Download Mesas: Their Role in the Formation of Widely Spaced Step Arrays

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Rheology wikipedia , lookup

Surface tension wikipedia , lookup

Tunable metamaterial wikipedia , lookup

Glass transition wikipedia , lookup

Low-energy electron diffraction wikipedia , lookup

Photoconductive atomic force microscopy wikipedia , lookup

Molecular nanotechnology wikipedia , lookup

History of nanotechnology wikipedia , lookup

Atomic force microscopy wikipedia , lookup

Societal impact of nanotechnology wikipedia , lookup

Nanogenerator wikipedia , lookup

Sessile drop technique wikipedia , lookup

Self-assembled monolayer wikipedia , lookup

Strengthening mechanisms of materials wikipedia , lookup

Microelectromechanical systems wikipedia , lookup

Semiconductor wikipedia , lookup

Nanotechnology wikipedia , lookup

Synthetic setae wikipedia , lookup

Colloidal crystal wikipedia , lookup

Sol–gel process wikipedia , lookup

Impact of nanotechnology wikipedia , lookup

Ultrahydrophobicity wikipedia , lookup

Nanomaterials wikipedia , lookup

Energy applications of nanotechnology wikipedia , lookup

Nanochemistry wikipedia , lookup

Industrial applications of nanotechnology wikipedia , lookup

Adhesion wikipedia , lookup

Transcript
Materials
Spontaneous Formation and Breakdown of Ridges on Si(111) Mesas:
Their Role in the Formation of Widely Spaced Step Arrays
CNF Project # 317-87
Abstract:
Mesa structures fabricated on Si(111) surfaces have been found
experimentally to develop step arrays with large spacing of the order of a
micron or more after annealing at temperatures where sublimation becomes
important. Ridges around the edges initially develop during annealing and
form barriers to step motion. When the ridges eventually break down, an array
of steps of the same sign with a few wide terraces are produced.
Summary:
Arrays of steps with controlled spacing in order of microns may have
applications as templates for nanowires or organic thin films. Fabrication of
such step arrays on Si(111) have been investigated by using the following
method:
Si(111) was patterned by standard optical lithographic methods using silicon
oxide as an etch mask. Mesa structures with different azimuthal orientations
and size were etched into the silicon by reactive ion etching. The etch depth
was around 2 µm for all of our samples. 15 mm x 5 mm samples were cut
from the wafer using a diamond tip scriber. The samples were then dipped in
HF for 3 minutes to remove any residual oxide and to hydrogen-terminate the
surface before loading into the vacuum chamber.
The samples were heated by direct current in a UHV system (base pressure
10-10 Torr) while the temperature was monitored with an optical pyrometer.
Some samples were subjected to flashing above 1200°C to quickly remove
the native oxide layer before annealing while others were annealed without
flashing. The annealing temperature ranged from 925°C to 1150°C. After
annealing, the samples were taken out of the vacuum chamber and imaged in
air by atomic force microscopy.
Image 1 shows an example of the typical height profile of a mesa after
annealing. Ridges have formed along most of the edges of the mesas. The
portion of the edge without a ridge act as a step source for the widely separated
step arrays that develop on the mesa tops. The step arrays formed through this
Cornell NanoScale Facility, Member: National Nanotechnology Infrastructure Network
Principal Investigator(s): Jack Blakely
process showed similar step spacing on other mesas that were of the same size
and orientation on the same sample.
From a series of mesas patterned at different azimuthal orientations on the
sample, we also find that the length of the mesa edge without a ridge increases
dramatically as the mesa edge becomes nearly parallel to the miscut axis as
shown in Image 2. Neither the crystallographic direction of the mesa edges nor
the direction of the DC heating current (and hence electromigration effects)
had any influence on the location or amount of ridge breakdown.
The formation of the step arrays involves two separate stages: (1) ridge
formation and its subsequent breakdown, and (2) the motion of steps across
the mesa surface. Below the crystal roughening temperature, ridge formation
can take place by the modification of the shape of the steps near the mesa edge
as illustrated in Image 3. Ridge formation is a transient phenomena which will
slow down as the sharp corners and edges of the fabricated mesa becomes more
rounded due to the migration of atoms. As annealing progresses, the ridges will
subsequently break down. The height of the ridge depends on the number of
steps at the mesa edge, which explains the more extensive ridge breakdown
for the mesa edges that are oriented parallel to the miscut axis.
The effect of the ridges is to create a crater on top of the mesa. The steps
density inside the crater is lowered during annealing through a step clearing
process that has also been investigated by our group [1]. When a ridge breaks
down and the steps move into the mesa top that has a low density of steps,
the distance between the steps is large enough so that the step dynamics is
dominated by the sublimation of adatoms between the steps. In this case, widely
spaced steps in the order of microns can persist on the mesa tops, even as steps
move in from a mesa edge and sweep across the mesa top during annealing.
References:
[1] D.H. Lee, J.M. Blakely, Surf. Sci. 445 (2000) 32.
page 132
Spontaneous Formation and Breakdown of Ridges on Si(111) Mesas:
Their Role in the Formation of Widely Spaced Step Arrays
CNF Project # 317-87
User(s):
Kee-Chul Chang
Principal Investigator(s):
Jack Blakely
Affiliation(s):
Materials Science and Engineering,
Cornell University
Primary Funding:
Cornell Center for Materials
Research (CCMR)
Principal Investigator(s): Jack Blakely
•
Fabricated mesas on silicon(111).
•
Annealed the patterned silicon under ultrahigh vacuum.
•
Observed the resulting pattern of steps with
AFM.
•
The process resulted in an array of widely
spaced steps.
Image 1: Height profile of a mesa on Si(111) that has been annealed
at 1000°C for 1 hour followed by flashing at 1220°C for 20 seconds.
Ridges have formed along most of the edges of the mesas. The portion of
the edge without a ridge acts as a step source for the widely separated
step arrays that develop on the mesa tops.
Contact Information:
[email protected]
[email protected]
Image 3: Chemical potential gradients set up by high curvature at the
mesa edges results in mass flow from the mesa edge to a step on the
mesa as illustrated on the top picture. The existence of inner steps is
critical for this process as the nucleation of a new island on the mesa
involves a high energy barrier. Reshaping of the inner steps results in
Cornell NanoScale Facility
Materials, page 133
Image 2: Deflection AFM image of step arrays formed on top of a
rectangular mesa on Si(111) after flashing at 1250°C followed by
annealing at 1100°C for 15 minutes.
National Nanotechnology Infrastructure Network
Materials
Leakage Current through Thin Silicon Oxide
Grown on Atomically Flat Silicon Surfaces
CNF Project # 317-87
Abstract:
Our research project involves building and characterizing MOS capacitors
starting with Si (111) surfaces that are almost atomically flat (with no atomic
steps). Atomically flat surfaces can be obtained by high-temperature annealing
in UHV of specially patterned silicon samples according to a procedure
developed previously [1]. Thin silicon oxide layers were then grown by dry
oxidation on three types of surfaces: (a) atomically flat surfaces, (b) normal
(stepped) surfaces cleaned in UHV by the same high-temperature annealing,
and (c) normal wafer surfaces which underwent just an RCA chemical cleaning
before oxidation. Atomic force microscopy (AFM) was performed to reveal
the topography of the surfaces. Aluminum pads were deposited on these
oxidized surfaces using photolithography techniques, obtaining arrays of MOS
capacitors with dimensions of 18 x 18 µm. The leakage current through the
oxide was measured for all three cases.
Summary:
The leakage current was measured for these capacitors, tracing I-V
characteristics and testing against the Fowler-Nordheim (FN) theory. The slope
of the FN plot is related to the effective thickness of the oxide. The rougher
stepped surface was expected to show a higher tunneling current according
to the model developed by Majkusiak and Strojwas [2].
Initial results obtained in our group showed that, for an oxide thickness of
about 5 nm, the ratio of the slope of the FN plots for the capacitors on areas
outside the craters to the slope for the capacitors on step-free surfaces is about
0.9 [3]. Our more recent investigations also indicated that the slope of the
FN plot is about 5% higher for capacitors built on step-free surfaces than on
stepped surfaces for 6 nm oxide and about 10% higher for 5 nm oxide. There
is no measurable difference if the oxide is thicker than 8 nm.
Cornell NanoScale Facility, Member: National Nanotechnology Infrastructure Network
Principal Investigator(s): Jack M. Blakely
While it was expected that this difference should increase for thinner oxide,
our results for 3.5 and 4 nm oxide thickness showed only a ~ 5% higher slope
for the capacitors built on step-free surfaces. A possible explanation for this
result is that the thinner oxide requires a lower growth temperature, which in
turn is known to create a rougher Si/SiO2 interface. Nevertheless, for a given
voltage, the tunneling current is significantly lower for the capacitors built on
step-free regions than for those on the UHV treated stepped surface.
A much bigger leakage current was measured for the capacitors built on the
unpatterned wafer, which did not undergo a UHV high temperature cleaning. In
that case, the slope of the FN plot is several times smaller than for the sample
cleaned in the UHV chamber by high temperature annealing. Also, the overall
tunneling current is much higher in the case of the normal wafer surface. We
are currently investigating in detail this very important observation.
References:
[1] D. Lee and J. Blakely, “Formation and stability of large step-free areas on Si(001) and
Si(111)”, Surface Science 445, 32 (2000).
[2] B. Majkusiak and A. Strojwas, “Influence of oxide thickness non-uniformities on the tunnel
current-voltage and capacitance-voltage characteristics of the metal-oxide-semiconductor
system”, J. Appl. Phys. 74, 5638 (1993).
[3] A. C. Oliver and J. M. Blakely, “A comparison of tunneling through thin oxide layers on
step-free and normal Si surfaces”, Mat. Res. Soc. Symp. Proc. 747, V 4.6.1 (2003).
page 134
Leakage Current through Thin Silicon Oxide
Grown on Atomically Flat Silicon Surfaces
CNF Project # 317-87
•
Principal Investigator(s):
Jack M. Blakely
Step-free surfaces obtained by heating
patterned samples in ultra-high vacuum
chamber.
•
Affiliation(s):
Materials Science and Engineering,
Cornell University
Thin silicon oxide layer grown by dry furnace
oxidation.
•
MOS capacitors built both on normal wafer
and on UHV cleaned, step-free and stepped
regions.
Primary Funding:
National Science Foundation
•
Electrical characterization of these MOS
capacitors.
User(s):
Valerian Ignatescu
Contact Information:
[email protected]
[email protected]
Principal Investigator(s): Jack M. Blakely
Figure 1: AFM scan over a large area (60x60 µm), showing a crater
& the surrounding, stepped region; the steps have been removed from
the crater bottom.
http://people.ccmr.cornell.edu/
~blakely/
Figure 2: Optical image showing the aluminum pads deposited after
oxidation on step-free surfaces (inside the craters) and on stepped
surfaces (outside the patterned area); the smaller pads in-between
craters represents areas with slightly reduced step density.
Cornell NanoScale Facility
Materials, page 135
Figure 3: Average leakage current for five capacitors built respectively
on (A) step-free surfaces, (B) stepped surfaces and (C) on normal (not
cleaned in UHV) wafer surface; the oxide thickness is 3.5 nm.
National Nanotechnology Infrastructure Network
Materials
Thin Film Nucleation and Growth using Energetic Neutral Species
CNF Project # 459-92
Introduction:
The deposition and growth of organic materials for applications in electronics
and photonics differs fundamentally from that of more conventional inorganic
materials such as metals, semiconductors and oxides. A key difference involves
the presence of strong covalent and ionic bonding in the latter class of materials,
whereas organic materials are often bound by rather weak dispersion (van
der Waals) forces. In addition, many organic materials are often known to
crystallize in different phases, separated in total energy by amounts on the
order of a few kT. As a consequence, considerable promise exists in the use
of energy tunable molecular beams for the deposition of organic thin films.
In this project, we have used supersonic molecular beams as sources for film
deposition, in particular, for the deposition of thin films of pentacene, an
organic semiconductor.
Summary:
Pentacene is a promising candidate for applications in organic thin-film
electronics owing to the ability to form highly ordered thin films near room
temperature with excellent electrical transport properties [1-2]. Pentacene,
which possesses a very low vapor pressure, presents a number of experimental
challenges that must be overcome in order to generate energy tunable
beams.
First, we designed an in-vacuum evaporator heatable to 200-300°C to
provide enough flux for supersonic molecular beams. Next, using time-offlight quadrupole mass spectrometry, we characterized supersonic molecular
beams of pentacene generated using carrier gases of N2, He and H2. With
these carrier gases, we were able to obtain kinetic energies over the range of
Ei = 1.4-10.6 eV whereas incident molecular fluxes were on the order of 1015
molecules-cm-2s-1 [3].
Then we carried out nucleation and thin film deposition experiments with
these energetic sources and the samples were characterized by atomic force
microscopy (AFM) and ellipsometry.
Cornell NanoScale Facility, Member: National Nanotechnology Infrastructure Network
Principal Investigator(s): James R. Engstrom
We find that beam energy affects a number of phenomena, namely nucleation
in the monolayer regime, and both the kinetics of thin film deposition and
the microstructure in the multilayer regime. Closer examination of the data
indicates that the deposition rate in the monolayer regime is determined by
the trapping probability of pentacene [3], which decreases with increasing
energy. However, in this same regime the trapping probability is found to
decrease with more glancing angles of incidence, a result inconsistent with socalled normal energy scaling. Analysis of the data indicates that both parallel
and perpendicular energy of incident pentacene molecules are critical to the
process of adsorption [3].
In the multilayer regime, trapping probability also decreases with
increasing incident energy, but does so at a rate less than that observed in the
submonolayer regime. In addition, grain size of multilayer pentacene films is
found to increase with increasing incident energy at normal incidence. These
results demonstrate that the morphological evolution of organic thin films can
be modified substantially at high incident kinetic energies.
We have also fabricated organic thin film transistors (OTFTs) with
these pentacene films with gold top contacts on bare thermal SiO2 and
hexamethyldisilizane (HMDS) coated SiO2 substrates to understand the
relationship between performance characteristics of organic thin film transistors
(OTFTs), film microstructure and film-dielectric interface properties. We find
that with increasing incident energy of pentacene molecules, the grains get
larger and hence the field effect mobility obtained for the corresponding
OTFTs increases. In addition, the devices with HMDS primed SiO2 showed
improved performance characteristics with respect to bare thermal SiO2, owing
to reduced trapping at the semiconductor-dielectric interface.
References:
[1] C.D. Dimitrakopoulos & P.R.L. Malenfant, Adv. Mater. (Weinheim,Ger.) 14, 99 (2002)
[2] S.R. Forrest, Chem. Rev. (Washington, D.C.) 97, 1793 (1997).
[3] A.S. Killampalli, T.W. Schroeder & J.R. Engstrom, Appl. Phys. Lett. in press (2005).
page 136
Thin Film Nucleation and Growth using Energetic Neutral Species
CNF Project # 459-92
Principal Investigator(s): James R. Engstrom
User(s):
Aravind S. Killampalli
Jared L. Mack
Principal Investigator(s):
James R. Engstrom
Affiliation(s):
Chemical and Biomolecular
Engineering, Cornell University
•
Pentacene thin films have been deposited
using energetic supersonic molecular beams.
•
Incident energy of pentacene molecules
strongly influences nucleation, film deposition
rate and thin film microstructure.
•
Organic thin film transistors fabricated
from these pentacene films show promising
performance characteristics.
Primary Funding:
CCMR
Contact Information:
[email protected]
[email protected]
Figure 1: Atomic force micrographs of pentacene on SiO2 as a function
of exposure to the supersonic beam (Ei = 6.7 eV, qi = 0˚). In each case
the field of view is 20 µm x 20 µm. A line scan (across the 75 s image)
yields the height of pentacene islands to be 1.59 nm. Pentacene grows
layer by layer for the first two layers on SiO2.
Cornell NanoScale Facility
Materials, page 137
Figure 2: Coverage-exposure relationships, deduced from AFM,
for the adsorption of pentacene on SiO2 at the four incident kinetic
energies indicated and normal incidence. The solid lines represent
linear least-squares fits to the data. Deposition rate decreases with
increasing energy characteristic of trapping mediated physisorption
of pentacene on SiO2.
National Nanotechnology Infrastructure Network
Materials
Low Heat Capacity Substrates for Calorimetry Measurements
CNF Project # 522-94
Abstract:
The goal of our group is to develop new thermal analysis characterization
techniques and devices for nanometer scale thin films of metals, polymers
and nanoliter volumes of proteins [1-8]. Our technique makes use of MEMS
fabrication [7] techniques using Si-N membranes. A variety of sample
configurations can be used including vacuum-deposited, spin-cast, or deposited
via solution such as samples of self-assembled monolayers of alkanethiols
(SAMS).
In this report we show work on two topics: (1) heat capacity of the melting/
order-disorder transitions in 2D and 3D C16 alkanethiols SAMs [8], and
(2) size-dependence melting point depression and possible superheating of
bismuth nanoparticles [9].
Summary:
Our technique is used to characterize the thermal properties of twodimensional (2D) self-assembled monolayers (SAMs) of C-16 alkanethiols
on a gold surface. The amount of SAMs is typically in the 10 picomole range
over an area of 0.025 cm-cm. The order-disorder transition characteristics
of C16 2D SAMs grown on polycrystalline Au surfaces are easily observed
via nanocalorimetry in their as-deposited state without desorption for scan
temperatures below 100°C. Melting for the 2D SAMs occurs over a broad
temperature range (100°C) with a melting temperature of about 70°C. The
thiol can be thermally desorbed from the surface by scanning the sample to
temperatures above 100°C (Figure 1).
The scanning nanocalorimeter is also used to investigate the melting of
Bi nanoparticles. The particles were formed by evaporating Bi onto a silicon
nitride substrate. The particles self-assemble into truncated spherical particles.
Below 5 nm average film thickness, the mean particle sizes increased linearly
with deposition thickness, but increased rapidly for films 10 nm thick. As
expected, small particles were found to exhibit size-dependent melting
temperatures less than the bulk melting temperature (e.g. 67 K lower for a 3
nm radius particle). The measured melting temperatures for particles under
~7 nm in radius, however, were ~50 K above the value predicted by the
Cornell NanoScale Facility, Member: National Nanotechnology Infrastructure Network
Principal Investigator(s): Leslie H. Allen
homogeneous melting model (Figure 2). We discuss this discrepancy in terms
of a possible size-dependent crystal structure change and the superheating of
the solid phase.
Results from a two-step deposition, which created a bimodal particle size
distribution are shown in Figure 3. A 10-nm Bi deposition was made in order
to form large particles, then an additional 0.8 nm deposited to form small
ones. (a) A TEM micrograph showing a few very large particles combined
with many smaller ones. (b) Heat capacity data from the 10-nm and the 10
nm + 0.8-nm films. The smaller particles have a lower Tm which can be seen
in the small melting peak at ~225°C thus demonstrating two different melting
temperatures for the same material.
References:
[1] MY Efremov, F Schiettekatte, M Zhang, EA Olson, AT Kwan, RS Berry & LH Allen,
“Discrete Periodic Melting Points in Nanostructure” Phys. Rev. Lett. 85, 3560 (2000).
[2] M. Zhang, M.Y.Efremov, F. Schiettekatte, E.A.Olson, A.T.Kwan, L.S.Lai, J. E. Greene & L.
H. Allen “Melting Point Depression of Nanostructures using Heat Capacity Measurements:
Nanocalorimetry” Phys. Rev. B., 62, 10548 (2000).
[3] M. Yu. Efremov, E. A. Olson, M. Zhang, S. L. Lai, F. Schiettekatte, Z. S. Zhang, and L.
H. Allen, “Thin-Film Differential Scanning Nanocalorimetry: Heat Capacity Analysis”
Thermochimica Acta, 412, 13 (2004).
[4] M. Y. Efremov, E. A. Olson, M. Zhang, Z. Zhang & L. H. Allen. “Glass transition in ultrathin polymer films: calorimetric study”, Phys. Rev. Lett., 91, 85703 (2003).
[5] M. Y. Efremov, E. A. Olson, M. Zhang, Z. Zhang, F. Schiettekatte, and L. H. Allen, “UltraSensitive Thin-Film Differential Scanning Calorimeter”, Rev. Sci. Inst., 75, 179 (2004).
[6] M. Y. Efremov, E. A. Olson, M. Zhang, Z. Zhang & L. H. Allen. “Glass transition in
ultra-thin polymer films: calorimetric study: Annealing Study”, Macromolecule, 37, 4607
(2004), see also Macromolecules, 35, 1481 (2002).
[7] E. A. Olson, M. Yu. Efremov, M. Zhang, Z. S. Zhang, and L. H. Allen “The Design and
Operation of a MEMS Differential Scanning Nanocalorimeter”, IEEE J. Microelectromech
Sys. (JMEMS) 13, 355 (2003).
[8] Z. S. Zhang, O.Wilson, M. Efremov, E. Olson, M. Zhang, P. Braun, C. Ober, & L. H.
Allen, “Heat Capacity Measurements of 2D Self-Assembled Monolayers on Polycrystalline
Gold”, Appl. Phys. Lett., 84, 5198 (2004).
[9] E. A. Olson, M. Yu. Efremov, M. Zhang, Z. Zhang, and L. H. Allen “Size-dependent
melting of Bi nanoparticles” J. of Appl. Phys. v97, p034304 (2005).
page 138
Low Heat Capacity Substrates for Calorimetry Measurements
CNF Project # 522-94
Principal Investigator(s): Leslie H. Allen
User(s):
Mike Efremov
Eric Olson
Ming Zhang
Zishu (Sue) Zhang
Liang Hu
Figure 1, top left: A series of heat capacity Cp(T) curves of multiple scans
with two different maximum scanning temperatures. The 200˚C shows the
gradual desorption of 2D SAMs. There is a 30% loss of mass and a 40%
decrease in the heat during this series. For samples scanned to 330˚C
most of the remaining alkanethiol desorbs in the first few pulses during
these high temperature scans [Appl. Phys. Lett., 84, 5198 (2004)].
Figure 2, below left: Melting temperatures as functions of particle sizes,
using eight films 0.3 nm to 10 nm thick. The horizontal dotted-line is
the bulk Tm for Bi—271˚C. The dashed line shows the expected Tm(r)
behavior calculated from values using the homogeneous melting model
and the standard accepted surface energies The upper dashed-line is a
linear fit to the our experimental values for Tm(r) for smaller particles.
The difference between the dashed lines represents superheating of the
bismuth or a difference in crystal structure.
Principal Investigator(s):
Leslie H. Allen
Affiliation(s):
Material Science, University of
Illinois at Urbana-Champaign
Primary Funding:
NSF Materials Research Dr. L. Hess
(0108694); American Chemical
Society-Petroleum Research Fund
(37027-AC7); NSF Materials
Research Dr. L. Hess (0304149)
Dr. William Nes
Figure 1
Figure 2
Figure 3, below right: C p measurements of a bimodal particle
size distribution (a) A TEM micrograph showing a few very large
particles combined with many smaller ones. (b) Heat capacity Cp
data demonstrating two different melting temperatures for the same
material.
Contact Information:
[email protected]
http://allen.mse.uiuc.edu/index.htm
Figure 3
Cornell NanoScale Facility
Materials, page 139
National Nanotechnology Infrastructure Network
Materials
Investigation of the Effect of Biaxial Strain on the Diffusivity of B in Si
CNF Project # 1007-02
Abstract:
The purpose of this project is to measure the effect of biaxial stress on the
in-plane diffusivity of B in Si. For non-hydrostatic stresses, the diffusivity is
affected by the product of the stress tensor and the activation strain tensor [1].
For a cubic system, this three-by-three tensor can be reduced through symmetry
considerations to only three independent parameters. Our measurement of the
effect of biaxial stress on the in-plane diffusivity of B in Si will contribute to
the complete determination of this tensor. This measurement will be performed
by examining the lateral spread of B-doped stripes in a strained Si film which
will be patterned through standard lithographic techniques at the Cornell
NanoScale Facility (CNF).
Summary:
Externally applied stress influences the diffusivity through the scalar
activation volume in the case of hydrostatic pressure and the activation strain
tensor in the case of non-hydrostatic stresses [1]. This activation strain is the
sum of a formation (Vf) and migration volume(Vm), which are the volume
changes necessary for the formation of a mobile point defect and the additional
deformation when that defect is at the saddle point in its migration path.
For a cubic system, this three-by-three activation strain tensor can be reduced
through symmetry considerations to only three independent parameters. If
the entire tensor were known, we would be able to predict the effect of any
arbitrary stress state on the diffusivity in any arbitrary direction. Determination
of three independent parameters requires three independent measurements,
two of which have been performed to date [2].
The samples we eventually plan to use to perform this experiment consist
of biaxially strained Si layers which have been grown on top of relaxed SiGe
buffer layers. We plan to create a series of B-doped stripes in these strained Si
layers, and we will then anneal these samples and observe the lateral spread of
Cornell NanoScale Facility, Member: National Nanotechnology Infrastructure Network
Principal Investigator: Michael Aziz
the B lines using Scanning Capacitance Microscopy (SCM). We are using the
CNF to actually create these samples. There are several steps of the sample
fabrication which require the use of the resources at the CNF, and which have
been performed in the past year
The first step was to use a mask to lithographically pattern some wafers.
We coated plain Si wafers with 98 nm of PECVD oxide, after which they
were spun with resist and exposed with the Autostep. We then etched these
films, creating a physical mask for the B implantation. The B implantation
was performed elsewhere, but the wafers were brought back into the CNF
and a second layer of patterning was performed. This second set of lines was
an outline of the major features of the sample, and these lines were etched
through the oxide and into the Si using the Unaxis deep-Si etcher. The oxide
was then removed with wet chemistry. The last processing step was laserannealing, performed elsewhere.
We used Scanning Capacitance Microscopy to measure the lateral boron
profile and fitted the data to extract a diffusion length. On a set of samples
made from plain Si wafers the measured diffusion length was found to agree
with the predicted value from literature [2] to within a factor of about 3.
This sample fabrication has been an iterative process. There have been
several generations of samples over the past three years, involving both plain
Si wafers and strained Si films. Each generation has required two trips to the
CNF. We are about to begin the process and create what should be the final
set of samples with strained films.
References:
[1] Aziz, APL 70, 2810 (1997).
[2] Zhao, APL 74, 31 (1999).
page 140
Investigation of the Effect of Biaxial Strain on the Diffusivity of B in Si
CNF Project # 1007-02
Principal Investigator: Michael Aziz
User(s):
Jennifer Sage
Principal Investigator(s):
Michael Aziz
Affiliation(s):
Division of Engineering & Applied
Sciences, Harvard University
Primary Funding:
NSF
Goal: Measure the effect of biaxial strain on the in-plane component
of B diffusivity in Si. Figure 1: Mask design. Numbers represent stripe
width, stripe spacing and stripe length in microns.
Figure 2: Cross section of final sample. B stripes in thin, strained Si film
grown epitaxially on Si1-xGex. Later: laser-melt, furnace anneal.
Contact Information:
[email protected]
[email protected]
Figure 3, above: Scanning Capacitance Microscopy (SCM) image of a
boron-doped stripe. Areas of lighter contrast have higher doping.
Figure 4, right: Plot of predicted diffusion length, based on known anneal
time and temp and lit value of D vs. the measured diffusion length from
SCM data. They agree to within a factor of 3.
Cornell NanoScale Facility
Materials, page 141
National Nanotechnology Infrastructure Network
Materials
Effects of Surface Properties on the Fracture
Strength of Materials at Nanoscale
CNF Project # 1154-03
Abstract:
In our work, we are investigating the effects of surface properties on the
fracture of nano-scale materials. For this purpose, single crystal silicon beams
with chemically modified surface step structures were fabricated and the
fracture strength statistics were determined by loading them with a carefully
calibrated AFM cantilever. The obtained mean strengths of 10.79 GPa and
12.93 GPa for different surface properties indicate that higher fracture strength
values, which are closer to the theoretical limit, are possible by surface
modification.
Summary:
Fracture strength of bulk materials is much smaller than the theoretical
cleavage strength, which only considers the separation of perfect crystal planes.
Existing bulk or surface flaws (small cracks or geometric singularities such
as surface steps) act as stress concentrators and decrease the fracture strength
of the material. The flaw density decreases with decreasing size. Hence, at
smaller length scales experimental values approach theoretical ones [1].
To investigate the surface effects, we fabricated doubly clamped beams
from (111) single crystal silicon wafers. In order to isolate effects of surface
properties, the samples were designed to have a tapered region avoiding
corner stress singularities. The beams were 5 µm long and 1 µm wide. They
were released by potassium hydroxide (KOH) and tetramethyl ammonium
hydroxide (TMAH) etching.
After release, their thicknesses were determined by resonance frequency
measurements in high vacuum and their surface properties were characterized
by an AFM scan. Surface step characteristics depend on the etchant type as
well as other parameters like concentration, temperature, stirring, etc. For
instance, TMAH released beams have smaller step height and step density
then the KOH etched ones.
Cornell NanoScale Facility, Member: National Nanotechnology Infrastructure Network
Principal Investigator(s): Alan T. Zehnder
Samples released with different etchants were tested with AFM in force
mode. Before the experiment, the geometric properties of the AFM cantilever
were obtained by SEM analysis and resonance frequency measurements, and
the stiffness was determined by a static Finite Element Model. The experiment
consisted of loading the samples with the AFM cantilever until they fractured.
The deflection signal from the photo detector, the cantilever stiffness, and the
piezo motion were used to obtain the force deflection curve and therefore the
stress value at fracture.
In order to analyze the effects of an individual step, the theoretical behavior
of the stress near a geometric singularity [2] is compared to Finite Element
Simulations. Since there is a statistical variation in size, number and orientation
of the steps for fixed etching parameters, the fracture strength values were
analyzed statistically using Weibull Probability distribution. The Weibull
analysis indicated that the KOH and TMAH released beams were 10.79 GPa
and 12.93 GPa respectively. Their Weibull Modulus, m (less scatter in data for
higher values of m), however, were 15.14 and 11.81 respectively. It should be
noted that both strength values are much larger than the strength of millimeter
sized samples (0.5 GPa [1]) and close to the approximate theoretical strength
(~E/10). Modifying the surface steps further, we are expecting to reach the
theoretical strength value.
References:
[1] T. Namazu, Y. Isono, T. Tanaka, “Evaluation of Size Effect on Mechanical Properties
of Single Crystal Silicon by Nanoscale Bending Test Using AFM”, Journal of
Microelectromechanical Systems, Vol. 9, pp. 450-459 (2000).
[2] W. Suwito, M. L. Dunn, and S. J. Cunningham, “Fracture initiation at sharp notches in
single crystal silicon”, J. Appl. Phys., 83, p. 3574 (1998).
page 142
Effects of Surface Properties on the Fracture
Strength of Materials at Nanoscale
CNF Project # 1154-03
User(s):
Tuncay Alan
Principal Investigator(s):
Alan T. Zehnder
Affiliation(s):
Theoretical and Applied Mechanics,
Cornell University
Primary Funding:
Cornell Center for Materials
Research (CCMR), a Materials
Science and Engineering Center of
the National Science Foundation
(DMR-0079992).
Principal Investigator(s): Alan T. Zehnder
•
5 µm long single crystal silicon beams
fabricated.
•
Fracture tests performed with AFM.
•
Surface steps act as stress concentrators,
decreasing the fracture strength.
•
Higher strength values were obtained by
chemical surface modification.
Figure 1: AFM image of the KOH released beam.
Contact Information:
[email protected]
[email protected]
Figure 3: Probability of failure of KOH and TMAH released beams.
Cornell NanoScale Facility
Materials, page 143
Figure 2: Local stress behavior near the corner
of a step during fracture. Inset: A fractured test beam.
National Nanotechnology Infrastructure Network
Materials
Ordering on Corrugated Glass Surfaces
CNF Project # 1163-03
Abstract:
Corrugations were formed on glass surfaces by off-normal ion beam erosion.
Small amounts of gold were deposited onto the corrugated glass surfaces. The
changes in the positions of the deposited gold nanoclusers as a function of
annealing conditions were investigated by atomic force microscopy (AFM).
The glass substrates were also used to form cells in which to test the influence
of the corrugations on the alignment of nematic liquid crystals (LC’s).
Experimental Details:
Pieces of a commercial boroaluminosilicate glass (Corning Code 1737)
were washed with acetone and isopropyl alcohol and blown with dry nitrogen.
The pieces then were placed in a Veeco Ion Mill with an incoming ion angle
of 45° where they were bombarded at 450 V and 80 mA for 20 minutes.
The resulting corrugations had a wavelength of approximately 50 nm, and
amplitude of 1.5 nm.
The corrugated samples were then placed in the CVC 4500 evaporator
where gold coverages with thicknesses ranging from 5 to 15Å were thermally
deposited onto the corrugated glass surface at normal incidence. Samples were
placed in an Rapid Thermal Anneal (RTA) oven for various times ranging
from 10-60 seconds at 450°C. AFM characterization was performed after
each annealing period.
Before depositing a liquid crystal onto the substrate, the glass was soaked in
acetone for approximately 15 minutes to remove any organic residues, rinsed
with methanol and dried. Then a plasma cleaner, running for ten minutes, was
used to clean the sample.
Two types of liquid crystal cell were fabricated: a parallel cell (corrugations
parallel to each other) and a twisted-nematic cell (corrugations perpendicular
to each other). The cell is made by sandwiching liquid crystals between two
Cornell NanoScale Facility, Member: National Nanotechnology Infrastructure Network
Principal Investigator(s): Christopher Umbach
sputtered glass substrates. The two pieces of sputtered glass are separated by
a uniform gap approximately 8 µm. Using a pipette, a small drop of liquid
crystal (Merck E7) is deposited onto the sputtered glass and pressed down
by the second piece of sputtered glass. The alignment of LC’s relative to the
corrugation directions is determined by the intensity of the polarized light
transmitted through the cell in a cross-polarizer geometry.
Results:
At coverages of 3 nm of gold, the corrugation morphology is significantly
obscured by the gold overlayer. At coverages of 1 nm and less, the corrugations
can be distinguished from the gold nanoparticles that are present on the surface.
With annealing, the nanoparticles coarsen in size; because of the limitations
of AFM in imaging the troughs of the corrugations, it is not possible to
determine if the gold exhibits any preference for the trough or crests of the
corrugation.
In the case of the liquid crystal cells, the parallel cell showed alignment that
was more uniform than that achieved with unsputtered glass and close to the
quality that can be achieved with the conventional rubbed polymer alignment
layer. The twisted nematic cell showed macroscopic regions of the alignment
that typically occurs in a conventional twisted nematic cell, but this alignment
was not achieved over the entire cell.
Summary:
Corrugated glass substrates can be used as substrates for thin-film or liquid
crystal devices. The corrugations strongly influence the alignment of the liquid
crystals, while the ordering of gold particles is not detectible with atomic
force microscopy.
page 144
Ordering on Corrugated Glass Surfaces
CNF Project # 1163-03
Principal Investigator(s): Christopher Umbach
•
Bombard glass with 500 eV ions at 45° offnormal incidence.
•
Bombardment produces surface corrugations
with 50 nm wavelength and 1.5 nm
amplitude.
Affiliation(s):
1. Electrical Engineering & Phyiscs,
Lehigh University;
2. Materials Science & Engineering,
Cornell University
•
Deposit 1 nm gold and anneal at 450˚C.
•
Align liquid crystals (LC’s) in a sandwich
of corrugated glass substrates.
Primary Funding:
CNF REU program (Saboktakin),
Cornell LIFE program (Huang)
Figure 1, top:
User(s):
Marjan Saboktakin1
Chunghung Michael Huang2
Principal Investigator(s):
Christopher Umbach2
Contact Information:
[email protected]
[email protected]
[email protected]
Atomic force microscopy of gold on corrugations.
Gold nanoparticles replicate corrugations before
annealing. After annealing, gold nanoparticles
coarsen and corrugations are obscured.
Figure 1: 1 nm of gold deposited on corrugated glass.
Figure 2, bottom:
Polarized light microscopy of LC’s. Bright
transmission at 45˚ shows LC’s are aligned
with corrugations.
Figure 2: Transmission of polarized light
through cross-polarizers corrugated glass.
Cornell NanoScale Facility
Materials, page 145
National Nanotechnology Infrastructure Network
Materials
Self-Assembly of Non-Spherical Colloids
CNF Project # 1177-03
Abstract:
Fabrication of three-dimensional (3D) structures with micron-scale features
presents a continuing challenge. Photonic crystals are one example of such 3D
structures which are of interest to the scientific community. While multilayer
and 3D lithography are potential routes to these structures, we attempt to
construct with micron scale building blocks which self-assemble in solution. Or
in other words, we seek to develop a chemistry of non-spherical colloids.
Summary:
Previously we demonstrated the ability to create colloidal particles using
contact photolithography. A layer of OmniCoat sacrificial layer was spin-coated
on a wafer before spin-coating a layer of SU-8-2002. Both of these products
are available through MicroChem. After exposure and development of the SU8, the OmniCoat layer is dissolved using MicroChem’s PG Remover which
releases the particles into solution.
This year we have optimized the process by using the GCA5x Autostep
instead of the EV620 contact aligner. By using the Autostep, we have achieved
greater reproducibility with higher resolution of features. We have achieved
yields of over one billion particles per four inch wafer using the Autostep.
Also, we are able to generate a variety of shapes. For releasing these high
density arrangements of particles, we have found that sonication of the wafer
with isopropanol is useful.
Cornell NanoScale Facility, Member: National Nanotechnology Infrastructure Network
Principal Investigator(s): Abraham Stroock
After releasing the particles into isopropanol, we wash them in isopropanol
by repeated pelleting with centrifugation to remove any residual amounts of
processing chemicals. Pellets of SU-8 colloids can then be redispersed into
isopropanol or aqueous surfactant solutions via sonication. It is important
to note that neither pelleting nor sonication damages the colloids. The SU-8
particles have been successfully stabilized in aqueous solutions of Sodium
Dodeclysulfate (SDS).
Currently we are exploring depletion interactions as a means for selfassembly of our particles. In an effort to precisely control concentrations in
small volumes, we fabricate microfluidic devices in polydimethlysiloxane
(PDMS) by casting in molds fabricated from SU-8 100 via contact lithography
in CNF.
Our initial results are promising and we plan to pursue the fabrication of
similarly shaped particles in different materials such as silicon, silicon oxide,
or other materials for which CNF has processing capabilities.
References:
[1] J.D. Joannopoulos, R.D. Meade, J.N. Winn, Photonic Crystals: Molding the Flow of Light
(Princeton University Press, Princton, 1995).
[2] S. Asakura, F. Oosawa, On interaction between two bodies immersed in a solution of
macromolecules, Journal of Chemical Physics, 22, 1255 (1954).
page 146
Self-Assembly of Non-Spherical Colloids
CNF Project # 1177-03
User(s):
Stephane Badaire1
Cecile Cottin-Bizonne1
Joseph Woody1
Allen Yang2
Principal Investigator(s): Abraham Stroock
Goals:
1) Investigate the role of shape in defining the interaction between colloidal particles and 2) develop a set
of tools with which to self assemble structures on the
100 nanometer to 100 micron scale.
Principal Investigator(s):
Abraham Stroock1
Affiliation(s):
1. Chemical and Biomolecular
Engineering, Cornell University;
2. Chemical Engineering,
Carnegie Mellon University
Primary Funding:
Cornell Center for Materials
Research, Cornell University
Contact Information:
[email protected]
[email protected]
[email protected]
[email protected]
Figure 1: Lithographic process for the fabrication of colloids. Colloids
patterned on a wafer patterned with the Autostep 200 are released to
form a colloidal dispersion.
Cornell NanoScale Facility
Materials, page 147
Figure 2: Scanning electron micrograph of cylindrical particles
fabricated in negative photoresist, released, stabilized, and redeposited.
Particles of aspect ratio 0.2, 0.5, and 1.2 are shown.
National Nanotechnology Infrastructure Network
Materials
Mimicry of Biological Adhesion Through Fabrication of Fibrillar Surfaces
CNF Project # 1225-04
Introduction:
Geckos and other lizards as well as many species of insects utilize fibrillar
contact surfaces to enhance their ability to adhere to a wide variety of natural
surfaces. The arrays of tiny setae found on the contact surfaces of these
organisms are typically between 0.5 and 5 µm in diameter and range in
length from 5 to nearly 100 µm. Previous researchers have determined that
the mechanism of adhesion for the fibrillar structures found in nature is noncovalent (predominantly van der Waals) interactions [1].
In theoretical work, we have shown that a fibrillar surface in contact with a
substrate can possess both larger adhesion force and larger fracture energy than
a corresponding flat interface of the same materials [2]. In addition, fibrillar
surfaces are locally more compliant at the interface than flat ones due to the
ability of fibrils to bend [3]. We have shown in other theoretical work that this
allows a fibrillar surface to adhere better to rough surfaces than a flat surface
[4]. Seeking to confirm these theoretical ideas with quantitative experimental
data and to understand the functioning of the natural surfaces more fully, we
built several model fibrillar arrays that mimic the biological exemplars.
Summary:
Our first attempt was at macroscopic length scales, using manual fabrication
methods. We constructed fibrils approximately 1 mm in diameter from the
rubbery polymer poly(vinyl butyral) and showed via tensile pull off tests that
the fibrillar samples have about an order of magnitude larger adhesion energy
than flat control samples [5]. The caveat with these samples is that uniform
adhesive contact is difficult to achieve, and heat treatment at the interface was
required to attain it. However, at smaller length scales, surface forces become
stronger relative to the elastic restoring forces of the fibrillar structures. Hence,
at small length scales, it should be possible to bring a fibrillar surface into
intimate adhesive contact simply by placing it in proximity to the desired
adherend.
Cornell NanoScale Facility, Member: National Nanotechnology Infrastructure Network
Principal Investigator(s): Anand Jagota
Because of this, and because our theory predicts even larger adhesion strength
as the diameter of the fibril is reduced, we set about to build microscopic fibrillar
arrays. Using tools at CNF, we carried out two separate processes to produce
arrays of fibrils with diameter between 1 and 15 µm and length between 5 and
100 µm. The first process used standard photolithography and deep reactive
ion etching to produce holes in Si with the desired cross-sections and depths.
This Si master surface was then used to mold poly(dimethylsiloxane) (PDMS),
which was cured and removed from the master to give fibrils of the same shape
and size as the holes in the Si surface.
In the second process, photolithography and a wet chemical etch were used
to pattern an array of circles on an Al layer atop a cured polyimide (PI) film. A
reactive ion etch was then used to remove PI except where masked by the Al
circles. The result was circular PI fibrils with approximately 1 µm diameter
and lengths between 5 and 10 µm, depending on the etch time.
In preliminary results for the adhesion of microscopic samples, we found
in a tensile test that the pull off force of the fibrillar PDMS samples was about
80% that of the flat control [2]. Although the adhesion force is slightly smaller,
one must note that the area of contact is about 20 times as large for the flat
sample. Future work will involve maximizing the contact area for microscopic
fibrillar samples, as well as adding more complex geometry to the ends of the
fibrils, to increase compliance and increase the contact area and adhesion.
References:
[1]
[2]
[3]
[4]
[5]
Autumn, K. et al. Proc. Natl. Acad. Sci. USA 99, 12252-12256 (2002).
Hui, C.-Y. et al. J. Roy. Soc., Interface 1, 35-48 (2004).
Glassmaker, N. J., et al. J. Roy. Soc., Interface 1, 23-33 (2004).
Hui, C.-Y., et al. J. Adhesion 81, in press (2005).
Glassmaker, N. J., et al. Acta Biomaterialia 1, in press (2005).
page 148
Mimicry of Biological Adhesion Through Fabrication of Fibrillar Surfaces
CNF Project # 1225-04
User(s):
Nicholas J. Glassmaker
Ajita Rajan
Principal Investigator(s):
Anand Jagota
Affiliation(s):
Chemical Engineering,
Lehigh University
Primary Funding:
Lehigh University Start-Up Funds
Contact Information:
[email protected]
[email protected]
[email protected]
Principal Investigator(s): Anand Jagota
•
Natural fibrillar surfaces provide improved
adhesion in animals (Figure 1).
•
Theory shows fibrillar surfaces are stronger
than flat surfaces.
•
Large aspect ratio fibrils more compliant
than flat surfaces, allowing better contact
with rough surfaces.
•
Synthetic fibrillar surfaces fabricated by
molding (Figure 2) or direct reactive ion
etching (Figure 3).
Figure 1: Fibrillar surface, lizard (Anolis carolinensis).
Figure 2: Silicon master for molding fibrils.
Cornell NanoScale Facility
Materials, page 149
Figure 3: Polyimide fibrillar array (1 µm diameter).
National Nanotechnology Infrastructure Network
Materials
Electronic Charge Trap States Dynamics and Effect
on Imprint Behavior in P(VDF-TrFE) Ferroelectric Thin Films
CNF Project # 1253-04
Abstract:
We investigate the effect of imprint, a time-dependent tendency for
ferroelectric materials to resist polarization reversal, on ferroelectric
Poly(vinylidene fluoride-trifluoroethylene) (P(VDF-TrFE)) thin films. Based
on experimental measurements of the time and temperature dependence of
imprint, we have developed a model that links imprint to electronic charge trap
states in the bulk or near the metal/polymer interfaces. A fast-ramp thermally
stimulated current (TSC) measurement technique has been developed to
monitor trap state dynamics within the seconds time-frame.
Thin films of P(VDF-TrFE) with Ti electrodes were fabricated on oxidized
Si substrates. Following controlled polarization and imprint initialization,
charging of trap states was monitored through temperature cycling at fixed
heating and cooling rates. The post-TSC final polarization state was also
determined. Thermally induced trap filling has been directly correlated with
post-cycling imprint conditions.
Future experiments will focus on correlating trap densities, energetics and
kinetics with polymer and electrode compositions.
Summary:
P(VDF-TrFE) is a potential material for use as a storage medium for threedimensional, stacked non-volatile memory [1]. Imprint, a time-dependent
tendency for ferroelectric materials to resist polarization reversal, often limits
such memory applications. In inorganic ferroelectrics, imprint is related to the
contact materials and microstructure. In contrast, for P(VDF-TrFE), imprint
is hypothesized to result from internal electric fields formed by charges at
trap sites distributed throughout the film, as well as at interfaces between
the polymer and electrodes. The origin, dynamics, and control of these traps
remain a key research challenge to fully understand this material.
This research focuses on understanding and quantifying the origin of the
trap states in the polymer. To characterize the density and dynamics of these
Cornell NanoScale Facility, Member: National Nanotechnology Infrastructure Network
Principal Investigator(s): Michael O. Thompson
traps, we have extended the thermally stimulated discharge current technique,
typically used to study dipole relaxation in electrets [2], by almost two orders
of magnitude in time. The fast-ramp thermally stimulated current (TSC)
measurement is suitable for monitoring the trap state dynamics within the
seconds time-frame.
Thin films of P(VDF-TrFE) with Ti electrodes were fabricated on oxidized
Si substrates. Trap states were thermally filled/emptied through temperature
cycling between 20-100°C with heating and cooling rates of 1-5°C/s, and
observed as a function of poling conditions and imprint times. In addition to
charge accumulated during thermal cycling, the final polarization state was also
determined. Imprint-induced filling of trap states has been verified from the
TSC in the 1-10000 second time-frame [3]. Similarly, thermally induced trap
filling has been directly correlated with post-cycling imprint conditions.
In addition, results show a significant asymmetry with field orientation (top
versus bottom electrodes), suggesting that the trap density and characteristics
are directly related to processing conditions and materials.
Future experiments will focus on correlating trap energetics and kinetics
with polymer and electrode composition in order to determine relationships
between polymer/metal interactions and trap behavior.
References:
[1] O. Auciello, J.F. Scott, and R. Ramesh. “The Physics of Ferroelectric Memories”, Physics
Today, 51 (7), (1998), 22.
[2] E.R. Neagu, J.S. Hornsby, D.K. Das-Gupta. “Polarization and Space Charge Analysis in
Thermally Poled PVDF”, J. Phys. D (Appl. Phys), 35 (11), (2002), 1229.
[3] C. Lew, M.O. Thompson, “Quantifying the Role of Electronic Charge Trap States on
Imprint Behavior in Ferroelectric Poly(vinylidene fluoride-trifluoroethylene) (P(VDFTrFE)) Thin Films”, MRS Proc. 830, (2005), D2.4.
page 150
Electronic Charge Trap States Dynamics and Effect
on Imprint Behavior in P(VDF-TrFE) Ferroelectric Thin Films
CNF Project # 1253-04
Principal Investigator(s): Michael O. Thompson
•
P(VDF-TrFE) ferroelectric polymer for nonvolatile memory.
•
Affiliation(s):
Materials Science and Engineering,
Cornell University
Correlate imprint behavior with model of
electronic charge trap states in polymer and
polymer-metal interfaces.
•
Primary Funding:
Intel, Thin Film Electronics,
and others
Developed fast-ramp thermally stimulated
current (TSC) technique to study trap
dynamics and energetics.
•
Proved and quantified trap states filling in
1-10000s imprint time-frame.
User(s):
Connie Lew
Principal Investigator(s):
Michael O. Thompson
Contact Information:
[email protected]
[email protected]
Figure 2
Figure 3
Figure 1
Figure 1, above: Measurement: (a) poling, (b) imprint, (c) TSC.
Figure 2, opposite top: TSC filling trap states during first heat cycle.
Figure 3, opposite bottom: TSC with increasing imprint time, showing
trap filling during imprint.
Cornell NanoScale Facility
Materials, page 151
National Nanotechnology Infrastructure Network