NIRT: GOALI An ElectronBeam Based Microscale NanoManufacturing Platform with 1nm Accuracy

1
NIRT GOALI - An Electron-Beam Based Microscale
Nano-Manufacturing Platform with 1-nm Accuracy
UNIVERSITY OF KENTUCKY
Yugu Yang1, Chandan Samantaray1, Euclid Moon2,
Henry Smith2, Francesco Stellacci2, James
Spallas3, Lawrence Muray3, and J. Todd Hastings1
1University of Kentucky, 2Massachusetts
Institute of Technology, and 3Novelx Inc.
Recent Advance 2 High signal-to-noise ratio
fiducial grids
Introduction
Nanometer-Level Precision using Spatial-Phase
Locking with a Single Electron Beam
Standard SEBL System

• Motivation Scanning electron-beam lithographys
  (SEBL) arbitrary patterning capability and
  sub-10-nm resolution would be ideal for nanoscale
  manufacturing if three problems could be solved.
• Problems in standard SEBL
• poor pattern-placement accuracy,
• low throughput, and
• high cost-of-ownership
• Key Challenges
• Open loop operation (no direct knowledge of beam
  position, yet many factors introduce position
  errors)
• Current-density limited electron beam and time
  consuming stabilization, calibration, and
  alignment
• Costly precision electronics, electron-optics,
  and stages. High cost of electromagnetic,
  thermal, and vibration isolated facilities.

• Motivation Residual pattern-placement error is
  inversely proportional to the grid
  signal-to-noise ratio (SNR). To minimize
  placement errors while also minimizing system
  design constraints, we seek an order of magnitude
  improvement in SNR over our existing thin metal
  grids.
• Key Innovations
• Form grid lines from nanoparticles to increase
  secondary electron (SE) escape probability and
  thus SE yield.
• Produce a nano-corrugated sub-structure with a
  spatial frequency comparable to nanoparticles
  within the grid lines to increase secondary
  electron yield.
• Self-assemble nanoparticles in the
  nano-corrugated structure to further enhance
  secondary-electron yield.

Nanometer-level pattern-placement precision using
spatial-phase locked e-beam lithography with a
single beam on a Raith 150 SEBL system converted
for raster-scan operation. (a) Scanning
electron micrograph of an 8-nm thick aluminum
fiducial grid. The grid is rotated with respect
to the scan direction to provide feedback control
for both axes. (b) Averaged spatial-frequency
spectrum obtained from scanning the fiducial grid
many times. The two critical frequency
components (kHI and kLO) for determining x- and
y- position errors are labeled. Inset sample
signal from a single scan. (c) Histogram of
measured x-axis stitching errors at the vertical
boundaries of each deflection field after a
SPLEBL exposure using the grid above. (d) Y-axis
stitching errors at the horizontal field
boundaries. The mean, m, and standard deviation,
s, of each measurement is given, and the insets
contain electron micrographs of a representative
pattern at the field boundary. The gap between
the lines was intentionally introduced to
identify the boundary.3 This technique was
also adapted to multi-level alignment4 and beam
shape determination.5
SEM of C60 molecules on a silicon substrate. The
substrate is cleaved and the edge is overlapping
a Faraday cup in the wafer stage. A
cross-sectional view is superimposed in the
image, showing the secondary-electron return from
agglomerations of C60 molecules, relative to the
substrate and the Faraday cup. If these
nanoparticles can be assembled onto 100-nm wide
grid lines, C60 promises high SE yield with low
forward scattering of electrons.
Standard SEBL systems deflect a focused
electron-beam over a small region to write
nanoscale patterns. This approach provides
arbitrary patterning and sub-10-nm resolution,
but suffers from poor pattern placement, low
throughput, and high cost of ownership.

• Key Innovations
• Spatial-phase locking for closed-loop control of
  beam position based on the signal from an
  electron - transparent, metrologically accurate,
  fiducial grid.
• Multiple micro-fabricated electron-optical
  systems writing arbitrary patterns in parallel.

1-nm placement accuracy. System cost minimized
by relaxed requirements for precision
engineering, components, isolation, and
stabilization. No longer limited by current in
single beam. Column cost minimized by batch
fabrication with wafer scale process.
Objectives for Multi-Column Spatial-Phase Locked
EBL

• A high signal-to-noise ratio grid that does not
  perturb the nano-patterning process.
• A means to transfer the grid to each work-piece
  while retaining nanometer accuracy.
• A phase-locking system suitable for a
  micro-column SEBL array.

(d)
Micro-Column Array SEBL with Spatial-Phase Locking
(a) Schematic of a grid line composed of
deposited nanoparticles. (b) Grid line formed by
imprinted nano-corrugations in the e-beam resist.
(c) Grid line with enhanced secondary-electron
yield from imprinted nano-corrugations and
self-assembled nanoparticles. A conformal metal
coating may be added to prevent charging. (d)
Nano-corrugations in resist, produced by
tap-imprint of a scanning probe. Such
corrugations, if replicated by large area
imprinting, could serve as a high SE yield
sub-structure within the fiducial grid as shown
in (b) and (c).
Recent Advance 3 Parallelizable Spatial-Phase
Locking for Micro-column Arrays
(a)
(b)
Objective 1. Electron-transparent grids that
emit secondary electrons or photons when struck
by a primary electron are under development.
Secondary electron based grids are potentially
applicable to all primary electron energies
however, scintillating grids seem best suited for
low primary electron energies.
Objective 2. (a) Inking and stamping using a
hybrid rigid/soft master grid offers one
possibility for transferring the fiducial grid to
each work piece while maintaining nanometer
accuracy. (b) Near-field optical patterning
techniques using rigid masks are also under
development. SEM micrographs shows a grid
transferred using such a technique.
Motivation The current spatial-phase locking
system exposes 8 Mpixels/sec using a general
purpose microprocessor and 10 Msample/sec DACs
and ADCs. For high-throughput nanomanufacturing
with micro-column arrays we need a parallelizable
system that can run at up to 100Mpixels/second. Ke
y Innovation Integrated, single board
implementation of spatial-phase locking using a
FPGA.
Recent Advance 1 Spatial-phase locking system
for vector-scan SEBL
Motivation Currently, real-time spatial-phase
locking relies on raster scan exposures, but
often sparse nanoscale patterns can be fabricated
more rapidly using vector scan exposures. Key
Innovation A novel 2D phase-locking algorithm
that allows arbitrary sampling sequences.6
Multiple-electron beam (two shown)
nanomanufacturing system based on spatial-phase
locked electron-beam lithography (SPLEBL). Each
electron optical system is microfabricated by
Novelx, Inc. in a batch MEMs process. As the
electron-beams scan across the substrate they
interact with an electron-transparent fiducial
grid. The grid produces a secondary electron or
photon signal whose phase can be detected to
provide feedback control of the beam position.
Vector-scan SPLEBL Strategy
Secondary electron signals for vector-scan SPLEBL
Novelx Inc. Micro-Column SEBL Array
Schematic of the integrated spatial-phase locking
system suitable for either raster or vector scan
exposures and compatible with high-throughput
multi-column systems. The system is implemented
on a single board using a Xilinx Virtex 4
field-programmable gate array (FPGA) and
accommodates the full exposure speed of a Novelx
micro-column.
(b)
(a)
(a) Experimental secondary electron signal and
Fourier transform from a 400 nm period SiO2 on Si
grid. SNR 0.39. The fundamental frequency
components. are circled.
Nanomanufacturing K-12 Outreach and Education
(a)
(b)
Spatial-phase locking for vector-scan SEBL using
an intentionally rotated fiducial grid. In
vector scan SEBL sparse patterns are be exposed
using complicated deflection strategies (spiral,
flyback, or serpentine shape filling). This
requires more sophisticated spatial-phase locking
algorithms to extract the beam position
information from the grid signal.
Normalized standard deviation (200 trials per
data point) of experimental x and y position
errors vs. SNR for both 200nm and 400nm fiducial
grids rotated by 20?. Regions where the signal
was primarily composed of secondary electrons
(SE) or backscattered electrons (BSE) are
indicated. Also shown are the Cramer-Rao bounds
(CRB) based on phase estimation in Gaussian white
noise.
References
1 J. P. Spallas, C. S. Silver, and L. P.
Muray, "Arrayed miniature electron beam columns
for mask making," Journal of Vacuum Science
Technology B, vol. 24, pp. 2892-2896, Nov-Dec
2006. 2 L. P. Muray, C. S. Silver, and J. P.
Spallas, "Sub-100-nm lithography with miniature
electron beam columns," Journal of Vacuum Science
Technology B, vol. 24, pp. 2945-2950, Nov-Dec
2006. 3 J. T. Hastings, F. Zhang, and H. I.
Smith, "Nanometer-level stitching in
raster-scanning electron-beam lithography using
spatial-phase locking," Journal of Vacuum Science
Technology B, vol. 21, pp. 2650-2656, Nov-Dec
2003. 4 A. V. Krishnamurthy, R. V. Namepalli,
and J. T. Hastings, "Subpixel alignment for
scanning-beam lithography using one-dimensional,
phase-based mark detection," Journal of Vacuum
Science Technology B, vol. 23, pp. 3037-3042,
Nov-Dec 2005. 5 J. T. Hastings, "Real-time
determination of electron-beam probe shape using
an in situ fiducial grid," Journal of Vacuum
Science Technology B, vol. 24, pp. 2875-2880,
Nov-Dec 2006. 6 Y. Yang and J. T. Hastings,
"Real-time Spatial Phase Locking for Vector-Scan
Electron Beam Lithography," accepted for
publication in the Journal of Vacuum Science
Technology B., 2007.
(a) One group of middle school students (from 200
total) outside the entrance to U.K.s Center for
Nanoscale Science and Engineering. The students
proceeded to take a group picture and watch to
process of printing it onto a 4 silicon wafer
using photolithography. (b) 9th grade student
from Kentuckys Appalachian region who patterned
their own silicon wafer using photolithography
(inset).
Acknowledgements
This material is based upon work supported by the
National Science Foundation under Grant No.
0601351. Facilities and technical assistance
were provided by the University of Kentucky
Center for Nanoscale Science and Engineering
(CeNSE) which is supported by National Science
Foundation EPSCoR award No. 0447479. We
acknowledge Prof. Lance DeLong and Dr. Wentao Xu
(University of Kentucky Dept. of Physics) for
their help with the electron-beam lithography
system and Dr. Timothy Savas and Thomas OReilly
(MIT Nanostructures Laboratory) for fabricating
the fiducial grids used for the vector-scan
experiments.