substrate maintaining their directional coherency


(Figure 34c). Position-dened pore generation during the
second anodization has been attributed to the relatively thin
native oxide at the bottom of each concave, where the electric
eld (E) is the highest and the resistance is the lowest.168 In
general, porous AAOs formed by two-step MA process exhibit a
polydomain structure (Figure 34d).226 The lateral size of the
defect-free domain increases with the anodizing time, but is
limited to several micrometers.
Since the development of the two-step process from oxalic
acid-based anodization, a lot of studies have been conducted
not only to produce porous AAOs with dierent pore sizes and
densities with an improved arrangement of pores, but also to
understand the mechanism responsible for the growth and selforganizing behavior of pores during anodization. Particular
eorts have been devoted to exploring the optimum conditions
for pore ordering, mainly with sulfuric, oxalic, or phosphoric
acid. Studies have indicated that the self-organized growth of
ordered pores occurs within a relatively narrow window
(known as the self-ordering regime) of anodizing conditions
(see Figure 36). For a given electrolyte system, in the case of
MA, there is an optimum anodizing potential (U) for the best
ordering of pores: (i) sulfuric acid (0.3 M H2SO4) at U = 25 V
106,227
for an interpore distance (DMA
(ii) oxalic acid
int ) = 65 nm;
MA
(0.3 M H2C2O4) at 40 V for Dint = 103 nm;6,106,168,227,228 (iii)
11
selenic acid (0.3 M H2SeO4) at 48 V for DMA
int = 112 nm; and
MA
(iv) phosphoric acid (0.3 M H3PO4) at 195 V for Dint = 500
nm.109,229 Considerable eorts have been made to explore new
self-ordering regimes in a wider range of DMA
int . With the idea of
controlling self-organization by adjusting the mechanical stress
between the metal/oxide interface, Shingubara et al. employed
a mixture solution of 0.3 M sulfuric acid and 0.3 M oxalic acid
(v/v = 1:1) to change the density of the resulting porous AAO
and obtained self-ordered porous AAO with DMA
int = 73 nm at U

of aluminum produces self-ordered porous AAOs with uniform


pore size (Dp) and interpore distance (Dint), which can be easily
tuned by an appropriate selection of anodization conditions.
Thus, that process has been intensively utilized in academic
research for a wide variety of nanotechnology applications. Yet
the MA process is slow (lm growth rate = 210 m h1), and
can be conducted within a narrow range of anodizing
conditions. In the following sections, we will discuss the
conventional MA and the newly developed HA processes,
factors governing the structure of porous AAO, and some
characteristics of porous AAO that are relevant to nanotechnology applications. In addition, recent attempts to
engineer the internal pore structure of AAO by pulse
anodization (PA) and cyclic anodization (CA) will be
discussed.
7.1. Mild Anodization (MA)

Masuda and Fukuda found that the bottom part of porous


AAOs lms produced by anodizing of aluminum in 0.3 M oxalic
acid at 40 V exhibits a self-ordered pore structure as a result of
the gradual rearrangement of the initially disordered pores.5 On
the basis of this experimental nding, Masuda and Satoh
developed the so-called two-step anodization process, by
which porous AAOs could be obtained with a highly ordered
arrangement of uniform nanopores (Figure 34).6 In a typical
two-step anodization, the rst anodization process is conducted
for more than 24 h. The resulting porous AAO with disordered
pores in its top part (Figure 34a) is selectively removed by the
so-called PC-etching process using an aqueous mixture of 0.5
M H3PO4 and 0.2 M CrO3 at 80 C.225 The surface of the
resulting aluminum is textured with arrays of almost hemispherical concaves (Figure 34b). The second anodization is
carried out with the textured aluminum under the same
condition employed for the rst anodizing. Pores nucleate at
the centers of each concave feature and grow normal to the
7514

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

= 37 V.230 More recently, Sun and co-workers231 reported an


H3PO4-based MA process using aluminum oxalate ((AlC2O4)2C2O4) as an additive to suppress breakdown of porous
AAO during anodization at high potential and temperature.
They showed that the interpore distance (DMA
int ) of porous
AAOs can be continuously tuned from 410 to 530 nm by
performing anodization with the mixture solution at an
anodization potential (U) ranging from 180 to 230 V. Ono
et al.17,232 demonstrated the fabrication of porous AAOs with
an interpore distance range of 300600 nm by MA in organic
acid electrolytes: DMA
int = 300 nm for 5 M malonic acid at 120 V,
MA
DMA
int = 500 nm for 3 M tartaric acid at 195 V, and Dint = 600
nm for 2 M citric acid at 240 V. Yet the spatial ordering and size
uniformity of pores of the resulting porous AAOs were by far
inferior to those formed by conventional MA processes using
sulfuric, oxalic, or phosphoric acid, which would limit the
practical applications of such porous AAOs to nanotechnology
research.
For self-ordered porous AAO formed by MA processes,
recent morphological studies indicated that anodizing potential
(U) determines the interpore distance DMA
int , barrier layer
MA 106,109,116
thickness tMA
The interpore
b , and pore diameter Dp .
MA
)
and
barrier
layer
thickness
(t
distance (DMA
int
b ) increase
linearly with anodizing potential (U) with proportionality
1
constants MA = 2.5 nm V1 for DMA
int and ARMA = 1.2 nm V
106,109,114
MA
for tMA
.
The
observed
potential
dependence
of
D
b
int
and tMA
is consistent with the earlier reports for disordered
b
porous AAO.7,100,114 Linear dependence of the pore diameter
(DMA
p ) on anodizing potential (U) has also been reported both
for disordered and for self-ordered porous AAOs. OSullivan
and Wood100 reported that pore diameter of disordered AAOs
increases with anodizing potential at a rate of p = 1.29 nm V1.
For self-ordered porous AAO, on the other hand, Nielsch et
al.109 proposed that self-ordering of pores requires a porosity
(PMA) of about 10% regardless of the specic anodizing
1
conditions by assuming DMA
int = 2.5 nm V . Because porosity
(PMA) of AAO is given by eq 53, 10% porosity requirement of
Nielsch et al. dictates a linear increase of the pore diameter with
the anodizing potential at a late of p = 0.83 nm V1,
irrespective of anodizing conditions. However, there are a
number of recent reports, indicating that the self-ordering of
pores can occur at other porosity (PMA) values ranging from
0.8% to 30% depending on the MA conditions, and thus that
the anodizing potential (U) is not the only parameter
determining the pore diameter (Dp).11,102,232,233 For example,
Nishinaga et al. have recently reported fabrication of porous
MA
AAO with DMA
p = 10.4 nm and Dint = 112 nm (porosity, PMA =
0.8% according to eq 53) by H2SeO4-based anodization at 48 V
and 0 C.11 The authors noted that the solubility of anodizing
acid electrolyte has important eects on pore diameter and
interpore distance, and suggested that the weak solubility of
selenic acid under the anodization condition employed caused
the formation of 10-nm-scale pores.11 In another example,
Chen et al. reported a continuous decrease of the pore
diameter, which did not aect either the interpore distance or
the barrier layer thickness, when increasing the concentration of
polyethylenglycol (PEG) additive in phosphoric acid electrolyte.233 Because the porosity (P) of AAO is determined by the
ratio of the pore diameter to the interpore distance (i.e., Dp/
Dint in eq 53), increasing the PEG concentration in the
anodizing electrolyte corresponds to decreasing the porosity of
AAO. The use of an organic additive to reduce the pore
diameter has recently been extended by Martin et al., who

demonstrated fabrication of porous AAOs with pore diameter


less than 15 nm by using an aqueous mixture of sulfuric acid
and ethylene glycol (EG).234 Chen et al. noted two possible
origins for the decreasing porosity with the addition of organic
additive:233 (i) the increase of the eective electric eld (E) due
to the reduction of the dielectric constant () of the anodizing
electrolyte upon addition of an organic modulator, and (ii) the
weak chemical dissolution of the pore wall oxide under the
protection of organic molecules. Among them, the former is in
line with the earlier report by Ono et al. that the Dp/Dint ratio
decreases with increasing electric eld strength (E) during
anodization.232 Su et al. have further explored the eect of
electric eld (E) on porosity (P). In a series of papers,235237
they proposed a mechanistic model, proposing that the porosity
(P) of AAO is directly governed by the relative rate of water
dissociation at the oxide/electrolyte interface, according to the
following relation:236
P = 3/(n + 3)

(66)

where n is the amount of water that dissociates per mole of


Al2O3 in the oxide dissolution reaction given in reaction 15.
The model regards both the dissociation of water and the
dissolution of oxide at the oxide/electrolyte interface as
important processes, although the latter has recently been
largely disputed, as discussed in sections 6.3.4 and 6.3.5. By
performing quantum-chemical model computations, the
authors showed that the electric eld (E) can signicantly
facilitate heterolytic dissociation of properly oriented water
molecules at the oxide surface, that is, eld-enhanced water
dissociation.41 Under a stronger electric eld (E), the
dissociation rate of water will be increased (i.e., an increase
of the n value in eq 66), and thus the porosity will be reduced.
As such, the model appears to adequately explain the
dependence of porosity (P) on electric eld (E). The above
relation was derived from a simple geometric consideration of
growing AAO and the mass balance of total oxide-forming
oxygen anions produced by the dissociation of water and
dissolution of Al2O3. In other words, the original model by Su
et al. postulates that all of the oxygen anions produced by water
dissociation and oxide dissolution should contribute to the
oxide formation at the metal/oxide interface.236 However, by
itself this assumption does not fulll the necessary condition
required for the formation of the steady-state pore morphology,
that is, the constant barrier layer thickness (tb) maintained by
the dynamic balance of the movement rates of metal/oxide and
oxide/electrolyte interface, as follows. A close inspection of
reaction 15 used in the authors model reveals there is a
dierent eld dependence for the water dissociation than for
the oxide dissolution process. To increase the n value under a
stronger electric eld (E), water dissociation should occur at a
greater rate than oxide dissolution. Under such circumstances,
one may expect a progressive increase of the barrier oxide layer
thickness (tb) as the anodiziation proceeds. Therefore, other
ionic processes may be required to explain the constant
thickness of the barrier layer during anodization. It was
assumed that Al3+ ions are directly ejected from the metal/
oxide interface without contributing to the oxide formation.235
When MA is conducted outside the self-ordering regime, the
degree of spatial ordering of nanopores decreases drastically.
For a given acid electrolyte, there is a breakdown potential
(UB), above which anodization is accompanied by local
thickening, burning, and cracking of the anodic oxide lm,
caused by a catastrophic ow of current (j) and consequent
7515

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

evolution of a large amount of heat:4,18,79,89,94,97,238,239 UB = 27,


50, and 197 V for sulfuric, oxalic, and phosphoric acid,
respectively. It has been known for the MA process that
anodizations just below the breakdown potentials (UB) yield
porous AAOs with the best self-ordered pore structures.18 Ono
and co-workers232,240 investigated the self-ordering of nanopores at the local areas of burnt protrusions with a large
number of cracks. They found that the locally thickened areas
formed by breakdown exhibit domains of highly ordered cell
arrangement. In addition, the cell size (Dint) and the barrier
layer thickness (tb) of porous AAO at the burnt area were
observed to be remarkably smaller than those at the burnt-free
area, which is in line with the earlier report by Tu et al.239
These experimental observations imply that current density (j)
primarily determines the degree of spatial ordering of pores and
the structural parameters (i.e., cell size Dint and the barrier layer
thickness tb) of porous AAOs.
7.2. Hard Anodization (HA)
Figure 35. (a,b) SEM images showing weak cell junction strength of
porous AAOs formed by H2SO4-hard anodization (HA). (c) A
schematic illustration showing two dierent fracture modes of porous
AAOs formed by H2SO4-hard (HA) and mild anodization (MA): the
AA cleavage plane for HA-AAO and the BB cleavage plane for
MA-AAO. Panel (a) reprinted with permission from ref 18. Copyright
2006 The Electrochemical Society. Panel (b) reprinted with
permission from ref 113. Copyright 2008 The American Chemical
Society.

As mentioned in section 4.1, stable anodization in a given


electrolyte is dicult to maintain over breakdown potential
(UB) due to the occurrence of burning or breakdown of the
growing anodic oxide lm. Classical hard anodization (HA)
processes adopted in industry have typically been conducted at
potentials (U) higher than breakdown value (UB) at the
expense of randomly occurring local breakdown of anodic oxide
lm. The processes take advantage of high speed growth
(typically 50100 m h1) of oxide lms due to the high
current density (j) at an increased anodizing potential (U).
However, the resulting porous anodic lms are characterized by
severely buckled (or uneven) surfaces with numerous cracks
and non-uniform (or distorted) pores. Control of the pore size
(Dp), interpore distance (Dint), and the aspect ratio of the
nanopores is also very dicult. For these reasons, the classical
HA processes have not been employed for nanotechnology
applications.
Various attempts have been made to overcome the problems
associated with local breakdown events during HA. The
research activity to date includes extending anodizing potential
(U) over breakdown potentials (UB) by appropriately tuning
the three major pore-forming acid electrolytes (i.e., H2SO4,
H2C2O4, and H3PO4), by searching for new anodizing
electrolytes, and by eciently removing the reaction heat.
The current density (j) in HA process is typically 1 or 2 orders
of magnitude higher than that of conventional MA processes.
Thus, Joules heating (Q) during HA is 2 or 4 orders of
magnitude greater than the ordinary MA processes. The
excessive heat during the HA process can not only promote
acidic dissolution of the resulting porous AAO, but also trigger
local breakdown events.4,239
Chu and co-workers18,110 reported that the breakdown
potential (UB = 27 V) in a sulfuric acid-based anodization
system can be increased up to 70 V by experimentally aging the
electrolyte after a long-term anodizing (i.e., pre-electrolyzing)
at 1020 A hours per liter. Because of the high anodizing
potential (U), the current density (j) for stable anodization
correspondingly rose signicantly, and thus led to a high-speed
lm growth. The authors were able to fabricate self-ordered
porous AAOs with Dint = 90130 nm at 4070 V and 160
200 mA cm2 at 0.110 C. That condition is far from those of
MA (H2SO4: 25 V and 215 mA cm2) but similar to those of
classical HA,221,223,241243 although the authors named this
process high-eld anodization. Porous AAOs formed by this

process exhibited poor mechanical stability, which greatly limits


their practical application as templates for various nanofabrications. The produced porous AAOs exhibited weak cell
junction strength (Figure 35a), and thus the individual alumina
nanotubes could be easily separated upon applying a weak
external stress (Figure 35b), unlike the case of porous AAOs
formed by MA processes (Figure 35c). Chu et al. pointed out
that the concentration of Al3+ ions dissolved in an aged solution
plays a key role in the stable growth of porous AAOs at high
potentials (U > Ub) and current densities (j), avoiding the
burning or breakdown of porous AAO lms.18,110 However, the
electrochemical action of the aged solution has not yet been
clearly understood. In an attempt to determine the eect of
aged solutions on breakdown potential (UB), Schwirn et
al.113,119 performed a series of HA experiments at the potential
range of U = 2780 V by using sulfuric acid containing
dierent concentrations of Al3+ ions. Their results have shown
that stable anodization at high potentials (U > UB) is actually
determined by the initial limiting current density (jlimit), not by
the solution state.
For an oxalic acid-based anodization system, Lee et al.111
have shown that the self-ordering regimes can be extended by
performing HA of aluminum. By introducing a thin (ca. 400
nm) porous oxide layer onto an aluminum substrate, gradually
increasing the anodizing potential (U) to a target value (100
160 V) at the rate of 0.50.9 V s1, and eectively removing
the reaction heat through direct thermal contact of the
aluminum substrate with an underlying cooling plate, the
authors could suppress the breakdown of oxide lms and grow
mechanically stable highly ordered porous AAOs at anodizing
potentials of U > 100 V. Their HA process established a new
self-ordering regime with widely tunable interpore distances:
DHA
int = 220300 nm at U = 110150 V and j = 30250 mA
cm2. They suggested that current density j (i.e., the electric
eld E) is a key parameter governing the self-ordering of pores
7516

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

agent for lowering the freezing point of the electrolyte down to


10 C, but also as a coolant for removing a large amount of
reaction heat through its vaporization from the metal/oxide
interface. The process allowed the fabrication of self-ordered
porous AAOs with various interpore distances: DHA
int = 70140
nm for H2SO4-HA at 3080 V, 225450 nm for H2C2O4-HA
at 100180 V, and 320380 nm for H3PO4-HA at 195 V.118,122
Microscopic investigations of HA-AAOs have shown that the
barrier layer thickness (tb) increases at a rate of ARHA = 0.61.0
nm V1 with respect to anodizing potential (U) depending on
the current density (j),18,111,113,118,119 which is smaller than
ARMA 1.2 nm V1 for MA-AAOs.7,100,114 Lee et al.111
attributed the reduced ARHA (i.e., anodizing ratio) to the high
current density (j) involved in the HA process in accordance
with the high eld conductivity theory, which predicts an
inversely proportional dependence of tb to the logarithm of
current density (j) at a given anodization potential (U) (see eq
1). Figure 36 summarizes the self-ordering potentials (U) and
corresponding interpore distances (Dint) of porous AAOs
formed by MA and HA in three major pore-forming
electrolytes (i.e., H2SO4, H2C2O4, and H3PO4). It was found
that HA-AAOs exhibit a reduced potential (U) dependence of
the interpore distance (Dint) with a proportionality constant
HA = 1.82.1 nm V1,16,110,111,113,118,122,244,245 as compared to
self-ordered MA-AAOs (i.e., MA = 2.5 nm V1).109 On the
other hand, HA experiments performed at potentiostatic
conditions (i.e., U = constant) have also shown that the
interpore distance (DHA
int ) of HA-AAOs decreases with the
current density (j), indicating that anodizing potential (U) is
not the only parameter determining the cell size of porous
AAOs.16,111,113,118 Lee et al. assumed that high mechanical
stress at the metal/oxide interface due to high current density j
(i.e., high electric eld strength, E) may be responsible for the
reduced HA.111 In other words, energetically unfavorable eldinduced mechanical stress (i.e., electrostriction pressure) was
suggested to be accommodated by reducing the cell size to
increase the surface area of the metal/oxide interface. This is
reminiscent of the decrease of critical wavelength (c) of surface
undulations upon increase in the internal stress () for an
initially at surface (i.e., c as 0) according to the
model of Srolovitz:246,247

Figure 36. (af) Representative SEM images of MA- and HA-AAOs.


Panel (a) reproduced with permission from ref 226. Copyright 2008
The American Chemical Society. Panel (b) reproduced with
permission from ref 106. Copyright 1998 AIP Publishing LLC.
Panel (c) reproduced with permission from ref 229. Copyright 1998
The Japan Society of Applied Physics. Panel (d) reprinted with
permission from ref 110. Copyright 2005 Wiley-VCH Verlag & Co.
KGaA, Weinheim. Panel (e) reproduced with permission from ref 111.
Copyright 2006 Nature Publishing Group. Panel (f) reproduced with
permission from ref 118. Copyright 2006 IOP Publishing. Reprinted
with permission from IOP Publishing. (g) Self-ordering regimes in MA
(lled symbols) and HA (open symbols) by using H2SO4 (black
symbols), H2C2O4 (red symbols), H2SeO4 (green symbol), and
H3PO4 (blue symbols). The black solid lines represent the linear
regressions of the data with correlation parameters of MA = 2.5 nm
V1 and HA = 1.82.0 nm V1. Data for H3PO4-HA () show
current density (j) dependence of the interpore distance (Dint) at a
xed anodizing potential (U = 195 V). Data derived from refs 106,227
for H2SO4-MA, refs 5,106,168 for H2C2O4-MA, ref 11 for H2SeO4, refs
109,229 for H3PO4-MA, ref 113 for H2SO4-HA, ref 111 for H2C2O4HA, and ref 118 for H3PO4-HA.

c = M / 2

(67)

where is the surface energy and M is the biaxial modulus of a


solid. On the other hand, on the basis of the experimentally
observed tensile to compressive stress transition during
galvanostatic anodization in 0.4 M H3PO4, Proost et al. have
reported that the factor controlling the interpore distance (Dint)
is not likely to be an internal stress-induced surface
perturbation, but rather an electrostatic energy-induced surface
instability with the critical perturbation wavelength (c) given
by the following equation:175

at a given anodizing potential (U), which is in line with the


suggestion by Ono et al.232,240 The authors also found that HA
process produces porous AAOs at 2535 times faster growth
rates, as compared to conventional MA processes. Interestingly,
the porosity (PHA) of HA-AAOs was found to be 3.33.4%,
which is about one-third of the porosity value (PMA 10%)
that was proposed as a requirement for self-ordered MA-AAOs
by Nielsch et al.109 The experimental method has also been
applied to sulfuric and malonic acid-based HAs.16,113
As mentioned above, HA of aluminum is accompanied by a
large amount of reaction heat. Accordingly, the reaction heat
should be properly removed for stable anodization at high
potential (U > UB) and current density (j). To this end, Li et
al.118,122 added ethanol (C2H5OH) to the aqueous anodizing
electrolytes (e.g., H2SO4, H2C2O4, and H3PO4) and conducted
stable HA at high potentials (U) and current densities (j). In
their HA processes, the added ethanol served not only as an

c = (4tb /0rE2)1/2

(68)

where tb is the barrier layer thickness, 0 and r are the vacuum


and the relative permittivities of the oxide, respectively, and E is
the electric eld.248 The instability equation predicts that the
interpore distance (Dint) is a function of (tb/E2)1/2 with the
slope of 2/(/0r)1/2. Interpore distances measured during
the growth of porous anodic alumina and titania agree closely
with those predicted by their electrostatic energy-based
perturbation criterion.175
7517

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 37. (a) Current (j)time (t) transient showing spontaneous current (j) oscillation during hard anodization of aluminum at 200 V with an
unstirred 0.3 M oxalic acid. An instant solution agitation was introduced at the time indicated by the arrow in (a). (b) An enlarged jt curve after the
electrolyte agitation, together with the corresponding pore structure of AAO. Reprinted with permission from ref 103. Copyright 2010 Wiley-VCH
Verlag GmbH & Co. KGaA, Weinheim.

Recently, Lee et al.103 accidently found that porous AAOs


that experienced spontaneous current oscillations (amplitude
800 mA cm2) during a potentiostatic HA under unstirred
electrolyte conditions exhibit modulated pore structures. From
anodizing experiments performed at potential (U) = 140200
V, the authors observed that peak proles of the oscillating
currents are symmetrically sinusoidal with relatively small but
increasing amplitudes at the early stage, while asymmetric with
larger and uniform amplitudes at the lager stage of anodization
(Figure 37a). The oscillation period was observed to increase
with time. Microscopic investigation of the resulting porous
AAOs revealed that the pattern of pore modulations matches
exactly with the detailed prole of oscillating current peaks
(Figure 37b); the pore diameter and the segment length of
modulated pores increase with the amplitude and the period of
oscillating currents, respectively. The authors attributed the
spontaneous current oscillations to the periodic concentration
change of anodizing electrolyte at the pore bottom under
unstirred electrolyte (i.e., a diusion-controlled anodic
oxidation of aluminum). On the basis of the observed
dependence of the pore diameter on the anodizing current
density, they suggested that the internal pore structures of
porous AAOs can be tailored by judiciously controlling the

current density (j) during anodization of aluminum under


potentiostatic conditions.
7.3. Pulse Anodization (PA)

Ordered porous AAOs with tailor-made internal pore structures


may provide a new degree of freedom in templated syntheses of
novel nanomaterials,249251 and also serve as ideal platforms for
investigating the adsorption and separation behaviors of diverse
particles, ions, or biologically important molecules.252254 On
the basis of the experimental nding that HA of aluminum
produces porous AAOs with one-third lower porosity (P) than
MA (i.e., PHA 3% for HA and PMA 10% for MA), Lee and
co-workers111 have demonstrated fabrication of AAOs having
periodically modulated nanopore diameters along the pore axes
by combining MA and HA processes (Figure 38). In their
method, each step for modulation of pore diameter (Dp)
required a tedious manual exchange of the anodizing
electrolytes to satisfy both MA and HA conditions. In an
attempt to avoid this problem, they have recently developed an
approach for continuous modulations of internal pore structure
of porous AAOs by pulse anodization (PA) of aluminum under
a potentiostatic condition using sulfuric or oxalic acid
solution.112 Historically, pulse anodization (PA) of aluminum
was developed in the early 1960s.255,256 The process has
7518

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 38. SEM image of porous AAO with modulated pore diameters prepared by cycling mild anodization (MA) and hard anodization (HA). A
magnied image of the area marked with a white rectangle is shown as an inset. Reproduced with permission from ref 111. Copyright 2006 Nature
Publishing Group.

popularly been employed in the aluminum industry to produce


anodic lms of high technical quality (i.e., improved corrosion
and abrasion resistance) at an ecient rate of production.257259 However, the process has been out of view in
academic research in the past four decades, and has not been
applied to the development of nanostructured materials due to
the non-uniform and disordered pore structures of the resulting
anodic alumina.
In a newly developed PA process, a low potential (UMA) and
a high potential (UHA) are alternately pulsed to achieve MA and
HA conditions, respectively. A typical pulse prole is
schematically shown in Figure 39. A current recovery behavior
is observed for MA pulses, similarly to the current evolution at
the early stage of ordinary MA processes (Figure 23a). Current
is high at the initial stage, drops to a minimum value, and then
gradually increases to reach a steady value after passing an
overshoot.112,260,261 On the other hand, upon applying an HA
pulse, the current density (j) increases steeply (typically, jHA
10102 jMA) for a short period of time and then decreases
exponentially, which is the typical anodization kinetics of HA of
aluminum under a continuous potentiostatic condition.111 In
the anodization of aluminum, Joules heat (Q) for a given
period of time (t) is proportional to the square of the current
density (j):

Q = Ujt = R bj 2 t

Figure 39. Schematics showing potential (U)current (j) relation


during potentiostatic pulse anodization (PA).

of the Joules heat. The heat generated by an HA-pulsing can be


eectively dispersed during the subsequent MA-pulsing.224,263
In a typical PA process, the current density (j) changes
periodically to the values determined by pulsed potentials (i.e.,
jMA for UMA and jHA for UHA; jMA < jHA). As a consequence, the
resulting porous AAO exhibits a lamellar structure with
alternate stacking of MA-AAO slab with a smaller pore
diameter and HA-AAO slab with a larger pore diameter
(Figure 40ac), where the thickness of each oxide layer is
determined by the pulse durations (i.e., MA for MA-pulse and
HA for HA-pulse).112,224 In addition to such a structural
modulation, porous AAOs formed by PA also exhibit a periodic

(69)

where Rb is the resistance of the barrier layer.262 HA of


aluminum produces a large amount of reaction heat due to the
high anodic current density (j). The generated heat would
increase the proton activity of the acid electrolyte, and thus may
cause undesired acidic dissolution of pore wall oxide or even
local burning of the growing porous AAO via catastrophic local
ow of current (i.e., electrolytic breakdown). Therefore, the
heat should be properly removed during anodization at high
current density (j). PA provides an eective way of dissipation
7519

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 40. (a) Scheme for the preparation of porous AAOs with modulated pore diameters by pulse anodization (PA). (b) False-colored SEM image
of AAO formed by H2SO4 PA. AAO slabs formed by MA and HA pulses are indicated by MA-AAO and HA-AAO, respectively. (c) SEM image
showing modulated pore diameter. (d) SEM image of 3D stacks of MA-AAO slabs, obtained by selective removal of HA-AAO slabs by chemical
etching. Reproduced with permission from ref 112. Copyright 2008 Nature Publishing Group.

ultrasonic treatment to separated individual nanotubes from the


sample.
As was discussed in section 7.2, MA and HA exhibit dierent
dependences of the barrier layer thickness (tb) on anodizing
potential (U); ARMA 1.2 nm V1 for MA and ARHA = 0.61.0
nm V1 for HA. According to eq 1, the anodizing current
density (j) is exponentially proportional to the inverse of the
barrier layer thickness (tb). In PA of aluminum, therefore, when
the anodizing potential is changed from a higher UHA to a lower
UMA, the current density drops abruptly from jHA to a minimum
value and then increases gradually to a value (jMA) determined
by UMA (i.e., current recovery). It was reported that the time
required for a complete recovery of anodizing current depends
on the chemical nature of the barrier oxide (i.e., the content of
anionic impurities), the electrolyte temperature, and the
potential dierence between UHA and UMA.112,264 For PA in
three major pore-forming acid electrolytes, Lee and Kim noted
that the time required for current recovery increases in the
order: H2SO4 < H2C2O4 < H3PO4.264 Further, they pointed out
that PA of aluminum in H2C2O4 or H3PO4 electrolyte is
dicult to achieve continuously within a reasonable period of
time because of the retarded current recovery, especially at a
low temperature and a large potential dierence.264 To resolve
the problem associated with the slow current recovery, they
increased gradually anodizing potential prior to pulsing a high
anodizing potential.264 Porous AAOs with tailor-made internal
pore geometries could be conveniently prepared by employing
potential pulses with deliberately designed periods and

compositional modulation along the pore axes. TEM-EDX


point analysis of porous AAO formed by sulfuric acid-based PA
revealed that the amount of anionic impurities (mostly SO2
4 )
in HA-AAO slabs is about 88% higher than that in MA-AAO
slabs, which was attributed to the high current density (jHA)
during HA-pulsing.112 As discussed in section 5.2, anodic oxide
contaminated with anionic impurities exhibits a poor chemical
stability against an oxide etchant (e.g., 5 wt % H3PO4). By
taking advantage of higher level of anionic impurities in HAAAO slabs, Lee et al.112 could completely separate the MAAAO slabs from a single as-prepared porous AAO by
performing selective chemical etching of HA-AAO slabs (Figure
40d).
Structural modulation of porous AAOs can also be achived
by galvanostatic PA, where current pulses satisfying MA and
HA conditions are periodically applied. As discussed in section
7.2, H2SO4-AAO formed under HA conditions shows fairly
weak junction strength between cells.110,112,113 On the basis of
this experimental nding, Lee and co-workers263 explored a
convenient route for the mass preparation of uniform alumina
nanotubes with prescribed lengths. The authors applied
periodically galvanic MA and HA pulses to achieve continuous
modulations of pore diameter and also to weaken the junction
strength between cells. Alumina nanotubes, the length of which
is determined by the HA-pulse duration (HA), could be
obtained by immersing the resulting porous AAO into an
aqueous mixture of 0.2 M CuCl2 and 6.1 M HCl, followed by
7520

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 41. Schematics showing (a) the experimental process for the fabrication of AAO with tailor-made pore structures by pulse anodization (PA)
and (b) a generalized form of a potential pulse employed in pulse anodizations. Uj and ij dene the repeating unit of potential waves, where Uj = the
potential at the time tj with U1 = U5 (j = 14), ij = tj+1 tj, i = the pulse number (i = 1, 2, 3...). Representative SEM image of porous AAOs with
modulated pores prepared by pulse anodization; (c) 11 = 36 s, (d) 11 = 144 s, (e) 11 = 21 = 36 s, 31 = 41 = 144 s, and (f) 11 = 21 = 144 s, 31 =
36 s, 41 = 144 s, 51 = 61 = 36 s. Other parameters were xed at U1 = 80 V, U2 = 140 V, U3 = U4 = 160 V, i2 = i4 = 0 s, i3 = 0.2 s. The repeating
units of pulses are shown as insets in the respective images. Reprinted with permission from ref 264. Copyright 2010 IOP Publishing.

beginning with rst cycle having a gradually increasing


amplitude of current, then a double-proled cycle, and last a
series of triangular galvanic cycles.265 AAOs with periodically
perforated pores (i.e., pores with nanoholes along horizontal
directions) were also fabricated by chemical etching of cyclic
anodized porous AAOs.266

amplitudes (Figure 41). This capability for engineering internal


pore geometry may provide a unique opportunity in templated
synthesis of nanowires and nanotubes with modulated
diameters, and also in utilization of porous AAOs with 3D
periodic pore structures in photonic applications.
7.4. Cyclic Anodization (CA)

7.5. Anodization of Thin Aluminum Films Deposited on


Substrates

The concept of structural engineering of porous AAO by


combining MA and HA has been put forward by Losic et
al.,265,266 who developed a new anodizing process, called cyclic
anodization (CA). They employed periodically oscillating
current signals with dierent cyclic parameters (i.e., period,
amplitude, and prole) during anodization of aluminum to
achieve structural modulations in porous AAOs (Figure 42).265
Porous AAOs with modulated pores of dierent shapes
(circular- or ratchet-type), diameters, and lengths were
prepared by applying current signals of deliberately chosen
cyclic parameters. Microscopic investigation of AAOs formed
by CA processes indicated that the structural details of pores
follow exactly the applied current proles. The authors pointed
out that the transitional anodization (TA) mode, the transition
from MA to HA condition (Figure 42d), is important for
structural engineering of pores in CA process. It was proposed
that the TA mode creates a pore, whose internal geometry is
directed by the characteristics (i.e., prole, period, and
amplitude) of the applied cyclic signal. They further showed
nanostructuring of porous AAOs with distinctive, hierarchical
internal pore structures by employing multiproled current
signals: for example, three dierent successive CA steps,

In addition to overcoming their brittle characteristics, AAOs


grown with thin Al lms deposited on substrates of choice
would potentially oer much broader application than those on
bulk aluminum foils. The substrate could be insulators (e.g.,
glass),267270 semiconductors (Si and TiN),271277 non-valvemetal (e.g., Cu, Ag, Au, Pt, etc.)270,278283 or valve-metal (e.g.,
Ti, W, Nb, Zr, Ta, etc.)-coated Si substrates,284290 and
transparent indium thin oxide (ITO).291298 AAOs formed by
anodizing thin Al lms on substrates have been utilized not
only as patterning masks, but also as templates for fabricating
various functional nanostructures, including arrays of Si
nanoholes,271 carbon nanotubes (CNT),287 magnetic nanowires,289,293 thermoelectric nanowires,270 photocatalytic nanowires or nanotubes,293,296 and valve-metal oxide nanodots or
nanorods.277,284,286,288
Previous works have indicated that the anodizing behavior of
the interfacial area of the Al and substrate material depends on
the underlying material in terms of anodization kinetics and
structure of the barrier layer (Figure 43). When anodizing Al
lms on insulating substrates (e.g., SiO2) in three major pore7521

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 42. (a) Schematics of the galvanostatic cyclic anodization (CA) for structural modulation of porous AAO. (b,c) SEM images of AAOs with
modulated pores fabricated by asymmetrical current signals (i.e., exponential saw-tooth) with two dierent amplitudes. (d) Inuence of current
amplitude on the pore shape and the length of modulated pore segments. Anodization modes (i.e., MA, TA, HA) associated with corresponding
anodization currents are marked on the pore structures and graphs. Reprinted with permission from ref 265. Copyright 2009 Wiley-VCH Verlag
GmbH & Co. KGaA, Weinheim.

Figure 43. Schematic current (j)time (t) curves during anodization of thin Al lms on (a) insulating substrates, (b) non-valve-metal-coated (or
ITO) substrates, and (c) valve-metal-coated substrates. (df) Schematic cross-sections of the respective porous AAOs. The solid curve in (b)
corresponds to jt curve during anodization of silicon substrate without conducting surface layer.

Nb, Zr, Ta, etc.) are lled with corresponding oxide nanodots
or nanorods, protruding from the barrier layer (Figure
43c,f).284290
Among the above-mentioned substrate materials, the
anodization of Al lm on a Si substrate has been the most
extensively investigated. When anodizing an Al/Si substrate,
after the complete anodization of the Al lm, oxidation of the Si
surface forms SiO2 nanodots under an inverted barrier oxide
layer. The thickness of the inverted barrier layer is signicantly
reduced as compared to that formed on bulk Al
foils.271,272,299,300 Seo et al. suggested that formation of
interfacial voids and the inversion of the barrier layer has a
mechanical origin, involving multiple process stages, as follows

forming electrolytes (i.e., H2SO4, H2C2O4, and H3PO4, Figure


43a,d), the completion of the anodization is marked by a sharp
decrease in current density (j) and color change.267270 The
barrier layer has a U-shaped morphology like that of AAO
formed from bulk Al foils. Al just underneath the cell boundary
remains unoxidized in the form of discrete nanometer-sized
particles, which are trapped between the alumina barrier layer
and the insulating substrate. Unlike bulk Al foils, however, the
barrier layers of AAOs formed by anodization of Al lms on
conductor-coated substrates (e.g., Si, Pt, Au, ITO, etc.) are
characterized by inverted morphology with interfacial voids
(Figure 43b,e).270274,276282,291297 On the other hand, pores
of AAOs formed on valve-metal-coated substrates (e.g., Ti, W,
7522

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 44. Cross-sectional SEM images (top) and a schematic (bottom) showing the process of void formation: (a) touching of the barrier layer
with Si surface, (b) attening of pore bottoms due to stress accumulation, (c) void nucleation to minimize the stress, and (d) formation of inverted
barrier layer. SEM images were adapted with permission from ref 276. Copyright 2007 Elsevier.

(Figure 44).276 When the barrier layer touches the Si surface,


the residual Al underneath the cell boundary region is laterally
conned and thus cannot accommodate the volume expansion
stress due to the rigid Si substrate without interfacial
restructuring to create the necessary additional space (i.e.,
void). The driving force for void nucleation is the stress
pushing the Si substrate downward. On the other hand, upward
stresses are exerted on the bottom of each pore, resulting in the
inversion of the barrier layer curvature (i.e., detachment of the
barrier layer). The electric eld for anodizing the remaining Al
is locally concentrated on the pore edges. As a result, a
dendritic branching occurs at the edge of each pore bottom.
The upward stress increases due to void growth during
anodizing of the residual Al. The inversion behavior allows the
barrier layer to be farther away from the Si substrate, which
relieves the stress burden while the barrier layer becomes
thinner upon further anodizing.276 From TEM nanoprobe
energy dispersive X-ray spectroscopy (EDS), Seo et al.276 found
that the inverted barrier layer surrounding a void has a
compositional bilayer structure with a thin Al-rich region near
the void and relatively thick Al-poor region near the pore
bottom, the origin of which is likely to be associated with the
detachment process of the barrier layer during void formation.
Anodizing Al lm/Si substrate for an extended period of time
results in a local oxidation of Si by electrolyte inltrating
through the channels between the pore bottoms and voids,
forming SiO2 nanodots just underneath the voids,276,300
accompanied by a violent evolution of oxygen gas bubbles.
As a result of the gas evolution, the porous AAO typically
delaminates from the Si substrate. Similar phenomena have also
been observed for anodization of Al lms on ITO/glass
substrates, although Chu et al.291,292,295,298 have successfully
anodized thin Al lms on ITO/glass substrate. This may be
ascribed to improved physical bonding of the AAO as a result
of the sputter deposition of highly energetic Al atoms on the
ITO/glass substrate. In the case of Al lms on non-valve-metalcoated substrates (e.g., Cu, PdAu, Ag, Pt), prolonged
anodization results in detrimental breakdown or dissolution
of metal during the anodization.270 Accordingly, anodization for
a prescribed time is required for each substrate.
For electronic applications, it is desirable to grow a barrierlayer-free AAO directly on conductor-coated substrates. Sander
and Tan demonstrated the fabrication of a barrier-layer-free
AAO membrane by anodizing Al lm deposited on an Aucoated substrate, followed by chemical etching of the barrier
layer in 5 wt % H3PO4.278 Yet their chemical etching process
resulted in enlargement of the pore diameter, due to the

isotropic nature of anodic alumina etching. Moreover, several


minutes of anodization of the sample resulted in detachment of
the AAO from the substrate surface. Yang et al. anodized an Al/
Au/Si substrate in an eort to circumvent the need to open the
barrier layer.281 Yet, the Al/Au bilayer system formed AuAl
intermetallic phases, which catalyzed a deleterious oxygen
evolution reaction, causing detachment of the AAO membrane
from the substrate.281,283 For anodization of Al/Pt/Si
substrates, a barrier layer thinning process based on stepwise
voltage reduction279 or the use of a reverse bias in KOH270
have been employed to minimize pore widening. However, the
stepwise voltage reduction process resulted in the bifurcation of
pores near the barrier layer. Under the reverse polarization
conditions, the Pt underlayer catalyzed the electrolysis of water,
violently evolving H2 gas and thus delaminating porous AAO
from the substrate. To improve physical bonding between the
porous AAO and a conductor-coated substrate, a thin interlayer
(typically, ca. 5-nm-thick Ti) between the Al and the conductor
layer has been introduced.282,283,301 Yasui et al. introduced a
thin layer of Ti (1.5 nm) between the Al and Pt layer to serve
both as an adhesion promoter and as a barrier eliminating thin
TiO2 in 5 wt % H3PO4.282 Oh and Thompson have successfully
demonstrated a selective barrier perforation process by
anodizing Al lms on a W(60 nm)/Ti(15 nm)-coated silicon
substrate.289,290 Their process is based on selective dissolution
of the anodized valve-metal oxide (i.e., WO3) in a pH 7
phosphate buer solution. During removal of the WO3,
isotropic pore wall etching did not take place due to the
neutral nature of the etching solution. After the selective
removal of WO3 from the pore base, Oh and Thompson could
directly grow Ni or Pt nanowires by electrodeposition into the
pores of AAO, in which the exposed base metal layers served as
cathode.289,290
Pringle215 rst theoretically analyzed the anodizing behaviors
of superimposed valve-metal layers (i.e., bilayered metal lms).
He predicted that the metal order would be conserved if the
anodic oxide of the superimposed metal was less resistive, while
the metal order would be partially inverted through nger
penetration of oxide by the underlying anodic oxide if the
superimposed anodic oxide was more resistive. He also pointed
out that these phenomena may be changed by the eects of
transport number, relative rate of cation migrations, oxide
structure, and PBR.215 Later, this model was experimentally
conrmed by Shimizu and co-workers,214,302304 who observed
that the random penetrations of lanceolate oxide ngers from
the less resistive underlying oxide layer into the more resistive
upper oxide layer take place. Recent works by Mozalev and co7523

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 45. SEM images of anodic tantala nanorods formed by multi-step anodization of Al/Ta/Si substrate: (a) before and (b) after removal of
AAO. (ce) Schematic cross-sectional views of (c) the Al/Ta bilayer anodized at 53 V in 0.2 M H2C2O4, (d) after potentiostatic re-anodizing from
53 to 310 V in 0.5 M H3BO3, and (e) after pore widening followed by re-anodizing from 53 to 310 V in 0.5 M H3BO3. Reprinted with permission
from ref 307. Copyright 2004 The Electrochemical Society.

stamp (mold) onto the aluminum by mechanical pressure (i.e.,


nanoimprinting), followed by anodization. The SiC imprint
stamps were fabricated with an array of convex features of
desired arrangements in a limited dimension by electron beam
lithography (EBL) technique.308311 Each shallow indentation
formed on the aluminum surface dened the position of pore
growth by initiating pore nucleation at the initial stage of
anodization, and thus led to a perfect arrangement of pores
within the patterned area (Figure 46a). Masuda et al. further
extended the method to fabricate pore array architectures with
square- or triangle-shaped pore openings in square or triangular
arrangements (Figure 46c,d).309 In a pre-pattern on aluminum,
the missing sites of the pattern can be compensated
automatically during anodization, if the distance between the
missing site and the nearest patterned sites satises the
potential (U)interpore distance (Dint) relation required for
the self-ordering of pores (i.e., MA = Dint/U = 2.5 nm V1,
section 7.1).311 The size of the pores formed at those missing
sites is smaller than that of pores formed at the patterned sites
(Figure 46e,f). By utilizing this self-compensation ability in
AAO growth, Smith and co-workers have demonstrated the
fabrication of porous AAOs with hybrid circular-diamond and
circular-triangular-diamond pore cross-sections.312 By anodizing aluminum with surface pre-patterns in nonequilibrium

workers have indicated that in certain electrolytes, anodizing of


thin Al deposited on a layer of valve-metals (e.g., W, Ti, Ta,
Nb) results in the lling of AAO pores with nanodots or
nanorods of the corresponding metal oxides (Figure
45).20,286,305307 More recently, Chu et al.288 obtained similar
experimental results from anodization of Al lm on a Zr-coated
glass substrate. Mozalev et al.307 pointed out that the lling of
AAO pores with other valve-metal oxides is possible due to the
higher PBR and cation transport number (t+), as compared to
those of Al; for example, the PBR for Ta/Ta2O5 = 2.5 and t+Ta
= 0.240.5.

8. LONG-RANGE ORDERED POROUS AAO


Porous AAOs formed under the self-ordering conditions exhibit
a poly-domain structure, where each domain contains
hexagonally ordered nanopores of an identical orientation
and is separated by the boundaries. The domain boundaries are
characterized by defect pores (white dots in Figure 34d). For
most nanotechnology applications, porous AAOs with uniform
pore size and long-range ordering of pores are required.
Masuda et al.308 rst reported fabrication of ideally ordered
porous AAOs with a single-domain conguration over a few
mm2 area. The process involved pre-patterning of the
aluminum surface by transferring the pattern of a hard SiC
7524

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 46. (a) Schematics showing process for ideally ordered porous AAO. (bd) SEM micrographs of porous AAOs with dierent hole array
architectures: (b) circular, (c) square, and (d) triangular pore openings. (ej) SEM micrographs showing self-compensated pore formation; (eg)
surface prepatterns with missing sites of pattern and (hj) the respective porous AAOs formed by anodization. The black arrows in (h) indicate
pores formed at the missing sites of pattern in (e). SEM micrographs shown in (f,g,i,j) highlight formation of porous AAOs with (f,i) hybrid circulardiamond and (g,j) hybrid cicular-diamond-triangle pore cross-sections; the scale bars are 500 nm. Panel (b) was reprinted with permission from ref
308. Copyright 1997 AIP Publishing LLC. Panels (c,d) were reprinted with permission from ref 309. Copyright 2001 Wiley-VCH Verlag GmbH &
Co. KGaA, Weinheim. Panels (e,h) were reprinted with permission from ref 311. Copyright 2001 AIP Publishing LLC. Panels (f,g,i,j) were reprinted
with permission from ref 312. Copyright 2008 AIP Publishing LLC.

cracking the underlying brittle substrate, high mechanical


pressure also causes structural damage of the imprint stamp
after several times of pattern transfer.
In attempts to resolve the above-mentioned problems, other
pre-patterning techniques have been proposed: evaporation of
thin aluminum lm on ordered arrays of self-assembled Fe2O3
nanoparticles (NPs) on silicon followed by mechanical
striping,318 resist-assisted or direct patterning by focused-ionbeam (FIB) lithography,319,328330 and direct nanoindentation
using a tip of a scanning probe microscope (SPM).320,331 Yet
these methods are not practical for pre-patterning over an
extended area due to the limited ordered area (ca. 2 m2) of
nanoparticles for the rst technique, and due to the serial
nature of the patterning process for the other two techniques.
For large-scale fabrications of long-range ordered porous AAOs
on fragile substrates, pre-patterning of thin aluminum lms by
holographic lithography,321 block-copolymer lithography,322
and nanosphere lithography (NSL)323 has been developed.
Fabrication of single domain porous AAOs over 2-in. silicon
wafer has been also realized by directly anodizing thin
aluminum lms deposited on a lithographically generated
SiO2 mask, in which ordered SiO2 holes underneath the
aluminum lm dene the position of pore growth by eectively
guiding the electric eld during anodization.332 Most of these
techniques have demonstrated their eectiveness in fabrications
of ideally ordered porous AAOs with tunable interpore distance
(Dint) on brittle substrates over an extended dimension.
However, they are inherently incapable of generating multiple
copies of surface pre-patterns on aluminum.
Soft lithography utilizing elastomeric poly-dimethylsiloxane
(PDMS) stamp or nanoimprint lithography (NIL) on
polymeric resist has been demonstrated as a versatile method
for multiple transfers of a master pattern onto various
substrates.325,333,334 Pattern transfer in these lithographic
techniques can be achieved at pressures 103105 times less
than those used in hard stamping directly onto aluminum. Lai

tessellation arrangement (Figure 46f,g), the authors found that


the cell geometries in the resulting porous AAOs are
determined by the arrangements of the unpatterned and
patterned pore sites and also direct the cross-sectional shape of
pores: circular, diamond, and triangular pores, respectively, in
regular, elongated, and partially compressed hexagonal cells
(Figure 46i,j). The authors attributed the evolution of diamond
and triangular pore cross-sections to a coupled eect between
the thick pore wall oxide and the longer segment length of cellboundary bands (section 5.2), suppressing or eliminating the
inuence of the smaller cell segments.312 Porous AAOs with
sharp-featured non-circular pore cross-sections can be used as
templates in synthesizing functional nanostructures for
enhanced sensing in localized surface plasmon resonance
(LSPR) and surface-enhanced Raman spectroscopy (SERS).313
Stimulated by the works of Masuda and co-workers, several
groups have developed various surface pre-patterning methods
to fabricate single-domain AAOs with tunable interpore
distances, Dint (Table 2). Earlier research in this direction was
mostly devoted to the development of an economic way of
producing hard imprint stamps with large lateral dimensions.
Surface pre-patterning by mechanical nanoindentations using
optical diraction grating,314 Si3N4 mold fabricated by deep-UV
lithography,315 wafer-scale Ni mold fabricated by laser
interference lithography (LIL),317 and self-assembled monoor multi-layer of nanospheres316,327 is eective in terms of
process cost and pre-patterning area. However, these
approaches have an intrinsic limitation, in that they rely on
the transfer of stamp patterns onto the aluminum surface under
high mechanical pressure (typically, 5100 kN cm2) and
therefore are not suited for pre-patterning of thin aluminum
layers deposited on technologically more relevant substrates for
device integration, such as fragile silicon or glass. In general, as
the pattern density on an imprint stamp increases, the higher
the required applied pressure must be in this pattern transfer
protocol. In addition to severely deforming the aluminum and
7525

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

pre-patterning methods

7526

100

10 cm area

wafer scale

100

277

2 cm 2 cm

O
2

84

>1 mm2 area

45

300

1 cm 1 cm

100

1 m 1 m

100

63
481
500
150
200
13

patterned area

3 mm 3 mm
5 mm 5 mm
4-in. wafer
>cm2 area
wafer scale
2 m 2 m

X
X
X
X
X
X

reported
min Dint
(nm)

Ni mold was replicated from a master pattern fabricated by EBL, and its pattern was transferred onto a thermoplastic resist; subsequently, the resulting
resist pattern was transferred onto an Al surface by Ar-ion beam milling; imprint pressure: 0.5 kN cm2
ordered arrays of nanoindents were generated by wet-etching of Al through pre-patterned polymer masks that were created by SFIL on Al surface;
imprint pressure: <103 kN cm2

pre-patterning of Al lm on a substrate was achieved by chemical etching of Al through the mask formed by sputter-deposition of tungsten on the
ordered arrays of polystyrene sphere
pre-patterning of Al was achieved by contacting with Al etchant-absorbed PDMS mold

thin Al lms with periodic surface undulations were generated by depositing Al onto the photoresist-grating-patterns, developed by holographic
lithography
highly ordered hole array pattern of thin block copolymer lm was transferred to the Al surface by reactive ion etching (RIE)

the surface of Al was directly patterned by nanoindentation using the tip of a scanning probe microscope (SPM); indentation load: 40 N

326

325

324

323

322

321

320

319

pre-patterning of Al surface was achieved by focused Ga+-ion beam

refs
308,310
314
315
316
317
318

remarks
SiC imprint stamp was fabricated using electron beam lithography (EBL); stamping pressure: 28 kN cm2
Al surface was pre-patterned by a two-step press-in procedure of optical diraction grating
stamp with arrays of Si3N4 pyramids was replicated from a silicon master; stamping pressure: 5 kN cm2
hexagonal array pattern of self-assembled nanosphere crystal was transferred onto Al surface; stamping pressure: 98 kN cm2
Ni imprint stamps were replicated from master patterns, fabricated by laser interference lithography (LIL); stamping pressure: 525 kN cm2
6-m-thick Al was deposited on self-assembled 2D arrays of Fe2O3 NPs on Si, and separated by using an adhesive tape for anodization

Pre-patterning capability of thin Al lm deposited on fragile substrates.

block-copolymer
lithography
nanosphere lithography
(NSL)
mold-assisted chemical
etching
nanoimprint lithography
(NIL)
step and ash imprint
lithography (SFIL)

SiC stamp
optical diraction grating
Si3N4 stamp
nanosphere crystals
Ni stamp
Fe2O3 nanoparticles
(NPs)
focused-ion-beam (FIB)
lithography
probe-tip-based direct
patterning
holographic lithography

Al lm on
substratea

Table 2. Methods of Surface Pre-patterning of Aluminum for Ideally Ordered Porous Anodic Aluminum Oxide (AAO)

Chemical Reviews
Review

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 47. (a) Schematic diagrams showing fabrication procedures of ideally ordered porous AAO. (be) SEM images of porous AAOs fabricated
by the procedures shown in (a): (b) an overview image, (c) an image showing clear distinction between the pre-patterned (left) and non-patterned
(right) pore arrangement, (d) an SEM image of porous AAO with square arrangement of pores, and (e) a cross-sectional image of porous AAO
grown on silicon wafer. (f) A photograph of AAO template on a 4-in. silicon wafer. Panels (af) were reproduced with permission from ref 326.
Copyright 2010 The American Chemical Society.

Figure 48. (a) Cross-sectional SEM image of porous AAO formed by anodization of pre-patterned aluminum at an outside self-ordering condition:
pattern interval = 200 nm, anodization at U = 80 V using 0.04 M H2C2O4 (17 C). (b) Dependence of the depth of ordered pores on the anodizing
potential (U): pattern intervals = 70, 100, 150, 200, and 250 nm for U = 28, 40, 60, 80, and 100 V, respectively. Anodizing electrolyte = 0.3 M
H2C2O4 for U = 28, 40, 60 V and 0.04 M H2C2O4 for U = 80 and 100 V. Reproduced with permission from ref 335. Copyright 2001 The
Electrochemical Society.

et al.324 reported a generic method of pre-patterning of an


aluminum surface by mold-assisted chemical etching. Their
technique is based on the reaction-diusion wet-stamping (RDWETS) process, which creates ordered arrays of shallow etch
pits on aluminum by the absorption/liberation of aluminum
etchant adsorbed in a PDMS stamp. Single-domain porous
AAOs with a square lattice conguration over 2 2 cm2 area
were fabricated by anodization of the patterned aluminum.
Recently, Kustandi and co-workers have demonstrated

fabrication of single-domain porous AAOs with arbitrary


interpore distances (Dint) and dierent pore lattice conguration on 4-in. silicon wafer (Figure 47).326 Their approach is
based on the step and ash imprint lithography (SFIL) using a
patterned quartz template. The SFIL was employed to pattern a
polymer mask layer on aluminum lm. Pre-patterning of the
aluminum surface was achieved by transferring the pattern of
the polymer mask onto the underlying aluminum lms by wetchemical etching. The authors claimed that the demonstrated
7527

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 49. General scheme for electrochemical deposition (ECD) of materials into porous AAO.

process provides signicant merits over existing surface prepatterning approaches in terms of patterning area, process
simplicity, robustness, and throughput, allowing up to 10 000
times of pattern transfers to the surface of aluminum lms
deposited on fragile substrates.326
As discussed above, anodization of pre-patterned aluminum
is indeed an eective way for obtaining ideally ordered porous
AAO. It should be noted, however, that there is a major
limitation of the process; to maintain the initially ideally
ordered pore conguration, the obtainable maximum aspect
ratio of uniform nanopores depends critically on the
anodization conditions. For anodization of aluminum of a
given pattern interval, one should properly select anodizing
parameters (e.g., potential and electrolyte) to fulll the selfordering requirement of pores. As discussed in section 7, for
MA and HA using three major pore-forming acid electrolytes,
there are specic ranges of anodizing potential (U), which give
self-ordered pore structure. Under other applied potentials, the
depth of nanopores is limited in terms of ideal ordering (see
Figure 48).335

interested readers to excellent recent review articles given in


refs313 and 336, that extensively cover the templated synthesis
and assembly of low-dimensional nanostructures and their
applications. As discussed in previous sections, self-ordered
porous AAOs with tightly controlled pore size, density, and
intervals can be obtained by anodizing aluminum substrate
under proper conditions. This provides many unique
opportunities in the templated synthesis of low-dimensional
functional nanostructures, allowing simple and cost-eective
preparations of extended arrays of structurally well-dened and
identical nanostructures and also overcoming many of the
drawbacks of conventional state-of-the-art lithographic techniques. During the last two decades, by taking advantage of its
highly ordered structural feature, porous AAO has intensively
utilized as a template, stencil mask, or scaold for functional
nanostructures or nanodevices. The approaches employed for
developing functional nanostructures include electrochemical
deposition (ECD), atomic layer deposition (ALD), chemical
vapor deposition (CVD), physical vapor deposition (PVD),
solgel deposition, surface modications with technologically
important (bio)molecules, polymers or nanoparticles, melt
impregnation of materials, reactive ion etching (RIE), etc. A
vast variety of nanostructures including carbon nanotubes
(CNTs), metal, semiconductor, or polymer nanowires/nanotubes, and ordered arrays of nanodots/nanoholes on various
substrates have been successfully fabricated by utilizing porous
AAO as a template or stencil mask. Their structureproperties
and functionalizations have been intensively investigated in an
attempt to utilize them for practical applications in safety,
energy, information, and biomedicine. In this section, we
discuss AAO template-based synthesis of low-dimensional
nanomaterials, their functionalizations, and applications.

9. AAO TEMPLATE-BASED SYNTHESIS OF


FUNCTIONAL NANOSTRUCTURES
While nanostructured materials are considered to have huge
application potential, cost-eective, high-throughput, and
reproducible synthetic strategies are an essential prerequisite.
The synthesis of both simple and complex low-dimensional
nanostructures (e.g., nanodots, wires, tubes, core/shell NPs,
organicinorganic nanohybrids, etc.) relies largely on templates, whose size, structure, and physicochemical properties are
predened. Templated synthesis provides robust ways of
precise control over the size, shape and conguration, and
growth direction and place of otherwise unattainable nanostructured materials. More importantly, it may also oer
opportunities for the in situ assembly of discrete nanostructures
(i.e., building blocks) into a well-dened hierarchical
architecture for practical device applications. In principle,
almost all existing synthetic approaches and their combination
can be applied to the templated synthesis of nanostructured
functional materials. Templates can be any substance with
nanostructured features, including DNA, protein, viruses, living
organisms, colloidal nanoparticles, nanowires, nanotubes, block
copolymers, porous materials, etc.; the present authors refer

9.1. Electrochemical Deposition (ECD)

Possin337 implemented rst the electrochemical deposition


(ECD) of metal into a track-etched mica template to obtain 40nm-thick and 15-m-long Sn, In, and Zn nanowires. The
template approach was drawn a renewed attention by the works
of Martin and co-workers and has been popularly employed as
a nanofabrication strategy.338342 The electrochemical deposition of materials into the pores of AAO template provides
marked advantages over other preparation methods of
nanowires or nanotubes. In comparison to other deposition
techniques, such as chemical vapor deposition (CVD), atomic
7528

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 50. Scheme of the nanodisk code method. (a) Synthesis and functionalization. (b) Thirteen possible 5-disk-pair nanodisk codes with the
corresponding binary codes. (c) SEM images of Au/Ni multisegmented nanorods before (top) and after (bottom) deposition of the backing layer
and Ni etching. (d) Two-dimensional (top) and three-dimensional (middle) scanning Raman microscopy images of a 11111 nanodisk code.
Representative Raman spectrum of methylene blue (bottom) taken from the center of the hot spot generated in the middle disk pair shown in the
Raman maps above. Panels (a,b,d) were adapted from ref 358 with permission. Copyright 2007 The American Chemical Society. Panel (c)
reproduced with permission from ref 359. Copyright 2009Macmillan Publishers Ltd.: Nature Protocols.

focus our discussion on multisegmented metallic nanowires


consisting of two or more dierent materials because of their
versatilities in terms of degree of freedom in functionalization
for various applications.
Multisegmented metallic nanowires can conveniently be
prepared by sequentially changing the electrolytic solution
during the electrodeposition. Judicious control of the length of
each segment allows one to obtain submicrometer barcodes,
which can be used as platforms for multiplexed bioassays.251,344,354,355 The dierential reectivity of adjacent
metal segments and the selective self-assembly of appropriate
molecules on specic metal segments enable the identication
of striping patterns by conventional optical microscopes. More
recently, multisegmented nanowires have been utilized for
developing nanogap devices. Qin et al.356 developed a generic
approach to lithographically process 1D nanowires. The
procedure, termed on-wire lithography (OWL), involves
selective removal of one of the components comprising a
multisegmented nanowire to create gaps. Before gap creation
via the wet-chemical etching of targeted segment(s), one side of
the multisegmented nanowires is subjected to the deposition of
a thin layer of insulting SiO2, which acts as a support for the
nanogap device. In this way, the authors could create
nanometer-sized gaps (5 to several hundred nm) on nanowires.
Further, by employing dip-pen nanolithography, they deposited
nanoscopic amounts of conducting polymer within the created
nanogaps to investigate the transport properties of nanogaps.
More recently, Chen et al.357 demonstrated heterometallic
nanogaps for molecular transport junctions (MTJs). Thiolterminated molecules were assembled into heterometallic
nanogaps (i.e., Pt/2-nm gap/Au) to observe their molecular
diode behavior. The on-wire lithography (OWL) technique was
further extended to develop a new encoding system by the
same group. By taking advantage of the facile control of the

layer deposition (ALD), and physical vapor deposition (PVD),


the electrochemical deposition into porous templates is simple
and inexpensive, and can be conducted without any special
equipment.
In general, to synthesize nanowires, the same general
procedure can be applied irrespective of the materials to be
deposited (Figure 49). First, a thin Ag layer is deposited onto
one face of an AAO membrane. This Ag layer serves as the
working electrode in the deposition of desired materials. Next, a
thin layer of sacricial Ag (or Ni) is electrochemically deposited
into the pores. This process is recommended to avoid a socalled puddling eect that typically causes nailhead shaping of
one end of the deposited nanowire due to the geometric details
of the pore mouth.342 After that, the desired material is
electrochemically deposited. The resulting nanowires/AAO
composite sample then is dipped into HNO3 solution to
remove the Ag working electrode layer and the sacricial layer
(Ag or Ni). The nanowires can be collected by dissolving the
AAO template using an appropriate AAO etchant (typically,
KOH or H3PO4). The choice of oxide etchant depends on the
material deposited; the etchant solution should not react with
the nanowire material. The diameter of the resulting nanowires
is determined by the pore size of the porous AAO template,
while their length is proportional to the total amount of charge
passed during the electrochemical deposition. Various metal
nanowires (e.g., Au,264,343,344 Ag,344,345 Pt,343,346 Ni,343,347
Pb,348 Cu,348 Zn,349 Co,350352 Sb353) have been synthesized
in porous AAO templates. These single component 1D metallic
nanowires have been used as model systems for systematically
investigating various research issues in chemistry and physics,
for example, the catalytic, magnetic, thermoelectric, and
plasmonic properties of 1D nanostructures. Although single
component metallic nanostructures (e.g., nanowire or nanotube) are useful for studying structureproperty relations, we
7529

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

(Figure 52).367373 In pulsed electrochemical deposition using


a two-component electrolytic solution, the cathodic potential is

length of segments in the nanowire, dispersible 1D objects


containing arrays of nanodisks were prepared and functionalized with Raman active chromophores.358,359 This allowed
encoding of individual nanodisks both physically and
spectroscopically (Figure 50). As proof-of-concept, the authors
demonstrated multiplexed DNA detection at target concentrations as low as 100 fM.
Component materials in the multisegmented nanowires do
not need to be metals. They could be semiconductors or
conducting polymers. Park and co-workers.360,361 have found
that two-component nanorods consisting of metal/conducting
polymer layer can behave as mesoscopic amphiphiles. They
synthesized segmented Au/polypyrrol (Ppy) nanorods by
electrochemical deposition of Au into porous AAO template,
followed by electrochemical polymerization of pyrrol.360 The
length of each segment could be conveniently tuned by
controlling the deposition time (i.e., the total amount of
charge). Au/Ppy nanorods released from the template exhibited
amphiphilic characteristics that originated from hydrophilic Au
segments and hydrophobic Ppy blocks. Individual Au/Ppy
nanorods self-assembled into microscale 2D sheets or 3D
bundles or tubes depending on the Au/Ppy segment length
ratio (Figure 51).

Figure 52. TEM analysis of a freestanding Fe/Au barcode nanowire:


(a) elemental line scanning of Fe and Au composition along the
nanowire (Au, red; Fe, green), (b) elemental mapping of Au, and (c)
elemental mapping of Fe. Reprinted with permission from ref 643.
Copyright 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

alternately pulsed between values above and below the


reduction potential of the less noble metal component.362
When the less negative potential is pulsed, more noble metal is
exclusively deposited. However, when depositing a less noble
metal by pulsing the more negative potential, the more noble
metal can be co-deposited. Therefore, the concentration of the
nobler metal ions should be held low enough as compared to
that of the less noble metal ions, so that the deposition of the
nobler metal ions during the pulsing of more negative potential
is limited by ion diusion.342,362 Recently, Liu and coworkers374 have demonstrated fabrication of tailor-made
inorganic nanopeapods (i.e., Co@CoAl2O4) nanowires by
pulsed electrochemical deposition of Co/Pt multilayers into
porous AAO template, followed by a solid-state reaction
between Co and Al2O3 at a high temperature.
Fabrication of multisegmented metal nanotubes has also
been demonstrated by Lee et al.375 Their method was based on
the preferential electrochemical deposition of metal along the
surface of pore walls decorated with metallic nanoparticles. The
authors immobilized Ag nanoparticles (AgNPs) on the pore
wall surfaces of an AAO template by utilizing spontaneous
reduction of Ag+ by Sn2+ as the following reaction:

Figure 51. SEM images of open superstructures formed by


kinetically controlled, shape-directed assembly of Au/polypyrol
(Ppy) nanorods: (a) a superstructure that has not fully closed, (b) a
fully closed superstructure, and (c) a physically cleaved area of the
superstructure showing the individual rods that have been assembled
into the superstructure. Scale bars for (a,b) = 100 m. Scale bar for (c)
= 50 m. Reproduced with permission from ref 361. Copyright 2008
Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

+
4+
2Ag +aq + Sn 2surface
2Ag 0surface + Sn surface

(70)

The process for AgNP immobilization is a simple variation of


the well-established sensitization-pre-activation protocol applied to AAO376,377 or track-etched polycarbonate template338,378 prior to the electroless deposition of metals (see
section 9.2). Sn2+ was rst chemisorbed on the pore wall
surfaces by dipping the AAO membrane into an aqueous
mixture solution of 0.02 M SnCl2 and 0.01 M HCl for 2 min.
After thorough rinsing with water and drying, the resulting
AAO template was immersed into 0.02 M AgNO3 solution.
This cycle, usually repeated about six times, resulted in uniform
deposition of AgNPs on the oxide surfaces. With the AgNPimmobilized AAO template, metallic nanotubes, embedded in
the AAO, were synthesized by electrochemical deposition. By
periodically changing the electrolytic solution during electrochemical deposition of the metal, the authors could synthesize
mutisegmented Au/Ni nanotubes (Figure 53).

As a more convenient approach, pulsed electrochemical


deposition can be used for the preparation of multisegmented
nanowires with prescribed periodicity. Yahalom and Zadok362
utilized pulsed electrochemical deposition technique to
fabricate Cu/Ni superlattice lms by using a single electrolyte
containing Cu and Ni salts. Pulsed electrochemical deposition
has been extended by other researchers to prepare multilayered
lms of magnetic/nonmagnetic metal couple to investigate the
giant magnetoresistance (GMR) eect.363366 The technique
has also been implemented in track-etched polycarbonate
membrane or porous AAO membrane to synthesize periodically multisegmented metal nanowires with sharp interface
7530

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Kohli et al.381 noted that the above electroless deposition


method applied to polycarbonate membranes does not work in
porous AAO templates because of insucient binding sites for
the Sn2+ on the pore wall surfaces. The authors solved this
problem by modifying the surfaces of porous AAO with 2(succinic anhydride)propyl trimethoxysilane (SAPT), which
forms covalent AlOSi bonds with OH functionalities on
the oxide surfaces. The anhydride in the SAPT molecule
hydroyzes to dicarboxylic acid, which chelates Sn2+ ions during
the sensitization process. As an alternative approach to increase
the density of oxygen groups on the oxide surface, Yu et al.394
incubated porous AAO membrane in 35% hydrogen peroxide
solution prior to Sn2+-sensitization. The rate of electroless Au
deposition was reported to increase with the pH of the plating
bath.381 Uniform Au nanotubes within pores of AAO templates
have been obtained by performing electroless deposition at the
pH < 10 and at temperature below 4 C.381,394 At pH > 10 or at
a high temperature, either porous AAO dissolves and/or the
pores of the AAO template become closed due to the higher
rate of metal deposition relative to the rate of mass transfer of
Au(I) and formaldehyde down to the pores. Yu et al.394
reported that Au nanotubes formed by electroless deposition
are characterized by nanoclustered morphology, in which the
size of the gold nanoclusters increases with the pH of the
plating bath. They showed that the size of the gold nanoclusters
determines the catalytic properties of Au nanotubes embedded
in AAO membrane. Further, they showed that Au nanotubes/
AAO composite membranes can be reused in the catalytic
conversion of 4-nitrophenol into 4-aminophenol in the
presence of NaBH4 as a reductant.
Activation of the porous AAO membranes can also be
accomplished by immobilizing PdNPs on the pore wall
surfaces.395397 Fabrications of arrays of Co,398 Ni,398
Cu,398,399 Pd,400 Pt,400 binary398 or ternary401 metal alloy
nanotubes, nanowires,402 and nanocones403 have been
demonstrated by electroless deposition of metals into PdNPimmobilized porous AAO membranes.

Figure 53. SEM images of multisegmented metal nanotubes with a


stacking conguration of Au/Ni/Au/Ni/Au along the nanotube axis:
(a) false-colored cross-sectional SEM image of as-prepared metal
nanotube-AAO composite. (b,c) SEM images of mutisegmented metal
nanotubes after removal of the AAO template. Adapted with
permission from ref 375. Copyright 2005 Wiley-VCH Verlag GmbH
& Co. KGaA, Weinheim.

9.2. Electroless Deposition (ELD)

Electroless deposition of metals into porous templates was


pioneered by Martin and co-workers.378381 Most of the earlier
works were devoted to electroless deposition of gold within the
pores of track-etched polycarbonate membranes. The typical
deposition process is composed of the following three general
steps:378,382,383
(1) Sensitization: This is accomplished by immersing the
porous polymeric template into an aqueous solution of SnCl2
to deposit Sn2+ onto the surfaces of the membrane.
(2) Activation: This is accomplished by dipping the Sn2+sensitized polymeric membrane into a AgNO3 solution, which
yields metallic AgNPs on the membrane surfaces through a
surface redox reaction (see reaction 70).
(3) Electroless deposition: This entails immersing the
activated polymeric membrane into a gold plating solution, in
which the surface-bound AgNPs act as catalyst for the
reduction of Au+ to yield AuNPs on the membrane surface
through the following reaction.
Au+aq + Ag 0surface Ag 0surface + Ag +aq

9.3. SolGel Deposition

Solgel processing has developed into a versatile protocol for


the stoichiometric synthesis of diverse nanocrystalline materials.
In general, solgel chemistry involves the hydrolysis of
precursor molecules under an acidic condition to prepare a
suspension of colloidal particles (i.e., sol) and the subsequent
condensation of the sol particles to obtain a gel. The resulting
gel is then calcined to obtain the desired material. Metal
alkoxides in organic media or inorganic salts in aqueous media
can be used as precursors. Martin and co-workers pioneered the
solgel porous AAO templating method for the synthesis of
various semiconductor or insulator oxide nanostructures
(nanowires and nanotubes) including TiO2, ZnO, WO3,
V2O5, MnO2, Co3O4, and SiO2.404406 Since then, many
investigators have employed the solgel deposition method to
prepare a vast variety of oxide nanostructures to investigate, for
example, photovoltaic (TiO2),407,408 gas sensing (SnO2,
ZnO),409,410 ferromagnetic (CoFe2O4),411,412 ferroelectric
(BiFeO3, SrBi2Ta2O9, PbZr0.52Ti0.48O3),413415 superconducting (YBa2Cu3O7-, Bi2Sr2CaCu2Oy),416,417 and luminescence
(Y2O3:Eu, TiO2, ZnO:Dy) properties.418420
Typically, when depositing solgel into oxide nanopores, an
AAO is directly dipped into a solution containing sol particles
for a given period of time. After thermal treatment, either
nanowires or nanotubes of inorganic oxide are formed within

(71)

The surface-bound AuNPs act as autocatalysts for the reduction


of Au+ to metallic Au0 in the presence of a reducing agent (e.g.,
formaldehyde). Au deposition starts at the entire surface of
membrane (i.e., pore walls and membrane faces). As a result,
the surfaces of pore walls and the faces of the membrane can be
coated with a thin Au layer, yielding Au nanotubes within the
pores and continuous Au lms on the faces of the membrane.
The resulting gold nanotube/polymer composite membranes
have been successfully used for diverse applications, for
example, for selective ion-transport,379,382387 (bio)molecule
separations,380,388,389 biosensing, and electroanalysis.390393
7531

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

the authors were able to synthesize very thin single-crystalline


TiO2 nanowires (diameter 10 nm) by performing heattreatment and subsequently removing the AAO template.
As a surface modication protocol, the solgel template
technique oers a large degree of freedom in nanotechnology
applications of porous AAO. Lee et al.425 demonstrated that
solgel-derived silica nanotubes/AAO composite can be used
as a synthetic bio-nanotube membrane for separating two
enantiomers of a chiral drug, in which individual silica
nanotubes act as nanometer-sized chromatography columns
in parallel. They deposited thin-walled (<3 nm) silica
nanotubes within porous AAO by adopting solgel template
synthesis approach. Subsequently, the inner wall surfaces of the
silica nanotubes were functionalized with aldehyde terminated
silane. This surface functionalization allowed the conjugation of
an antibody that selectively binds to one enantiomer of a drug
4-[3-(4-uorophenyl-2-hydroxyl-1-[1,2,4]triazol-1-yl-propyl]benzonitrile, which has two chiral centers and four stereoisomers: RR, SS, SR, and RS. The silica nanotubes can be
individually released from the porous AAO template to obtain
silica nano test tubes.426428 In another study, the inner and
outer surfaces of silica nano test tubes were selectively
derivatized with dierent functional groups using silane
chemistry.426 Silica nano test tubes with hydrophilic outer
and hydrophobic inner surfaces were demonstrated to be ideal
for extracting lipophilic molecules from aqueous solution.426 In
addition, they can be capped with either polystyrene latex
nanoparticles429 or gold nanoparticles427 to create nanocapsules. Their facile surface functionalization, loadable large
hollow interior, low level of cytotoxicity, and ease of dispersion
make the silica nano test tubes an eective drug/gene delivery
system.428,430
Yamaguchi et al.431 demonstrated that a porous AAO
template can used to fabricate a silicasurfactant hybrid
membrane containing nanochannels oriented in parallel with
respect to the pore axis of AAO. They used a precursor solution
containing tetraethoxylsilane (TEOS) as the silica source and
cationic cetyltrimethylammonium bromide (CTAB) as a
structure guiding agent. By applying moderate aspiration, the
authors were able to introduce precursor solution into the
pores of the AAO. Plastschek et al.432,433 demonstrated that the
orientation of mesoporous silica structure in porous AAO can
be systematically adjusted through combination of solgel
process and evaporation-induced self-assembly. Systematic
variations of the type and amount of structure guiding agent
and also the amount of added inorganic salts allowed the
authors to investigate systematically the connement eect of
the interfacial interaction on the orientation of silica nanochannels. Their experiments revealed that ionic CTAB
produces a columnar hexagonal 2D structure parallel to the
vertical nanochannels of the porous AAO membrane, whereas
nonionic surfactants, such as Pluronic P123 [poly(ethylene
oxide)20-poly(propylene oxide)70-poly(ethylene oxide)20] and
Brij 56 [decaethylene glycol hexadecyl ether], prefer the
formation of either a circular hexagonal 2D structure normal to
the nanopores of AAO or a structure with phase mixtures of
circular and columnar orientations (Figure 54). Platschek et al.
found that the moisture content of the deposited sample and
the relative humidity during the drying process inuence the
microscopic separation.433
The cylindrical pore morphology of AAO, and the pores
dimensional tuneability, make porous AAO an ideal model
system for systematically investigating the connement eect

the pores. In general, immersion of porous AAO template into


a sol solution for long periods of time yields nanowires, while
immersion for short periods of times yields nanotubes. The
formation of nanotubes indicates that the sol particles are
adsorbed onto the pore wall surfaces due to electrostatic
interaction between the negatively charged pore wall surfaces
and the positively charged sol particles.404 Lakshmi et al.404
reported that the gelation occurs at a faster rate within the
nanopores than in the bulk reservoir, due possibly to local
increase in sol concentration at the pore wall surfaces. In their
SiO2 nanotube synthesis, Zhang et al.421 found that the
viscosity of the sol solution is a key factor determining the
morphology of the resulting nanotubes. Viscosity of sol
solution increases with aging time. They obtained nanowires
from a porous AAO templated dipped for 1 min in a sol
solution aged at room temperature for 2 days. From a solution
aged for 30 days, they obtained SiO2 nanotubes connected with
nanowires. They also found that the formation of the
nanotubes depends strongly on the temperature of the sol
solution. For a short dipping time less than 1 min, bamboo-like
nanowires were prepared from a sol solution at 50 C. On the
other hand, perfect nanotubes with sharp walls were
synthesized from a solution at a lower temperature (5 C).
As the temperature of the sol solution decreased, the inside wall
of SiO2 nanotube became smoother.
Limmer et al.422 pointed out that lling of sol particles into
oxide nanopores is mainly driven by the capillary force. Heattreatment of the sol-coated AAO templates often results in 1D
porous nanostructures or hollow tubes due to insucient
packing of sol particles with the oxide nanopores. Limmer and
co-workers422,423 developed an electrophoretic solgel process
to resolve the problems associated with low particle packing
density. In their approach, the electrophoretic motion of the
positively charged sol particles under an electric eld is utilized.
The authors demonstrated successful syntheses of technologically important oxide nanowires (e.g., TiO2, BaTiO3, SiO2,
SrNb2O6, Pb(Zr,Ti)O3; diameter = 70200 nm) by electrophoretic deposition of sol particles into nanopores of tracketched polycarbonate membranes followed by heat-treatments.422,423
Although the electrophoretic sol deposition process has
distinct merit, it is rather dicult to apply the process to porous
templates with smaller pore diameters (<50 nm).422 The
problem may arise due to the decreased diusivity of sol
particle in narrower pores or the size of the sol particles
(typically, 10100 nm). To resolve this issue, Maio et al.424
employed a cathodically driven solgel process and prepared
single crystalline TiO2 nanowires. The authors employed a Ti
precursor solution containing NO3 ions. The local increase in
pH at the surface of electrode under a cathodic bias gives rise to
the hydrolyzation of the Ti precursor through the following
reactions:424
NO3 + 6H 2O + 8e NH3 + 9OH

(72)

TiO2 + + 2OH TiO(OH)2 (sol)

(73)

The resulting TiO(OH)2 sol particles are converted into a 3D


network of titanium oxyhydroxide gel within the nanopores of
AAO template via the following electrochemical reaction:
TiO(OH)2 (sol) x H 2O TiO1 + x (OH)2 2x (gel)

(74)

As such, both the sol formation and the subsequent gelation


occur within the nanopores. By taking advantage of this fact,
7532

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

a rather complex evolution of mesochannels with the geometric


connements. Figure 55 summarizes the evolution of the
conned mesostructures as a function of the channel diameter
of AAO template. When the AAO nanochannel diameter was in
the range 5573 nm, the conned silicasurfactant composite
mesophase consisted of three coaxial layers: a straight inner
mesochannel and two outer concentric mesochannels with
diverse morphologies of stacked doughnuts, single helix (Shelix), or double helix (D-helix). For the connement
dimension of 4954 nm, a coaxial double-layer helix
mesostructure was evolved; the inner core = S-helix and the
outer layer = S-helix or D-helix. On reducing the connement
dimension down to 3445 nm, the inner core became a
straight mesochannel (diameter = 1316 nm), and the outer
layer consisted of stacked doughnuts, S-helix, or D-helix. For
further reduced connement dimension of 31 nm, a singlelayered D-helix mesostructure was observed. For connement
dimension below 30 nm, inverted peapod-like mesoporous
morphology, rather than helical mesostructures, was observed.
For the connement dimension below 18 nm, a line of aligned
mesocages was formed. The authors pointed out that the
mesoporous silica frameworks in their composite mesophases
can be preserved even after the surfactant removal, which is
markedly dierent from that case of conned polymers.
9.4. Surface Modication

Coupling appropriate molecules to the pore wall surface of


AAO can greatly expand the application capability of porous
AAO for the synthesis of new functional nanomaterials, sensing,
recognition, and the controlled release of biologically relevant
molecules. Surface functionalization with appropriate molecules
or particles provides an opportunity for tuning of the surface
properties of porous AAO, such as hydrophilicity, surface
charge, reectivity, membrane selectivity, antifouling resistance,
etc. The surfaces of porous AAO can be functionalized with a
wide variety of molecules, such as n-alkanoic or uorinated
organic acid,435437 organosilanes,438,439 and phosphonic
acids.438,440 On the other hand, most studies on surface
modications of porous AAO are based on silanization
chemistry. Organosilane compounds can be covalently bound
on the surfaces porous AAO membranes, by taking advantage
of the rich hydroxide groups on the oxide surfaces. Organo-

Figure 54. Plane view TEM images of (a) the columnar oriented
hexagonal mesostructure templated with CTAB, (b) the sample
templated with Brij at 60% humidity, and (c) the sample templated
with P123 at 60% humidity. Reproduced with permission from ref 432.
Copyright 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

on molecular organization. We et al.434 synthesized silicaP123


composite mesostructures within cylindrical nanochannels of
AAO by employing a solgel method. It is well-known that this
precursor forms hexagonal silica mesostructures (i.e., SBA-15),
in which the long axes of the mesochannels are aligned parallel
to the surface of at substrates. However, the authors observed

Figure 55. Summary of the experimentally observed evolution of mesostructures conned within AAO nanochannels of varying diameters.
Reproduced with permission from ref 434. Copyright 2009 Macmillan Publishers Ltd.: Nature Materials.
7533

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 56. Schematic procedure for fabricating an AAO membrane with three dierent layers displaying distinct chemical functionalities. Adapted
with permission from ref 442. Copyright 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

explosive molecule, 2,4-dinitrotoluene, the AuNP-decorated


AAOs with 3D nanochannel arrays exhibited Raman enhancement with a factor of about 105, as compared to that observed
on a AuNP monolayer with identical surface densities of
AuNPs. The authors suggested that the AAO membrane itself
could contribute extra enhancement of Raman signals, not only
providing additional sites for molecule adsorption, but also
guiding the light by the vertically aligned nanopores. A similar
line of experimental results was also reported for AgNPdecorated AAOs.449,450
Rubinstein et al.446 prepared nanoparticle-decorated nanotubes (NPNTs) by passing a citrate-stabilized AuNP solution
through the pores of the AAO membrane, the surface of which
was modied with APTMS (or APTES). The formation of Au
NPNTs was assumed to involve the aggregation of surfaceconned AuNPs accompanied by spontaneous room-temperature coalescence upon drying. With the same approach,
multiwalled bimetallic Au/Pd NPNTs were also prepared.451 In
another study, Wang et al.398 utilized the terminal amine groups
of surface-attached APTES to complex with Sn2+-sensitizer in
electroless depositions of Co, Ni, and Cu nanotubes within
pores of AAO template. The surface-bound Sn2+ reduced Pd2+
into metallic PdNPs, which acted as catalyst during the
electroless deposition of metals. Recently, Tanvir et al.452
immobilized covalently recombinant human cytochrome
(CYP2E1) and glucose-6-phosphate dehydrogenase (G6PD)
on the pore wall of APTES-functionalized AAO membranes
through N-succinimidyl-3-maleimidoproprionate and glutaraldehyde cross-linker, respectively. The authors showed that the
immobilized enzymes can retain 100% of the activity of free
enzyme. A miniaturized heterogeneous membrane reactor was
realized by stacking two independently modied membranes
and placing them in a uidic device, in which the rst
membrane regenerates the cofactor NADPH from NADP+ and
glucose-6-phosphate and the second one utilizes it for CYP2E1based catalysis. Likewise, a cell adhesion peptide has also been
grafted on APTMS-derivatized porous AAOs to improve
osteoblast adhesion.453 Swan et al.453 reacted the APTMSfunctionalized AAO membranes with N-succinimidyl-3-maleimidopropionate and subsequently with a cellular adhesion
peptide, arginine-glycine-aspartic acid-cysteine (RGDC).

silanes on AAO surfaces can act as linkers to immobilize


nanoparticles, polymers, proteins, DNA, and other molecules
for additional functionalities. Among a variety of organosilane
compounds, aminosilanes (e.g., 3-aminoproyl trimethoxysilane
(APTMS) or 3-aminopropyl triethosysilane (APTES)) have
most frequently been employed as coupling agents.
Losic and co-workers441 have recently reported a new
approach for controlling the surface architecture of porous
AAO membranes, which generates layered, silane-based surface
chemistries and yields distinctly dierent functionalities on the
pore openings and the internal pore surface. The method was
based on the remarkable stability of the silanized surface of
AAO during anodization of aluminum. After a short-term rst
anodization of aluminum and removal of most of the resulting
porous oxide, the surface of the remaining thin oxide layer was
derivatized with ATPES. The resulting silanized sample was
subsequently reanodized for a longer period of time. The
surfaces of the pore wall formed by the second-step anodization
were functionalized with hydrophobic pentauorophenyldimethylchlorosilane (PFPES) with distinctly dierent properties
from the APTES present in the pore opening. By extending the
work, they have also demonstrated that dierent functionalities
and wettabilities can be imparted to the pore channels of AAO.
Multilayered surface modications of porous AAOs were
achieved by repeating anodization and subsequent silanization
of the pore wall oxide with dierent silane molecules, such as
APTES, PFPTES, and N-triethoxysilylpropyl-O-poly(ethylene
oxide) urethane (PEGS) (Figure 56).442 The thickness of
individual functional layers could be conveniently varied by
controlling the anodization time. Deliberate combination of
silane molecules with dierent chemical properties allowed
selective membrane transports of small molecular compounds.
Decoration of inorganic nanoparticles on the pore wall
surface of AAO often imparts new properties to porous AAO.
This decoration can be facilitated by surface modication of the
pore wall surface with an organic monolayer or polymer,443447
but can also be achieved directly on the AAO.375,448,449 Ko and
Tsukruk443 immobilized AuNPs on pore walls modied with
poly(diallydimethylammonium chloride) (PDDA) polyelectrolyte by passing cetyltrimethylammonium bromide (CTAB)stabilized AuNP solution through the AAO membrane. On an
7534

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 57. (Left) Derivatization of PHEMA with NTA-Ni2+ prior to protein adsorption. (Right) Binding of his-tagged protein to an NTA-Ni2+
derivatized PHEMA brush inside pores of AAO. Adapted with permission from ref 456. Copyright 2007 The American Chemical Society.

Figure 58. Schematic diagram showing (a) the formation of thin polymer lms using layer-by-layer (LbL) deposition of polyelectrolytes, (b) a
nanoltration membrane prepared by the deposition of a multilayer polyelectrolyte lm on AAO, (c) a membrane for protein purication created by
growing polymer brushes within pores of AAO, and (d) a catalytic membrane prepared by LbL deposition of polyelectrolytes and charged metal
nanoparticles. Adapted with permission from ref 462. Copyright 2008 The American Chemical Society.

counterparts.143 Polymers grafted on AAOs can be also


functionalized with biologically important molecules, such as
proteins or DNAs. Recently, Li et al.454 prepared thermoresponsive gating membranes with tuneable length and density

Polymer brushes can also be grafted on the pore wall surfaces


of porous AAOs. Polymer-grafted porous AAOs exhibit an
improved binding capacity, stability, selectivity, biocompatibility, and lubricating characteristics as compared to bare
7535

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

by activating with GPTMS. The pH-valve eect was observed


at pH between 3 and 5. Electrochemical impedance spectroscopy (EIS) showed that the membrane resistance increased
from 4.3 105 cm2 at pH 2 to 1.3 106 cm2 at pH 6.
The layer-by-layer (LbL) deposition method, which involves
the alternate adsorption of dierently charged polyelectrolytes,
can be applied to the substrate surfaces that allow the
adsorption of an initial polymer layer (Figure 58a).462466
The resulting multilayer lms can be further functionalized with
nanoparticles or biomolecules. The most used polyelectrolytes
are commercially available ones, such as poly(ethylenimine)
(PEI), poly(allylamine hydrochloride) (PAH), poly(diallydimethylammonium chloride) (PDADMAC), poly(styrenesulfonate) (PSS), poly(vinylsulfate) (PVS), and poly(acrylic acid) (PAA). LbL deposition of polyelectrolyte
multilayer lms on the surfaces of porous AAO membranes
was rst demonstrated by the Bruening group (Figure 58b
d).467 LbL deposition of PAH and Cu2+-complexed PAA on the
surfaces of AAO membranes and subsequent controlled
removal of Cu2+ and deprotonation allowed control of the
xed charge density within PAA/PAH multilayer lms (Figure
58b). The resulting Cu2+-templated PAA/PAH-AAO memtransport
branes exhibited a 4-fold increase in Cl/SO2
4
selectivity in comparison to Cu2+-free PAA/PAH-AAO
membranes deposited under similar condition.467 Cross-linking
of Cu2+-templated PAA/PAH-AAO membranes further increased Cl/SO2
4 selectivity as high as 610. It was suggested
that this selectivity might be due to both Donnan exclusion and
diusivity dierences among ions. Size-selective transport of
neutral solute through PSS/PAH has also been reported by the
same group.468 The number of bilayers combined with dierent
compositions of multilayered lms turned out to signicantly
inuence the rejection, ux, and selectivity of charged solutes
through the membranes.468470 In addition, deposition
conditions such as pH, ionic strength of the polyelectrolyte
solutions, and the charge of the outermost polyelectrolyte layer
were also found to inuence the molecular separation of
cations.470 Polyelectrolyte lms composed of PSS/PAH or
PSS/PDADMAC+PSS/PAH lms were reported to be eective
in the selective removal of Mg2+ from aqueous solutions
containing NaCl and MgCl2 salts (i.e., water softening), while
pure PSS/PDADMAC lms and commercial NF270 membranes showed only relatively low rejections of Mg2+.470 For
PAH-terminated multilayer lms, Mg2+ rejection and Na+/Mg2+
selectivity increased with the increasing charge near the surface
of the polyelectrolyte multilayer lms, by increasing the ionic
strength of the PAH deposition solution. LbL lm-based
catalytic AAO membranes were also demonstrated by
Bruenings group.444 By utilizing the charge on the LbL lms,
citrate-stabilized AuNPs were immobilized within the pores of
AAO. Au-immobilized LbL lms within the pores of AAO
membranes exhibited a similar rate constant for Au-catalyzed 4nitrophenol reduction in solution and in membranes (Figure
58d). Covalent immobilization of antibodies on LbL lms (i.e.,
[PAA/PAH]3PAH) in the pores of AAO membranes has also
been realized via carbodiimide coupling between COOH
groups of PAA amine groups of antibodies.471 Detection limits
in the analysis of Cy5-labeled IgG were reported to be 0.02 ng
mL1 because of the high surface area of the porous AAO.471 In
addition, resistance against nonspecic protein adsorption for
these LbL lm-AAO composites has also been reported.
Combination of the LbL technique and the AAO templating
approach has been demonstrated its versatility in fabricating

of poly(N-isopropylacrylamide) (PNIPAM). The authors


reacted an APTMS-functionalized AAO membrane with the
initiator 2-bromoisobutyryl bromide (BBIB), and then the
resulting membrane with Br groups was reacted with Nisopropylacrylamide (NIPAM) monomers during atom-transfer
radical polymerization (ATRP) to yield PNIPAM-grafted AAO
membrane (PNIPAM-g-AAO). The density of PNIPAM
grafted on the pore wall surfaces could be controlled by
changing the number density of Br groups (i.e., ATRP
initiator). The themo-responsive characteristics of the
PNIPAM-g-AAO membranes were investigated by tracking
the diusional permeation of vitamin B12 at temperatures
below and above the lower critical solution temperature
(LCST). The results indicated that thermo-responsive
characteristics are heavily aected by both the length and the
density of grafted PNIPAM chains in the pores of AAO, and
the eect of the length of grafted PNIPAM chains is more
signicant than that of the density. Purications of proteins
based on reusable metal anity membranes have been
demonstrated by the Bruening group. The use of ATRP to
grow poly(2-hydroxyethyl methacrylate) (PHEMA) brushes in
the pores of AAO, followed by functionalization of the PHEMA
with nitrilotriacetate-M 2+ (NTA-M2+; M = Cu or Ni)
complexes, yielded AAO membranes that adsorbed proteins
via coordination of M2+ to his-tagged proteins (Figures 57,
58c).455,456 In another study by the same group,457 two types of
ultrathin (50 nm) polymer brushes, linear PHEMA and crosslinked poly(ethylene glycol dimethacrylate) (PEGDMA), on
the pore wall surfaces of AAOs were synthesized using the
ATRP method. Gas permeation studies showed that the
PEGDMA-AAO membrane had a CO2/CH4 selectivity of 20
and an O2/N2 selectivity of 2, whereas uncross-linked
PHEMA-AAO membrane showed very low selectivity.
However, the CO2/CH4 selectivity of PHEMA improved to
8 after esterication of OH groups of PHEMA with
pentadecauorooctanoyl chloride. Further derivatization of
PHEMA-grafted AAO membranes with octyl, hexadecyl, or
pentadecauorooctyl side chains made the membranes hydrophobic, allowing selective removal of volatile organic
compounds (VOCs) from water via pervaporation.458 In
another study by Bruenings group,459 poly(ethylene glycol)
(PEG)-grafted AAO (PEG-g-AAO) membranes were prepared
through ATRP of poly(ethylene glycol methyl ether methacrylate) with dierent lengths of PEG chain from initiatormodied porous AAO. The resulting PEG-g-AAO membranes
contained a mixture of short and long PEG side chains, in
which the shorter PEG chain (89 ethylene oxide units)
prevented crystallization, while the longer side chains (2324
ethylene oxide units) allowed the membranes to maintain a
CO2/H2 selectivity of 12 at room temperature.459 Shi et al.460
prepared metal anity membranes with uniform diameter and
distribution of pores for protein separation and purication.
Chitosan(CS)-AAO membranes were prepared by activating
the hydroxylated AAO with 3-glycidoxypropyl trimethoxysilane
(GPTMS), followed by CS grafting.460 Cu2+-attached anity
membranes were then obtained by immobilizing Cu2+ on the
CS-AAO, and were utilized for the separation and purication
of hemoglobin from red cell lysate. Song et al.461 have reported
preparation of pH-responsive poly(acrylic acid)-grafted porous
AAO membranes. Silica-AAO composite membrane was rst
prepared by depositing silica onto the AAO membrane through
the solgel method. The pH-responsive poly(arcrylic acid)
(PAA) was grafted onto the silica-AAO composite membranes
7536

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 59. (a) Schematic illustration of a lipid bilayer formed by 1,2-diphytanoyl-sn-glycero-3-phosphocholin (DPhPC) onto a self-assembled 1,2diphytano-dipalmitoyl-sn-glycero-3-phosphothioethanol (DPPTE) monolayer chemisorbed on a gold-coated surface of a porous AAO substrate. (b)
Impedance analysis of a lipid bilayer bathed in 0.1 M Na2SO4 recorded in a frequency range of = 101106 Hz before () and after ()
gramicidin addition. The membrane resistance was dropped by >3 106 from Rm = 8.73 106 to Rm = 5.45 106 . The solid line is the result
of a tting routine using the equivalent circuit shown in inset. Adapted with permission from ref 496. Copyright 2004 The Biophysical Society.

Figure 60. Separation of attoliter-sized compartments employing pore spanning lipid bilayers. (A) composite z-stack image (total z-distance: 15 m)
of pore-spanning lipid bilayer preventing avidin entrance into the underlying pores (black area). Main image: Lipid membrane is located at the
interface between the AAO and the bulk solution, showing both red uorescence from the membrane and green avidin uorescence in the overlayer.
The total image size is 67 m 67 m. Top and right side images are line proles along the z-direction, manifesting that the membrane prevents the
uorescent labeled proteins from entering the pores. (B) A z-stack image of the membrane encapsulating pyranine dye molecules. The image size is
34 m 34 m. Adapted with permission from ref 500. Copyright 2011 The American Chemical Society.

exhibit long-term stability. However, the close proximity


(typically 0.22 nm) of a lipid bilayer to the surface of solid
support limits lateral lipid mobility, insertions of large
transmembrane proteins to the lipid bilayer, or the generation
of electrochemical gradient across lipid bilayer membranes,
which is required for selective transport of specic ions, for
instance, in ion pumps and ligand- or voltage-gated ion
channels.486 Several approaches to decouple the lipid bilayer
membrane from its underlying support have been developed to
increase the lateral mobility. These include utilization of lipids
with long hydrophilic spacers,487 water or polymer cushions
between lipid membrane and solid support,488491 and surface
patterns with dierent thiol-components.492 On the other hand,
free-standing bilayer lipid membranes (also known as black

composite tubular nanostructures with various functional


properties. Non-electrostatic interactions, such as hydrogen or
covalent bonding and hybridization, have been utilized to
assemble quantum dots (QDs), nanoparticles, polymers, and
biologically relevant molecules or materials into functional
nanotubes.445,462,472481
Lipid bilayers are novel mimics of biological cell
membranes.143 Free-standing bilayer lipid membranes reconstituted with ion-channel proteins provide an excellent system
not only for drug screening,482,483 but also for designing highly
sensitive biosensors.484,485 Conventional solid supported
membranes (SSMs) methods utilize gold-thiol, silanization of
OH-groups, and electrostatic interactions to immobilize lipid
bilayers on solid supports.486 The immobilized lipid bilayers
7537

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

template into a solution containing the desired monomers and


the polymerization initiator or the electropolymerization within
the cylindrical nanopores.338,406,506509 Electropolymerization
has been utilized to synthesize conducting polymer nanotubes
within the pores of AAO or track-etched polycarbonate
membranes. Electrically insulating polymer nanotubes have
also been prepared chemically. For example, polyacrylonitril
(PAN) nanotubes with controlled inner diameters were
synthesized by dipping a porous template into a solution
containing acrylonitrile and a polymerization reagent, in which
polymerization time determined the inner diameter of
nanotubes.510
With an alternative approach, Steinhart et al.511 developed an
arguably simple, yet highly versatile technique for the
fabrication of functional polymer nanotubes. The method is
based on the spontaneous spreading (or wetting) of a polymer
melt or its solution on a surface with high surface energy to
form the so-called precursor lm.512514 Surface wetting
behavior of a droplet of a liquid is described by the spreading
parameter S:513

lipid membranes, BLMs) have been formed across micrometersized apertures in an attempt to eliminate the steric congestion
in SSMs.485 Yet the fragility of the bilayer lipid membranes on
micrometer-sized apertures often prevents their practical
applications. The mechanical stability of BLMs can be increased
by reducing the aperture size.493,494 However, the increased
stability comes at the expense of reduced eective membrane
area to which proteins can be incorporated.
Pioneering works by Steinem et al.486 have led to the
development of methods for suspending lipid bilayers on the
pores of porous AAO, bridging the technical gap between
conventional SSMs and BLMs. First, the authors selectively
functionalized the top surface of porous AAO (Dp = 60 10
nm) by chemisorbing 3-mercaptopropionic acid (MPA) onto
10-nm-thick Au layer that was pre-deposited onto the AAO.
Next, the MPA-modied surface was negatively charged by
treating with 10 mM tris(hydroxymethyl)aminomethane)
solution (pH 8.0). Finally, a pore-spanning lipid bilayer was
obtained by adsorbing positively charged vesicles of N,Ndimethyl-N,N-dioctadecylammonium bromide (DODAB) onto
the negatively charged MPA monolayer. Pores of AAO spanned
by bilayer lipid membranes are tunable in a nanometer range.
Moreover, because the pore channels of AAO are separated by
a semipermeable lipid membrane from the outer solution, they
can serve as nanocontainers for bio-reaction and for establishment of electrochemical gradients.495 Extending their previous
works, Steinem et al.496499 showed that pore-spanning lipid
bilayers on porous AAO substrates are electrically insulating,
and also demonstrated that the ion-channel proteins reconstituted in bilayer membranes are fully functional (Figure 59).
More recently, the same group of authors employed porespanning lipid bilayers formed by spreading a giant unilamellar
vesicle (GUV) on porous AAO substrates to create attolitersized compartments, which were used not only for entrapment
but also for exclusion of materials. By taking advantage of the
optically transparent nature of porous AAO substrates,
uorescent molecules inside the compartments (i.e., AAO
pores) as well as the lipid membranes on top of the pores could
be visualized by using confocal laser scanning uorescence
microscopy (Figure 60).500 This technique will become
powerful in biosensing applications, if it is combined with the
recently developed label-free nanoporous optical waveguide
sensor by Hotta et al., in which changes in reection spectra of
the AAO/Al multilayer lms were measured in the
Kretschmann conguration, similarly to the conventional
surface plasmon resonance (SPR) sensor.501
Smirmov and Poulektov502 demonstrated that phospholipid
bilayers can self-assemble to form nanotubes within the pores
of AAO membrane by exposing one side of membrane to an
aqueous dispersion of phospholipid. The formation of lipid
nanotubes was been conrmed by spin-labeling electron
paramagnetic resonance (EPR),502 solid-state NMR spectroscopy,503 and uorescence microscopy.504 In another study,
Deme and Marchal505 prepared polymer-cushioned lipid
bilayers within the pores of AAO. Their preparation approach
consisted of direct fusion of lipid vesicles into the poly(ethylene
glycol) (PEG)-derivatized pores of AAO.

S = sg sl lg

(75)

where sg, sl, and lg, respectively, are the solidgas, solid
liquid, and liquidgas interfacial tensions. If S is negative, a
liquid drop on the substrate adopts an equilibrium shape
corresponding to a nite contact angle e dened by Youngs
condition cos e = (sl sg)/lg. If S is positive, spontaneous
spreading of the liquid occurs because the adhesion forces
between the liquid and the solid substrate dominate the
cohesive forces within the wetting liquid, and the equilibrium
situation corresponds to complete coverage of the substrate by
the precursor lm with a thickness ranging from several
angstroms to tens of nanometers (i.e., e = 0).513,515
Accordingly, direct contact of porous inorganic templates
(i.e., high-surface-energy materials) with polymer melts or
solutions (low-surface-energy materials) results in the immediate wetting of pore wall surfaces with thin precursor lm
(Figure 61a). Steinhart et al.515 suggested that pore wall wetting
is kinetically stable (but thermodynamically unstable), when the
strong adhesive forces between the liquid and solid are
neutralized upon complete surface wetting, due to the nite
surface area of an individual pore of the template. To attain a
thermodynamically stable state (i.e., complete pore lling), the
cohesive force should dominate the viscous forces of the
wetting liquid. Thus, pore wall wetting and complete lling
occur on dierent time scales; the latter, if it would happen at
all, will require several months or even several years of
time.515,516 Therefore, polymer nanotubes can be obtained by
solidifying the wetting liquid before complete pore lling
(Figure 61bd). According to Steinhart et al.516 complete pore
lling occurs when the pore diameter of the porous template is
smaller than the thickness of wetting layer (1030 nm). The
wetting layer thickness may depend on the nature of polymer
and the surface state of wetting solid. It was reported that
injection of polystyrene (PS) melt into the pores of AAO (Dp =
40 nm) results in arrays of PS nanowires, rather than
nanotubes.517
On the basis of the experimental observations of
spontaneous surface spreading of highly viscous melts of
polymers with ultrahigh molecular weights (typically, 106 and
107 g mol1), Steinhart et al.516 suggested that wetting of
polymer melt or solution would occur through surface
diusion. Therefore, individual polymer chains should align

9.5. Template Wetting

Pores of AAO template can be utilized as nanoreactors for the


synthesis of a variety of nanotubular materials. The templatebased chemical syntheses of polymer nanotubes, pioneered by
C. R. Martin, typically involve either the immersion of a porous
7538

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

mixture solutions containing Pd(OAc)2 and poly(D,L-lactide)


(PDLLA) at 350 C, in which the PDLLA acts as a reducing
agent for PdII at 160 C and pyrolyzes at a higher 350 C.521
The method was extended to prepare Pt nanotubes from
Pt(acac)2/PDLL composite nanotubes.522
9.6. Mask Techniques

Two-dimensional (2D) periodic arrays of nanostructures (e.g.,


semiconductors, metal nanoparticles, or complex oxides arrays)
have attracted tremendous research attention because of their
many unique properties and promising applications in
electronic/optoelectronics, bio-sensing, and high-density data
storage media. In 1996, Masuda and Satoh6 reported for the
rst time the fabrication of arrays of gold nanodots on silicon
substrate by using ultrathin AAO as a stencil mask for
depositing gold. Since then, thin AAO-based surface patterning
has been popularized. The interested reader is referred to
review articles on surface nanopatterning using ultrathin AAO
masks given in refs 523 and 524. AAO-based surface patterning
methods are highly attractive in terms of throughput,
accessibility, cost, and compatibility with high-temperature
nanofabrication processes, in comparison to other surface
nanopatterning techniques (for example, electron-beam direct
writing (EBDW), focused ion beam (FIB) milling, nanoimprinting lithography (NIL), and scanning probe microscope
(SPM)-based writing).
Ultrathin AAO membranes can be transferred onto desired
substrates by supporting AAO membranes with appropriate
polymers (typically, poly(methyl methacrylate) (PMMA) or
PS).6,523 In this method, a thin polymer layer is coated on one
face of the ultrathin porous AAO lm. Afterward, the aluminum
substrate and the barrier layer at the bottom of the pores are
removed. Next, the resulting AAO/polymer composite lm is
transferred onto a substrate of choice. Finally, the polymer is
removed by immersing the sample into an appropriate solvent
or by performing oxygen plasma treatment. However, this
method often suers from the diculty in transferring a largescale ultrathin AAO membrane onto a substrate without
introducing structural damages caused by folding, cracking, or
ripping of the ultrathin AAO. Recently, Meng et al.525
demonstrated the successful transfer of wafer-scale ultrathin
AAO membranes (thickness = 200 nm) onto hydrophilized
substrates. The method involved selective etching of the
underlying aluminum of a two-step anodized AAO lm to leave
a rigid aluminum frame surrounding the AAO lm, followed by
transfer of the resulting aluminum-frame-supported AAO lm
onto a desired substrate (Figure 62). The aluminum frame
stabilized the ultrathin AAO lm during transfer. The
hydrophilic surface of the substrate enabled a conformal
contact between the ultrathin AAO lm and substrate via
dropping water on the substrate. After removal of the
surrounding aluminum frame, the barrier oxide layer could be
removed by conducting argon milling to obtain a through-hole
AAO membrane.
A variety of materials have been fabricated into ordered
arrays of nanostructures on substrates of choice, including
arrays of semiconductor,526535 metal,536539 and oxide
nanodots540546 or arrays of nanoholes,547562 nanorings,563
nanopillars,548 and nanowires.525,564 Figure 63ad schematically shows the general procedures for the fabrication of 2D
extended arrays of (a) nanoholes, (b) nanodots, (c) nanopillars,
and (d) nanowires by using ultrathin AAO membranes as
masks. Ordered arrays of nanoholes have been fabricated on

Figure 61. (a) Wetting of porous templates with polymer melts or


solutions: (i) The uid comes into contact with the template. (ii)
Within a second, the pore walls are covered with a thin lm of the
liquid. By freezing this stage, nanotubes are preserved. (iii) A complete
lling of the pore space was not observed. (be) SEM images of
polymer nanotubes obtained by melt-wetting: (b) polystyrene (PS)
nanotubes embedded in porous AAO, (c) PS nanotubes after removal
of AAO template, (d) array of polytetrauoroethylene (PTFE)
nanobues, (e) poly(methyl methacrylate) (PMMA) nanotubes. Panel
(a) was reproduced with permission from ref 515. Copyright 2003,
Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. Panels (be)
reprinted with permission from ref 511. Copyright 2002 AAAS.

on the wetting surface. The cylindrical geometry of pores in


AAO imposes a spatial connement with respect to the
movement of a polymer chain, in which the length of the pore
channel can be regarded as innite, while the circumference and
thickness of the wetting polymer are nite with a curvature.
Poly(vinylidene diuoride) (PVDF) nanotubes prepared by
melt-wetting of porous AAO (Dp = 400 nm) exhibited a
curvature-directed crystallization to form -PVDF phase, in
which the crystallographic b-axis of -PVDF is parallel to the
long axis of the nanotubes, the only direction to minimize
curvature constriction.518 The eects of geometric connement
and the pore wall on the mesophase formation of liquid
crystalline materials have also been reported.519,520
Template wetting is versatile in terms of the facile variation
of tube wall materials. Almost all polymer solutions with a low
surface energy can be used to synthesize functional or
composite nanotubes. Preparations of polymer nanotubes
with high technical importance (e.g., polyether ether ketone
(PEEK) or polytetrauoroethylene (PTFE)), which are very
dicult or practically impossible to process by conventional
methods (e.g., extrusion or injection molding of polymer), have
been demonstrated by wetting of polymer melt impregnation
into porous templates.516 Multicomponent composite nanotubes have also been utilized to prepare inorganic nanotubes. In
this case, the polymers are mixed with inorganic precursors and
act as a carrier in the wetting process. Chemical transformation
of the inorganic precursors within the composite nanotube
walls yields inorganic nanotubes. For example, Pd nanotubes
were prepared by the wetting of porous AAO templates with
7539

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

also been used either as etching masks in dry etching processes


to generate ordered arrays of nanopillars (Figure 63c)548,553 or
as catalysts in the chemical vapor deposition (CVD) growth of
arrays of carbon nanotubes (CNTs),565 the MBE growth of
GaAs nanowires,566 and the vaporliquidsolid (VLS) growth
of Si,564 MgO,525 and ZnO567 nanowires (Figure 63d).
Thick porous AAO membranes have been utilized as
replication masters for preparing metallic stencil masks. Lee
et al.568 demonstrated that Au nanotube membranes replicated
from porous AAO can be utilized as deposition masks for
fabricating 2D periodic arrays of catalyst nanodots for sitespecic VLS growth of semiconductor nanowires (see Figure
64)569,570 or ordered arrays of ferroelectric nanostructures.568,571,572 Their fabrication of Au nanotube membranes
involved the sputter deposition of thin Au lm on the pore wall
surfaces of top part of AAO membrane, as well as on the top
membrane surface, followed by electrochemical deposition of
Au, to further thicken the metal layer. Subsequent removal of
the AAO membrane by oating the sample on 30 wt % H3PO4
solution resulted in a free-standing Au nanotube membrane.
Finally, Au nanotube membrane was transferred onto desired
substrate. The replicated Au nanotube membranes exhibited a
high degree of exibility, allowing conformal contact even with
curved surfaces. In the process, the inner diameter of the Au
nanotubes can be easily tuned by controlling the electrodeposition time. Therefore, nanodots with sizes smaller than
the pore diameter of the starting AAO could be fabricated.568
With a similar approach, replications of ultrathin (thickness
20 nm) metal meshes with arrays of ordered nanoholes have
also been reported. The replicated metal meshes have been
utilized not only as stencil masks for the patterning of polymer
substrates,573 but also as catalysts for the metal-assisted
chemical etching of silicon wafers to prepare arrays of silicon
nanowires with controlled size, surface morphology, and
crystallographic orientations (Figure 65).573578

Figure 62. (a) Schematic procedure for the fabrication of ultrathin


AAO membrane on a substrate. Photographs of 200-nm-thick AAO
membrane (b) supported by Al frame and (c) transferred onto a 3-in.
silicon wafer. Reproduced with permission from ref 525. Copyright
2012 The Royal Society of Chemistry.

various substrates by conventional dry processes, such as


plasma etching, reactive ion etching (RIE), and ion milling
(Figure 63a). Because of high nanohole regularity and density,
the resulting nanostructures exhibited many interesting properties that can be exploited in the development of eective
antireection structures,555,556 high performance electrodes or
capacitors,559,560 waveguides and photonic crystals,561 and
optical or lasing devices.549,562 On the other hand, arrays of
nanodots can simply be obtained by depositing a desired
material through an ultrathin AAO membrane (Figure 63b), in
which a 2D array pattern, and size and interdistance of the
nanodots are dened by those of the ultrathin AAO mask. Most
physical vapor deposition (PVD) methods can be employed to
create surface nanostructures, including e-beam evaporation,6,533,534,539,548 thermal evaporation,527,528,540 sputter deposition,535 pulsed laser deposition (PLD),541546 and molecularbeam epitaxy (MBE).526,530532 The generated nanodot arrays
have been used as model systems for investigating the
structureproperty relations of materials (e.g., magnetism,
plasmonics, photoluminescence, cathode luminescence, ferroelectricity) at nanometer-length scales. Arrays of nanodots have

9.7. Chemical Vapor Deposition (CVD)

A major challenge facing the chemical vapor deposition (CVD)


of materials into the pores of AAO involves achieving a uniform
deposition of materials on the entire surface of the pore walls. A
fast rate of deposition may cause blockage of pores before the
precursor chemical vapor traverses pores of high aspect

Figure 63. Schematic procedure for the fabrication of 2D extended arrays of (a) nanoholes, (b) nanodots, (c) nanopillars, and (d) nanowires by
using ultrathin AAO membranes as mask.
7540

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

emitters,585 electrodes for supercapacitors,588 and platforms for


intracellular delivery.589
Porous AAO has also been utilized as a structure-guiding
template in CVD growth of nanowires. It has been well
established for the VLS growth of SiNWs that the nanowires
grow preferentially along the 111, 112, or 110 directions
depending on the diameter.590,591 At the same time, vertically
aligned [100] SiNWs on Si(100) substrate are highly desired
for most practical applications in current complementary-metaloxide-semiconductor (CMOS) technology. Moreover, tight
control over the diameter and spacing of SiNWs in conventional VLS growth process is dicult to achieve without
employing electron beam lithography to dene the size and
location of the catalyst nanoparticles. Shimizu et al.592
demonstrated homoepitaxial growth of SiNWs on Si(100)
substrate by utilizing porous AAO as an orientation guiding
template. The method entailed the combination of electroless
deposition of Au catalyst into AAO nanopores and VLS growth
of Si (Figure 67). This approach has recently been extended by
Gorisse et al.593 to prepare extended arrays of vertically aligned
[100] SiNWs with tightly controlled diameters and separations.
Fabrications of other semiconductor nanostructures by CVD
techniques have also been reported. Cheng et al.594 obtained
single crystalline GaN NWs in porous AAO templates by
reacting Ga2O3 vapor with NH3 gas at 1000 C. The authors
suggested that the defects in the pore walls act as GaN
nucleation centers under a saturated reaction atmosphere, and
the multiplication of screw dislocations is responsible for the
unidirectional growth and formation of GaN NWs. Later, this
method was modied by Zhang et al.,595 who employed indium
nanoparticles electrodeposited in AAO pores as catalyst, to
prepare ordered arrays of GaN nanowires. VLS growth of
hexagonal wurtzite GaN nanowires with very miscible droplets
of InGaN was also suggested. Jung et al.596 reported
fabrication of GaN nanotubes by metallorganic chemical vapor
deposition (MOCVD) using trimethylgallium (Ga(CH3)3) and
ammonia (NH3) as precursors. Because of the random
nucleation of GaN nanoparticles on the surfaces of the inner
pore wall of AAO, the fabricated GaN nanotubes consisted of
GaN nanoparticles with sizes of 1530 nm. In another study, Li
et al.597 reported the controlled growth of Ge nanowires or
nanotubes inside a porous AAO template by low-temperature
CVD assisted by an electrodeposited metal nanorod catalyst.
Depending on the types of catalytic metals (Au, Ni, Cu, Co)
and germane (GeH4) concentration during CVD, Ge nanowires or nanotubes could be synthesized at 310370 C. The
authors also demonstrated the synthesis of multi-segmented
nanowire junctions of Au1xGex and Ge inside the nanochannels of porous AAO template by Au nanorod-catalyzed
VLS growth process from GeH4 precursor.598 Polycrystalline
CdS,599 In2O3,600 and ZnS601 nanotube arrays have also been
synthesized within the pores of AAO membranes by lowtemperature CVD methods using, respectively, cadmium
bis(diethyldithiocarbamate) [Cd(S2CNEt2)2] at 400 C,
indium acetlacetonate [In(acac)3] at 350 C, and zinc
bis(diethyldithiocarbamate) [Zn(S2(NEt2)2] at 200 C. Metalcatalyzed VLS growths of single crystalline CdS602 or ZnO603
nanowires within the pores of AAO membrane have recently
been reported.

Figure 64. (a) SEM image of Au catalyst nanodots deposited using an


Au nanotube membrane as a shadow mask. (b) Top-view SEM image
of VLS-grown ZnO nanorod arrays using Au catalyst nanodots shown
in (a). Reproduced with permission from ref 569. Copyright 2006
Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

ratio.579,580 The uniform deposition of carbon was rst reported


by Kyotani et al.,581 who performed the thermal decomposition
of propylene at 800 C. Pyrolytic carbon deposition from
propylene resulted in carbon tubes within the pores of AAO, in
which the wall thickness of the carbon tubes was dependent on
the deposition period. The carbon tubes obtained by the
method exhibited low crystalline quality. In another study, Chu
et al.582 employed the pyrolytic carbon deposition method for
the preparation of carbon tubes, in which porous AAO
membrane was exposed to either ethylene or pyrene gas
stream at 900 C for 10 min. They showed that the
temperature needed for CVD of carbonaceous materials can
be lowered to 500 C by decorating pore wall surfaces of AAO
template with Ni-catalyst nanoparticles.582 Apart from lowering
the CVD temperature, it was also found that the metallic
catalysts improve the crystalline quality of the resulting carbon
nanostructures. Multiwalled carbon nanotubes (CNTs) embedded in porous AAO have also been obtained by CVD of a
gaseous precursor (typically, acetylene) in the temperature
range of 550650 C in the presence of electrodeposited Co
catalyst at the bottom of AAO pores (Figure 66).287,583587
The ordered arrays of CNTs have been utilized as eld

9.8. Atomic Layer Deposition (ALD)

Atomic layer deposition (ALD) has drawn much attention as a


versatile methodology for thin lm deposition due to conformal
7541

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 65. Metal-assisted chemical etching of (100)-oriented silicon wafer by using ultrathin metal mesh for the preparation of SiNWs with
controlled axial orientation and morphology. Reproduced with permission from refs 575,576. Copyright 2011 The American Chemical Society.

nature, avoiding CVD-like lm growth. By virtue of such a selfterminating lm growth process, the conformal and uniform
deposition of lm with atomic scale precision is achievable over
a large area.
Ott et al.610 rst performed ALD of Al2O3 on porous AAO
membrane to narrow the pores, and demonstrated the gas
separation capability of the pore-modulated AAO membrane.
In another study, Chen et al.611 reported a strategy for
fabricating a single-molecule (i.e., DNA molecule) sensor
through surface modication of the pore walls of AAO and
precise control of the pore diameter at the single angstrom
scale. Since then, ALD on porous AAO has been intensively
utilized as a facile method not only for surface modication of
oxide nanopores,612 but also for fabricating 1D functional
nanostructures, such as nanowires,613 nanorods,614,615 and
nanotubes.616625
Single-614,616,624628 or multi-layered617,621 nanotubes composed of dierent materials can conveniently be fabricated
through a deliberate choice of precursor molecules during ALD
on the pore wall surfaces of AAO. The ability to precisely
control lm thickness is benecial in tuning the thickness of
tube walls and the gap between each tube. Fabrications of
multi-segmented nanotubes have also been demonstrated
recently through the combination of 3D site-selective ALD
and anodization of aluminum. Bae et al.618 have shown that
cylindrical nanopores of AAO can be continuously anodized
upon coating with thin organic and/or inorganic layers such as
octadecyltrichlorosilane (OTS)-self-assembled monolayers and
ALD-grown TiO2, ZnO, and ZrO2. Nanowires or nanorods can
also be obtained through the so-called super-lling of the
pores of AAO by virtue of the superior capability of ALD
techniques for conformal lm deposition over other conventional thin lm deposition techniques.613,615 3D device
architectures can be realized by the ALD of appropriate
materials within pores of AAO, and the approach may be
benecial to energy storage and energy conversion systems

Figure 66. (a) Schematic procedure for the fabrication of carbon


nanotubes (CNTs). (b) SEM image of the ordered arrays of CNT
fabricated by the method in (a). Reprinted with permission from ref
583. Copyright 1999 AIP Publishing LLC.

and uniform deposition of thin lms on substrates with


complicated 3D morphology.143,604609 In ALD, a thin lm of a
desired material is grown in a layer-by-layer manner by
repeating a unit cycle composed of four consecutive steps: (i)
precursor exposure, (ii) purging (or evacuation) of unreacted
excess gas molecules and reaction byproducts, (iii) reactant
exposure, and (iv) subsequent purging (Figure 68). Because of
the alternating exposure of gaseous precursor and reactant at
the separate steps, the lm growth process is self-limiting in its
7542

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 67. (ad) Schematic cross-sections, showing the growth of SiNWs in porous AAO using Au catalyst: (a) after pre-annealing at 900 C, (b)
after HF etching to remove SiO2, (c) electroless deposition of Au catalyst, and (d) VLS growth of SiNWs. (e) Cross-sectional TEM image of a SiNW
in AAO template along the [011] zone axis. The dashed line shows the interface between the original Si(100) substrate and SiNW. The magnied
TEM image of the area marked with a white rectangle is shown in (f), which demonstrates homoepitaxial growth of SiNW. Reprinted with
permission from ref 592. Copyright 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

have been independently reported by Comstock et al.619 and


Lee et al.635 Comstock et al.619 reported a fabrication strategy
for a nanotubular glucose sensor by using ALD-grown Pt thin
lms within the pores of AAO. Lee et al.635 realized fast
response H2 sensors by transferring TiO2-coated AAO onto a
glass substrate followed by selective removal of the AAO.
Nanotube arrays synthesized by a similar approach have been
used as an ultraviolet (UV) sensor,636 membranes for gas and
molecule separations,612,637,638 catalytic membranes,639,640 and
photocatalysts.641,642

10. CLOSING REMARKS AND OUTLOOK


The anodization of aluminum and the resulting porous anodic
aluminum oxide (AAO) are currently the subject of extensive
research with nearly a thousand papers published annually in
the eld. In this Review, we have presented fundamental
electrochemical processes associated with the porous-type
anodization of aluminum, the recent progress on aluminum
anodization for the fabrication of self-ordered porous AAO, and
applications of porous AAO to templated synthesis of
functional nanostructures for current nanotechnology research.
Over the past near 100 years, many electrochemical aspects
of aluminum anodization have been disclosed. Ionic migrations
within the oxide have been explained in the framework of higheld conduction theory. The type of acid electrolytes and
anodizing potential have direct relevance not only on the
structural parameters (e.g., pore diameter, interpore distance,
and the barrier layer thickness) of the porous AAO, but also on
the incorporation of anionic species and their distribution in
the anodic oxide. Empirical relations between the parameters
dening the geometric structure of porous AAO and the
anodizing potential have also been established. Pore formation

Figure 68. Schematic illustration of a typical atomic layer deposition


(ALD) process of Al2O3 ALD from trimethyl aluminum (TMA) and
water (H2O).

such as Li batteries629631 and solar cells.407,626 Banergee et


al.632 have recently demonstrated fabrication of densely packed
3D nanocapacitors by performing the sequential ALD of a TiN
bottom electrode (BE)/dielectric Al2O3/TiN top electrode
(TE) within the pores of AAO (Figure 69). Because of the large
interfacial area of TiN/Al2O3 inside the open volume of porous
AAO, their 3D nanocapacitors exhibited considerably higher
capacitance as compared to equivalent 2D planar capacitors:
100 F cm2 for 3D nanocapacitors formed in 10-m-thick
AAO.
ALD-assisted modication of AAO surfaces has also been
considered a promising approach to molecule and gas
sensing.619,633,634 Two dierent kinds of sensor congurations
7543

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Figure 69. Arrays of supercapacitors fabricated by sequential ALD of TiN, Al2O3, and TiN within pores of AAO: (a) cross-sectional SEM images of
the capacitors at the pore mouth (upper panel) and at the pore bottom (lower panel) and (b) schematic drawing of a unit cell of the capacitor.
Reprinted with permission from ref 632. Copyright 2009 Macmillan Publishers Ltd.: Nature Nanotechnology.

has long been attributed to the eld-assisted dissolution of


anodic alumina. This dissolution-based pore formation model
appears to be operative for the initial stage of pore formation.
However, for steady-state pore formation, there is a growing
number of results from recent experiments and theoretical
modelings, disproving the eld-assisted oxide dissolution
model. These include, for instance, the direct ejection of Al3+
ions from the metal/oxide interface through the barrier oxide to
the anodizing electrolyte (from 18O isotope studies) and the
viscous ow of oxide materials from the pore base toward the
cell boundary (from W-tracer studies and nite-elemental
analysis). However, more systematic experimental investigations using various electrolytes/tracer elements and theoretical
modelings considering the mobility of tracer cations are
required to fully verify the oxide ow model. Furthermore,
recent studies have revealed that the pore initiation and selforganized formation of porous AAO need to be understood in
terms of mechano-electrochemistry. Studies have indicated that
stresses and their gradients within anodic oxide have profound
implications on ionic migration within the anodic oxide, on
morphological instability associated with pore initiation at the
early stage of anodization, breakdown of growing oxide, as well
as the viscous ow of oxide materials and self-organization of
pores during the steady-state anodic oxidation. Although some
progress in this direction has recently been made, satisfactory
correlation between stresses and all experimental observations
has yet to be achieved. Internal stresses of growing anodic oxide
need to be systematically evaluated in situ under controlled
anodization conditions. In addition, the eect of external
stresses on the anodization kinetics as well as on the selfordering behavior of pores needs to be further investigated.
Studies on these issues would do much to advance our
knowledge on the mechanism governing the self-organized
formation of pores during anodization of aluminum. Thorough
understanding of the mechano-electrochemical processes may
provide a solid foundation for exploring new electrolyte
systems and novel porous architectures as well as for
developing ordered porous structures from other valve-metals,
such as Mg, Zr, Nb, Sn, Hf, Ta, W, Bi, etc.
Aluminum anodizing under properly controlled conditions
produces highly ordered porous AAOs. On the basis of recent
developments of various anodization methods (e.g., mild, hard,

pulse, cyclic, and guided anodization), the diameter, density,


and aspect ratio of pores, and even internal pore structures can
be tightly controlled by appropriate selection of the anodizing
conditions. These capabilities may oer large degrees of
freedom for the templated syntheses of low-dimensional
functional nanostructures, and also in the development of
AAO-based advanced devices, allowing simple and costeective non-lithographic fabrication of extended arrays of
structurally well-dened and identical nanostructures. Indeed,
over the last two decades, we have witnessed fascinating
applications of porous AAO membranes as templates for the
synthesis of various nanowires and nanotubes, as masks for
extended arrays of structurally well-dened surface nanopatterns, and also as platform materials for (bio) molecule
separations, catalysts, drug delivery, photonic, and energy
storage devices. All of these applications have been achieved
through deliberate control over the dimensions of pores and
the thickness of AAO membranes, and also through the
appropriate engineering of surface properties of porous AAO. It
is very likely that evolving experimental techniques for
engineering of the internal pore structures and for programmed
functionalizing of the surface properties of porous AAO will
further expand the application eld. Accordingly, the future
prospects for nanotechnology applications of porous AAO are
very promising.

AUTHOR INFORMATION
Corresponding Author

*E-mail: woolee@kriss.re.kr.
Notes

The authors declare no competing nancial interest.


7544

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

Biographies

the former director of Max-Planck-Institut fur Mikrostrukturphysik in Halle, Germany.

ABBREVIATIONS
AAO
anodic aluminum oxide
U
anodizing potential
j
current density
E
electric eld
E*
threshold electric eld
t+
cation transport number
t
anion transport number
j
current eciency (i.e., oxide formation
eciency)
PBR
PillingBedworth ratio
U
potential drop
tb
barrier layer thickness

hopping attempt frequency

density of mobile charge in C cm3


a
hopping interdistance
W
hopping activation energy at zero eld

parameter describing the asymmetry of the


activation barrier at non-zero eld
z
valence of the mobile ions
F
Faradays constant
m/o
potential drop at the metal/oxide interface
o/e
potential drop at the oxide/electrolyte
interface
RBS
Rutherford backscattering spectrometry
TEM
transmission electron microscopy
AR
anodizing ratio (in nm V1)
P
porosity of porous AAO
i
ionic chemical potential
Ji
ux of ion i
Ci
concentration of ion i
uio
pre-exponential velocity of ion i
a
migration jump distance in the oxide

mean stress

electrical potential
uio
standard chemical potential of ion i
zi
charge number of ion i
i
molar volume of ion i
UB
breakdown potential
e
electrolyte resistivity
CA
anion concentration
EB
electric breakdown
MB
mechanical breakdown
UEB
electric breakdown potential
UMB
mechanical breakdown potential
CB
conduction band
tox
oxide thickness
je
electronic current
x
travel distance of electrons
(E)
the impact ionization coecient at the
electric eld E
i
the threshold energy for impact ionization
r
recombination constant (r < 1)
tox,B
critical oxide thickness at the moment of
breakdown
T
temperature
jt
total current density
j1
oxidation current
j2
current density consumed by the incorporated electrolyte species

Woo Lee is a principal researcher at Korea Research Institute of


Standards and Science (KRISS) and a professor of the Department of
Nano Science, University of Science and Technology (UST), Korea.
He received his Ph.D. from Seoul National University (2003). He
worked with the late Prof. Ulrich Gosele as a postdoctoral research
fellow and later a group leader at Max-Planck-Institut fu r
Mikrostrukturphysik in Halle, Germany, until he joined KRISS in
2008. He brought a renewed attention to the academic research on
pulsed anodization as well as hard anodization of aluminum by
establishing new self-ordering regimes and also by implementing them
for the structural engineering of porous AAO. His research interests
focus on the anodization of aluminum and template-based synthesis of
low-dimensional functional nanostructures for memory and energy
harvesting applications.

Sang-Joon Park did his graduate work in Materials Science &


Engineering at the Pohang University of Science and Technology
(POSTECH) under the co-supervision of Woo Lee and Prof. Sunggi
Baik. He earned his Ph.D. with a main focus on resistive switching of
capacitors with TiO2 active layer, grown by atomic layer deposition
(ALD) in 2014. He then joined Woo Lees group at KRISS as a
postdoctoral researcher. His research interest is atomic layer
deposition (ALD) for functional nanostructures.

ACKNOWLEDGMENTS
Support from KRISS project Anodization Research Laboratory
(KRISS-2013-13011082) and in part from the Future-based
Technology Development Program (Nano Fields) through the
National Research Foundation of Korea (NRF) funded by the
Ministry of Science, ICT, and Future Planning (Grant no.
2010-0029332) is greatly acknowledged. This Review is
dedicated to the memory of the late Professor Ulrich Gosele,
7545

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

ox
M1
M2
x1, x2
y1, y2
K

C
P

A
Qc
Dint
Dp
tw
p
p
MA
HA
PA
CA
MA
ARMA
ARHA
PL
tetch
EXAFS
NMR
XRD
TG
DSC
MS
Vox
Vm
n
m
ox
AAO
kV
hAAO
hAl
kv

ES
0
Y
k

Review

EM
M
L
jlimit
K
PSD
1D
2D
3D
R
w

the ratio j2/j1


oxide density
molecular weight of the oxide
molecular weight of the species incorporated
into the oxide
anion valences
cation valences
unitary rate of anodization for oxide without
electrolyte incorporation
ratio of the equivalent weight of the
incorporated species to that of oxide, that
is, = (M2/x2y2)/(M1/x1y1)
electrolyte concentrations
stress accumulated in the oxide
oxide permittivity
surface tension
anion density at the oxide surface at the
breakdown potential (UB)
total charge
interpore distance
pore diameter
pore wall thickness
pore density
ratio of the pore diameter (Dp) to the
anodization potential (=Dp/U)
mild anodization
hard anodization
pulse anodization
cyclic anodization
proportionality constant between Dint and
mild anodization (MA) potential (UMA)
anodizing ratio for mild anodization (MA)
anodizing ratio for hard anodization (HA)
photoluminescence
pore wall etching time
extended X-ray absorption ne structure
nuclear magnetic resonance
X-ray diraction
thermogravimetry
dierential scanning calorimetry
mass spectrometry
molar volume of oxide grown during the
anodization
molar volume of metal consumed during the
anodization
number of atoms of metal per one formula
of the oxide
density of metal
density of oxide
density of anodic aluminum oxide (AAO)
thickness ratio of the AAO formed during
anodization to the aluminum consumed
during anodization
vertical height of the AAO
vertical height of the aluminum consumed
during anodization
volume expansion factor
compressive stress normal to the oxide
surface
electrostatic stress
vacuum permittivity
Yongs modulus
radius of curvature

jtot
jox
jloss
jAl
jdec

M
UMA
UHA
U
MA-AAO
HA-AAO
DMA
int
DMA
int
DMA
p
tMA
b
tHA
b
PMA
PHA
HA
HA
Q
Rb
TA
ITO
CNT
EDS
EBL
LIL
FIB
SPM
NSL
PDMS
NIL
RD-WETS
SFIL
ECD
ALD
CVD
PVD
RIE
MTJ
7546

elastic modulus of metal strip


Poisson ratio of metal strip
metal strip length
limiting current
the substrate curvature in km1
power spectral density
one dimension
two dimension
three dimension
radius curvature of the metal/oxide interface
angle subtended from the center of curvature
to the pore bases
total ionic current density
current density linked with the number of
oxygen in porous AAO
current density due to the loss of Al3+ ions
into the electrolyte
current density caused by direct ejection of
Al3+ ions
current density stemming from the outward
movement of Al 3+ ions produced by
decomposition inside the barrier oxide
toward the electrolyte
critical wavelength
surface energy
biaxial modulus of solid
mild anodization potential
hard anodization potential
UHA UMA
AAO formed by mild anodization (MA)
AAO formed by hard anodization (HA)
interpore distance of MA-AAO
interpore distance of HA-AAO
pore diameter of MA-AAO
barrier layer thickness of MA-AAO
barrier layer thickness of HA-AAO
porosity of MA-AAO
porosity of HA-AAO
proportionality constant between Dint and
hard anodization (HA) potential (U)
pulse duration for hard anodization (HA)
Joules heat
resistance of the barrier layer
transition between mild and hard anodization
indium tin oxide
carbon nanotube
energy dispersive X-ray spectroscopy
electron beam lithography
laser interference lithography
focused-ion-beam
scanning probe microscope
nanosphere lithography
poly-dimethylsiloxane
nanoimprint lithography
reaction-diusion wet-stamping
step and ash imprint lithography
electrochemical deposition
atomic layer deposition
chemical vapor deposition
physical vapor deposition
reactive ion etching
molecular transport junction
dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

S
sg
sl
lg
e
PVDF
PEEK
PTFE
PDLLA
EBDW
PMMA
PS
PLD
MBE
VLS
SiNW
MOCVD
TMA
OTS
BE
TE
UV

OWL
Ppy
GMR
NP
SAPT

on-wire lithography
polypyrole
giant magnetoresistance
nanoparticle
2-(succinic anhydride)propyl trimethoxysilane
CTAB
cetyltrimethylammonium bromide
TEOS
tetraethoxylsilane
Pluronic P123
poly(ethylene oxide) 20 -poly(propylene
oxide)70-poly(ethylene oxide)20
Brij 56
decaethylene glycol hexadecyl ether
S-helix
single helix
D-helix
double helix
APTMS
3-aminoproyl trimethoxysilane
APTES
3-aminopropyl triethosysilane
PFPES
pentauorophenyldimethylchlorosilane
PFPTES
pentauorophenyldimethylpropylchlorosilane
PEGS
N-triethoxysilylpropyl-O-poly(ethylene
oxide) urethane
PDDA
poly(diallydimethylammonium chloride)
NPNT
nanoparticle-decorate nanotubes
CYP2E1
cytochrome P450 2E1
G6PD
glucose-6-phosphate dehydrogenase
NADPH
nicotinamide adenine dinucleotide phosphate
NADP+
2-oxoaldehyde dehydrogenas
RGDC
arginine-glycine-aspartic acid-cysteine
PNIPAM
poly(N-isopropylacrylamide)
BBIB
2-bromoisobutyryl bromide
NIPAM
N-isopropylacrylamide
ATRP
atom-transfer radical polymerization
PNIPAM-g-AAO PNIPAM-grafted AAO membrane
LCST
lower critical solution temperature
PHEMA
poly(2-hydroxyethyl methacrylate)
NTA-M2+
nitrilotriacetate-M2+, where M is Cu or Ni
PEGDMA
poly(ethylene glycol dimetrhacrylate)
VOC
volatile organic compound
PEG
poly(ethylene glycol)
PEG-g-AAO
poly(ethylene glycol) (PEG) grafted AAO
CS
chitosan
GPTMS
3-glycidoxypropyl trimethoxysilane
PAA
poly(arcrylic acid)
EIS
electrochemical impedance spectroscopy
LbL
layer-by-layer
PEI
poly(ethylenimine)
PAH
poly(allylamine hydrochloride)
PDADMAC
poly(diallydimethylammonium chloride)
PSS
poly(styrenesulfonate)
PVS
poly(vinylsulfate)
QD
quantum dot
SSM
solid supported membranes
BLM
black lipid membranes
MPA
3-mercaptopropionic acid
DODAB
N,N-dimethyl-N,N dioctadecylammonium
bromide
DPhPC
1,2-diphytanoyl-sn-glycero-3-phosphocholin
DPPTE
1,2-diphytano-dipalmitoyl-sn-glycero-s-phosphothioethanol
GUV
giant unilamellar vesicle
SPR
surface plasmon resonance
EPR
electron paramagnetic resonance
PAN
polyacrylonitril

spreading parameter
solidgas interfacial tensions
solidliquid interfacial tensions
liquidgas interfacial tensions
nite contact angle
poly(vinylidene diuoride)
polyether ether ketone
polytetrauoroethylene
poly(D,L-lactide)
electron-beam direct writing
poly(methyl methacrylate)
polystyrene
pulsed laser deposition
molecular-beam epitaxy
vaporliquidsolid
Si nanowire
metallorganic chemical vapor deposition
trimethyl aluminum
octadecyltrichlorosilane
bottom electrode
top electrode
ultraviolet

REFERENCES
(1) Buff, H. Liebigs Ann. Chem. 1857, 3, 265.
(2) Bengough, G. D.; Stuart, J. M. Improved process of protecting
surfaces of aluminium of aluminium alloys. U.K. Patent 223,994,
August 2, 1923.
(3) Diggle, J. W.; Downie, T. C.; Goulding, C. W. Chem. Rev. 1969,
69, 365.
(4) Sheasby, P. G.; Pinner, R. The Surface Treatment and Finishing of
Aluminum and Its Alloys, 6th ed.; Finishing Publications Ltd. & ASM
International: Materials Park, OH, and Stevenage, UK, 2001.
(5) Masuda, H.; Fukuda, K. Science 1995, 268, 1466.
(6) Masuda, H.; Satoh, M. Jpn. J. Appl. Phys. 1996, 35, L126.
(7) Keller, F.; Hunter, M. S.; Robinson, D. L. J. Electrochem. Soc.
1953, 100, 411.
(8) Despic, A. R. J. Electroanal. Chem. 1985, 191, 417.
(9) Takahashi, H.; Saito, Y.; Nagayama, M. J. Surf. Finish. Soc. Jpn.
1982, 33, 225.
(10) Thompson, G. E. Thin Solid Films 1997, 297, 192.
(11) Nishinaga, O.; Kikuchi, T.; Natsui, S.; Suzuki, R. O. Sci. Rep.
2013, 3, 2748.
(12) Kape, J. M. Electroplat. Met. Finish. 1961, 14, 407.
(13) Furneaux, R. C.; Rigby, W. R.; Davidson, A. P. Nature 1989,
337, 147.
(14) Thompson, G. E.; Wood, G. C. Nature 1981, 290, 230.
(15) Kape, J. M. Metallugia 1959, 60, 181.
(16) Lee, W.; Nielsch, K.; Gosele, U. Nanotechnology 2007, 18,
475713.
(17) Ono, S.; Saito, M.; Asoh, H. Electrochim. Acta 2005, 51, 827.
(18) Chu, S. Z.; Wada, K.; Inoue, S.; Isogai, M.; Katsuta, Y.;
Yasumori, A. J. Electrochem. Soc. 2006, 153, B384.
(19) Mozalev, A.; Surganov, A.; Magaino, S. Electrochim. Acta 1999,
44, 3891.
(20) Mozalev, A.; Mozaleva, I.; Sakairi, M.; Takahashi, H. Electrochim.
Acta 2005, 50, 5065.
(21) Guntherschulz, A.; Betz, H. Z. Phys. 1934, 92, 367.
(22) Verwey, J. W. Physica 1935, 2, 1059.
(23) Mott, N. F. Trans. Faraday Soc. 1947, 43, 429.
(24) Cabrera, N.; Mott, N. F. Rep. Prog. Phys. 1948, 12, 163.
(25) Lohrengel, M. M. Mater. Sci. Eng., R 1993, 11, 243.
(26) Davies, J. A.; Domeij, B.; Pringle, J. P. S.; Brown, F. J.
Electrochem. Soc. 1965, 112, 675.
(27) Davies, J. A.; Pringle, J. P. S.; Graham, R. L.; Brown, F. J.
Electrochem. Soc. 1962, 109, 999.
7547

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

(28) Shimizu, K.; Thompson, G. E.; Wood, G. C.; Xu, Y. Thin Solid
Films 1982, 88, 255.
(29) Cherki, C.; Siejka, J. J. Electrochem. Soc. 1973, 120, 784.
(30) Siejka, J.; Ortega, C. J. Electrochem. Soc. 1977, 124, 883.
(31) Wu, Z.; Richter, C.; Menon, L. J. Electrochem. Soc. 2007, 154, E8.
(32) Skeldon, P.; Thompson, G. E.; Garcia-Vergara, S. J.; IglesiasBubianes, L.; Blanco-Pinzon, C. E. Electrochem. Solid-State Lett. 2006,
9, B47.
(33) Shimizu, K.; Kobayashi, K.; Thompson, G. E.; Wood, G. C.
Philos. Mag. A 1992, 66, 643.
(34) Shimizu, K.; Kobayahi, K.; Thompson, G. E.; Wood, G. C. J.
Surf. Finish. Soc. Jpn. 1991, 42, 645.
(35) Guntherschulz, A.; Betz, H. Z. Phys. 1932, 68, 145.
(36) Dewald, J. F. Acta Metall. 1954, 2, 340.
(37) Dewald, J. F. J. Electrochem. Soc. 1955, 102, 1.
(38) Chao, C. Y.; Lin, L. F.; Macdonald, D. D. J. Electrochem. Soc.
1981, 128, 1187.
(39) Olsson, C.-O. A.; Hamm, D.; Landolt, D. J. Electrochem. Soc.
2000, 147, 4093.
(40) Brown, F.; Mackintosh, W. D. J. Electrochem. Soc. 1973, 120,
1096.
(41) Su, Z.; Buhl, M.; Zhou, W. J. Am. Chem. Soc. 2009, 131, 8697.
(42) Davies, J. A.; Domeij, B. J. Electrochem. Soc. 1963, 110, 849.
(43) Khalil, N.; Leach, J. S. L. Electrochim. Acta 1986, 31, 1279.
(44) Thompson, G. E.; Xu, Y.; Skeldon, P.; Shimizu, K.; Han, S. H.;
Wood, G. C. Philos. Mag. B 1987, 55, 651.
(45) Dekker, A.; Middelhoek, A. J. Electrochem. Soc. 1970, 117, 440.
(46) Takahashi, H.; Nagayama, M. Corros. Sci. 1978, 18, 911.
(47) Houser, J. E.; Hebert, K. R. Phys. Status Solidi A 2008, 205,
2396.
(48) Hebert, K. R.; Houser, J. E. J. Electrochem. Soc. 2009, 156, C275.
(49) Battaglia, V.; Newman, J. J. Electrochem. Soc. 1995, 142, 1423.
(50) Bradhurst, D. H.; Leach, J. S. L. J. Electrochem. Soc. 1966, 113,
1245.
(51) Tajima, S.; Shimizu, K.; Baba, N.; Matsuzawa, S. Electrochim.
Acta 1977, 22, 845.
(52) Tajima, S. Electrochim. Acta 1977, 22, 995.
(53) Shimizu, K.; Tajima, S. Electrochim. Acta 1980, 25, 259.
(54) Kasalica, B.; Belca, I.; Stojadinovic, S.; Sarvan, M.; Peric, M.;
Zekovic, L. J. Phys. Chem. C 2007, 111, 12315.
(55) Stojadinovic, S.; Belca, I.; Kasalica, B.; Zekovic, L.; Tadic, M.
Electrochem. Commun. 2006, 8, 1621.
(56) Ikonopisov, S. Electrochim. Acta 1975, 20, 783.
(57) Kasalica, B.; Stojadinovic, S.; Zekovic, L.; Belca, I.; Nikolic, D.
Electrochem. Commun. 2005, 7, 735.
(58) Stojadinovic, S.; Belca, I.; Zekovic, L.; Kasalica, B.; Nikolic, D.
Electrochem. Commun. 2004, 6, 1016.
(59) Stojadinovic, S.; Zekovic, L.; Belca, I.; Kasalica, B. Electrochem.
Commun. 2004, 6, 427.
(60) Vermilyea, D. A. J. Electrochem. Soc. 1957, 104, 542.
(61) Jackson, N. F. J. Appl. Electrochem. 1973, 3, 91.
(62) Habazaki, H.; Ogasawara, T.; Konno, H.; Shimizu, K.; Nagata,
S.; Skeldon, P.; Thompson, G. E. Corros. Sci. 2007, 49, 580.
(63) Habazaki, H.; Uozumi, M.; Konno, H.; Shimizu, K.; Skeldon, P.;
Thompson, G. E. Corros. Sci. 2003, 45, 2063.
(64) Yahalom, J.; Zahavi, J. Electrochim. Acta 1970, 15, 1429.
(65) Leach, J. S. L.; Pearson, B. R. Corros. Sci. 1988, 28, 43.
(66) Li, Y.; Shimada, H.; Sakairi, M.; Shigyo, K.; Takahashi, H.; Seo,
M. J. Electrochem. Soc. 1997, 144, 866.
(67) Proost, J.; Vanhumbeeck, J.-F.; Van Overmeere, Q. Electrochim.
Acta 2009, 55, 350.
(68) Matykina, E.; Arrabal, R.; Skeldon, P.; Thompson, G. E.;
Habazaki, H. Thin Solid Films 2008, 516, 2296.
(69) Ikonopisov, S.; Girginov, A.; Machkova, M. Electrochim. Acta
1979, 24, 451.
(70) Santway, R. W.; Alwitt, R. S. J. Electrochem. Soc. 1970, 117, 1282.
(71) Burger, F. J.; Wu, J. C. J. Electrochem. Soc. 1971, 118, 2039.
(72) Wood, G. C.; Pearson, C. Corros. Sci. 1967, 7, 119.
(73) Alwitt, R. S.; Vijh, A. K. J. Electrochem. Soc. 1969, 116, 388.

(74) Di Quarto, F.; Piazza, S.; Sunseri, C. J. Electroanal. Chem. 1988,


248, 99.
(75) Di Quarto, F.; Piazza, S.; Sunseri, C. J. Electrochem. Soc. 1984,
131, 2901.
(76) Yahalom, J.; Hoar, T. P. Electrochim. Acta 1970, 15, 877.
(77) Kato, M.; Uchida, E.; Kudo, T. J. Met. Finish. Soc. Jpn. 1984, 35,
475.
(78) Arifuku, F.; Yoneyama, H.; Tamura, H. Electrochim. Acta 1980,
25, 863.
(79) Albella, J. M.; Montero, I.; Martnez-Duart, J. M. J. Electrochem.
Soc. 1984, 131, 1101.
(80) Albella, J. M.; Montero, I.; Martnez-Duart, J. M. Thin Solid
Films 1985, 125, 57.
(81) Albella, J. M.; Montero, I.; Martnez-Duart, J. M. Electrochim.
Acta 1987, 32, 255.
(82) Zahavi, J.; Yahalom, J. Electrochim. Acta 1971, 16, 89.
(83) Di Quarto, F.; Piazza, S.; Sunseri, C. Corros. Sci. 1986, 26, 213.
(84) Dyer, C. K.; Leach, J. S. L. J. Electrochem. Soc. 1978, 125, 1032.
(85) Jouve, G.; Derradji, N. E. J. Less-Common Met. 1982, 86, 161.
(86) Vermilyea, D. A. Acta Metall. 1954, 2, 476.
(87) Albella, J. M.; Montero, I.; Martnez-Duart, J. M.; Parkhutik, V.
J. Mater. Sci. 1991, 26, 3422.
(88) Yahalom, J.; Zahavi, J. Electrochim. Acta 1971, 16, 603.
(89) Ikonopisov, S.; Girginov, A.; Machkova, M. Electrochim. Acta
1977, 22, 1283.
(90) Ikonopisov, S. Electrochim. Acta 1977, 22, 1077.
(91) Nigam, R. K.; Singh, K. C.; Maken, S. Thin Solid Films 1987,
155, 115.
(92) Ikonopisov, S. Electrochim. Acta 1969, 14, 761.
(93) Ikonopisov, S.; Elenkov, N. J. Electroanal. Chem. 1978, 86, 105.
(94) Shimizu, K. Electrochim. Acta 1981, 26, 1691.
(95) Albella, J. M.; Montero, I.; Fernandez, M.; Gomez-Aleixandre,
C.; Martnez-Duart, J. M. Electrochim. Acta 1985, 30, 1361.
(96) Montero, I.; Albella, J. M.; Martnez-Duart, J. M. J. Electrochem.
Soc. 1985, 132, 814.
(97) Sato, N. Electrochim. Acta 1971, 16, 1683.
(98) Garcia-Vergara, S. J.; Skeldon, P.; Thompson, G. E.; Habazaki,
H. Electrochim. Acta 2006, 52, 681.
(99) Sulka, G. D. Highly ordered anodic porous alumina formation
by self-organized anodizing. In Nanostructured Materials in Electrochemistry; Eftekhari, A., Ed.; Wiley-VCH Verlag GmbH & Co. KGaA:
Weinheim, 2008; pp 820.
(100) OSullivan, J. P.; Wood, G. C. Proc. R. Soc. London, Ser. A 1970,
317, 511.
(101) Parkhutik, V. P.; Shershulsky, V. I. J. Phys. D: Appl. Phys. 1992,
25, 1258.
(102) Sulka, G. D.; Parkoa, K. G. Electrochim. Acta 2007, 52, 1880.
(103) Lee, W.; Kim, J.-C.; Gosele, U. Adv. Funct. Mater. 2010, 20, 21.
(104) Ebihara, K.; Takahashi, H.; Nagayama, M. J. Met. Finish. Soc.
Jpn. 1983, 34, 548.
(105) Ebihara, K.; Takahashi, H.; Nagayama, M. J. Met. Finish. Soc.
Jpn. 1982, 33, 156.
(106) Li, A. P.; Muller, F.; Birner, A.; Nielsch, K.; Gosele, U. J. Appl.
Phys. 1998, 84, 6023.
(107) Hwang, S.-K.; Jeong, S.-H.; Hwang, H.-Y.; Lee, O.-J.; Lee, K.H. Korean J. Chem. Eng. 2002, 19, 467.
(108) Ono, S.; Masuko, N. Surf. Coat. Technol. 2003, 169170, 139.
(109) Nielsch, K.; Choy, J.; Schwirn, K.; Wehrspohn, R. B.; Gosele,
U. Nano Lett. 2002, 2, 677.
(110) Chu, S.-Z.; Wada, K.; Inoue, S.; Isogai, M.; Yasumori, A. Adv.
Mater. 2005, 17, 2115.
(111) Lee, W.; Ji, R.; Gosele, U.; Nielsch, K. Nat. Mater. 2006, 5, 741.
(112) Lee, W.; Schwirn, K.; Steinhart, M.; Pippel, E.; Scholz, R.;
Gosele, U. Nat. Nanotechnol. 2008, 3, 234.
(113) Schwirn, K.; Lee, W.; Hillebrand, R.; Steinhart, M.; Nielsch, K.;
Gosele, U. ACS Nano 2008, 2, 302.
(114) Hunter, M. S.; Fowle, P. J. Electrochem. Soc. 1954, 101, 481.
(115) Vrublevsky, I.; Parkoun, V.; Schreckenbach, J. Appl. Surf. Sci.
2005, 242, 333.
7548

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

(116) Vrublevsky, I.; Parkoun, V.; Schreckenbach, J.; Marx, G. Appl.


Surf. Sci. 2004, 227, 282.
(117) Vrublevsky, I.; Parkoun, V.; Sokol, V.; Schreckenbach, J. Appl.
Surf. Sci. 2005, 252, 227.
(118) Li, Y.; Zheng, M.; Ma, L.; Shen, W. Nanotechnology 2006, 17,
5101.
(119) Schwirn, K. Harte Anodisation von Aluminium mit Verdunnter
Schwefelaure. Ph.D. Thesis, Martin-Luther-Universitat Halle-Wittenberg, Halle (Saale), 2008.
(120) Yamamoto, Y.; Baba, N.; Tajima, S. Nature 1981, 289, 572.
(121) Xu, W. L.; Zheng, M. J.; Wu, S.; Sheng, W. Z. Appl. Phys. Lett.
2004, 85, 4364.
(122) Li, Y. B.; Zheng, M. J.; Ma, L. Appl. Phys. Lett. 2007, 91,
073109.
(123) Vojkuvka, L.; Santos, A.; Pallares, J.; Ferre-Borrull, J.; Marsal,
L. F.; Celis, J. P. Surf. Coat. Technol. 2012, 206, 2115.
(124) Roa, J. J.; Gaston-Garca, B.; Garca-Lecina, E.; Muller, C.
Ceram. Int. 2012, 38, 1627.
(125) Thompson, G. E.; Wood, G. C. Anodic Films on Aluminium.
In Treatise on Materials Science and Technology, Corrosion: Aqueous
Process and Passivation; Scully, J. C., Ed.; Academic Press: New York,
1983; pp 286287.
(126) Ono, S.; Ichinose, H.; Masuko, N. Corros. Sci. 1992, 33, 841.
(127) Mason, R. B. J. Electrochem. Soc. 1955, 102, 671.
(128) Wood, G. C.; Skeldon, P.; Thompson, G. E.; Shimizu, K. J.
Electrochem. Soc. 1996, 143, 74.
(129) Han, H.; Park, S.-J.; Jang, J. S.; Ryu, H.; Kim, K. J.; Baik, S. ACS
Appl. Mater. Interfaces 2013, 5, 3441.
(130) Thompson, G. E.; Furneaux, R. C.; Wood, G. C. Corros. Sci.
1978, 18, 481.
(131) Ono, S.; Masuko, N. Corros. Sci. 1992, 33, 503.
(132) Ono, S.; Ichinose, H.; Kawaguchi, T.; Masuko, N. Corros. Sci.
1990, 31, 249.
(133) Thompson, G. E.; Furneaux, R. C.; Wood, G. C. J. Electrochem.
Soc. 1978, 125, 1480.
(134) Thornton, M. C.; Furneaux, R. C. J. Mater. Sci. Lett. 1991, 10,
622.
(135) Le Coz, F.; Arurault, L.; Data, L. Mater. Charact. 2010, 61, 283.
(136) Oka, Y.; Takahashi, T.; Okada, K.; Iwai, S.-i. J. Non-Cryst. Solids
1979, 30, 349.
(137) El-Mashri, S. M.; Jones, R. G.; Forty, A. J. Philos. Mag. A 1983,
48, 665.
(138) Farnan, I.; Dupree, R.; Forty, A. J.; Jeong, Y. S.; Thompson, G.
E.; Wood, G. C. Philos. Mag. Lett. 1989, 59, 189.
(139) Brown, I. W. M.; Bowden, M. E.; Kemmitt, T.; MacKenzie, K.
J. D. Curr. Appl. Phys. 2006, 6, 557.
(140) Kirchner, A.; MacKenzie, K. J. D.; Brown, I. W. M.; Kemmitt,
T.; Bowden, M. E. J. Membr. Sci. 2007, 287, 264.
(141) Farnan, I.; Dupree, R.; Jeong, Y.; Thompson, G. E.; Wood, G.
C.; Forty, A. J. Thin Solid Films 1989, 173, 209.
(142) Ingham, C. J.; ter Maat, J.; de Vos, W. M. Biotechnol. Adv. 2012,
30, 1089.
(143) Jani, A. M. M.; Losic, D.; Voelcker, N. H. Prog. Mater. Sci.
2013, 58, 636.
(144) Mardilovich, P. P.; Govyadinov, A. N.; Mukhurov, N. I.;
Rzhevskii, A. M.; Paterson, R. J. Membr. Sci. 1995, 98, 131.
(145) Levin, I.; Brandon, D. J. Am. Ceram. Soc. 1998, 81, 1995.
(146) Burgos, N.; Paulis, M.; Montes, M. J. Mater. Chem. 2003, 13,
1458.
(147) Le Coz, F.; Arurault, L.; Fontorbes, S.; Vilar, V.; Datas, L.;
Winterton, P. Surf. Interface Anal. 2010, 42, 227.
(148) Ozao, R.; Ochiai, M.; Ichimura, N.; Takahashi, H.; Inada, T.
Thermochim. Acta 2000, 353354, 91.
(149) Bocchetta, P.; Sunseri, C.; Chiavarotti, G.; Quarto, F. D.
Electrochim. Acta 2003, 48, 3175.
(150) Mata-Zamora, M. E.; Saniger, J. M. Rev. Mex. Fis. 2005, 51,
502.
(151) McQuaig, M. K., Jr.; Toro, A.; Van Geertruyden, W.; Misiolek,
W. Z. J. Mater. Sci. 2011, 46, 243.

(152) Chang, Y.; Ling, Z.; Liu, Y.; Hu, X.; Li, Y. J. Mater. Chem. 2012,
22, 7445.
(153) Pilling, N. B.; Bedworth, R. E. J. Inst. Met. 1923, 29, 529.
(154) Phull, B. Evaluating stress-corrosion cracking. In ASM
Handbook, Corrosion: Fundametals, Testing, and Protection; Cramer, S.
D., Covino, B. S., Jr., Eds.; ASM International: Materials Park, OH,
2003; pp 575616.
(155) Burgers, W. G.; Claasen, A.; Zernike, J. Z. Phys. 1932, 74, 593.
(156) Bernard, W. J.; Cook, J. W. J. Electrochem. Soc. 1959, 106, 643.
(157) Dorsely, G. A., Jr. J. Electrochem. Soc. 1969, 116, 466.
(158) Garcia-Vergara, S. J.; Habazaki, H.; Skeldon, P.; Thompson, G.
E. Nanotechnology 2007, 18, 415605.
(159) Garcia-Vergara, S. J.; Skeldon, P.; Thompson, G. E.; Habazaki,
H. Corros. Sci. 2007, 49, 3696.
(160) Mercier, D.; Van Overmeere, Q.; Santoro, R.; Proost, J.
Electrochim. Acta 2011, 56, 1329.
(161) Baron-Wiechec, A.; Ganem, J. J.; Garcia-Vergara, S. J.; Skeldon,
P.; Thompson, G. E.; Vickridge, I. C. J. Electrochem. Soc. 2010, 157,
C399.
(162) Arurault, L. Trans. Inst. Met. Finish. 2008, 86, 51.
(163) Vrublevsky, I.; Parkoun, V.; Schreckenbach, J.; Marx, G. Appl.
Surf. Sci. 2003, 220, 51.
(164) Zhou, F.; Al-Zenati, A. K. M.; Baron-Wiechec, A.; Curioni, M.;
Garcia-Vergara, S. J.; Habazaki, H.; Skeldon, P.; Thompson, G. E. J.
Electrochem. Soc. 2011, 158, C202.
(165) Jessensky, O.; Muller, F.; Gosele, U. Appl. Phys. Lett. 1998, 72,
1173.
(166) Surganov, V. F.; Mozalev, A. M.; Mozaleva, I. I. Russ. J. Appl.
Chem. 1997, 70, 254.
(167) Vrublevsky, I.; Parkoun, V.; Sokol, V.; Schreckenbach, J.; Marx,
G. Appl. Surf. Sci. 2004, 222, 215.
(168) Li, F.; Zhang, L.; Metzger, R. M. Chem. Mater. 1998, 10, 2470.
(169) Nelson, J. C.; Oriani, R. A. Corros. Sci. 1993, 34, 307.
(170) Vanhumbeeck, J.-F.; Proost, J. Electrochim. Acta 2008, 53,
6165.
(171) Stoney, G. G. Proc. R. Soc. London, Ser. A 1909, 82, 172.
(172) Vermilyea, D. A. J. Electrochem. Soc. 1963, 110, 345.
(173) Moon, S.-M.; Pyun, S.-I. J. Solid State Electrochem. 1998, 2, 156.
(174) Moon, S.-M.; Pyun, S.-I. Electrochim. Acta 1998, 43, 3117.
(175) Van Overmeere, Q.; Blaffart, F.; Proost, J. Electrochem.
Commun. 2010, 12, 1174.
(176) C apraz, O . O .; Hebert, K. R.; Shrotriya, P. J. Electrochem. Soc.
2013, 160, D501.
(177) Van Overmeere, Q.; Proost, J. Electrochim. Acta 2010, 55, 4653.
(178) Van Overmeere, Q.; Proost, J. Electrochim. Acta 2011, 56,
10507.
(179) Van Overmeere, Q.; Vanhumbeeck, J.-F.; Proost, J. Rev. Sci.
Instrum. 2010, 81, 045106.
(180) Sulka, G. D.; Stroobants, S.; Moshchalkov, V. V.; Borghs, G.;
Celis, J.-P. J. Electrochem. Soc. 2004, 151, B260.
(181) Park, S.-H.; Lee, Y.; Lee, J.-K.; Kim, K.-B. Electrochem. SolidState Lett. 2006, 9, D31.
(182) Cojocaru, C. S.; Padovani, J. M.; Wade, T.; Mandoli, C.;
Jaskierowicz, G.; Wegrowe, J. E.; i Morral, A. F.; Pribat, D. Nano Lett.
2005, 5, 675.
(183) Xiang, Y.; Lee, W.; Nielsch, K.; Abstreiter, G.; i Morral, A. F.
Phys. Status Solidi RRL 2008, 2, 59.
(184) Masuda, H.; Nishio, K.; Baba, N. Appl. Phys. Lett. 1993, 63,
3155.
(185) Gowtham, M.; Eude, L.; Cojocaru, C. S.; Marquardt, B.; Jeong,
H. J.; Legagneux, P.; Song, K. K.; Pribat, D. Nanotechnology 2008, 19,
035303.
(186) Oh, J.; Thompson, C. V. J. Electrochem. Soc. 2011, 158, C71.
(187) Hoar, T. P.; Yahalom, J. J. Electrochem. Soc. 1963, 110, 614.
(188) Albella, J. M.; Montero, I.; Jimenez, M. C.; Martnez-Duart, J.
M. Electrochim. Acta 1991, 36, 739.
(189) Van Overmeere, Q.; Mercier, D.; Santoro, R.; Proost, J.
Electrochem. Solid-State Lett. 2012, 15, C1.
7549

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

(190) Van Overmeere, Q.; Nysten, B.; Proost, J. Appl. Phys. Lett.
2009, 94, 074103.
(191) Garcia-Vergara, S. J.; Skeldon, P.; Thompson, G. E.; Habazaki,
H. Corros. Sci. 2007, 49, 3772.
(192) Singh, G. K.; Golovin, A. A.; Aranson, I. S. Phys. Rev. B 2006,
73, 205422.
(193) Hoar, T. P.; Mott, N. F. J. Phys. Chem. Solids 1959, 9, 97.
(194) Hebert, K. R.; Albu, S. P.; Paramasivam, I.; Schmuki, P. Nat.
Mater. 2012, 11, 162.
(195) Raja, K. S.; Misra, M.; Paramguru, K. Electrochim. Acta 2005,
51, 154.
(196) Sample, C.; Golovin, A. A. Phys. Rev. E 2006, 74, 041606.
(197) Stanton, L. G.; Golvin, A. A. Phys. Rev. B 2009, 79, 035414.
(198) Thamida, S. K.; Chang, H.-C. Chaos 2002, 12, 240.
(199) Houser, J. E.; Hebert, K. R. J. Electrochem. Soc. 2006, 153,
B566.
(200) Houser, J. E.; Hebert, K. R. Nat. Mater. 2009, 8, 415.
(201) Hunter, M. S.; Fowle, P. J. Electrochem. Soc. 1954, 101, 514.
(202) Leach, J. S. L.; Neufeld, P. Corros. Sci. 1969, 9, 413.
(203) Nagayama, M.; Tamura, K. Electrochim. Acta 1968, 13, 1773.
(204) Chowdhury, P.; Thomas, A. N.; Sharma, M.; Barshilia, H. C.
Electrochim. Acta 2014, 115, 657.
(205) Singh, G. K.; Golovin, A. A.; Aranson, I. S.; Vinokur, V. M.
Europhys. Lett. 2005, 70, 836.
(206) Friedman, A. L.; Brittain, D.; Menon, L. J. Chem. Phys. 2007,
127, 154717.
(207) Oh, J.; Thompson, C. V. Electrochim. Acta 2011, 56, 4044.
(208) Baron-Wiechec, A.; Burke, M. G.; Hashimoto, T.; Liu, H.;
Skeldon, P.; Thompson, G. E.; Habazaki, H.; Ganem, J.-J.; Vickridge, I.
C. Electrochim. Acta 2013, 113, 302.
(209) Garcia-Vergara, S. J.; Clere, D. L.; Hashimoto, T.; Habazaki,
H.; Skeldon, P.; Thompson, G. E. Electrochim. Acta 2009, 54, 6403.
(210) Garcia-Vergara, S. J.; Skeldon, P.; Thompson, G. E.; Habazaki,
H. Appl. Surf. Sci. 2007, 254, 1534.
(211) Habazaki, H.; Shimizu, K.; Skeldon, P.; Thompson, G. E.;
Wood, G. C. J. Electrochem. Soc. 1996, 143, 2465.
(212) Habazaki, H.; Shimizu, K.; Skeldon, P.; Thompson, G. E.;
Wood, G. C.; Zhou, X. Trans. Inst. Met. Finish. 1997, 75, 18.
(213) Garcia-Vergara, S. J.; Hashimoto, T.; Skeldon, P.; Thompson,
G. E.; Habazaki, H. Electrochim. Acta 2009, 54, 3662.
(214) Iglesias-Rubianes, L.; Skeldon, P.; Thompson, G. E.; Habazaki,
H.; Shimizu, K. J. Electrochem. Soc. 2002, 149, B23.
(215) Pringle, J. P. S. Electrochim. Acta 1980, 25, 1423.
(216) Garcia-Vergara, S. J.; Skeldon, P.; Thompson, G. E.; Habazaki,
H. Surf. Interface Anal. 2007, 39, 860.
(217) Bolger, C. T.; Fois, G.; Petkov, N.; Sassiat, N.; Burke, M.;
Quinn, A. J.; Cross, G. L. W.; Holmes, J. D. Nanotechnology 2012, 23,
175602.
(218) Garcia-Vergara, S. J.; Skeldon, P.; Thompson, G. E.; Habazaki,
H. Corros. Sci. 2008, 50, 3179.
(219) Forgazza, M.; Santamaria, M.; Di Quarto, F.; Garcia-Vergara, S.
J.; Molchan, I.; Skeldon, P.; Thompson, G. E.; Habazaki, H.
Electrochim. Acta 2009, 54, 1070.
(220) Herrera-Erazo, A. E.; Habazaki, H.; Shimizu, K.; Skeldon, P.;
Thompson, G. E. Corros. Sci. 2000, 42, 1823.
(221) Csokan, P. Metalloberflache 1961, 15, B49.
(222) Csokan, V. P.; Hollo, M. Werkst. Korros. 1961, 12, 288.
(223) Csokan, P.; Sc., C. C. Electroplat. Met. Finish. 1962, 15, 75.
(224) Lee, W. JOM 2010, 62, 57.
(225) Schwartz, G. C.; Platter, V. J. Electrochem. Soc. 1975, 122, 1508.
(226) Hillebrand, R.; Muller, F.; Schwirn, K.; Lee, W.; Steinhart, M.
ACS Nano 2008, 2, 913.
(227) Masuda, H.; Hasegawa, E.; Ono, S. J. Electrochem. Soc. 1997,
144, L127.
(228) Zhang, L.; Cho, H. S.; Li, F.; Metzger, R. M.; Doyle, W. D. J.
Mater. Sci. Lett. 1998, 17, 291.
(229) Masuda, H.; Yada, K.; Osaka, A. Jpn. J. Appl. Phys. 1998, 37,
L1340.

(230) Shingubara, S.; Morimoto, K.; Sakaue, H.; Takahagi, T.


Electrochem. Solid-State Lett. 2004, 7, E15.
(231) Sun, C.; Luo, J.; Wu, L.; Zhang, J. ACS Appl. Mater. Interfaces
2010, 2, 1299.
(232) Ono, S.; Saito, M.; Ishiguro, M.; Asoh, H. J. Electrochem. Soc.
2004, 151, B473.
(233) Chen, W.; Wu, J.-S.; Xia, X.-H. ACS Nano 2008, 2, 959.
(234) Martn, J.; Manzano, C. V.; Caballero-Calero, O.; MartnGonzalez, M. ACS Appl. Mater. Interfaces 2013, 5, 72.
(235) Su, Z.; Hahner, G.; Zhou, W. J. Mater. Chem. 2008, 18, 5787.
(236) Su, Z.; Zhou, W. Adv. Mater. 2008, 20, 3663.
(237) Su, Z.; Zhou, W. J. Mater. Chem. 2011, 21, 357.
(238) Hoar, T. P. Corros. Sci. 1967, 7, 341.
(239) Tu, G. C.; Chen, I. T.; Shimizu, K. J. Jpn. Inst. Light Met. 1990,
40, 382.
(240) Ono, S.; Saito, M.; Asoh, H. Electrochem. Solid-State Lett. 2004,
7, B21.
(241) Csokan, P. Trans. Inst. Met. Finish. 1964, 41, 51.
(242) Arrowsmith, D. J.; Clifford, A. W.; Moth, D. A. J. Mater. Sci.
Lett. 1986, 5, 921.
(243) Wada, K.; Shimohira, T.; Yamada, M.; Baba, N. J. Mater. Sci.
1986, 21, 3810.
(244) Kashi, M. A.; Ramazani, A.; Noormohammadi, M.; Zarei, M.;
Marashi, P. J. Phys. D: Appl. Phys. 2007, 40, 7032.
(245) Li, Y.; Ling, Z. Y.; Chen, S. S.; Wang, J. C. Nanotechnology
2008, 19, 225604.
(246) Srolovitz, D. J. Acta Metall. 1989, 37, 621.
(247) Li, F. Nanostructure of anodic porous alumina lms of interest
in magnetic recording. Ph.D. Thesis, The University of Alabama, AL,
1998.
(248) Gill, V.; Guduru, P. R.; Sheldon, B. W. Int. J. Solids Struct. 2008,
45, 943.
(249) Matthias, S.; Schilling, J.; Nielsch, K.; Muller, F.; Wehrspohn,
R. B.; Gosele, U. Adv. Mater. 2002, 14, 1618.
(250) He, B.; Son, S. J.; Lee, S. B. Langmuir 2006, 22, 8263.
(251) Brunker, S. E.; Cederquist, K. B.; Keating, C. D. Nanomedicine
2007, 2, 695.
(252) Matthias, S.; Muller, F. Nature 2003, 424, 53.
(253) Casanova, F.; Chiang, C. E.; Li, C.-P.; Schuller, I. K. Appl. Phys.
Lett. 2007, 91, 243103.
(254) Bruschi, L.; Mistura, G.; Liu, L.; Lee, W.; Gosele, U.; Coasne,
B. Langmuir 2010, 26, 11894.
(255) Miller, M. A. Anodizing process and system. U.S. Patent
2,920,018, January 5, 1960.
(256) Schaedel, F. C. Aluminum anodizing method. U.S. Patent
3,418,222, December 24, 1968.
(257) Yokoyama, K.; Konno, H.; Takahashi, H.; Nagayama, M. Plat.
Surf. Finish. 1982, 7, 62.
(258) Tu, G. C.; Huang, L. Y. Trans. Inst. Met. Finish. 1987, 65, 60.
(259) Raj, V.; Rajaram, M. P.; Balasubramanian, G.; Vincent, S.;
Kanagaraj, D. Trans. Inst. Met. Finish. 2003, 81, 114.
(260) Murphy, J. F.; Michelson, C. E. A Theory for the Formation of
Anodic Oxide Coatings on Aluminium. Proceedings of Anodizing
Aluminium; Nottingham, England, 1961; pp 8395.
(261) Takahashi, H.; Nagayama, M.; Akahori, H.; Kitahara, A. J.
Electron Microsc. 1973, 22, 149.
(262) Rasmussen, J. Plat. Surf. Finish. 2001, 88, 42.
(263) Lee, W.; Scholz, R.; Gosele, U. Nano Lett. 2008, 8, 2155.
(264) Lee, W.; Kim, J.-C. Nanotechnology 2010, 21, 485304.
(265) Losic, D.; Lillo, M.; Losic, D., Jr. Small 2009, 5, 1392.
(266) Losic, D.; Losic, D., Jr. Langmuir 2009, 25, 5426.
(267) Wang, B.; Xu, G.; Cui, P. Mater. Lett. 2013, 93, 336.
(268) Yang, C.-J.; Liang, S.-W.; Wu, P.-W.; Chen, C.; Shieh, J.-M.
Electrochem. Solid-State Lett. 2007, 10, C69.
(269) Miney, P. G.; Colavita, P. E.; Schiza, M. V.; Priore, R. J.;
Haibach, F. G.; Myrick, M. L. Electrochem. Solid-State Lett. 2003, 6,
B42.
(270) Rabin, O.; Herz, P. R.; Lin, Y.-M.; Akinwande, A. I.; Cronin, S.
B.; Dresselhaus, M. S. Adv. Funct. Mater. 2003, 13, 631.
7550

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

(271) Crouse, D.; Lo, Y.-H.; Miller, A. E.; Crouse, M. Appl. Phys. Lett.
2000, 76, 49.
(272) Asoh, H.; Matsuo, M.; Yoshihama, M.; Ono, S. Appl. Phys. Lett.
2003, 83, 4408.
(273) Shiraki, H.; Kimura, Y.; Ishii, H.; Ono, S.; Itaya, K.; Niwano, M.
Appl. Surf. Sci. 2004, 237, 369.
(274) Tian, M.; Xu, S.; Wang, J.; Kumar, N.; Wertz, E.; Li, Q.;
Campbell, P. M.; Chan, M. H. W.; Mallouk, T. E. Nano Lett. 2005, 5,
697.
(275) Kokonou, M.; Nassiopoulou, A. G.; Giannakopoulos, K. P.
Nanotechnology 2005, 16, 103.
(276) Seo, H.-S.; Jung, Y.-G.; Jee, S.-W.; Yang, J. M.; Lee, J.-H. Scr.
Mater. 2007, 57, 968.
(277) Chen, P.-L.; Kuo, C.-T.; Tsai, T.-G.; Wu, B.-W.; Hsu, C.-C.;
Pan, F.-M. Appl. Phys. Lett. 2003, 82, 2796.
(278) Sander, M. S.; Tan, L.-S. Adv. Funct. Mater. 2003, 13, 393.
(279) Choi, J.; Sauer, G.; Goring, P.; Nielsch, K.; Wehrspohn, R. B.;
Gosele, U. J. Mater. Chem. 2003, 13, 1100.
(280) Zhao, X.; Lee, U.-J.; Seo, S.-K.; Lee, K.-H. Mater. Sci. Eng., C
2009, 29, 1156.
(281) Yang, Y.; Chen, H.; Mei, Y.; Chen, J.; Wu, X.; Bao, X. Solid
State Commun. 2002, 123, 279.
(282) Yasui, N.; Imada, A.; Den, T. Appl. Phys. Lett. 2003, 83, 3347.
(283) Crouse, M. M.; Miller, A. E.; Crouse, D. T.; Ikram, A. A. J.
Electrochem. Soc. 2005, 152, D167.
(284) Vorobyova, A. I.; Outkina, E. A. Thin Solid Films 1998, 324, 1.
(285) Mozalev, A.; Surganov, A.; Magaino, S. Electrochim. Acta 1999,
44, 3891.
(286) Mozalev, A.; Khatko, V.; Bittencourt, C.; Hassel, A. W.;
Gorokh, G.; Llobet, E.; Correig, X. Chem. Mater. 2008, 20, 6482.
(287) Iwasaki, T.; Motoi, T.; Den, T. Appl. Phys. Lett. 1999, 75, 2044.
(288) Chu, S.-Z.; Wada, K.; Inoue, S.; Segawa, H. J. Electrochem. Soc.
2011, 158, C148.
(289) Oh, J.; Thompson, C. V. Adv. Mater. 2008, 20, 1368.
(290) Oh, J.; Shin, Y. C.; Thompson, C. V. J. Electrochem. Soc. 2011,
158, K11.
(291) Chu, S. Z.; Wada, K.; Inoue, S.; Todoroki, S. J. Electrochem. Soc.
2002, 149, B321.
(292) Chu, S. Z.; Wada, K.; Inoue, S.; Todoroki, S. Electrochim. Acta
2003, 48, 3147.
(293) Inoue, S.; Todoroki, S.; Suehara, S.; Konishi, T.; Chu, S. Z.;
Wada, K.; Kikkojin, T.; Isogai, M.; Katsuta, Y.; Sakamoto, T.;
Yasumori, A. J. Non-Cryst. Solids 2006, 352, 632.
(294) Musselman, K. P.; Mulholland, G. J.; Robinson, A. P.; SchmidtMende, L.; MacManus-Driscoll, J. L. Adv. Mater. 2008, 20, 4470.
(295) Foong, T. R. B.; Sellinger, A.; Hu, X. ACS Nano 2008, 2, 2250.
(296) Schierhorn, M.; Boettcher, S. W.; Kraemer, S.; Stucky, G. D.;
Moskovits, M. Nano Lett. 2009, 9, 3262.
(297) Hill, J. J.; Haller, K.; Ziegler, K. J. J. Electrochem. Soc. 2011, 158,
E1.
(298) Chu, S.-Z.; Wada, K.; Inoue, S.; Todoroki, S.-i. Chem. Mater.
2002, 14, 4595.
(299) Masuda, H.; Yasui, K.; Sakamoto, Y.; Nakao, M.; Tamamura,
T.; Nishio, K. Jpn. J. Appl. Phys. 2001, 40, L1267.
(300) Kimura, Y.; Shiraki, H.; Ishibashi, K.-i.; Ishii, H.; Itaya, K.;
Niwanoa, M. J. Electrochem. Soc. 2006, 153, C296.
(301) Holubowitch, N.; Nagle, L. C.; Rohan, J. F. Solid State Ionics
2012, 216, 110.
(302) Shimizu, K.; Habazaki, H.; Skeldon, P.; Thompson, G. E.;
Wood, G. C. J. Surf. Finish. Soc. Jpn. 1999, 50, 2.
(303) Skeldon, P.; Shimizu, K.; Thompson, G. E.; Wood, G. C.
Philos. Mag. B 1990, 61, 927.
(304) Habazaki, H.; Skeldon, P.; Shimizu, K.; Thompson, G. E.;
Wood, G. C. Corros. Sci. 1995, 37, 1497.
(305) Mozalev, A.; Sakairi, M.; Saeki, I.; Takahashi, H. Electrochim.
Acta 2003, 48, 3155.
(306) Tatarenko, N. I.; Mozalev, A. M. Solid-State Electron. 2001, 45,
1009.

(307) Mozalev, A.; Sakairi, M.; Takahashi, H. J. Electrochem. Soc.


2004, 151, F257.
(308) Masuda, H.; Yamada, H.; Satoh, M.; Asoh, H.; Nakao, M.;
Tamamura, T. Appl. Phys. Lett. 1997, 71, 2770.
(309) Masuda, H.; Asoh, H.; Watanabe, M.; Nishio, K.; Nakao, M.;
Tamamura, T. Adv. Mater. 2001, 13, 189.
(310) Asoh, H.; Nishio, K.; Nakao, M.; Yokoo, A.; Tamamura, T.;
Masuda, H. J. Vac. Sci. Technol., B 2001, 19, 569.
(311) Masuda, H.; Yotsuya, M.; Asano, M.; Nishio, K.; Nakao, M.;
Yokoo, A.; Tamamura, T. Appl. Phys. Lett. 2001, 78, 826.
(312) Smith, J. T.; Huang, Q.; Franklin, A. D.; Janes, D. B.; Sands, T.
D. Appl. Phys. Lett. 2008, 93, 043108.
(313) Jones, M. R.; Osberg, K. D.; Macfarlane, R. J.; Langille, M. R.;
Mirkin, C. A. Chem. Rev. 2011, 111, 3736.
(314) Mikulskas, I.; Juodkazis, S.; Tomasiunas,
R.; Dumas, J. G. Adv.

Mater. 2001, 13, 1574.


(315) Choi, J.; Nielsch, K.; Reiche, M.; Wehrspohn, R. B.; Gosele, U.
J. Vac. Sci. Technol., B 2003, 21, 763.
(316) Fournier-Bidoz, S.; Kitaev, V.; Routkevitch, D.; Manners, I.;
Ozin, G. A. Adv. Mater. 2004, 16, 2193.
(317) Lee, W.; Ji, R.; Ross, C. A.; Gosele, U.; Nielsch, K. Small 2006,
2, 978.
(318) Matsui, Y.; Nishio, K.; Masuda, H. Small 2006, 2, 522.
(319) Liu, C. Y.; Datta, A.; Wang, Y. L. Appl. Phys. Lett. 2001, 78,
120.
(320) Masuda, H.; Kanezawa, K.; Nishio, K. Chem. Lett. 2002, 31,
1218.
(321) Sun, Z.; Kim, H. K. Appl. Phys. Lett. 2010, 81, 3458.
(322) Kim, B.; Park, S.; McCarthy, T. J.; Russell, T. P. Small 2007, 3,
1869.
(323) Lipson, A. L.; Comstock, D. J.; Hersam, M. C. Small 2009, 5,
2807.
(324) Lai, K.-L.; Hon, M.-H.; Leu, I.-C. Nanoscale Res. Lett. 2011, 6,
157.
(325) Oshima, H.; Kikuchi, H.; Nakao, H.; Itoh, K.-i.; Kamimura, T.;
Morikawa, T.; Matsumoto, K.; Umada, T.; Tamura, H.; Nishio, K.;
Masuda, H. Appl. Phys. Lett. 2007, 91, 022508.
(326) Kustandi, T. S.; Loh, W. W.; Gao, H.; Low, H. Y. ACS Nano
2010, 4, 2561.
(327) Masuda, H.; Matsui, Y.; Yotsuya, M.; Matsumoto, F.; Nishio,
K. Chem. Lett. 2004, 33, 584.
(328) Liu, N. W.; Datta, A.; Liu, C. Y.; Wang, Y. L. Appl. Phys. Lett.
2003, 82, 1281.
(329) Chen, B.; Lu, K.; Tian, Z. Electrochim. Acta 2010, 56, 435.
(330) Chen, B.; Lu, K.; Tian, Z. Langmuir 2011, 27, 800.
(331) Shingubara, S.; Murakami, Y.; Morimoto, K.; Takahagi, T. Surf.
Sci. 2003, 532535, 317.
(332) Nasir, M. E.; Allsopp, D. W. E.; Bowen, C. R.; Hubbard, G.;
Parsons, K. P. Nanotechnology 2010, 21, 105303.
(333) Xia, Y.; Whitesides, G. M. Annu. Rev. Mater. Sci. 1998, 28, 153.
(334) Guo, L. J. Adv. Mater. 2007, 19, 495.
(335) Asoh, H.; Nishio, K.; Nakao, M.; Tamamura, T.; Masuda, H. J.
Electrochem. Soc. 2001, 148, B152.
(336) Liu, Y.; Goebl, J.; Yin, Y. Chem. Soc. Rev. 2013, 42, 2610.
(337) Possin, G. E. Rev. Sci. Instrum. 1970, 41, 772.
(338) Martin, C. R. Science 1994, 266, 1961.
(339) Martin, C. R. Adv. Mater. 1991, 3, 457.
(340) Martin, C. R. Chem. Mater. 1996, 8, 1739.
(341) Penner, R. M.; Martin, C. R. Anal. Chem. 1987, 59, 2625.
(342) Hurst, S. J.; Payne, E. K.; Qin, L.; Mirkin, C. A. Angew. Chem.,
Int. Ed. 2006, 45, 2672.
(343) Kovtyukhova, N. I.; Mallouk, T. E. Chem.Eur. J. 2002, 8,
4355.
(344) Nicewarner-Pena, S. R.; Freeman, R. G.; Reiss, B. D.; He, L.;
Pena, D. J.; Walton, I. D.; Cromer, R.; Keating, C. D.; Nata, M. J.
Science 2001, 294, 137.
(345) Choi, J.; Sauer, G.; Nielsch, K.; Wehrspohn, R. B.; Gosele, U.
Chem. Mater. 2003, 15, 776.
7551

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

(346) Su, Y.-K.; Qin, D.-H.; Zhang, H.-L.; Li, H.; Li, H.-L. Chem.
Phys. Lett. 2004, 388, 406.
(347) Nielsch, K.; Muller, F.; Li, A. P.; Gosele, U. Adv. Mater. 2000,
12, 582.
(348) de Menten de Horne, F.; Piraux, L.; Michottea, S. Appl. Phys.
Lett. 2005, 86, 152510.
(349) Wang, J.-G.; Tian, M.-L.; Kumar, N.; Mallouk, T. E. Nano Lett.
2005, 5, 1247.
(350) Chaure, N. B.; Stamenov, P.; Rhen, F. M. F.; Coey, J. M. D. J.
Magn. Magn. Mater. 2005, 290291, 1210.
(351) Yasui, K.; Morikawa, T.; Nishio, K.; Masuda, H. Jpn. J. Appl.
Phys. 2005, 44, L469.
(352) Xu, X.; Xu, Y. Mater. Lett. 2006, 60, 2069.
(353) Zhang, Y.; Li, G.; Wu, Y.; Zhang, B.; Song, W.; Zhang, L. Adv.
Mater. 2002, 14, 1227.
(354) Tok, J. B.-H.; Chuang, F. Y. S.; Kao, M. C.; Rose, K. A.; Pannu,
S. S.; Sha, M. Y.; Chakarova, G.; Penn, S. G.; Dougherty, G. M. Angew.
Chem., Int. Ed. 2006, 45, 6900.
(355) Walton, I. D.; Norton, S. M.; Balasingham, A.; He, L.;
Dominador, F.; Oviso, J.; Gupta, D.; Raju, P. A.; Natan, M. J.;
Freeman, R. G. Anal. Chem. 2002, 74, 2240.
(356) Qin, L.; Park, S.; Huang, L.; Mirkin, C. A. Science 2005, 309,
113.
(357) Chen, X.; Yeganeh, S.; Qin, L.; Li, S.; Xue, C.; Braunschweig,
A. B.; Schartz, G. C.; Ratner, M. A.; Mirkin, C. A. Nano Lett. 2009, 9,
3974.
(358) Qin, L.; Banholzer, M. J.; Millstone, J. E.; Mirkin, C. A. Nano
Lett. 2007, 7, 3849.
(359) Banholzer, M. J.; Qin, L.; Millstone, J. E.; Osberg, K. D.;
Mirkin, C. A. Nat. Protoc. 2009, 4, 838.
(360) Park, S.; Lim, J.-H.; Chung, S.-W.; Mirkin, C. A. Science 2004,
303, 348.
(361) Ciszek, J. W.; Huang, L.; Wang, Y.; Mirkin, C. A. Small 2008,
4, 206.
(362) Yahalom, J.; Zadok, O. J. Mater. Sci. 1987, 22, 499.
(363) Cohen-Hyams, T.; Plitzko, J. M.; Hetherington, C. J. D.;
Hutchison, J. L.; Yahalom, J.; Kaplan, W. D. J. Mater. Sci. 2004, 39,
5701.
(364) Uhlemann, M.; Gebert, A.; Herrich, M.; Krause, A.; Cziraki, A.;
Schultz, L. Electrochim. Acta 2003, 48, 3005.
(365) Marsza, M.; Jaworski, J.; Michalik, A.; Prokop, J.; Stachura, Z.;
Voznyi, V.; Bolling, O.; Sulkio-Cleff, B. J. Magn. Magn. Mater. 2001,
226230, 1735.
(366) Urbaniak, M.; Lucinski, T.; Stobiecki, F. J. Magn. Magn. Mater.
1998, 190, 187.
(367) Piraux, L.; George, J. M.; Despres, J. F.; Leroy, C.; Ferain, E.;
Legras, R.; Ounadjela, K.; Fert, A. Appl. Phys. Lett. 1994, 65, 2484.
(368) Blondel, A.; Meier, J. P.; Doudin, B.; Ansermet, J.-P. Appl. Phys.
Lett. 1994, 65, 3019.
(369) Liu, K.; Nagodawithana, K.; Searson, P. C.; Chien, C. L. Phys.
Rev. B 1995, 51, 7381.
(370) Tourillon, G.; Pontonnier, L.; Levy, J. P.; Langlais, V.
Electrochem. Solid-State Lett. 2000, 3, 20.
(371) Guo, Y.-G.; Wan, J.-J.; Zhu, C.-F.; Yang, D.-L.; Chen, D.-M.;
Bai, C.-L. Chem. Mater. 2003, 15, 664.
(372) Chen, M.; Searson, P. C.; Chien, C. L. J. Appl. Phys. 2003, 93,
8253.
(373) Dubois, S.; Marchal, C.; Beuken, J. M.; Piraux, L.; Duvail, J. L.;
Fert, A.; George, J. M.; Maurice, J. L. Appl. Phys. Lett. 1999, 70, 396.
(374) Liu, L.; Lee, W.; Scholz, R.; Pippel, E.; Gosele, U. Angew.
Chem., Int. Ed. 2008, 47, 7004.
(375) Lee, W.; Scholz, R.; Nielsch, K.; Gosele, U. Angew. Chem., Int.
Ed. 2005, 44, 6050.
(376) Yuan, X. Y.; Wu, G. S.; Xie, T.; Geng, B. Y.; Lin, Y.; Meng, G.
W.; Zhanga, L. D. Solid State Sci. 2004, 6, 735.
(377) Yuan, X. Y.; Wu, G. S.; Xie, T.; Lin, Y.; Zhang, L. D.
Nanotechnology 2004, 15, 59.
(378) Menon, V. P.; Martin, C. R. Anal. Chem. 1995, 67, 1920.

(379) Nishizawa, M.; Menon, V. P.; Martin, C. R. Science 1995, 268,


700.
(380) Jirage, K. B.; Hulteen, J. C.; Martin, C. R. Science 1997, 278,
655.
(381) Kohli, P.; Wharton, J. E.; Braide, O.; Martin, C. R. J. Nanosci.
Nanotechnol. 2004, 4, 605.
(382) Jirage, K. B.; Hulteen, J. C.; Martin, C. R. Anal. Chem. 1999,
71, 4913.
(383) Martin, C. R.; Nishizawa, M.; Jirage, K.; Kang, M. J. Phys.
Chem. B 2001, 105, 1925.
(384) Martin, C. R.; Nishizawa, M.; Jirage, K.; Kang, M.; Lee, S. B.
Adv. Mater. 2001, 13, 1351.
(385) Lee, S. B.; Martin, C. R. Chem. Mater. 2001, 13, 3236.
(386) Lee, S. B.; Martin, C. R. J. Am. Chem. Soc. 2002, 124, 11850.
(387) Siwy, Z.; Heins, E.; Harrell, C. C.; Kohli, P.; Martin, C. R. J.
Am. Chem. Soc. 2004, 126, 10850.
(388) Yu, S.; Lee, S. B.; Kang, M.; Martin, C. R. Nano Lett. 2001, 1,
495.
(389) Wirtz, M.; Yu, S.; Martin, C. R. Analyst 2002, 127, 871.
(390) Kobayashi, Y.; Martin, C. R. Anal. Chem. 1999, 71, 3665.
(391) Wirtz, M.; Parker, M.; Kobayashi, Y.; Martin, C. R. Chem. Rec.
2002, 2, 259.
(392) Steinle, E. D.; Mitchell, D. T.; Wirtz, M.; Lee, S. B.; Young, V.
Y.; Martin, C. R. Anal. Chem. 2002, 74, 2416.
(393) Kohli, P.; Wirtz, M.; Martin, C. R. Electroanalysis 2004, 16, 9.
(394) Yu, Y.; Kant, K.; Shapter, J. G.; Addai-Mensah, J.; Losic, D.
Microporous Mesoporous Mater. 2012, 153, 131.
(395) Volpe, M.; Inguanta, R.; Piazza, S.; Sunseri, C. Surf. Coat.
Technol. 2006, 200, 5800.
(396) Inguanta, R.; Amodeo, M.; DAgostino, F.; Volpe, M.; Piazza,
S.; Sunseri, C. J. Electrochem. Soc. 2007, 154, D188.
(397) Wu, J. P.; Brown, I. W. M.; Bowden, M. E.; Kemmitt, T. Solid
State Sci. 2010, 12, 1912.
(398) Wang, W.; Li, N.; Li, X.; Geng, W.; Qiu, S. Mater. Res. Bull.
2006, 41, 1417.
(399) Li, N.; Li, X.; Yin, X.; Wang, W.; Qiu, S. Solid State Commun.
2004, 132, 841.
(400) Chu, S. Z.; Kawamura, H.; Mori, M. J. Electrochem. Soc. 2008,
155, D414.
(401) Azizi, A.; Mohammadi, M.; Sadrnezhaad, S. K. Mater. Lett.
2011, 65, 289.
(402) Yuan, X.; Du, C.; Sun, G.; Pan, N. Appl. Surf. Sci. 2007, 253,
4546.
(403) Nagaura, T.; Takeuchi, F.; Yamauchi, Y.; Wada, K.; Inoue, S.
Electrochem. Commun. 2008, 10, 681.
(404) Lakshmi, B. B.; Dorhout, P. K.; Martin, C. R. Chem. Mater.
1997, 9, 857.
(405) Lakshmi, B. B.; Patrissi, C. J.; Martin, C. R. Chem. Mater. 1997,
9, 2544.
(406) Hulteen, J. C.; Martin, C. R. J. Mater. Chem. 1997, 7, 1075.
(407) Foong, T. R. B.; Shen, Y.; Hu, X.; Sellinger, A. Adv. Funct.
Mater. 2010, 20, 1390.
(408) Kang, T.-S.; Smith, A. P.; Taylor, B. E.; Durstock, M. F. Nano
Lett. 2009, 9, 601.
(409) Wang, G. X.; Park, J. S.; Park, M. S.; Gou, X. L. Sens. Actuators,
B 2008, 131, 313.
(410) O zturk, S.; Klnc, N.; TasAltin, N.; O zturk, Z. Z. Thin Solid
Films 2011, 520, 932.
(411) Ji, G.; Tang, S.; Xu, B.; Gu, B.; Du, Y. Chem. Phys. Lett. 2003,
379, 484.
(412) Wu, Z.; Niu, P.; Yang, B.; Yu, R. Appl. Phys. A: Mater. Sci.
Process. 2011, 105, 177.
(413) Xu, X.; Qian, T.; Zhang, G.; Zhang, T.; Li, G.; Wang, W.; Li, X.
Chem. Lett. 2007, 36, 112.
(414) Feng, M.; Wang, W.; Zhou, Y.; Jia, D. J. Sol-Gel Sci. Technol.
2009, 52, 120.
(415) Min, H.-S.; Lee, J.-K. Ferroelectrics 2006, 336, 231.
(416) Zhang, G.; Lu, X.; Zhang, T.; Qu, J.; Wang, W.; Li, X.; Yu, S.
Nanotechnology 2006, 17, 4252.
7552

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

(417) Lai, S. H.; Hsu, Y. C.; Lan, M. D. Solid State Commun. 2008,
148, 452.
(418) Wu, G. S.; Lin, Y.; Yuan, X. Y.; Xie, T.; Cheng, B. C.; Zhang, L.
D. Nanotechnology 2004, 15, 568.
(419) Lei, Y.; Zhang, L. D.; Meng, G. W.; Li, G. H.; Zhang, X. Y.;
Liang, C. H.; Chen, W.; Wang, S. X. Appl. Phys. Lett. 2001, 78, 1125.
(420) Wu, G. S.; Zhuang, Y. L.; Lin, Z. Q.; Yuan, X. Y.; Xie, T.;
Zhang, L. D. Physica E 2006, 31, 5.
(421) Zhang, M.; Bando, Y.; Wada, K. J. Mater. Res. 2000, 15, 387.
(422) Limmer, S. J.; Seraji, S.; Forbess, M. J.; Wu, Y.; Chou, T. P.;
Nguyen, C.; Cao, G. Adv. Mater. 2001, 13, 1269.
(423) Limmer, S. J.; Seraji, S.; Wu, Y.; Chou, T. P.; Nguyen, C.; Cao,
G. Adv. Funct. Mater. 2002, 12, 59.
(424) Miao, Z.; Xu, D.; Ouyang, J.; Guo, G.; Zhao, X.; Tang, Y. Nano
Lett. 2002, 2, 717.
(425) Lee, S. B.; Mitchell, D. T.; Trofin, L.; Nevanen, T. K.;
Soderlund, H.; Martin, C. R. Science 2002, 296, 2198.
(426) Mitchell, D. T.; Lee, S. B.; Trofin, L.; Li, N.; Nevanen, T. K.;
Soderlund, H.; Martin, C. R. J. Am. Chem. Soc. 2002, 124, 11864.
(427) Son, S. J.; Lee, S. B. J. Am. Chem. Soc. 2006, 128, 15974.
(428) Buyukserin, F.; Martin, C. R. Appl. Surf. Sci. 2010, 256, 7700.
(429) Hillebrenner, H.; Buyukseri, F.; Kang, M.; Mota, M. O.;
Stewart, J. D.; Martin, C. R. J. Am. Chem. Soc. 2006, 128, 4236.
(430) Chen, C.-C.; Liu, Y.-C.; Wu, C.-H.; Yeh, C.-C.; Su, M.-T.; Wu,
Y.-C. Adv. Mater. 2005, 17, 404.
(431) Yamacuchi, A.; Uejo, F.; Yoda, T.; Uchida, T.; Tanamura, Y.;
Yamashita, T.; Teramae, N. Nat. Mater. 2004, 3, 337.
(432) Platschek, B.; Petkov, N.; Bein, T. Angew. Chem., Int. Ed. 2006,
45, 1134.
(433) Platschek, B.; Petkov, N.; Himsl, D.; Zimdars, S.; Li, Z.; Kohn,
R.; Bein, T. J. Am. Chem. Soc. 2008, 130, 17362.
(434) Wu, Y.; Cheng, G.; Katsov, K.; Sides, S. W.; Wang, J.; Tang, J.;
Fredrickson, G. H.; Moskovits, M.; Stucky, G. D. Nat. Mater. 2004, 3,
816.
(435) Chang, C.-S.; Suen, S.-Y. J. Membr. Sci. 2006, 275, 70.
(436) Cheow, P.-S.; Liu, L.; Toh, C.-S. Surf. Interface Anal. 2007, 39,
601.
(437) O berg, K.; Persson, P.; Shchukarev, A.; Eliassona, B. Thin Solid
Films 2001, 397, 102.
(438) Liakos, I. L.; Newman, R. C.; McAlpine, E.; Alexander, M. R.
Surf. Interface Anal. 2004, 36, 37.
(439) Chen, Y.-F.; Hu, Y.-H.; Chou, Y.-I.; Lai, S.-M.; Wang, C.-C.
Sens. Actuators, B 2010, 145, 575.
(440) Yildirim, O.; Gang, T.; Kinge, S.; Reinhoudt, D. N.; Blank, D.
H. A.; Van der Wiel, W. G.; Rijnders, G.; Huskens, J. Int. J. Mol. Sci.
2010, 11, 1162.
(441) Jani, A. M. M.; Anglin, E. J.; McInnes, S. J. P.; Losic, D.;
Shapter, J. G.; Voelcker, N. H. Chem. Commun. 2009, 3062.
(442) Jani, A. M. M.; Kempson, I. M.; Losic, D.; Voelcker, N. H.
Angew. Chem., Int. Ed. 2010, 49, 7933.
(443) Ko, H.; Tsukruk, V. V. Small 2008, 4, 1980.
(444) Dotzauer, D. M.; Dai, J.; Sun, L.; Bruening, M. L. Nano Lett.
2006, 6, 2268.
(445) He, Q.; Cui, Y.; Ai, S.; Tian, Y.; Li, J. Curr. Opin. Colloid
Interface Sci. 2009, 14, 115.
(446) Lahav, M.; Sehayek, T.; Vaskevich, A.; Rubinstein, I. Angew.
Chem., Int. Ed. 2003, 42, 5576.
(447) Hobler, C.; Bakowsky, U.; Keusgen, M. Phys. Status Solidi A
2010, 207, 872.
(448) Thormann, A.; Teuscher, N.; Pfannmcller, M.; Rothe, U.;
Heilmann, A. Small 2007, 3, 1032.
(449) Lu, Z.; Ruan, W.; Yang, J.; Xu, W.; Zhao, C.; Zhao, B. J. Raman
Spectrosc. 2008, 40, 112.
(450) Ji, N.; Ruan, W.; Wang, C.; Lu, Z.; Zhao, B. Langmuir 2009, 25,
11869.
(451) Sehayek, T.; Lahav, M.; Popovitz-Biro, R.; Vaskevich, A.;
Rubinsein, I. Chem. Mater. 2005, 17, 3743.
(452) Tanvir, S.; Pantigny, J.; Boulnois, P.; Pulvin, S. J. Membr. Sci.
2009, 329, 85.

(453) Swan, E. E. L.; Popat, K. C.; Desai, T. A. Biomaterials 2005, 26,


1969.
(454) Li, P.-F.; Xie, R.; Jiang, J.-C.; Meng, T.; Yang, M.; Ju, X.-J.;
Yang, L.; Chu, L.-Y. J. Membr. Sci. 2009, 337, 310.
(455) Sun, L.; Dai, J.; Baker, G. L.; Bruening, M. L. Chem. Mater.
2006, 18, 4033.
(456) Jain, P.; Sun, L.; Dai, J.; Baker, G. L.; Bruening, M. L.
Biomacromolecules 2007, 8, 3102.
(457) Balachandra, A. M.; Baker, G. L.; Bruening, M. L. J. Membr. Sci.
2003, 227, 1.
(458) Sun, L.; Baker, G. L.; Bruening, M. L. Macromolecules 2005, 38,
2307.
(459) Grajales, S. T.; Dong, X.; Zheng, Y.; Baker, G. L.; Bruening, M.
L. Chem. Mater. 2010, 22, 4026.
(460) Shi, W.; Shen, Y.; Ge, D.; Xue, M.; Cao, H.; Huang, S.; Wang,
J.; Zhang, G.; Zhang, F. J. Membr. Sci. 2008, 325, 801.
(461) Song, C.; Shi, W.; Jiang, H.; Tu, J.; Ge, D. J. Membr. Sci. 2011,
372, 340.
(462) Bruening, M. L.; Dotzauer, D. M.; Jain, P.; Ouyang, L.; Baker,
G. L. Langmuir 2008, 24, 7663.
(463) Decher, G. Science 1997, 277, 1232.
(464) Jin, W.; Toutianoush, A.; Tieke, B. Langmuir 2003, 19, 2550.
(465) Toutianoush, A.; Tieke, B. Mater. Sci. Eng., C 2002, 22, 135.
(466) Peyratout, C. S.; Dahne, L. Angew. Chem., Int. Ed. 2004, 43,
3762.
(467) Balachandra, A. M.; Dai, J.; Bruening, M. L. Macromolecules
2002, 35, 3171.
(468) Liu, X.; Bruening, M. L. Chem. Mater. 2004, 16, 351.
(469) Hong, S. U.; Ouyang, L.; Bruening, M. L. J. Membr. Sci. 2009,
327, 2.
(470) Ouyang, L.; Malaisamy, R.; Bruening, M. L. J. Membr. Sci.
2008, 310, 76.
(471) Dai, J.; Baker, G. L.; Brening, M. L. Anal. Chem. 2006, 78, 135.
(472) Liang, Z.; Susha, A. S.; Yu, A.; Caruso, F. Adv. Mater. 2003, 15,
1849.
(473) Lee, D.; Nolte, A. J.; Kunz, A. L.; Rubner, M. F.; Cohen, R. E. J.
Am. Chem. Soc. 2006, 128, 8521.
(474) Lee, D.; Cohen, R. E.; Rubner, M. F. Langmuir 2007, 23, 123.
(475) Ai, S.; Lu, G.; He, Q.; Li, J. J. Am. Chem. Soc. 2003, 125, 11140.
(476) Feng, C.-L.; Zhong, X.; Steinhart, M.; Caminade, A.-M.;
Majoral, J.-P.; Knoll, W. Adv. Mater. 2007, 19, 1993.
(477) Tian, Y.; He, Q.; Tao, C.; Li, J. Langmuir 2006, 22, 360.
(478) Alem, H.; Blondeau, F.; Glinel, K.; Demoustier-Champagne, S.;
Jonas, A. M. Macromolecules 2007, 40, 3366.
(479) Cho, Y.; Lee, W.; Jhon, Y. K.; Genzer, J.; Char, K. Small 2010,
6, 2683.
(480) Hou, S.; Harrell, C. C.; Trofin, L.; Kohli, P.; Martin, C. R. J.
Am. Chem. Soc. 2004, 126, 5674.
(481) Quinn, J. F.; Johnston, A. P. R.; Such, G. K.; Zelikin, A. N.;
Caruso, F. Chem. Soc. Rev. 2007, 36, 707.
(482) Kongsuphol, P.; Fang, K. B.; Ding, Z. Sens. Actuators, B 2013,
185, 530.
(483) Hirano-Iwata, A.; Taira, T.; Oshima, A.; Kimura, Y.; Niwano,
M. Appl. Phys. Lett. 2010, 96, 213706.
(484) Bally, M.; Bailey, K.; Sugihara, K.; Grieshaber, D.; Voros, J.;
Stadler, a. B. Small 2010, 6, 2481.
(485) Hirano-Iwata, A.; Niwano, M.; Sugawara, M. Trends Anal.
Chem. 2008, 27, 512.
(486) Hennesthal, C.; Steinem, C. J. Am. Chem. Soc. 2000, 122, 8085.
(487) Naumann, R.; Jonczyk, A.; Kopp, R.; van Esch, J.; Ringsdorf,
H.; Knoll, W.; Griiber, P. Angew. Chem., Int. Ed. 1995, 34, 2056.
(488) Cremer, P. S.; Boxer, S. G. J. Phys. Chem. B 1999, 103, 2554.
(489) Radler, J.; Strey, H.; Sackmann, E. Langmuir 1995, 11, 4539.
(490) Sackmann, E. Science 1995, 271, 43.
(491) Sackmann, E.; Tanaka, M. Trends Biotechnol. 2000, 18, 58.
(492) Heyse, S.; Ernst, O. P.; Dienes, Z.; Hofmann, K. P.; Vogel, A.
H. Biochemistry 1998, 37, 507.
(493) Mayer, M.; Kriebel, J. K.; Tosteson, M. T.; Whitesides, G. M.
Biophys. J. 2003, 85, 2684.
7553

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

(530) Mei, X.; Blumin, M.; Kim, D.; Wu, Z.; Ruda, H. E. J. Cryst.
Growth 2003, 251, 253.
(531) Jung, M.; Lee, H. S.; Park, H. L.; Mho, S.-i. Curr. Appl. Phys.
2006, 6S1, e187.
(532) Mei, X.; Blumin, M.; Sun, M.; Kim, D.; Wu, Z. H.; Ruda, H. E.;
Guo, Q. X. Appl. Phys. Lett. 2003, 82, 967.
(533) Chen, Z.; Lei, Y.; Chew, H. G.; Teo, L. W.; Choi, W. K.; Chim,
W. K. J. Cryst. Growth 2004, 268, 560.
(534) Xu, W. L.; Zheng, M. J.; Ding, G. Q.; Shen, W. Z. Chem. Phys.
Lett. 2005, 411, 37.
(535) Guo, Q.; Mei, X.; Ruda, H.; Tanaka, T.; Nishio, M.; Ogawa, H.
Jpn. J. Appl. Phys. 2003, 42, L508.
(536) Masuda, H.; Yasui, K.; Nishio, K. Adv. Mater. 2000, 12, 1031.
(537) Liu, K.; Nogues, J.; Leighton, C.; Masuda, H.; Nishio, K.;
Roshchin, I. V.; Schuller, I. K. Appl. Phys. Lett. 2002, 81, 4434.
(538) Nam, W.; Seo, H.; Park, S. C.; Bae, C. H.; Nam, S. H.; Park, S.
M.; Ha, J. S. Jpn. J. Appl. Phys. 2004, 43, 7793.
(539) Kossyrev, P. A.; Yin, A.; Cloutier, S. G.; Cardimona, D. A.;
Huang, D.; Alsing, P. M.; Xu, J. M. Nano Lett. 2005, 5, 1978.
(540) Lei, Y.; Chim, W.-K. J. Am. Chem. Soc. 2005, 127, 1487.
(541) Gao, X.; Liu, L.; Birajdar, B.; Ziese, M.; Lee, W.; Alexe, M.;
Hesse, D. Adv. Funct. Mater. 2009, 19, 3450.
(542) Lee, W.; Han, H.; Lotnyk, A.; Schubert, M. A.; Senz, S.; Alexe,
M.; Hesse, D.; Baik, S.; Gosele, U. Nat. Nanotechnol. 2008, 3, 402.
(543) Rodriguez, B. J.; Gao, X. S.; Liu, L. F.; Lee, W.; Naumov, I. I.;
Bratkovsky, A. M.; Hesse, D.; Alexe, M. Nano Lett. 2009, 9, 1127.
(544) Kim, Y.; Han, H.; Lee, W.; Baik, S.; Hesse, D.; Alexe, M. Nano
Lett. 2010, 10, 1266.
(545) Han, H.; Park, Y. J.; Baik, S.; Lee, W.; Alexe, M.; Hesse, D.;
Gosele, U. J. Appl. Phys. 2010, 108, 044102.
(546) Han, H.; Kim, Y.; Alexe, M.; Hesse, D.; Lee, W. Adv. Mater.
2011, 23, 4599.
(547) Cheng, G.; Moskovits, M. Adv. Mater. 2002, 14, 1567.
(548) Liang, J.; Chik, H.; Yin, A.; Xu, J. J. Appl. Phys. 2002, 91, 2544.
(549) Liang, J.; Hong, S.-K.; Kouklin, N.; Beresford, B.; Xu, J. M.
Appl. Phys. Lett. 2003, 83, 1752.
(550) Shingubara, S.; Okino, O.; Murakami, Y.; Sakaue, H.; Takahagi,
T. J. Vac. Sci. Technol., B 2001, 19, 1901.
(551) Nakao, M.; Oku, S.; Tamamura, T.; Yasui, K.; Masuda, H. Jpn.
J. Appl. Phys. 1999, 38, 1052.
(552) Guo, Q.; Takaka, T.; Nishio, M.; Ogawa, H.; Mei, X.; Ruda, H.
Jpn. J. Appl. Phys. 2002, 41, L118.
(553) Kouklin, N.; Chik, H.; Liang, J.; Tzolov, M.; Xu, J. M.; Heroux,
J. B.; Wang, W. I. J. Phys. D: Appl. Phys. 2003, 36, 2634.
(554) Liang, J.; Luo, H.; Beresford, R.; Xu, J. Appl. Phys. Lett. 2004,
85, 5974.
(555) Kanamori, Y.; Hane, K.; Sai, H.; Yugami, H. Appl. Phys. Lett.
2001, 78, 142.
(556) Tian, L.; Ram, K. B.; Ahmad, I.; Menon, L.; Holtz, M. J. Appl.
Phys. 2005, 97, 026101.
(557) Masuda, H.; Watanabe, M.; Yasui, K.; Tryk, D.; Rao, T.;
Fujishima, A. Adv. Mater. 2000, 12, 444.
(558) Masuda, H.; Yasui, K.; Watanabe, M.; Nishio, K.; Nakao, M.;
Tamamura, T.; Rao, T. N.; Fujishima, A. Electrochem. Solid-State Lett.
2001, 4, G101.
(559) Honda, K.; Rao, T. N.; Tryk, D. A.; Fujishima, A.; Watanabe,
M.; Yasui, K. J. Electrochem. Soc. 2001, 148, A668.
(560) Honda, K.; Rao, T. N.; Tryk, D. A.; Fujishima, A.; Watanabe,
M.; Yasui, K.; Masuda, H. J. Electrochem. Soc. 2000, 147, 659.
(561) Nakao, M.; Oku, S.; Tanaka, H.; Shibata, Y.; Yokoo, A.;
Tamamura, T.; Masuda, H. Opt. Quantum Electron. 2002, 34, 183.
(562) Wang, Y. D.; Chua, S. J.; Sander, M. S.; Chen, P.; Tripathy, S.;
Fonstad, C. G. Appl. Phys. Lett. 2004, 85, 816.
(563) Hobbs, K. L.; Larson, P. R.; Lian, G. D.; Keay, J. C.; Johnson,
M. B. Nano Lett. 2004, 4, 167.
(564) Lombardi, I.; Hochbaum, A. I.; Yang, P.; Carraro, C.;
Maboudian, R. Chem. Mater. 2006, 18, 988.
(565) Lei, Y.; Yeong, K.-S.; Thong, J. T. L.; Chim, W.-K. Chem.
Mater. 2004, 16, 2757.

(494) Simon, A.; Girard-Egrot, A.; Sauter, F.; Pudda, C.; DHahan, N.
P.; Blum, L.; Chatelain, F.; Fuchs, A. J. Colloid Interface Sci. 2007, 308,
337.
(495) Hennesthal, C.; Drexler, J.; Steinem, C. ChemPhysChem 2002,
10, 885.
(496) Romer, W.; Steinem, C. Biophys. J. 2004, 86, 955.
(497) Schmitt, E. K.; Vrouenraets, M.; Steinem, C. Biophys. J. 2006,
91, 2163.
(498) Schmitt, E. K.; Nurnabi, M.; Bushby, R. J.; Steinem, C. Soft
Matter 2008, 4, 250.
(499) Schmitt, E. K.; Weichbrodt, C.; Steinem, C. Soft Matter 2009,
5, 3347.
(500) Lazzara, T. D.; Carnarius, C.; Kocun, M.; Janshoff, A.; Steinem,
C. ACS Nano 2011, 5, 6935.
(501) Hotta, K.; Yamaguchi, A.; Teramae, N. ACS Nano 2013, 6,
1541.
(502) Smirnov, A. I.; Poluektov, O. G. J. Am. Chem. Soc. 2003, 125,
8434.
(503) Chekmenev, E. Y.; Gorkov, P. L.; Cross, T. A.; Alaouie, A. M.;
Smirnov, A. I. Biophys. J. 2006, 91, 3076.
(504) Li, R.-Q.; Marek, A.; Smirnov, A. I.; Grebel, H. J. Chem. Phys.
2008, 129, 095102.
(505) Deme, B.; Marchal, D. Eur. Biophys. J. 2005, 34, 170.
(506) Parthasarathy, R. V.; Martin, C. R. Nature 1994, 369, 298.
(507) Cai, Z.; Lei, J.; Liang, W.; Menon, V.; Martin, C. R. Chem.
Mater. 1991, 3, 960.
(508) Parthasarathy, R. V.; Martin, C. R. Chem. Mater. 1994, 6, 1627.
(509) Martin, C. R. Acc. Chem. Res. 1995, 28, 61.
(510) Parthasarathy, R. V.; Phani, K. L. N.; Martin, C. R. Adv. Mater.
1995, 7, 896.
(511) Steinhart, M.; Wendorff, J. H.; Greiner, A.; Wehrspohn, R. B.;
Nielsch, K.; Schilling, J.; Choi, J.; Gosele, U. Science 2002, 296, 1997.
(512) de Gennes, P. G. Rev. Mod. Phys. 1985, 57, 827.
(513) Ausserre, D.; Picard, A. M.; Leger, L. Phys. Rev. Lett. 1986, 57,
2671.
(514) Heslot, F.; Cazabat, A. M.; Levinson, P. Phys. Rev. Lett. 1989,
62, 1286.
(515) Steinhart, M.; Wendorff, J. H.; Wehrspohn, R. B.
ChemPhysChem 2003, 4, 1171.
(516) Steinhart, M.; Wehrspohn, R. B.; Gosele, U.; Wendorff, J. H.
Angew. Chem., Int. Ed. 2004, 43, 1334.
(517) Lee, W.; Jin, M.-K.; Yoo, W.-C.; Lee, J.-L. Langmuir 2004, 20,
7665.
(518) Steinhart, M.; Senz, S.; Wehrsphon, R. B.; Gosele, U.;
Wendorff, J. H. Macromolecules 2003, 36, 3646.
(519) Steinhart, M.; Zimmermann, S.; Goring, P.; Schaper, A. K.;
Gosele, U.; Weder, C.; Wndorff, J. H. Nano Lett. 2005, 5, 429.
(520) Steinhart, M.; Murano, S.; Shaper, A. K.; Ogawa, T.; Tsuji, M.;
Gosele, U.; Weder, C.; Wendorff, J. H. Adv. Funct. Mater. 2005, 15,
1656.
(521) Steinhart, M.; Jia, Z.; Schaper, A. K.; Wehrspohn, R. B.; Gosele,
U.; Wendorff, J. H. Adv. Mater. 2003, 15, 706.
(522) Luo, Y.; Lee, S. K.; Hofmeister, H.; Steinhart, M.; Gosele, U.
Nano Lett. 2004, 4, 143.
(523) Lei, Y.; Cai, W.; Wilde, G. Prog. Mater. Sci. 2007, 52, 465.
(524) Sulka, G. D.; Zaraska, L.; Stpniowski, W. J. Anodic Porous
Alumina as a Template for Nanofabrication. In Encylopedia of
Nanoscience and Nanotechnology, 2nd ed.; Nalwa, H. S., Ed.; American
Scientic Publishers: Califonia, 2011; Vol. 11, pp 261349.
(525) Meng, G.; Yanagida, T.; Nagashima, K.; Yanagishita, T.; Kanai,
M.; Oka, K.; Klamchuen, A.; Rohong, S.; Horprathum, M.; Xu, B.;
Zhuge, F.; He, Y.; Masuda, H.; Kawai, T. RSC Adv. 2012, 2, 10618.
(526) Mei, X.; Kim, D.; Ruda, H. E.; Guo, Q. X. Appl. Phys. Lett.
2003, 81, 361.
(527) Lei, Y.; Chim, W.-K. Chem. Mater. 2005, 17, 580.
(528) Lei, Y.; Chim, W. K.; Sun, H. P.; Wilde, G. Appl. Phys. Lett.
2005, 86, 103106.
(529) Cloutier, S. G.; Guico, R. S.; Xu, J. M. Appl. Phys. Lett. 2005,
87, 222104.
7554

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

(566) Wu, Z. H.; Mei, X. Y.; Kim, D.; Blumin, M.; Ruda, H. E. Appl.
Phys. Lett. 2002, 81, 5177.
(567) Chik, H.; Liang, J.; Cloutier, S. G.; Kouklin, N.; Xu, J. M. Appl.
Phys. Lett. 2004, 84, 3376.
(568) Lee, W.; Alexe, M.; Nielsch, K.; Gosele, U. Chem. Mater. 2005,
17, 3325.
(569) Fan, H. J.; Lee, W.; Hauschild, R.; Alexe, M.; Le Rhun, G.;
Scholz, R.; Dadgar, A.; Nielsch, K.; Kalt, H.; Krost, A.; Zacharias, M.;
Gosele, U. Small 2006, 2, 561.
(570) Fan, H. J.; Lee, W.; Scholz, R.; Dadgar, A.; Krost, A.; Nielsch,
K.; Zacharias, M. Nanotechnology 2005, 16, 913.
(571) Lee, S. K.; Lee, W.; Alexe, M.; Nielsch, K.; Hesse, D.; Gosele,
U. Appl. Phys. Lett. 2005, 86, 152906.
(572) Lee, S. K.; Hesse, D.; Alexe, M.; Lee, W.; Nielsch, K.; Gosele,
U. J. Appl. Phys. 2005, 98, 124302.
(573) Park, S.-J.; Han, H.; Rhu, H.; Baik, S.; Lee, W. J. Mater. Chem.
C 2013, 1, 5330.
(574) Han, H.; Kim, J.; Shin, H. S.; Song, J. Y.; Lee, W. Adv. Mater.
2012, 24, 2284.
(575) Kim, J.; Han, H.; Kim, Y. H.; Choi, S.-H.; Kim, J.-C.; Lee, W.
ACS Nano 2011, 5, 3222.
(576) Kim, J.; Kim, Y. H.; Choi, S.-H.; Lee, W. ACS Nano 2011, 5,
5242.
(577) Kim, J.; Rhu, H.; Lee, W. J. Mater. Chem. 2011, 21, 15889.
(578) Huang, Z.; Zhang, X.; Reiche, M.; Liu, L.; Lee, W.; Shimizu, T.;
Senz, S.; Gosele, U. Nano Lett. 2008, 8, 3046.
(579) Hulteen, J. C.; Martin, C. R. Template synthesis of
nanoparticles in nanoporous membranes. In Nanoparticles and
Nanostructured Films: Preparation, Characterization and Applications;
Fendler, J. H., Ed.; Wiley-VCH: New York, 1998; pp 235262.
(580) Shingubara, S. J. Nanopart. Res. 2003, 5, 17.
(581) Kyotani, T.; Tsai, L.-f.; Tomita, A. Chem. Mater. 1996, 8, 2109.
(582) Che, G.; Lakshmi, B. B.; Martin, C. R.; Fisher, E. R. Chem.
Mater. 1998, 10, 260.
(583) Li, J.; Papadopoulos, C.; Xu, J. M. Appl. Phys. Lett. 1999, 75,
367.
(584) Suh, J. S.; Lee, J. S. Appl. Phys. Lett. 1999, 75, 2047.
(585) Jeong, S.-H.; Hwang, H.-Y.; Lee, K.-H.; Jeong, Y. Appl. Phys.
Lett. 2001, 78, 2052.
(586) Yanagishita, T.; Sasaki, M.; Nishio, K.; Masuda, H. Adv. Mater.
2004, 16, 429.
(587) Meng, G.; Jung, Y. J.; Cao, A.; Vajtai, R.; Ajayan, P. M. Proc.
Natl. Acad. Sci. U.S.A. 2005, 102, 7074.
(588) Chen, Q.-L.; Xue, K.-H.; Shen, W.; Tao, F.-F.; Yin, S.-Y.; Xu,
W. Electrochim. Acta 2004, 49, 4157.
(589) Park, S.; Kim, Y.-S.; Kim, W. B.; Jon, S. Nano Lett. 2009, 9,
1325.
(590) Schmidt, V.; Wittemann, J. V.; Gosele, U. Chem. Rev. 2010,
110, 361.
(591) Schmidt, V.; Senz, S.; Gosele, U. Nano Lett. 2005, 5, 931.
(592) Shimizu, T.; Xie, T.; Nishikawa, J.; Shingubara, S.; Senz, S.;
Gosele, U. Adv. Mater. 2007, 19, 917.
(593) Gorisse, T.; Dupre, L.; Gentile, P.; Martin, M.; Zelsmann, M.;
Buttard, D. Nanoscale Res. Lett. 2013, 8, 287.
(594) Cheng, G. S.; Zhang, L. D.; Zhu, Y.; Fei, G. T.; Li, L.; Mo, C.
M.; Mao, Y. Q. Appl. Phys. Lett. 1999, 75, 2455.
(595) Zhang, J.; Zhang, L. D.; Wang, X. F.; Liang, C. H.; Peng, X. S.;
Wang, Y. W. J. Appl. Phys. 2001, 115, 5714.
(596) Jung, W.-G.; Jung, S.-H.; Kung, P.; Razeghi, M. Nanotechnology
2006, 17, 54.
(597) Li, X.; Meng, G.; Xu, Q.; Kong, M.; Zhu, X.; Chu, Z.; Li, A.-P.
Nano Lett. 2011, 11, 1704.
(598) Li, X.; Meng, G.; Qin, S.; Xu, Q.; Chu, Z.; Zhu, X.; Kong, M.;
Li, A.-P. ACS Nano 2012, 6, 831.
(599) Shen, X.-P.; Yuan, A.-H.; Wang, F.; Hong, J.-M.; Xu, Z. Solid
State Commun. 2005, 133, 19.
(600) Shen, X.-P.; Liu, J.-J.; Fan, X.; Jiang, Y.; Hong, J.-M.; Xu, Z. J.
Cryst. Growth 2005, 276, 471.

(601) Shen, X.-P.; Han, M.; Hong, J.-M.; Xue, Z.; Xu, Z. Chem. Vap.
Deposition 2005, 11, 250.
(602) Fan, Z.; Razavi, H.; Do, J.-w.; Moriwaki, A.; Ergen, O.; Chueh,
Y.-L.; Leu, P. W.; Ho, J. C.; Takahashi, T.; Reichertz, L. A.; Neale, S.;
Yu, K.; Wu, M.; Ager, J. W.; Javey, A. Nat. Mater. 2009, 8, 648.
(603) Fan, Z.; Dutta, D.; Chien, C.-J.; Chen, H.-Y.; Brown, E. C.
Appl. Phys. Lett. 2006, 89, 213110.
(604) Bae, C.; Yoo, H.; Kim, S.; Lee, K.; Kim, J.; Sung, M. M.; Shin,
H. Chem. Mater. 2008, 20, 756.
(605) Detavernier, C.; Dendooven, J.; Pulinthanathu Sree, S.;
Ludwig, K. F.; Martens, J. A. Chem. Soc. Rev. 2011, 40, 5242.
(606) George, S. M. Chem. Rev. 2009, 110, 111.
(607) Knez, M.; Nielsch, K.; Niinisto, L. Adv. Mater. 2007, 19, 3425.
(608) Marichy, C.; Bechelany, M.; Pinna, N. Adv. Mater. 2012, 24,
1017.
(609) Narayan, R.; Monteiro-Riviere, N. A.; Brigmon, R. L.; Pellin,
M. J.; Elam, J. W. JOM 2009, 61, 12.
(610) Ott, A. W.; Klaus, J. W.; Johnson, J. M.; George, S. M.;
McCarley, K. C.; Way, J. D. Chem. Mater. 1997, 9, 707.
(611) Chen, P.; Mitsui, T.; Farmer, D. B.; Golovchenko, J.; Gordon,
R. G.; Branton, D. Nano Lett. 2004, 4, 1333.
(612) Velleman, L.; Triani, G.; Evans, P. J.; Shapter, J. G.; Losic, D.
Microporous Mesoporous Mater. 2009, 126, 87.
(613) Kim, W.-H.; Park, S.-J.; Son, J.-Y.; Kim, H. Nanotechnology
2008, 19, 045302.
(614) Chang, Y.-H.; Wang, S.-M.; Liu, C.-M.; Chen, C. J. Electrochem.
Soc. 2010, 157, K236.
(615) Yang, C.-J.; Wang, S.-M.; Liang, S.-W.; Chang, Y.-H.; Chen, C.;
Shieh, J.-M. Appl. Phys. Lett. 2007, 90, 033104.
(616) Bachmann, J.; Jing, J.; Knez, M.; Barth, S.; Shen, H.; Mathur,
S.; Gosele, U.; Nielsch, K. J. Am. Chem. Soc. 2007, 129, 9554.
(617) Bae, C.; Yoon, Y.; Yoo, H.; Han, D.; Cho, J.; Lee, B. H.; Sung,
M. M.; Lee, M.; Kim, J.; Shin, H. Chem. Mater. 2009, 21, 2574.
(618) Bae, C.; Zierold, R.; Montero Moreno, J. M.; Kim, H.; Shin,
H.; Bachmann, J.; Nielsch, K. J. Mater. Chem. C 2013, 1, 621.
(619) Comstock, D. J.; Christensen, S. T.; Elam, J. W.; Pellin, M. J.;
Hersam, M. C. Adv. Funct. Mater. 2010, 20, 3099.
(620) Gaspard, P.; Hithesh, K. G.; Goran, S.; Wouter van der, W.;
Niclas, R. Nanotechnology 2013, 24, 015602.
(621) Gu, D.; Baumgart, H.; Abdel-Fattah, T. M.; Namkoong, G.
ACS Nano 2010, 4, 753.
(622) Marianna, K.; Emma, H.; Viljami, P.; Mikko, R.; Markku, L.
Nanotechnology 2010, 21, 035301.
(623) Pitzschel, K.; Moreno, J. M. M.; Escrig, J.; Albrecht, O.;
Nielsch, K.; Bachmann, J. ACS Nano 2009, 3, 3463.
(624) Shin, H.; Jeong, D. K.; Lee, J.; Sung, M. M.; Kim, J. Adv. Mater.
2004, 16, 1197.
(625) Yoon, J.; Kim, S.; No, K. Mater. Lett. 2012, 87, 124.
(626) Martinson, A. B. F.; Elam, J. W.; Hupp, J. T.; Pellin, M. J. Nano
Lett. 2007, 7, 2183.
(627) Sander, M. S.; Cote, M. J.; Gu, W.; Kile, B. M.; Tripp, C. P.
Adv. Mater. 2004, 16, 2052.
(628) Bachmann, J.; Zierold, R.; Chong, Y. T.; Hauert, R.; Sturm, C.;
Schmidt-Grund, R.; Rheinlander, B.; Grundmann, M.; Gosele, U.;
Nielsch, K. Angew. Chem., Int. Ed. 2008, 47, 6177.
(629) Bae, C.; Yoon, Y.; Yoon, W.-S.; Moon, J.; Kim, J.; Shin, H. ACS
Appl. Mater. Interfaces 2010, 2, 1581.
(630) Chen, X.; Pomerantseva, E.; Banerjee, P.; Gregorczyk, K.;
Ghodssi, R.; Rubloff, G. Chem. Mater. 2012, 24, 1255.
(631) Panda, S. K.; Yoon, Y.; Jung, H. S.; Yoon, W.-S.; Shin, H. J.
Power Sources 2012, 204, 162.
(632) Banerjee, P.; Perez, I.; Henn-Lecordier, L.; Lee, S. B.; Rubloff,
G. W. Nat. Nanotechnol. 2009, 4, 292.
(633) Du, X.; George, S. M. Sens. Actuators, B 2008, 135, 152.
(634) Moreno i Codinachs, L.; Birkenstock, C.; Garma, T.; Zierold,
R.; Bachmann, J.; Nielsch, K.; Schoning, M. J.; Fontcuberta i Morral,
A. Phys. Status Solidi A 2009, 206, 435.
(635) Lee, J.; Kim, D. H.; Hong, S.-H.; Jho, J. Y. Sens. Actuators, B
2011, 160, 1494.
7555

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

Chemical Reviews

Review

(636) Liu, C.-M.; Chen, C.; Cheng, H.-E. Electrochem. Solid-State Lett.
2011, 14, K33.
(637) Cameron, M. A.; Gartland, I. P.; Smith, J. A.; Diaz, S. F.;
George, S. M. Langmuir 2000, 16, 7435.
(638) Tran, T. H. Y.; Haije, W. G.; Longo, V.; Kessels, W. M. M.;
Schoonman, J. J. Membr. Sci. 2011, 378, 438.
(639) Pellin, M. J.; Stair, P. C.; Xiong, G.; Elam, J. W.; Birrell, J.;
Curtiss, L.; George, S. M.; Han, C. Y.; Iton, L.; Kung, H.; Kung, M.;
Wang, H. H. Catal. Lett. 2005, 102, 127.
(640) Stair, P. C.; Marshall, C.; Xiong, G.; Feng, H.; Pellin, M. J.;
Elam, J. W.; Curtiss, L.; Iton, L.; Kung, H.; Kung, M.; Wang, H. H.
Top. Catal. 2006, 39, 181.
(641) Kemell, M.; Pore, V.; Tupala, J.; Ritala, M.; Leskela, M. Chem.
Mater. 2007, 19, 1816.
(642) Tan, L. K.; Kumar, M. K.; An, W. W.; Gao, H. ACS Appl.
Mater. Interfaces 2010, 2, 498.
(643) Lee, J. H.; Wu, J. H.; Liu, H. L.; Cho, J. U.; Cho, M. K.; An, B.
H.; Min, J. H.; Noh, S. J.; Kim, Y. K. Angew. Chem., Int. Ed. 2007, 46,
3663.

7556

dx.doi.org/10.1021/cr500002z | Chem. Rev. 2014, 114, 74877556

You might also like