DNA KINETICS IN MICROFABRICATED DEVICES

Yick Chuen Chan1, Rosie Ming Sum Ma1, Maria Carles2,3, Nikolaus J. Sucher2,4, Man Wong5 and Yitshak Zohar1
1
Department of Mechanical Engineering, 2Biotechnology Research Institute, 3Department of Biochemistry,
4
Department of Biology, 5Department of Electrical & Electronic Engineering
Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong
Tel: +852 23587194, Fax: +852 23581543, Email: mezohar@ust.hk

ABSTRACT to minimize the MCE chip size while maintaining a

long separation channel [2,3,7]. Thus far, though, the
The DNA kinetics in micro-capillary motion of a DNA plug through a bend has been
electrophoresis is presented. The mobility and analyzed only qualitatively. Splitting an injection plug
diffusion coefficient of 14bp-DNA fragments as a into several channels to carry out different tests is
function of concentration in two types of separation another attractive feature. However, the DNA mass
sieving matrices, hydroxyethylcellulose (HEC) fraction injected to each downstream branch is not
polymer solution and agarose gel, are extracted known. These issues indicate that despite the
through a series of experiments performed in micro- widespread utilization of micro-capillary
fabricated devices. In addition, the motion of a DNA electrophoresis, detailed studies of the DNA kinetics
plug through a miter bend and splitting a plug in a in microfabricated devices with various designs are
branch are quantitatively characterized. The concept still needed.
of equivalent length is introduced to quantify the Recently, we presented a novel and robust
effect of a bend on the DNA plug motion. In a integrated micro-capillary electrophoresis system with
branching system, a simple kinematic relationship was feed-through electrodes, fabricated using amorphous-
discovered relating the quantity of DNA in each silicon-assisted glass-to-silicon anodic bonding
downstream branch to its relative channel cross- technology [8]. The use of both silicon and glass
sectional area. substrates with integrated circuits facilitates the
design of an automated lab-on-chip. In the present
INTRODUCTION work, this microsystem has been used to study DNA
kinetics in a variety of microchannels.
Powered by major advances in microfabrication
technology, the development of a fully automated lab-
on-chip for genetic assays has been attracting major
research and commercial interest. In recent years,
numerous miniaturized capillary electrophoresis
Wire-bonds Lead-frame
systems with high performance have been designed
and fabricated. The majority of those microsystems Reservoirs
were fabricated with bonded, bulk micromachined
glass substrates with external electrodes [1,2,3].
Although higher voltages can be applied to those
microsystems, the heat dissipation problem and large- Separation
scale detection systems limited their potential of channel Injection channel
becoming good candidates for a lab-on-chip. Fig. 1: A picture of a packaged integrated micro-
Although DNA analysis using capillary system showing the micro-channels, reservoirs, lead-
electrophoresis has been exercised for more than a frame and wire-bonds.
decade, detailed characterization of the fundamental
processes involved in DNA kinetics, such as diffusion
DEVICE DESIGN AND FABRICATION
or separation, is still lacking. In slab gel
electrophoresis, DNA electrophoretic mobility was
The detailed design and fabrication of the
already well characterized [4], and the mobility of
integrated MCE device is described elsewhere [8].
DNA fragments in polymer solutions was also
The injection and separation channels were etched
measured [5,6]. However, little has been said about
either in a 100mm, Corning 7740 glass wafer or in a
DNA kinetics in microfabricated capillary
100mm, P-type, (100) Si wafer. Platinum electrodes,
electrophoresis (MCE) devices. Various designs
with TiW as a glue layer, were sputtered and
incorporating bends along the channel were adopted
patterned on a second wafer using the lift off process.

0-7803-7185-2/02/$10.00 ©2002 IEEE 60

 A thick thermal oxide layer was grown on the Si This equation facilitates the calculation of the
wafer either before the Pt-electrode sputtering (if the mobility, P=xm/Et, and diffusion coefficient, D=V2/2t,
channels were etched in the glass wafer) or after from the measurements of the time-dependent peak
channel etching (if the channels were formed in the Si location xm and standard deviation V of the DNA
wafer) for electrical insulation. This was followed by distribution in the separation channel. Several images
the deposition of a very thin amorphous-silicon layer were recorded as the plug travelled along the
to significantly improve the yield of the glass to oxide separation channel, under applied electric field, in
anodic bonding. The two processed wafers were then either the HEC polymer solution or the agarose gel.
anodically bonded under 900V at 335oC for 1 hour. A typical set of velocity measurements of 14bp-
The process was finally completed with wafer cutting DNA fragments as a function of the electric field for
and device packaging. A picture of a packaged device different agarose gel concentration is plotted in Figure
is shown in Figure 1. 2. Indeed, the velocity increases linearly with the
Prior to their use, the capillary walls of the electric field strength. The constant slope, which is the
fabricated devices were treated using a modified mobility, increases with decreasing concentration. The
Hjerten procedure [9,10], which consists of a 0.1M experiments were repeated for HEC, and the extracted
NaOH rinsing and a polyacrylamide coating. The mobility is shown in Figure 3 as a function of the
NaOH rinsing renders the capillary wall hydrophilic concentration of both matrices. The mobility
and, hence, facilitates the loading of the sieving decreases exponentially as the matrix concentration C
matrices. The polyacrylamide coating suppresses increases, P=P0˜exp(-KrC); Kr is known as the
electro-osmotic flow (EOF) for the subsequent retardation coefficient, which is a characteristic of a
experiments. This is especially important when using given molecular species in a particular molecular
polymer solution, rather than gel, as the sieving system. Interpolation of the curves to zero
matrix because EOF results in a bulk motion of both concentration, yields the constant P0=3.7˜10-4 and
the sieving matrix and the DNA, which in turn affects
3.3˜10-4cm2/V/s for agarose and HEC, respectively,
the experimental measurements.
which is the free-solution mobility.
EXPERIMENTAL SET-UP 400

350 Agarose 0.5%

The packaged MCE device was loaded with the Agarose 1%
300
appropriate amount of sieving matrix and DNA Agarose 2%
Speed ( P m/s)

labeled with FITC, and connected to a voltage divider 250

Agarose 3%
as described in [8]. Upon the application of an electric 200

field in the injection mode, a pinched DNA plug was 150

obtained. Instantaneous switching to separation mode 100

resulted in a well-defined DNA plug moving into the 50
separation channel. The voltage divider was then 0
disconnected, and direct electric field was applied to 0 20 40 60 80 10 0 12 0
the separation channel to conduct the experiments. E lectric field (V/cm )

The experiments were viewed using a CCD camera Fig. 2: Plug peak velocity dependence on the electric
mounted on a microscope. Image acquisition and field for various concentrations of agarose (14bp).
analysis software captured the DNA motion. The
location and concentration distribution of DNA could 0.0005
then be extracted from the light-intensity distribution agarose
M obility [(cm /s)/(V/cm )]

observed in the still pictures. 0.0004 H EC

0.0003
RESULTS AND DISCUSSION
0.0002
The motion of a DNA plug in a microchannel,
under an applied electric field E, is governed by the 0.0001
one-dimensional convective diffusion equation, which
admits a normal-distribution solution for the DNA 0
0 1 2 3
concentration C(x,t) as follows:
Concentration [% ]
M ª  x  x m 2 º Fig. 3: Mobility dependence on the concentration of
C  x, t  exp « » (1)
S 4SDt 4 Dt »¼ the agarose and HEC matrix (14bp).
«¬

0-7803-7185-2/02/$10.00 ©2002 IEEE 61

 Diffusion Coefficient [P m /s] 400 150
2

agarose

Standard deviation [ P m]
130
300
H EC
110
200
90

100
70

0
50
0 1 2 3 5000 7000 9000 11000 13000 15000
C oncentration [% ] Disan ce from intersection [ P m ]

Fig. 4: Diffusion-coefficient dependence on the Fig. 6: Evolution of the 14bp-DNA standard deviation
concentration of the HEC and agarose matrix (14bp). as the plug moves through a miter bend, showing the
equivalent length, Le, of the straight channel that
Another set of experiments was conducted to would result in the same band broadening as the
determine the diffusion coefficient. Similar to the bend.
mobility experiments, a DNA plug was injected into
the separation channel. However, in this case, the In this concept, the band broadening due to the
electric field was turned off once the plug cleared the plug motion in a straight microchannel is compared
channel intersection. The variance of the DNA with the band broadening due to the bend. The
distribution in the channel increased linearly with standard deviation of the DNA concentration, as the
time, and the diffusion coefficient is half the slope. plug moves through the miter-bend, is plotted and
Consistent with the mobility measurements, the compared with the standard-deviation evolution along
estimated diffusion coefficient for agarose is higher a straight channel in Figure 6. The standard deviation
than that of HEC as shown in Figure 4 as a function increases linearly with the plug travel distance in the
of the matrix concentration. straight section before and after the bend with almost
Several designs have incorporated bends along the the same slope. The equivalent length of the bend is
channel to lengthen the separation channel without the length of the straight channel that would result in
increasing the MCE chip size. However, the motion of identical bend-induced band broadening. In this
a DNA plug around a bend has been analyzed only example, this equivalent length of the bend is about
visually up to now. The plug motion through a miter 4000Pm. The equivalent length may depend on the
bend, a sharp turn of 90o, is shown in the picture arbitrary decision at which location the bend effect is
series in Figure 5, where the DNA distribution is over. This location is the point the band-broadening
significantly altered due to the bending motion. The slope thereafter is equal to the slope before the bend.
concept of equivalent length, Le, is proposed in order In a similar fashion, the effect of various
to quantify the effect of the bend on the plug motion geometrical features, like local contraction or
and, thus, allow a systematic comparison between the expansion of the separation channel, on the band
effects of different bends. broadening can be quantified for design purposes.

Fig. 7: Plug motion through a branch in the

separation channel: (a) upstream and (b) downstream
Fig. 5: The DNA plug motion around a bend in the of the branch.
separation channel; the time sequence is a-b-c-d.

0-7803-7185-2/02/$10.00 ©2002 IEEE 62

 CONCLUSIONS
200 A B-1 B-2
Measurements of the diffusion coefficient and
mobility of DNA short fragments (14bp) in a
Intensity [arb]

150

microfabricated channel are reported using either

100
agarose or HEC as the sieving matrix. The average
50
free-solution mobility of 3.5˜10-4cm2/V/s is in good
agreement with measurements published in standard
0 electrophoresis systems. The mobility and diffusion
-350 -250 -150 -50 50 150 250 350 coefficient in agarose are higher than those in HEC.
Distance [ P m ] The concept of equivalent length is introduced to
quantitatively characterize the plug motion through a
Fig. 8: 14bp-DNA light intensity distributions
miter bend. Using this concept, the effect of various
upstream (A), and downstream (B-1 and B-2) of the
geometric features on band broadening within the
branch.
separation channel can be evaluated. Finally, a general
kinematic relation for predicting the DNA mass
Splitting an injection plug into several channels, as
fraction injected into several downstream from a
demonstrated in Figure 7, to simultaneously carry out
single upstream branch is proposed.
several tests on the same DNA sample is another
attractive geometrical feature. However, the mass
fraction injected into each downstream branch from
ACKNOWLEDGMENTS
the upstream branch is not known. The light intensity
This work is supported by the Hong Kong
distribution, observed in the pictures of Figure 7, is
Research Grant Council through RGC grant
directly proportional to the DNA concentration. Based
HKUST6082/00E, the Industry Department of the
on these pictures, the light intensity distributions in
Hong Kong SAR (AF/150/99) and the Hong Kong
the upstream branch (A) and the two downstream
Jockey Club.
branches (B-1 & B-2) are plotted in Figure 8.
According to Equation 1, the DNA concentration in
each straight branch is proportional to the DNA mass REFERENCES
per unit cross-section area, M/S. Since these are
normal distributions, the area under each curve is [1] D.J. Harrison, K. Fluri, K. Seiler, Z. Fan, C.S.
directly proportional to the total mass within each Effenhauser & A. Manz, Science, vol. 261, pp.
branch divided by its cross-section area, Mi/Si. 895-897, 1993.
Conservation of mass requires MA=MB-1+MB-2. [2] S.C. Jacobson, R. Hergenroder, L.B. Koutny, R.J.
Indeed, we found that SAAA=SB-1AB-1+SB-2AB-2, where Warmack & J.M. Ramsey, Analytical Chemistry,
A denotes the area under the distribution curve. vol. 66, pp. 1107-1113, 1994.
Furthermore, we discovered that the area under the [3] D. Schmalzing, L. Koutny, A. Adourian, P.
distribution of each downstream branch is the same, Belgrader, P. Matsudaira & D. Ehrlich,
AB-1=AB-2. Namely, the average ‘density’ in each Proceedings of the National Academy of Sciences
downstream branch is the same, MB-1/SB-1=MB-2/SB-2. of the United States of America, vol. 94, pp.
Therefore, at least for the current system of two 10273-10278, 1997.
branches, the DNA mass injected into each [4] K. Strutz & N.C. Stellwagen, Electrophoresis,
downstream branch is proportional to its relative vol. 19, pp. 635-642, 1998.
cross-sectional area. A general kinematic relationship [5] W.M. Sunada & H.W. Blanch, Electrophoresis,
for the mass injected into each downstream branch vol. 19, pp. 3128-3136, 1998.
from a single upstream branch can be expressed as: [6] A.E. Barron, H.W. Blanch & D.S. Soane,
n Electrophoresis, vol. 15, pp. 597-615, 1994.
[7] F.E. Regnier, B. He, S. Lin & J. Busse, Trends in
¦M
i 1
i
Biotechnology, vol. 17, pp. 101-106, 1999.
Mj n
Sj (2) [8] Y.C. Chan, R. Lenigk, M. Carles, N.J. Sucher, M.
Wong & Y. Zohar, Transducers’01, pp. 1166-
¦S i 1
i
1169, 2001.
where n is the number of downstream branches. [9] S. Hjerten, Journal of Chromatography, vol. 347,
Equation 2 was shown to be valid only for two pp. 191-198, 1985.
downstream branches, and more experiments are [10] S. Hjerten & M. Kiessling-Johansson, Journal of
needed to verify it for more branches. Chromatography, vol. 550, pp. 811-822, 1991.

0-7803-7185-2/02/$10.00 ©2002 IEEE 63