Content uploaded by Michel Demierre
Author content
All content in this area was uploaded by Michel Demierre on May 08, 2014
Content may be subject to copyright.
Content uploaded by Michel Demierre
Author content
All content in this area was uploaded by Michel Demierre
Content may be subject to copyright.
4D2.06P
SIMPLE AND LOW COST FABRICATION OF EMBEDDED MICRO-
CHANNELS
BY
USING A NEW THICK-FILM PHOTOPLASTIC
L.
J.
Gukrin, M. Bossel, M. Demierre,
S.
Calmes, Ph. Renaud
Institute of Microsystems, EPFL - Swiss Federal Institute of Technology
CH
-
1015
Lausanne, Switzerland
Tel:
+41
(0)
21
693 58 57; fax:
+41
(0)
21 693 66
70
email: 1ouis.guerin @ims.dmt.epfl.ch
SUMMARY
In this paper we describe new ways to realise
micro-channels. The used technologies are based
on
a novel thick-film photoplastic, the SU-8. This
photoresist has outstanding properties, allowing the
realisation of high aspect ratio microstructures. It is a
simple and low-cost fabrication process. The micro
channels developed in this work it will be used for
the realisation of thermal flow-sensors.
Keywords: Photoplastic, microfluidic channels.
INTRODUCTION
Different procedures may be used for the
fabrication of closed embedded channels for
microfluidic devices. A classical method is to
surface-micromachine a silicon wafer. The channels
are
then covered by bonding the latter onto a second
silicon or glass wafer. Other techniques to realise
micro- channels may be LIGA technologies or micro-
stereolithography processes. All these methods are
rather complex, expensive and/or need special
equipment. In order to get commercially viable
products, new and inexpensive techniques have to be
found. In this work such a low-cost and easy-to-use
method will be described.
THE
SU-8
PHOTORESIST
The material used in this work is a new, negative
photoresist known under the name of SU-8
[l],
developed at IBM-Yorktown, IBM-Zurich and
EPFL-Institute for Microsystems
[2].
It exhibits
outstanding properties, allowing the realisation of
thick microstructures
(>
200
pm) with aspect ratios
of about
20.
This photoresist is a low-cost material,
which may be used as a photoplastic for permanent
use. It consists of a commercially epoxy (EPONTM
SU-8 from Shell Chemical), an organic solvent and a
photoinitiator. It is deposited on
a
surface by
spinning or by other techniques, i.e. screen printing.
0-7803-3829-4/97/$10.00
01
997
IEEE
Depending on the viscosity of the resist, films with
thicknesses of
1
to
500
pm
can be produced in one
single step. The deposition is followed by a softbake
evacuating the solvent. The illumination is done with
a standard UV i-line mask-aligner. After the
exposure, the polymer is crosslinked by a postbake
of several minutes to
e
few hours, depending on the
film thickness and the curing temperature.
The development of the photoresist is done at
room temperature using a PGMEA (propylenglygol-
monomethylether-acetate) solution. An ultrasonic
bath may be employed in order to reduce the
development time.
FABRICATION PROCESS
The use of the SU-8 photoplastic allows the
fabrication of monolithic auto-assembled channels for
microfluidic applications. In this work we discuss
two of the technologies we have developed. The first
one is called the mask-process whereas the second is
named fill-process.
The first step in both technologies is the
deposition of a first layer of SU-8 on a thermoplastic
substrate (figure 1.a). This allows the structure to be
released easily once its fabrication is finished. After
the exposure and the postbake of this first layer, a
second SU-8 layer is spun over the surface,
illuminated and polymerised (figure
1
.b).
In the fill-process, the structure is now
developed. The resulting channel is filled with a
material, that can easily be dissolved (figure 1 .cl).
A
third layer is deposited and processed. In a last step,
the SU-8 is developed, the filling material is
dissolved, and the structure with the micro-channel is
released from the substrate. Different materials have
been tested as filling material. These were
thermoplastics, waxes and epoxies. The latter exhibit
the best behaviour. In a first attempt we tried non-
photosensitive SU-8. The results are displayed on
figure 2.a. It was possible to open the channel,
TRANSDUCERS
'97
1997
lnfernafional
Conference
on
SoMSfate
Sensors
and
Acfuafors
Chicago,
June
16-19, 1997
1419
4
D2.06
P
a
b
cl
c2
I
1
I
I
I
1
dl
d2
1
1
el
e2
Figure
I:
Fill- and mask-process-flow for micro-
channel fabrication. Light grey: nonpolymerized
SU-
8;
middle guey: polymerised
SU-8;
dark grey: filling
material; black: metal.
but we observe that the photoinitiator has penetrated
into the filling material, thus decreasing the height of
the channel
(-75
pm
instead of 100 pm). By using
other epoxies (for example the AralditeTM GT6063
from Ciba-Geigy), it was possible to open the
channel in its whole height (see figure 2.b for a
channel of 50x50
pm2)
For the mask-process, the second
SU-8
layer is
not developed after step b) (see figure
l),
but a metal
is deposited over the whole structure and is patterned
in order to stay just on the top of the channel (figure
1.~2).
A third SU-8 layer is spun on the surface and
illuminated. The metal mask prevents the
photosensitive
SU-8
inside the channel to be exposed
and crosslinked during this process-step (figure
1
.d2). Finally the photoresist is developed outside
the structure and inside the micro-channel. The last
step is the liberation of the structure from the
substrate (figure 1 .e2). Figure
3
illustrates the results
of the mask-process. Figure 3.a shows a completely
opened channel with a cross section of 50x100 pm2.
The metal mask is still present. But the mask is
weakly fixed and it is easy, by using a ultrasonic bath
during the development, to remove it.
With both technologies, channels of cross-
sections as small as 25x50 pm2 and lengths of more
than 10 mm have been realised. The mask-process
is
more complicated and more expensive than thefill-
Figure
3:
Micro channels realised by the mask-
process. (a) opened channel with metal mask;
(b)
opened channels without metal mask.
Figure
2:
Micro channels realised
by
the fill-
process. (a)
fill
material: nor-photosensitive
SU-8;
(b)
$11
material: Araldite
TM
GT4043.
Figure
4:
2-level micro-channel structure.
1420
TRANSDUCERS
'97
1997 lnfernational Conference on Solid-state Sensors and Actuators
Chicago, June 16-19. 1997
4D2.06P
cation process. This assembly procedure is an
additional and sometimes complicated manufacturing
step. By using a thick-film, high aspect ratio
photoplastic, monolithic flow sensors which require
no assembly, can be produced. This greatly reduces
the fabrication time and costs, thus allowing
inexpensive high volume production of such devices.
Figure
8:
Photoplastic flow-sensor: (a) opened
channel with the electrical resistor;
(b)
cross-section
of the channel.
SU-8
layer (step (b) of figure l), a metal layer is
deposited and patterned in order to realise the
electrical resistors. The materials used are chromium
(100
nm) and gold
(50
nm). The resulting resistors
are about
20
to
60
Ohms. Figure 9 shows a finished
flow-sensor.
Silicon is a material often used in the fabrication
of thermal flow sensors. Due to the high thermal
conductivity k of the silicon, a lot
of
generated heat is
lost into the substrate, which strongly reduces the
sensor's sensitivity. Special techniques such as
suspended resistor bridges with the resistors or the
deposition of the resistors
on thermally isolating
layers (Si02) have to be used. By employing a
polymer substrate with a very low thermal
conductivity, the heat losses can be greatly
decreased. Numerical simulations (ANSYS) have
shown the required heating power given by the
resistors are at least of one order of magnitude lower
in the case of a photoplastic
(k=0.2
W/(mK)) flow-
sensor, compared with a similar device made of
silicon (k=130 W/(mK)).
Furthermore, classical flow-sensors made of silicon
and/or glass are normally realised in different parts
which have to be assembled at the end of the fabri-
Figure
9:
A finished monolithic low-cost
photoplastic thermal
flow-sensor
(grad.:
1
mm)
The monolithic photoplastic flow-sensor has been
realised and is currently being tested. First
preliminary electrical tests and measurements show a
good behaviour of this photoplastic sensor. More
details of its characteristics will be described
elsewhere.
CONCLUSION
In this work we present a simple and low-cost
technology for the realisation of embedded micro
channels. We proved the feasibility of the method by
the production of a thermal flow-sensor. In the
future, we will apply this technique to the design and
fabrication of other devices such as micropumps or
such as the microfluidic part of a bio-chip.
ACKNOWLEDGEMENTS
This work couldn't have been done without the
enormous development work on the SU-8
photoresist done by the researchers of IBM-
Yorktown (Ms.
N.
LaBianca), IBM-Zurich (Dr. M.
Despont) and EPFL-IMS (Mr.
H. Lorenz). The
authors thank N. Panchaud for the many useful
discussions.
This paper is dedicated to the memory of Dr. M.
Dutoit (1941-1996).
REFERENCES
[l]
H.
Lorenz et al, EPON
SU-8:
A
Low
Cost
Negative Resist
for
MEMS,
Proceedings of
MME'96 (Micro Mechanics Europe),
Barcelona, Spain, October 1996.
M. Despont et al, High-Aspect-Ratio,
Ultrathick, Negative- Tone Near-
UV
Photoresist
for
MEMS
Applications
Proceedings of MEMS'97, Nagoya, Japan,
January 1997.
[2]
TRANSDUCERS
'97
1997 lnternational Conference
on
Solid-State Sensors and Actuators
Chicago,
June 16-19. 1997
1421
4D2.06P
proccss because of the additional metallisation of the
device. But unlike the latter it allows the realisation
multilevel channels. Figure
4 shows
a
structure with
channels on
2
different levels.
The dissolution of the material inside the channel
is an important step. We modelize this step by
expressing the length of developed channel as a
function of the development time.
where
L
is
the developed length at a time t,
D
and
C,,,
are respectively the diffusion constant and the
maximum concentration of the photoresist in the
developer; p is the density of the photoresist and
a
is
a constant depending on the geometry of the
channel’s cross-section and taking into account the
flow of a liquid inside the channel. For a large micro-
channel (cross-section
>
100x100
pm2)
a
is
equal to
1. Figure
5
illustrates experimental measurements.
They are in good agreement with the above simplified
model.
During the polymerisation, the
SU-8
photoresist
undergoes a shrinkage due to the crosslinking of the
molecules. This shrinkage can be observed on figure
3
at the top of channel wall. A more detailed
schematic view
is
shown on figure
6.
After
illumination and before the postbake the two surfaces
of the polymerised and the nonpolymerised
photoresists
are
on the same level. After the
4000
k
3000
1000
75v
/
L=340.7
50
pn
c
/
yo
L=331.2t04’
25
pn
L=71.6t03’
0
50
100
150
time (minutes)
Figure
5:
Simplified model (lines) and experimental
data (symbols)
for the developed channel length as a
jimction
of
time. The height
of
the channels are
100
pm
and the respective widths are
25,
50
and
75
pn.
surface
of
illuminated
SU-8
/
surface of illuminated interface
and postbaked
SU-8
Figure
6:
Shrinkage
of
polymerised photoresist.
postbake, the surface of the polymerised
EPON
is
below the nonpolymerised one. The difference
between these heights is a direct indication
of
the
shrinkage of the polymerised photoplastic. In our
case we found
a
shrinkage
of
about
7.5
%.
APPLICATIONS
In order to show the feasibility
of
the new micro-
channel fabrication technology described in this
work, a thermal flow-sensor has been designed and
realised (see figure
7).
The working principle of this
device is well known. Three electrical resistors are
placed inside a channel. The inner resistor is heated
and the two other ones measure the temperature
difference due to a liquid flow. The channel of the
prototype is
5
mm long and has a cross-section
of
300
x
300
pm2
(see figure 8).
The mask-process used for the fabrication of this
device is almost identical to the one described above.
However before the spinning of the second
-Photoresist layer
3
Photoresist layer
2
’hotoresist layer
1
Microchannel Emb2dded Resistor
Figure
7:
Schematic view
of
a monolithic
photoplastic
flow
sensor.
TRANSDUCERS
’97
1997
international Conference on Solid-state Sensors and Actuators
Chicago, June
16-19, 1997
1422