D
. G
. S
C
H
LO
M
E
T
A
L.
J.
A
M
. C
ER
A
M
. S
O
C.
9
1,
2
42
9–
24
54
(2
0
0
8)
.
Thin films of oxygen-bearing compounds could have myriad practical
applications, finds Joerg Heber, if a few problems can be overcome.
In late 1996, a young Bell Labs physicist named Harold Hwang told his lab director that he wanted to start a radical programme of research into oxides — the ubiquitous,
oxygen-bearing compounds found in every-
thing from granite and glass to ceramics, chalk
and rust. Hwang was convinced that even the
most familiar oxides might show surprising
and useful properties if different ones could
be stacked up into ‘heterostructures’: layer-
cake-like arrangements in which each level is
an ultrathin film just a few atoms thick.
The lab director, Horst Störmer, was slightly
dubious — not about the potential, but about
the practicality. “Have you ever grown a thin
film in your life?” he asked. He knew all too well
what Hwang was getting himself into. Störmer
had made his own reputation by growing and
studying thin films of a very different class of
materials: semiconductors. Those films had
shown some remarkable properties — including
a phenomenon called the fractional quantum
Hall effect, in which the free-roaming electrons
inside a layer condense into a liquid-like state.
That discovery would later earn Störmer a share
of the 1998 Nobel Prize in Physics. But such
phenomena appeared only if the layers were
absolutely uniform in height, with a crystalline
structure that was so pure and defect-free that
electrons could race along without crashing
into imperfections. It had taken Störmer and
his colleagues at Bell Labs more than 10 years to
invent and perfect the techniques for fabricat-
ing such films. And oxides, he knew, would be
even more difficult to master. The compounds,
which form 99% of Earth’s outer crust, typically
consist of a larger number of chemical elements
than semiconductors, and have more complex
crystal structures.
Still, Störmer told Hwang to go ahead,
and the younger man did not disappoint. In
time, Hwang and others doing
research in the field succeeded
in growing high-quality oxide
thin films with the same atomic
precision as semiconductors.
And those films do indeed
exhibit interesting phenomena.
In 2004, for example, Hwang co-discovered
the existence of a two-dimensional (2D) elec-
tron gas, in which electrons at the interface
of two oxide thin films show an extremely
high mobility1 — an effect that is particularly
striking because the two oxides involved are
electrical insulators.
Now oxide thin films are at roughly the same
stage of development as semiconductor thin-
films were in the early 1970s — a period when
researchers were finally learning how to work
with them well enough to fabricate devices
such as the thin-film lasers, which would later
have their commercial breakthrough in com-
pact-disc players. For example, the 2D electron
gas that Hwang and his colleagues discovered
is being explored for use in a new type of fast
transistor, a device that can amplify or switch
electronic signals. Another use of oxide films
could be as the basis for very high-density
data-storage devices in which the magnetic
information is controlled with electrical fields.
And that’s just the beginning, says Hwang.
“The great opportunity we
have now is to design and
grow artificial thin film struc-
tures down to the atomic scale
— using multilayers of super-
conductors, ferromagnets, or
even a combination — and to
engineer systems that may one day be used for
electronics or sensing applications.”
The power of oxygen
The rich array of phenomena found in oxides
is largely due to the oxygen, says Yoshinori
Tokura, a physicist from the University of
Tokyo who has worked in this field for more
than 20 years. Oxygen tends to pull electrons
away from other atoms in the compound, says
Tokura, resulting in strong electrical fields at
the interatomic scale. These fields can give
“Have you ever
grown a thin film in
your life?”
— Horst Störmer
ENTER THE OXIDES
28
NATURE|Vol 459|7 May 2009NEWS FEATURE
Oxide Heterostructure MH CNS.indd 28 5/5/09 10:34:29
© 2009 Macmillan Publishers Limited. All rights reserved

rise to substantial correlations in behaviour
between the electrons of one atom and those
of its neighbours. And the correlations in turn
can lead to effects such as ferromagnetism, in
which a material’s electrons spontaneously line
up and produce a magnetic field.
Nevertheless, for many years researchers
tended to shy away from using oxides in
advanced applications, because they are far
more difficult to fabricate than metals and
semiconductors. This situ-
ation changed in 1986 with
the discovery of high-tem-
perature superconductivity
in certain oxides. The work
kicked off an intense, world-
wide focus on oxides that led
to other discoveries. In 1993,
for example, researchers
encountered ‘colossal mag-
netoresistance’, in which a slight shift in the
external magnetic field causes certain oxides
to undergo an orders-of-magnitude change in
electrical resistance2.
Another example is the 2D electron gas
that Hwang and his co-worker Akira Ohtomo
stumbled on when they were studying the
interface between two insulators3, lanthanum
aluminate (LaAlO3) and strontium titanate
(SrTiO3). “We started to fabricate very crude-
looking transistors that should not have been
conducting by themselves, but found they
were already conducting,” recalls Hwang, who
is now at the University of Tokyo. “We started
thinking, ‘What is going on here?’.”
They soon found that everything depended
on the precise crystalline structure of the inter-
face: only when the right atomic layers met
would the internal electrical fields on each side
push electrons towards the junction, so that
they could form the electron gas. Otherwise,
no charge layer develops4. The interface elec-
trons also turned out to be surprisingly mobile.
In fact, as discovered in 2007 by the groups of
Jochen Mannhart from the University of Augs-
burg in Germany and Jean-Marc Triscone from
the University of Geneva in Switzerland5, these
structures can become superconducting,
meaning that the electrons can travel without
resistance — albeit only below the very low
temperature of about 200 millikelvin.
Researchers have also been studying potential
applications that would exploit the thin-film
interface. One way to do that would be to place
a ferromagnetic oxide next to an insulating
oxide that isn’t ferromagnetic. If an external
electrical field is applied, it causes an electri-
cal polarization to develop at the interface. But
the field also shifts the number of electrons in
the ferromagnetic material, which changes the
magnetic field. As a result, electrical polariza-
tion and magnetism are both controlled by the
same electrical field, and are therefore cross-
linked — a coupling of properties that defines
multiferroic materials6. Such materials are of
interest both as magnetic field sensors and as
memory devices, in which information is writ-
ten by electrical voltages and read by magnetic
read head — with the benefit that no electrical
current flows through the device, significantly
reducing heat generation. Indeed, says Tokura,
“the route via thin films offers
the most straightforward fab-
rication method to realize a
multiferroic material”.
As well as looking at the
coupling of two different prop-
erties at an oxide interface,
researchers are looking for
applications in which a single
property, such as magnetic
field7,8 or electrical conductivity, is controllably
turned on and off. “Controlling conductivity
as a whole, rather than electrical current itself,
in some sense is the most exciting area,” says
Stuart Parkin, a physicist at IBM’s Almaden
Research Center in San Jose, California. “Con-
trary to conventional transistors, the required
current densities could be quite small, and this
is what you want for applications.”
Consider, for example, the superconducting
2D electron gas. Through the application of
an electrical field it is easy to push electrons
away from the interface, destroying the super-
conducting state and making it impossible for
current to flow9.This is analogous to what hap-
pens in conventional transistors, in which the
flow of electrons can likewise be switched on
or off by an external electrical field. Conceiv-
ably, researchers could use local voltages to
write complex patterns into this 2D electron
gas. Where a voltage is applied, the interface
would be insulating, and elsewhere it would
be superconducing — potentially allowing the
definition of entire electronic circuits. “It will be
exciting to see the realization of small devices
such as logic and memory circuits, or even small
amplifiers,” says Mannhart. Amplifiers written
into the superconducting film could enable fast
switching with extremely low noise levels and
thus could detect and amplify weak electronic
signals. Even the logic gates used for quantum
information processing could be etched into the
superconducting layer this way.
Just one small push
Unfortunately, the switching of superconduc-
tivity in the LaAlO3/SrTiO3 system occurs at
temperatures far too low to be relevant for
most applications. So one alternative is to
look at the different phases many oxides show
at various temperatures or pressures. Con-
ductivity often changes dramatically at the
transition from one phase to another. “If you
go to phase boundaries, that’s where you often
get extremely large instabilities,” says Parkin.
“Then you can imagine controlling those
states by small modifications.” Such a small
trigger impulse can push the system from one
phase to the other. This is what happens in
colossal magnetoresistance — a small external
magnetic field induces huge variations in elec-
trical resistance during a phase transition.
To realize high-quality oxide heterostruc-
tures for applications, researchers have had to
overcome substantial obstacles to the develop-
ment of suitable thin-film growth techniques.
Oxides often have complex crystal structures,
The interface between lanthanum aluminate and
strontium titanate.
“You can imagine
controlling the states
at phase boundaries
by making small
modifications.”
— Stuart Parkin
IB
M

J.
M
A
N
N
H
A
RT
/G
ER
M
A
N
H
A
M
M
ER
L
Stuart Parkin: excited by
controlling conductivity.
29
NATURE|Vol 459|7 May 2009 NEWS FEATURE
Oxide Heterostructure MH CNS.indd 29 5/5/09 10:34:33
© 2009 Macmillan Publishers Limited. All rights reserved

and films refuse to grow properly unless the
right crystal layer is exposed on the top sur-
face. Otherwise, the incoming atoms will not
be able to stick to the proper chemical bonds.
In 1994, this problem was identified and solved
for SrTiO3 by Masashi Kawasaki, a materials
scientist then at the Tokyo Institute of Technol-
ogy, now at Tohoku University in Japan, who
developed a pre-growth treatment involving
various acids that strip the crystal substrate
down to the desired atomic layer10. With this
advance, says Kawasaki, “people could finally
grow complex oxides”.
Unfortunately, it is still impossible to obtain
oxide heterostructures anywhere near as large
as those used in silicon technology. “Scaling up
SrTiO3 wafers to realistic sizes is out of the ques-
tion,” says Darrell Schlom, a materials scientist
from Cornell University in Ithaca, New York.
So, many researchers are now trying to integrate
oxide heterostructures into silicon wafers. “The
plan is not only to integrate oxides with silicon
electronics, but even more importantly to take
advantage of the processing infrastructure of
silicon technology,” says Schlom.
This is an arduous task, not least because
there are substantial differences in crystal
structure between most oxides and silicon.
And worse, oxide thin films are grown by
condensing a high-temperature vapour that
includes oxygen — which can turn silicon into
silicon dioxide at the slightest contact. This can
be avoided only by carefully adjusting growth
temperatures and supplying just the right
amount of oxygen at precisely the right time.
Still, progress has been made and the quality
of oxide films on silicon has been improving
steadily11. “Even advanced oxide films such as
LaAlO3/SrTiO3 heterostruc-
tures have now been grown
on silicon,” says Mannhart.
Not so crazy
Particularly promising in this
regard is zinc oxide (ZnO),
which is itself a semicon-
ductor with a wide range of
potential applications. “In ZnO, electrons can
travel up to a micrometre without scattering,”
says Kawasaki. Kawasaki and his colleagues
have even observed the quantum Hall effect
in ZnO — a first for an oxide12. The presence
of such quantum phenomena suggests the use
of ZnO for ‘spintronics’ applications, which
promise ultrahigh-density storage and ultrafast
processing of information using the electron’s
tiny magnetic moment, or spin.
This isn’t the end of the possible uses of
oxides. “This might be a very crazy idea, but we
are wondering whether these heterostructures
can be applied to new types of solar cells,” says
Tokura. Solar cells are currently made of semi-
conductors, he explains, and function through
the absorption of light with energies larger than
a certain threshold known as the band gap. If
the light has an energy much larger than this
band gap, the excess is wasted into heat. But if
electrons are confined, for example, in semi-
conductor nanoparticles, they begin to inter-
act strongly with each other, which amplifies a
process in which the excess energy is not wasted
but rather used to excite multiple electrons. The
entire process becomes more efficient.
In complex oxides, with their strong electron
correlations, such an amplification could be
very strong, says Tokura. Indeed, researchers
already know of certain oxides in which light
can excite so many electrons that the material
becomes metallic. But that would still leave
the problem of extracting these electrons from
the oxide to put their energy to use. Even here,
oxide thin-film structures
may offer a solution. The
layers are generally very
thin, which means that elec-
trons generated in one film
could easily be extracted to
an adjacent layer. “If we can
make this work, it would
be really exciting,” says
Kawasaki, who is investigating this idea with
Tokura.
Parkin has an even more ambitious idea. He
is looking for layered oxide systems in which
superconductivity sets in at unprecedentedly
high temperatures. “Room temperature is,
of course, the ultimate goal,” Parkin says. “In
my mind this is entirely feasible.” He thinks
that such superconductivity might be found
at interfaces similar to LaAlO3/SrTiO3, and
might also involve the use of oxide compounds
that do not normally exist in nature and can
only be stabilized as thin films.
After more than 20 years of research into
oxide thin films, efforts are bearing fruit.
Progress is becoming fast-paced. Thin-film-
growth technology has been adapted for oxide
compounds, suitable substrates have been devel-
oped and complex heterostructures are being
studied for new functionality. “Although what
we have achieved as a community is still at the
very early stages, we now know a lot more about
the basic rules of engagement,” says Hwang.
At the same time, Hwang sounds a note of
caution. “Now the hard questions come,” he
says. Even seemingly mundane issues such
as sample quality need to be tackled. “Oxide
heterostructures are still loaded with defects.
Understanding how to control these is key to
taking oxide heterostructures from scientific
curiosity — their current position in various
scientific sandboxes — to real technologies,”
says Schlom.
Nevertheless, the achievements so far are
strong testament to the fact that researchers
in the field have begun to predict and con-
trol the phenomena that can exist in oxide
heterostructures. Whether as new electronic
compounds, as sensors, as memory devices, as
solar cells or simply for their exciting science,
oxide heterostructures are here to stay. The
journey has merely begun. ■
Joerg Heber is a senior editor of Nature
Materials.
1. Ohtomo, A. & Hwang, H. Y. Nature 427, 423–426 (2004).
2. Jin, S. et al. Science 264, 413–415 (1994).
3. Ohtomo, A., Muller, D. A., Grazul, J. L. & Hwang, H. Y. Nature
419, 378–380 (2002).
4. Nakagawa, N., Hwang, H. Y. & Muller, D. A. Nature Mater. 5,
204–209 (2006).
5. Reyren, N. et al. Science 317, 1196–1199 (2007).
6. Ramesh, R. & Spaldin N. A. Nature Mater. 6, 21–29 (2007).
7. Kanki, T., Tanaka, H. & Kawai, T. Appl. Phys. Lett. 89, 242506
(2006).
8. Salvador, P. A., Haghiri-Gosnet, A.-M., Mercey, B., Hervieu, M.
& Raveau, B. Appl. Phys. Lett. 75, 2638–2640 (1999).
9. Caviglia, A. D. et al. Nature 456, 624–627 (2008).
10. Kawasaki, M. et al. Science 266, 1540–1542 (1994).
11. Warusawithana, M. P. et al. Science 324, 367–370 (2009).
12. Tsukazaki, A. et al. Science 315, 1388–1391 (2007).
J.
M
A
N
N
H
A
RT
Electronic circuits made from thin oxide layers are only starting to tap the potential of oxides.
“This might be a crazy
idea, but maybe these
heterostructures can be
applied to solar cells.”
— Yoshinori Tokura
30
NATURE|Vol 459|7 May 2009NEWS FEATURE
Oxide Heterostructure MH CNS.indd 30 5/5/09 10:34:38
© 2009 Macmillan Publishers Limited. All rights reserved