performing microchannel temperature cycling reactions using reciprocating reagent shuttling along a...
TRANSCRIPT
Performing microchannel temperature cycling reactions using reciprocatingreagent shuttling along a radial temperature gradient
Ji-Yen Cheng,* Chien-Ju Hsieh, Yung-Chuan Chuang and Jing-Ru Hsieh
Received 21st January 2005, Accepted 4th April 2005
First published as an Advance Article on the web 22nd April 2005
DOI: 10.1039/b501061f
This study develops a novel temperature cycling strategy for executing temperature cycling
reactions in laser-etched poly(methylmethacrylate) (PMMA) microfluidic chips. The developed
microfluidic chip is circular in shape and is clamped in contact with a circular ITO heater chip of
an equivalent diameter. Both chips are fabricated using an economic and versatile laser scribing
process. Using this arrangement, a self-sustained radial temperature gradient is generated within
the microfluidic chip without the need to thermally isolate the different temperature zones. This
study demonstrates the temperature cycling capabilities of the reported microfluidic device by a
polymerase chain reaction (PCR) process using ribulose 1,5-bisphosphate carboxylase large
subunit (rbcL) gene as a template. The temperature ramping rate of the sample inside the
microchannel is determined from the spectral change of a thermochromic liquid crystal (TLC)
solution pumped into the channel. The present results confirm that a rapid thermal cycling effect
is achieved despite the low thermal conductivity of the PMMA substrate. Using IR thermometry,
it is found that the radial temperature gradient of the chip is approximately 2 uC mm21. The
simple system presented in this study has considerable potential for miniaturizing complex
integrated reactions requiring different cycling parameters.
1 Introduction
The polymerase chain reaction (PCR) technique represents a
major advance in modern life sciences. This technique has been
applied to a diverse range of research fields, including
molecular biology, forensic science, and genomics. In recent
years, much effort has been spent in miniaturizing the PCR
technique in order to enhance reaction throughputs while
simultaneously minimizing reagent consumption.1–5 It is anti-
cipated that the successful miniaturization of the PCR process
will enable the extension of this particular technique to a
broader range of clinical analysis processes.
The published literature contains various references relating
to the construction and configuration of miniaturized PCR
devices. For example, studies have addressed different chip
substrate materials and PCR devices with various reaction
vessel format.1,2,6 Briefly, reaction vessels are categorized as
having either a microchannel format or a micro-well format.
Regarding the substrate materials, glass, quartz and silicon (Si)
represent the most commonly employed inorganic substrates,
while organic substrates and reaction vessels are typically
fabricated using poly(tetrafluoroethylene) (PTFE), poly-
carbonate (PC), or poly(dimethylsiloxane) (PDMS). Some of
these substrates provide good optical transparency, while
others have good thermal conductivity.
Many reports have been presented describing different
temperature cycling approaches. There are two basic tempera-
ture cycling strategies. The first strategy involves the direct
heating and cooling of the entire reaction vessel by means of a
temperature-controlled heater and cooler. Typically, strategies
of this type adopt either contact4,7,8 or non-contact9–12
heating/cooling approaches. In the second strategy, sometimes
referred to as spatial-temperature cycling, the sample is heated
and cooled by flowing the sample through a designated
sequence of temperature-controlled blocks.4,13,14
The discrete micro-well format4,8–10,13,14 has the advantage
that the number of temperature cycles can be varied while the
same chip design is used. In addition, extending a single-well to
a multiple-well reaction chip in order to increase throughout is
straightforward15 and can be achieved simply by arranging
multiple reaction wells in an array format. By contrast,
adjusting the number of temperature cycles is more proble-
matic when using a flow-through microchannel reaction chip,
i.e. the microchannel format, since the number of cycles is
determined by the physical configuration of the device.
Although the closed-loop microchannel PCR device reported
in ref. 13 has circumvented this disadvantage, it is nevertheless
necessary to increase the number of heating blocks when
additional temperature zones are required. Moreover, in the
flow-through microchannel format, each sample occupies a
greater chip area than in the well format. Therefore, multi-
plexing the microchannels of flow-through PCR chips greatly
increases the necessary chip size. However, since the heat
exchange zones of the flow-through microchannel PCR format
are maintained at preset constant temperatures, it is not
necessary to heat and cool the reaction chamber repeatedly
during the temperature cycling process. Therefore, tempera-
ture cycling in a flow-though microchannel device is more
efficient than in devices with a micro-well format. However,
incorporating additional temperature zones complicates the
design of flow-through chip. For example, a recent report
integrating reverse-transcription with PCR has increased the*[email protected]
PAPER www.rsc.org/analyst | The Analyst
This journal is � The Royal Society of Chemistry 2005 Analyst, 2005, 130, 931–940 | 931
number of heating block from three to four in the micro-
channel chip.14
The objective of the current study is to develop a tem-
perature cycling reaction microchip that combines the benefits
of the several microchip PCR formats described above. That
is, the efficient temperature cycling can be performed, as is in
the flow-through microchannel PCR chip, while the flexibility
of varying the cycle number and varying the number of
temperature zones is maintained, as is in the micro-well PCR
chip. Very recently, a chip PCR scheme that shuttles sample
back and forth in a microchannel has been reported.16,17
An annular or circular continuous temperature distribution
was generated in the reported microfluidic chip and the liquid
sample plug was pumped back and forth along the radial
direction of the chip to achieve rapid temperature cycling.
Since a continuous temperature distribution is sustained in the
chip, the number of temperature zones is not limited. Although
a similar spatial temperature cycling strategy has been utilized
previously in the literature,4,13,14 the temperature zones have
been distributed discretely rather than continuously as in the
present case.
This study integrated a microfluidic chip fabricated from
poly(methylmethacrylate) (PMMA) with a resistive heater chip
fabricated from transparent indium tin oxide (ITO)-coated
glass. The two chips were both machined by CO2 laser, which
enables the chip to be rapidly re-designed and re-fabricated as
required. The integrated chip/heater assembly provides an
economic and flexible means of generating a radial tempera-
ture gradient for temperature cycling reactions.
In this study, non-contact IR thermometry was utilized to
assess the radial temperature gradient on the chip surface. The
distribution symmetry of the temperature distribution was
then quantitatively examined. Meanwhile, the literature has
reported several methods for internal temperature measure-
ment, including fluorescence quenching,18 fluorescent mole-
cular beacons,19 and thermochromic liquid crystal (TLC).20 In
this study, the internal temperature of the microchannel was
calibrated on the basis of the spectral change of a TLC
solution pumped through the microchannel to monitor the
temperature ramping rate inside the microchannel.
Although many suitable candidates exist for the microfluidic
chip substrate material, there are several advantages of
adopting PMMA. For example, PMMA is transparent, and
hence enables the application of conventional optical detection
methods when evaluating the microfluidic chip’s heating
characteristics.21 Furthermore, PMMA is an economic mate-
rial. Using the CO2 laser micromachining technique,22,23
economic and versatile microfluidic PMMA-based chips
can be fabricated in a matter of minutes. This particular
fabrication approach offers great flexibility for microfluidic
chip development. PMMA also has favorable biocompatibility
properties and has been used for many different biomedical
applications24 in recent decades. It has been reported that
protein adsorption on PMMA is less than on other forms of
plastics.25–27 Particularly, the adsorption on PMMA is less
than that which occurs on polycarbonate (PC) and polystyrene
(PS) for contact times of less than 30 min.25 Some widely used
substrate materials such as poly(dimethylsiloxane) (PDMS)
and silicon (Si) are known to result in substantial protein
adsorption,3,28 which may cause the PCR amplification
process to fail.29 Moreover, microchannels in Si chips can
not be integrated with photo-detection methods involving the
use of optical paths perpendicular to the substrate plane
because Si is not transparent to visible light. Glass and quartz
are also frequently used as microfluidic chip substrate
materials.30–32 However, the fabrication processes required
for these substrates are not as convenient as those for plastics.
Accordingly, PMMA was chosen as the substrate material of
the developed microfluidic device.
Addressing a perceived lack in the literature, this study
documents the performance of PMMA as a microchip PCR
substrate material. This study also examines the temperature
equilibrium rate attained in the reported device and confirms
that the design facilitates a rapid heat transfer from the chip to
the liquid sample. Although the present study takes the PCR
process as an example to demonstrate the temperature cycling
process, the execution of other applications requiring addi-
tional temperature zones can also be accomplished very easily
using the reported device.
2 Experimental procedure
2.1 Design and fabrication of heater chip and microfluidic chip
The reported reaction system comprises two basic compo-
nents, namely a heater chip and a microfluidic chip. The heater
chip is fabricated from ITO-coated glass, while the micro-
fluidic chip is fabricated from a PMMA sheet. In the reported
design, the two components are clamped firmly together to
ensure a close contact between them.
Since ITO is a transparent conductor, its use enables the
integration of an optical detection technique through the chip
plane. A deposited ITO film has been used as the chip heater
material in a previous study.32 However, the present study
differs in that it uses a piece of commercially available ITO-
coated glass (243735X0, Merck) as the heater substrate. In the
current fabrication process, a circular chip of diameter 85 mm
was cut from the ITO glass blank. It is noted that this diameter
was chosen merely to enable convenient manual handling. In
practice, the diameter can be modified to suit a particular
application. The thickness of the ITO film was y80 nm.
Discrete ITO strips were prepared on the chip surface to serve
as resistive heaters by scribing the ITO glass surface using a
CO2 laser scriber (M-300, Universal Laser). The sheet
resistance of the ITO film was 22.5 ohm per square (V %21).
Since the ITO film has a uniform thickness, the resistance of
each strip is governed by its width and length. In general, the
position, geometry, and resistance of each resistor strip can be
designed using computer software such as CorelDraw or
AutoCAD. As an example of the typical power consumption
of the heating device, a 3 mm wide circular strip with radius of
30 mm (hence the resistance of 2025 V) has heating power of
3.5 W. The maximal temperature generated by this heating
ring is y100 uC, as shown in the result below. The positional
accuracy of the scriber used in the present study was 10 mm
and the minimum attainable width of the ITO strip when using
the CO2 laser scriber was approximately 100 mm. In this study,
the ITO heating strips (indicated by the shaded regions in the
ITO chip in Fig. 1) were ring-shaped in order to establish a
932 | Analyst, 2005, 130, 931–940 This journal is � The Royal Society of Chemistry 2005
radial temperature gradient. As shown, two concentric heating
rings, connected in parallel by a pair of wide ITO strips of low
resistance, were designed on the chip. During fabrication, the
location of each resistor strip was specified by the computer
drawing software used during the scribing process. Clearly,
when a constant voltage is applied to the ITO strips via a pair
of conductive pins, a lower resistance results in a higher power,
and vice versa. When the heating ring with a lower power
was positioned near the center of the chip and the ring with
a higher power was located near the circumference, as with
the ITO pattern shown in Fig. 1, a circular bowl-shaped
temperature distribution was generated (see below, Fig. 3(b)).
Alternatively, positioning the high-power heating ring in the
center and the low-power ring near the circumference
produced a cone-shaped temperature distribution (Fig. 3(a)).
In summary, in the reported microfluidic device, the ITO
resistors can be dimensioned and positioned at an appropriate
pre-determined location to obtain the required temperature
distributions. Alternatively, a similar effect can be obtained
without modifying the chip design but simply by modifying the
voltages applied to the two ring heaters.
The use of a CO2 laser for PMMA surface machining has
been reported previously.22,23 The CO2 laser micromachining
of polycarbonate (PC) has also been studied.33,34 Using these
techniques, trench widths as low as y80 mm can be rapidly
prepared on plastic surfaces. In the present study, the required
trench pattern was designed using commercial computer
software and a CO2 laser scriber was then used to etch the
required trenches onto the PMMA substrate (thickness 5 2 mm,
Yen-Nan Acrylic Co. Taiwan). On-chip via holes for sample
loading were also etched into the chip using the laser scriber.
Various trench widths and depths were obtained by controlling
the laser beam diameter and the laser power. A higher power
was used to cut a circular PMMA chip of the same diameter as
the heater chip (85 mm) from a PMMA sheet. The trenched
PMMA chip was then thermally bonded with a blank PMMA
substrate (a cover plate) to form a chip with sealed channels, as
described in our previous study.22
The PMMA chip drawn in Fig. 1 shows the microfluidic
channel configuration used to perform the PCR process. It can
be seen that the overall configuration comprises a total of three
interconnected arcs. In this particular example, the inner arc is
used to perform the annealing step, the intermediate arc is for
the extension step, and the outer arc is for the denaturing step.
Although the minimal feature attainable by the current laser
scriber was as low as y80 mm, the microchannel in this study
was intentionally enlarged to increase the volume capacity so
that enough sample was collected for gel electrophoresis
analysis. The channel width and depth were specified as 1.2
and 0.6 mm, respectively. It was found that the laser ablation
process yielded a smooth and transparent channel wall.
Previous studies in the literature have employed a CNC milling
to fabricate channels of similar dimensions.25,35 However, the
current CO2 laser scribing process is suitable for the machining
of both the ITO glass used for the heater chip and the PMMA
used for the microfluidic chip substrate. The use of a single
machining process simplifies the overall fabrication procedure
considerably. In the fabricated PMMA microfluidic chip, the
channel length is approximately 70 mm. However, the channel
length can be easily re-fabricated using the CO2 laser scriber.
After bonding the trenched PMMA chip with the cover
plate, an adaptor was glued to the sample-loading hole on the
chip to facilitate the attachment of a microtube (OD 1.6 mm).
The microchannel walls were then coated with amorphous
Teflon (PFC802A, Cytonix) to reduce their surface energy in
order to ensure that the liquid sample would form a unified
plug when pumped through the channel. The microtube was
then connected to a computer-controlled syringe pump. It is
noted that the microchannel had only one opening for sample
loading. Hence, connecting this opening to the syringe pump
sealed the entire channel.
To heat the microfluidic chip using the heater chip, the two
chips were stacked together and held securely in place by
means of clamps and screws to ensure a good thermal contact.
A power supply is used to provide a stable voltage to the
heater chip.
Fig. 1 Microfluidic system setup showing assembly of microfluidic chip and heater chip. The two chips are clamped together and secured by
screws. All components other than the glass chip and the conductive pins are made of plastic. The heater chip is fabricated from ITO-coated glass.
The shaded area denotes the energized strips used for heating. The microfluidic chip is fabricated from PMMA. Note that only one opening exists.
In this example, three interconnected arcs are fabricated at three different radii. The radius for both chips is 85 mm.
This journal is � The Royal Society of Chemistry 2005 Analyst, 2005, 130, 931–940 | 933
Following sample loading, a syringe pump was connected to
the opening and air was pumped into the channel. In the current
design, the injection of compressed air pushes the liquid plug
forward into the microchannel. Conversely, when the pressure
of the compressed air in the syringe pump is reduced, the liquid
plug is pulled back towards the sample loading opening.
Hence, the liquid plug can be positioned in the desired tempera-
ture zone of the microchannel simply by controlling the
position of the syringe piston. The plug can then be shuttled
back and forth between the desired temperature locations by
varying the syringe pressure appropriately. The actual pressure
increase is calculated by the plug position. Note that since in
the current microfluidic device, the temperature gradient is
radial and continuous, the local temperature of any region can
be deduced from its radial position on the chip.
2.2 Measurement of chip surface and microchannel temperatures
To eliminate perturbations caused by the use of contact-type
thermal probes, this study employed a non-contact IR
thermometry technique to assess the surface temperature of
the chip. A mid-IR (8–12 mm) two-dimensional temperature
distribution thermal image was acquired by an IR camera
(120 6 120 pixels, IR SnapShot, Model 525, IRCON). The
temperature measurement accuracy of the IR camera was 2 uCand the spatial resolution of the IR image was 0.9 mm.
However, the internal temperature of the chip can not be
obtained using the same IR thermometry approach. Conse-
quently, in the present study, this temperature was determined
by the spectral change of a TLC solution passed through the
channel. Two different TLC fluids were used, one with a color
changing temperature of 71 uC (blue to transparent, band-
width 3 uC, Product No. PD Blue 71, New Prismatic Enterprise,
Taiwan) and another with a color changing temperature of
60 uC (black to transparent, bandwidth 3 uC, Product No. PD
Black 60, New Prismatic Enterprise, Taiwan). In both cases,
the TLC was suspended in water with 1% Tween 20 (X251-07,
J. T. Baker) added to enhance dispersion. The concentration of
the TLC solution was 5% w/v.
The static internal temperature was calibrated using the
following procedure. Initially, the TLC solution was flowed
into a microchannel having two openings (Fig. 2). Once
thermal equilibrium conditions had been achieved (y10 s), the
TLC color in the regions where the temperature exceeded the
color changing temperature became transparent. Meanwhile,
the color of the TLC in the regions of the microchannel having
a temperature lower than the color changing temperature
remained unchanged, as shown in Fig. 2. Finally, the position
of the microchannel at which color transition took place
corresponded to the point at which the local temperature was
the same as the TLC operation temperature.
2.3 Temperature ramping rate in microchannel
A TLC solution was also used to determine the temporal
temperature change in the microchannel. In determining the
temperature ramping rate, the measurement setup was the
same as that shown in Fig. 2 and a fiber spectrometer
(USB2000, Ocean Optics) was used. As described above, the
chip incorporated a microchannel with two openings, one of
which was used to introduce the reagent and another which led
to the waste buffer. Initially, a syringe was used to introduce
the TLC solution until the entire microchannel had been filled.
In the microchannel regions having a temperature lower than
71.5 uC, the liquid crystal appeared blue in color (dark liquid
in Fig. 2). The low-temperature absorption was measured by
the fiber spectrometer and the maximum absorption wave-
length was then determined for temporal detection. The
maximum absorption wavelength was found to be 600 nm.
Once temperature equilibrium conditions had been achieved,
the visible absorption of the liquid crystal inside the micro-
channel was continuously monitored by the fiber spectrometer
aligned with the region of the microchannel within which the
TLC was transparent and was located at a distance of approxi-
mately 1 mm from the color transition point, as shown in
Fig. 2. At the beginning of the temporal temperature change
measurement process, 200 ml of liquid crystal at room tempera-
ture was pumped through the channel at a flow velocity of
y2 m s21. The liquid flow was then stopped and the liquid
maintained in a fixed position aligned with the fiber head of the
spectrometer. The liquid crystal absorption increased signifi-
cantly when the TLC solution was initially pumped through the
channel. However, the absorption then decayed to its initial
value when the liquid crystal became transparent. The temporal
change of the absorption in the range 600 ¡ 30 nm was
integrated, recorded and digitized by the spectrometer. The
temporal resolution was set to be 0.3 s. It is noted that the volume
of the TLC solution consumed in the temporal measurement
(200 ml) process was much larger than that used for the
temperature cycling reaction process (4.5 ml), which ensured a
complete heat transfer for sample volumes smaller than 200 ml.
2.4 Conventional PCR, microchannel PCR and gel
electrophoresis
The PCR cocktail consisted of 100 pg ml21 ribulose 1,5-
bisphosphate carboxylase large subunit (rbcL) template DNA,
Fig. 2 Setup for measuring TLC spectral change in microchannel. A
miniature fiber spectrometer is used to monitor the static absorption
and the temporal absorption change of the TLC solution. Cold TLC
(dark liquid in the syringe) is pumped into the microchannel at a speed
of y2 m s21.
934 | Analyst, 2005, 130, 931–940 This journal is � The Royal Society of Chemistry 2005
0.16 mM of each dNTP base (1969064, Roche Boehringer
Mannheim Indianapolis, IN, USA), 10 mM betaine (20424-
1000, Acros, Morris Plains, NJ, USA), 0.2 mM of each primer
(forward primer: 59-GCGTGTTGTTGGAGAAAAAGA-39;
reverse primer: 59-GGACGACCATACTTGTTCAATTT-39),
5 mg ml21 bovine serum albumin (BSA; B-4287, Sigma, St.
Louis, MO, USA), 0.2 units ml21 FastStart taq (2032953,
Roche Boehringer Mannheim), and a reaction buffer (50 mM
Tris-HCl, 10 mM KCl, 5 mM (NH4)2SO4 and 2 mM MgCl2,
pH 8.3 at room temperature). The anticipated product length
was 240 bp, calculated by the primer alignment to the template
sequence (Accession number: Nc001879). The reaction cocktail
was loaded into the microchannel manually using a micro-
pipette. The temperature cycling process was then performed
using the procedure described below. Conventional PCR is
performed on a commercial PCR cycler (Bio-Rad, MyClear
Thermal Cycler). The total volume is 100 ml.
The concept of temperature cycling using the preset
temperature gradient involves pumping the sample plug along
the microchannel and shuttling it back and forth between
microchannel regions having the desired temperatures. In the
PCR process, 4.5 ml of the reaction mixture was loaded into the
well and then pumped into the channel. Surface tension effects
acting on the aqueous sample caused the formation of a short
unified liquid plug in the microchannel. It was found that
without the Teflon coating in the inner wall of the
microchannel, the microchannel PCR failed. The specific
reason was not investigated in this study. A syringe pump
was used to pump air into the channel, thereby driving the
PCR sample plug into the microchannel. In the initial
denaturing step, the PCR sample plug was retained in the
denaturing arc (95 uC) for 180 s. The subsequent temperature
cycling steps were as follows. The sample plug was held in the
annealing arc (57 uC) for 30 s and then pumped to the
extension arc (72 uC) and incubated for 30 s. The plug was then
pumped to the denaturing arc and incubated for 30 s. To
complete the cycle, the sample plug was moved back to the
annealing arc. The pumping speed is about 10.8 mm s21 (flow
rate 5 9.7 ml s21). Note that both the pumping speed and the
number of temperature cycles were controlled by the program
driving the syringe pump. Following temperature cycling, 2 ml
of the product was retrieved from the via hole by the micro-
pipette and incubated with 1 ml of loading dye, Bromophenol
Blue (B-5525, Sigma, St. Louis, MO, USA), and fluorescence
dye, SYBR Green I (S-7585, Molecular Probes, Engene, OR,
USA), for 15 min under dark conditions. A DNA ladder
(Invitrogen) was treated with the same dyes for comparison.
The PCR products were then analyzed by 2% agarose gel
electrophoresis in 0.5 6 TAE (Tris/acetate/EDTA) buffer.
3 Results and discussion
3.1 Radial surface temperature distribution of microfluidic chip
In investigating the radial surface temperature distribution of
the microfluidic chip, the chip was securely clamped to the
heater chip and then heated. The radial temperature distribu-
tion was evaluated on the basis of the observed chip surface
temperature. An IR camera was used to measure the chip
surface temperature. The IR thermometry method deduces the
substrate temperature by measuring the IR emissions (mid-IR,
wavelength 5 8–12 mm) from the substrate. Hence, a limitation
of this technique is that only surface temperatures can be
obtained since most materials are not transparent to mid-IR
emissions.
Depending on the relative geometrical positioning of the
high-power and low-power heating rings, two different types
of radial temperature distribution can be established on the
chip. When the high-power heating ring is placed near the chip
center, a cone-shaped temperature distribution, as shown in
the three-dimensional plot in Fig. 3(a). Alternatively, an
annular type temperature distribution is obtained when the
high-power heating ring is positioned near the circumference
of the chip, as shown in Fig. 3(b). It is noted that the present
design does not involve the use of active cooling devices to
maintain the temperature gradient. Rather, the distribution is
sustained by natural thermal conduction mechanisms. The
omission of a cooling device enables a reduction in the
footprint of the completed chip assembly.
The two spatial temperature distributions shown in Fig. 3(a)
and (b) are both circularly symmetric. However, it can be
seen that there exists a non-symmetric sector subtending
an angle of approximately 40 degrees at the center of the
chip. This deviation from the radial symmetry is caused by the
non-symmetric hardware structure in this region of the chip
comprising of the probe pins and the two ITO strips
connecting the two heating rings. The non-symmetric region
can be minimized by adopting an optimized heating coil
pattern. Alternatively, a non-contact inductive heating
strategy12 can be applied to eliminate the use of the contacting
pins and connecting ITO strips. In the current study, this
problem is circumvented by specifically designing the micro-
fluidic channel within a region of the chip positioned away
from this non-symmetric sector.
The distribution symmetry can be quantitatively expressed
in terms of the roundness of the isothermal zone. Using the
data presented in Fig. 3(b), IR image pixels corresponding to a
surface temperature ranging from 55 to 57 uC are analyzed.
The corresponding region is an annular temperature zone. The
roundness of this zone can then be examined. The radius
values around the isothermal circumference are found to be 23
pixels with a standard deviation of 0.37 pixels. Since the IR
imaging resolution is 0.9 mm/pixel, the corresponding spatial
deviation is 0.33 mm. These results indicate that the roundness
of the isothermal region is highly satisfactory. The deviation is
smaller than the image resolution of the IR imager. At this
radius, the temperature gradient is 2 uC/pixel (based on the
data in Fig. 3(b)) and therefore, the temperature uncertainty is
0.74 uC (2 uC/pixel 6 0.37 pixel). This analysis takes account
of the non-symmetric sector described above and indicates
that the temperature distribution asymmetry effect appears
mainly in the high-temperature regions. It is therefore reason-
able to estimate the surface temperature of extension and
annealing regions of the chip on the basis of its radial distance
from the center.
Several significant characteristics were observed in the
developed microfluidic system. For example, the radial
temperature distribution discussed above indicates that the
reported simple design is capable of generating a large
This journal is � The Royal Society of Chemistry 2005 Analyst, 2005, 130, 931–940 | 935
temperature span. The temperature ranges of the temperature
distributions shown in Fig. 3(a) and (b) are found to be 25 uCto 81 uC and 47 uC to 82 uC, respectively. It is noted that these
values refer to surface temperatures rather than internal
temperatures. Expanding the range to a higher temperature
domain can be achieved by using a suitable heat resistant
substrate material, such as PC or glass. Furthermore,
additional cooling devices such as annular thermoelectric
coolers can be integrated with the reported device to extend its
operating range towards the lower temperature domain. The
miniaturized microfluidic heater/cooler presented in Guijt’s
work19 can also be used without any significant increase in the
device size. Additionally, in the reported design, the tempera-
ture changes gradually and continuously along the radial
direction. Hence, there is no need to adopt thermal isolation
structures such as those micro-structures reported in some
delicate temperature cycling chips8 to isolate the different
temperature zones. Furthermore, the temperature at any
local chip position can be predicted solely on the basis of its
radial position. Therefore the reported design allows for
the incorporation of additional temperature zones without
the need to add further heating elements. For example, the
telomerase activity assay (i.e. Telomerase-PCR-ELISA, or the
TRAP assay),36 which requires two additional temperatures of
30 uC and 37 uC, can be integrated with PCR simply by driving
the sample to the desired temperature zones. Additionally, the
sample temperature overshooting phenomenon reported for
some micro-well PCR chip33 does not occur when the sample is
pumped correctly through the desired temperature zones.
Moreover, multiple sectors can be arranged on the same chip
in order to perform either independent or integrated reactions.
These sectors can be assigned identical or independent
temperature cycling parameters as required. For example, the
effects of different temperature cycling parameters, including
the incubation temperature, the incubation time, and the
temperature ramping rate, can all be tested using a single
microfluidic chip.
In order to utilize the continuous temperature gradient for
the execution of microfluidic temperature cycling reactions, it
is necessary to obtain the temperature distribution inside the
microchannel in addition to the surface temperature distribu-
tion discussed above. It is anticipated that the internal tem-
perature distribution will also maintain a radial symmetry,
although the gradient may vary slightly from that of the
surface temperature distribution.
3.2 Microchannel interior temperature distribution
When detecting the temperature of an object using a detection
probe, the probe absorbs heat from the object and conducts
heat away from it. Hence, the original temperature is
perturbed and hence can not be measured accurately. The
use of thermochromic liquid crystals (TLCs) provides an
appropriate alternative for internal temperature measure-
ment.33,37 This technique is highly effective in identifying the
Fig. 3 3-D plots of surface temperature distribution of circular
PMMA microchip. (a) Shows a cone-shaped temperature distribution.
The temperature varies from 81 to 25 uC from the center to the
circumference. (b) Shows bowl-shaped temperature distribution. The
temperature varies from 82 to 47 uC from the rim to the center. In (a)
and (b), a brighter color indicates a higher temperature. (c) Shows the
internal microchannel temperature profile, calculated by eqn. (2), of
cone-shaped and bowl-shaped temperature distribution along the chip
radius. The range for the cone-shape type extends from 25 to 101 uC,
while that for the bowl-shape type extends from 55 to 102 uC.
936 | Analyst, 2005, 130, 931–940 This journal is � The Royal Society of Chemistry 2005
regions having the same temperature as the TLC’s operation
temperature. In measuring the internal temperature distribu-
tion, this study first determines the temperature distribution
by IR thermometry and then correlates the internal tempera-
ture to the surface temperature by means of a thermo-
conducting model calibrated by the TLC internal temperature
measurement.
In developing the thermo-conducting model, the thermal
conductance through the chip is first simplified into the form
of a one-dimensional thermo-resistor system.38 This system
can then be transferred to the equivalent thermal circuit. In
this circuit, the heat transfer rate is expressed as:
qx~Ts{T?
1=hA~
Tc{Ts
L=kA(1)
Hence:
Tc~Lh
kz1
� �Ts{T?ð ÞzT? (2)
where Ts is the surface temperature of the PMMA microfluidic
chip, Tc is the temperature in the channel inside the chip,
and T‘ is the ambient temperature. Furthermore, L is the
thickness of the trenched PMMA chip, k denotes the thermal
conductivity of PMMA, and h is the convection heat transfer
coefficient of air. Finally, A denotes the area of the PMMA
surface perpendicular to the direction of heat transfer. Since
the model simplifies the heat conduction to a one-dimensional
form, the approximation is valid at positions distant from the
chip circumference.
Eqn. (2) allows the temperature inside the channel of the
PCR chip to be estimated on the basis of the measured surface
temperature. However, it is important to compare the
calculated channel temperature to that indicated by the TLC
method. Initially, a TLC solution with an operation tempera-
ture of 71.5 uC is used to identify the region of the
microchannel having the equivalent internal temperature.
The distance from this point to the circumference is found to
be 21 mm. This position is reasonably distant from the chip
circumference. The surface temperature at this point is then
established from the IR thermometry image and the corres-
ponding internal temperature is calculated from eqn. (2) as
Tc 5 73.5 uC. This temperature corresponds very well to that
indicated by the TLC approach (71.5 uC); particularly when it
is considered that the TLC color is known to deviate from its
original hue under the influence of ambient light.37 Therefore,
this study uses eqn. (2) in conjunction with the IR thermo-
metry technique to deduce the two-dimensional internal
temperature distribution.
3.3 Reciprocating temperature cycling
The following paragraph describes the use of the two-
dimensional temperature distribution in the temperature
cycling process. The second type of heater design is taken
as an example. In this design, the temperature distribution
exhibits a high-temperature rim near the chip circumference
(Fig. 3(b)). The surface temperature of this high-temperature
zone is 75 uC and the calculated channel temperature is 102 uC.
This temperature is slightly higher than the glass transition
point of the PMMA substrate. However, no substrate
deformation is observed in the current short-term applications
(,40 min). Inside the high-temperature rim, the temperature
decreases with decreasing radius. In this example, the lowest
channel temperature is 55 uC. As described above, the internal
channel temperature can be calculated from the measured
surface temperature by eqn. (2). The internal temperature
profile along the chip radius is shown in Fig. 3(c) for both
heater configurations. Utilizing the self-sustained temperature
gradient, it can be seen that the reported chip is capable of
providing temperature zones ranging continuously from 55 to
y100 uC.
Using the radially symmetric temperature gradient, tem-
perature cycling can be achieved by moving the sample
backward and forward along the chip radius. Since the local
temperature can be calculated on the basis of the distance from
the chip center, the microfluidic channel can be designed such
that it passes through the required temperature zones. When
the liquid sample is pumped into the channels and moved to
the desired temperature zones, it assumes the local temperature
via heat transfer.
Since the microfluidic chip and the heater chip are fabricated
separately, various channel designs can be fitted to the same
heater chip to obtain different temperature cycling parameters.
Alternatively, different temperature cycling parameters can be
obtained without modifying either chip by means of an
appropriate sample pumping scheme. Under the latter
approach, when performing the temperature cycling reaction
process, the sample plug can be reciprocated at different
pumping speeds in order to change the temperature ramping
rate. The sample plug can also be maintained in a stationary
position so that it is incubated at the constant temperature
associated with that particular location.
3.4 Temperature ramping rate in microchannel
It is important to examine the maximum temperature ramping
rate of the sample in order to enhance the control of the
temperature cycling parameters. As discussed previously, the
temperature change inside the microchannel can not be
measured by IR thermometry and is therefore monitored by
the temporal spectral change of the TLC.
TLCs are a class of organic compounds that change their
light reflectivity characteristics with temperature. Although the
temporal color change can be recorded by a video camera, it is
more conveniently monitored spectrally since this avoids hue
interference caused by the light source. The temporal absorp-
tion change is easily recorded and digitized by a miniature
spectrometer.
Fig. 4 shows the abrupt absorption increase which occurs
when 200 ml of the TLC solution is pumped through the
detection point at a rapid velocity of 2 m s21. The absorption
then decays gradually as the temperature of the TLC rises
and its color becomes transparent. As shown, the temperature
of the liquid inside the microchannel reaches equilibrium con-
ditions at 71.5 uC in approximately 10 s from room tem-
perature (25 uC). The corresponding temperature ramping rate
is y4.5 uC s21. PMMA, as with other plastics, is recognized
as a poor thermal conductor. Its thermal conductivity is
This journal is � The Royal Society of Chemistry 2005 Analyst, 2005, 130, 931–940 | 937
0.193 W m21 K21.39 However, in the present study, temperature
equilibrium is attained very rapidly, which indicates that a
rapid heat transfer to the sample takes place inside the micro-
channel. The surface to volume ratio of the channel with the
width of 1.2 mm and the depth of 0.6 mm is 8 mm2 ml21. Hence,
the rapid heat transfer to the sample is reasonably attributed to
the large surface to volume ratio of the microchannel.
Although in the experiment above, temperature equilibrium
is attained after approximately 10 s, the time required for the
actual temperature cycling process is far less when a smaller
sample volume is used. An almost instantaneous color
bleaching was observed when the TLC was pumped through
the heated channel at a slow speed of 10 mm s21 (data not
shown). There are several reasons for this. First, the
temperature equilibrium observation in Fig. 4 is performed
by pumping a large volume of TLC solution at room
temperature into the channel in a very short time (i.e. 200 ml
in y0.3 s). This procedure is adopted in order to ensure that
the temperature of the TLC solution at the spectrum detection
point rises from room temperature so that color change can be
recorded. The heat required to raise the TLC temperature is
provided solely by the channel wall at the detection point and
the limited heating area prolongs the temperature rise time.
However, when sample is pumped through the microchannel
at low speed and the temperature cycling is performed slowly,
the temperature equilibrium time is far shorter. The estimation
of the fastest temperature ramping rate is as follows. Assuming
the sample is pumped at 10 mm s21 from the chip center to the
circumference, for which the corresponding temperature
change is from 50 to 100 uC, and the chip radius is 43 mm,
the corresponding temperature ramp rate is y12 uC s21. It can
be concluded that the large surface to volume ratio and the
absence of a large heat mass yield a rapid temperature cycling
of the sample in the microchannel.
3.5 Sample evaporation inside channel
Sample evaporation is often problematic in microfluidic
applications because the sample volumes are very small.
Hence, this study devised a sample-pumping scheme to reduce
sample loss during the temperature cycling process. The
reported method significantly reduces the sample evaporation,
particularly for samples processed in high-temperature zones.
Under standard atmospheric pressure, water evaporation is
vigorous at 100 uC. In order to avoid sample loss through
evaporation in the high-temperature zones during the tem-
perature cycling process, this study intentionally increases the
internal pressure when the sample plug is pumped to high-
temperature zones. In the reported approach, a single opening
is fabricated for each microfluidic channel. This opening serves
for the loading of the sample and for connection to the syringe
pump. When the opening is connected to the pump, the
microchannel is completely sealed by the syringe. Air is then
compressed into the channel in order to drive the sample plug.
The air behind the distal end of the plug is also compressed
when the plug is pushed forward. Since the opening is
deliberately positioned in the low-temperature region, the
channel pressure increases as the sample plug is driven toward
the distal end of the channel in the direction of the high-
temperature zone. The internal pressure can be calculated by
Boyle’s law. Alternatively, it can be determined simply by
dividing the cross-sectional area of the syringe pump into the
force required to push the sample plug (Pressure 5 Force/
Area). In the present example, the pressure is calculated to be
y7 kg cm22 when the sample is driven to the high-temperature
zone. The water vapor pressure is 1/7 that of the compressed
air. At this pressure, water boils at a temperature of 166 uC.40
Therefore, the sample evaporation is greatly reduced where the
local temperature is only around 100 uC. As shown in Fig. 5, in
the current PCR temperature cycling example, the reduction in
sample volume varies as a function of the number of PCR
cycles. It can be seen that y70% of the sample remains as a
unified plug following 30 temperature cycles. The addition of
betaine is at low concentration (10 mM) compared to that in
the literature (1–2.5 M,41) and does have significant effect on
boiling temperature. The evaporation reduction is therefore
considered to be resulted from pressure increase. Without the
high pressure achieved in the current design, the sample would
boil in the denaturing zone and would dry in seconds.
Fig. 5 Minimized sample evaporation inside microchannel vs.
number of temperature cycles. Evaporation is minimized by pressuriz-
ing the sample during heating.
Fig. 4 Temporal spectral change of TLC (color changing tempera-
ture 71.5 uC) solution inside microchannel. The color change is
measured by visible absorption spectroscopy. The inset shows the
absorption spectrum of the liquid crystal.
938 | Analyst, 2005, 130, 931–940 This journal is � The Royal Society of Chemistry 2005
3.6 Microchannel PCR using radial temperature gradient
This study takes the well-known PCR process as an example
in order to demonstrate the successful temperature cycling
reaction achieved using the reported microfluidic device.
However, it should be noted that the same chip could also
be extended to the execution of processes other than PCR.
Fig. 6 presents the agarose gel electrophorogram of the
amplified PCR product. The first lane indicates the DNA
ladder, while the third lane is the product from the
microchannel PCR. It can be seen that the anticipated 240 bp
product amplified from the rbcL gene template has been
obtained. The reaction took a total of 38 min for the micro-
channel PCR with it is 3.5 h for the conventional PCR. The
injected sample volume was 4.5 ml and the amplification was
performed over 23 cycles. As with the well-based microchip
PCR,6,15,42–44 the cycle number can be adjusted as required in
the reported chip and is not limited by the original channel
configuration, as is the case for flow-through PCR chip.
Furthermore, the reported microfluidic chip can be divided
into a number of discrete sectors such that multiple channels
can be configured on the same chip. Approximately 10 PCR
channels can be fabricated on the current chip with a diameter
of 85 mm diameter. However, further optimization is currently
under development with the intention of increasing this
number in future versions of the chip. Currently with the
limited channel number but with the low power consumption
(3.5 W), the reported device may be useful as a portable device,
e.g. a lab-on-a-chip PCR machine.
4 Conclusion
This paper presented a simple and versatile PMMA micro-
fluidic system for temperature cycling reactions utilizing a self-
sustained radial temperature gradient. It has been shown that
this device can generate a large temperature span ranging
from room temperature to y100 uC without the need for a
cooling device or active temperature control mechanisms.
Furthermore, the temperature gradient can be maintained
without the need for complex temperature control electro-
nics.31 Furthermore, no thermal isolation structures are
necessary to separate zones of different temperature. The
completed chip assembly provides a powerful tool for
integrating biological and chemical reactions on a single chip.
The PMMA microfluidic chip and the ITO resistor heater
chip were both fabricated using an economic CO2 laser
scriber. The entire manufacturing process is straightforward
and highly versatile. The overall approach is eminently suitable
for low-cost rapid prototyping.
A non-invasive full-chip temperature characterization pro-
cedure has been presented. IR thermometry and TLC
spectral change observation techniques have been combined
to calibrate the temperature distributions on the chip surface
and within the microchannels, respectively. Based upon the
observed absorption change of the TLC, it has been shown
that the liquid sample in the channel attains a temperature
equilibrium condition in less than 10 s. For low sample volume
(y4.5 ml) the temperature ramping rate is estimated to be
y12 uC s21. It should be emphasized that a spectral method
rather than a colorimetric method is employed to monitor
the TLC color change. Hence, the potential interference of
ambient light on the TLC color is avoided.
A novel strategy for the temperature cycling process has
been presented, together with a scheme designed to reduce
sample evaporation. Although the microchannel PCR process
has been taken as an example to demonstrate the capability
of the chip to perform temperature cycling reactions, the chip
can readily be extended to other processing operations. The
reported scheme successfully combines the benefits of the well-
based PCR and continuous flow microchannel PCR devices on
a single chip.
Abbreviations
PCR, polymerase chain reaction; PMMA, poly(methylmetha-
crylate); ITO, indium tin oxide; PTFE, poly(tetrafluoroethyl-
ene); PC, polycarbonate; PDMS, poly(dimethylsiloxane).
Acknowledgements
This work is partially supported by the National Science
Council, Taiwan under the contract number NSC 93-2113-M-
001-037.
Ji-Yen Cheng,* Chien-Ju Hsieh, Yung-Chuan Chuang andJing-Ru HsiehResearch Center for Applied Sciences, 128 Sec. 2 Academia Rd., Taipei,11529 Taiwan E-mail: [email protected];Fax: 886-2-2782-6680; Tel: 886-2-2789-8000-34
References
1 E. Verpoorte, Electrophoresis, 2002, 23, 677–712.2 I. Schneegass and J. M. Kohler, J. Biotechnol., 2001, 82, 101–121.3 J. Cheng, M. A. Shoffner, G. E. Hvichia, L. J. Kricka and
P. Wilding, Nucleic Acids Res., 1996, 24, 380–385.4 M. U. Kopp, A. J. de Mello and A. Manz, Science, 1998, 280,
1046–1048.5 P. Wilding, M. Shoffner and L. Kricka, Clin. Chem., 1994, 40,
1815–1818.6 L. J. Kricka and P. Wilding, Anal. Bioanal. Chem., 2003, 377,
820–825.
Fig. 6 Agarose gel electrophorogram showing successful PCR
amplification in PMMA microchannel. The template is rbcL gene.
The expected product length is 240 bp. M denotes DNA ladder. Cv
and Ch denote conventional and microchannel PCR, respectively.
This journal is � The Royal Society of Chemistry 2005 Analyst, 2005, 130, 931–940 | 939
7 M. A. Northrup, B. Benett, D. Hadley, P. Landre, S. Lehew,J. Richards and P. Stratton, Anal. Chem., 1998, 70, 918–922.
8 S. Poser, T. Schulz, U. Dillner, V. Baier, J. M. Koehler, D. Schimkat,G. Mayer and A. Siebert, Sens. Actuators A, 1997, 62, 672–675.
9 A. F. Huhmer and J. P. Landers, Anal. Chem., 2000, 72, 5507–5512.10 M. N. Slyadnev, Y. Tanaka, M. Tokeshi and T. Kitamori, Anal.
Chem., 2001, 73, 4037–4044.11 R. P. Oda, M. A. Strausbauch, A. F. R. Huhmer, N. Borson,
S. R. Jurrens, J. Craighead, P. J. Wettstein, B. Eckloff, B. Klineand J. P. Landers, Anal. Chem., 1998, 70, 4361–4368.
12 D. Pal and V. Venkataraman, Sens. Actuators A, 2002, 102, 151–156.13 J. Chiou, P. Matsudaira, A. Sonin and D. Ehrlich, Anal. Chem.,
2001, 73, 2018–2021.14 P. J. Obeid, T. K. Christopoulos, H. J. Crabtree and
C. J. Backhouse, Anal. Chem., 2003, 75, 288–295.15 S. Shandrick, Z. Ronai and A. Guttman, Electrophoresis, 2002, 23,
591–595.16 F. Jiang, K. S. Drese, S. Hardt, M. Kupper and F. Schonfeld,
AIChE J., 2004, 50, 2297–2305.17 P. A. Auroux, Y. Koc, A. de Mello, A. Manz and P. J. R. Day,
Lab Chip, 2004, 4, 534–546.18 D. Ross, M. Gaitan and L. E. Locascio, Anal. Chem., 2001, 73,
4117–4123.19 R. M. Guijt, A. Dodge, G. W. K. v. Dedem, N. F. de Rooij and
E. Verpoorte, Lab Chip, 2003, 3, 1–4.20 J. Liu, M. Enzelberger and S. Quake, Electrophoresis, 2002, 23,
1531–1536.21 F. Dang, L. Zhang, H. Hagiwara, Y. Mishina and Y. Baba,
Electrophoresis, 2003, 24, 714–721.22 J.-Y. Cheng, C.-W. Wei, K.-H. Hsu and T.-H. Young, Sens.
Actuators B, 2004, 99, 186–196.23 H. Klank, J. P. Kutter and O. Geschke, Lab Chip, 2002, 2, 242–246.24 V. Rebeix, F. Sommer, B. Marchin, D. Baude and M. D. Tran,
Biomaterials, 2000, 21, 1197–1205.25 M. J. Madou, Y. Lu, S. Lai, C. G. Koh, L. J. Lee and B. R. Wenner,
Sens. Actuators A, 2001, 91, 301–306.26 R. P. Baptista, A. M. Santos, A. Fedorov, J. M. Martinho,
C. Pichot, A. Elaissari, J. M. Cabral and M. A. Taipa,J. Biotechnol., 2003, 102, 241–249.
27 P. Valette, M. Thomas and P. Dejardin, Biomaterials, 1999, 20,1621–1634.
28 G. Ocvirk, M. Munroe, T. Tang, R. Oleschuk, K. Westra andD. J. Harrison, Electrophoresis, 2000, 21, 107–115.
29 M. A. Shoffner, J. Cheng, G. E. Hvichia, L. J. Kricka andP. Wilding, Nucleic Acids Res., 1996, 24, 375–379.
30 B. C. Giordano, E. R. Copeland and J. P. Landers, Electrophoresis,2001, 22, 334–340.
31 I. Schneegass, R. Brautigamb and J. M. Kohlera, Lab Chip, 2001,1, 42–49.
32 K. Sun, A. Yamaguchi, Y. Ishida, S. Matsuo and H. Misawa, Sens.Actuators B, 2002, 84, 283–289.
33 Y. Liu, C. B. Rauch, R. L. Stevens, R. Lenigk, J. Yang, D. B. Rhineand P. Grodzinski, Anal. Chem., 2002, 74, 3063–3070.
34 J. Yang, Y. Liu, C. B. Rauch, R. L. Stevens, R. H. Liu, R. Lenigkand P. Grodzinski, Lab Chip, 2002, 2, 179–187.
35 P. K. Yuen, L. J. Kricka, P. Fortina, N. J. Panaro, T. Sakazumeand P. Wilding, Genome Res., 2001, 11, 405–412.
36 C. Nemos, J. P. Remy-Martin, P. Adami, F. Arbez-Gindre,J. P. Schaal, M. Jouvenot and R. Delage-Mourroux, Clin.Biochem., 2003, 36, 621–628.
37 A. M. Chaudhari, T. M. Woudenberg, M. Albin andK. E. Goodson, J. Microelectromech. Syst., 1998, 7, 345–355.
38 P. I. Frank and P. D. David, Fundamentals of Heat and MassTransfer, Wiley, New York, 1996.
39 J. Brandrup and E. H. Immergut, Polymer Handbook, Wiley,New York, 1989.
40 D. R. Lide, CRC Handbook of Chemistry and Physics, CRC Press,LLC, 2001.
41 W. Henke, K. Herdel, K. Jung, D. Schnorr and S. A. Loening,Nucleic Acids Res., 1997, 25, 3957–3958.
42 H. Nagai, Y. Murakami, K. Yokoyama and E. Tamiya, Biosens.Bioelectron., 2001, 16, 1015–1019.
43 H. Nagai, Y. Murakami, Y. Morita, K. Yokoyama and E. Tamiya,Anal. Chem., 2001, 73, 1043–1047.
44 P. Belgrader, W. Benett, D. Hadley, G. Long, R. Mariella, Jr.,F. Milanovich, S. Nasarabadi, W. Nelson, J. Richards andP. Stratton, Clin. Chem., 1998, 44, 2191–2194.
940 | Analyst, 2005, 130, 931–940 This journal is � The Royal Society of Chemistry 2005