1

One-dimensional ferroelectrics: Nanowires and nanotubes

M. Alexe and D. Hesse

Max Planck Institute of Microstructure Physics, D-06120 Halle, Germany

Abstract

An overview of some general features of the fabrication of one-dimensional (1D)

ferroelectric structures is given. Included are positive and negative template-assisted

fabrication methods, along with some examples and properties of the obtained

ferroelectric nanowires and nanotubes, in particular from the group of the authors. The

contribution of Professor James F. Scott to the development of the field of nano-

ferroelectrics at MPI-Halle is briefly highlighted.

2

Considerable attention has been paid in the last 15 years to nanowires and nanorods as

one-dimensional nano-systems. Most of the effort was directed towards the synthesis of

one-dimensional nanostructures comprising carbon nanotubes1, semiconductor, metallic,

and simple oxide nanowires.2,3,4,5 A somewhat smaller amount of work has been

performed to synthesize one-dimensional nanostructures from functional complex oxide

materials. This is most probably due to the chemical and structural complexity of these

materials, such as BaTiO3 (BTO) or PbTiO3 (PTO) which are perovskite oxides, or

YBa2Cu3O7-δ - a high-Tc superconductor, in all of which stoichiometry and structure play

a most important role in achieving functional properties.

At the general level, there are two main methods to fabricate nanoscopic and

mesoscopic one-dimensional structures: self-assembly, or bottom-up methods, used for

instance by Urban et al. to fabricate BTO ferroelectric nanowires 6, and template-

mediated approaches described, e.g., by Martin et al.7 In principle, self-assembly

fabrication methods offer a higher crystalline quality and smaller dimensions mainly due

to the bottom-up building of the structure using atoms and molecules. On the other side,

the template-mediated methods offer a number of advantages in terms of tunability of the

characteristic dimensions, as well as a better control of stoichiometry and a higher

flexibility. Most of the template methods to obtain tubular or rod-like structures use

porous templates in conjunction with infiltration or conformal deposition to partially or

entirely fill the pores. This paper will address some general features of the fabrication of

one-dimensional (1D) ferroelectric structures along with some examples and properties of

obtained ferroelectric nanowires and nanotubes, in particular from the group of the

authors. For more comprehensive reviews on one-dimensional ferroelectric and

3

perovskite nanostructures, see, e.g., refs. 8,9 Before turning to 1D ferroelectrics, however,

a historical remark referring to the significant role James F. Scott played in initiating the

studies of nanosize ferroelectrics will be given.

1. Historical remark

It was as early as in 1997 that Professor James F. Scott joined the Max Planck

Institute of Microstructure Physics (MPI) in Halle, Germany, for a half-a-year research

stay within the framework of a prestigious Alexander von Humboldt Research Award.

At that time the authors of the present paper were coworkers of Professor Ulrich Gösele's

department at MPI, working with ferroelectric thin films and heterostructures, mainly

grown on silicon wafers. This research had been initiated by Gösele who had a strong

background in silicon materials science and microelectronics, and who clearly saw the

promising perspectives of high-k and ferroelectric complex-oxide materials for the future

downsizing of silicon-based thin-film devices. Jim Scott brought along not only an

important experience in ferroelectric thin films and integration of them into silicon

technology, but, most importantly, his joy of discovery and open mind. Soon after his

arrival at Halle, he immediately noticed that the department had not only state-of-the art

deposition equipment available (like MBE, PLD, MOD, etc.), but also quite some

equipment to structure and analyze thin films down to the nanosize region as, e.g.,

electron-beam lithography, reactive ion etching, as well as powerful infrastructure and

knowledge in electron microscopy. In the discussions with the present authors, Professor

Scott turned their attention to the fact that – different from the semiconductor area, where

the micro and nano aspects had already found considerable attention – ferroelectric

4

nanostructures had not yet been in the proper focus of international research at that time.

He suggested moving into this direction, and – apart from participating in the running

activities of the group10,11 – he actively took part in the first steps done by the Halle

group into the nano area. The title lines of the first three papers12,13,14 reporting on

corresponding results of this joint activity in 1998 and 1999, clearly indicate the direction

chosen: „Self-patterning nano-electrodes on ferroelectric thin films for gigabit memory

applications“ (Ref. 12), „Nano-phase SBT-family ferroelectric memories“ (Ref. 13), and

“Switching properties of self-assembled ferroelectric memory cells” (Ref. 14), with SBT

meaning SrBi2Ta2O9. Whereas that work was related to self-assembled nanostructures, i.e.

a bottom-up path, the subsequent activities of the group were also focussed on the top-

down approach, viz. the preparation of ferroelectric nanosize islands by help of electron-

beam direct writing, and the investigation of their structural and physical properties15,16.

Later, among the various activities of the group related to the preparation and

investigation of many different ferroelectric nanostructures (see, e.g., Refs. 17,18,19,20)

there were also various one-dimensional ferroelectric nanostructures as described below.

The cooperation with Jim Scott continues until present days addressing different topics

such as nanotubes, which are within the scope of the present paper, or vortex structures.21

The present authors acknowledge with great respect and gratefulness the impact received

by Professor James F. Scott.

2. Ferroelectric nanowires

A good example for the growth of perovskite nanowires using bottom-up

fabrication methods is the preparation of BaTiO3 and SrTiO3 nanowires by Park's group

5

at Harvard University. The synthesis of perovskite nanowires is performed by solution-

phase decomposition of complex alkoxide precursors in the presence of coordinating

ligands. 6,22 In a typical reaction, an excess of H2O2 was added at high temperature

(100°C) to a heptadecane solution containing the Ba-Ti or Sr-Ti bimetallic alkoxide and

oleic acid. By heating up the resulting solution to 280°C for 6 hours, anisotropic

nanorods grow most likely due to decomposition and crystallization in an inverse micelle

medium formed by the precursor and the oleic acid. The reaction yields well-isolated

nanorods with diameters varying from 6 to 30 nm and lengths of over 10 µm.

Ferroelectric domains as small as 10 nm in diameter have been written on a single wire

using ultra-high vacuum electric force microscopy (EFM), i.e. by applying an external

field via a conductive tip.

More recent examples for the growth and investigation of ferroelectric nanowires

comprise PbTiO323, KNbO3

24, BaTiO325

, and lead-free bismuth-based complex

perovskites26. It should be noticed that also theorists have discovered ferroelectric

nanowires as an attractive research subject, see, e.g. refs. 27, 28, 29, with the latter

reference also extending into the area of multiferroic nanowires.

3. Ferroelectric nanotubes

From fundamental physics as well as from the application point of view the tubular

geometry, respectively nanotubes, might be more interesting than the simple wire

geometry. The intricate tubular geometry along with a high aspect ratio and one

additional variable parameter, i.e. the wall thickness, as well as the larger surface-to-bulk

6

ratio might influence the ferroelectricity in such structures. There are fundamental

questions that immediately arise:

− Is the polarization direction influenced by the geometry?

− Which would be the aspect ratio at which the surface effects would become

important?

− If the polarization is perpendicular to the tube axis, what would be the

ferroelectric domain pattern?

− In which way the obvious non-planar geometry and stress distribution influence

the polarization switching?

The most interesting aspect of the nanotube system is the characteristic geometry,

viz. one dimension that can extend into microscopic sizes along with the toroid-type

cross-section, which in certain cases might have the aspect ratio of a thin film. This

abnormal geometry correlated with the actual crystalline orientation will define the

ferroelectric properties of the tube in terms of polarization orientation and switching

ability. It has to be stressed that the "top" and the "bottom" surface of that "thin film", i.e.

the outer and inner surfaces of the tube shell, have different curvatures. This might not be

a real problem in the case of bulk tubes, but in the nanotube case this might generate a

highly non-uniform field. To answer the above questions properly it is necessary to

handle proper fabrication processes which allow an easy and convenient way to fabricate

nanotubes of different materials, geometries, and sizes. In the following we will

summarize fabrications methods of ferroelectric nanotubes available up to the date.

7

3.1. Hydrothermal synthesis

The hydrothermal method to obtain ferroelectric nanotubes follows the same general

route of hydrothermal synthesis that allows the preparation of ferroelectric perovskites at

low temperatures, i.e. the reaction of oxide or hydroxide precursors in an aqueous

solution under controlled pH value and temperature to obtain an ABO3-type compound.

One strategy to obtain perovskite nanotubes is to use simple-oxide nanotubes as bona fide

precursor and to let these react, under hydrothermal conditions, with a compound which

contains the second cation. For a ferroelectric perovskite a good precursor is TiO2 which

allows to obtain the important ferroelectric perovskite oxides such as ATiO3, where A can

be Ba, Sr, Pb, etc. Using TiO2 nanotubes rather than "normal" crystalline TiO2 for a

reaction in a hydrothermal environment might be an easy route to fabricate perovskite

nanotubes. There are several ways to achieve TiO2 nanotubes. One of these uses a simple

hydrothermal treatment of crystalline TiO2 with an NaOH aqueous solution.30 Another

approach uses a nanocrystalline atanase phase, which can be easily obtained by

hydrolysis of titanium alkoxide. Titania nanotubes down to 10 nm diameter can be

obtained by hydrothermal reaction of this nanocrystalline phase in an NaOH solution at

about 150°C. TiO2 nanotubes can also be obtained by rolling up crystalline TiO2 sheets

exfoliated from TiO2 crystals during hydrothermal annealing as it is proposed by Yao et

al.31 To obtain perovskite nanotubes one can simply let react these titania nanotubes with

Ba(OH)2 and/or SrCl2 to obtain (Ba,Sr)TiO3.32 The hydrothermal reaction is performed at

ambient pressure by refluxing the solution for a period of 20 to 60 h. BaTiO3 and SrTiO3

nanotubes with 8 to 15 nm outer diameter, 4 to 7 nm inner diameter, and several hundred

nanometer length were obtained. In a similar way, PbTiO3 nanotubes were recently

8

prepared by a reaction of prefabricated TiO2 nanotubes with PbO vapour at between 300

and 600 °C.33

An alternative approach is proposed by Padture and Wei.34 The starting material is

self-organized TiO2 nanotubes obtained by anodization of Ti metal sheets. The

anodization process is similar to the one used to obtain ordered porous alumina.35,36,37

Briefly, a polished Ti foil is anodized in an aqueous HF solution at 16°C by applying a

direct voltage between the Ti foil and a Pt cathode.38 A honeycomb-type array of

amorphous TiO2 nanotubes of about 100 nm diameter and 200-300 nm length is obtained.

These nanotubes can lateron be reacted with Ba(OH)2 under hydrothermal conditions (at

a pH value of 13.2 and a temperature of 200°C for 80 min) to obtain an array of

polycrystalline nanotubes. Remarkable is the preservation of the nanotubular shape after

the reaction of the amorphous titania with the Ba ions and the transformation of the

initially amorphous walls into polycrystalline BaTiO3 or (Ba,Sr)TiO3.39 The above

fabrication method along with the methods to fabricate large-area, perfectly ordered

nanoporous alumina arrays 40 might develop into an interesting way to fabricate well-

ordered nanoporous ferroelectrics, regular composite materials, ordered nanotubes, etc. -

in other words a new class of materials.

3.2. Template-based synthesis

A general approach to fabricate nanoscale structures of complex materials is to use

existing structures that are easily obtained from simple materials such as silicon or

aluminium, and to deposit, react and/or etch them in order to obtain new materials which

would keep the geometry of the initial structure, the latter usually being called "template".

In other words, the existing nanoscale materials are used as skeleton for the fabrication of

9

new structures with new materials that are more complex in structure and eventually rich

in functional properties.

Any adroit technique is here allowed. Any material from well-known silicon to

biological materials such as viruses41 can be suitable to play the template role. Generally,

the initial template can be negative or positive, i.e. holes in a substrate, or wires grown on

a substrate, respectively. The fabrication process shown below comprises three different

stages:

− Template fabrication

− Coating of the template

− Crystallization annealing

− Etching away the host template

3.2.1. Negative (porous) templates

There are two major templates used for nano-fabrication, viz. macroporous silicon

and nanoporous alumina. Different approaches are being used to obtain the templates,

depending on the material of choice. For instance, for the preparation of macroporous

silicon, which is one of the most used templates due to various reasons (including the

availability of high-quality silicon and the eventual compatibility with silicon

microelectronic technologies, as well as possible applications in photonics, e.g., as bi-

dimensional photonic band-gap material), there is now a relatively well-established

technology, and good reviews are published.42

Template fabrication from macroporous silicon is a different field and we will not

detail this, but we can briefly say that a careful control of the etching parameters can

make the silicon pores absolutely straight up to a depth of several hundreds of microns or

10

can modulate or tune the pore diameter in a desired way as it is shown in Figure 1 (from

ref. 43).

Figure 1: Diameter-modulated pores etched into silicon (left), and a corresponding gold

wire obtained by electrochemical deposition (right).43

The pore diameter in macroporous silicon can vary from 0.4 µm up to 2-3 µm and

the depth can range from several microns up to 250-300 µm. The main advantage of

macroporous silicon is the possibility of wafer-scale processing. Wafers of four-inch

diameter are nowadays standard in macroporous silicon processing, yielding perfect pore

arrays over the whole wafer.

Nanoporous alumina is as well an established template used in the fabrication of

one-dimensional nanosize structures made out of different materials including multi-

segment nanotubes. Nanoporous alumina is obtained by anodization of an aluminium foil

using different electrolytes. The anodization of aluminium has been studied for a century

now, but only in the last 15 years - after Masuda and Fukuda 35 discovered that the pores

are self-assemblying into ordered arrays after long anodization times - it became really

intensively studied and applied in photonic crystals and other nanotechnology-related

11

fields. Self-ordered alumina pores are arranged in a hexagonal pattern with the lattice

constant varying from 60 nm to 500 nm, depending on the actual etching conditions. The

size of the self-organized domains is in the range of few microns. For example, Choi et

al.40 using an imprint technique to induce pore nucleation in the form of a regular array

have successfully obtained large–area (in the centimeter range) single domain, well-

ordered nanoporous alumina. By etching away the aluminium substrate, a nanoporous

Al2O3 membrane with straight and well-aligned pores from the top to the bottom can be

relatively easily obtained.

It is important that these porous materials, macroporous silicon and nanoporous

alumina, can act as templates to fabricate one-dimensional functional materials. Taken

together, these two types of templates offer a wide range of diameters ranging from 40

nm to several microns and a high aspect ratio, sometimes reaching several orders of

magnitude.

3.2.2. Nanotubes and nano-shell tubes obtained by infiltration

Tubes of different materials and dimensions can be relatively easily obtained by

infiltration of porous substrates such as porous silicon, nanoporous alumina, or other

porous materials, with polymers or other precursor solutions. For instance, polymer

nanotubes have been fabricated by infiltration of macroporous silicon with a polymer

material at the glass transition temperature. The infiltration of the high-aspect ratio

porous template is performed by wetting the inner surface of the silicon pores with a

polymer.44

Inorganic nanotubes or nanowires are similarly obtained, if a precursor solution is

used to infiltrate the template, followed by an appropriate thermal annealing. In such a

12

way, nano-shell tubes of BaTiO3, Pb(Zr,Ti)O3 (PZT), and SrBi2Ta2O9 (using a “mist”

solution deposition approach) and corresponding arrays were fabricated. 45,46 Figure 2

shows an array of nano-shell tubes obtained by partially etching the host silicon template

away, and also a single wire showing the high aspect ratio.

Figure 2: Array of BaTiO3 nano-shell tubes obtained by partially etching the host silicon

template away (left), and a single BaTiO3 tube arranged on a silicon wafer (right).45

A single-step infiltration yields tubes with about 90 nm wall thickness. The outer

diameter and the length correspond to the template, i.e. 400 nm and 150 µm, respectively.

Piezoelectric properties and ferroelectric switching of PZT and BTO nanotubes were

measured by PFM after they had been arranged onto a Pt-coated silicon substrate.

Individual ferroelectric tubes of PZT and BTO were probed by a conductive tip and

electrically characterized by measuring the local piezoelectric hysteresis. The as-prepared

nanotubes showed only weak ferroelectric properties; however by a one-hour annealing

at 700°C the switching properties were significantly improved, resulting in rectangular

hysteresis loops with sharp ferroelectric switching at a coercive voltage of about 2 V (see

Figure 3).

13

-10 -5 0 5 10

-100.0

-50.0

0.0

50.0

100.0

d eff (

pm/V

)Applied Bias (V)

Figure 3: Piezoresponse hysteresis loop of a single PZT nanoshell tube after

annealing.

The effective remanent piezoelectric coefficient was estimated to about 90 pm/V

which is higher than in thin films. Morozovska et al. have theoretically shown that the

transition temperature could be higher in long ferroelectric nanotubes than the one of the

bulk material for a negative electrostriction coefficient, and predicted preservation and

enhancement of the polarization.47 The possible reason of the enhancement of the polar

properties in these confined ferroelectric nanotubes and nanowires is the radial stress

coupled with polarization via the electrostriction effect under the decrease of the

depolarization field for long cylindrical nanoparticles.

Alternative solution-based deposition methods might also be used in uniform

coating of the template. Morrison et al. have used “mist” (liquid-source) deposition and

similar macroporous silicon templates to obtain SrBi2Ta2O9 nano-shell tubes 46.

Alternatively, nanoporous alumina templates and a similar infiltration process can be

used to obtain either nanotubes 48 or nanowires 49 depending on the precursor solution

wetting properties and infiltration time.50

14

Various physical properties of negative template-made ferroelectric nanotubes are

now under investigation worldwide. For example, mechanical fracture in relation to

domain structure33, resistive switching51, and retention-loss dynamics52 have recently

been investigated on PZT and PbTiO3 nanotubes, respectively, made by applying

nanoporous alumina templates.

3.2.3. Positive templates: nanowires

However, the negative-template approach described above has a number of drawbacks

when real applications, mostly in microelectronics, are envisaged. One of the most

serious ones is the relatively difficult control of the wall thickness of the tube.

Additionally, real devices need electrodes on both inner and outer sides. This would

require several precursor infiltrations of the porous templates in the above process. These

turned out to be more difficult and more laborious than a single-step infiltration.

An alternative approach to the fabrication of one-dimensional nanoscale

ferroelectric structures has been proposed a few years ago.53 Instead of porous templates,

positive templates, such as epitaxially grown nanowires, are used. One layer or many

layers can be easily deposited on the nanowires by conventional vapor deposition

methods, as typically used in microelectronics industry. Finally, tubular structures can be

obtained by selectively etching the initial nanowires away. An important advantage

offered by the vapor deposition methods compared to the infiltration methods is a better

control of the wall thickness due to the inherently better control of the film thickness in

vapor deposition processes. For many reasons, including geometric and hydrodynamic

reasons, it is much easier to perform vapor deposition on skyscraper-type structures than

into deep trenches.

15

Multi-wall nanotubes have been prepared using two types of templates, i.e. silicon

nanowires and ZnO nanowires (Figures 4 and 5). The silicon nanowires (SiNW) used

have an average diameter ranging from 20 nm to 50 nm and a length of 200 nm up to 600

nm, whereas ZnO nanowires have a hexagonal cross-section with an equivalent diameter

of 100 nm up to 150 nm and a length in the micron range. Details on nanowire template

fabrication were published elsewhere.54,55 Platinum and SrRuO3 were used as metals and

BaTiO3 and PZT as ferroelectric materials in a sequential deposition to obtain metal-

ferroelectric-metal (MFM) structures.

Figure 4: SEM image of a cross-section sample prepared by focused ion beam

thinning of a Pt-PZT-Pt tri-layer structure deposited on a silicon nanowire template. The

inset shows a zoom-in image of a coated single Si nanowire.

16

Figure 5: SEM image of a ZnO nanowire template covered with a SrRuO3(10

nm)/PZT(60 nm) bi-layer film (left); top-view of a PZT nanotube obtained by etching the

silicon template away (right).

Ferroelectric switching of a single PZT/Pt nanotube of about 250 nm outer diameter

was measured by piezoresponse scanning probe microscopy (PFM). Hysteresis loops

acquired on a tube as well as on the free surface of a ferroelectric film similarly processed

as the tubes are shown in Figure 6.

-6 -4 -2 0 2 4 6

-20

-10

0

10

20

30

40

50

nanotube film

d zz (p

m/V

)

Applied Bias (V)

Figure 6: Piezoelectric hysteresis loop acquired on a PZT/Pt nanotube arranged on

a Pt-coated Si wafer (triangular dots) and on a 60 nm thick PZT film deposited on a Pt-

coated Si wafer and processed in the same run as the nanotubes (square dots).53

17

The hysteresis loop shows relatively good switching properties of the ferroelectric

thin film, whereas the effective piezoelectric coefficient is about ten times lower

compared to the values acquired using the same measurement setup on high-quality

epitaxial films. 56 A lower quality of the polycrystalline ferroelectric layer due mostly to

the low deposition temperature and the thickness below 50 nm might be responsible for

these low values of the effective piezoelectric coefficient. The cylindrical geometry of the

tube, in which the inner electrode is at a floating potential, gives a relatively intricate

field distribution within the ferroelectric shell. This effect along with the mechanical

boundary conditions increases the effective piezoelectric coefficient as well as the

measurement noise, deteriorating the shape of the hysteresis loop. Nevertheless, a

switching process together with a strong imprint in the positive direction is confirmed.

The fabrication method based on nanowires is appropriate to fabricate three-

dimensional ferroelectric capacitors (see Fig. 7) for storage devices. This is mostly due to

the fabrication processes and materials which are similar to the ones used in nowadays

microelectronic industry. The gain in the cell surface of a ferroelectric capacitor made

from nanowires in the stacked architecture is simply the ratio between the surface with

and without nanowires:

G=Snw/S=π/2 Ra

Where Snw and S are the surface with and without nanowires, respectively, the latter

being given by L2, and Ra is the aspect ratio of a single nanowire. For nanowires with 100

nm diameter and 1 µm length the aspect ratio is 10 and the gain is about 15, which

practically means that the active surface is 15 times larger than in the planar case.

18

The nanowire-based fabrication method in combination with the template-assisted

large-scale ordered growth of nanowires (viz. ZnO nanopillars 57) might be an alternative

to the trench-based fabrication of 3D ferroelectric capacitors.

Figure 7: Proposed memory cell based on 3D ferroelectric structures built on

silicon nanowires.

4. Conclusions

In summary, we have shown here the possibility to obtain ferroelectric tubular

structures, including multilayer structures such as metal-ferroelectric-metal capacitors,

with a high aspect ratio and characteristic dimensions (diameters) ranging from about 100

nm up to several micrometers. The presented methods are suitable to fabricate single

nanotubes and ordered nanotube arrays. Potential applications are manifold and comprise

micro- and nano-fluidics, mesoscopic piezoelectric actuators and scanners, storage

devices and tunable photonic crystals, 58,59 or, as recently shown by Jim Scott, terahertz

emitters.60

19

Acknowledgements

Work partly supported by German Research Foundation (DFG) via SFB 762.

References

1 W. Liang, M. Bockrath, D. Bozovic, J.H. Hafner, M. Trinkham, and H. Park, Nature 411, 665 (2001).

2 A.M. Morales and C.M. Lieber, Science 287, 1471 (2000).

3 M. H. Huang, Y. Wu, H. Feick, N. Tran, E. Weber, and P. Yang, Adv. Mater. 13, 113 (2001).

4 V. F. Puntes, K. M. Krishnan, and A. P. Alivisatos, Science 291, 2115 (2001).

5 P. Yang, H. Yan, S. Mao, R. Russo, J. Johnson, R. Saykally, N. Morris, J. Pham, R. He, and H. J. Choi,

Adv. Funct. Mater. 12, 323 (2002)

6 J. J. Urban, J. E. Spanier, O. Y. Lian, W. S. Yun, and H. Park, Adv. Mater. 15, 423 (2003).

7 C.R. Martin, Science 266, 1961 (1994).

8 X. Zhu, Z. Liu, and N. Ming, Nanoscience and Nanotechnology 10, 4109 (2010).

9 M. Roervik, T. Grande, and M.-A. Einarsrud, Adv. Mater. 23, 4007 (2011).

10 M. Alexe, J. F. Scott, A. Pignolet, D. Hesse, and U. Gösele, Integrated Ferroelectrics 19 95 (1998).

11 M. Alexe, S. Senz, A. Pignolet, J. F. Scott, D. Hesse, and U. Gösele, Mat. Res. Soc. Symp. Proc. 493,

517 (1998).

12 M. Alexe, J. F. Scott, C. Curran, N. D. Zakharov, D. Hesse and A. Pignolet, Appl. Phys. Lett. 73, 1592

(1998).

13 J. F. Scott, M. Alexe, N. D. Zakharov, A. Pignolet, C. Curran, and D. Hesse,

Integrated Ferroelectrics 21, 1 (1998).

14 M. Alexe, A. Gruverman, C. Harnagea, N. D. Zakharov, A. Pignolet, D. Hesse, and J. F. Scott,

Appl. Phys. Lett. 75, 1158 (1999).

15 M. Alexe, C. Harnagea, D. Hesse, and U. Gösele, Appl. Phys. Lett. 75, 1793 (1999).

16 M. Alexe, C. Harnagea, W. Erfurth, D. Hesse, and U. Gösele, Applied Physics A 70, 247 (2000).

20

17 S. K. Lee, W. Lee, M. Alexe, K. Nielsch, D. Hesse, and U. Gösele, Appl. Phys. Lett. 86 152906 (2005).

18 W. Lee, H. Han, A. Lotnyk, M. A. Schubert, S. Senz, M. Alexe, D. Hesse, S. Baik, and U. Gösele,

Nature Nanotechnology 3, 402 (2008).

19 B. J. Rodriguez, X. S. Gao, L. F. Liu, W. Lee, I. I. Naumov, A. M. Bratkovsky, D. Hesse, and M. Alexe,

Nano Letters 9, 1127 (2009).

20 H. Han, Y. S. Kim, M. Alexe, D. Hesse, and W. Lee, Adv. Mater. 23, 4599 (2011).

21 A. Gruverman, D. Wu, H.-J. Fan, I. Vrejoiu, M. Alexe, R. J. Harrison and J. F. Scott, J. Phys.: Condens.

Matter 20, 342201 (2008).

22 M. H. Huang, Y. Wu, H. Feick, N. Tran, E. Weber, and P. Yang, Adv. Mater. 13, 113 (2001).

23 B. Im, H. Jun, K. H. Lee, S.-H. Lee, I. K. Yang, Y. H. Jeong, and J. S. Lee, Chem. Mater. 22, 4806

(2010).

24 L. Louis, P. Gemeiner, I. Ponomareva, L. Bellaiche, G. Geneste, W. Ma, N. Setter, and B. Dkhil, Nano

Letters 10, 1177 (2010).

25 R. Boucher, P. Renz, C. Li, T. Fuhrlich, J. Bauch, K. H. Yoon, and D. C. Lupascu, J. Appl. Phys. 110,

064112 (2011).

26 Y.-D. Hou, L. Hou, J.-L. Zhao, M.-K. Zhu, and H. Yan, J. Electroceramics 26, 37 (2011).

27 Y. Zhang, B. Liu, and D. Fang, J. Appl. Phys. 110, 054109 (2011).

28 Y. Q. Chen, Y. F. En, Y. Huang, X. D. Kong, X. J. Zheng, and Y. D. Lu, Appl. Phys. Lett. 99, 203106

(2011).

29 A. N. Morozovska, M. D. Glinchuk, R. K. Behera, B. Zaulychny, C. S. Deo, and E. A. Eliseev, Phys.

Rev. B 84, 205403 (2011).

30 T. Kasuga, M. Hiramatsu, A. Hoson, T. Sekino, and K. Niihara, Langmuir 14, 3160 (1998).

31 B.D. Yao, Y. F. Chan, X. Y. Zhang, W. F. Zhang, Z. Y. Yang, and N. Wang, Appl. Phys. Lett. 82, 281

(2003).

32 Y. Mao, S. Banerjee, and S. S. Wong, Chem. Commun. 2003 (3), 408.

33 H. Choi, S. Hong, Y. Kim, M. Kim, T.-H. Sung, H. Shin, and K. No, Phys. Stat. Sol. RRL 5, 59 (2011).

34 N. P. Padture and X. Wei, J. Am. Ceram. Soc. 86, 2215 (2003).

21

35 H. Masuda and K. Fukuda, Science 268, 1466 (1995).

36 A. P. Li, F. Müller, and U. Gösele, Appl. Phys. Lett. 84, 6023 (1998).

37 W. Lee, R. Ji, U. Gösele, and K. Nielsch, Nature Materials 5, 741 (2006).

38 D. Gong, C. A. Grimes, O. K. Varghese, Z. Chen, W. Hu, and E. C. Dickey, J. Mater. Res. 16, 3331

(2001).

39 X. Wei, A. Vasiliev, and N.P. Padture, J. Mater. Res. 20, 2140 (2005).

40 J. Choi, R. B. Wehrspohn, and U. Gösele, Adv. Mater. 15, 1531 (2003).

41 M. Knez, A. Kadri, C. Wege, U. Gösele, H. Jeske, and K. Nielsch, Nano Letters 6, 1172 (2006).

42 R. B. Wehrspohn (Ed.), Ordered porous nanostructures and applications, Springer, New York, 2005.

43 S. Matthias, J. Schilling, K. Nielsch, F. Müller, R.B. Wehrspohn, and U. Gösele, Adv. Mater. 14, 1618

(2002).

44 M. Steinhart, J. H. Wendorff, A. Greiner, R. B. Wehrspohn, K. Nielsch, J. Schilling, J. Choi, and U.

Gösele, Science 296, 1997 (2002).

45 Y. Luo, I. Szafraniak, N. D. Zakharov, V. Nagarajan, M. Steinhart, R. B. Wehrspohn, J. H. Wendorff,

R. Ramesh, and M. Alexe, Appl. Phys. Lett. 83, 440 (2003).

46 F. D. Morrison, L. Ramsay, and J. F. Scott, Journal of Physics: Condensed Matter 15, L527 (2003).

47 A.N. Morozovska, M.D. Glinchuk and E.A. Eliseev, Phys. Rev. B 73, 214106 (2006).

48 B. A. Hernandez, K. S. Chang, E. R. Fisher, and P. K. Dorhout, Chem. Mater. 14, 480 (2002).

49 X. Y. Zhang, X. Zhao, C. W. Lai, J. Wang, X. G. Tang, and J. Y. Dai, Appl. Phys. Lett. 85, 4190

(2004).

50 M. Steinhart, R. B. Wehrspohn, U. Gösele, J. H. Wendorff, Angew. Chem. Int. Ed. 43, 1334, (2004).

51 S. S. Nonnenmann, E. M. Gallo, and J. E. Spanier, Appl. Phys. Lett. 97, 102904 (2010).

52 H. Choi, Y. Kim, S. Hong, T.-H. Sung, H. Shin, and K. No, Phys. Stat. Sol. RRL 5, 289 (2011).

53 M. Alexe, D. Hesse, V. Schmidt, S. Senz, H. J. Fan, M. Zacharias, and U. Gösele, Appl. Phys. Lett. 89,

172907 (2006).

54 V. Schmidt, S. Senz, and U. Gösele, Nano Letters 5, 931 (2005).

22

55 H. J. Fan, R. Scholz, M. Zacharias, U. Gösele, F. Bertram, D. Forster, and J. Christen, Appl. Phys. Lett.

86, 023113 (2005).

56 I. Vrejoiu, G. Le Rhun, L. Pintilie, D. Hesse, M. Alexe, and U. Gösele, Adv. Mater., 18, 1657 (2006).

57 H. J. Fan, W. Lee, R. Hauschild, M. Alexe, G. Le Rhun, R. Scholz, A. Dadgar, K. Nielsch, H. Kalt, A.

Krost, M. Zacharias, and U. Gösele, Small 2, 561 (2006).

58 M. Alexe, Y. Luo, I. Szafraniak, R. B. Wehrspohn, and M. Steinhart, EU Patent No. EP1439546, 2004.

59 F. D. Morrison and J. F. Scott, UK Patent No. GB2414246, 2005.

60 J. F. Scott, H. J. Fan, S. Kawasaki, J. Banys, M. Ivanov, A. Krotkus, J. Macutkevic, R. Blinc, V. V.

Laguta, P. Cevc, J. S. Liu, and A. L. Kholkin, Nano Letters 8, 4404 (2008).