Transverse Electrodes for Improved

DNA Hybridization in Microchannels

Siddhartha Das and Suman Chakraborty
Dept. of Mechanical Engineering, Indian Institute of Technology, Kharagpur Kharagpur 721302, India

DOI 10.1002/aic.11144
Published online March 19, 2007 in Wiley InterScience (www.interscience.wiley.com).

The present study examines the modality, in which localized transverse electric fields
can be successfully employed, to augment the rate of DNA hybridization at the capturing
probes that are located further downstream relative to the inlet section of a rectangular
microchannel. This is in accordance with an enhanced strength of convective transport
that can be achieved, on account of increments in the wall zeta potential at the trans-
verse electrode locations. In the present model, the overall convective transport, which
is an implicit function of the magnitude and the location of the transverse electrical field
being employed, is essentially coupled with the surface kinetics of the bare silica wall
and also the kinetics that are involved in the dual mechanisms of DNA hybridization.
Parameters that govern the overall transport phenomena, such as the pH of the inlet
buffer, the length of the transverse electrodes, and the voltages at which these electrodes
are maintained are critically examined, in an effort to obtain an optimized wall pH dis-
tribution, which in turn can ensure favorable DNA hybridization rates at the capturing
probe locations. Practical constraints associated with the upper limits of the strength of
the transverse electrical fields that can be employed are also critically analyzed, so as
to ensure that an optimized rate of DNA hybridization can be achieved from the bio-
microfluidic arrangement, without incurring any adverse effects associated with the
overheating of the DNA molecules leading to their thermal denaturation. Ó 2007 Ameri-
can Institute of Chemical Engineers AIChE J, 53: 1086–1099, 2007
Keywords: electro-osmosis, microchannel, DNA hybridization, surface kinetics

Introduction special mention. Erickson et al.4 proposed a comprehensive

Research investigations on DNA hybridization in presence model of dynamic solid-phase oligonucleotide hybridization
of microfluidic transport have attracted serious attention over kinetics, based on an approach that accounted for both the
recent times, primarily motivated by an ever-increasing direct hybridization from the bulk phase and hybridization
necessity of optimizing the hybridization performance, in after an initial nonspecific adsorption stage. Das et al.,5 in a
tune with the stringent biotechnological demands. Starting more recent study, have extended the earlier models to incor-
from the pioneering efforts of Chan et al.,1 several research- porate the effects of electro-osmotic transport on the DNA
ers have come up with reaction-diffusion models of DNA hybridization mechanisms. Utilization of electrokinetic body
hybridization in microfluidic channels. In this respect, the re- forces in such bio-microfluidic applications has primarily
ceptor–ligand model of Axelrod and Wang2 and the dual been motivated by the possibility of building efficient flow
mechanism based hybridization model of Zeng et al.3 need control devices, without involving any moving components.
Irrespective of the details of the flow actuation mechanisms,
it also needs to be mentioned in this context that in principle,
Correspondence concerning this article should be addressed to S. Chakraborty the DNA hybridization rates could be significantly enhanced
at suman@mech.iitkgp.ernet.in.
by arranging for more efficient mixing mechanisms in the
Ó 2007 American Institute of Chemical Engineers associated fluidic systems. This is because of the fact that a

1086 May 2007 Vol. 53, No. 5 AIChE Journal

better mixing ensures faster reaction kinetics in reactive model, simulation experiments are undertaken, so as to
microfluidic systems. Micromixing augmentation mecha- depict the influence of the significant system parameters,
nisms, such as the use of pulsatile flows,6–8 serpentine such as the pH of the buffer solution at the inlet, voltages at
channels,9,10 creation of slanted wells at the floor of the the transverse electrodes, and the electrode dimensions, on
channel,11 presence of surface inhomogenities,12 surface pat- the resultant DNA hybridization rates. An upper limit of the
terning,13 inhomogeneous zeta potential14 etc, can therefore extent to which the transverse electric fields can be exploited
be potentially explored to achieve faster DNA hybridization to augment the rate of DNA hybridization is also established,
kinetics in microchannels. Such arrangements, in effect, can so as to constrain the associated Joule heating effects within
overcome the technological constraints associated with the permissible standards, to avoid a further thermal denaturation
stringent limitations in employing large pressure gradients of the double stranded (hybridized) DNA macromolecules.
and/or electric fields under practical operating conditions, To establish the afore-mentioned basic principles from a fun-
leading to an optimized DNA hybridization performance, damental mathematical perspective, a relatively simple con-
consistent with the chosen bio-microfluidic configuration. figuration is chosen, which assumes only a pair of capturing
It is well known that the axial variations in the wall zeta DNA probes, positioned symmetrically on both sides of a
potential, established on account of local pH gradients, can central electrode located on the bottom wall of a parallel
be utilized to modulate electro-osmotic transport in micro- plate microchannel, with another electrode of the same length
channels to a significant extent in many of the biomicroflui- being located on the upper wall, at the same axial location.
dic applications.15,16 However this effect is yet to be well For more involved designs, this arrangement can be consid-
exploited to obtain favorable DNA hybridization patterns. ered as a repeating monoblock, the number of which depends
Fundamentally, localized variations in the wall zeta potential on the desired complexity of the system chosen, in accord-
can alter the number of hydrogen ions that can be adsorbed ance with the DNA hybridization rates sought. Underlying
at the channel walls, thereby altering the electrical potential principles of this simple model, therefore, can be conven-
distribution in the vicinity. Utilizing this basic principle, sev- iently extended to obtain favorable hybridization patterns in
eral techniques, including the popular method of isoelectric more complicated situations as well, in which there are pro-
focusing,17 have been successfully employed by the research visions to employ larger numbers of DNA probes and match-
community, for the purpose of separation of analytes on the ing transverse electrode pairs in the microfluidic system.
basis of their respective isoelectric points, as a consequence
of the development of transverse potential gradients estab-
lished across a pair of surface electrodes.18 In a pioneering Mathematical Modeling
study, Molloy and Leighton demonstrated the effectiveness We consider electro-osmotic transport of an ionic solution
of employing transverse electric fields to perform binary os- of water through a parallel plate microchannel, made of bare
cillatory cross flow electrophoresis.19 Amongst the other im- silica, of length L0, height H, and width w, with w
H and
portant studies in related applications, the investigations of a subsequent hybridization of transported ss-DNA molecules
Chan and Chauhan on the employment of transverse electro- with their complementary pairs that are immobilized at the
des for performing cyclic EFFF20 and the use of EFFF for two different probe locations along the bottom wall of
size-based separation of DNA molecules21 also worth special the channel. A potential gradient is applied along the axis of
mention. However, in case of DNA transport, systematic the- the channel, which provides the necessary driving force for
oretical guidelines are yet to be put forward, so as to exploit electrokinetic transport. The magnitude of this potential gra-
a deliberate generation of pH gradients through a designed dient is ascertained by assigning a specified potential, f0,
configuration of transverse electrode pairs, in an effort to (given in Table 1) at the channel inlet and a zero potential at
augment the rate of hybridization in electrokinetically driven the channel exit. Typically, f0 is so chosen that the axial
microfluidic systems. Such attempts, in reality, can bear sig- electric field becomes of the order of 105 V/m, as a common
nificant consequences, leading towards a maximization of technological practice. In addition to this axial field, a local
DNA hybridization rates, within the constraints of a maxi- transverse electric field is applied by maintaining two trans-
mum permissible potential gradient that can be imposed on verse electrodes at different voltages, the transverse field
the microfluidic system. being confined over a limited portion of the channel length
The aim of the present work is to theoretically demonstrate only. The magnitude of this transverse field is varied in the
how a strong local pH gradient, obtained by employing a range of 105–106 V/m, by altering the cross-electrode poten-
pair of transverse electrodes, can be effectively utilized to tials. The effects of variations of the transverse potential gra-
achieve a faster rate of DNA hybridization in a microchan- dients on the DNA hybridization are illustrated later in this
nel. The conceptual system design is exercised by avoiding article. We assume that in the hybridization channel no bub-
any direct interference between the positioning of transverse bles are being generated due to the electrochemical reaction
electrodes and surface DNA probes on the microchannel on these transverse electrodes We consider the two DNA
walls, so as to provide a technological feasibility to the entire hybridization probes to be located at equal distances on ei-
arrangement. In other words, the DNA probes are considered ther side of the transverse electrode. The entire arrangement
to be located sufficiently downstream or upstream to the is schematically represented in Figure 1. In effect, the DNA
transverse electrodes. The model, in essence, couples the sur- molecules are transported by the combined actions of the
face kinetics, as modified by the presence of the transverse electro-osmotic and the electrophoretic migration mecha-
electrodes, with the bulk transport of momentum, heat, ionic nisms in the axial as well as in the transverse directions.
species, and DNA macromolecules, from the fundamental Because of a negative charge on the DNA molecules, the
physico-chemical principles. Based on the mathematical electrophoretic influences on those effectively oppose their

AIChE Journal May 2007 Vol. 53, No. 5 Published on behalf of the AIChE DOI 10.1002/aic 1087
 Table 1. Summary of Boundary Conditions
Governing Equation Inlet (x ¼ 0) Outlet (x ¼ L0) Bottom Wall (y ¼ 0) Top Wall (y ¼ H)
c¼0 qc c ¼ z (zeta Potential) c ¼ z (zeta Potential)
Poisson-Boltzmann eq. (Eq. 5)
qx ¼0

u ¼ uin qu u¼0 u¼0

Fluid flow eqs. (eq. 1–3) ¼0
qx
n¼0 n¼0 n¼0

qT
Energy conservation eq. (Eq. 13) T ¼ T?
qx ¼0 k qT
qy ¼ q
00
k qT
qy ¼ q
00



qcHþ cHþ ¼ cfilm exp 
ec
cHþ ¼ cfilm 
ec
cHþ ¼ ðcHþ Þinter ¼0 Hþ H þ exp
Conservation equation for qx

kB T


kB T

cMnþ ¼ ðcMnþ Þinter qcMnþ nec nec
concentration of the cations (Eq. 8) ¼0 cMnþ ¼ cfilm exp  cMnþ ¼ cfilm
Mnþ exp 
qx Mnþ
kB T kB T



qcOH cOH ¼ cfilm
ec
cOH ¼ cfilm
ec
cOH ¼ ðcOH Þinlet ¼0 OH exp OH exp
Conservation eq. for concentration qx

kB T


kB T

cXn ¼ ðcXn Þinlet qcXn nec nec
of anions (Eq. 8) ¼0 cXn ¼ cXn exp
film
cXn ¼ cXn exp
film
qx kB T kB T

qcDNA qcDNA qcDNA

Conservation eq. for DNA cDNA ¼ ðcDNA Þinlet qx ¼0 qy ¼0 qy ¼0
concentration (Eq. 8)

electro-osmotic migration. However, the electro-osmotic mo- X-Momentum Equation.

bility of the DNA molecules being greater than their electro- 8 9
q ~ ¼  qp þ r  ðmruÞ  re > qf qc
phoretic mobility, this retarding effect gets effectively over- ðruÞ þ r  ðrVuÞ : þ > ;
weighed, and the DNA molecules move in the direction qt qx qx qx
of electro-osmotic transport. This aspect is further elucidated (2)
through the numerical simulation predictions, as discussed where re is the net electric charge density, f is the potential
later. distribution due to an externally imposed electric field and c
is the surface potential distribution.
Convection–diffusion transport equations Y-Momentum Equation.
The governing transport equations, under these circumstan- 8 9
q qp > qf qc>
ces, can be described as follows. ðrnÞ þ r  ðrVnÞ ¼  þ r  ðmrnÞ  re : þ >
~ > ;
qt qy qy qy
Continuity Equation.
(3)
qr ~ ¼0 In both Eqs. 2 and 3, the dynamic viscosity, m, is a func-
þ r  ðrVÞ (1)
qt tion of temperature, given as22

Figure 1. The flow configuration.

1088 DOI 10.1002/aic Published on behalf of the AIChE May 2007 Vol. 53, No. 5 AIChE Journal


6 1713 amounts when compared with Mnþ or Xn, and hence, the
m ¼ 2:761  10 exp (4) effects of their presence are neglected in the present study.
T
We also consider that only a single type of ss-DNA mole-
where T is in K and m is in Pa s. cules is being transported. While considering transport of the
Distribution of the net electric charge density, re, appear- ss-DNA species, the diffusion coefficient Di, appearing in
ing in the momentum conservation equations, can be ascer- Eq. 8 represents a generalized diffusion coefficient, which is
tained by solving the Poisson–Boltzmann equation for sur- the liquid phase diffusion coefficient (D3) in the bulk fluid
face potential distribution as and surface phase diffusion coefficient (D2) at the nonspecific
adsorption sites (It can be noted here that DNA hybridization
re
r  ðercÞ ¼  (5) can take place directly from the bulk phase (3D hybridiza-
e0 tion) or along the wall surface (2D hybridization) after an
initial nonspecific adsorption step.). By borrowing analogy
where c denotes the electric double layer (EDL) potential, e0
from diffusion transport of polymeric chains, the diffusion
is the permittivity of free space, and e is the dielectric con-
coefficients for DNA molecules can be expressed as the
stant of the electrolyte, which is a function of temperature,
functions of the size and concentration of the ss-DNA mole-
given as22
cules, along with the ionic strength of the solution. Typically,
8 9
T >
e ¼ 305:7 exp>
one can employ certain well-established scaling relationship
: ; (6)
219 for this purpose, for example24

where T is in K. In Eq. 5, re, is described as Di  

¼ exp acv Rd (9)
Di;0
X
N
re ¼ 103 Na e zi ci (7) where Di is the effective diffusion coefficient of a DNA mac-
i¼1 romolecule of radius R and concentration c, and Di,0 is the
corresponding diffusion coefficient for c ¼ 0. Experiments
Here zi and ci are the valency and the concentration of the on a wide range of polymer systems, both biological and
ith ionic species, respectively, Na is the Avogadro number synthetic,25–27 have shown that such scaling laws can
and N is the total number of ionic species in the solution. adequately describe the diffusion of a polymer over a wide
The concentration of the ith ionic species is governed by the range of solution concentrations. From a series of investiga-
species-conservation equation, given by tions, it has been revealed that the parameter v, as appearing
in Eq. 9, typically falls in the range of 0.5–1.0, with the
qðrci Þ ~ i Þ ¼ r  ðrDi rci Þ þ r  ðrmep;i ci rfÞ þ rRi lower and the upper limits signifying the relative contribu-
þ r  ðrVc
qt tions of hydrodynamic and direct interactions, respectively.
(8) The parameter d varies from 0 to 1, depending upon the
polymer system. The parameter a is a function of the ionic
In Eq. 8, Ri is the term responsible for any reaction that strength of the solution given as a ; Ib, where I is the ionic
the ionic species may undergo in the solution. The term mep,i strength and b is a function of the nature of the polymer so-
represents the electrophoretic mobility of the ith species, lution. In the present study, we estimate the earlier parame-
which depends on the nature of species being transported. ters by considering a total of 10 nucleotide units of ss-DNA
For the ss-DNA molecules, mep,i is given by23 molecules with a hydrodynamic radius of around 1 nm, by
referring to the scaling estimates developed by Dwyer and
mep;i ¼ ½f ðkai Þ=ð1 þ kai Þezi oH;i ; (8a) Bloomfield.24 Diffusivities of the other ionic species, how-
ever, can be estimated in a more straightforward manner as
In Eq. 8a, k is the inverse of the Debye length, zi is the
valence of the ith type of the ss-DNA molecule (which can Di ¼ oH;i kB T (9a)
be obtained through Eq. 11, as described subsequently), ai is
its Stokes radius, oH,i is its hydrodynamic mobility (which It also needs to be noted here that an accurate solution of
can be expressed as oH;i ¼ 6pma1
i
and f ðkai Þ is the Henry’s f- the species conservation equation (Eq. 8) not only requires a
function. However, for the other ionic species, the electro- physically-consistent specification of the diffusion coeffi-
phoretic mobility can be more simply expressed as cients, but also necessitates a correct specification of the
electro-osmotic mobilities of the ionic species, which in turn
mep;i ¼ ezi oH;i (8b) depend on the respective charges of the ionic species being
transported. For that purpose, it can be noted that unlike the
In the present model, we consider simple ionic species in simple ionic species, charges of the ss-DNA molecules do
the solution, with the following constituents: Hþ, OH, Mnþ, not remain constant for all values of the buffer pH. A quanti-
Xn, where Xn is the anion corresponding to the added me- tative description of the electrostatic characteristics of a ss-
tallic cation, Mnþ. The other kind of ionic species that might DNA molecule, as a function of the buffer pH, can be
be expected to be present in the solution will include traces derived from the ionization of the phosphodiester linkages
of some phosphates or sulfates, which are routinely added to (resulting in one negative charge per nucleotide) and also
alter certain properties of the solution, such as its conductiv- from ionization of bases (making the charge on the DNA
ity. However, these components are present in much smaller base-dependent) in an aqueous solution. A phosphodiester is

AIChE Journal May 2007 Vol. 53, No. 5 Published on behalf of the AIChE DOI 10.1002/aic 1089
a strong acid whose pK is around 1. Nucleotides which make In Eq. 12a, Kw is the dissociation constant for water. In
up the DNA molecules can be of four types differing only in the earlier formulation, we assume that in totality there are
the nitrogenous base: Adenine, Thymine, Cytosine and Gua- MP types of different ss-DNA molecules in the solution, and
nine. Adenine (A), and Cytosine (C) bases can be found as (N-2-MP) types of ionic species, other than the Hþ and OH
neutral or positive forms, whereas thymine (T) is neutral or ions. For the present model, MP ¼ 1 and N ¼ 5.
negative, and guanine (G) can exist in all the three states.28
The ionization of these bases can be represented as Thermal energy conservation equation
Since various physico-chemical characteristics pertinent to
KA
Adenine: ½AHþ  !½A þ ½Hþ  KA ¼ 103:5 (10a) the present model are very much temperature sensitive, a so-
lution of the energy equation also turns out to be necessary.
KC The earlier equation can be cast in the following general
Cytosine: ½CHþ  !½C þ ½Hþ  KC ¼ 104:2 (10b) form:

KT q ~ ¼ r  ðkrTÞ þ j þ q
Thymine: ½TH !½T  þ ½Hþ  KT ¼ 109:2 (10c) ðrCP TÞ þ r  ðrCP VTÞ (13)
qt
K G1
Guanine: ½GHþ þ
2  !½GH þ ½H  KG1 ¼ 102:1 (10d) where k is the thermal conductivity of the electrolyte solu-
tion, which is a function of temperature, given as22
K G2
½GH !½G  þ ½Hþ  KG2 ¼ 109:2 (10e) k ¼ 0:6 þ 2:5  105 T (14)

Now, assuming the probability of presence of each base where T is in K and k is in W/m2 K. In Eq. 13, j is the vol-
type (A, T, C, G) to be 25% per nucleotide, the average net umetric heat generation due to viscous dissipation, given as
charge per nucleotide, C/Nnu, may be estimated as28   

qu 2  qn 2 qu qn 2
j ¼ 2m þ þm þ (15)
1 1 qx qy qy qy
C=Nnu  þ 0:25
1 þ ½Hþ =Kph 1 þ KA =½Hþ 
Further, q̇ is the volumetric heat generation due to Joule
1 1
þ 0:25 þ  0:25
effects, which, according to Ohm’s law, can be given as
1 þ KC =½H  1 þ ½Hþ =KT
½Hþ 2  KG1 KG2  I2
þ 0:25 ð11Þ q¼ (16)
KG1 KG2  KG1 ½Hþ  þ ½Hþ 2 s

where I is the electrical current density and s is the electrical

where Kph (¼101) corresponds to the ionization of the phos- conductivity. It can be noted here that the electrical current
phodiester groups. The valency term zi appearing in the density includes two specific contributions, one due to the
species conservation equation for the ss-DNA molecule is applied electric field imposed on the conducting solution
identical to the average net charge per nucleotide C/Nnu and ~ and the other due to the net charge density moving
(sE),
is thus given by Eq. (11). !
with the fluid flow ðre V Þ. Therefore, the electrical current
It can be noted here that an accurate modeling of the sec- !
density, I , can expressed as
ond term in the right hand side of Eq. 8 not only demands a
consistent description of the electro-osmotic mobilities of the ~ ~ þ sE
~
I ¼ re V (17)
constituents, but also requires a specification of the distribu-
tion of the potential field, f, which can be obtained by con-
Using Eqs. 16 and 17, one can obtain an expression for q̇
sidering the charge conservation in the liquid, in accordance
as
with the requirement of r I ¼ 0. This finally leads to a dif-
ferential equation governing the distribution of f as
 ~ þ sEÞ
ðre V ~  ðre V
~ þ sEÞ
~
q¼ (18)
s
r  ðsrfÞ ¼ r ~
iD (12)

where iD is the diffusion current density and s is the electri- Coupling of transport modeling with DNA
cal conductivity, given as23 hybridization kinetics
"M The model of species (DNA) transport adopted here is
X p
f ðkai Þ 2 X
N2 based on two basic mechanisms of hybridization, namely,
s ¼ 10 Na e
3 2
zi oH;i ci þ z2i oH;i ci direct (specific) hybridization from the bulk phase to the sur-
i¼1
1 þ kai i¼MP þ1
# face-bound probes, and by indirect (nonspecific) hybridiza-
Kw tion, in which the target is initially nonspecifically adsorbed
þ oH;Hþ cHþ þ oH;OH ð12aÞ on the surface and then diffuses along the surface before
c Hþ
reaching an available target probe molecule. Details of these

1090 DOI 10.1002/aic Published on behalf of the AIChE May 2007 Vol. 53, No. 5 AIChE Journal
mechanisms are outlined in Erickson et al.,4 and are not Dependence of zeta potential on the surface kinetics
repeated here for the sake of brevity. However, a summary At the surface of the bare silica walls, the following reac-
of the pertinent formulation is presented here, for the sake of tions take place.
completeness. Effectively, we implement the hybridization (1)
boundary conditions at the bottom wall by invoking the K1
generic source term, Ri, appearing in Eq. 8 such that, for the SiOH ! SiO þ Hþ ; (22)
bulk phase, Ri ¼ 0, and for the reactive bottom wall (where
the capturing probes are attached) where K1 is the equilibrium constant for the earlier reaction,
given as28


qc2;s qc2;ns ½SiO ½Hþ 
Ri ¼  þ (19) K1 ¼ ¼ 106
qt q ½SiOH
(22a)

where, c2;s and c2;ns are the surface-phase concentration of (2)

specifically and nonspecifically adsorbed target molecules, K
respectively. Along the nonreacting surfaces, however, Ri ¼ SiOHþ
2 ! SiO þ 2Hþ (23)
0. The terms appearing in Eq. 19 can further be expressed as
a set of coupled two-dimensional kinetic equations as where K is the equilibrium constant for the earlier reaction,
given as28
qc2;s
¼ ½k31 c3;m ðc2;s;max  c2;s Þ  k31 c2;s  ½SiO ½Hþ 2
qt K¼ ¼ 102pzc (23a)
½SiOHþ 
þ ½k21 c2;ns ðc2;s;max  c2;s Þ  k21 c2;s  ð20Þ 2

Here, ‘pzc’ or the point of zero charge, which is the pH at

and which the surface density of the positive charges is equal to
the surface density of the negative charges.29 A standard
qc2;ns practice for measuring the pzc for a given surface is to calcu-
¼ ½ka c3;m ðc2;ns;max  c2;ns Þ  kd c2;ns 
qt late the pH of a water drop at which the drop makes the
 ½k21 c2;ns ðc2;s;max  c2;s Þ  k21 c2;s  ð21Þ largest wetting angle with the concerned surface.28 A typical
experimentally measured value29,30 for pzc for a bare silica
where, c2;s;max is the maximum concentration of the hybri- surface is 3, which gives K ¼ 106.
dized targets (equivalent to the local concentration of the sur- (3)
K2
face bound probes available for hybridization), c2;ns;max is the SiOHþ 2 ! SiOH þ Hþ (24)
maximum concentration of the nonspecifically adsorbed mol-
ecules, c3;m is the bulk-phase concentration of the targets in where K2 is the equilibrium constant for the earlier reaction,
surface film, k31 is the kinetic association constant for direct given as28
hybridization (from solution phase), k31 is the kinetic disso-
ciation constant for direct hybridization (from solution ½SiOH½Hþ  K
phase), k21 is the kinetic association constant for indirect K2 ¼ ¼ ¼1 (24a)
½SiOHþ
2  K1
hybridization of the nonspecifically adsorbed targets (from
surface phase), k21 is the kinetic dissociation constant for (4) Like the other ionic species, the metallic ions present
indirect hybridization of the nonspecifically adsorbed targets in the solution also give rise to a certain chemical reaction,
(from surface phase), ka is the kinetic association constant as31
for nonspecific adsorption of the targets to the surface, and
kd is the kinetic dissociation constant for nonspecific adsorp- KM
tion of the targets to the surface. It is to be noted here that Mnþ þ SiO ! Mnþ SiO (25)
the values of k21 and k31 may differ from one another, since
they denote dissociation constants from two different phases, where KM is the equilibrium constant for the earlier reaction,
i.e., the surface and the solution phases, respectively. Physi- given as
cally, Eq. 20 describes that the rate of change in surface con-
centration of hybridized species is a combination of the rate ½Mnþ SiO 
KM ¼ (25a)
of change of targets getting hybridized directly from the bulk ½Mnþ ½SiO 
phase and the rate of targets getting hybridized after an ini-
tial nonspecific adsorption. Equation 21 implies that the rate The value of KM depends on the specific nature of the me-
of change in surface concentration of nonspecifically tallic ion. Higher the value of KM, greater is the protonation
adsorbed targets is increased by the rate of adsorption from of free silanol groups by these cations, and larger is the low-
the bulk phase, but is decreased by the rate at which the non- ering of the zeta potential and the consequent electro-osmotic
specifically adsorbed targets become hybridized. For detailed force (EOF). It needs to be noted here that although the me-
calculation of the kinetic constants appearing in Eqs. 20 and tallic cations protonate the silanol group and thereby reduce
21, Erickson et al.4 can be referred to. the wall zeta potential, a presence of excess of such ions in
The boundary conditions consistent with various conserva- the bulk provides the necessary driving force for the electro-
tion equations described as earlier are summarized in Table 1. kinetic flow to take place. Accordingly, the presence of such

AIChE Journal May 2007 Vol. 53, No. 5 Published on behalf of the AIChE DOI 10.1002/aic 1091
metallic ions is virtually indispensable for an efficient con- the transverse electrodes also induces a ‘field effect,’ which
vective transport to take place in the buffer solution. modifies the local wall zeta potential. For a quantitative
For an effective estimation of the concentration of the assessment of the same, the transverse electrodes can be
SiO group appearing in the afore-mentioned reactions, we assumed to constitute a parallel plate capacitor, which is
first note that the surface site density, g, can be given as completely filled with a dielectric material. Therefore, con-
sidering the electric field between the parallel plates, a Gaus-
g ¼ NA ð½SiO  þ ½SiOH þ ½SiOHþ nþ 
2  þ ½M SiO Þ (26) sian surface can be imagined around the interface of the
dielectric and each plate. Consequently, when Gauss’ law is
where NA is the Avogadro number. A typical value of g for applied, the total charge enclosed by the Gaussian surface
glass walls is around 4.5 sites/nm2.28 Combining Eqs. 22a, includes both the free and the induced charges, which even-
23a, and 24a, along with Eq. 26, we get tually affects the surface zeta potential.33 This effect is
known as the ‘field effect,’ and the resultant zeta potential is
g given by33
½SiO  ¼   (27)
½Hþ  ½Hþ 2
NA 1 þ K1 þ K1 K2 þ KM ½Mnþ 
CT
zfield effect ¼ VT (34)
Again, the surface charge per unit area is given as CDiffuse

sSi ¼ NA eð½SiO   ½SiOHþ

2 Þ (28) where VT is the potential of the nearest transverse electrode
and CT is the total capacitance of the system. In Eq. 34,
(where e is the electronic charge), which can be expressed CDiffuse is the capacitance of the diffused EDL formed at the
(using Eqs. 22a, 23a, and 24a) as wall, which, for a unit channel width can be estimated as33
!
 ½Hþ 2 CDiffuse ¼ 7:23  103 cos h½ð19:46 V1 ÞzL0 (34a)
sSi ¼ NA e½SiO  1  (29)
K1 K2
with z the wall zeta potential, L0 is the channel length (in
Combining Eqs. 27 and 29, we get centimeters), and CDiffuse is given in the units of Faraday.
0 1 Assuming the capacitance of the diffuse layer, CDiffuse, and
þ 2
1  ½H  the capacitance of the silica channel walls, Cwall
sSi ¼ g e@ A (Cwall ¼ esilica
K K L0
twall , where esilica is the permittivity of silica and
1 2
þ þ 2
(30)
1 þ ½HK1  þ ½H 
K1 K2 þ KM ½M nþ
 twall is the thickness of the silica channel walls, width of the
channel being taken as unity) to be in series, we can obtain
The net charge density, stotal, which can be considered an expression governing the net capacitance, CT, as
approximately same as sSi, determines the effective zeta
potential as32 1 1 1
¼ þ (35)
 1=2 ! CT CDiffuse Cwall
2kB T 1 500p
z¼ sin h sSi
e ee0 RTð½Hþ Surface þ ½Mnþ Surface Þ Substituting the earlier expression for CT in Eq. 34, an
estimation of the local zfield effect can be obtained, which,
(31) physically is the zeta potential that is induced over and above
the nominal value of wall zeta potential, due to an ionization
The parameters ½Hþ Surface and ½Mnþ Surface , as appearing in
of the silica wall surface and the resultant charge distribution
Eq. 31, can be described through Boltzmann partitioning as31
within the EDL. Accordingly, the overall zeta potential at the

 location of transverse electrodes is given as
ec
½Hþ Surface ¼ ½Hþ Film exp  (32)
kB T
zoverall ¼ zwall þ zfield effect (36)


nec
½Mnþ Surface ¼ ½Mnþ Film exp  (33) Effectively, it can be inferred from Eq. 36 that an axial
kB T
gradient of the zeta potential (and a consequent variation of
the local pH field) is induced along the channel walls,
where ½Hþ Film and ½Mnþ Film are the bulk hydrogen ion and
because of a selective placement of the transverse electrodes
metallic ion concentrations, respectively, in a thin film adja-
on the same. Such a nonuniform zeta potential, in turn,
cent to the channel surface. Eqs. 32 and 33, in effect, suffice
ensures a nonuniform body force at different channel cross-
as the necessary boundary conditions at the channel walls for
sections, resulting in dissimilar flow patterns at different
the species conservation equations for the hydrogen and the
axial locations of the channel.34 In general, it is expected
metallic ions, respectively.
that an augmented rate of electro-osmotic transport, on
account of enhanced zeta potential values at the transverse
Effect of transverse electrodes on the electrode locations, may be responsible for an augmented
overall transport phenomena rate of DNA hybridization at the probe locations in further
Apart from creating a local body force along the y-direc- downstream sections, a direct quantification of which hap-
tion, the localized electrical potential gradient produced by pens to be the central focus of attention behind this study.

1092 DOI 10.1002/aic Published on behalf of the AIChE May 2007 Vol. 53, No. 5 AIChE Journal
 Table 2. Simulation Data vicinity of the transverse electrodes. At the channel sections
Parameter Value
where there are no transverse electrodes, the identical values
of the zeta potential at the two walls effectively ensure that
L0 1.0  102 m
LP 1.0  104 m
the velocity profile has a plug-like shape, which is typical to
LD 2.0  103 m any fully developed electro-osmotic flow. However, for the
H 50  106 m sections containing the transverse electrodes, the wall zeta
e0 8.854  1012 (C/VM) potential does not remain the same for the top and the bot-
e 1.6  1019 C tom walls (Figure 3), because of the fact that the top and
f0 5000 V
uinitial 2.0 m/s bottom electrodes are maintained at different voltages.
aDNA (Stokes radius for DNA) 1 nm Unequal voltages at the top and the bottom electrodes ensure
KW 1014 that the ‘field effect’ attenuates the zeta potential to different
T? 300 K extents at the two walls. The consequence of this becomes
r 998 kg/m3
kB 1.38  1023 J/K
evident in the variations of the ionic concentration distribu-
ZKþ þ1 tions at the top and the bottom channel walls (Figure 4). At
ZCl 1 the locations where there are no electrodes, the distribution
aKþ (radius of potassium ion) 1.33 Å of a particular type of ion (either anion or cation) remains
aCl (radius of chlorine ion) 1.67 Å identical at both the top and the bottom walls, leading to the
q@ 10 W/m2
(CDNA)inlet 1.0  106 M typical plug-like and symmetrically-shaped electro-osmotic
C3,m 1.9  107 M velocity profiles (Figure 5). However, at the channel sections
C2,s,max 2.0  107/mol/m2
C2,ns,max 1.98  107/mol/m2
k31 1  106 (1/Ms)
k31 0.49 (1/s)
k21 1  106 m/Ms
k21 0.51 (1/s)
ka 9  103 (1/Ms)
kd 0.3 (1/s)
D3,0 1.3  1010 m2/s
D2,0 5.0  1013 m2/s
twall 20  106 m
eSilica (Dielectric constant for silica) 4.3

Results and Discussions

To obtain a quantitative prediction from the present model,
we investigate the case of electro-osmotic flow of a KCl
ionic solution, with an inlet buffer pH of 4 and a uniform
inlet velocity profile (uin). The transverse electrodes are each
taken to be of length L0/5, and are maintained at þ15 V (top
electrode) and þ20 V (bottom electrode). The other relevant
problem data are listed in Table 2.
Figure 2a gives the variation of the wall pH (at the bottom
wall) along the channel length, for the afore-mentioned case.
From the plot it is evident that there is a significant deviation
of the wall pH when compared with the inlet buffer pH, over
that section of the channel length where the transverse elec-
trodes are placed. Such significant changes in the pH cannot
be induced by means of a mere convective transport. This
can be attributed to the short-ranged electrostatic forces of
attraction, by which the positively charged hydrogen ions
remain bound to the negatively charged silica ions of the Figure 2. (a) Variation of the bottom wall pH with the
wall, which effectively happen to be too strong to be dis- axial position for the case in which the inlet
rupted by the flow field alone. However, the presence of buffer pH is 4.0, lengths of the transverse
localized transverse electric fields, as in the present case, electrodes are L0/5, and the electrodes are
modifies the Hþ ion distribution within the Stern layer, maintained at þ15 V (top electrode) and þ20
thereby altering the wall zeta potential in the vicinity. This is V (bottom electrode); (b) variation of the bot-
clearly manifested by the variation of wall zeta potential tom wall zeta potential with the axial position
along the channel length, which is depicted in Figure 2b, for for the case in which the inlet buffer pH is
the bottom wall. This variation in the local zeta potential 4.0, lengths of the transverse electrodes are
bears a significant consequence in modifying the relative L0/5, and the electrodes are maintained at
ionic distribution at the top and the bottom walls of the chan- þ15 V (top electrode) and þ20 V (bottom
nel, which effectively govern the local velocity variations in electrode).

AIChE Journal May 2007 Vol. 53, No. 5 Published on behalf of the AIChE DOI 10.1002/aic 1093
Figure 3. Variation of the wall zeta potential (f) with the Figure 5. Comparison of the velocity profiles at differ-
axial position for the case in which the inlet ent axial locations for the case in which the
buffer pH is 4.0, lengths of the transverse inlet buffer pH is 4.0, lengths of the trans-
electrodes are L0/5, and the electrodes are verse electrodes are L0/5, and the electrodes
maintained at þ15 V (top electrode) and þ20 are maintained at þ15 V (top electrode) and
V (bottom electrode). þ20 V (bottom electrode).
Here the upper plate is the top wall and the lower plate the
bottom wall of the channel.
erential excess charge density gradient that is directed from
the top wall towards the bottom wall. The combination of
containing the transverse electrodes, the zeta potential at the this charge density gradient and the transverse electric field
bottom wall is observed to be more than that at the top wall, creates a transverse electro-osmotic body force, resulting in a
corresponding to the buffer-electrode arrangements employed transverse component of the electro-osmotic velocity being
in the present study. This leads to the development of a pref- directed towards the bottom electrodes.

Effects on DNA hybridization

The present section discusses the influence of various elec-
trical and flow parameters on the rate of DNA hybridization,
consistent with the specific configuration employed in this
study. The transverse velocity components being imparted by
the cross-electrodes, indeed, play a pivotal role in dictating
the rate of DNA hybridization in the microfluidic channel, as
governed by the important system parameters. For illustrating
this aspect, Figure 6 is plotted, which depicts the velocity
fields within the microchannel, for an inlet buffer pH of 4.0,
transverse electrode length of L0/5, and the voltages of the
top and the bottom electrodes as þ15 V and þ20 V, respec-
tively. From Figure 6, it is evident that at sections upstream
to the transverse electrodes, the velocity field is predomi-
nantly axial. On contrary, a significant transverse component
(downward) velocity is imparted because of the effect of the
presence of cross-electrodes in the further downstream sec-
Figure 4. Variation of the ionic distribution for the posi- tions. The net effect is an enhanced advective transport of
tive (Kþ) and negative ions (Cl) at the chan- the DNA molecules towards the bottom wall probe (Probe 2)
nel top and bottom walls with the axial posi- that is located immediately downstream to the section con-
tion, for the case in which the inlet buffer pH taining the transverse electrodes. This ensures that despite
is 4.0, lengths of the transverse electrodes the surface adsorption of the ssDNA molecules at the
are L0/5, and the electrodes are maintained upstream probe location (Probe 1), considerable rates of
at þ15 V (top electrode) and þ20 V (bottom migration of the ssDNA molecules from the buffer solution
electrode). to the downstream probe location (Probe 2) can effectively
Here the upper plate forms the top wall and the lower plate be maintained. Thus, with the present configuration, the rate
forms the bottom wall of the channel. of DNA hybridization at the location of the Probe 2 is sub-

1094 DOI 10.1002/aic Published on behalf of the AIChE May 2007 Vol. 53, No. 5 AIChE Journal
 (a). The pH of the inlet buffer
(b). The length of the transverse electrodes, and
(c). The voltages at which the top and the bottom electrodes
are maintained.
Effect of variation of the pH of the inlet buffer. To ex-
hibit the effect of variation of the inlet buffer pH on the rate
of DNA hybridization at the capturing probes, we plot
(Figure 8) the temporal variation of the concentration of the
hybridized targets at both the probes, with variable pH values
of the inlet buffer, all other conditions remaining unaltered.
In general, it is observed that higher the inlet pH value,
faster is the DNA hybridization rate, as apparent from Figure
8. This can be attributed to the fact that an increased value
of the inlet buffer pH necessarily implies an increase in the
wall pH, over and above the local pH augmentation effects
induced by the transverse electrodes. This, in turn, leads to
an enhancement in the effective zeta potential, thereby ensur-
ing an augmented rate of convective transport of ss-DNA
molecules, resulting in a improved DNA hybridization char-
Figure 6. Velocity vectors at a specified portion of the acteristics at both the capturing probes. However, the extent
flow domain that includes the transverse of enhancement in the rate of DNA hybridization, as
electrodes for the case in which the inlet achieved by using an inlet buffer of higher pH, is not identi-
buffer pH is 4.0, lengths of the transverse cal for the two probes. It is critical to recognize in this
electrodes are L0/5, and the electrodes are respect that the negative y-gradient of the transverse compo-
maintained at þ15 V (top electrode) and þ20 nents of the velocities near the cross-electrodes (due to the
V (bottom electrode). introduction of an additional vertically downward velocity
component) also result in a positive x-gradient of the axial
component of the flow velocity, so as to satisfy the continu-
stantially enhanced, when compared with the case without ity requirements (conservation of mass). As a consequence,
transverse electrodes, as depicted in Figure 7. an accelerated migration of the ssDNA molecules takes place
To further elucidate the effects of various system parame- towards the location of Probe 2. What follows from this
ters on the rate of DNA hybridization, as dictated by the argument is the fact that the enhancement effects in the rate
mechanisms explained earlier, the influences of the following of DNA hybridization, because of an augmented convective
aspects are analyzed in further details, in the subsequent dis- transport, are much more significant at the location of Probe
cussions: 2, when compared with Probe 1. This relative enhancement

Figure 8. Temporal variation of the concentration of

Figure 7. Variation of the film concentration with nor- the hybridized targets at the two probe loca-
malized axial coordinate for the case in tions, for different inlet values of the buffer
which the inlet buffer pH is 4.0, lengths of pH. Lengths of the transverse electrodes are
the transverse electrodes are L0/5, and the L0/5 and the electrodes are maintained at
electrodes are maintained at þ15 V (top elec- þ15 V (top electrode) and þ20 V (bottom
trode) and þ20 V (bottom electrode). electrode).

AIChE Journal May 2007 Vol. 53, No. 5 Published on behalf of the AIChE DOI 10.1002/aic 1095
 Effect of the Variation of the Voltages of the Transverse
Electrodes. To exhibit the effect of the variation of voltages
at the transverse electrodes on the rate of DNA hybridization,
we plot Figure 10, which depicts the temporal variation of
the concentration of the hybridized targets at the two probe
locations, for different values of the transverse electrode vol-
tages, all other conditions remaining unaltered. To demon-
strate these aspects, following three cases are considered: (a)
Case A: no transverse electrode, (b) Case B: þ15V and
þ20V at the top and the bottom electrodes, respectively, (c)
Case C: þ15V and þ65V at the top and the bottom electro-
des, respectively. In each of three cases, the buffer pH is
taken to be 4 and the transverse electrodes for Cases B and
C are taken to be of length L0/5. As expected, the hybridiza-
tion pattern at Probe 1 is insensitive to the variations
amongst the three different cases, since the Probe 1 is
located upstream to the transverse electrodes. However, at
Figure 9. Temporal variation of the concentration of the location of Probe 2, effects of voltages of the transverse
the hybridized targets at the two capturing electrodes are strongly felt. Higher the electrode potentials,
probe locations, for different values of the more significant is the corresponding enhancement in the
transverse electrodes lengths. wall zeta potential, and stronger is the augmentation in the
The electrodes are maintained at þ15 V (top electrode) rate of electrokinetic convective transport and the effective
and þ20 V (bottom electrode), and the inlet buffer pH is downward ‘push,’ leading to a richer supply of DNA mole-
taken to be 4. cules to the downstream probe location (i.e., Probe 2),
thereby ensuring improved hybridization rates. As a result,
for higher values of the transverse electrode voltages, a
is only possible because of the local increments in the mag- nearly uniform rate of DNA hybridization can effectively be
nitudes of the wall zeta potentials, on account of the trans- achieved at the different capturing probe locations, thereby
verse electrode effects. virtually nullifying the counteracting effects of a continuous
Effect of the Variation of the Length of the Transverse removal of the ss-DNA molecules from the bulk because of
Electrodes. To analyze the effects of variation of the length the specific and nonspecific hybridization effects at the fur-
of the transverse electrodes on the rate of DNA hybridization
at the probes, we plot Figure 9, which depicts the temporal
variation of the concentration of the hybridized targets for
different lengths of the transverse electrodes (i.e., different
values of LE), keeping all other conditions unaltered. By
altering the lengths of the transverse electrodes, the portion
of the channel length over which the zeta potential is modi-
fied due to the ‘field effect’ is effectively varied. From Fig-
ure 9, it is evident that the rate of DNA hybridization at
Probe 1, as indicated by the temporal variation of the con-
centration of the hybridized targets at that location, is insen-
sitive to the changes in lengths of the transverse electrodes,
because of the fact that the Probe 1 is located upstream rela-
tive to the transverse electrodes, and therefore carries no in-
formation of alteration of the wall zeta potential by virtue of
their presence. At the location of Probe 2, however, influence
of upstream transverse electrodes can be strongly felt. In
general, it is revealed that longer the transverse electrodes,
longer is the portion of the channel over which the ‘field Figure 10. Temporal variation of the concentration of
effect,’ causing an enhanced localized axial transport and an the hybridized targets at the two capturing
effective ‘push’ towards the bottom wall leading to destruc- probe locations, for the following cases,
tion of depletion layer, on account of enhanced zeta poten- Case A: no transverse electrodes; Case B:
tial, is felt. The result is the improvement in the rate of DNA the transverse electrodes are maintained at
hybridization at Probe 2. However, it needs to be remem- þ20 V (the bottom electrode) and þ15 V
bered at this point that the length of transverse electrodes (top electrode); and Case C: the transverse
cannot be indiscriminately increased so as to exploit a maxi- electrodes are maintained at þ65 V (the
mum advantage of the consequent augmentation in the rate bottom electrode) and þ15 V (top elec-
of convective transport, mainly because of the technological trode).
constraints associated with the placement of the DNA probes Lengths of the transverse electrodes are L0/5 and the inlet
devoid of any direct interference with the wall electrodes. buffer pH is taken to be 4.

1096 DOI 10.1002/aic Published on behalf of the AIChE May 2007 Vol. 53, No. 5 AIChE Journal
ther upstream sections. Such a uniform concentration distribu-
tion is often useful for the cases, in which one employs a sin-
gle nucleic acid hybridization system, to accomplish multiple
diagnostic purposes. However, it needs to be remembered here
that the transverse electrode voltages cannot be incremented at
will, in an effort to obtain the favorable effects mentioned as
earlier. This is because of the fact that the resultant electric
field needs to be constrained within stringent limits in practice,
so as to ensure that the associated Joule heating effects are not
strong enough to result in a denaturation of the thermally
labile DNA samples. We present a critical assessment of this
aspect in the subsequent subsection.
Other than the effects of these three significant operating
parameters mentioned as earlier, it is important to recognize
here that the channel height is also intuitively expected to
bear significant consequences on the rate of DNA hybridiza-
tion. However, it has already been demonstrated in the litera-
ture that the effect of the channel height on the rate of DNA
hybridization becomes important35 only if the transport is
predominantly pressure-driven, since in such a case the ve-
locity variations are strongly dependent on the transverse
coordinates. On the other hand, for a pure electro-osmotic
flow field, as considered in this study, the axial velocity com-
ponents remain virtually uniform over the channel cross sec-
tions, provided that the channel dimensions are large enough
(typically of the order of 1 mm or more) to ensure nonover-
lapped EDL fields, so that the channel height does not play a
critical role in influencing the rate of DNA hybridization
under these conditions. However, in case of nanochannels,
such effects can be of critical importance in determining the
net rate of macromolecular transport and hybridization. For
instance, in case of nanochannels, the net rate of macromo-
lecular transport may be significantly reduced because of a
hindered diffusive transport originated out of the confinement
effects. Narrower the channel, stronger becomes this effect.
However, since the present study is only confined to the elec-
tro-osmotically-driven DNA hybridization in microfluidic
channels, analysis of the height-dependent macromolecular
transport in nanochannels falls beyond the present scope, and
accordingly, is not discussed here.
Limitations in Employing Transverse Electric Fields for
Enhancing the Rate DNA Hybridization. ‘Melting’ or
‘denaturation’ of DNA is characterized by the separation of
two DNA strands from an existing hybridized state. This
splitting occurs at the melting temperature, Tm, defined as
the temperature at which 50% of the oligonucleotides and
their perfect complements are in duplex. To avoid problems
like inappropriate duplex formation, primer mismatch etc., Figure 11. Temporal variation of temperature along the
the hybridization is typically carried out 58  108C below bottom wall of the channel in the axial direc-
Tm. Hence it is important that the temperature rise due to tion, for (a) Case A, (b) Case B, and (c) Case C.
Joule heating during hybridization is not more than around Detailed particulars of these three cases are mentioned in
58C. This, in turn, imposes serious restraints to the upper the caption of Figure 7.
limit of the electric field that can be employed to actuate the
electrokinetic flows in the DNA hybridization assays. To
assess this issue critically, it is imperative to obtain the tem- DNA samples. So far as the Case B is concerned, it is
perature distributions along the channel bottom wall, for the revealed that the axial temperature profile along the bottom
Cases A, B, and C mentioned in the previous subsection, wall, at a given instant of time, exhibits the trend of an ab-
which are depicted in Figures 11a–c. For Case A (no trans- rupt rise near the locations of placement of the transverse
verse electrodes), the maximum temperature rise is found to electrodes. However, for this case too, the maximum temper-
be well within the permissible limits prescribed earlier, ature rise at any portion of the channel wall is safely within
thereby avoiding any thermal denaturation of the hybridized the permissible limits to avoid DNA melting. Therefore, so

AIChE Journal May 2007 Vol. 53, No. 5 Published on behalf of the AIChE DOI 10.1002/aic 1097
far as the DNA hybridization is concerned, the arrangement 2. Axelrod D, Wang MD. Reduction-of-dimensionality kinetics at reac-
for Case B (with transverse electrodes) turns out to be more tion-limited cell surface receptors. Biophys J. 1994;66:588–600.
3. Zeng J, Almadidy A, Watterson J, Krull UJ. Interfacial hybridization
favorable than the same employed for Case A (without trans- kinetics of oligonucleotides immobilized onto fused silica surfaces.
verse electrodes). However, for Case C, it is found that the Sens Actuators B. 2003;90:68–75.
localized temperature rise on account of the placement of 4. Erickson D, Li D, Krull UJ. Modelling of DNA hybridization
transverse electrodes turns out to be in the tune of 358C kinetics for spatially resolved biochips. Anal Biochem. 2003;317:
186–200.
(refer to Figure 11c), which is much above the safe permissi-
5. Das S, Das T, Chakraborty S. Modeling of coupled momentum, heat
ble limit of about 58C. This can be attributed to the existence and solute Transport during DNA hybridization in a microchannel in
of much stronger localized transverse potential gradients in presence of electro-osmotic effects and axial pressure gradients.
Case C, when compared with that for the Cases A and B. Microfluidics Nanofluidics. 2006;2:37–49.
For a physical assessment of the prevailing situation, a scal- 6. Hitt DL, McGarry M. Numerical simulations of laminar mixing sur-
faces in pulsatile microchannel flows. Math Comput Simul. 2004;65:
ing argument in favor of such behaviour can be put forward, 399–416.
by noting that the maximum temperature rise, DT, can be 7. Glasgow I, Aubry N. Enhancement of microfluidic mixing using
estimated as time pulsing. Lab Chip. 2003;3:114–120.
8. Glasgow I, Batton J, Aubry N. Electroosmotic mixing in microchan-
sðE2x þ E2y ÞDt nels. Lab Chip. 2004;4:558–562.
DT  (37) 9. Liu RH, Stremler MA, Sharp KV, Olsen MG, Santiago JG, Adrian
rCp RJ, Aref H, Beebe DJ. Passive mixing in a three-dimensional ser-
pentine microchannel. J Microelectromech Syst. 2000;9:190–207.
where Dt is the time elapsed. In the earlier expression, 10. Xia HM, Wan SYM, Shu C, Chew YT. Chaotic micromixers using
Ey  DVH , where DVT is the potential difference between the
T
two-layer crossing channels to exhibit fast mixing at low Reynolds
transverse electrodes. Clearly, DVT for Case C (i.e., 40 V) numbers. Lab Chip. 2005;5:748–755.
turns out to be much higher than that for Case B (i.e., 5 V), 11. Johnson TJ, Ross D, Locascio LE. Rapid microfluidic mixing. Anal
Chem. 2002;74:45–51.
resulting in much greater temperature increments in the for- 12. Ajdari A. Generation of transverse fluid currents and forces by an
mer case than the later. As a consequence, although the rate electric field: electro-osmosis on charge-modulated and undulated
of DNA hybridization is clearly more favorable in Case C surfaces. Phy Rev E. 1996;53:4996–5005.
than in Case B, the former case does not offer with a techno- 13. Kuksenok O, Yeomans JM, Balazs AC. Using patterned substrates
to promote mixing in microchannels. Phys Rev E. 2002;65:031502.
logically viable option for augmenting the hybridization per- 14. Zhang J, He G, Liu F. Electro-osmotic flow and mixing in heteroge-
formance in a DNA assay. neous microchannels. Phys Rev E. 2006;73:056305.
15. Tseng WL, Hsieh MM, Wang SJ, Chang HT. Effect of ionic
strength, pH and polymer concentration on the separation of DNA
Conclusions fragments in the presence of electroosmotic flow. J Chromatogr A.
The present model discusses a novel approach to ensure a 2000;894:219–230.
faster rate of DNA hybridization, by creating a localized pH 16. Peng XY, Li PCH. A three-dimensional flow control concept for
single-cell experiments on a microchip. II. Fluorescein Diacetate
gradient on employment of local transverse electric fields metabolism and calcium mobilization in a single yeast cell as
over selected portions of the channel length. Arrangement of stimulated by glucose and pH changes. Anal Chem. 2004;76:5282–
such transverse electrodes is primarily motivated because of 5292.
the following reason. A progressive depletion of the ss-DNA 17. Righetti P, Bossi A. Isoelectric focusing of proteins and peptides in
gel slabs and in capillaries. Anal Chim Acta. 1998;372:1–19.
molecules from the bulk (owing to nonspecific adsorption 18. Lee GB, Fu LM, Lee CY, Yang RJ. Dispersion control in microflui-
and 3D hybridization at the probes that are located at further dic chips by localized z potential variation using the field effect.
upstream sections), during their transport through the micro- Electrophoresis. 2004;25:1879–1887.
channel, is likely to result in a reduced rate of DNA hybrid- 19. Molloy RF, Leighton DT Jr. Binary oscillatory cross-flow elec-
ization at channel sections that are located further down- trophoresis: theory and experiments. J Pharm Sci. 1998;87:1270–
1281.
stream. However, on employment of transverse electrode 20. Chen Z, Chauhan A. Taylor dispersion in cyclic electric field-flow
pairs, localized enhancements in the wall zeta potential can fractionation. Phys Fluid. 2006;18:067105–067116.
be achieved, which in turn, can promote a stronger convec- 21. Chen Z, Chauhan A. DNA separation by EFFF in a microchannel.
tive transport, and accordingly, a faster rate of DNA hybrid- J Colloid Interface Science. 2005;285:834–844.
22. Tang GY, Yang C, Chai JC, Gong HQ. Joule heating effect on elec-
ization can be achieved at capturing probes that are located troosmotic flow and mass species transport in a microcapillary. Int.
further downstream relative to the channel inlet section. J. Heat Mass Trans. 2004;47:215–227.
However, strength of the transverse electric field cannot be 23. Sonuart TL, Baygents JC. Electrically-driven fluid motion in chan-
selected at will, but needs to be constrained within permissi- nelswith streamwise gradients of the electrical conductivity. Colloid
ble limits, so as to ensure that no further thermal denatura- Surf A. 2001;195:59–75.
24. Dwyer JD, Bloomfield VA. Brownian dynamics simulation of probe
tion of the hybridized DNA molecules takes place. The pres- diffusion in DNA: effects of probe size, charge and DNA concentra-
ent model provides with a scientific guideline for solution of tion. Biophys Chem. 1995;57:55–64.
the resultant optimization problem, so as to achieve the fast- 25. Phillies GDJ. Universal scaling equation for self-diffusion by macro-
est rate of DNA hybridization without violating the upper molecules in solution. Macromolecules. 1986;19:2367–2376.
26. Tracy MA, Pecora R. Macromolecular synthesis, characterization
limits of temperature rise that can be allowed in practice. and dynamics of a rod/sphere composite Liquid. Macromolecules.
1992;25:337–354.
27. Wattenbarger MR, Bloomfield VA, Bu Z, Russo P. Tracer diffusion
Literature Cited of proteins in DNA solutions. Macromolecules. 1992;25:5263–5265.
1. Chan V, Graves DJ, McKenzie SE. The biophysics of DNA hybridiza- 28. Carré A, Lacarrière V, Birch W. Molecular interactions between
tion with immobilized oligonucleotide probes. Biophys J. 1995;69: DNA and an aminated glass substrate. J Colloid Interface Sci.
2243–2255. 2003;260:49–55.

1098 DOI 10.1002/aic Published on behalf of the AIChE May 2007 Vol. 53, No. 5 AIChE Journal
29. Parks GA. Isoelectric points of solid oxides solid hydroxides and 33. Lee CS, McManigill D, Wu CT, Patel B. Factors affecting direct
aqueous hydroxo complex systems. Chem Rev. 1965;65:177–198. control of electroosmosis using an external electric field in capillary
30. Carré A, Roger F, Varinot C. Study of acid/base properties of oxide, oxide electrophoresis. Anal Chem. 1991;63:1519–1523.
glass and glass-ceramic surfaces. J Colloid Interface Sci. 1992;154:174–183. 34. Wu HY, Liu CH. A novel electrokinetic micromixer. Sens Actuators
31. Kirby BJ, Hasselbrink EF. Jr. The z potential of microfluidic sub- A. 2005;118:107–115.
strates. I. Theory, experimental techniques, and effects on separa- 35. Kim JH, Marafie A, Jia X, Zoval JV, Madou M. Characterization of
tions. Electrophoresis. 2004;25:187–202. DNA hybridization kinetics in a microfluidic flow channel. Sens
32. Hayes MA, Ewing AG. Electroosmotic flow control and monitoring Actuators B. 2006;113:281–289.
with an applied radial voltage for capillary zone electrophoresis.
Anal Chem. 1992;64:512–516. Manuscript received July 6, 2006, and revision received Jan. 16, 2007.

AIChE Journal May 2007 Vol. 53, No. 5 Published on behalf of the AIChE DOI 10.1002/aic 1099