Electrophoresis 2003, 24, 1253–1260 1253

Lung-Ming Fu Low-voltage driven control in electrophoresis

Ruey-Jen Yang
microchips by traveling electric field
Department of
Engineering Science, This paper presents the use of a physical model and numerical simulation in the inves-
National Cheng Kung University, tigation of traveling electric fields on capillary electrophoresis (CE) chips. The principal
Tainan, Taiwan material transport mechanisms of electrokinetic migration, ionic concentration, fluid
flow, and diffusion are all taken into consideration. Traditionally, the high electric field
strength required for the separation of biological samples by microfluidic devices has
involved the application of high external voltages. In contrast, this study presents a
proposal for samples separation by means of a moving electric field within a low volt-
age-driven CE chip. Under this proposal, the separation channel is partitioned into a
series of smaller separation zones by means of electrode pairs. This paper considers
two different electrode configurations, namely arranged along a single side of the
separation channel, and arranged on two sides of the separation channel. The quality
of the separation achieved with these two configurations is then compared with the
traditional straight separation channel approach. The results confirm that the pro-
posed method is successful in maintaining an adequate field strength for separation
purposes in a low-voltage driven CE chip. Furthermore, it is determined that the best
separation results are obtained using electrodes arranged along both sides of the
separation channel.

Keywords: Double-banked electrode channel / Capillary electrophoresis chips / Miniaturization /

Traveling electric field EL 5343

1 Introduction ments. Generally speaking, DNA separation requires an

Recent years have seen an increasing interest in the
development of microfluidic devices for chemical or bio-
logical applications, including the separation of DNA for
genetic engineering purposes. Separation techniques
electric field strength in the range of 300–800 V/cm, and
an applied voltage in the order of 1–3 kV.
When AC-type traveling electric fields (dielectrophoresis)
are used in the separation of cells and dielectric particles,

for the microdevices used in such applications typically
include capillary electrophoresis (CE) [1–5], synchronized
cyclic electrophoresis (SCE) [6], free-flow electrophoresis
[7, 8], gel electrophoresis [9, 10], and electroosmotic flow
[11–14]. These techniques exploit the fact that dielectric
particles move in the opposite direction to an applied
electrical potential under the influence of the resulting
electric field. Furthermore, different dielectric particles
exhibit different mobility characteristics. Regarding DNA
applications, the traditional CE chip-based approach
has involved establishing a potential difference across
the two ends of the separation channel. This voltage
must be high enough to create an electric field of suffi-
cient intensity to bring about separation of the DNA frag-
Correspondence: Prof. Ruey-Jen Yang, Department of Engi-
neering Science, National Cheng Kung University, Tainan, 70101
Taiwan
E-mail: rjyang@mail.ncku.edu.tw
Fax: +886-6-276-6549
Miniaturization
they have the ability to drive the cells at a rate of around
25–30 mm/s [15–17]. The use of DC-type moving electric
fields in the CE process was also considered [18]. As
stated above, the conventional separation process using
CE microchips involves the application of a voltage in the
range of 1–3 kV to the separation channel. Recently,
external voltage flow control has been extended to micro-
fluidic devices [19–23]. They show measurable flow con-
trol with the application of the voltage about 100–150 V.
Though the driving potential is lower, the necessary
field strength for separation purposes is maintained. This
is achieved by partitioning the separation length with
arrayed electrodes into a series of smaller separation
zones. The principle of the separation approach is to use
a DC voltage to energize particular sets of these elec-
trodes in a controlled sequence such that the sample is
driven along the channel and becomes separated.
We propose to partition the separation channel into many
smaller separation zones using arrayed electrodes. Fig-
ure 1 provides a schematic view of the separation channel

 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 0173-0835/03/07-0804–1253 $17.501.50/0
1254 L.-M. Fu and R.-J. Yang
Figure 1. Schematic layout of the subdivision of the
separation channel showing the principal design param-
eters.
subdivision. The principal design parameters are clearly
shown, namely the interval between two successive elec-
trodes, D, and the span of the separation zone, which lies
between any two active electrodes, S. Figure 2 provides
two schematic views of the arrayed electrode design, and
indicates how sample driving and DNA fragment separa-
tion are achieved by moving the electric field sequentially
along the channel. Switching the electrode pairs on and
off successively causes the sample to be driven pro-
gressively through the individual separation zones, which
are established within the currently active electrode pair.
When a particular electrode pair is active, an electric field
is established in the interval between the two electrodes.
The intensity of the field within this zone is sufficient to
bring about DNA separation. Therefore, the proposed
method both drives and separates the sample.
It is known that the velocities of the DNA fragments are
determined by the electric field strength and by certain
buffer solution properties. By considering the charge and
weight of the DNA fragments, the electric field strength,
and the required detected signal resolution of the sepa-
rated DNA fragments, it is possible to determine an ap-
propriate design for the separation subdivisions, and the
optimal arrayed-electrode configuration.
Electrophoresis 2003, 24, 1253–1260
Figure 2. Schematic diagram of an arrayed-electrode
design and the mechanism for driving and separating
DNA fragments using a moving electric field with (a) sin-
gle-sided electrodes; (b) double-banked electrodes.
In the present investigation, two electrode configurations
are considered, namely electrodes arrayed along one side
of the separation channel, and electrodes arrayed along
opposite sides of the channel. These two configurations
are presented in Figs. 2a and b, respectively. The current
study provides a detailed description of the design of low-
voltage driven CE microchips using the traveling electric
field method, and compares the results with those ob-
tained from the traditional straight separation channel
approach. With the moving electric field technique, the
applied voltage for generating the necessary electric field
to drive and separate samples can therefore be lowered.
2 Formulation and numerical method
Pressure-driven flow through a separation channel re-
quires the use of a micropump. The production cost of
such a separation system is expensive, and the operation
process is difficult to control. A more convenient alterative
is to drive fluid motion by applying an electric field to the
net fluid charge in order to produce a body force. This
method is commonly referred to as electroosmotic flow.
Although most previous studies regarding eletroosmotic
flows [24–28] assumed the velocity profile to be fully
developed in the microchannels, and considered the
charge density to conform to the Boltzmann equilibrium
distribution, these assumptions are not strictly valid since
Electrophoresis 2003, 24, 1253–1260
electroosmotic flow involves an entrance region within
which neither condition is satisfied. To take account of
this entrance effect, the current authors have previously
developed physical models based on (i) the Poisson
equation for the electric potential and zeta potential,
(ii) the Nernst-Planck equations for ionic concentration,
and (iii) the full Navier-Stokes equations modified to
include the effects of the body force due to the electric
and the charge density [11, 12, 29–31].
In order to simulate the distribution of a sample plug, it
is necessary to determine an equation which expresses
its concentration. The computational domain considered
in the current simulation extends from the entry region to
the fully developed region. Since the flow rate is very
small, the entrance length is very short, and has there-
fore only a marginal influence upon the separation re-
sults. By introducing the following reference quantities,
Lref, Uref, pref, and rref and defining re ˆ re =n0 ze ˆ
…n‡ n †=n0 , c ˆ zec=k b T, t ˆ tUref =Lref , u ˆ u=Uref ,


v ˆ v=Uref , x ˆ x=Lref , y ˆ y=Lref , p ˆ …p pref †= rU2ref ,
r ˆ rf =rref , n‡ ˆ …n‡ n0 †=n0 , and n ˆ …n n0 †=n0 ,
the dimensionless forms of the governing equations after
dropping the head symbols can be written as:
k2
r2 c ˆ r (2)
2 e
qni * 1 1
‡ u
rni ˆ r2 ni ‡ ‰r…ni rc†Š (3)
qt ScRe ScRe
*
r
u ˆ 0
* lengths of the injection channel and the separation chan-
qu * * 1 2*
‡ u
ru ˆ rp ‡ r u Gx re rc
qt Re
qC * 1 The numerical study assumed that the microchannel was
‡ u
rC ˆ r2 C
qt ScRe
In these expressions, k = WK, where K is the Debye-
Huckel parameter and is given by K = (2n0z2e2/ee0kbT)1/2,
1/K is the characteristic thickness of the charge density,
re = (n12 n2)ze is the charge density, n1 and n2 are the
concentrations of the positive and negative ions, e is the
dielectric constant of the medium, e0 is the permittivity
of a vacuum, n0 is the bulk concentration of the ions,
kb is the Boltzmann’s constant, T is the absolute temper-
ature, Lref = W, where W is the channel height, Uref =
Cinletee0uzu/mL, Cinlet is the activated potential at the inlet,
Re is the Reynolds    and is given by Re ˆ
numbers
rf Uref Lref =m ˆ rf cinletmLee0 jzj Lmref , Sc is the Schmidt number
and is given by Sc = m/rfDi, m is the liquid viscosity,
rf is the fluid density, Di is the diffusion coefficient of
sample, z is the surface zeta potential, p is the pressure,
Gx is the ratio of the electrical double layer (EDL) energy
to the mechanical kinetic energy and is given by, Gx =
2n0kbTrfW2/mRe2 and finally, C is the sample concentra-
tion. For the flow conditions considered in the current
Traveling electric field within CE chips 1255
investigation, the value of Sc in the equations given
above is of the order of 105, while Re is of the order of
one. Although this implies that the diffusion term is very
small in comparison to the convection term, diffusion is
one of the primary mechanisms for sample dispersion,
and accordingly, the term is retained in the numerical
simulation. Since the sample concentration is carried
passively by the flow field, Eq. (6) can be solved sepa-
rately once the flow field has been determined.
In most electrophoresis experiments, the microchannel is
filled with an aqueous electrolyte solution. The sample is
injected into the channel, and an electric potential is then
activated in order to carry out sample separation. Conse-
quently, the assumed initial conditions are that the veloc-
ity is static throughout the system, and that the zeta
potential of the wall is given by Cuwall = z. Furthermore,
the initial ionic concentration and potential are assumed
to conform to the Boltzmann equilibrium condition. The
detailed expressions of the initial conditions and the
boundary conditions are provided in [29].
3 Results and discussion
In the present investigation, a CE chip was fabricated
onto a glass substrate of dimensions 7562562 mm,
and was composed of two layers of micromachined glass
bonded together. The microchannels were fabricated
(4) using standard microfabrication technology [31, 32]. The
(5) nels were 10 mm and 40 mm, respectively, and the cross-
section of each channel was 80 mm wide by 40 mm deep.
(6)
made of silica glass and that the buffer fluid was sodium
borate. The physical and electrical properties of the liquid
were taken to be as follows: dielectric constant e = 80,
zeta potential of the channel wall z = 275 mV, electro-
kinetic diameter k = 32, fluid viscosity m = 1023 Ns/m2,
Schmidt number of the charge density Sc = 105, and diffu-
sion coefficient of the sample D = 6.9610211 m2/s.
3.1 Traditional straight separation channel
A traditional electrophoresis separation employs a straight
separation channel and a large driving voltage since this
arrangement provides a better sample band shape in the
detection area for resolution purposes. In this type of
microchannel flow, an applied electric field acts on the
net fluid charge near the wall to produce a body force
which drives fluid motion. Figure 3 shows typical plots of
the electric potential field and the velocity vector profile
in the traditional straight separation channel. It can be
seen that the electric field is uniformly distributed in the
1256 L.-M. Fu and R.-J. Yang
Figure 3. (a) Electric potential distribution; (b) velocity
vector profile in the traditional straight separation chan-
nel.
separation channel, and that the velocity vectors have a
flat profile. These characteristics provide excellent condi-
tions for sample flow and separation. Figure 4 shows the
typical distribution of a sample band as it flows along a
straight separation channel, i.e., from location (a) to the
detection area at location (e). Despite a slight dispersion
of the sample at the detection area, it can be seen that its
shape remains suitable for detection purposes.
Figure 4. Sample band concentration distribution along
a traditional straight separation channel with a voltage of
1.2 kV.
The DNA sample used in the current straight separation
channel investigation was a HaeIII-digested fX-174 DNA
consisting of eleven fragments, i.e., 72, 118, 194, 224,
271, 281, 310, 603, 872, 1078, and 1353 base pairs (bp).
A 0.5 mL DNA sample with a concentration of 1 mg/mL was
injected into a conventional cross-type CE chip. CE buffer
Electrophoresis 2003, 24, 1253–1260
was 1.2% HPMC (hydroxypropylmethylcellulose) in TBE
(Tris-borate-EDTA) with 1% Cy5 fluorescence dye. The
injection voltage was 300 V for a duration of 30 s. The
separation voltage was 1.2 kV for a duration of 2 min.
The resulting migration velocity served later as a bench-
mark for the design of the traveling electric field chips.
The separated DNA fragments were detected using a
fluorescent microscope (BX60; Olympus, Japan). The
excitation light was generated using a mercury lamp,
which was filtered by a 530–550 bandpass filter, and
then focused on the detection spot of the CE chip through
the fluorescent microscope. The induced fluorescence
from the DNA sample was filtered using a 580 nm high-
pass filter, and then detected through a fluorescence
microscopy technique using a photomultiplier tube (R928;
Hamamatsu, Japan). Figure 5 presents electrophero-
grams generated from an image collected at the detec-
tion area of a straight separation channel. It can be seen
that the fX-174 DNA separation has 11 obvious peaks.
These experimental results provided in this figure serve
as a benchmark against which to compare the results of
the two traveling electric field CE chips which are de-
scribed in the following subsections.
Figure 5. Electropherograms of the separation result for
the fX-174 DNA (HaeIII digested) sample generated from
an image collected at the detection area of a traditional
straight separation channel with a voltage of 1.2 kV.
3.2 Single-sided electrode separation channel
A numerical simulation was conducted in order to analyze
the electric potential distribution, the velocity vector pro-
file, and the travel characteristics of a single sample band
within a low-voltage driven CE chip having a single-sided
electrode separation channel. The simulation considered
a separation channel of 40 mm length and 40 mm height.
Twenty-one electrode pairs were arranged along the bot-
tom of the separation channel with a pitch, D, of 2 mm.
The span between active electrodes, S, was taken to be
6 mm. Accordingly, the voltage required to drive and
Electrophoresis 2003, 24, 1253–1260
separate the DNA fragments was reduced to about 30%
of the voltage required for the straight separation channel
case.
Figure 6 shows the resulting electric potential distribution
and velocity vector profiles between two active electrodes
separated by a span of 6 mm with a voltage of 360 V.
The electric potential distribution is nonuniform at the
two extremities of the span (Fig. 6a). The nature of the
field distribution would be expected to result in an asym-
metric velocity distribution since it is the effect of an
electric field acting on the net fluid charge near the wall
which produces the necessary body force to drive fluid
motion. This is confirmed by the results presented in
Fig. 6b, which indicates a skewed velocity distribution in
the region of the two active electrodes. Figure 7 shows
the computed sample band concentration distribution
along a single-sided electrode separation channel. At
location (a), which indicates the first separation zone, the
Figure 6. (a) Electric potential distribution; (b) velocity
vector profile within an interval of one span S = 6 mm in a
single-sided electrode separation channel with a voltage
of 360 V.
Figure 7. Sample band concentration distribution along
a single-sided electrode separation channel.
Traveling electric field within CE chips 1257
sample band concentration is already skewed. It can be
seen that the band becomes increasingly distorted as it
flows through the subsequent separation zones within
the partitioned separation channel, i.e., towards the
detection area at location (e).
Figure 8 compares the typical sample band shape and
peak intensity at the detection area for the cases of a
straight separation channel and a single-sided electrode
separation channel. It is possible to define a shape error
value, Ê, of the sample in the single-sided electrode chan-
nel by comparing it with the benchmark result obtained in
the straight separation channel case, i.e., Ê = [(detection
shape area – maximum overlap area)/the shape area
of straight separation channel]100%. A small value of Ê
implies that the shapes of the two samples are similar,
and that detection results will be satisfactory. However,
the results in Fig. 8a suggest that the band shape ob-
tained with the single-sided electrode separation channel
is unsuitable for detection purposes, and indeed, analysis
shows the shape error value, Ê, to be approximately
80.3%. Figure 8b compares the peak intensity of the
sample band within a single-sided electrode separation
channel with the intensity in the straight separation chan-
nel. While the peak in concentration intensity achieved
Figure 8. Comparison of (a) band shape and (b) peak
intensity at the detection area between a single-sided
electrode separation channel and a traditional straight
separation channel.
1258 L.-M. Fu and R.-J. Yang
with the straight channel is obvious, it can be seen that
the intensity within the single-sided electrode channel is
extremely weak and widely dispersed.
Figure 9 presents electrophoerograms generated from
images collected at the detection area of the traditional
straight separation channel and numerical simulation re-
sults for a single-sided electrode separation channel. The
sample consists of 1024 M rhodamine B and 6.561023 M
Cy3 fluorescent dye. From experimental results, the sam-
ple separation exhibits four obvious peaks, which corre-
spond to the three fragments of Cy3 in the hydrolysate
and the single fragment of rhodamine B within the sam-
ple. However, from numerical simulations, for a single-
sided electrode separation channel, Fig. 9 (dashed line)
shows the peak concentration intensity at the detection
area by simulation for sample fragments with concentra-
tion values of 2 (rhodamine B), 0.45 (Cy3(a)), 1.5 (Cy3(b)),
and 1.1 (Cy3(c)), respectively. The electroosmotic mobility
ratio of four fragments are rhodamine B : Cy3(a) : Cy3(b) :
Cy3(c) = 1 : 0.75 : 0.54 : 0.47. Injecting the fragments at
the same location and with the different electroosmotic
mobility simulates the convection of the fragments. The
peak intensity of the sample concentration in detection
PN
area is defined as C ˆ Ci =N, where C is the average
iˆ1
concentration intensity in the detection area, Ci is the
local concentration intensity in the detection area, and N
is the total number of points in the detection area. Figure 9
(dashed line) indicates that the computed concentration
value of the rhodamine B peak is about 1.07, and the
Cy3(a) is unobvious. This corresponds to a dispersive
ratio of 46.5% (i.e.,(221.07)/26100%). It is noted that
this ratio is very high, and therefore the detection results
are unlikely to be satisfactory from the current single-
sided electrode separation channel.
Figure 9. Comparison of electropherograms generated
from an image collected in a traditional straight separa-
tion channel by experimental (solid line) and intensity of
concentration fragments in a single-sided electrode
separation channel with numerical simulation (dashed
line) at the detection area.
Electrophoresis 2003, 24, 1253–1260
3.3 Double-banked electrodes separation
channel
The effect of employing double-banked electrodes along
the separation channel is to increase the electric potential
strength in the upper regions of the channel. Figure 10
shows the electric potential distribution and velocity
vector profiles which result when using double-banked
electrodes separated by span S = 6 mm and a voltage of
180 V. The current simulation considered a separation
channel of 40 mm length, a span, S, of 6 mm, a pitch be-
tween successive electrodes, D, of 2 mm, and a total of
21 electrode pairs located along the upper and lower
regions of the separation channel. The voltage required
to drive and separate the sample is reduced to approxi-
mately 15% of the voltage required within the straight
separation channel. The simulated results in Fig. 10a
show that the double-banked electrodes create a sym-
metric electric potential distribution in the separation
channel between the two active electrode pairs. There-
fore, as would be expected, Fig. 10b indicates that the
resulting velocity profiles are also flat and symmetric.
Figure 11 shows the typical concentration distribution of a
sample as it passes along a double-banked electrode
separation channel. It can be seen that the sample con-
Figure 10. (a) Electric potential distribution; (b) velocity
vector profile within an interval of one span S = 6 mm in a
double-banked electrode separation channel with a volt-
age of 180 V.
Figure 11. Sample band concentration distribution along
a double-banked electrode separation channel.
Electrophoresis 2003, 24, 1253–1260 Traveling electric field within CE chips 1259

centration distribution is uniform and flat within each of Figure 13 presents electrophoerograms generated from
the intermediate separation zones, i.e., there is very little images collected at the detection area of the traditional
tilting or distortion of the sample at locations (a)–(d). When straight separation channel and numerical simulation
the sample arrives at the detection area, i.e., at location results for a double-banked electrode separation chan-
(e), its band shape is sufficiently good to ensure satis- nel. The peak intensities of the four samples are very
factory detection results. Figure 12 shows the band shape clearly distinguishable. The dispersive ratio is determined
and peak intensity of the sample at the detection area of a to be less than 5.2%. The results indicate that a good
double-banked electrode separation channel, and com- quality of separation detection will be provided by this
pares the results to those achieved with the traditional particular arrayed-electrode configuration. Figure 14
straight separation channel. Figure 12a compares the presents the relationship between the applied voltage
two band shapes at the detection area. It is observed required for driving and separation purposes and the
that there is good correlation between the two results, active electrode interval. The figure considers the case of
and in fact the shape error value of the sample in the a double-banked electrode separation channel of 40 mm
double-banked separation channel is very low, i.e., Ê =
6.5%. Figure 12b compares the peak intensity of a sam-
ple band in the double-banked separation channel with
the results observed for the straight separation channel.
Compared to the previous case of the single-sided
electrode channel, it is noted that there is now a distinct
peak intensity within the double-banked electrodes sepa-
ration channel, and that it corresponds very closely to the
results obtained in the benchmark straight channel case.
The results presented in Figs. 12a and b suggest that the
use of a double-banked electrode separation channel will
yield a sample of satisfactory shape and intensity for
detection purposes.

Figure 13. Comparison of electropherograms generated

from an image collected in a traditional straight separa-
tion channel by experimental (solid line) and intensity of
concentration fragments in a double-banked electrode
separation channel with numerical simulation (dashed
line) at the detection area.

Figure 12. Comparison of (a) band shape and (b) peak

intensity at the detection area between a double-banked Figure 14. Relationship between voltage ratio and active
electrode separation channel and a traditional straight electrode span for a double-banked electrode separation
separation channel. channel.
1260 L.-M. Fu and R.-J. Yang
length with 21 electrode pairs separated by a pitch of
2 mm. The results show a direct relationship between
the electrode span and the required driving voltage. The
voltage can be as low as 10% of the original value if the
interval of the span S is long enough for the sample plug
to be separated.
4 Concluding remarks
This paper has presented an investigation into the gen-
eration of traveling electric fields within CE chips. This
study provides valuable information in designing elec-
trode layouts which maintain a sufficient electric field
strength to carry out separation, while requiring driving
voltages which are orders of magnitude lower than those
used in conventional straight separation channel CE
chips. The simulated results have shown that the use of
a single-sided electrode separation channel generates
the nonuniform electric field distribution and asymmetric
velocity profile within the channel. This results in tilting
and distortion of the sample band as it is driven along
the channel, which results in poor detection of the DNA
separation. The use of a double-banked electrode chan-
nel greatly improves the uniformity of the electric field and
the symmetry of the velocity profile. The resulting sample
band distribution is very similar to that achieved in the
traditional straight separation channel, and is ideal for
detection purposes. Furthermore, this electrode config-
uration allows the original driving voltage to be reduced
by 85%.
The authors thank Prof. Che-Hsin Lin of Southern Taiwan
University of Technology for comments on the experi-
ment. This project is partially supported by the Taiwan
National Research Council under NSC-90-2212-E006-
109.
Received September 11, 2002
5 References
[1] Behnke, B., Metzger, J. W., Electrophoresis 1999, 20, 80–83.
[2] Xin, B., Lee, M. L., Electrophoresis 1999, 20, 67–73.
[3] Mistry, K., Krull, I., Grinberg, N., J. Sep. Sci. 2002, 25, 935–
958.
Electrophoresis 2003, 24, 1253–1260
[4] Russel, W. B., Saville, D. A., Schowalter, W. R., Colloidal Dis-
persion, Cambridge University Press, New York 1989.
[5] Culbertson, C., Ramsey, R., Ramsey, J. M., Anal. Chem.
2000, 72, 2285–2291.
[6] Burggraf, N., Manz, A., Verpoorte, E., Effenhauser, C. S.,
Sens. Actuators B 1994, 80, 103–110.
[7] Hymer, W., Kirshnan, K., Kraemer, W., Welsch, J., Lanham,
W., Electrophoresis 2000, 21, 311–317.
[8] Junkers, J., Kopplin, P. S., Munch, J., Kettrup, A., Electro-
phoresis 2002, 23, 2872–2879.
[9] Rizzi, C., Rossini, K., Bruson, A., Sandri, M., Peruffo, A. D. B.,
Carraro, U., Electrophoresis 2002, 23, 3266–3269.
[10] Compton, M. M., Lapp, S. A., Pedemonte, R., Electrophore-
sis 2002, 23, 3262–3265.
[11] Yang, R.-J., Fu, L.-M., Lin, Y.-C., J. Colloid Interface Sci.
2001, 239, 98–105.
[12] Yang, R.-J., Fu, L.-M., Hwang, C.-C., J. Colloid Interface Sci.
2001, 244, 173–179.
[13] Jacobson, S. C., Hergenroder, R., Moore, A. W., Ramsey,
J. M., Anal. Chem. 1994, 66, 4127–4132.
[14] Jacobson, S. C., Koutny, L. B., Hergenroder, R., Moore,
A. W., Ramsey, J. M., Anal. Chem. 1994, 66, 3472–3476.
[15] Xu, C., Wang, Y., Cao, M., Lu, Z., Electrophoresis 1999, 20,
1829–1831.
[16] Schnelle, T., Müller, T., Gradl, G., Shirley, S. G., Fuhr, G.,
Electrophoresis 2000, 21, 66–73.
[17] Cui, L., Holmes, D., Morgan, H., Electrophoresis 2001, 22,
3893–3901.
[18] Ying, F., Mastrangelo, C. H., Burkes, D. T., Burns, M. A.,
Electrophoresis 1998, 19, 1388–1393.
[19] Schasfoort, R. B. M., Schlautmann, S., Hendrikse, J., van
den Berg, A., Science 1999, 286, 942–945.
[20] Polson, N. A., Hayes, M. A., Anal. Chem. 2000, 72, 1088–
1092.
[21] McKinght, T. E., Culbertson, C. T., Jacobson, S. C., Ramsey,
J. M., Anal. Chem. 2001, 73, 4045–4049.
[22] Hartley, N. K., Hayes, M. A., Anal. Chem. 2002, 74, 1249–
1255.
[23] Heller, C., Electrophoresis 2001, 22, 629–643.
[24] Grabtree, H. J., Cheong, C. S., Tilroe, D. A., Backhouse,
C. J., Anal. Chem. 2001, 73, 4079–4086.
[25] Patankar, N. A., Hu, H. H., Anal. Chem. 1998, 70, 1870–
1881.
[26] Arulanandam, S., Li, D., Colloids Surf. A 2000, 161, 89–102.
[27] Griffiths, S., Nilson, R. H., Anal. Chem. 2001, 73, 272–278.
[28] Dutta, P., Beskok, A., Anal. Chem. 2001, 73, 1979–1986.
[29] Fu, L.-M., Yang, R.-J., Lee, G.-B., Electrophoresis 2002, 23,
602–612.
[30] Yang, R.-J., Fu, L.-M., Lee, G.-B., J. Sep. Sci. 2002, 25, 996–
1010.
[31] Fu, L.-M., Yang, R.-J., Lee, G.-B., Liu, H.-H., Anal. Chem.
2002, 74, 5084–5091.
[32] Lin, C. H., Lee, G. B., Lin, Y. H., Chang, G. L., J. Micromech.
Microeng. 2001, 11, 726–732.