PAPER www.rsc.

org/loc | Lab on a Chip

Flow rate analysis of a surface tension driven passive micropump{{

Erwin Berthiera and David J. Beebe*b
Received 21st May 2007, Accepted 6th July 2007
First published as an Advance Article on the web 30th July 2007
DOI: 10.1039/b707637a

A microfluidic passive pumping method relying on surface tension properties is investigated and a
physical model is developed. When a small inlet drop is placed on the entrance of a microfluidic
channel it creates more pressure than a large output drop at the channel exit, causing fluid flow.
The behavior of the input drop occurs in two characteristic phases. An analytical solution is
proposed and verified by experimental results. We find that during the first phase the flow rate is
stable and that this phase can be prolonged by refilling the inlet drop to produce continuous flow
in the microchannel.

Introduction contact angle to a lower dynamic contact angle. In the second

phase, the contact angle remains constant as the contact area
Microfluidics has been an active area of research over the last decreases until it matches the area of the input port.
decade and promises to bring improvements in areas such as
drug discovery, diagnostics, biosensing and chemical synthesis,
where handling small volumes is critical.1 Pumping fluids is an
Flow rate calculation
ubiquitous requirement and many approaches have been used Consideration of dimensionless numbers of microscale fluid
to drive fluids.2 Tubing connection issues and the loss of dynamics allow for simplification of the hypothesis. The
analytes in the dead volume of the tubing has limited the use of Reynolds number takes a typical value of Re = rU0L0/g # 1,
traditional external pumping devices in some applications. meaning that the flow is very laminar, allowing the use of the
Several interesting integrated on-chip pumping methods Washburn law for linear pressure drop in a channel with a
exist but require specialized external equipment.3,4 Passive rectangular cross section. A capillary number of Ca = gU0/c #
pumping methods provide for semi-autonomous5 or autono- 1024 means that surface tension dominates inertia, therefore at
mous6,7 integrated pumping systems, though most of these every instant the system can be considered static. The Bond
require additional fabrication steps (beyond those required for number Bo = rgR2/c # 0.01 justifies neglecting gravitational
a simple microchannel). forces on the drop and allows the hypothesis that the input
A method requiring only a traditional micropipette that drop is spherical at all times. In addition we assume that the
utilizes the surface tension of fluids has been shown to create
flow through a microfluidic channel.5 A large drop is placed on
the outlet port of the device and a smaller drop, with a smaller
radius, on the input port (Fig. 1). According to the Laplace
Law, the smaller curvature radius of the smaller drop will
cause a larger pressure and thus create flow from the small
drop towards the large drop. To enable efficient use of this
method and to facilitate devices design, we present an
analytical model incorporating design parameters, such as
the geometry of the device and the properties of materials used
to fabricate the device.
Guided by experimental results and similar behaviors
observed in evaporating droplets,8 we have developed a two
phase model (Fig. 2) for the decrease in volume of the inlet
drop. During the first phase, the contact area of the drop
remains constant as its contact angle varies from a high static

a
MacDiarmid Institute for Advanced Materials and Nanotechnology,
Department of Electrical and Computer Engineering, University of
Canterbury, Christchurch, NZ. E-mail: erwin.berthier@gmail.com
b
Department of Biomedical Engineering, University of Wisconsin –
Madison, Madison, WI, 53706, USA. E-mail: djbeebe@wisc.edu
{ The HTML version of this article has been enhanced with colour
images.
{ Electronic supplementary information (ESI) available: Appendix 1: Fig. 1 Passive pumping chip containing a small input drop and a
trigonometric relationships in a spherical cap and Appendix 2: refill. large output drop. (A) Before the placement of a drop. (B) and (C)
See DOI: 10.1039/b707637a During the flow.

This journal is ß The Royal Society of Chemistry 2007 Lab Chip, 2007, 7, 1475–1478 | 1475
 a(t) = Rw. From eqn (4) and (5) we find a differential equation
in H, the height of the drop (Appendix 1, ESI{):
dH H
~Kl
2 (6)
dt H zR2
2
w

pKw
with Kl ~ (7)
8c
By integrating eqn (6) with the boundary condition being the
initial height of the drop H0 at time t0 = 0 we can write a
function t(H):
 4  
H {H04
 H
t~{Kl zR2w H 2 {H02 zR4w ln (8)
4 H0
When the contact angle reaches the critical value of Hdyn,
Fig. 2 Recession of a drop during passive pumping modeled in two the model enters phase 2. The transition from phase 1 to phase
phases, each pictured in a different shade. 2 occurs at a value of H1 = Rwtan(Hdyn/2) at a time t1.

output drop is sufficiently large to neglect its surface tension Second phase
pressure. Using the Laplace Law for an input drop of radius R
The contact angle Hdyn then remains constant and the
and surface tension c we write:
drop reduces in wetted radius (Fig. 2). Similarly, from eqn (4)
2c and (5), using different trigonometric relationships to make
Pinlet drop {Poutput drop ~ (1)
RðtÞ Hdyn appear instead of Rw, we deduce a differential equation
on H:
The pressure drop in a channel is the sum of components
from different origins. Due to viscosity, a linear pressure drop dH 1
~Kh 3 (9)
occurs along the channel. In addition a punctual pressure drop dt H
occurs at corners or when the width of the channel changes
abruptly.9 We consider here a simple rectangular channel of
4
2c 1{coshdyn
length L0, width LD, height Lh and flow rate Q, with l = LD/ with Kh ~
 (10)
pKw 2{3coshdyn zcos3 hdyn zcos3 hdyn
Lh. For a liquid of viscosity g, the Washburn law10 is written:

8gL0 gðlÞ We integrate eqn (9) using the the boundary conditions
DPin channel ~ QðtÞ~Kw QðtÞ (2)
L3D Lh specified above, i.e. H1 and t1, and find a function H(t):

8 H(t) = [24Kh(t 2 t1) + H14]1/4 (11)

> 3
>
< 2 if lw4:45
with gðlÞ~ (3) When the wetted radius of the drop matches the radius of
>
> ð1zlÞ2
: if lv4:45 the port (a(t) = Rp), the contact area will stop receding and the
l2 dynamics will enter the last phase, during which the height of
From eqn (1) and (2), we obtain a simple differential the drop continues to diminish until the pressures at the inlet
equation that we will use throughout the modeling process to and outlet match. Experimental observations tend to suggest
find an expression of H, the height of the drop. Trigonometric that the drop collapses slightly earlier. Flow ends at a
relations specific to spherical caps are used to evaluate the theoretical value of H2 = Rptan(Hdyn/2) at a time t2.
volume of the inlet drop V in function of the other parameters
in the system (Appendix 1, ESI{): Global solution
2c The initial drop placed on the port of radius Rp = 375 mm has a
~Kw QðtÞ (4)
RðtÞ volume of V0 = 3.5 mL. The dimensions of the channel are L0 =
dV 35 mm, LD = 210 mm and Lh = 125 mm. Analytical solutions
with Q~{ (5) for the height, the width, the contact angle and the velocity are
dt
plotted in Fig. 3. The velocity was numerically derived from
the variations of the height, wetted radius and contact angle
First phase
(Appendix 1). Two remarks can be made concerning the
The drop placed on the input starts with a static contact angle solutions to the differential equations: (i) the solution found
Hstat (Fig. 2) and its contact area will not start diminishing can be extended to other channel geometries;9 (ii) as we neglect
until the dynamic contact angle Hdyn is reached. The value of dynamics and inertial effects, a drop placed on the device can
this angle is a property of the materials and the liquid used. be ‘‘refilled’’ during phase 1 without changing its motion
The constant parameter in this phase is the wetted radius equation (Fig. 4).

1476 | Lab Chip, 2007, 7, 1475–1478 This journal is ß The Royal Society of Chemistry 2007

Fig. 3 Experimental measures (black crosses) and analytical solution
(solid curve) of H, W, h and the velocity. Full vertical lines shows the
transition between the phases. Top: height of the drop. Middle top:
width. Middle bottom: contact angle. Bottom: velocity in the channel.
Fig. 4 Analytical solution of the pressure in the input drop in full
black line. Full vertical lines shows the transition between the phases.
During most of the first phase the pressure remains well bounded, as
illustrated by the two dashed horizontal lines.
Experimental results
Microchannels were fabricated using soft lithography techni-
ques and polydimethylsiloxane (PDMS, Sylgaard 184, Dow
Corning).11 The molds were created by coating SU-8 100
(Microchem) photoresist on a silicon wafer. A first 120 mm
thick layer defines the channels and a second 500 mm thick
layer defines the ports. The process is described in more detail
elsewhere.12 After developing (Microchem SU-8 Developer)
and cleaning the mold, PDMS was poured over of the wafer. A
transparency film, a thin rubber sheet and a 7 kg iron wieght
are placed on top to allow the SU-8 ports to pierce through the
PDMS layer and the stack is baked on a hot plate at 85 degrees
This journal is ß The Royal Society of Chemistry 2007
for 1 hour 30 minutes. The PDMS is then peeled off the mold
and laid on a microscope slide.
The height, H, the width, W = 2a(t), and the contact angle,
H, of the drop were measured using a Rame-hart goniometer
(Netcong, USA); the drops are dispensed using a 20 mL
micropipette. The values are recorded every second by the
integrated software DROPImage. Fig. 3 plots H, W and H of
the drop and the average velocity in the channel.
Results and discussion
The model and experiments suggest several points for
discussion. The analytical solution plotted in Fig. 3 shows a
good fit with the experimental data for phase 1 and 2. The
initial theoretical width does not match exactly the experi-
mental measurement because the drop is not perfectly
spherical, and by fixing the contact angle and the volume,
the spherical cap is completely defined. Discrepancies seen
at the end of fluid motion are likely due to the following:
(i) our negligence of the output pressure which becomes
increasingly relevant at the end of the motion and will tend
to slow down the flow, and (ii) the limitation of the
goniometer measurement capabilities, due to reflected light
and mirage effects, prevent accurately measuring small
spherical caps about a millimetre wide and only several
hundred micrometres tall.
Moreover, it has been observed that the end of phase 2
happens earlier than predicted, possibly due to border effects
of the port, as the fabrication process leaves a small ridge
around the port. It should be noted that a phenomena
presenting a striking similarity is observed during a droplet
evaporation process.8,13,14 An additional source of error in the
measured velocity in the channel is the lack of precision for a
macroscopic observation (i.e. drop height) to reveal informa-
tion on such minute variations.
Of this analysis, perhaps the most interesting result is the
stability of the velocity during the first phase. As we can see in
Fig. 4, the pressure in the input drop varies less than 10%
around an average pressure of 150 N m22, the fluid velocity
has an even smoother variation. For many laminar flow based
devices, constant velocity improves device performance.15 It is
therefore best to run it in this regime and this model provides
an accurate timing method to know when to refill the drop
before it switches into a phase 2 behavior.
It has been noticed previously (Fig. 4), that refilling the inlet
drop during the first phase of motion allows for continuous
perfusion. It is even possible in certain conditions to refill with
a volume of liquid maintaining the pressure constant
(Appendix 2, ESI{). This can be done only if the contact
angle at the time of the refill is lower than 90 degrees and that
the static contact angle is higher than 90 degrees, implying that
the substrate must be hydrophobic. In this case the volume of
the refill is written:
2pa3

Vrefill ~ 3coshðtÞ{cos3 hðtÞ (12)
3sin3 hðtÞ
Of course the refill doesn’t have to be done in this manner,
however this would lead to a discontinuity in the pressure, and
would be represented by an oblique arrow on Fig. 4.
Lab Chip, 2007, 7, 1475–1478 | 1477
 Furthermore, the prediction enables the creation of devices
with tuned flow resistances to (i) control the maximum velocity
in the channels, for slow flow applications, or (ii) to passively
direct the flow to a specific channel, for autonomous
microfluidic chips.
Finally, additional interest in surface tension passive
pumping can be envisaged in the interfacing with electro-
wetting on dielectric (EWOD) systems.16 Electrocapillarity or
electrowetting methods have the advantage of being integrable
in a microdevice, and deliver calibrated droplets at high speed,
which could be utilized to refill the ports. This method can be
envisaged to be a solution for interfacing continuous flow
microfluidics and digital microfluidics.
Conclusions
In summary, we have shown here that the suggested
mechanism of passive pumping based only on surface tension
provides a predictive analytical model that accurately char-
acterizes the passive pumping phenomena. The drop placed on
the input port goes through two phases of reduction, the first
during which the contact surface is constant and the contact
angle reduces from a static contact angle to a dynamic contact
angle. In the second phase, only the contact area recedes until
it reaches the area of the port, and finally a last phase in which
the height of the drop will continue reducing until the pressure
in the channel equilibrates.
The analytical solution not only allows one to characterize
the intensity of the flow and the duration of the relatively
stable first phase, but also shows that refilling of the inlet drop
is possible without change of pressure. This can provide for a
practical, long lasting and inexpensive method to obtain flow
in a channel without the traditional tubes and pumping
1478 | Lab Chip, 2007, 7, 1475–1478
systems. Moreover it is very appealing to cell studies as it
requires only the traditional manual or automatic pipettors.
Acknowledgements
This work was supported by funding from the MacDiarmid
Institute for Advanced Materials and Nanotechnology. The
authors thank Jean Berthier for his valued contribution.
References
1 T. M. Squires and S. R. Quake, Rev. Mod. Phys., 2005, 77,
977–1026.
2 D. J. Laser and J. G. Santiago, J. Micromech. Microeng., 2004, 14,
R35–R64.
3 M. A. Unger, H. P. Chou, T. Thorsen, A. Scherer and S. R. Quake,
Science, 2000, 288, 113–116.
4 S. Ghosal, Electrophoresis, 2004, 25, 214–228.
5 G. M. Walker and D. J. Beebe, Lab Chip, 2002, 2, 131–134.
6 C. H. Ahn, J. W. Choi, G. Beaucage, J. H. Nevin, J. B. Lee,
A. Puntambekar and J. Y. Lee, Proc. IEEE, 2004, 92, 154–173.
7 K. Hosokawa, M. Omata, K. Sato and M. Maeda, Lab Chip, 2006,
6, 236–241.
8 G. McHale, S. M. Rowan, M. I. Newton and M. K. Banerjee,
J. Phys. Chem. B, 1998, 102, 1964–1967.
9 I. Idel’Cik, Memento des pertes de charge, Collection de la
Direction des Etudes et Recherches d’Electricite de France,
Eyrolles, Paris, 1969.
10 R. K. Shah, Laminar Flow Forced Convection in Ducts, Academic
Press, New York, 1978.
11 D. C. Duffy, J. C. McDonald, O. J. A. Schueller and
G. M. Whitesides, Anal. Chem., 1998, 70, 4974–4984.
12 B. H. Jo, L. M. Van Lerberghe, K. M. Motsegood and D. J. Beebe,
J. Microelectromech. Syst., 2000, 9, 76–81.
13 N. Zhang and W. J. Yang, J. Heat Transfer, 1982, 104, 656–662.
14 S. J. Gokhale, J. L. Plawsky and P. C. Wayner, J. Colloid Interface
Sci., 2003, 259, 354–366.
15 J. Atencia and D. J. Beebe, Nature, 2005, 437, 648–655.
16 S. K. Cho, H. J. Moon and C. J. Kim, J. Microelectromech. Syst.,
2003, 12, 70–80.
This journal is ß The Royal Society of Chemistry 2007