Anal. Chem.
2003, 75, 288-295
Microfabricated Device for DNA and RNA
Amplification by Continuous-Flow Polymerase
Chain Reaction and Reverse
Transcription-Polymerase Chain Reaction with
Cycle Number Selection
Pierre J. Obeid,† Theodore K. Christopoulos,*,†,‡ H. John Crabtree,§ and Christopher J. Backhouse|
Department of Chemistry, University of Patras, Patras, Greece GR-26500, Institute of Chemical Engineering and
High-Temperature Chemical Processes, P.O. Box 1414, Patras, Greece GR-26500, Micralyne Inc., 1911 94th Street,
Edmonton, Alberta, T6N 1E6, Canada, and Department of Electrical and Computer Engineering, University of Alberta,
Edmonton, Alberta, T6G 2V4, Canada
We have developed a high-throughput microfabricated,
reusable glass chip for the functional integration of reverse
transcription (RT) and polymerase chain reaction (PCR)
in a continuous-flow mode. The chip allows for selection
of the number of amplification cycles. A single micro-
channel network was etched that defines four distinct
zones, one for RT and three for PCR (denaturation,
annealing, extension). The zone temperatures were con-
trolled by placing the chip over four heating blocks.
Samples and reagents for RT and PCR were pumped
continuously through appropriate access holes. Outlet
channels were etched after cycles 20, 25, 30, 35, and
40 for product collection. The surface-to-volume ratio for
the PCR channel is 57 mm-1 and the channel depth is
55 µm, both of which allow very rapid heat transfer. As a
result, we were able to collect PCR product after 30
amplification cycles in only 6 min. Products were col-
lected in 0.2-mL tubes and analyzed by agarose gel
electrophoresis and ethidium bromide staining. We stud-
ied DNA and RNA amplification as a function of cycle
number. The effect of the number of the initial DNA and
RNA input molecules was studied in the range of 2.5 ×
106-1.6 × 108 and 6.2 × 106-2 × 108, respectively.
Successful amplification of a single-copy gene (β-globin)
from human genomic DNA was carried out. Furthermore,
PCR was performed on three samples of DNA of different
lengths (each of 2-µL reaction volume) flowing simulta-
neously in the chip, and the products were collected after
various numbers of cycles. Reverse transcription was also
carried out on four RNA samples (0.7-µL reaction volume)
flowing simultaneously in the chip, followed by PCR
amplification. Finally, we have demonstrated the concept
of manually pumped injection and transport of the reac-
tion mixture in continuous-flow PCR for the rapid genera-
tion of amplification products with minimal instrumenta-
tion. To our knowledge, this is the first report of a
288 Analytical Chemistry, Vol. 75, No. 2, January 15, 2003
monolithic microdevice that integrates continuous-flow RT
and PCR with cycle number selection.
Nucleic acid analysis techniques have a major impact in diverse
areas, such as molecular diagnosis of disease and assessment of
therapy, environmental testing, food technology, agriculture, and
forensic science.1 The polymerase chain reaction (PCR)2,3 entails
the in vitro exponential enzymatic amplification of specific DNA
sequences to levels that are several orders of magnitude higher
than the starting material. RNA amplification is accomplished by
synthesizing a copy of complementary DNA, using reverse
transcriptase, and then performing PCR. DNA or RNA amplifica-
tion has become an essential step of most procedures used for
nucleic acid detection, quantification, and sequencing. The Human
Genome Project including projects on various model organisms
have provided a vast amount of sequence data and therefore
initiated a new era of nucleic acid-based tests. High throughput
and automation of DNA/RNA analysis techniques, however, is
required in order to exploit the accumulated genetic information.
In recent years, significant advances have emerged in the area of
miniaturization of analytical instrumentation and development of
micro total analysis systems (µTAS).4-6 Key benefits of miniatur-
ization include high throughput, low consumption of sample and
reagents, portability, and the prospect of integration of all steps
* To whom correspondence should be addressed: (tel) (+30) 2 610 997130;
(fax) (+30) 2 610 997118; (e-mail) tkc@chemistry.upatras.gr.
† University of Patras.
‡ Institute of Chemical Engineering and High-Temperature Chemical Pro-
cesses.
§ Micralyne Inc.
| University of Alberta.
(1) Christopoulos, T. K. Anal. Chem. 1999, 71, 425R-438R.
(2) Mullis, K. B.; Faloona, F. A. Methods Enzymol. 1987, 155, 335-350.
(3) Christopoulos, T. K. Polymerase Chain Reaction and Other Amplification
Systems. In Encyclopedia of Analytical Chemistry; Meyers, R. A., Ed.; Wiley:
Chichester, 2000; pp 5159-5173.
(4) Reyes, D. R.; Iossifidis, D.; Auroux, P. A.; Manz, A. Anal. Chem. 2002, 74,
2623-2636.
(5) Auroux, P. A.; Iossifidis, D.; Reyes, D. R.; Manz, A. Anal. Chem. 2002, 74,
2637-2652.
(6) Kopp, M. U.; Crabtree, H. J.; Manz, A. Curr. Opin. Chem. Biol. 1997, 1,
410-419.
10.1021/ac0260239 CCC: $25.00 © 2003 American Chemical Society
Published on Web 12/12/2002
of an analytical procedure into a single device. The manufacturing
of miniaturized and integrated analytical systems is based on the
technologies developed in the microelectronics industry for
production of integrated circuits. A major part of the research
activity in this area is directed toward biochemical analysis.7-9
Microfabricated devices for cell separation,10,11 electrophoresis,12,13
DNA amplification,14-22 and DNA sequencing23,24 have been
developed.
Chip-based PCR systems reported so far have been based on
two configurations. In one configuration, the reaction mixture is
stationarysenclosed in a reaction chamber that is typically
micromachined in Si or glass. The chamber is then placed on a
conventional thermocycling instrument for heating and cooling.14-16
Alternatively, heaters and temperature sensors are integrated into
the device.17 Noncontact heating by infrared radiation has also
been employed for fast thermocycling of microstructures.18
Further developments in this basic configuration aim at the
analysis of amplification products. This has been accomplished
by integrating optical windows in the silicon chamber allowing
for real-time fluorometric monitoring of the reaction.17 There are
also several reports on the combination of PCR with capillary
electrophoresis (CE).19-21
More recently,22 a second configuration of chip-based PCR was
reported that involves a continuous-flow amplification system in
which the reaction mixture, instead of being stationary in a
thermocycling chamber, is pumped continuously through the chip
and flows repeatedly through zones of different temperatures to
ensure denaturation, annealing, and extension. The system allows
for very rapid heat transfer and thermal cycling of the minute
fluidic element due to both the high surface-to-volume ratio and
the short linear thermal diffusion distance in the microfluidic
channels. This design advantage is best achieved by maintaining
the zones at preset constant temperatures. The continuous-flow
system is a true microfluidic device and is more suitable for high-
throughput analysis and coupling with upstream and downstream
(7) Kricka, L. J. Clin. Chim. Acta 2001, 307, 219-223.
(8) Sanders, G. H. W.; Manz, A. Trends Anal. Chem. 2000, 19, 364-378.
(9) Burke, D. T.; Burns, M. A.; Mastrangelo, C. Genome Res. 1997, 7, 189-
197.
(10) Cheng, J.; Sheldon, E. L.; Wu, L.; Uribe, A.; Gerrue, L. O.; Carrino, J.; Heller,
M. J.; O’Connell, J. P. Nat. Biotechnol. 1998, 16, 541-546.
(11) Wilding, P.; Kricka, L. J.; Cheng, J.; Hvichia, G. E.; Schoffner, M. A.; Fortina,
P. Anal. Biochem. 1998, 257, 95-100.
(12) Effenhauser, C. S.; Paulus, A.; Manz, A. Widmer, M. M. Anal. Chem. 1994,
66, 2949-2953.
(13) Woolley, A. T.; Mathies, R. A. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 11348-
11352.
(14) Wilding, P.; Schoffner, M. A.; Kricka, L. Clin. Chem. 1994, 40, 1815-1818.
(15) Schoffner, M. A.; Cheng, J.; Hvichia, G. E.; Kricka, L. J.; Wilding, P. Nucleic
Acids Res. 1996, 24, 375-379.
(16) Cheng, J.; Schoffner, M. A.; Hvichia, G. E.; Kricka, L. J.; Wilding, P. Nucleic
Acids Res. 1996, 24, 380-385.
(17) Northrup, M. A.; Benett, B.; Hadley, D.; Landre, P.; Lehew, S.; Richards, J.;
Stratton, P. Anal. Chem. 1998, 70, 918-922.
(18) Giordano, B. C.; Ferrance, J.; Swedberg, S.; Huhmer, A. F. R.; Landers, J.
P. Anal. Biochem. 2001, 291, 124-132.
(19) Woolley, A. T.; Hadley, D.; Landre, P.; deMello, A. J.; Mathies, R. A.;
Northrup, M. A. Anal. Chem. 1996, 68, 4081-4086.
(20) Waters, L. C.; Jacobson, S. C.; Kroutchinina, N.; Khandurina, J.; Foote, R.
S.; Ramsey, J. M. Anal. Chem. 1998, 70, 5172-5176.
(21) Khandurina, J.; McKnight, T. E.; Jacobson, S. C.; Waters, L. C.; Foote, R.
S.; Ramsey, J. M. Anal. Chem. 2000, 72, 2995-3000.
(22) Kopp, M. U.; deMello, A. J.; Manz, A. Science 1998, 280, 1046-1048.
(23) Liu, S.; Shi, Y.; Ja, W. W.; Mathies, R. A. Anal. Chem. 1999, 71, 566-573.
(24) Backhouse, C.; Caamano, M.; Oaks, F.; Nordman, E.; Carrillo, A.; Johnson,
B.; Bay, S. Electrophoresis 2000, 21, 150-156.
flow-through devices. However, a significant limitation of this
device is that the number of cycles is fixed by the channel layout
at 20 cycles. Therefore, only a limited control of the reaction
conditions is possible through changes of the flow rate and zone
temperatures.
Despite the wide spectrum of applications of RNA amplification
in research and molecular diagnosis, there is limited work on the
coupling of reverse transcription (RT) and PCR on a single
microdevice. In a previous report, RT was performed in a
stationary reaction mixture (50 µL) placed in a polypropylene tube
liner that was inserted in a silicon thermal cycled cavity.17
Following one round of PCR, the sample was manually transferred
to another chamber for a second round of amplification (nested
PCR).
In this work we report a microfabricated device for continuous-
flow DNA and RNA amplification, which offers the following
distinct advantages: (a) functional integration of reverse transcrip-
tion and polymerase chain reaction (RT-PCR) onto a single
monolithic chip for direct RNA amplification and (b) cycle number
selection at 20, 25, 30, 35, and 40 cycles. We first study the effect
of cycle number on the amount of amplification product. The
influence of the input DNA and RNA molecules on the amplified
product is also studied. Moreover, the high-throughput capability
of continuous-flow PCR and RT-PCR is demonstrated, for the first
time, by running simultaneously multiple DNA samples of different
sizes followed by product collection at different cycle numbers.
Similarly, multiple small-volume RNA samples are subjected to
RT-PCR in a continuous-flow format. The amplification of a single-
copy gene from human genomic DNA isolated from whole blood
is performed successfully. Finally, it is shown that continuous-
flow PCR can be performed rapidly in this chip by a simple hand-
driven sample and reagent transport without the use of pumps.
EXPERIMENTAL SECTION
Microfabrication. The microfabricated chip used throughout
this study was fabricated at Micralyne (Edmonton, AB, Canada)
and is depicted in Figure 1. This device is composed of two
Corning 0211 borosilicate glass plates (each 40 × 45 × 0.55 mm)
that were fusion-bonded together to form a closed structure. A
continuous channel network was etched into the bottom plate by
standard photolithographic and wet chemical (HF) etching
techniques. The channels were 194 µm wide (at the top) and 55
µm deep for the reverse transcription section of the device, and
120 µm wide and 55 µm deep for the PCR section (Figure 1).
The cover plate contains access holes (400-µm diameter, drilled
ultrasonically) that were used for sample injection and product
collection. These were aligned with the inlet and outlet channels
on the bottom plate. After cleaning, the cover and bottom plates
were fusion bonded to form the sealed microfluidic device.
Both the RT and PCR sections contain two inlets each and
are marked on the chip by RT1, RT2, PCR1, and PCR2. The PCR
channel incorporates 40 identical cycles that pass over three
temperature-controlled zones, to perform DNA denaturation,
primer annealing, and enzymatic extension steps of amplification
at time ratios of 4:4:9, respectively. The channel allows for a 3-fold
extended denaturation step prior to entering the first cycle. Outlet
channels were designed after cycles 20, 25, 30, 35, and 40 for
product collection. The capacity of the PCR channel (from cycles
Analytical Chemistry, Vol. 75, No. 2, January 15, 2003 289
Figure 1. (A) Schematic diagram of the microfabricated device for continuous-flow PCR and RT-PCR with cycle number selection. There are
two inlets for PCR (PCR1 and -2), two inlets for RT (RT1 and -2), and five outlets for product collection at 20, 25, 30, 35, and 40 cycles. The
chip is placed over four heating blocks, a-d, with appropriate temperatures for denaturation, extension, and annealing, respectively. I, intersection
between the RT and the PCR channels. (B) A simple holding apparatus for the chip. This configuration provides quick leak-proof connections
between capillaries and the access holes of the chip (inlets and outlets) by using silicone disks. It also facilitates the accurate positioning of the
chip over the heated blocks and eliminates the direct gluing of the capillaries to the access holes. (C) Diagram of the etched channels. For the
RT channel, the radial depth (D) is 55 µm and the width (W) is 84 µm. This yields a 194-µm channel width at the top. The PCR channel has a
radial depth of 55 µm and width of 10 µm. Hence, the channel width at the top is 120 µm.

1 to 40) is 15.3 µL. The RT channel has a capacity of 5 µL and is
also in contact with its own temperature-controlled zone.
Assembly of the Chip. To introduce samples through the chip
and collect amplification products, we have constructed a simple
holding apparatus that facilitates the connection between capillary
tubes and the inlets and outlets of the device (Figure 1). A piece
of clear Plexiglas (80 × 55 × 5 mm) was used, which contained
0.5-mm drilled holes that matched exactly those of the micro-
device. Capillary tubes (3 cm long, 375-µm o.d., 100-µm i.d.,
Polymicro Technologies, Phoenix, AZ) were cut and glued with
epoxy to the Plexiglas holes such that 1 mm of their ends
protruded out from the holes. To provide leak-proof connections,
disks (2-4-mm diameter and 0.8 mm thick) were prepared from
silicone glue and were perforated with the protruding ends of the
capillaries. For assembly, the capillary ends carrying the disks
were inserted carefully into the access holes of the device and
the apparatus was held tight with two screws. The free ends of
the capillaries, therefore, become the inlets and outlets for the
channels. There are four inlet and five outlet capillaries. Cycle
selection was accomplished by blocking the ends of all outlet
capillaries with Teflon caps except of the one corresponding to
the selected cycle number. Sample collection was accomplished
by simply placing an inverted plastic 0.2-mL tube on top of the
uncapped capillary. The caps used for blocking the free ends of
unutilized capillaries were prepared by using a 5-mm-long Teflon
tubing (∼380-µm i.d.). One end of the tubing was plugged with
epoxy glue while the open end was tightly fitted over the capillary.
Prior to the first PCR of the day, the whole chip was filled with
water. Thus, the unused (capped) outlet channels prior and after
the selected cycle number contain water. This ensures that the
reaction mixture exits only from the uncapped capillary. No losses
of sample into the capped outlets occur.
Chip Peripherals. Fluids were pumped through the chip by
hydrostatic pressure. Two PC-controlled precision syringe pumps
(model 50300, Kloehn, Las Vegas, NV), each with a syringe
capacity of 25 µL, were used for sample delivery. A 2-µm in-line
microfilter (model M-532, Upchurch Scientific, Oak Harbor, WA)
was connected between the chip and the pumps.
Temperature control was achieved by placing the chip as-
sembly on top of a suitable heating base constructed in-house by
using four copper blocks, 80 × 20 × 10 mm each. Inside each
block, and along its length, we placed a heating cartridge (Omega,
CSS-03365/110) and a temperature sensor (Omega, 1PT 100
kn2528). The block temperatures were controlled by using four
digital PID temperature controllers (Jumo iTron microprocessor
controller, model 702040 from Enercorp, Islington, ON, Canada)
housed in a single unit containing a power supply and electric
circuits for the heating cartridges and sensors. The blocks were
then positioned, with a 5-mm spacing between each other, on a
nonthermally conducting base (85 × 85 × 10 mm) and their
surfaces were evened out with respect to each other. This ensured
that the chip made a good contact throughout, when it was placed
on top of the heating blocks. To minimize heat transfer between
the blocks, a small fan was placed in front of them, allowing air
to pass through the spacing. Temperature control accuracy was
within (0.1 °C.
Channel Surface Passivation. Although some work has been
done to ascertain the steps necessary to prevent poisoning of the
PCR by contaminants from the microfabrication process,25 we have
developed a more extensive method that effectively prevents
contamination in our devices. The walls of the microchannels were
passivated by silanization as follows. The channels were cleaned
(25) Taylor, T. B.; Harvey, S. E.; Albin, M.; Lebak, L.; Ning, Y.; Mowat, I.;
Scluerlein, T.; Principe, E. Biomed. Microdevices 1998, 1, 65-70.

290 Analytical Chemistry, Vol. 75, No. 2, January 15, 2003

thoroughly with filtered double-distilled water to remove the
presence of chromium and other contaminants due to the micro-
fabrication process. The water was then removed by washing with
acetone and the channel dried by applying vacuum. The channel
was filled with a filtered (0.45-µm filter) 5% dichlorodimethylsilane
(DDMS) solution in chloroform. The chip along with a small
beaker containing 1 mL of the DDMS solution was placed in a
desiccator. Vacuum was applied until the solution in the beaker
started boiling. The vacuum inlet was then shut, and the desiccator
was left for 1 h. During this period, the DDMS solution in the
chip and beaker evaporates completely and the vapors form a thin
film on the walls of the channels. The chip was removed and
washed twice with 25 µL of chloroform, twice with 25 µL of filtered
acetone, and twice with 25 µL of filtered water. Washings were
performed by applying vacuum.
DNA and RNA Templates. The DNA template was a 0.23-
kbp fragment synthesized by amplifying the prostate-specific
antigen (PSA) mRNA as described elsewhere.26 The RNA template
was a 1-kb fragment synthesized by in vitro transcription of PSA
cDNA. The primers used for continuous-flow PCR and RT-PCR
of these templates are as in ref 26. The upstream and downstream
primers used for PCR amplification of the human β-globin gene
from genomic DNA correspond to sequences 62 253-62 272
(exon 1) and 62 468-62 487 (exon 2) of the gene and generate a
0.23-kbp product. The plasmid used as a template was a 4-kbp
GFP control vector, and the primers were the T7 promoter and
terminator primers (plasmid from kit Catalog No. 3186148; primers
from kit Catalog No. 3186237; Roche Applied Sci., Mannheim,
Germany). This generates a 1.02-kbp PCR product.
Continuous-Flow PCR. The PCR mixture contained 10
mmol/L Tris-HCl, pH 9.0, 2.5 mmol/L MgCl2, 50 mmol/L KCl, 1
mL/L Triton X-100, 1 mL/L Tween 20, 0.2 g/L bovine serum
albumin (BSA; New England BioLabs, MA), 2.8 µmol/L poly-
(vinylpyrrolidone) (MW 44 000), 0.2 mmol/L each of the dNTPs
(MBI, Fermentas, Lithuania), 0.5 µmol/L each of the upstream
and downstream primers, 0.5 unit/µL thermostable DNA poly-
merase (HT Biotechnology, Cambridge, England), and DNA
template. The heating block temperatures were set at 95 °C for
denaturation, 58 °C for primer annealing, and 72 °C for extension.
Between samples, the PCR channel was washed with a volume
of water equal to the sample volume.
Continuous-Flow RT-PCR. The reverse transcription mixture
contained 50 mmol/L Tris-HCl, pH 8.3, 8 mmol/L MgCl2, 30
mmol/L KCl, 10 mmol/L dithiothreitol, 0.2 g/L BSA, 2.8 µmol/L
poly(vinylpyrrolidone), 0.5 µmol/L downstream primer, 0.5 mmol/L
each of the dNTP, 2 units/µL human placental ribonuclease
inhibitor (HT Biotechnology), 10 units/µL reverse transcriptase
(M-MuLV from Finzyme), and RNA template. The RNA template
and the primer were first heated at 70 °C for 5 min, to reduce
any secondary structures of the RNA, and then cooled on ice.
Afterward the rest of the components were added and the mixture
was introduced into the RT inlet of the chip.
Two protocols were used for performing RT-PCR. In one
protocol (protocol A), the RT reaction mixture was pumped into
the RT zone of the chip at a fast flow rate until it reached the
channel intersection with the PCR reagents. At this point, the flow
was stopped and the RT was allowed to incubate for 30 min at 42
(26) Verhaegen, M.; Christopoulos, T. K. Anal. Chem. 1998, 70, 4120-4125.
°C. At the end of this period, the RT heating block temperature
was raised to 95 °C for 5 min to inactivate the reverse tran-
scriptase. Subsequently, both the RT mixture and the PCR
reagents were pumped continuously and simultaneously with a
ratio of volumetric flow rates set at 1:9, respectively. Therefore, a
10-µL total volume of RT-PCR contains 1 µL of the RT product.
The RT-PCR product was collected at the desired cycle number.
It should be noted that, after mixing of the two solutions, the final
concentrations of the PCR components are as above. In the second
RT-PCR protocol (protocol B), the RT mixture was pumped
continuously into the RT channel at a flow rate of 2 nL/s. Under
these conditions, 30 min is required for the mixture to reach the
intersection with the PCR channel. At this point, the pump
delivering the PCR reagents is activated at 18 nL/s to carry out
the amplification. Between samples, the RT channel and the PCR
channel are washed once with a volume of water eaqual to the
sample volume.
Analysis of Amplification Products. PCR and RT-PCR
products were analyzed by agarose gel (2%) electrophoresis and
ethidium bromide staining. DNA markers were from a φX174
HaeIII digest with fragment sizes of 1353, 1078, 872, 603, 310,
281, 271, 234, 194, 118, and 72 bp (MBI, Fermentas). The products
were quantitated by densitometry of pictures taken with a Kodak
DC120 digital camera while the gels were under UV illumination.
The markers were used as standards for construction of calibration
curve.
RESULTS AND DISCUSSION
The 4 × 4.5 cm RT-PCR chip (Figure 1) enables the functional
integration of reverse transcription and polymerase chain reaction
using a single continuous channel 343.2 cm long. The lengths of
channel segments corresponding to RT and PCR (for 40 cycles)
are 54.4 and 288.8 cm, respectively, covering most of the chip. In
the design process, the needs for compact device design,
minimum linear thermal diffusion distance (e.g., minimum etch
depth) for a given channel volume, and higher cycle count and
channel volumes were balanced with microfabrication yield
challenges associated with densely packed channel meanders and
having only one device per substrate if the design was not
sufficiently compact. Microfabrication of this chip required that
particular care be paid to reduce surface damage to the glass
during subsequent processing. Careful handling minimized the
occurrence of significant channel defects upon etching. Microchips
with smaller active areas are considerably less affected by
occasional scratches or other surface damage. The 4 × 4.5 cm,
40-cycle design used afforded relatively shallow channels (55 µm),
sufficient volume (5.0 and 15.5 µL for the RT and PCR sections,
respectively), yet well-spaced channels and four devices per 4-in.
substrate to enhance yield. The surface-to-volume ratio for the
PCR channel is 57 mm-1 as opposed to a surface-to-volume ratio
on the order of 1 mm-1 and a diffusion distance of 1-2 mm for a
typical 25-µL macroscale PCR. The length and the volume for each
PCR cycle are 7.2 cm and 0.38 µL, respectively. Reaction mixtures
for RT and PCR are introduced into the chip through appropriate
access holes, and the amplification products are collected from
various outlets, thus allowing for cycle number selection. For
example, at a flow rate of 21 nL/s, the observed cycling time is
13 s/cycle. Also, at this flow rate, product collection starts at 5, 6,
7, 8, and 9 min for cycles 20, 25, 30, 35, and 40, respectively. The
Analytical Chemistry, Vol. 75, No. 2, January 15, 2003 291
Figure 2. (Left panel) Effect of the cycle number on the concentration of amplification product in continuous-flow PCR with on-chip cycle
selection. The input DNA was 107 molecules. (Right panel) Study of the concentration of the amplification product as a function of the input DNA
molecules. All products were collected at 30 cycles. The 0.23-kbp PSA DNA template was used. M, DNA markers. N, negative (no DNA template).
respective times for completion of the collection of a 10-µL product
are 13, 14, 15, 16, and 17 min.
Sample introduction into the chip as well as product collection
is accomplished through capillaries connected to the access holes.
Epoxy glue is most commonly used to connect microfluidic
devices with capillary tubes. However, we have found it to be both
time-consuming and often unsuccessful as the epoxy seeps around
the capillary into the access holes, clogging the channel, and
making the chip unusable. Even if a successful connection is made
with epoxy glue, it is still troublesome to remove the glued
capillaries for cleaning the chip. The simple apparatus described
in the Experimental Section and shown in Figure 1b ensures both
an easy connection (within minutes) of the capillaries to the access
holes and convenient disassembly of the chip for cleaning.
The effect of the number of PCR cycles on the amount of
amplification product was studied by using 107 molecules of DNA
template (0.23 kbp of PSA DNA) in 10 µL of reaction mixture.
The products were collected at 20, 25, 30, 35, and 40 cycles, and
the flow rate was set at 21 nL/s. A negative control, containing
no DNA template, was also subjected to PCR for 40 cycles to
ensure the absence of contamination. An electropherogram of the
products and a plot of product concentration as a function of the
number of cycles are presented in Figure 2. In all cases, a single
band was observed at the expected size (0.23 kbp). Initially the
amplification is exponential, but at high cycle numbers the plateau
phase of PCR was reached.
The dependence of the amount of amplification product on the
input DNA template molecules was studied in the range of 2.5 ×
106-160 × 106 molecules (10-µL reaction) at a constant number
of 30 cycles and a flow rate of 21 nL/s. The results are presented
in Figure 2. We observe a linear relationship between the
concentration of accumulated product and the number of input
DNA molecules. Nonspecific amplification products due either to
mispriming or to primer dimers were not observed, even at a high
number of template molecules.
292 Analytical Chemistry, Vol. 75, No. 2, January 15, 2003
The reproducibility of continuous-flow PCR was assessed by
running four identical mixtures containing 107 DNA template
molecules. The products were collected at 30 cycles, analyzed by
agarose gel electrophoresis, and quantified densitometrically. The
calculated CV was 5.7% (n ) 4).
To investigate the capability of the chip for high-throughput
PCR, three samples, containing DNA templates of different sizes,
were run simultaneously on the chip by continuous-flow and the
products were collected at different cycle numbers. Sample 1
contained 107 molecules of a 0.23-kbp DNA template (PSA), and
the amplification product was collected at 25 cycles. Sample 2
contained 109 molecules of a 4-kbp supercoiled plasmid, and the
product was collected at 30 cycles. Sample 3 also contained 107
molecules of the 0.23-kbp template (PSA) and was collected at 35
cycles. The PCR volume was only 2 µL. The samples along with
their appropriate primers were introduced at 5 nL/s from the
PCR1 inlet. The rest of the PCR components were introduced at
5 nL/s from the PCR2 inlet (total flow rate of 10 nL/s). A 2-µL
water plug was used as a wash between samples. A small air
bubble (0.4 µL) was introduced between the sample and the wash.
It was observed that the length of the bubble remained constant
as the solutions were flowing through the chip. When a wash was
encountered, the pump delivering the PCR mix was stopped, and
the flow rate of the other pump was increased accordingly in order
to maintain the 10 nL/s flow rate. After the wash was complete,
the flow rate was decreased back to 5 nL/s and the PCR mix pump
was activated for the next sample. The whole process from
injection of samples to the collection of products after amplification
took 35 min to be completed. The products (2 µL) were subjected
to agarose gel electrophoresis and the results are presented in
Figure 3. The three samples are amplified independently, without
any evidence of cross-contamination. Therefore, the chip offers
the advantage of multiple sample processing.
The amplification of a single-copy gene (β-globin) from human
genomic DNA was also tested by using continuous-flow PCR on
Figure 3. (a) Schematic diagram of the PCR channel of the chip containing reaction mixtures (2 µL) from three samples (black segments) run
simultaneously. Intervening washes (2 µL) are shown as gray segments. A small air bubble (0.4 µL) is introduced between sample and wash.
The sample (with its primers) and the wash are introduced through PCR inlet 1. PCR reagents (enzyme, dNTPs, buffer) are introduced through
PCR inlet 2. (b) Agarose gel (2%) electrophoresis of the amplification products (2 µL) from three continuous-flow PCRs run simultaneously. M,
DNA markers. S1, sample 1 contains 107 DNA molecules (PSA DNA, 0.23 kbp) and the product is collected at 25 cycles. S2, sample 2 contains
109 supercoliled plasmid DNA molecules and the product is collected at 30 cycles. S3, sample 3 contains 107 DNA molecules (PSA DNA, 0.23
kbp) and the product is collected at 35 cycles. (c) Agarose gel (2%) electrophoresis of the amplification product from PCR of 2 (lane 1) and 4
µL (lane 2) of the β-globin gene from human genomic DNA isolated from whole blood. M, DNA markers. N, negative (no DNA template).

the chip. Genomic DNA was isolated from 200 µL of whole blood
using the QIAamp DNA blood mini kit (Qiagen, Germany).
Aliquots of 2 and 4 µL were used for PCR in a total reaction volume
of 10 µL and a flow rate of 10 nL/s. Products were collected at 40
cycles and subjected to agarose gel electrophoresis (Figure 3c).
A single band at the expected size (0.23 kbp) was observed.
RNA amplification by RT-PCR is most often performed as a
two-step reaction. First, cDNA is synthesized by reverse transcrip-
tion. Then, the cDNA is mixed with the PCR reagents for
amplification. The difficulty in integrating the two processes on a
single chip arises from the fact that the components of the RT
mixture interfere with PCR. The Mg2+ concentration of the RT
mixture is 4 times higher than the optimum for PCR. Also, the
reverse transcriptase enzyme inhibits PCR27 and must be inacti-
vated at high temperature (95 °C) prior to amplification. As a
consequence, the RT mixture should constitute about 10% of the
total PCR volume. This was overcome by using two pumps
simultaneously, for RT and PCR, with flow rates at a ratio of 1:9.
We studied the influence of the cycle number on the concen-
tration of amplification product in flow-through RT-PCR (protocol
A) by using a reverse transcription mixture containing 108 RNA
template molecules (1 kb)/µL. The total volume of RT-PCR was
10 µL and contained 1 µL of the RT product (cDNA). A negative
control (with no RNA template) was also included. Figure 4
presents an electropherogram of the amplified products and a
graph of the product concentration as a function of the number
of cycles. The product increases exponentially until the plateau
phase of PCR is reached. A single band at the expected size was
observed in all cases. Nonspecific amplification products were not
observed.
The effect of input RNA template molecules on the amount of
RT-PCR product was studied in the range of 6.25 × 106-200 ×
(27) Chandler, D. P.; Wagnon, C. A.; Bolton, H. Appl. Environ. Microbiol. 1998,
64, 669-677.
106 molecules. Products were collected at 30 cycles. The results
are presented in Figure 4. The amount of product is linearly
related to the input RNA molecules. However, a plateau is reached
at a high number of RNA molecules because of saturation of
reverse transcription and PCR.
We performed various experiments to demonstrate high-
throughput RT-PCR on the chip. Four RT reaction mixtures, 0.7
µL each, were pumped in a continuous-flow manner (protocol B)
into the RT channel at a flow rate of 2.6 nL/s. A 0.7-µL water plug
was used as a wash between samples. Upon reaching the
intersection of RT and PCR channels, the solution was mixed with
the PCR reagents that were flowing at 21 nL/s. When a water
plug was encountered, the PCR reagent pump was stopped,
allowing the RT pump to introduce the wash. Once the next RT
sample reached the intersection, the PCR reagent pump was
reactivated. The products (6.3 µL) were collected after 30 PCR
cycles and then were subjected to agarose electrophoresis. The
results are shown in Figure 5. Samples 1 and 3 contain 4.2 × 107
RNA molecules and samples 2 and 4 contained 2.8 × 106 RNA
molecules. We observe (Figure 5b) clear characteristic bands for
each sample.
It is well known that a major problem of PCR and RT-PCR is
that of false positives arising from the contamination of the
reaction mixture from previously amplified sequences. In continu-
ous-flow PCR and RT-PCR, the same reaction vessel (chip) is used
for amplification of various samples containing a wide range of
DNA and RNA molecules. As a consequence, we were concerned
that the chip might become a source of contamination, i.e., that
amplification products might remain in the chip and be carried
over to the next sample. To examine this, we included a negative
control after a series of samples with high DNA/RNA template
numbers. In all cases, no contamination was observed. Therefore,
washing the chip with an equal volume of water after each sample
is sufficient to avoid carryover. Generally, contamination in PCR
Analytical Chemistry, Vol. 75, No. 2, January 15, 2003 293
Figure 4. (Left panel) Effect of the cycle number on the concentration of amplification product in RT-PCR. The input RNA (1 kb) was 108
molecules. The RT mixture was mixed with the PCR mixture at a flow rate ratio of 1:9. (Right panel:) Study of the concentration of the amplification
product as a function of the input RNA molecules. All products were collected at 30 cycles. M, DNA markers. N, negative (no RNA template in
the RT mixture).

Figure 5. (a) Schematic diagram of the RT channel of the chip containing reverse transcription mixtures (0.7 µL) from four samples (black
segments) run simultaneously. Intervening washes (0.7 µL) are shown as gray segments. A small air bubble is introduced between sample and
wash. Upon reaching the intersection of RT and PCR channels (indicated by an asterisk), the RT solution is mixed with the PCR reagents. (b)
Agarose gel (2%) electrophoresis of the amplification products (6.3 µL) from four continuous-flow RT-PCRs run simultaneously. M, DNA markers.
S1, S3, samples 1 and 3 containing 4.2 × 107 RNA molecules. S2, S4, samples 2 and 4 containing 2.8 × 106 RNA molecules. The products were
collected at 30 cycles. (c) Hand-driven continuous-flow PCR. Electropherogram including DNA markers (M), a negative control (N), and the
product (10 µL) of a PCR mixture containing 108 input DNA molecules. (d) Organic material clogging a channel of the chip after prolonged
usage. (e) Burnt organic material after heating the chip at 300 °C. (f) Cleaned channels after treating the chip with concentrated HNO3. Pictures
in (d-f) were taken by using the Zeiss Axioscope 2 equipped with a Sony video camera model SSC-DC58AP.

depends on the amount of accumulated product. A more sensitive
detection system requires less amplification product, thereby
decreasing the possibility of contamination and enabling the use
of lower ranges of starting DNA/RNA copies (or lower number
of cycles).
An advantage of the flow-through chip for PCR and RT-PCR is
that it can be reused. No special treatment is required after the
294 Analytical Chemistry, Vol. 75, No. 2, January 15, 2003
initial silanization. However, channel clogging is observed after
two weeks of continuous daily usage. The clogs always occur at
the high-temperature zone (95 °C) and are attributed to the
accumulation of denatured protein. The blockage most frequently
occurs during either the extended denaturation step or during
the first few PCR cycles. Perhaps the initial rapid rise of
temperature in these regions contributes to the deposition to the
channel wall. Figure 5d depicts a clogged channel wall obstructing
the flow. To eliminate the problem, we have developed a simple
procedure for removing the clogs effectively. The chip is placed
on a ceramic base into a furnace (Thermolyne, type 48000), and
the temperature is raised slowly (within 1 h) to 300 °C. This
temperature is maintained for 1 h, and the furnace is turned off.
The chip is removed when ambient temperature is reached. This
process burns the organic material inside the channel. A clogged
chip, after baking, is shown in Figure 5e. The heating process
does not damage the chip. Subsequently, 65% HNO3 is aspirated
into the channel by applying negative pressure to remove the
burnt organic material. The acid-filled chip is left overnight at room
temperature and then washed thoroughly with chloroform. Figure
5f shows a chip section after the cleaning process. The clean chip
was silanized before use.
Finally, we investigated performing PCR in the flow-through
microdevice by a simple hand-driven injection and transport of
the reaction mixture, without using any pumps. The reaction
mixture (10 µL) was aspirated into a 0.4-mm (i.d.) Teflon tubing
connected to a 1-mL syringe containing water. PCR was then
performed by applying moderate hand pressure to the syringe
and collecting the amplification products at the desired cycle.
Figure 5c presents an electropherogram of the products from two,
hand-driven, continuous-flow PCRs including a negative control
and a positive containing 108 DNA template molecules. Products
were collected at 30 cycles. The reactions were completed in 6
min (including injection and collection of 10 µL). This is 2.5 times
faster than injection and collection by pump-driven PCR (21 nL/
s) for 30 cycles, which requires a total time of 15 min. Despite
the possible inconsistency of the pressure in the hand-driven flow,
a clear band is observed at the expected size. We have thus
demonstrated a simple and robust system for rapid PCR, and we
envisage future implementations that are based on a flow-through
chip or a cartridge and that are operated by hand-driven injection
and transport of the reaction mixture without pumps. This system
may be particularly useful for the rapid generation of amplified
DNA in the molecular biology research laboratory. Furthermore,
it could be used in field testing, especially for detection of the
presence of a specific target sequence.
In a previous report, RT and PCR were performed on a
microdevice.17 The RNA sample and the reagents, in a final volume
of 50 µL, were placed in a polypropylene tube liner that was
inserted into a thermally cycled silicon cavity. A layer of mineral
oil was added over the PCR mixture to prevent evaporation during
cycling. Following RT-PCR, an aliquot of the mixture was
transferred manually to another tube for a second PCR (nested
PCR). On the contrary, in the present work, continuous-flow RT
and PCR are truly integrated for the first time on a single
monolithic device. No intervention is required between the RT
and PCR steps due to automated introduction and mixing of
samples and reagents. Considerably faster cycling times are
achieved. For instance, at a typical flow rate of 21 nL/s, the cycling
time is 13 s/cycle compared to 118 (“normal” protocol) and 70
s/cycle (“fast” protocol) reported in ref 17. No mineral oil is
required for PCR. Moreover, we have demonstrated that RT can
be performed in final reaction volumes as low as 0.7 µL. Several
RNA samples can be introduced into the chip in a continuous-
flow mode for high-throughput analysis. Additionally, RT may
precede PCR, or PCR can be chosen alone and to different
numbers of cycles. This is a significant advantage because it allows
control of the amount of amplification product depending on the
abundance of starting RNA or DNA copies in the sample. It was
demonstrated that multiple samples containing DNA sequences
of different sizes can be run simultaneously on the chip and the
products can be collected at different cycle numbers. Because it
is a continuous-flow system, the sample volume can be varied in
order to optimize amplification.
Throughout this work, the amplification products were ana-
lyzed by electrophoresis and ethidium bromide staining. It is
conceivable that the chip may be coupled in-line with other
separation and detection systems such as capillary electrophoresis
with laser-induced fluorescence detection.
ACKNOWLEDGMENT
This work was supported by a grant to T.K.C. from the Institute
of Chemical Engineering and High Temperature Chemical Pro-
cesses (ICE/HT), Patras, Greece, and the General Secreteriat of
Research and Technology, Greece. We gratefully acknowledge
the support of the Natural Sciences and Engineering Research
Council of Canada to C.J.B.. We thank Martin Kopp and Andreas
Manz for helpful discussions. We also acknowledge the financial
support of Micralyne Inc.
Received for review August 5, 2002. Accepted November
11, 2002.
AC0260239
Analytical Chemistry, Vol. 75, No. 2, January 15, 2003 295