Sensors and Actuators A 108 (2003) 103–107

3-D microarrays biochip for DNA amplification

in polydimethylsiloxane (PDMS) elastomer
Xiaomei Yu a,∗ , Dacheng Zhang a , Ting Li a , Lin Hao b , Xiuhan Li a
a Institute of Microelectronics, Peking University, Beijing 100871, China
b Department of Hematology, Peking University First Hospital, Beijing 100034, China
Received 29 July 2002; received in revised form 3 July 2003; accepted 9 July 2003

Abstract

In this paper, a rapid and low-cost fabrication technique for microarrays biochips is presented, and the fluorescence-based polymerase
chain reactions (PCR) were performed in the chip for DNA amplification. The chip is consisted of 1064 column chambers. Each chamber
has a volume of 25 nl, with the diameter of 460 ␮m and depth of 200 ␮m. The PCR chips were made of inexpensive silicone elastomer—
polydimethylsiloxane (PDMS) and rapidly fabricated with micro-molding technique. The three-dimensional (3-D) PDMS replicas were
molded directly from 3-D negative silicon master, which was fabricated by deep etching technique under the inductively coupled plasma
(ICP) system. We demonstrated that the fluorescence-based PCR could be performed in the PDMS chip with nanoliter-volumes.
© 2003 Elsevier B.V. All rights reserved.

Keywords: Polymerase chain reaction (PCR); Polydimethlysiloxane (PDMS); Inductively coupled plasma (ICP)

1. Introduction Northrup et al. [1,2], scientists have made some advance-

Recent developments in miniaturization technology
have already had a great impact on the biomedical re-
search field. The MEMS-based polymerase chain reaction
(PCR) chamber is one of the great progresses. PCR is a
well-characterized method for the selectively identical repli-
cation of DNA molecules. By an enzymatic amplification
process, the concentration of a DNA species is nearly dou-
bled through three different temperatures. In this way, The
DNA concentration can be multiplied more than a million
times by about 30 cycles of temperature. It has become an
important analytical method for the detection of diseases.
As we know, the conventional PCR systems use plastic-
based tubes with relatively big volume (>25 ␮l) as reaction
chamber. Therefore they always waste much more reagents
and need longer time to reach thermal balance because their
big volume tubes and slow thermal cycling. Micro-electro-
mechanical systems (MEMS) technology have been applied
in analytical chemistry and biomedical-related fields for
a number of years. The microliter or nanoliter PCR chips
have been developed by using microfabrication techniques.
Since the first report on the microfabricated PCR chip by
∗ Corresponding author. Tel.: +86-10-62759370;
fax: +86-10-62751789.
E-mail address: xmyu@ime.pku.edu.cn (X. Yu).
ment for the DNA amplification with MEMS-based PCR
reactors [3–6]. The potential merits of miniaturized reactor
are automation, rapid temperature cycle, smaller amounts
of sample and reagent, high sensitivity, portability and
mass-productivity.
The conventional substrates of making three-dimensional
(3-D) PCR chip are silicon, glass or quartz, which origi-
nally developed for integrated circuit industry. Therefore,
the fabrications of the chip are a little difficult, time con-
suming and high-cost. Recently, the replica molding tech-
nique using organic polymers has been widely used to
simplify the fabrication process of micro-fluidic devices
[7–10], such as the plastic, the Epon SU-8 photoresist, the
polymethylmethacrylate (PMMA), and the polydimethyl-
siloxane (PDMS). Among these polymers, PDMS is one
of the most promising materials for the microscale mold-
ing. PDMS is a low-cost material. The advantages of
using PDMS to fabricate chip include the submicrometer
replica fidelity, easy fabrication and bonding (self-adhesion
property), mass replication, better optical properties (trans-
parent at 230–700 nm wavelength) and inactive to most
bio-molecules. Therefore, PDMS is particularly suitable to
be used in the fabrication of biochips. The commonly used
procedures fabricating PDMS chip are: (1) fabricating a
negative master; (2) PDMS mixture is molded against a
microfabricated negative master; (3) curing and peeling off

0924-4247/$ – see front matter © 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0924-4247(03)00370-4
104 X. Yu et al. / Sensors and Actuators A 108 (2003) 103–107
the replica; (4) the PDMS replica is then sealed with a flat
substrate.
This paper describes a fabrication of three-dimensional
microarrays PCR chip in PDMS elastomer directly from a
3-D negative silicon master, and this work was going at In-
stitute of Microelectronics, Peking University. Raman spec-
troscopy detected the results of fluorescence-based PCR on
the PDMS chip, which was performed at Department of
Hematology, Peking University First Hospital.
2. Design and fabrication
2.1. Design
Many clinical DNA diagnostics require multiple patient
samples to be tested for certain diseases using microarrays
technique. The ability to perform massively assays with
nanoliter quantities of sample/reagent per PCR chamber
would provide substantial cost saving, and hence is highly
desirable. We designed our PCR chip with 1064 reaction
chambers in a 31.5 mm × 25.7 mm chip. In order to make
the reagents to flow inside the chips smoothly, and vent the
air bubbles out of the chambers as many as possible, column
chambers and inclined sub-channels have been designed.
The layouts of two designed chips are shown in Fig. 1.
The reaction chambers are designed in column with the
diameter of 460 ␮m and the depth of 200 ␮m. The volume of
each chamber is calculated to be 25 nl. The interval between
the two chambers is 800 ␮m. Main channels join the ports
and the all sub-channels, and the sub-channels connect all
chambers. One port is designed for the liquid inlet and two
ports are for the liquid and air outlet. The inlet and outlet
sub-channels drain the solution into or out off the chambers.
The sub-channels are designed in line (Fig. 1a) or parallel
(Fig. 1b), with the flow direction in an angle of 45◦ (Fig. 1a)
or 60◦ (Fig. 1b) with the main channels, respectively. The
width of the main channels and the sub-channels are 160 ␮m
and 70 ␮m, respectively, and the depth of all channels is
50 ␮m. The capillary force and syringe pressure draw the
reagents into the reaction chambers in the loading process.
The outlet ports were designed open to air, therefore the air
main-channel
chamber
sub-channel
(a) (b)
Fig. 1. Layouts of two designed biochips. The sub-channels are (a)
designed in line and in an angle of 45◦ with the main channels; (b)
parallel with each other and in an angle of 60◦ with the main channels.
inside the chambers and the redundant liquid can be vented
out easily in the unloading process.
2.2. Fabrication of negative silicon master
PDMS micro-molding techniques have been used to fab-
ricate micro-fluidic devices for a few years. The PDMS
structures are usually molded PDMS against SU-8 master.
The SU-8 master with a thickness up to 100 ␮m was fab-
ricated using SU-8 thick film photolithography. With this
technique, 2-D patterned layers are usually fabricated. In or-
der to realize 3-D structures, two or more 2-D patterned lay-
ers need to be stacked together and aligned very accurately.
Apart from the accurate aligning, the thickness limitation
of SU-8 master is also a problem in making deep reaction
chambers, which can avoid the contamination among cham-
bers during the PCR. The 3-D silicon master can be etched
sufficiently deep with inductively coupled plasma (ICP)
etching system, and the aligning of two 2-D chips is also
avoided.
The fabrication of silicon master starts from 525 ␮m thick,
4 in. silicon wafers in two mask processes. A schematic il-
lustration of fabricating negative silicon master is shown in
Fig. 2(a)–(e). After wafer cleaning, thermal silicon dioxide
has been grown as the first mask layers. The thermal sili-
con dioxide has a selectivity of about 200 to silicon under
our ICP system, therefore 400 nm SiO2 had been grown
to support the channel etching of 50 ␮m. After the first
photolithography, which forms the patterns of both chan-
nels and chambers, the second mask layer was deposited.
Aluminum and photoresist had been tested as the second
mask layer in the master fabrication. It was found that
aluminum mask layer gives rise to roughening of silicon
surface during the ICP deep etching due to the formation of
Al-protected regions on the surface of Al by mask erosion.
Photoresist was tested to be a good mask layer although it
has a low selectivity of 70 to silicon under our ICP etch-
ing system. The selected thickness of 3 ␮m photoresist is
tested to be sufficient to support the chamber etching of
200 ␮m.
After the second photolithography, the silicon apart from
those, which are on the chambers and the channels, were
etched away to a thickness of 50 ␮m. In order to obtain
highly anisotropic and high-aspect-ratio etching, ICP etch-
ing system (multiplex ICP STS) was used in the silicon
etching. The silicon etching parameters of the ICP system
are: SF6 of 130 sccm, O2 of 19.5 sccm, C4 F8 of 100 sccm,
coin rf forward power of 600 W, platen rf forward power of
13 W, chamber process pressure of 0.2 Pa. After above etch-
ing, the residual SiO2 on the channel surface was removed
in BHF. Then the silicon etching was carried continuously
until the needed chamber depth of 200 ␮m was reached. Re-
move the residual photoresist and SiO2 on chamber surface.
The finished negative silicon master was further thermally
oxidized to obtain relatively smooth surface. Fig. 3 gives a
SEM photograph of a negative silicon master.
 X. Yu et al. / Sensors and Actuators A 108 (2003) 103–107
SiO2
(a)
(b)
photoresist
(c)
(d)
(e)
PDMS
(f)
chamber channel
(g)
Fig. 2. Fabrication processes of silicon master and PDMS replica. (a)
The silicon wafer was thermally oxidized to form a 400 nm SiO2 . (b)
First photolithography, the SiO2 both on channels and chambers were left
over. (c) Three micrometer thick photoresist layer was spun on wafer,
and then was patterned with chamber mask. (d) The silicon was etched in
ICP system, the etching depth is 50 ␮m. (e) After removing the SiO2 on
the surface of the channels, and another 150 ␮m silicon was etched away
by ICP system. (f) The PDMS prepolymer mixture was poured onto the
master. (g) After curing and peering off the PDMS, the PDMS replica
and the PDMS plate were bonded together.
2.3. Fabrication of PDMS replica
Fig. 2(f) and (g) illustrate the molding procedure of PDMS
replica. In order to peeled off the PDMS replica easily, the
negative silicon master was first coated with a fluorocarbon
layer polymerized by CHF3 plasma in a reactive ion etch-
ing (RIE) system for 5 min under the following conditions:
CHF3 gas flow of 60 sccm, pressure of 2 Pa, and power of
100 W. The silicon master was then placed in a petri dish of
Fig. 3. The SEM photograph of a negative silicon master.
105
Fig. 4. The SEM photograph of a PDMS replica.
about 2 mm in depth to ensure the PDMS chip to be thick
and rigid enough. The PDMS prepolymer (SYLGARD 184
Silicone Elastomer Kit, Dow Corning) and the curing agent
were thoroughly mixed in a 10:1 weight ratio and was de-
gassed with a mechanical vacuum pump for 10 min to re-
move any air bubbles in the mixture and to insure complete
mixing between the two parts.
After the PDMS was poured onto the silicon master, the
patterned PDMS was put into the vacuum system again
for another 10 min to degas the possible bubbles between
the interface of the silicon master and the PDMS mixture.
The PDMS prepolymer was cured for 1 h at 65 ◦ C in oven
and then the PDMS replica was peeled off from the sili-
con master. Fig. 4 shows the SEM photograph of a finished
3-D PDMS replica. Another PDMS plate was molded in a
smooth dish as the top cover using the total same fabrication
processes as the PDMS replicas. The soft PDMS allows in-
let/outlet ports to be created easily by pushing syringe nee-
dles into the top PDMS plate.
Four chips have been designed in a 4 in. silicon wafer, so
the peeled PDMS chips need to be cut after the whole wafer
molding. Narrow dicing lines have been designed on the
silicon masters; therefore the replica can be diced accurately
along the dicing lines with a sharp-edged razor. The accurate
dicing of the PDMS chips will also confirm the later accurate
and automatic dispersing of DNA samples by an automatic
spotter system.
2.4. Bonding and hydrophilicity
Oxygen plasma treatments for PDMS surface activation
were carried out using RIE system in order to receive the ir-
reversible bonding between the cured top and bottom PDMS
parts. Experiments were performed many times to find the
optimum condition of plasma treatment, which are: rf power
of 100 W, oxygen gas flow of 80 sccm, pressure of 0.8 Pa,
and treatment time of 40 s. Plasma-treated samples were
immediately bonded together with activated surface facing
each other, and then baked in oven at 100 ◦ C for 2 h. Since
no patterns exist on the top cover, the two PDMS plates
need not to be aligned accurately. During the baking, small
external force was applied on the chip to assist the bonding.
106 X. Yu et al. / Sensors and Actuators A 108 (2003) 103–107
The bonding quality was examined via pulling the PDMS
pieces apart from each other. Fractures took place most on
PDMS top plate or replica, a little at the interface of the
two PDMS plates. We conclude that under the optimum
treatment condition of oxygen plasma and the baking, the
two PDMS plates can be bonded together successfully.
PDMS is a hydrophobic material after peering off. The
hydrophobicity of PDMS makes the DNA reagent difficult
to flow inside the channels. It is found that the PDMS
elastomer rapidly lost its hydrophobicity if it is treated
in oxygen plasma treatment for a short time. In our ex-
periment, the oxygen plasma activation on the surface of
PDMS was performed at the same time with the bond-
ing processes. Water flow test shows that the surfaces of
PDMS chips were changed from hydrophobicity into hy-
drophilicity easily under above oxygen plasma treatment
condition.
3. Testing and discussion
Flow characteristics of the PDMS chip were first evaluated
with color water. The liquid was directly injected into inlet
port from the top PDMS plate by syringe because the PDMS
is soft enough to be pierced. The treatment of oxygen plasma
makes the liquid flowing into most of the chambers smoothly
under the syringe pressure in 1–2 s. The outlet ports can vent
the bubbles out of the chips and make the liquid flow easily.
No significant difference of flow characteristics was found
for the two designed chips.
To prove fluorescence-based PCR can be performed
through the PDMS chips, we carried out both the fluore-
scence-based PCR in the PDMS chips and routine PCR in
tubes. The samples selected to do the comparison are the
DNA from a six members family and two male newborns.
The purpose of the experiment is to identify point muta-
tions of two patients and diagnose hereditary sideroblastic
anemia in this family. The point mutations of G514A in
the exon 5 of ALAS2 in a hereditary X-linked sideroblas-
tic anemia family was confirmed by PCR single-stranded
conformational polymorphism (SSCP) and sequencing.
The temperature cycle was provided with a commercial
thermalcycler controller (Gene Amp PCR System 2400,
Perkin Elman, USA). The contrast experiments between
the tubes and the PDMS chips were performed many times
with the same probes and reagents. It is found that when the
temperature display on the controller is selected the same,
the same reaction results were obtained for both methods.
That is to say the display temperature are the temperature
inside reaction chamber of the PDMS chip, and the tem-
perature uniformity inside the reaction chamber and from
chamber to chamber can be certain although the PDMS
is not a good thermally conduction materials. With this
thermalcycler controller, the circle time for our chip were
decreased down to satisfied with the requirement of smaller
amounts of reagent and sample.
The PCR reaction mixture contained 1 ␮l DNA template,
2 ␮l MgCl2 , 1 ␮l of each dNTP, 1 ␮l of each primer, and
2.5 ␮l Tris–HCl. In the experiment, a Taqman probe specif-
ically to detect a point mutation was designed and used
in fluorescence-based PCR. The concentration of Taq poly-
merase was 5 U/␮l. The primer set used to the amplifica-
tion of DNA was 5 -CCATAAGTCAATGACTGAGC-3 and
5 -AAGTGGTGATAGAACCCAAG-3 . The PCR reaction
was first performed in tubes. The PCR products were then
analyzed by conventional slab gel electrophoresis. The elec-
trophoresis was carried out 7 h at 300 V. Ten percent poly-
acrylamide gel (PAGE) was used. Analyzing the ALAS2
gene of the six members of sideroblastic anemia family by
PCR–SSCP analysis, the point mutation of G514A in the
exon 5 of ALAS2 for the patients, two brothers, were con-
firmed comparing their PCR products with normal male
newborns. Their father and grandfather have the same elec-
trophoresis bands with the normal male newborns. But their
mother and grandmother electrophoresis bands are different
from the normal male newborn. That is to say the abnormal
gene come from maternity.
To verify the performance of the PDMS chip array, the
same reaction mixtures were loaded into the PCR chip
simultaneously. The PCR of both units were set at 30 tem-
perature cycles. The temperature steps used in fluorescence-
based PCR consisted of following three steps: denaturation
at 94 ◦ C for 30 s, primer annealing at 65 ◦ C for 30 s and
primer extension at 72 ◦ C for 30 s. The amplified products
were analyzed by Renishaw 1000 cofocusing Raman spec-
troscopy (Renishaw, USA) after the thermal cycling. The
fluorescence-based PCR from the PDMS chip show the
same results as routine PCR. The detected Raman signals
from the two patients, the mother, and the grandmother
are positive, which show absorb at 570 nm. No absorb
peaks were observed for the PCR results from the father,
the grandfather, and the normal male newborns, the point
mutations were not identified for them. One of the com-
parison results was show in Fig. 5. In this figure, number
1 is a positive result from a brother, number 3 is a negative
Fig. 5. Scanned fluorescent PCR spectra on a PDMS chip.
 X. Yu et al. / Sensors and Actuators A 108 (2003) 103–107
result from the father, and the line number 0 is a negative
comparison from a male newborn.
A shorter temperature cycle time was tested for the
fluorescence-based PCR on PDMS chip. When the holding
time at each temperature step was set to 15 s to accomplish
the 30 temperature cycles, the DNA was successfully am-
plified, and the reaction efficiency of the fluorescence-based
PCR almost did not change. When the temperature step was
set to 8 s to accomplish the cycles, the signal peak intensi-
ties get reduced quickly. Based on our fluorescence-based
PCR experiment, it is demonstrated that the PDMS poly-
mer is biocompatible in a DNA PCR assay and the Taq
fluorescence-based PCR can be finished in PDMS chip.
4. Conclusions
We describe a fabrication technique of 3-D microar-
rays biochip in PDMS for thermal cycling of DNA by
fluorescence-based PCR. The PDMS mixture was molded
directly from 3-D silicon master, which was etched under an
ICP system to obtain sufficient deep reaction chamber. With
the techniques of silicon MEMS, PDMS micro-molding
and PDMS bonding, PDMS chip composed of 1064 re-
action chambers, each of the chambers has about 25 nl in
volume, were made. The identification of point mutations
in the anemic family and two normal male newborns using
the Taqman probe and the fluorescence-based PCR on the
PDMS chip show the same results as those performed with
routine PCR method. This proves that the PDMS is a bio-
compatible material and the fluorescence-based PCR could
be successfully carried out with the PDMS chip. The PDMS
microarrays chips can perform massively assays of DNA
amplification with nanoliter quantities of sample/reagent
per assay, which would provide substantial cost saving,
time saving and powerful analysis. Considering the many
advantages of microarrays PDMS chip, this fabrication
technique is suited for a wide range application in biochip
and micro-total analysis system (␮-TAS).
Acknowledgements
This work is founded by National Natural Science Foun-
dation of China (founded no. 90207013) and 985 project of
Peking University, China.
References
[1] M.A. Northrup, M.T. Ching, R.M. White, R.T. Watson, DNA am-
plification with a microfabricated reaction chamber, in: Proceedings
107
of the Seventh International Conference on Solid-State Sensors and
Actuators, Yokohama, Japan, June 1993, p. 923.
[2] M.A. Northrup, C. Gonzalez, D. Hadley, R.F. Hills, P. Landre,
S. Lethew, et al., A MEMS-based miniature DNA analysis sys-
tem, in: Proceedings of the Eighth International Conference on
Solid-State Sensors and Actuators, Stockholm, Sweden, June 1995,
p. 764.
[3] A.T. Wooley, D. Hadley, P. Landre, A.J. de Mello, R. Mathies, M.A.
Northrup, Functional integration of PCR amplification and capillary
electrophoresis in a microfabricated DNA analysis device, Anal.
Chem. 68 (1996) 4081–4086.
[4] J. Cheng, M.A. Shoffner, G.E. Hvichin, L.J. Kricka, P. Wilding,
Chip PCR. II. Investigation of different PCR amplification systems
in microfabricated silicon-glass chip, Nucleic Acids Res. 24 (1996)
380.
[5] M.U. Koop, A.J. de Mello, A. Manz, Chemical amplifications:
continuous-flow PCR on a chip, Science 280 (1998) 1046.
[6] S. Kondo, K. Kano, E. Shinohara, Reaction chamber for PCR made
by microfabrication, in: Proceedings of the 10th International Con-
ference on Solid-State Sensors and Actuators, Japan, June 1999,
p. 1718.
[7] E. Delamarche, A. Bernard, H. Schmid, B. Michel, H. Biebuyck,
Patterned delivery of immunoglobulins to surface using micro-fluidic
networks, Science 276 (1997) 779.
[8] G.J.M. Bruin, A. Paulus, M. Ehrat, Integrated capillary electrophore-
sis on flexible silicone microdevices: analysis of DNA restriction
fragments and detection of single DNA molecules on microchips,
Anal. Chem. 69 (1997) 3451.
[9] D.C. Duffy, O.J.A. Schueller, S.T. Brittain, G.M. Whitesides, Wafer
level batch transfer process of RF MEMS passive device using
PDMS, J. Micromech. Microeng. 9 (1999) 211.
[10] B.-H. Jo, L.M. Van Lerberghe, K.M. Motsegood, D.J. Beebe,
Three-dimensional micro-channel fabrication in polydimethyl-
siloxane (PDMS) elastomer, J. Microelectromech. Syst. 9 (2000)
76.
Biographies
Xiaomei Yu, Postdoctoral, Institute of Microelectronics, Peking Uni-
versity, China. PhD was obtained from Beijing University of Areo-
nautics and Astronautics in 2001. Her current research interests in-
clude the design and fabrication of PCR biochip and micromechanical
biosensor.
Dacheng Zhang, Professor, Institute of Microelectronics, Peking Univer-
sity, China, received MS degree in 1998 at Peking University. His work
is focused on processes and techniques of MEMS.
Ting Li, Associate Professor, Institute of Microelectronics, Peking Uni-
versity, China, received BS degree in 1991 at Beijing University of Tech-
nology. His work is focused on processes and techniques of MEMS.
Lin Hao is a graduate student at Peking University First Hospital, China,
majoring in molecular biology.
Xiuhan Li is a graduate student at Peking University, China, majoring in
microelectronics.