Sensors and Actuators A 136 (2007) 178183

Thermal control of micro reverse transcription-polymerase

chain reaction systems
Nan-Chyuan Tsai , Chung-Yang Sue
Department of Mechanical Engineering, National Cheng Kung University, No.1, Ta-Hsueh Road, Tainan 701, Taiwan
Received 8 June 2006; received in revised form 13 October 2006; accepted 24 October 2006
Available online 4 December 2006

Abstract
Being beneficial from dramatic progress in biochemical and micro-electromechanical technologies, traditional DNA manipulation devices tend to
be miniaturized to speed up detection processing. Polymerase chain reaction (PCR) and reverse transcription-polymerase chain reaction (RT-PCR)
are two typical examples of them. In this report, micro RT-PCR (RT-PCR) is designed to quantitatively detect viruses. Test samples reservoirs,
RT-PCR, and capillary electrophoresis are integrated on a SU-8 based monolithic chip. A high-precision temperature control unit is well developed
by embedding amplification circuits and an Intel 8051 microprocessor. The integrated system exhibits high efficacy of heat transfer, exemption
of fluid flow clogging and adequate sensitivity to, by feedback, control the cycle of detection for the infected malignant tissues via intensive
simulations.
2006 Elsevier B.V. All rights reserved.
Keywords: RT-PCR; Capillary electrophoresis; SU-8; Feedback control

1. Introduction
Polymerase chain reaction (PCR) in DNA amplification was
ever profoundly studied and reported by Mullis et al. in 1985 [1].
PCR is mainly to duplicate a certain fragment sequence of DNA
in million times precisely. The amplification process of PCR
can be divided into three consecutive temperature sectors: (A)
denaturation: the double-strand DNA segment is disassembled
into two singles at high temperature 94 C, (B) annealing: the
single-strand DNA is attached to a specific primer at 54 C, (C)
extension: the associated complementary DNA (cDNA) singles
with primers are extended into double-strand DNAs. This paper
is an extended research report from PCR to reverse transcription
PCR (RT-PCR). With RNA as the template of RT-PCR, the first
strand of the corresponding cDNA is synthesized by a set of
artificial primer and certain particular enzyme. Right afterward,
the cDNA is used as the template for PCR, thus a great deal of
DNA fragments are duplicated via appropriate transformation
triggered by Taq DNA polymerase. RT-PCR can be applied
for detection of RNA virus, mass production of cDNA, and

Corresponding author. Tel.: +886 6 2757575x62137; fax: +886 6 2369567.

E-mail address: nortren@mail.ncku.edu.tw (N.-C. Tsai).

0924-4247/$ see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.sna.2006.10.057

quantitative analysis of infection caused by DNA virus. A few

illnesses do not stem from DNA virus (e.g., Hepatitis C [2]) so
that PCR detection process cannot be directly employed. They
have to be reversely transcribed to the first-strand cDNA at first,
before PCR process can be undertaken. The manipulation types
of Chip-PCR can be categorized into continuous-flow [27] and
stationary [8,9] PCR. The type of stationary PCR is processed
in a chamber that can provide heating and cooling cycles but
samples are not flowing at all. The samples for the type of
continuous-flow PCR dose flow through a few heated zones,
individually with specified temperature curves, to complete the
full reaction cycle. The first Chip-PCR was reported by Northrup
et al. and it was basically a stationary type [9]. They utilized
silicon as the substrate and included heater, sensor, and reaction
chamber in a single chip by MEMS technologies. In 1998, Kopp
and Manz proposed a continuous-flow PCR by heating three
reaction zones from externally-added heating blocks [7]. The
continuous-flow micro RT-PCR (RT-PCR) chips, reported in
this paper, are to integrate internally-embedded heaters, thermal
sensors, RT-PCR module, and capillary electrophoresis into
a single detection unit so that it is more compact, efficient,
and reliable. In RT-PCR, it is extremely crucial to control
temperature to follow the required trajectory precisely. Only if
the required temperature is satisfied in each reaction zone, can

N.-C. Tsai, C.-Y. Sue / Sensors and Actuators A 136 (2007) 178183

179

the expected DNA amplification be achieved. In contrast, if the

required reaction temperature is deviated far away, the desired
reaction in each zone would not be completed and the examination result might be incorrect. Since the measured signal,
provided by the thermal sensor, is very weak in magnitude, an
efficient amplification circuit is needed to overcome the possibly
lost fidelity. In addition, the embedded heater also needs an
efficient control circuit to regulate required drive current that is
quantitatively decided by a microprocessor Intel 8051. Finally,
theoretical analysis and simulation results are addressed and
discussed.
2. Design and fabrication
2.1. Design concept
The sample and the reagent together are designed to flow
through four different temperature zones. Because these four
zones require different time durations to complete individual
chemical reaction, the net channel length of each zone is purposely designed to define the time duration for the fluid to pass
through. In this work, the time duration ratio for the four reaction zones is set as 4.5:1:1:2 for RT, denaturation, annealing, and
extension, respectively.
Since the reported RT-PCR unit is composed of two layers
bonded together (the heater in the upper layer but micro fluidic
channels in lower layer), sufficient thermal conductivity, from
heater to fluid, has to be verified and examined. The heater is
fixed but the fluid in the channel flows so that the heat energy is
transferred by both conduction and convection. Before the RTPCR unit is really fabricated, it has to be qualified, in advance,
by commercial software IntelliSuite to ensure the heat transfer
efficacy from heater to microfluidic channels. The simulation
results will be shown and discussed later.

Fig. 1. Schematic diagram of designed Chip-RT-PCR: (a) reservoir of RT, (b)

reservoir of PCR, (c) reaction meanders of RT, (d) reaction meanders for denaturing, (e) reaction meanders for annealing, and (f) reaction meanders for extension.

3. Temperature control system

The desired temperature curve for RT-PCR is shown in Fig. 2.
The so-called temperature control system of RT-PCR consists
of heaters, thermal sensors, amplifiers, a microprocessor, and
associated control circuits. The schematic diagram of control
system is shown in Fig. 3. The heaters and thermal sensors are
fabricated in the top layer, as shown in Fig. 4, to control the temperature of the fluidic flowing in the microchannels fabricated
in the lower layer.

2.2. Fabrication process

Roughly speaking, the fabrication can be divided into two
portions: (A) heaters and sensors and (B) reservoirs and reaction meanders. At first, poly-methyl methacrylate (PMMA) is
assigned to serve as the material of the substrate of reservoirs and
reaction meanders and the SU-8 photo-resist is used to make the
main frame of structure. SU-8 is coated on the PMMA surface
and then patterned to construct the reservoirs and microchannels. Heaters and thermal sensors are made of Ti and Pt/Cr,
respectively on the Pyrex 7740 wafer. Both of them are fabricated by means of lift-off process. Finally, SU-8 photo-resist
is used to serve as the isolation between individual heater blocks
and sensor blocks as well. The fabrication addressed above is
for the upper layer of the RT-PCR unit. Once the lower layer
is also finished, the SU-8 photo-resist would serve as the bonding material between two layers. Traditionally, thermal diffusion
bonding process is undertaken at 650 C to bond two individual layers. However, severe damage to micro devices, such as
crack or fracture, might be seeded. In order to avoid the risk
of high-temperature sandwiching, SU-8 bonding is employed in
this work Fig. 1.

Fig. 2. Desired temperature control curve.

Fig. 3. Schematic diagram of temperature control system

180

N.-C. Tsai, C.-Y. Sue / Sensors and Actuators A 136 (2007) 178183

Fig. 4. Allocation of heaters and sensors.

3.1. Amplication circuit for sensors

Fig. 6. Amplification circuit of the thermal sensor.

The electrical resistance of any material is varying as temperature changes. The relation of them can be described by:
RT = R0 [1 + (T T0 )]

(1)

where is the temperature coefficient of resistance (TCR) in

1/ C. R0 is the reference resistance at T0 . In this work, the temperature sensors are made of Pt/Cr and the dimension each is
2400 m 100 m 1500 m.
The TCR curve of thermal sensor is calibrated by experiments
and shown in Fig. 5(a). The associated amplification circuit is
designed and shown in Fig. 6. The resistor R4 is behaving as
the thermal sensor. The heater is made of titanium (Ti) with
dimension 1 cm 3.5 cm 1500 m. The corresponding TCR
curve of the heater is shown in Fig. 5(b). The ideal heating power
can be expressed as follows:

P=

V2
R

(2)

where V is the applied voltage and R is the resistance of heater.

R is dependent of material and dimensions:
R=

L
A

(3)

where is the resistivity ( cm); L, the length of heater and

A is its cross section area. In general, the thickness of heaters is
constant, therefore Eq. (3) can be rewritten as follows:
R=

L
L
= Rs
t W
W

(4)

where Rs is the sheet resistance. From Eq. (4), it can be

observed that the resistance is mainly related to geometry of
the heater. Therefore, the resistance can be designed in geometry in advance. On the other hand, the thermal transfer can be
obtained by the finite element (FE) model:
 


{T }
kt
[0]
{V }
[0] [k ]



 
 
 
{Qnd }
{Qc }
{Qg }
{Qc }
{Qj }
=
+
+
+
{0}
{0}
{0}
{0}
{I nd }
(5)

Fig. 5. (a) R/T curve of the thermal sensor (Pt) and (b) R/T curve of the heater
(Ti).

where [kt ]: thermo-conduction matrix, [kv ]: electric conductivity

matrix, {T}: node temperature vector, {Qnd }: node temperature
vector, {Ind }: node current vector, {Qc }: surface convection,
{V}: node potential vector.
In order to evaluate the performance of designed heaters, Eq.
(5) is used to examine and verify the transformation from applied
control current to resulting temperature by commercial software,
ANSYS.
A high-resolution thermal sensor amplification circuit is presented and shown in Fig. 6. The circuit consists of an OP AD620
and a voltage regulator IC AD581 that provides a precise output
at 10 V from an unregulated input level ranging from 12 to 30 V.
This sensor amplification circuit plays the role of feedback to
microcontroller by an ADC. The output voltage versus resistance

N.-C. Tsai, C.-Y. Sue / Sensors and Actuators A 136 (2007) 178183

181

Fig. 7. Resolution of thermal sensor circuit.

curve is shown in Fig. 7. The resolution of sensor amplify circuit

is about 0.1 (V/ C). This resolution is far higher (five times) than
conventional design such as electrical bridge. In addition, it is
sufficiently sensitive for a 8-bit ADC (the minimum resolution
requirement of 8-bit ADC is about 0.02 V/Step).
4. Simulations
Since the reported RT-PCR is a continuous-flow type and
the chain reactions are subjected to specified temperatures,
effective flow of the fluid, carrying DNA samples, and correct
temperature distribution of four heated zones are the most important two issues to be ensured. The designed RT-PCR, whose
realistic picture is shown in Fig. 8.
The temperature distribution of the top layer is simulated and
shown in Fig. 9, where four heated zones present different tem-

Fig. 10. Thermal entry length of three reaction zones.

perature portraits in color. For each reaction zone, there exists

a so-called thermal entry length before the desired temperature is saturated, as shown in Figs. 10 and 11. The thermal
entry length is mainly function of temperature and flow rate.
The higher temperature to be saturated, the longer the thermal
entry length will be. The thermal entry length becomes shorter
if the flow rate is lower. If the heat transfer from the heaters to
the fluid is concerned, the bounding layer, made of photo-resist
SU-8, has to be examined for its transfer efficacy. From Fouriers
law:
q = kc Ac

T
n

(6)

where q is the heat flux; kc , thermal conductivity of SU-8 (about

0.2 W/mK); Ac , the cross section of heat flux; and T/n is the
temperature gradient in normal direction. Eq. (6) can be further
rewritten in matrix form:
{Qnd } = [Kt ]{T }

(7)

Fig. 8. RT-PCR Chip.

where {Qnd } is heat flux vector, [Kt ] is thermal conductivity

matrix and {T} is the temperature vector. Commercial software
IntelliSuite is used to simulate the distribution of temperature,
based on proper boundary conditions and parameters.
A few assumptions have been made for simulations, such as
(a) flow rate: 50 nl/s, (b) thermal conductivity: 0.643 W/mm C,
(c) density of fluid: 1000 kg/m3 , and (d) viscosity of fluid:
0.0005 kg/ms. After the heat is transferred transversely, heat
convection then succeeds along the microchannels by Newton
Cooling Law:

Fig. 9. Temperature distribution of the top layer of RT-PCR unit.

Fig. 11. Temperature distribution in heated zone.

182

N.-C. Tsai, C.-Y. Sue / Sensors and Actuators A 136 (2007) 178183

Fig. 12. Micrograph of microchannels.

Fig. 15. Gel electrophoresis for Y40 PCR amplification product (marker: 3 l;
sample: 3 l).

the flow is slowed down at the corner but retains a practically

acceptable flow speed. That is, clogging problem can be avoided.
5. Conclusion
Fig. 13. Sideview of flow field within microchannels.

q = hAs (T s T )

(8)

where q is the heat flux; h, the thermal conductivity; As , the

surface of fluid; Ts , the surface temperature; and T is the
steady temperature input. The simulation on heat convection
is conducted by commercial software CFD-RC.
Now, let us consider the fluid flow. The worst thing to conduct
flowing in microchannels is clogging that often results from
incorrect design of channel curvature at corners. The dimensions
of microchannel of RT-PCR are 100 m in width and 50 m
in height respectively, as shown in Fig. 12. The simulation, from
sideview, of flow field in the microchannels is shown in Fig. 13.
The flow field at the turning corners is shown in Fig. 14, where

A full RT-PCR unit is designed, fabricated and verified

by intensive experiments simulations. The advantage of the
reported RT-PCR is that the fluid flow, carrying DNA or
RNA samples, can continuously travel along purposely designed
microchannels that are heated and regulated at specified temperatures, controlled by a microprocessor. In this work, a bit
of test sample Y40 is to be amplified. The preliminary experimental result of RT-PCR is shown in Fig. 15. It is noted
that M represents the marker which behaves like the scale
or ruler. From this gel electrophoresis analysis (only Column
1 is used in this work), it can be clearly seen how serious
the virus defection is by quantity. Intensive simulations are
conducted via commercial softwares, ANSYS, CFD-RC, and
IntelliSuite to ensure efficient heat transfer and effective fluid
flow, especially at the turning corners of microchannels. The
over-heated and insufficient heat transfer are both carefully
examined and ruled out. The clogging problems in microchannels can be avoided since the flow speed is retained above a
certain level. A temperature control system, that consists of
sensors, heaters, of a microprocessor, and related amplification
circuits, is integrated as a compact module. The overall throughputs of the continuous-flow RT-PCR can be greatly increased
in addition to much improvement of precise temperature
control.
Acknowledgements

Fig. 14. Flow field around turning corner.

The authors would like to thank the Center for Micro/Nano

Technology Research, National Cheng Kung University, Tainan,
Taiwan, and National Nano Devices Laboratory (NDL) for
equipment access and technical support. This research was par-

N.-C. Tsai, C.-Y. Sue / Sensors and Actuators A 136 (2007) 178183

tially supported by National Science Council (Taiwan) with

Grant NSC94-2212-E-006-054.
References
[1] R.K. Saiki, S. Scharf, F. Faloona, K.B. Mullis, G.T. Horn, H.A. Erlich, N.
Arnheim, Enzymatic amplification of beta-GLOBIN genomic sequence and
restriction site analysis for diagnosis of sickle cell anemia, Science 230
(1985) 13501354.
[2] Y.C. Lin, M.Y. Huang, K.C. Young, T.T. Chang, C.Y. Wu, A rapid micropolymerase chain reaction system for hepatitis C virus amplification, Sens.
Actuators B 71 (2000) 28.
[3] E.T. Lagally, C.A. Emrich, R.A. Mathies, Fully integrated PCR-capillary
electrophoresis microsystem for DNA analysis, Lab. Chip 1 (2) (2001)
102107.
[4] M. Bu, T. Melvin, G. Ensell, J.S. Wilkinson, A.G.R. Evans, Design and
theoretical evaluation of a novel microfluidic device to be used for PCR, J.
Micromech. Microeng. 13 (2003) S125S130.
[5] C.G.J. Schabmueller, J.R. Pollard, A.G.R. Evans, J.S. Wilkinson, G. Ensell,
A. Brunnschweiler, Integrated diode detector and optical fibres for in
situ detection within micromachined polymerase chain reaction chips, J.
Micromech. Microeng. 11 (1990) 329333.
[6] M.A. Northrup, B. Benett, D. Hadley, P. Landre, S. Lehew, J. Richards,
P. Stratton, A miniature analytical instrument for nucleic acids based on
micromachined silicon reaction chambers, Anal. Chem. 70 (1998) 918922.

183

[7] M.U. Kopp, A. Manz, Chemical amplification: continuous-flow PCR on a

chip, Science 280 (1998) 10461048.
[8] P.J. Obeid, T.K. Christopoulos, Continuous-flow DNA and RNA amplification chip combined with laser-induced fluorescence detection, Anal. Chim.
Acta 494 (2003) 19.
[9] S. Li, S. Chen, Design, simulation, and microfabrication of a heat-conduction
DNA chip with integrated microheaters, J. Manuf. Proc. 6 (1) (2004).

Biographies
Nan-Chyuan Tsai was born in Taiwan in 1963. He received his BS degree
from National Cheng Kung University in 1986 and MS degree in mechanical engineering and electrical engineering from PENN STATE University in
1991 and 1993, respectively. In 1995 he received his PhD degree in Mechanical
Engineering from PENN STATE University. He has been an assistant professor
at National Cheng Kung University since 2003. His research interests include
MEMS/NEMS Technology, Mechatronics, Control Engineering, Active Magnetic Bearings and Biochip applications.
Chung-Yang Sue was born in Taiwan in 1980. He received his BS degree
from Kun Shan University in 2003 and MS degree from National Cheng Kung
University 2005 both in mechanical engineering. He has been in the PhD program
in the field of Bio-MEMS technologies at National Cheng Kung University
since 2005. His research interests include design, fabrication and experiments
of MEMS/NEMS sensors and actuators.