Fabrication of microfluidic devices using
polydimethylsiloxane
James Frienda͒
and Leslie Yeo
Department of Mechanical and Aerospace Engineering, MicroNanophysics Research
Laboratory, Monash University, Melbourne VIC 3800 Australia
͑Received 6 October 2009; accepted 16 October 2009; published online 15 March 2010͒
Polydimethylsiloxane ͑PDMS͒ is nearly ubiquitous in microfluidic devices, being
easy to work with, economical, and transparent. A detailed protocol is provided
here for using PDMS in the fabrication of microfluidic devices to aid those interested
in using the material in their work, with information on the many potential
ways the material may be used for novel devices. © 2010 American Institute of
Physics. ͓doi:10.1063/1.3259624͔
I. INTRODUCTION
Polydimethylsiloxane ͑PDMS͒ has come to be the most widely used material in microfluidic
device fabrication due to its many advantages in fabrication ease, physical properties, and
economy. Using the material to create even complex structures is a simple affair, although achieving
good results requires experience with a reliable technique. The purpose of this brief article is
not to review the many uses of PDMS in its application to microfluidics—this was provided by the
group of Whitesides1
some time ago, not to mention the incredible variety of articles that apply the
material in specific devices2–13
—nor is it to review the many ways PDMS may be used in the
larger field of micro- to nanofabrication, certainly well covered by past researchers.14,15
Recognizing
that entry into microfluidics is predicated upon having fabrication experience and that many
lack it, a simple protocol for fabricating a microfluidic device with PDMS is provided.
II. METHOD
Although there are many methods for fabricating structures in PDMS, we focus on the most
common approach here and encourage the reader to refer to other works16,17
for special needs. The
process outlined here and in Fig. 1 presumes the work is being conducted in a clean environment;
the use of a clean room is not absolutely necessary except for the most demanding fabrication
requirements. However, avoiding surface contamination is essential; Pirahna etch with subsequent
rinse using de-ionized water is ideal for cleaning tools and substrates.
A. PDMS
The classic choice of PDMS for microfluidics is Dow and Corning’s Sylgard 184, a two-part
system with mix ratio of cross-linker/curing agent A: siloxane B=1:10. Increasing the cross-linker
ratio in the mix is known to increase the rigidity of the PDMS produced,18
and some changes to
the cure cycle can be performed to alter the PDMS’ mechanical properties. Further, mixing in
other materials can alter the mechanical properties of the PDMS ͑Ref. 19͒ or offer novel features
not possible with standard PDMS.20
Coating with other media gives similarly broad possibilities.21
a͒
Author to whom correspondence should be addressed. URL: http://mnrl.monash.edu. Electronic mail:
james.friend@eng.monash.edu.au.
BIOMICROFLUIDICS 4, 026502 ͑2010͒
4, 026502-11932-1058/2010/4͑2͒/026502/5/$30.00 © 2010 American Institute of Physics
Idea
PDMS
Mold (for each
mask)
Cast
Mix Sylgard 184
at A:B of 10:1
Degas in vacuum Print mask
Spin-coat SU-8
onto substrate
Pre-bake
UV exposure
Post-bakeThicker?
Develop
Draw up mask
design(s) (B/W or
grayscale), one
layer or multilayer
Pour PDMS into
mold with overlay
(and ports)
Cure and release
Silanize mold
Bonding
Methanol to
delay bonding
Selective
silanization
O2 plasma etch
Contact bond
Integrate inlet/
outlet ports,
additional
functionality
Complete
FIG. 1. Process of making microfluidic devices with PDMS. ͑Enhanced online.͒ ͓URL: http://dx.doi.org/10.1063/
1.3259624.1͔
026502-2 J. Friend and L. Yeo Biomicrofluidics 4, 026502 ͑2010͒
Upon mixing, the PDMS must be degassed in a vacuum chamber for best results, and an
Erlenmeyer flask or large beaker with cover is recommended to avoid a mess from the typically
violent boiling off of the gas from the PDMS. The PDMS may be worked with for around 2 h after
mixing, although its viscosity will gradually increase over this time.
B. Casting
PDMS, as an elastomer, is known for its mold-release properties and ability to replicate
features down to the nanoscale, with low shrinkage during cure ͑around 1%͒ and excellent elastic
properties. Myriad structures may be fabricated casting the PDMS into preformed molds, the
construction of which is an important first step in the process.
1. Molds
Using photolithography to form a mold for casting the PDMS structure requires a mask and
light source to pattern a photosensitive resist polymer to match the features of the mask. Depending
on resolution needs, the mask may be simply a laser-printed overhead transparency ͑1200 dpi
or 250 ␮m resolution͒, a high-resolution print on thin polymer transparency film 10000
dpi/30 ␮m, or a laser quartz photomask print in chromium deposited on the quartz ͑420 000
dpi/600 nm͒. The highest resolution masks are several orders of magnitude more expensive and
restrict the size of the overall microfluidic device to around 100 mm on a side.
Typically the mold is a negative mold, with the PDMS poured into it and filling the regions
left open by the mold. This limits the potential configuration of the cast structure in some specific
ways.
• The cast PDMS must be contiguous if the mold is to be reused. If the mold is to be used and
then sacrificed ͑dissolved away͒, topologies with multiple holes and three-dimensional structures
may be cast.22
• The size of feature’s cast must accommodate the resolution limits of the mask and illumination
system.
• Feature height-to-width aspect ratios of 1:1 are straightforward, while higher aspect ratios
may be achieved either through multiple spin-coat/exposure cycles before final development
of the mold or very expensive Lithographie, Galvanoformung, Abformung ͑LIGA͒.17
Features
with lower aspect ratios than about 1:4 are difficult due to sagging of the PDMS in use;
unless the PDMS is made more rigid it is likely that the feature will not appear properly.
The mold for casting the PDMS may be made of a wide variety of materials, even for
photolithography with many different photoresists available, but we will choose SU-8 ͑SU-8 2000,
MicroChem, Newton MA USA͒; note that even in selecting SU-8, there are many formulations
suitable for forming different layer thicknesses. The SU-8 is the most common molding media
used for PDMS-based microfluidic structure fabrication. Using SU-8 with a single spin coat limits
our maximum mold thickness to around 250 ␮m; here we will aim for a thickness of 200 ␮m.
The process is as follows.
͑1͒ Spin coat SU-8 polymer photoresist as poured onto a wafer substrate; Si is ideal for this
purpose. Spin curves available from the vendor are useful for estimating the appropriate spin
rate to obtain the desired thickness of 200 ␮m; here we ramp up to 500 rpm spin rate and
hold for 15 s to flatten the resist, followed by a 30 s spin at 1250 rpm which gives us a final
film thickness of 205Ϯ3 ␮m. The spin acceleration is 100 rpm/s throughout.
͑2͒ Any bubbles seen in the photoresist must be removed, preferably by degassing the photoresist
prior to use. Heating the resist to 50–60 °C will help.
͑3͒ Prebake ͑soft bake͒ the SU-8 to evaporate its solvent in preparation for exposure. Here we
prebake the SU-8 coated wafer at 90 °C on a polished Al hot plate for 75 min; we ramp up
the temperature to this temperature at 5 °C/min to give improved film adhesion. Generally
the thicker the film, the longer it takes to complete evaporation of the solvent; this represents
one limit in the maximum thickness of the spun-on film.
026502-3 PDMS fabrication Biomicrofluidics 4, 026502 ͑2010͒
͑4͒ Mount SU-8 covered wafer with mask atop it and expose it with UV radiation with a
wavelength of 350–400 nm. The vendor should provide an exposure energy estimate versus
film thickness graph, but it will be at best an estimate and requires some trial and error to
obtain good results. Our approach uses 500 mJ/cm2
of exposure energy on a standard mask
aligner ͑MA-6/UV400, SUSS Microtec, Garching, Germany͒ with 350–400 nm UV light
source.
͑5͒ Making thicker structures can be accomplished by repeating steps 1 to 4 and then continuing
onward.
͑6͒ Postexposure baking aids in cross-linking the exposed portions of the SU-8 in preparation for
its development. This step will also require some trial and error; we ramp up to a bake of 15
min at 90 °C at 5 °C/min and ramp down after this time at the same rate. Care in ramping
the temperature up and down in baking will reduce the appearance of cracks and bowing
from internal stresses.
͑7͒ Development using MicroChem’s SU-8 developer is straightforward, requiring about 16 min
for immersion development, leaving the finished mold for use in casting.
2. Casting, curing, and releasing PDMS from the mold
With the appropriate PDMS on hand, the casting process is as simple as placing the mold in
a heat-tolerant plastic tray or Al boat and pouring PDMS onto the mold. Control of the thickness
may be made by using spacers of the appropriate thickness ͑or the SU-8 itself͒ to come into
contact with a glass or Si overlay placed atop the freshly poured PDMS with a weight placed on
top of the stack to squeeze out excess PDMS. Placing the result in an oven at 65 °C for 24 h cures
the PDMS, which may then be peeled away from the mold and cut as desired. Silanization of the
mold may be needed in some cases where the PDMS sticks to the SU-8 or Si, for example,
exposure to the vapor of dimethyloctadecylchlorosilane in a desiccator for 15 min.23
C. Bonding
The PDMS as cast will not stick to any other structures and it will be strongly hydrophobic.
There are methods for treating the PDMS to reduce its hydrophobicity in addition to other characteristics,
such as gas permeability, but to aid bonding the most common approach is to expose
the surface to be bonded to oxygen plasma for about 10 min. The PDMS may then be placed—as
soon as possible and in less than a minute—against a similarly cleaned Si, glass, or another PDMS
layer to form a permanent bond. To aid in alignment, a small amount of methanol can be used to
keep the surfaces separate for a short time. Selective exposure to liquid silanes can make it
possible to have effective PDMS release in specific regions, i.e., for valves.22
D. Interfacing and integration
Naturally the structure will require inlet and outlet connections; many researchers pierce
especially large open regions within the bonded PDMS structure using needles for temporary
applications, but more permanent applications require hose bibs or grommets cocured into the
PDMS structure. An interesting alternative is the use of vacuum to draw the PDMS against a plate
with prealigned inlet and outlet ports;24
such a bond is strong yet easily released. Integrating
microheaters, sensors, electro-osmotic fluid pumps, and many other devices may be accomplished
with a few additional steps25
to provide inexpensive and complete microfluidic devices with
remarkable capabilities.
1
J. Ng, I. Gitlin, A. Stroock, and G. Whitesides, Electrophoresis 23, 3461 ͑2002͒.
2
J. Wu, W. Cao, W. Wen, D. Chang, and P. Sheng, Biomicrofluidics 3, 012005 ͑2009͒.
3
S. Vanapalli, M. Duits, and F. Mugele, Biomicrofluidics 3, 012006 ͑2009͒.
4
D. Trahan and P. Doyle, Biomicrofluidics 3, 012803 ͑2009͒.
5
D. Luo, S. Pullela, M. Marquez, and Z. Cheng, Biomicrofluidics 1, 034102 ͑2007͒.
6
W. Ng, Y. Lam, and I. Rodríguez, Biomicrofluidics 3, 022405 ͑2009͒.
7
N. Lewpiriyawong, C. Yang, and Y. Lam, Biomicrofluidics 2, 034105 ͑2008͒.
8
L. Liu, W. Cao, J. Wu, W. Wen, D. Chang, and P. Sheng, Biomicrofluidics 2, 034103 ͑2008͒.
026502-4 J. Friend and L. Yeo Biomicrofluidics 4, 026502 ͑2010͒
9
W. Ebina, A. Rowat, and D. Weitz, Biomicrofluidics 3, 034104 ͑2009͒.
10
Y. Srivastava, M. Marquez, and T. Thorsen, Biomicrofluidics 3, 012801 ͑2009͒.
11
X. Gong and W. Wen, Biomicrofluidics 3, 012007 ͑2009͒.
12
S. Das, S. Chung, I. Zervantonakis, J. Atnafu, and R. Kamm, Biomicrofluidics 2, 034106 ͑2008͒.
13
A. Martel, M. Burghammer, R. Davies, E. DiCola, P. Panine, J. Salmon, and C. Riekel, Biomicrofluidics 2, 024104
͑2008͒.
14
M. Bender, U. Plachetka, J. Ran, A. Fuchs, B. Vratzov, H. Kurz, T. Glinsner, and F. Lindner, J. Vac. Sci. Technol. B 22,
3229 ͑2004͒.
15
Y. Xia and G. Whitesides, Annu. Rev. Mater. Sci. 28, 153 ͑1998͒.
16
C. Tamanaha, L. Whitman, and R. Colton, J. Micromech. Microeng. 12, N7 ͑2002͒.
17
K. Kim, S. Park, J. Lee, H. Manohara, Y. Desta, M. Murphy, and C. Ahn, Microsyst. Technol. 9, 5 ͑2002͒.
18
D. Campbell, K. Beckman, C. Calderon, P. Doolan, R. Ottosen, A. Ellis, and G. Lisensky, J. Chem. Educ. 76, 537
͑1999͒.
19
J. Mackenzie, Q. Huang, and T. Iwamoto, J. Sol-Gel Sci. Technol. 7, 151 ͑1996͒.
20
Y. Jung, S. Kar, S. Talapatra, C. Soldano, G. Viswanathan, X. Li, Z. Yao, F. Ou, A. Avadhanula, R. Vajtai, S. Curran, O.
Nalamasu, and P. M. Ajayan, Nano Lett. 6, 413 ͑2006͒.
21
L. Chen, P. Degenaar, and D. Bradley, Adv. Mater. ͑Weinheim, Ger.͒ 20, 1679 ͑2008͒.
22
B. Jo, L. Van Lerberghe, K. Motsegood, and D. Beebe, J. Microelectromech. Syst. 9, 76 ͑2000͒.
23
C. Liu, D. Cui, H. Cai, X. Chen, and Z. Geng, Electrophoresis 27, 2917 ͑2006͒.
24
J. Atencia, J. Morrow, and L. Locascio, Lab Chip 9, 2707 ͑2009͒.
25
D. Erickson and D. Li, Anal. Chim. Acta 507, 11 ͑2004͒.
026502-5 PDMS fabrication Biomicrofluidics 4, 026502 ͑2010͒