Available online at www.sciencedirect.

com
Colloids and Surfaces A: Physicochem. Eng. Aspects 323 (2008) 2835
A coupled electro-hydrodynamic numerical modeling
of droplet actuation by electrowetting
Alborz Arzpeyma, Shan Bhaseen,
Ali Dolatabadi

, Paula Wood-Adams
Department of Mechanical and Industrial Engineering, Concordia University, Montreal, Quebec, Canada
Received 1 July 2007; received in revised form 9 December 2007; accepted 10 December 2007
Available online 23 December 2007
Abstract
A numerical study is performed to investigate droplet actuation under electrowetting forces. A volume of uid, VOF, technique is applied to
track the 3D interface. The Laplace equation with proper boundary conditions is solved to predict the time- and space-dependent electric eld
in the domain. The droplet actuation with respect to the powered electrode, the contact angle distribution at the three-phase line, and the droplet
morphology are studied. Additionally, other issues such as droplet velocity, contact angle hysteresis, and droplet location with respect to the
powered electrode (switching frequency) are investigated to characterize the droplet behavior between two parallel substrates.
2008 Elsevier B.V. All rights reserved.
Keywords: Electrowetting; Droplet actuation; Volume of uid; Switching frequency
1. Introduction
Recently, there has been an incredible advance in microfabri-
cation technologies. For example, currently the fabrication and
development of an enormous variety of so-called lab-on-a-chip
(LOC) devices is one of the most studied areas in biotechnol-
ogy. A LOC device is a small chemical or biological laboratory
mounted on a credit card-sized chip consisting a network of
microchannels, electrodes, electric circuits, and sensors. One of
the most challenging issues in dealing with this class of devices
is the handling of uids on this scale. The work described here
is focused on this issue.
Contrary to bulk ows, in microuidic devices surface ten-
sion plays a signicant role and can in fact be harnessed to drive
the motion of the uid. Several techniques have been developed
to drive liquids using the concept of varying interfacial tension
to create an imbalance of forces within the uid. Three exam-
ples are the use of thermocapillarysurfactants, dielectrophoresis,
and electrowetting. Among these, electrowetting is particularly
interesting due to advantages such as the absence of heat gen-

Corresponding author. Tel.: +1 514 848 2424x5188; fax: +1 514 848 3175.
E-mail address: [email protected] (A. Dolatabadi).
eration, rapid switching response, and low power consumption.
In this work, the movement of single droplets via electrowetting
is studied.
Electrowetting works by changing the interfacial energy
between the solid and the liquid; going from hydrophobic or
non-wettingconditions tohydrophilic or wettingconditions after
electrical activation. The external electric eld rearranges the
ions and counter ions in the liquid leading to a change in the sur-
face energy. Atypical liquid for such experiments is water with a
small quantity of a salt like KCl. By using an array of electrodes
and switching them successively, one can make a droplet move
inside a microchannel with a signicant velocity.
Pollack et al. [1] performed experiments to investigate the
phenomenon. They studied the manipulation of 900 nl droplets
using electrowetting as the actuation force. Their experiments
showed that the method is capable of performing several
laboratory tasks such as droplet merging, splitting, mixing,
and dispensing. They managed to reach a droplet velocity of
100 mm/s with a rather low voltage of 60 V. This velocity is
about 40 times more than velocities achieved by other methods.
Also, the fabrication of a micro-pump using the surface tension-
based actuation by Lee and Kim [2] showed the high range of
velocity that can be achieved with considerably small power
consumption. The range of actuation voltage in electrowetting
0927-7757/$ see front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.colsurfa.2007.12.025
A. Arzpeyma et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 323 (2008) 2835 29
is limited by two phenomena; contact angle saturation and elec-
trolysis. Contact angle saturation means that after a certain point,
a further increase in the voltage does not result it a change in
the contact angle [4]. The occurrence of electrolysis depends on
the surface properties. At high voltages, if the coated insulator is
defective in any locations, electrical current can pass through the
liquid [3] and cause electrolysis. Therefore when increasing the
voltage, one of these two phenomena will occur eventually and
provide a practical limit to the process. Surface characteristics
determine which happen rst.
In addition to the experiments in this eld, many analyti-
cal and numerical studies have been done. Numerical studies
are particularly important because conducting experiments at
such small scales is difcult and costly. Shapiro et al. [4] used
the concept of energy minimization to model the electrowetting
effect on sessile droplets. Zeng and Korsmeyer [5] investigated
analytically and numerically the electro-hydrodynamic forces
which encompasses electrowetting on dielectric (EWOD) and
dielectrophoresis and their application in droplet generation and
manipulation. The effect of electric charge distribution on the
droplet leading and trailing edge was studied by Baird et al.
[6]. The electrostatic force on both liquidgas interface and the
three-phase line was studied analytically and numerically. In
addition, they investigated the effect of substrate material on the
electrostatic force on the interface.
The above studies were used as the basis for carrying out the
numerical modeling by Dolatabadi et al. [7]. They decoupled the
electric and hydrodynamic elds using a one-dimensional model
for the electric eld. The effect of electric eld on the interfacial
energies was consideredthrougha direct change incontact angle.
That model, though simple, showed good agreement with the
experimental results.
However, to capture the effect of the electric force more pre-
cisely, one needs to solve the electric eld throughout the whole
domain and nd the electric force for each point. These forces
then are used to move the droplet on the electrode path. This is
the goal of the current work. The electric force is solved for each
gridcell andat everytime step. Ineffect, the contact angle change
is considered a variable phenomenon in time and space leading
to a more accurate modeling of droplet transport. The droplet
actuation is discussed in detail. In addition, the droplet velocity
and the effect of electrode switching frequency are investigated.
2. Electrowetting concept
In electrowetting, an external electric eld is applied over a
portion of the liquid droplet. The electric eld leads to a higher
charge concentration at the three-phase line changing the energy
balance in that area and thus leading to a decrease in surface
energies. As a result, the liquidsolid contact angle is smaller in
the area of higher charge concentration.
Fig. 1 is a schematic of a droplet between two plates which
has been activated by an electrode on one side of the droplet.
In order to achieve such droplet morphology, both top and bot-
tomplates are covered with Teon rendering themhydrophobic.
On the top plate there is a continuous ground electrode. On the
bottomplate, there is an array of electrodes that can be red con-
Fig. 1. Schematic of droplet actuation under electrowetting (the darker electrode
represents the powered electrode).
secutively to induce continuous motion of the droplet. A layer
of insulator, for example Parylene C, is placed on the bottom
plate, between the Teon layer and the electrodes. This layer
enhances the energy storage of the systemand prevents electrol-
ysis under higher voltages. This type of electrowetting is called
electrowetting on dielectric, EWOD.
To develop the theory describing electrowetting we begin by
considering the force balance at the three-phase line which may
be described by the Young equation:

SG
=
SL
+
LG
cos (1)
where
SG
,
SL
, and
LG
are the surface energies at solidgas,
solidliquid, and liquidgas interfaces, respectively. The energy
level at this line is decreased by an increased charge concen-
tration resulting from the application of an electric eld. In
particular, the solidliquid interfacial energy is changed while
the liquidgas and solidgas surface energies are almost unaf-
fected. The change in solidliquid surface energy is given by
d
SL
d
= (2)
where represents the electric potential and is the charge
density at the interface. The system of electrodes, droplet, and
insulator form a capacitor. Since the droplet is considered to be
conductive, it effectively becomes a part of the conductive plates
of the capacitor. The relation between the charge density and the
system capacitance is then
= C =

0

l
d
A
SL
(3)
where
0

l
is the dielectric constant of the insulating layer, d is
the insulator thickness, A
SL
is the contact area of solidliquid, C
the systemcapacitance, and is the applied voltage between the
liquid and the electrode. The energy stored in the systemis given
simply by the capacitor energy as (1/2)C
2
. By combining the
above equations, we can transform the Young equation into the
YoungLippmann equation:
cos
V
= cos
0
+

0

l
2
LG
d

2
(4)
where
0
and
V
are solidliquid contact angles before and after
actuation, respectively. Here the effect of the electric double
layer, EDL, has been neglected when we considered the stored
energy. Inreality, this layer forms another capacitor inseries with
the insulating layer. Since the EDL thickness is much smaller
30 A. Arzpeyma et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 323 (2008) 2835
than the insulator thickness, it does not signicantly affect the
capacitance of the whole system.
3. Numerical techniques
The liquid droplet is assumed to be incompressible, three-
dimensional, Newtonian, and the ow is laminar. The mass and
momentum conservation equations are then


V = 0 (5)

V
t
+ (

V

V) =
1

p +
1

+ g +

F
b
(6)
in which,

V represents the velocity vector, p the pressure, the
liquiddensity, the stress tensor, gthe gravitational acceleration,
and

F
b
any body force (per unit volume) acting on the uid. As
the uid is Newtonian the stress tensor is
= (

V + (

V)
T
) (7)
A modied version of the volume of uid (VOF) method devel-
oped by Bussmann et al. [8] is used. VOF is based on averaging
the phases, in which the volume fraction is treated as a single
eld that is advected in an Eulerian way. To keep track of the
interface, the advection equation of volume fraction is solved:
f
t
+ (

V )f = 0 (8)
where f is the volume fraction.
The above equations are discretized on a Cartesian mesh.
The marker and cell (MAC) method is used in which, veloci-
ties are specied at the center of the cell edges and pressure at
each cell center [9,10]. The time discretization of the momen-
tum equation is divided into two steps. First, an interim velocity
is computed explicitly from convective, viscous, gravitational,
and body forces for a time step t. Then, the pressure is calcu-
lated implicitly. As momentum cannot be advected more than a
grid per time step, the Courant number should be less than one.
The same condition applied for the volume tracking as it can be
only advected to the neighboring cells. In the current work, the
Courant number is considered to be 0.3 [8].
The geometry and boundary conditions were chosen based on
the experiments conducted by Pollack et al. [1] in order to facili-
tate comparison. The droplet diameter is 1900 m, the electrode
length is 1500 m (see Fig. 2) and the contact angle on Teon
in the absence of actuation is 104

.
Along with the hydrodynamic equations presented previ-
ously, the Laplace equation is solved on the entire domain to
calculate the electric potential in every grid cell at each time
step.

2
= 0 ( = (x, y, z)) (9)
Several boundary conditions are involved in solving the Laplace
equation. There is an array of electrodes on the bottom plate
and only one is switched on at any time. The top electrode
is grounded; its potential is zero. Since the droplet itself is a
conductive solution in contact with the top electrode, there is
no potential dissipation in it. As the droplet moves inside the
Fig. 2. Schematic of the physical model used in numerical simulations; H is the
gap between the two plates and D is the droplet diameter.
channel, the boundary conditions and consequently, the poten-
tial distribution change. The electric eld is solved with respect
to the boundary conditions in time and space. The effect of the
potential in each cell at the three-phase line is described using a
discretized formof Eq. (4). The contact angle fromthis equation
is then incorporated in each cell which changes the surface ten-
sion forces. The resultant force is then fed to the NavierStokes
equation, Eq. (6), which is solved simultaneously with the other
equations. To consider the time dependency of the voltage, it is
solvedat eachcycle andthe relevant electric forces are calculated
at every time step.
Following the experiments of Pollack et al. [1], the droplet
volume is considered to be 900 nl, between two parallel sub-
strates with a 300 mgap. The droplet therefore has a cylindrical
shape with a diameter of 1900 m. The liquid is considered to
be conductive, i.e. the water contains the ions from a salt. The
electrode length is 1500 m unless otherwise stated. The con-
tact angle before actuation is 104

. In practice, interdigitated
electrodes are used for actuation. Such electrodes have zigzag
sides which are placed one after the other to minimize the gap
between the two adjacent electrodes.
We have neglected gravity forces as we are dealing with
micro-scale ows, in which the Bond number is relatively small.
The Bond number is the ratio of gravitational to surface tension
forces:
Bo =
gL
2

LG
(10)
With our geometry, the Bond number is about 0.012 indicating
that the gravitational energy is about two orders of magnitudes
smaller than the surface energy and therefore is negligible.
4. Results and discussions
As explained previously, the electric eld is solved in the
whole domain. Fig. 3 shows the distribution of equipotential
surfaces for an actuation voltage of 50 V. The electric energy
dissipates from the high voltage region (the powered electrode
at the right) to the zero potential level (ground electrode at the
top). This electric eld affects both the liquidgas interface and
the three-phase line.
A. Arzpeyma et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 323 (2008) 2835 31
Fig. 3. Electric eld distribution inside the channel with the actuation voltage of 50 V.
Zeng and Korsmeyer [5] investigated the electrostatic forces
on the droplet. Based on their study, the force acting on the
liquidgas interface is negligible when the ratio of insulator
thickness to channel gap is small. The only force term that
remains is the one acting on the three-phase line. This force
has a greater value where the contact line is closer to the elec-
trode. Away from the electrode position the voltage drops and
the force decreases as well. This force leads to a decrease in
the contact angle at the three-phase interface. Droplet move-
ment is initiated by the imbalance in surface energies at the
droplet leading and trailing edges. This mechanism is illus-
trated in Fig. 4 by examining the droplet top and side views.
The droplet stretches in the direction of the motion while its
diameter decreases in the direction perpendicular to the motion.
The non-actuated droplet has a circular top view, while after
actuation this shape becomes oval. As the droplet advances on
the electrode, a decrease in contact angle at the leading edge can
be observed.
Fig. 5 shows the contact angle distribution while the droplet
approaches the actuated electrode and advances on it. In the
graphs, the contact angle at each point is plotted with respect to
the radial position of that point. In other words, the solidliquid
contact angle is presented in terms of the polar angle of . The
symmetry boundary condition is used to simulate the droplet.
Hence, only half of the domain is modeled. The contact angles
in the graphs are for the modeled half (the upper section of the
Fig. 4. Droplet morphology and advancement on the electrode: (a) top viewand
(b) side view (plane of symmetry); leading and trailing edge contact angles are
shown by
A
and
R
, respectively.
32 A. Arzpeyma et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 323 (2008) 2835
Fig. 5. Various steps of droplet actuation: (a) 200 m and (b) 50 m separation of droplet leading edge with the powered electrode, (c) 50 m and (d) 300 m droplet
on the electrode; the equipotential lines are shown in each case.
center line). The other half is a mirror image and is not shown
in the line graphs. In Fig. 5a there is a separation of 200 m
between the droplet leading edge and the actuated electrode. In
this case, there is a very small contact angle change due to the
weak inuence of the electric potential at the interface. When
the droplet approaches the electrode, the contact angle change
becomes noticeable. In Fig. 5b, the droplet is about to touch the
electrode and the separation is 50 m. At this point, a 28

portion
of the droplet experiences a change of up to about 60

in contact
angle, while the rest of the droplet has the initial contact angle of
104

. The equipotential lines are also shown in the gure. Fig. 5c

is associated with the time right after the droplet has touched the
electrode edge. Interestingly, the largest portion of the droplet
is actuated under this condition; about 82

are affected, which

is also larger than when the droplet is completely covering the
width of the electrode (Fig. 5d). In case Fig. 5c, the electric
lines inuence more area of the droplet than in the case where
the droplet is touching the electrode. That is because when the
conductive droplet sits on the electrode, the electric potential
pattern around the contact line undergoes an acute change from
the electrode potential to zero.
In case Fig. 5d, only the portion of the three-phase line that
is on the electrode is actuated. Any further advancement of the
droplet on the electrode results in the same actuated area rep-
resented in Fig. 5d. Since a larger portion of the contact line is
actuated for an approaching droplet (as shown in Fig. 5c), one
may conclude that an array of electrodes with smaller length
would better actuate the droplet. In such a case, a small portion
of the droplet is placed on the electrode and the inuence of
the electric potential is present in a larger portion of the leading
edge.
The sudden change of the plotted contact angle comes from
the fact that the effect of the electric eld is very strong in the
proximity of the powered electrode. The electric potential has
a very small value further away as it drops rapidly with space.
Therefore, the points that are very close to the powered elec-
trode or located on it will face a large decrease in the contact
angle while the ones which are further will maintain the non-
actuated contact angle value. Using a very ne computational
mesh would capture the spatial variation of contact angle more
smoothly; however, this would need an incomparably longer
computational time. In reality, the liquid cannot experience such
a drastic change in contact angle between to adjacent points in
the frontier. It is important to mention that the plotted values for
contact angle are the values at the beginning of the each time
step before solving for the hydrodynamic equations. At each
time step these values will be smoothed in the VOF solver using
the continuous surface force (CSF) model [11]. Hence, at the
end of each time step there is no sharp change in contact angle
and surface tension forces.
It may also be noticed that if the electrode size increases,
the effect of potential on the trailing edge contact angle also
increases. In turn, this will cause an elongation of the droplet in
directions other than that of the motion. This case is unfavorable
for droplet transport, as the opposing force tends to reduce the
droplet velocity.
A. Arzpeyma et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 323 (2008) 2835 33
In order to evaluate the reliability of our results, two differ-
ent grid sizes of 25 and 50 m or 38 and 76 cells per droplet
diameter have been tested. A difference of 7% in droplet veloc-
ity is attained for actuation voltages up to 55 V. The applied
VOF technique has the best result when the cells are uni-
form cubes. Therefore, to decrease the mesh size by half, the
number of cells increases by a factor of 8 which signicantly
increases the computational time. Since the difference in results
between the coarser and the ner meshes are small, in this
work all the reported case was simulated with a grid size of
50 m.
Next, the steady state droplet velocity is calculated under
the inuence of different actuation voltages ranging from 40
to 60 V. Experimental data show that there exists a threshold
voltage below which there is no droplet movement. We can
understand this by considering the series of events that occur
as the actuation voltage increases from zero. At a voltage below
the threshold voltage the contact angle at the three-phase line
starts to decrease, but does not lead to movement. In order to
induce movement the actuation force must be large enough to
overcome inertia and contact angle hysteresis. A real droplet on
a real surface cannot be characterized by a single contact angle.
In fact the contact angle is normally different around the droplet,
depending on the surface roughness and heterogeneity [12,13].
The largest contact angle is called the advancing angle and the
smallest is called the receding angle. Therefore, a measured con-
tact angle is always between these two advancing and receding
angels. The difference between the two angles is called contact
angle hysteresis.
There are some case specic empirical models in the liter-
ature to predict hysteresis, yet the phenomenon itself is still
not well understood. In this study, at rst, we solved the
system of equations without considering the effect of con-
tact angle hysteresis. In our primary results the droplet did
not start moving before an actuation voltage of 35 V. This is
mainly due to droplet inertia. Then the results were shifted
with respect to voltage to best agree with the experimental data
[1] where the velocity was directly measured. Fig. 6 shows
the shifted results compared to experimental data from Ref.
[1].
The difference between the numerical and experimental
results observed in Fig. 6 is believed to be due to the assumption
of a constant contact angle hysteresis. In reality, the contact hys-
teresis depends on the position of the droplet on the surface and
also the surface characteristics. Although there is a small differ-
ence in the slopes, the numerical results are in good agreement
with the experimental ones. In this graph the droplet steady state
velocity is shown.
An important issue in controlling the droplet velocity is the
electrode switching frequency. At steady state the optimum fre-
quency in terms of having smooth consistent motion is dened
with respect to the time the droplet needs to travel along a single
electrode. In other words, we can calculate the optimum steady
state switching frequency from:
f =
V
L
(11)
Fig. 6. Droplet average velocity with air as the surrounding medium, under
different actuation voltages. The experimental data are taken from Ref. [1].
where V, f, and L are the droplet average velocity, switching
frequency, and electrode length, respectively.
The situation is different in transient conditions such as when
initiating motion or accelerating/decelerating the droplet. In this
case the optimum would be to power, the subsequent electrode
when the droplet leading edge contact line reaches the end of
the current electrode. Nowwe have to consider an instantaneous
velocity with a corresponding optimum instantaneous switch-
ing frequency. An example of this case is shown in Fig. 7. The
optimum switching frequency and the velocity increases while
traveling on the rst electrodes and reaches the steady state
condition eventually where the switching frequency is 14.3 Hz.
Of course in practice it might not be possible to sense the
current location of the droplet and apply the optimum switching
frequency at all times. Therefore we have also considered the
more realistic case of a constant switching frequency (Fig. 8).
Here we have studied the effect of four different constants
Fig. 7. Droplet velocity and frequency for the ideal case, V=50 V.
34 A. Arzpeyma et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 323 (2008) 2835
Fig. 8. The effect of ring frequency on the droplet velocity; the x-axis is the
position of droplet leading edge over the electrode length V=50 V.
switching frequencies, all at 55 V. In the study shown in Fig. 8,
all the cases start with the same condition and therefore at the
beginning, the velocities are the same. By the end of the rst
electrode we begin to see the effect of the different switching
frequencies. At any point, if the frequency is lower than ideal,
the current electrode remains switched longer than ideal and
the leading edge of the droplet moves onto the subsequent, still
un-powered electrode. Simultaneously, the powered electrode
stretches the trailing edge contact line in the direction opposite
to the motion, slowing down the droplet. If we choose a fre-
quency that is too high, the droplet starts to move but it may
stop after some time. Two such cases are shown in Fig. 8 where
the frequency is 16.7 and 13.3 Hz causing the droplet to stop
moving before reaching steady state. It is important to note that
while one of these cases, 13.33 Hz, is actually lower than the
optimum steady state frequency of 14.3 Hz it is higher than
the optimum frequency in much of the transient region. This
illustrates clearly the importance of understanding the transient
behavior of droplet actuation in designing and controlling such
digital uidic systems. The results in Fig. 8 also demonstrate
Fig. 9. Droplet velocity variation for three electrode lengths of 1400, 1500, and
1600 m; the x-axis is the position of droplet leading edge over the electrode
length.
that it is possible to nd a constant frequency under which the
droplet can be accelerated without stopping and reach close to
the velocity of the ideal case.
Fig. 9 shows the effect of electrode size on the droplet velocity
under the ideal, variable switching frequency at an actuation
voltage of 55 V. In all cases the velocity reaches to a steady state
condition after a while. These results show that by increasing
the electrode size the steady state velocity increases as well.
However, it shouldbe notedthat if the electrode size is larger than
the droplet diameter, a bigger area of the droplet is affectedby the
electric eld including a portion the trailing edge of the droplet.
The presence of an electric force at that portion is unfavorable
as it is opposes the droplet motion and slows the droplet down.
Therefore, there is a practical limit for increasing the electrode
size.
5. Conclusions
A numerical model was built to investigate the electrowet-
ting phenomenon by coupling the hydrodynamic and electrical
equations. These equations were solved simultaneously in three
dimensions to predict the droplet behavior under electric forces.
Simulated droplet velocities for various actuation voltages were
in agreement with experimental data from the literature con-
rming the validity of our model. The effect of droplet location
relative to the powered electrode on the droplet morphology was
investigated demonstrating that optimum actuation is achieved
just as the droplet touches the powered electrode. In this sit-
uation the largest fraction of the leading edge portion of the
three-phase contact line is wetting the surface. It was also shown
that the effect of contact angle hysteresis on the droplet veloc-
ity can approximated by a single threshold voltage for droplet
movement. Although this works satisfactorily for the conditions
simulated here it is likely that a more sophisticated spatially
dependant model for contact angle hysteresis is needed to
describe the details of the electrowetting actuation process. The
effects of electrode switching frequency and electrode length
on the droplet motion were studied. The results indicate that
the actuation of droplets by electrowetting is a complex phe-
nomenonespeciallyunder transient conditions andtherefore will
benet from additional numerical studies. The results also pro-
vide some initial guidelines for developing and operating digital
microuidics systems.
Acknowledgments
The authors would like to thank Simulent Inc., Toronto,
Ontario for providing access to the SIMULENT code which
is modied for the present simulations. They would also like
to acknowledge support from Natural Sciences and Engineer-
ing Research Council (NSERC) and Le Fonds qu eb ecois de la
recherche sur la nature et les technologies (FQRNT).
References
[1] M.G. Pollack, A.D. Shenderov, R.B. Fair, Electrowetting-based actuation
of droplets for integrated microuidic, Lab Chip 2 (2002) 96101.
A. Arzpeyma et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 323 (2008) 2835 35
[2] J. Lee, C.J. Kim, Surface-tension-driven microactuation based on continu-
ous electrowetting, J. Microelectromech. Syst. 9 (2) (2000).
[3] H. Moon, S.K. Cho, R.L. Garrell, C.-J. Kim, Low voltage electrowetting-
on-dielectric, J. Appl. Phys. 92 (7) (2002) 4080.
[4] B. Shapiro, H. Moon, R. Garrell, C.J. Kim, Equilibriumbehaviour of sessile
drops under surface tension applied external elds and material variations,
J. Appl. Phys. 93 (9) (2003).
[5] J. Zeng, T. Korsmeyer, Principles of droplet electrohydrodynamics for lab-
on-a-chip, Lab Chip 4 (2004) 265277.
[6] E. Baird, P. Young, K. Mohseni, Electrostatic force calculation for an
EWOD-actuated droplet, in: Microuidics and Nanouidics, Springer-
Verlag, 2007.
[7] A. Dolatabadi, K. Mohseni, A. Arzpeyma, Behaviour of a moving droplet
under electrowetting actuation: numerical simulation, Can. J. Chem. Eng.
84 (1) (2006) 1721.
[8] M. Bussmann, J. Mostaghimi, S. Chandra, On a three-dimensional vol-
ume tracking model of droplet impact, Phys. Fluids 11 (6) (1999) 1406
1417.
[9] F.H. Harlow, J.E. Welch, Numerical calculation of time-dependent viscous
incompressible ow of uid with free surface, Phys. Fuids 8 (12) (1965)
21822189.
[10] J.E. Welch, F.H. Harlow, J.P. Shannon, B.J. Daly, The MAC method, Tech-
nical Report LA-3425, LANL, 1966.
[11] J.U. Brackbill, D.B. Kothe, C. Zeamch, A continuum method for modeling
surface tension, J. Comput. Phys. 100 (1992) 335.
[12] G.O. Berim, E. Ruckenstein, microscopic interpretation of the depen-
dence of the contact angle on roughness, Langmuir 21 (2005) 7743
7751.
[13] N.A. Patankar, On the modeling of hydrophobic contact angles on rough
surfaces, Langmuir 19 (2003) 12491253.