Georgia Tech postdoctoral fellow Suenne Kim holds a sample of flexible polyimide substrate used in research on a new technique for producing ferroelectric nanostructures. Assistant professor Nazanin Bassiri-Gharb points to a feature on the material, while graduate research assistant Yaser Bastani observes. Photo: Gary Meek |
Using
a technique known as thermochemical nanolithography (TCNL), researchers have
developed a new way to fabricate nanometer-scale ferroelectric structures directly
on flexible plastic substrates that would be unable to withstand the processing
temperatures normally required to create such nanostructures.
The
technique, which uses a heated atomic force microscope (AFM) tip to produce
patterns, could facilitate high-density, low-cost production of complex ferroelectric
structures for energy harvesting arrays, sensors, and actuators in
nano-electromechanical systems (NEMS) and micro-electromechanical systems
(MEMS). The research was reported in Advanced Materials.
“We can directly create piezoelectric materials of the shape
we want, where we want them, on flexible substrates for use in energy harvesting
and other applications,” says Nazanin Bassiri-Gharb, coauthor of the paper
and an assistant professor in the School
of Mechanical Engineering
at the Georgia Institute of Technology. “This is the first time that structures
like these have been directly grown with a CMOS-compatible process at such a
small resolution. Not only have we been able to grow these ferroelectric
structures at low substrate temperatures, but we have also been able to pattern
them at very small scales.”
The research was sponsored by the National Science
Foundation and the U.S. Department of Energy. In addition to the Georgia Tech
researchers, the work also involved scientists from the University
of Illinois Urbana-Champaign and the University of Nebraska Lincoln.
The researchers have produced wires approximately 30 nm wide
and spheres with diameters of approximately 10 nm using the patterning
technique. Spheres with potential application as ferroelectric memory were
fabricated at densities exceeding 200 GB per square inch—currently the record
for this perovskite-type ferroelectric material, says Suenne Kim, the paper’s
first author and a postdoctoral fellow in laboratory of Professor Elisa Riedo
in Georgia Tech’s School
of Physics.
Image shows the topography (by atomic force microscope) of a ferroelectric PTO line array crystallized on a 360 nm thick precursor film on polyimide. The scale bar corresponds to one micron. Image: Suenne Kim |
Ferroelectric materials are attractive because they exhibit
charge-generating piezoelectric responses an order of magnitude larger than
those of materials such as aluminum nitride or zinc oxide. The polarization of
the materials can be easily and rapidly changed, giving them potential
application as random access memory elements.
But the materials can be difficult to fabricate, requiring
temperatures greater than 600 C for crystallization. Chemical etching
techniques produce grain sizes as large as the nanoscale features researchers
would like to produce, while physical etching processes damage the structures
and reduce their attractive properties. Until now, these challenges required
that ferroelectric structures be grown on a single-crystal substrate compatible
with high temperatures, then transferred to a flexible substrate for use in
energy-harvesting.
The thermochemical nanolithography process, which was
developed at Georgia Tech in 2007, addresses those challenges by using
extremely localized heating to form structures only where the
resistively-heated AFM tip contacts a precursor material. A computer controls
the AFM writing, allowing the researchers to create patterns of crystallized
material where desired. To create energy-harvesting structures, for example,
lines corresponding to ferroelectric nanowires can be drawn along the direction
in which strain would be applied.
“The heat from the AFM tip crystallizes the amorphous
precursor to make the structure,” Bassiri-Gharb explains. “The patterns are
formed only where the crystallization occurs.”
To begin the fabrication, the sol-gel precursor material is
first applied to a substrate with a standard spin-coating method, then briefly
heated to approximately 250 C to drive off the organic solvents. The
researchers have used polyimide, glass and silicon substrates, but in
principle, any material able to withstand the 250-degree heating step could be used.
Structures have been made from Pb(ZrTi)O3—known as PZT, and PbTiO3—known as
PTO.
Scanning electron microscope (SEM) image shows a large PZT line array crystallized on a 240 nm thick precursor film on a platinized silicon wafer. Image: Yaser Bastani |
“We still heat the precursor at the temperatures required to
crystallize the structure, but the heating is so localized that it does not
affect the substrate,” explains Riedo, a coauthor of the paper and an associate
professor in the Georgia Tech School of Physics.
The heated AFM tips were provided by William King, a
professor in the Department of Mechanical Science and Engineering at the University of Illinois at Urbana-Champaign.
As a next step, the researchers plan to use arrays of AFM
tips to produce larger patterned areas, and improve the heated AFM tips to operate
for longer periods of time. The researchers also hope to understand the basic
science behind ferroelectric materials, including properties at the nanoscale.
“We need to look at the growth thermodynamics of these
ferroelectric materials,” says Bassiri-Gharb. “We also need to see how the
properties change when you move from the bulk to the micron scale and then to
the nanometer scale. We need to understand what really happens to the extrinsic
and intrinsic responses of the materials at these small scales.”
Ultimately, arrays of AFM tips under computer control could
produce complete devices, providing an alternative to current fabrication
techniques.
“Thermochemical nanolithography is a very powerful
nanofabrication technique that, through heating, is like a nanoscale pen that
can create nanostructures useful in a variety of applications, including
protein arrays, DNA arrays, and graphene-like nanowires,” Riedo explains. “We
are really addressing the problem caused by the existing limitations of
photolithography at these size scales. We can envision creating a full device
based on the same fabrication technique without the requirements of costly
clean rooms and vacuum-based equipment. We are moving toward a process in which
multiple steps are done using the same tool to pattern at the small scale.”