83
14:15-14:30
A
Simple
Process
to Fabricate
Self-Aligned,
High-Performance
Torsional
Microscanners;
Demonstrated
Use
in
a Two-Dimensional
Scanner
H.
Choo,
D.
Gannire,
1.
Demmel,
and R.
S. Muller
Berkeley
Sensor
&
Actuator Center
CBSAC)
University
of California
at Berkeley
Using
a
n
ew
,
simple,
CMOS-compatible
process
carried
out
on SOl
w
af
e
r
s
,
we
have
built
high-performance
tcrsional
microscalUlers
ha
v
in
g
vertically
offset
interdigitated-comb
actuators.
Our microscanner-f
abrication
process
requires
three
p
ho
tclith
o
gra
p
h
y
masks:
two to
form
the
f
ro
n
t
·
s
id
e
microscanner
structures
and a third
to
defme
the
b
a
c
ks
i
d
e
o
p
enings
(
Figure
I).
Both
m
ov
in
g and
fixed
combs
are
f
a
b
ri
c
at
ed
in
the same
device
l
a
y
e
r
(30
!1ill
in
thickness),
and
t
h
e
offset
combs
are created
b
y reducing
the thickness
of the fixed
combs,
but not
that of
the
mo
v
in
g
combs.
Our
process
begins
by
removing
the 1-!1ill thick
thermal
oxide
selectively
to
open
rectangular
windows
at
locations
where
the fixed
combs
are
to
be
defined
(
F
igu
r
e
I-I).
In the following
step,
both
fixed-
and
moving-comb
sets
are defmed
s
i
m
u
l
tan
eo
u
sl
y with a
single
photclitho
graphy
mask;
this is followed
by
DRIE-etch
(Figure
1-2
&
1-3).
We then perform
a
timed-etch
in the DRIE-etcher
to
obtain
the desired
vertical
thickness
for the fixed
combs
(Figure
1-
5) without
affecting
the
moving
combs.
The
minimum
gap between
comb
f
m
ge
r
s
can be as
s
m
a
l
l
as twice
the
alignment
accuracy
of
the
photclithography
process,
which
i
s
,.;;0.4
Jlm
for state-of-the-art photolithography steppers.
The
u
n
if
o
rm
i
ty
of
the offset
heights
by
the
t
im
ed
-
e
tc
h
process
depends
on
the performance
of the DRIE
machine.
Using
our STS etcher,
we
have
been
able to keep
the overall
variation
in
offset
heights
across
the 4-inch
wafer
below
1.02%
(,.;;O.15-Jlill)
of the average
value
(15 Jlm) (T
a
bl
e
I). This
value
is
sm
al
l
er
than
the
ty
p
i
ca
l
thickness
fluctuations
(0.5-2
Jlm)
found
in
the
device
layers
of SOl wafers.
The simplicity
of this
fabrication
method
and its straightforward
use
of
well-established
IC-processing
tools,
provide
excellent
u
n
i
f
o
rm
i
t
y
,
performance,
and reliability
in
the
comb
drive
c
hara
ct
eris
t
i
c
s
as
well
as excellent
yields.
We have
designed
and built
m
i
c
rosc
ann
e
r
s
having
tcrsional
resonant
frequencies
between
0.8 and 16.4
kHz
and
maximum
optical-scanning
angles
between
8 and 23.30
with actuation
v
ol
ta
g
e
s
ranging
from
10 to 72,8
Vac_
rms.
Figure
3
s
h
o
w
s
the
frequency
responses
measured
for five
different
microscalUlers.
We have
made
two separate
fabrication
runs and
achieved
yields
of
operating
devices
better
than
75 %.
Damages
to
microscanners
m
o
s
tl
y occurred
during
the HF-releaselrinsinglcritical-point-drying
steps
as a consequence
of rough
handling.
We
consider
these
micro
scanners
as e
s
p
e
c
i
a
l
l
y
well
a
da
pt
e
d for a
pp
l
i
c
a
t
i
o
n
s
tc
refractive
laser
surgery
of
ocular
corneas
where
small
spot
size
and high
scan
speeds
are
important
assets.
To
d
e
mo
ns
tr
a
te
this application,
w
e
assembled a
two-dimensional
s
c
ann
ing system
by orienting
two identical
micro
scanners
at right
an
g
l
e
s
to one
another
(M
i
rr
o
r
#3 in
F
i
g
ur
e
3, mirror
diameter
=
1
rum,
resonant
frequency
=
6.01
kHz)
and
sc
an
n
e
d
a pulsed
laser
beam
(67Onm
wavelength).
The cross-coupled
scanners
w
er
e
driven
by t
w
o
6.01-kHz
s
in
e waves
that were
90° out of phase,
producing
circul
ar patterns
having
ra
di
i fixed
by
the
amplitude
of
the driving
voltage.
The laser
spots
forming
the
pattern
p
e
r
s
is
ted
for
OA)lsec
and
had a 220-)lm
diameter
(full
widthlhalf
maximum)
as
measured
with
a CCD
o
p
t
ical
sensor.
The
CCD
sensor,
used
in
p
l
a
c
e
of
an
ocular
c
o
rnea
,
allows
us to assess
performance
of the system. Refractive
laser
surgery
is a cumulative
a
b
l
a
t
i
o
n
process
[1] so to mimic
the
real
process,
we capture
the scanning
pattern
at
each
CCD
frame
and then
sum
the
in
t
e
n
s
i
ty
profiles
which
are proportional
to
the
f
m
a
l
ablation
pattern.
The
usual
period
of
time
for o
pt
i
c
al
l
a
se
r
surgery
is shorter
tban
20 minutes
so
w
e
measured
the repeatability
and stability
of our s
y
st
e
m
over
a period
of30
minutes
(Table
I
I
)
.
Our
s
y
s
tem
shows
excellent
r
ep
e
at
a
b
i
l
i
ty
in
pulse
position
(standard
deviation
less
than
0.56)lm)
as
w
e
l
l
as in
pulse diameter
(standard
deviation
less
than
O.68)lm)
around
the
ablation
zone
(Table II).
To demonstrate
the
versatility
of
our area
scanner,
we have
emulated
a small
pattern
from
the surface
topography
f
o
un
d
on a
US
R
o
o
s
e
v
el
t
dime
(Figure
4)
and
have
built
up
an
ablation
pattern
over
a 40-miuute
interval
(
F
i
g
ur
e 5).
The
resultant
pattern
compares
favorably
with
s
imi
l
a
r emulations
of such
ablation
p
a
t
t
e
rn
s
in
th
e
literature
[2].
The
high
p
erformance
and
e
xc
e
l
l
en
t
yield
of our self-alig
ned
vertically
offset
scanners
result
from
our
improved
and
s
im
p
l
ifi
ed
fabrication
technology
as compared
to technologies
reported
in
p
r
e
v
i
o
us
research.
Fabrication
challenges
enc
o
un
t
e
r
e
d
in
th
i
s
earlier
work
in
c
l
u
d
e
:
(a) the need
for
critical-aligmnent
steps
in a
t
w
o
-
w
a
f
e
r
process
[3};
(b) a
need
to control
and r
e
p
l
i
ca
t
e
the
pr
o
p
e
rt
ie
s
of materials
like
p
hot
o
r
e
s
ist o
r
bi-
m
o
rph layers
when
they
are used
for
hinges
[4]; (c)
a
need
for post-process-
annealing
in
a high-temperature
furnace
following
the
hand
assembly
of lid and
device chips
[5];
Cd)
a
need
to deposit
multiple-masking
layers
(composed
of silicon
dioxide
and
silicon
nitride)
to
create
0
ffset
combs
r 6].
Rderences
[11
J. F,
Bille,
C. F.
H.
Harner,
and
P. H. Loose!,
"Aborration-Froe
Refractive
Surgery,"
2"" E
d
i
t
i
o
n
.
Springer-Verlag,
2004, Chap.1
0, New
York,
USA
[21
op.
cil.
Chap. 10, page S2
[3].
R. A. Conant,
J, T. Nee, K.
Y. Lau. and
R.
S. Muller,
"A flat high-frequency
scanning
micrOlrurror."
Hilton
Head
Solid-State
Sensor
and
Actuator
Workshop
2000,
pp.6-9,
T
r
a
ns
d
uc
e
r
Research
Foundation.
Cleveland.
OH,
USA
[4].
P.
R. P
a
t
t
e
rs
o
n
; D.
Hahj
H.
Nguyen,
H.
Toshiyoshi,
R. Chao, a
n
d M. C. Vlu,
"A
scanning
nUcTomirmr
....vith
angular
comb
drive
actuation.".
International
Conference
on
Micro
Electro
Mechanical
Systems
2002,
pp.544-7,
Las Vegas,
NY,
USA
[5]
J. Kim, H. Choo.
L.
L
i
n
.
and
R.
S,
Muller,
"Micro
fabricated
tOfliional
a
c
t
u
a
t
o
r using self-aligned
p
l
as
ti
c
defonnation,"
IEEE Transducers 2003,
ppIOtS-IOI8.
Boston,MA,
USA
[6J
D. T.
McCormick
and
N. C.
Tlen.
"Multiple
Layer
Asymmetric
Vertical
Comb-Drive
Actuated
Trussed
Scanning
Mirrors,"
IEEEILEOS
International
Conference
on
Optical
MEMS 2003. pp.l2-13,
Waikoloa,
Hawaii.
USA
0-7B03-927B-7/0S/S20.00©200S
IEEE
21
Figure 1
Fabrication
Process:
T
h
e
left
column
shows
top
views
while
the right
column
shows
cross-sectional
views
along
the dotted
lines.
L
Grow
O.S-l1m
thennal
or low
temperature
oxide
(L TO)
for complet
e
ly
CMOS-compatible
process.
Using
the
photolitho-graphy
mask
#1,
pattern
and
remove
the thermal
oxi
d
e selectively
where
fixed c
o
mb
s
will
be
later
f
ab
r
i
c
at
ed
; 2.
Using
mask
#2, create patterns
of
microscalll1ers
including
moving
&
fixed
combs,
flexures,
and
mirrors;
3. Use
deep-reactive-ion-etch
(DRIE)
to
d
e
f
m
e
the
microscanners
in
the
device
layer;
4.
Remove
the
photoresist
layer
and
deposit
a very
thin
layer
(�O.2
Jim)
of L
TO; 5. Use
timed-anisotropic-plasma
etch
to remove
O.2-Jim
thick
LTO
from
the top-facing
surfaces.
Then
use
timed-anisotropic
or
isotropic
silicon-etch
to
c
r
eat
e
a
s
et of vertically
thinned
combs;
and
6.
Usin
g
mask #3,
pattern
and
op
e
n the
b
ack
s
i
d
e
of
the
microscanners.
Release
the
d
evices
in
HF.
-8
.f
0.8
Ii
�
0.6
"
�
0
4
.jij
E 0.2
o
Z
0
_Mirror
1
(fr
�
3.23 kHz)
-:-
Mirror
2
(rr
�
4.68 kHz)
-/ilittor �
(fr
=
6.01 kHz) -Mirror
4
(fr= 7.91 kH.)
---/II irror
5
(fr
=
B.89 kHz)
�
\ A
�
j \
$ l
�'-./
_V� E�
2 3
4
5
678
Frequency
(kHz)
9
10
Figure
3
Frequency
response
of five
different
rnicroscalll1ers:
The
amplitude
of
each
microscanner
is normalized
with
respect
to its
own
peak
amplitude.
Fig
u
re
2 SEM
pictures
of
micro
scanners
(device
layer
thickness
=
30
fIm):
The inset
in
the
p
i
c
ture (b) dearly
s
h
o
w
s
the
formation
of olfset
combs.
(The gap between
combs
is
511m
wide.
The
a
ffset
height
is 15
Jim.)
Ablation
Circles
(Max. Diameter
=
5 mm)
Location
1
2
Centroids
O.06Jim O.49Jim
Diameters
O.58pffi
O.64Jiffi
3
4
5
O.55Jiffi
O.47Jim
O.56Jiffi
O.68Jim
O.6311m
O.60Jirn
Table II
Repeatability
of
the ablation-pulse
locations (
c
entr
o
id
s
,
standard
deviations)
and
stability
of the p
u
l
s
e
diameters
at five
different
locations
in
the
c
i
rcular
ablation
zone
Figure 4
(a)
Picture
of a
US
dime.
(b)
gr
a
y
-sc
al
e
image
and
(c)
3-dimeI1sional
proftle
of the region
of
interest
indicated
by
the
do
tt
e
d cirde
(measurements
done
by using
WYKO
NT 3300)
Tahle I
Deviations
from
the
average
offset-he
ight (I5 Ji
m
)
Fig
u
re
5 L
e
ft: Orig
i
nal 3-D
surface
profile
of
a US
dime,
Right:
across
the
4-inch
wafer:
The 'Bottom'
region
c
o
rre
s
po
n
ds
t
o
3
-
D
di
a
g
r
am
o
f
small-spo
t
ablation
of
the dime's
surface
usin
g
the
wafer's flat
side.
(Overall
non-unifonnity
=
1.02%)
our microsc
anners
(peak-to-valley
height
difference
�
93
Jim)
22