Sensors and Actuators B 111–112 (2005) 385–389

Energy from streaming current and potential

Wouter Olthuis ∗ , Bob Schippers, Jan Eijkel, Albert van den Berg
MESA+ Institute for Nanotechnology, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands

Available online 30 April 2005

Abstract

It is investigated how much energy can be delivered by a streaming current source. A streaming current and subsequent streaming potential
originate when double layer charge is transported by hydrodynamic flow. Theory and a network model of such a source is presented and initial
experimental results are given, showing a supplied power of 20 nW obtained by a pressure difference of 1 atm over a glass plug of 60 mm in
diameter and 3.5 mm thickness with pore sizes of 1.0–1.6 ␮m, using a 1 mM KCl solution. An expression for the power transfer efficiency is
derived and discussed.
© 2005 Published by Elsevier B.V.

Keywords: Streaming potential; Zeta potential; Streaming current

1. Introduction
Recently, a paper by Yang et al. [1] about energy from
water attracted the attention of even the popular press. In
that paper it was theoretically investigated how streaming
current and potential from a water flow could be used to
create electrical energy. In this paper we elaborate on this
well-known concept [2] by giving experimentally obtained
results, and indicate ways to optimize the supplied energy.
When a solution is pressed through a porous plug (with
pores in the micrometer or nanometer range, e.g., a sin-
tered glass plug), a so-called streaming current is generated.
This stems from the fact that generally the wall material is
charged and a counter charge resides in the solution as shown
in Fig. 1a. When the mobile part of this counter charge is
dragged along by the hydrodynamic flow, Q, caused by the
pressure difference,
P, over the porous plug, a net trans-
port of charge occurs through the plug, the streaming cur-
rent Is , as depicted in Fig. 1b. Thus, charge polarization over
the plug is created, resulting in an electric field, E, causing
charge to move in the opposite direction of the streaming
current; the so-called conduction current, Ic . At a given hy-
drodynamic flow, equilibrium is reached: the net transport of
∗ Corresponding author.
E-mail address: w.olthuis@utwente.nl (W. Olthuis).
charge through the porous plug is zero. The potential differ-
ence that is present over the plug in this zero-current situation
is the (maximum) streaming potential as indicated in Fig. 1c.
The streaming current can be treated as source, which can be
connected to an external load.
In this paper, some theoretical details of the origin and
magnitude of the internal and external currents, as well as
the streaming potential are given, resulting in an equivalent
circuit. Regarding this circuit as a description of an energy
source, its optimized energy transfer to an external load is
discussed. In addition, an expression for the hydrodynamic
input power is given, enabling the calculation of the power
transfer efficiency. The results of some initial experiments,
focused on optimized transferred power to an external load,
are presented and discussed.
2. Theory
The origin of the charge that forms the basis of the stream-
ing potential stems from the chemical reactions taking place
at the interface between the wall material of the porous plug
and the aqueous solution flowing through it. Glass, for exam-
ple, will dissociate protons at an intermediate solution pH ac-
cording to SiOH ↔ SiO− + H+ . This deprotonation results in
an immobile negatively charged glass wall. This wall charge
is compensated for by positive counter charge, present in the

0925-4005/$ – see front matter © 2005 Published by Elsevier B.V.

doi:10.1016/j.snb.2005.03.039
386 W. Olthuis et al. / Sensors and Actuators B 111–112 (2005) 385–389

Fig. 2. Voltage-contolled current source network model of the streaming

current and potential.

force acting on the charges, resulting in a conduction current,

Ic , in the opposite direction (Fig. 1b):
Vs
Ic = Aσ (2)
l
where σ is the solution’s conductivity and Vs the voltage
difference over the plug. Soon, a steady state is reached and
the net transport of charge through the plug is zero: Ic + Is = 0.
The potential difference, present over the plug in that case is
the maximum streaming potential, Vs , from Eqs. (1) and (2)
(Fig. 1c):
ε0 εr
Vs =
Pζ (3)
ση
In order to investigate the energy supply of this source
to an external load, Rext , an equivalent circuit is presented,
that fits the operational principle: a voltage-controlled current
source, shown in Fig. 2.
Fig. 1. (a) Drawing of a pore with length l and cross-section A in a solid
material, filled with an electrolyte. At the pore wall, immobile, negative
It is known form network theory that the maximum power
charge is present, accompanied by partly mobile positive counter charge. transfer occurs at Rext = Rplug . In that case the power dissi-
(b) A pressure difference causes a flow Q and a transport of charge; the pated in Rext is, using Eqs. (1) and (3):
streaming current Is . (c) The charge polarization causes an E-field and thus

a conduction current, Ic . Due to the field, a streaming potential Vs is present 1
Pε0 εr ζ 2 A
Pext,max = Vs Is = (4)
over the pore. 4 η 4σl
solution as ions. This arrangement of wall charge balanced by Now that the maximum output power of the streaming
charges in the solution is referred to as the electrical double current source is known, it is interesting to compare this with
layer. A well-accepted description of this double layer is by the hydrodynamic input power in order to calculate the power
the Gouy–Chapman–Stern model, basically subdividing the transfer efficiency. The flow resistance Rh of a channel with
charge in the solution in two parts: an inner layer (the Stern cross-section A and length l (like in Fig. 1):
layer) including immobile ions that are relatively strongly 8πηl
bound to the wall surface, and an outer, or diffuse layer, con- Rh = (5)
A2
taining loosely bound mobile ions [3]. The distribution of
charge in the diffuse layer causes an electrical potential with With (laminar) flow Q through the channel, the hydrody-
respect to the bulk of the solution, having a specific value, the namic power is
zeta potential, ζ, being the potential at the plane of shear. A (
P)2 (
P)2 A2
pressure difference,
P, over the porous plug will cause a hy- Phydro =
PQ = = (6)
Rh 8πηl
drodynamic flow dragging along the mobile charge past the
shear plane. The electrical current thus created is the stream- Using Eqs. (4) and (6), the power transfer efficiency can
ing current [4–7] (Fig. 1a): be calculated:
Aε0 εr Pext,max 2π(εr ε0 ζ)2
Is =
Pζ (1) = (7)
ηl Phydro σηA
where ε0 εr and η are the dielectric constant and viscosity of It is interesting to observe that this efficiency is no longer
the solution, and l and A the pore length and cross-section a function of the pore length l as both the output power (Eq.
of all pores, respectively. The electric field created by the (4)) and the input power (Eq. (6)) scale with the same order
resulting polarization of charge along the plug axis causes a of the length l.
 W. Olthuis et al. / Sensors and Actuators B 111–112 (2005) 385–389
Optimization of the efficiency is possible by choosing a
small cross-sectional area, A, of the channel. There is, how-
ever, a lower limit determined by the fundamental operational
mechanism of streaming current, i.e., the convectional trans-
port of charge from the diffuse layer. The radius of a cylindri-
cal channel should be larger than the thickness of this diffuse
layer, as described by Debye [3]:

kTε0 εr
rD = (8)
2q2 [c]Na
for a simple 1 ÷ 1 salt, where q is the electron charge and
Na Avogadro’s number. For [c] in mol/m3 in an aqueous so-
lution at room temperature, rD equals 9.7 × 10−9 /[c]1/2 m.
Thus, the minimum cross-section A of the channel scales
with constant/[c]. Also present in the denominator of Eq. (7)
is the solution’s conductivity, σ. This conductivity scales for a
wide range linearly with the concentration: σ ÷ [c]. From this
reasoning, it is concluded that the optimal σA is not depend-
ing on [c]. It is, however, not possible to quantify the optimal
value for σA because on the one hand the minimal thickness
of the layer, past the shear plane, in which the mobile charge
that contributes to the streaming current, is present, is not pre-
cisely known. On the other hand, the distribution of charged
species over the channel radius disqualifies the use of one,
the constant conductivity σ. In addition, for channels in the
nanometer range, the value for the fluid viscosity, η, should
be adapted.
As an example, for a relatively large channel diameter of
1 ␮m and a concentration of [KCl] = 1 mM, the non-optimal
efficiency is ≈0.3% (for ζ = 0.1 V).
In the rest of this paper, some experiments dedicated to
the transfer of power to an external load are described and
initial results are presented and discussed.
3. Experimental
The set-up consisted of a 125 ml Duran Schott glass porous
filter of porosity five, indicating a nominal pore size of
1.0–1.6 ␮m. The diameter of the filter was 60 mm and its
thickness approximately 3.5 mm. The outlet of the funnel
was kept at 0.03 atm by a water jet pump. At both sides
of the glass porous plug Ag/AgCl electrodes were posi-
tioned to record the streaming potential with a PC-controlled
HP34401A multimeter. It was independently determined that
the charge transfer resistance of the Ag/AgCl electrodes to the
Table 1
Influence of [KCl] and Rext on Pext
[KCl] (M) Rext ()
10−3 1 × 1010
24 × 103
10−1 1 × 1010
1 × 104
387
Fig. 3. The measurement set-up.
electrolyte solution was approximately 250 , much lower
than the porous plug resistance and therefore, this resistance
could be neglected. The measurement set-up is schematically
shown in Fig. 3.
In order to increase the streaming potential, the pH of
the electrolyte, sucked through the plug, was kept at pH 10
with NaOH, thus increasing the zeta potential as explained
in the previous section. The electrolyte consisted of either 1
or 100 mM KCl.
4. Results
First, the influence of the electrolyte conductivity, σ, was
determined and the results are given in Table 1.
In this table, both the open circuit streaming potential (very
high Rext ) is presented and the value of Rext where Vs drops to
about half its maximum, indicating the value of Rplug . In this
situation, maximum power transfer to Rext is expected. In a
qualitative sense, all expectations are met: a lower electrolyte
concentration means a lower conductivity, σ, resulting in both
a higher (maximum) Vs and a higher supplied power to Rext ,
as expected from Eqs. (3) and (4), respectively. It is difficult
to draw quantitative conclusions both from the fact that the
nature and precise pore dimensions of the glass filter are not
known and that the zeta potential itself is also, via the diffuse
layer, a function of the electrolyte concentration. Assuming
a constant hydrodynamic input power for both tested KCl
concentrations, the ratio in efficiency, as expected from Eq.
Vs (V) Iext (A) Pext (W)
40 × 10−3 ∼0 ∼0
22 × 10−3 9.2 × 10−7 20 × 10−9
1.2 × 10−3 ∼0 ∼0
7.5 × 10−4 7.5 × 10−8 5.6 × 10−11
388 W. Olthuis et al. / Sensors and Actuators B 111–112 (2005) 385–389
Fig. 4. Streaming potential Vs and supplied power Pext as a function of Rext
for 1 mM KCl.
(7), would be 100, for the conductivity σ scales linearly with
the concentration. The experimentally found ratio is, from
Table 1: ≈350. Indeed, other variables, like the zeta potential,
ζ, in Eq. (7) are apparently also changing as a function of
[KCl].
Finally, a series of measurements was performed with dif-
ferent loads, Rext , using the lower concentration value, 1 mM
KCl. The results of both Vs and Pext are presented in Fig. 4.
These curves nicely show the behaviour of the streaming
source under load conditions, indicating a porous plug resis-
tance of about 20 k, where Vs = 1/2Vs,max and Pext indeed
obtains its maximum value (20 nW) as expected from Eq. (4).
The energy conversion efficiency from the mechanical to the
electrical domain is in this case far less than 1%, as expected
from Eq. (7) for the given pore dimensions. An even lower
electrolyte concentration through this plug will increase the
efficiency.
5. Concluding remarks
It has been shown that energy can be obtained from
a streaming current source, according to the theoretical
description, resulting in both a network model (voltage-
controlled current source) and an expression for the power
transfer efficiency. This description enables the proper de-
sign of a micromachined porous plug for optimized energy
transfer, e.g., for Lab-on-a-Chip applications. Initial experi-
mental results are in agreement with theory.
References
[1] J. Yang, F. Lu, L.W. Kostiuk, D.Y. Kwok, Electrokinetic microchannel
battery by means of electrokinetic and microfluidic phenomena, J.
Micromech. Microeng. 13 (2003) 963–970.
[2] J.F. Osterle, Electrokinetic energy conversion, J. Appl. Mech. 31
(1964) 161.
[3] A.J. Bard, L.R. Faulkner, Electrochemical Methods, Fundamentals and
Applications, Wiley, 1980.
[4] P. Fievet, M. Sbaı̈, A. Szymczyk, A. Vidonne, Determining the ζ-
potential of plane membranes from tangential streaming potential
measurements: effect of the membrane body conductance, J. Membr.
Sci. 226 (2003) 227–236.
[5] A. Yaroshchuk, V. Ribitsch, Role of channel wall conductance in
the determination of ζ-potential from electrokinetic measurements,
Langmuir 18 (2002) 2036–2038.
[6] C. Werner, R. Zimmermann, T. Kratzmüller, Streaming potential and
streaming current measurements at planar solid/liquid interfaces for
simultaneous determination of zeta potential and surface conductivity,
Colloids Surf. 192 (2001) 205–213.
[7] X. Xuan, D. Li, Analysis of electrokinetic flow in microfluidic net-
works, J. Micromech. Microeng. 14 (2004) 290–298.
Biographies
Wouter Olthuis (1960) received the MSc degree in electrical engineering
from the University of Twente, Enschede, The Netherlands in 1986 on
the subject of thermally excited resonating silicon membrane pressure
sensors. In that year, he joined the Center for MicroElectronics, Enschede
(CME) doing research on inorganic electret materials for subminiature
silicon microphones. In 1987 he started his PhD-project and received the
PhD degree from the Faculty of Electrical Engineering, University of
Twente, in 1990. The subject of his dissertation was the use of iridium
oxide in ISFET-based coulometric sensor–actuator devices. Since 1991,
he is working as an assistant professor in the Laboratory of Biosensors,
the Lab-on-a-Chip group, part of the MESA+ Research Institute, of the
University of Twente and as such co-supervising many projects on both
physical and (bio)chemical sensors and sensor systems for medical and
environmental applications.
Bob Schippers (1978) is enrolled as student at the Faculty of Electri-
cal Engineering, Mathematics and Computer Science. At the end of the
Bachelor phase, he did his individual research assignment at the Chair
of BIOS, the Lab-on-a-Chip group. The results of this assignment form
the basis of this paper. Currently, he is enrolled as master student in the
same faculty at the chair of Semiconductor Components.
Jan Eijkel (1955) studied pharmacy at the University of Amsterdam.
After completing his degree he worked in a public pharmacy in Utrecht
while studying theology at the University of Utrecht. He then went to
the University of Twente for a PhD study with Prof. Piet Bergveld on
a potentiometric protein detection. Subsequently he joined the Analytical
Centre at the Department of Chemistry at the Imperial College in London
to work with Andreas Manz. His research projects in London were the
development of a dc plasma on a chip for atomic emission detection
and the development of a circular chromatograph. He now works at the
University of Twente in the group of Albert van den Berg on nano-fluidic
research.
Albert van den Berg was born in Zaandam, The Netherlands, in 1957. He
graduated from the University of Twente, Enschede, The Netherlands, in
1983, and received the PhD degree in technical sciences on ion-sensitive
field-effect transistors (ISFETs) from the University of Twente, in 1988.
Thereafter he became project leader at the Swiss Center of Microme-
chanics (CSEM), where he was responsible for several sensor projects.
From 1991 to 1993, he was a research scientist with the University of
Neuchâtel, where he was involved with silicon-based electrochemical sen-
sors and microsystems for chemical analysis. From 1993 to 1999 he was
research coordinator of micrototal analysis systems (␮TAS) at MESA Re-
 W. Olthuis et al. / Sensors and Actuators B 111–112 (2005) 385–389 389

search Institute, University of Twente, a topic that was recently extended
to miniaturized chemical systems (MiCS). In 1998, he was appointed as
full professor on miniaturized systems for (bio)chemical analysis at the
Faculty of Electrical Engineering, Mathematics and Informatics at the
University of Twente. He is member of the ␮TAS steering committee
and editor of the section ␮TAS of Sensors and Actuators B. His current
research interests focus on theory, technologies, new devices and applica-
tions of micro- and nano-fluidics for miniaturized (bio)chemical synthesis
and analysis systems.