Sie sind auf Seite 1von 84


AC Dielectrophoresis Lab-on-Chip found in his classic text [1]. The advancement of

Devices microfabrication techniques and the demand of
Lab-on-a-Chip technologies have led to the
Stuart Williams development of dielectrophoresis techniques for
Mechanical Engineering, Purdue University, particulate, biological, and pharmaceutical appli-
West Lafayette, IN, USA cations. Dielectrophoresis was initially used to
manipulate particles and cells in the micrometer
range (1 mm to 1 mm). Since the early 1990s,
Synonyms nanotechnology has incorporated dielectro-
phoresis for the manipulation of viruses, DNA,
DEP; Dielectrophoretic force protein molecules, and other nanoparticles
(diameters of 1 nm to 1 mm). This essay provides
a brief background of dielectrophoresis followed
Definition by its basic manipulation of particles using
translation, rotation (electrorotation), orientation
Dielectrophoresis is the translational motion of (electro-orientation), and traveling wave
a neutral particle by induced polarization in dielectrophoresis. This entry approaches design
a nonuniform electric field. The magnitude and considerations and modeling techniques for the
direction of the induced dielectrophoretic force micrometer length scale; it does not specifically
are dependent on the characteristics of the address all dielectrophoresis complexities at the
applied electric field as well as the dielectric nanometer scale. The majority of this entry deals
properties of the surrounding medium and of the with popular applications of dielectrophoresis
particle itself. including novel techniques to induce these
forces; by no means does it cover all of the
existing applications. For a more extensive expla-
Overview nation of dielectrophoresis, the reader is referred
to texts by Jones [2], Morgan and Green [3], and
Herbert Pohl was one of the first to study particle Hughes [4], the latter of which addresses tech-
electrokinetics in the 1950s, particularly the niques for nanometer-sized particles.
manipulation of polarizable particles with For most dielectrophoresis cases, the applied
nonuniform fields. He coined the term dielectro- electric field is an alternating current (AC) signal,
phoresis, and details of his investigations can be created with a common frequency generator.

D. Li (ed.), Encyclopedia of Microfluidics and Nanofluidics,

DOI 10.1007/978-1-4614-5491-5, # Springer Science+Business Media New York 2015
A 2 AC Dielectrophoresis Lab-on-Chip Devices

In typical experimentation, frequencies are gen-

erally greater than 100 kHz with magnitudes
below 20 V peak to peak. The shape of the signal
is typically sinusoidal, but pulse signals have also
been used in dielectrophoresis applications. This
signal is applied to electrode geometries, the
micrometer-sized features of which can be cre-
ated using simple microfabrication techniques.
Electrodes are typically fabricated on the surface
of silicon wafers or glass substrates, including
microscope slides. For most biological applica-
tions, it is recommended that the species of inter-
est be visually observed by optical means, which
leaves glass or other transparent materials as
favorable substrates. In the case where the elec-
AC Dielectrophoresis Lab-on-Chip Devices, Fig. 1 A
trodes themselves hinder visual observation, they
polarizable particle in a nonuniform electric field
are patterned out of indium tin oxide (ITO),
a transparent conducting material. Microfluidic
channels and other fabricated features can easily parameters, typical dielectrophoretic manipula-
be incorporated into the system. It is possible to tion techniques, particle modeling consider-
manipulate, separate, or group targeted cells with ations, and dielectrophoresis-induced effects are
novel electrode geometry design and fabrication. discussed below.
Due to its simplicity in fabrication and its suscep-
tibility to visual observation and analysis, Dielectrophoresis
dielectrophoresis is a favorable technique for bio- Dielectrophoresis is the translational motion
logical experimentation. of a particle by induced polarization in a
Dielectrophoretic forces, though, can be nonuniform electric field. When conductive,
induced by means other than an applied electric uncharged particles are exposed to an electric
signal through electrodes. Optical tools can be field, they will polarize, inducing a dipole on the
implemented to modify an applied electric field, particle. The magnitude and charge orientation of
making these methods more susceptible for the induced dipole are dependent on the permit-
dynamic as opposed to static manipulation of elec- tivities and conductivities of the medium and of
tric fields with surface electrodes. Dielectro- the particle. If the electric field is uniform, the
phoresis applications are not limited to particulate induced charges on both sides of the particle are
manipulation either. With properly configured equal, creating no net force on the particle. How-
surface-electrode geometry, it is possible to induce ever, if the electric field is nonuniform (Fig. 1),
fluid motion and create nanoliter-sized droplets. there will be a net force greater than zero. The
Additionally, dielectrophoretic forces can be uti- general expression for the dielectrophoretic force
lized to manipulate particles to build micro- and of a homogeneous sphere is expressed as
nanostructures such as wires.
FDEP 2pe0 em r 3 ReK oE2 (1)

Basic Methodology where e0 is the permittivity of free space, em is the

relative permittivity of the medium, r is the radius
Before we can incorporate dielectrophoresis of the particle, Re[K(o)] is the real part of the
into Lab-on-a-Chip systems, it is important to Clausius-Mossotti factor, and E2 is the gradient
have a grasp of the theory behind dielectro- of the magnitude of the electric field squared.
phoresis. Basic theory, common experimental This equation assumes that the there is no applied
AC Dielectrophoresis Lab-on-Chip Devices 3 A
AC Dielectrophoresis
Lab-on-Chip Devices,
Fig. 2 Dielectrophoretic
manipulation techniques: A
(a) electrorotation,
(b) electro-orientation,
(c) particle trapping, and
(d) traveling wave

phase gradients to the electric field. The Clausius- For a positive value of Re[K(o)], the particle is
Mossotti factor is defined as attracted to high electric field gradients with pos-
itive dielectrophoresis (pDEP). With dielectro-
ep  em phoresis it is possible to manipulate particles
K o (2)
ep 2em using a variety of techniques and applications.

where subscripts p and m are assigned to the Particle Manipulation

particle and medium, respectively, and e is The majority of dielectrophoretic manipulation
called the complex permittivity. The complex of particles includes translating
permittivity is (dielectrophoresis), rotating (electrorotation),
orienting (electro-orientation), trapping, and
s using traveling wave dielectrophoresis. Figure 2
e e0 er  j (3) provides an illustrative representation of each
p technique. In electrorotation a torque is applied
where j is 1 , er is the materials relative to a particle that is subjected to a rotating electric
permittivity, s is the materials conductivity, field (Fig. 2a). The induced dipole takes a finite
and o = 2pf (where f is the applied frequency). amount of time to polarize in a neutral dielectric
The Clausius-Mossotti factor is a function of particle, which attempts to orient itself with the
frequency, and, depending on the dielectric prop- direction of the electric field. This dipole, though,
erties of the medium and particle, this factor can lags behind the applied rotating electric field. The
be either positive or negative with a possible reorientation of the dipole with the electric field
range of +1.0 to 0.5. If Re[K(o)] is negative, induces a torque on the particle, rotating it. Next,
the particle experiences negative dielectro- electro-orientation involves the alignment of
phoresis (nDEP) and is repelled from gradients a nonspherical particle in a uniform electric
in the nonuniform electric field (Fig. 1). field (Fig. 2b). When an ellipsoidal particle
A 4 AC Dielectrophoresis Lab-on-Chip Devices

polarizes, the dipole moment will align the parti- previously mentioned dielectrophoretic manipu-
cle with its longest nondispersed dipole parallel lation techniques can be applied to determine
to the field lines. Its orientation is a function of particle dielectric characteristics. By varying
the electric field frequency and the dielectric both the medium conductivity and the applied
properties of the medium and particle. frequency, the changes in a particles induced
Dielectrophoretic forces, though, can be used dielectrophoretic behavior can be visually
to not only rotate a particle, but trap it as well. observed.
There are two types of particle trapping, those The electric field can induce higher order
that utilize pDEP or nDEP forces. For example, poles, called multipoles, instead of the assumed
four electrodes can be positioned in a quadrupole dipole. This occurs when the electric field is
arrangement and, when the appropriate electric highly nonuniform, for example, when the elec-
field is applied, a particle or particles are trapped trode geometry is on the same length scale as the
in the electric field null at its center (Fig. 2c). particle or when a particle is in a field null. For
Particle trapping will be revisited later in this most applications, though, the dipole assumption
entry. Traveling wave dielectrophoresis is the accurately depicts the behavior of the particle.
linear application of electrorotation (Fig. 2d). However, multipoles can effect the particles
An AC electric wave is produced by applying interaction with the electric field and will react
an electric field that travels linearly along differently from an induced dipole. The electric
a series of electrodes. The particle will translate field can also induce some negative physiological
in the same or opposite direction as the traveling effects to the cell by inducing cell heating and
wave depending on the properties of the applied influencing transmembrane voltage. Transmem-
signal frequency and the dielectrics of the particle brane voltage can affect ion movement and dam-
and medium. age voltage-sensitive proteins. The electric field
itself can heat the surrounding medium in what is
Modeling and Dielectrophoretic Effects called Joule heating. Joule heating creates tem-
When incorporating dielectrophoresis in Lab-on- perature gradients that can directly heat the cell or
a-Chip systems, additional modeling parameters create localized convection currents. By working
and electro-physiological interactions need to be at the micrometer length scale, some of these
considered. Obviously, biological cells and some negative effects are minimized; however, they
particulates are not all completely spherical nor may still influence the system and cannot be
are they homogeneous. Typically cells are ignored.
modeled as multi-shelled particles with each
shell having its own respective conductivity and
permittivity. Additionally, nonspherical particles Key Research Findings
are modeled as ellipsoids. For each of these situ-
ations, the Clausius-Mossotti factor is extended This dielectrophoresis background only serves as
to include these geometrical and layered effects a brief overview; these technologies need to be
by applying polarization factors. These modifica- explored in more detail before incorporating
tions are described in detail elsewhere [24]. The them into a Lab-on-a-Chip system. Lab-on-a-
dielectric properties of some particles and cells Chip systems integrate techniques of small fluid
are unknown; however, dielectric techniques can and sample handling with detection or process
be applied to determine these characteristics. For capabilities. Dielectrophoresis can be incorpo-
example, when an induced particle changes from rated into these systems to manipulate, separate,
pDEP to nDEP, or vice versa, this is called its or trap cells as well as control small amounts of
crossover frequency. Similarly, a particle in fluid. This technology can be used to trap cells for
electrorotation that changes in rotational velocity additional analysis, separate cell types based on
or direction under different conditions will give dielectric properties, dispense picoliter droplets,
insight into its dielectric properties. Many of the or used for similar manipulative applications.
AC Dielectrophoresis Lab-on-Chip Devices 5 A
Electric Potential
dielectrophoretic trapping force can be estimated
with the hydrodynamic drag necessary to release
the particle. Similarly, interdigitated electrodes A
can be used to separate two cells that experience
particle particle
opposite pDEP/nDEP forces due to their different
V V dielectric characteristics. Interdigitated electrodes
are one of the simpler electrode geometries, com-
AC Dielectrophoresis Lab-on-Chip Devices, pared to other trapping arrangements.
Fig. 3 Dielectrophoresis of particles with interdigitated
Other electrode geometries have been
electrodes (side view). Particles experiencing pDEP are
attracted to the edges of the electrode, while nDEP forces designed to capture individual cells using either
repel the particle to the middle of the electrode pDEP or nDEP forces. A number of these geom-
etries have been explored by Voldman and his
research group [5] and will be discussed here. For
However, dielectrophoresis itself cannot be used pDEP, geometries are created that generate
as a sensor, except to determine dielectric prop- extreme electric field gradients in the trapping
erties of cells or detect such changes in response region. A geometry that creates such a field gra-
to stimuli. The integration of sensory technolo- dient is illustrated in Fig. 4a. An electrode smaller
gies with dielectrophoresis is not discussed. than the diameter of the particle has the signal
Instead, specific novel dielectrophoresis applica- applied through it to a larger, receiving electrode.
tions are addressed in detail. The generated electric field would be similar to
that illustrated in Fig. 1, with a high gradient in
Trapping Particles proximity to the smaller electrode. Electric field
One of the more popular dielectrophoretic manip- nulls have also been generated with electrode
ulation techniques involves the trapping of indi- geometries for nDEP traps. Recall the quadrupole
vidual or groups of cells. Dielectrophoresis can be geometry in Fig. 2c. When an appropriate electric
thought of as electrical tweezers that will grab field is applied, a field null is created in the center
onto and position cells. Trapping can be accom- of the trap. The trapped particle is repelled from
plished by means of novel electrode geometries the surrounding regions of field gradients.
that utilize pDEP or nDEP forces to selectively Another nDEP trap is illustrated in Fig. 4b. The
capture a particle or cell. Some electrode geome- electric field is applied from the trapping elec-
tries induce both types of forces simultaneously; trode to the receiving electrode strip. A particle is
an example of such an arrangement involves inter- trapped within the generated null inside the
digitated electrodes (Fig. 3). Planar electrode fin- boxed electrode geometry. These previous
gers, whose width and gap separation are a few geometries will trap a particle near the surface
times larger than the diameter of the particles of of the substrate. Electrodes, though, can be
interest, have alternating applied voltages (either arranged three dimensionally to create traps that
180 out of phase or alternating + V/ground). This capture particles and cells in suspension. One
geometry creates a strong pDEP force at the edges example is a nDEP trap using two stacked quad-
of the electrodes and a field null above the center rupole geometries, resulting in an octopole trap
of each electrode strip. This arrangement provides (Fig. 4c). The electric field null is generated in the
both trapping capabilities and the ability to deter- midplane between these two electrode substrates,
mine unknown particle dielectric properties from and thus, a particle experiencing nDEP is cap-
visual observation. For example, at a particular tured in suspension. These are just a few exam-
frequency and medium conductivity, a particle ples of established trapping geometries, as others
will experience a pDEP force. However, for could be designed to suit a particular trapping
a different applied frequency and/or a change in application.
medium conductivity, the same particle could There are numerous trap geometries that exist
undergo nDEP. The strength of the for both pDEP and nDEP. These electrode
A 6 AC Dielectrophoresis Lab-on-Chip Devices

AC Dielectrophoresis
Lab-on-Chip Devices,
Fig. 4 Various electrode
geometries for
dielectrophoretic traps:
(a) pDEN trap (side view),
(b) nDEP planar trap
(top view), and (c) nDEP
octopolar trap

arrangements and cell concentration can be


varied to trap both single cells and groups of

cells alike. However, there are dielectrophoretic
interactions that cannot be ignored for trapping

applications. Multipoles may need to be consid-
ered for single-particle traps since the generated
field will have inhomogeneities on the scale of
Flow Direction
the particle itself. Also, particle-particle
dielectrophoretic interactions will influence the
AC Dielectrophoresis Lab-on-Chip Devices,
capabilities of multi-particulate traps. Dielectro-
Fig. 5 An illustration of a dielectrophoretic microsystem
phoresis, therefore, needs to be accurately that can selectively sort particles (top view). Recall that
modeled and thoroughly investigated prior to the electrodes are on the top and bottom surfaces of the
use in a specific application. microchannel and to not mechanically manipulate the

Electrode-Based Systems
Microsystems have incorporated dielectro- selective biological investigations. An example
phoretic forces into microchannels to manipulate of a dielectrophoresis system and a description of
particles and cells with various electrode geome- its potential for biological analysis are explained
tries. Electrodes can be used to align target cells in [6]. These dielectrophoresis techniques can be
or separate particulate cells from the rest of the coupled with a variety of existing biological tools
sample, an example of which is illustrated in for innovative applications including fluores-
Fig. 5. These systems use a three-dimensional cence evaluation of cells in small populations,
arrangement of electrodes, with mirrored elec- cell sorting, long-term investigations of single
trode geometries aligned and separated by cells, cellular kinetics, and other similar
microfluidic structures. Thus the electrodes in Lab-on-a-Chip analyses.
Fig. 5 are on the top and bottom surfaces of the
microchannel and do not significantly impede or Insulating Dielectrophoresis
manipulate the fluid flow. This system is opti- Conventionally, the nonuniform electric fields
mized for a particular set of cells such that the for dielectrophoresis applications are produced
applied signals generate the strongest dielectro- directly with an arrangement of conductive elec-
phoretic forces; this system will operate with the trodes. However, a nonuniform electric field can
maximum possible applied flow rate for more be generated in microchannels without direct
effective processing. The system, though, will interaction with electrodes. For example, Cum-
be selective to a set of cells with the same or mings generated nonuniform electric fields using
similar dielectric properties. This process is obvi- remote electrodes placed at the inlet and outlet of
ously advantageous for an application that selec- the microchannel [7]. The electric field is applied
tively separates cells based on their dielectric across the channel and is deformed by the varia-
properties. Additionally, this system can simulta- tions in the channel geometry or by insulating
neously separate and trap targeted cells for structures or posts within the microchannel.
AC Dielectrophoresis Lab-on-Chip Devices 7 A
These mechanical structures within the channel Illumination
will affect the generated electric field as well as Source
dynamically change the flow field; these can be A
successfully coupled to produce an efficient par-
ticle concentrator within the channel. By using
remote electrodes, electrochemical reactions and Glass/ITO
other unwanted dielectrophoretic effects will not Photoconducter +++ i
occur within the section of particle collection or g
concentration. However, this technique relies on n
the application of higher voltages to produce the Channel a
necessary electric field gradient, with magnitudes l
in the hundreds to thousands of volts (outside the
range of conventional waveform generators).
AC Dielectrophoresis Lab-on-Chip Devices,
The driving mechanism for fluid flow for these Fig. 6 An example for a simple setup for optically
systems is electrokinetic in that it combines fea- induced dielectrophoresis. The optical source is illumi-
tures of electroosmotic and electrophoretic flows. nated through the ITO/glass substrate onto the photocon-
Insulating dielectrophoresis Lab-on-a-Chip sys- ductive layer, creating an electric field gradient in which
particles experiencing pDEP are attracted
tems use a coupled arrangement of electrokinetic
and dielectrophoretic phenomena to separate or
capture targeted particles. Consider such the photoconductive layers have higher conduc-
a dielectrophoresis system that has an array of tivities under illumination, thus distorting the
insulating posts within a microchannel. At a low otherwise uniform electric field. An illustration
applied electric field, there will be induced elec- of a simplified optically induced dielectro-
trokinetic motion but minimal particle manipula- phoresis system is shown in Fig. 6. The illumina-
tion effects from dielectrophoresis. However, at tion of the photoconductive layer produces
higher magnitudes the influence of insulating a virtual electrode and locally induced dielectro-
posts becomes apparent, inducing dielectro- phoretic forces. These forces can be used in
phoretic forces on the particles. At a certain pDEP or nDEP applications and have the distinct
threshold in the applied electric field, dielectro- advantage of being temporary, as opposed to
phoresis forces on the particles are greater than permanent microfabricated electrode features.
the electrokinetic forces, resulting in captured For example, trapped particles can be moved
targeted particles at the insulating posts. Releas- with the translation of the illumination pattern
ing these particles is just a simple decrease in and released by simply turning off the light.
signal magnitude. Structures other than posts These optical methods, though, have been
have been successfully used to similarly manip- developed to include image projectors to manip-
ulate and separate particles of different dielectric ulate cells using larger, patterned illumination
properties. areas. For example, virtual structures can be
used to direct cells or can be alternatively used
Optical Dielectrophoresis to pattern trapped cells. An investigation of opti-
An electrode-less approach to induce localized cally induced dielectrophoresis for cellular
dielectrophoretic forces can be created by light manipulation is found in detail elsewhere
illumination in optically induced dielectro- [8]. This investigation characterizes the control-
phoresis. This technique utilizes low-power opti- lability of these optical techniques. This tech-
cal beams to manipulate particles and cells. nique, though, should not be confused with
Consider an AC electric signal applied across optical tweezers (also called optical traps),
two parallel electrodes with a film of photocon- which manipulate cells with forces generated by
ductive layer in between. This generates a highly focused laser beam. Optical tweezers use
a uniform electric field across the film; however, high optical intensity that may cause damage to
A 8 AC Dielectrophoresis Lab-on-Chip Devices

the cell or induce localized heating. In contrast, S

optically induced dielectrophoretic techniques i
have about 1,000 times less optical intensity. g
Liquid Manipulation
Outside of particle manipulation, dielectro- Electrode
phoresis has also been used to control small vol- Liquid
umes of liquid [9]. Liquids themselves also Droplet Manipulated
(source) Liquid
polarize and will, therefore, respond to
nonuniform electric fields by being attracted to Insulating
high-field-intensity regions. Liquid dielectro- Layer
phoresis influences hydrostatic equilibrium and
does not directly manipulate the fluid using
dynamic mechanisms typically associated with
electroosmotic flow, electroconvection, or other
electrohydrodynamic methods. As with all
AC Dielectrophoresis Lab-on-Chip Devices,
dielectrophoresis techniques, liquid dielectro- Fig. 7 Liquid dielectrophoresis with a set of planar elec-
phoretic forces become stronger at smaller length trode fingers with an insulating layer (top view and side
scales, typically the micrometer length scale. view). A signal is applied to manipulate the liquid to travel
along the gap between the electrodes
Simple applications of this technique require
a fluid running between two closely spaced planar
electrodes (Fig. 7). A signal is applied between Molecular Dielectrophoresis
these two electrodes inducing the collection of Dielectrophoresis has also been used to manipu-
individual molecules to the area of highest field late macromolecules such as DNA, viruses, pro-
intensity, along the edges of the electrodes. The teins, and carbon nanotubes. The term colloids
liquid will translate and fill the area along these will be used here to generally describe a particle
electrode fingers until equilibrium is achieved. between 1 and 1,000 nm. At this scale we need to
When the signal is removed, the liquid will take into consideration additional parameters that
retract and return to its original hydrostatic equi- will affect the efficiency and application of
librium form. This retraction can be coupled dielectrophoresis. The first is Brownian motion,
together with surface structures to generate or the random chaotic movement of molecules,
nanoliter droplets. However, typical voltage which will introduce another destabilizing vari-
magnitudes are in the hundreds of volts, resulting able if we were to trap colloids. Second, electro-
in Joule heating and electrolysis. Electrolysis can static effects at the surface of colloids, created by
be easily be avoided by coating the electrode the electrical double layer, will influence particle-
surface with an insulating layer. Heating is particle interactions. Factors such as hydrody-
typically insignificant when liquids are namic drag, buoyancy, electrothermal effects,
nonconductive; however, water and other highly and a particles double layer interactions need to
conductive liquids will undergo Joule heating. To be considered when applying dielectrophoresis to
cope with these issues in present liquid dielectro- colloids.
phoresis systems, very short applications of volt- The importance of mechanically controlling
age (<0.1 s) are applied to avoid heating. Liquid these structures in Lab-on-a-Chip systems with
dielectrophoresis involves a complex relation- dielectrophoretic techniques will aid in the
ship between electrohydrodynamics, fluid understanding of their biological interactions
dynamics, surface effects, Joule heating, and with other subcellular entities. For example, min-
dielectrophoresis, but this technique provides iature quadrupole traps can be used to determine
a rapid actuation of liquid that can be utilized in the crossover frequency of specific proteins and
Lab-on-a-Chip systems. viruses. Capturing viruses at a particular set of
AC Dielectrophoresis Lab-on-Chip Devices 9 A
experimental parameters will give insight into Pearl chains of nanoparticles can be arranged
their identification as well as provide a means to and fused together to create a nanowire or other
concentrate or separate these particles. The cross- similar structures. However, other dielectro- A
over frequency of a protein is not only a function phoresis techniques can be utilized to assemble
of medium conductivity, but it is dependent on a variety of geometries. Dielectrophoretic traps
the pH as well. Thus, dielectrophoresis can be and optical dielectrophoresis can assemble groups
coupled with chemistry-oriented techniques in of like particles. Electro-orientation techniques
order to approach the experimentation of single have been used to align and connect nanowires
molecules. DNA has also been manipulated using between electrodes. Functionalized particles can
dielectrophoresis. In one study [10], DNA was be implemented with dielectrophoresis Lab-
uncoiled from its natural bundled condition and on-a-Chip systems to create biological sensory or
stretched between two electrodes, aligning itself assembly systems. Dielectrophoresis, therefore, is
along the field lines. It was shown that it took a very versatile engineering tool.
about one-third of 1 s to uncoil a 48,500 base pair
DNA to a length of 17 mm. The applications of
DNA manipulation include electrical measure- Future Directions for Research
ments between electrodes and positioning of
DNA for structural purposes or modification. As a nonmechanical and minimally invasive pro-
Carbon nanotubes have also been manipulated cess, dielectrophoresis presents as a tool to be used
similarly. These manipulative techniques have in Lab-on-a-Chip systems to manipulate particles
contributed to the assembly of nanometer-sized and cells. These techniques will continue to evolve
structures. as the size of microfabricated features continue to
decrease past the nanometer scale. This miniatur-
Dielectrophoretic Assembly ization will give rise into more comprehensive,
Dielectrophoresis can be utilized in assembly subcellular microsystems which need novel
procedures to induce particle-particle interac- methods to manipulate specific entities to suit
tions or to use attractive forces between elec- a biological, chemical, or sensory purpose.
trodes to position a component and complete Dielectrophoresis phenomena at these decreasing
a circuit. When two polarized particles come length scales will continue to be characterized,
into close proximity to each other, they will leading to newly developed dielectrophoresis phe-
undergo an attractive force due to their dipole nomena and subsequent innovative manipulation
interactions. This is referred to as dipole-dipole techniques. The future of dielectrophoresis is with
interaction, mutual dielectrophoresis, or pearl the continued investigation and development of
chaining because this phenomenon creates novel applications of dielectrophoresis techniques.
strings of particles. Recall that a particles With the integration of these technologies,
induced dipole aligns itself to the electric field a complete dielectrophoresis-driven Lab-on-a-
and that like particles will always have the same Chip diagnostics system is possible.
dipole orientation. These particles will have an
attractive force since their opposite charges are
aligned facing each other. Additionally, these Cross-References
particle chain formations can also be attributed
to the distorted electric field caused by the parti- AC Electro-Osmotic Flow
cles induced dipole. These field disturbances can Dielectrophoresis
cause a localized dielectrophoretic force, increas- Dielectrophoretic Motion of Particles and
ing the strength of these particle-particle interac- Cells
tions. Pearl chains are typically observed near Electrokinetic Motion of Cells and
electrode edges where the strength of the electric Nonpolarizable Particles
field is the greatest. Electroosmotic Flow (DC)
A 10 AC Electro-Osmotic Flow

Electrophoresis Definition
Electrothermal Effects
Joule Heating in Electrokinetic Flow: AC electroosmosis (ACEO) is a nonlinear elec-
Theoretical Models trokinetic phenomenon of induced-charge elec-
Lab-on-a-Chip Devices for Particle and Cell troosmotic flow around electrodes applying an
Separation alternating voltage.
Lab-on-a-Chip (General Philosophy)

Classical electrokinetic phenomena, such as elec-
1. Pohl HA (1978) Dielectrophoresis. Cambridge Uni- troosmotic flow and electrophoresis, are linear in
versity Press, Cambridge
2. Jones TB (1995) Electromechanics of particles. Cam-
the applied voltage and thus cannot produce any
bridge University Press, Cambridge net flow under alternating current
3. Morgan H, Green N (2003) AC electrokinetics: col- (AC) conditions. A variety of nonlinear electro-
loids and nanoparticles. Research Studies Press, kinetic phenomena, which persist in AC fields,
have been known for decades in colloid science,
4. Hughes MP (2003) Nanoelectromechanics in engi-
neering and biology. CRC Press, Boca Raton but the focus has been on electrophoretic mobil-
5. Rosenthal A, Taff BM, Voldman J (2006) Quantita- ity and particle interactions. The advent of
tive modeling of dielectrophoretic traps. Lab Chip microfluidics has stimulated interest in the use
of electric fields to drive fluid flows, without
6. Muller T, Pfennig A, Klein P, Gradl G, Jager M,
Schnelle T (2003) The potential of dielectrophoresis any moving parts. In this context, nonlinear elec-
for single-cell experiments. IEEE Eng Med Biol Mag trokinetics offers some unique advantages, such
22:5161 as the reduction of unwanted electrochemical
7. Cummings EB (2003) Streaming dielectrophoresis
for continuous-flow microfluidic devices. IEEE Eng
reactions (using AC voltages) and the ability to
Med Biol Mag 22:7584 drive fast, programmable flows at low voltages
8. Chiou PY, Ohta AT, Wu MC (2005) Massively par- (using closely spaced microelectrodes).
allel manipulation of single cells and microparticles In the late 1990s, Ramos et al. discovered
using optical images. Nature 436:370372
steady electroosmotic flow over a pair of micro-
9. Jones TB (2001) Liquid dielectrophoresis on the
microscale. J Electrost 5152:290299 electrodes applying an AC voltage and dubbed
10. Washizu M, Kurosawa O (1990) Electrostatic manip- the effect AC electroosmosis [1]. Around the
ulation of DNA in microfabricated structures. IEEE same time, Ajdari predicted ACEO flow over
Trans Ind Appl 26:11651172
periodic electrode arrays and showed how the
effect could be used for long-range pumping
[2]. As the performance of ACEO pumps has
advanced [3, 4], ACEO has also been exploited,
AC Electro-Osmotic Flow in conjunction with dielectrophoresis (DEP), in
different geometries to manipulate particles and
Martin Z. Bazant cells in microfluidic devices [57].
Department of Mathematics and Institute of
Soldier Nanotechnologies, Massachusetts
Institute of Technology, Cambridge, MA, USA Basic Methodology

Local Flow Generation

Synonyms ACEO is a phenomenon of induced-charge elec-
troosmosis (ICEO), where flow is generated by
AC pumping of liquids; Traveling-wave the action of an electric field on its own induced
electroosmosis diffuse charge near a polarizable surface.
AC Electro-Osmotic Flow 11 A

Current tubes, width D z

Double layer

Glass substrate Dz

Electrodes Distance, z

AC Electro-Osmotic Flow, Fig. 1 Equivalent RC circuit resistance (due to shorter current tubes) and thus charge
model for double-layer charging over a pair of electrodes. more quickly than the outer edges (Reproduced from [1])
The inner edges of the electrodes encounter less bulk

The main difference with other examples of charges would both change sign, resulting in the
ICEO, such as flows around metal colloids, is same ICEO flow.
that ACEO involves electrode surfaces, which Under AC forcing, the flow peaks when the
supply both the electric field and the induced oscillation period is comparable to the charging
screening charge, in different regions at different time (Fig. 2b). ACEO flow decays at higher fre-
times. For this reason, ACEO is inherently time- quencies, since there is not enough time for
dependent (as the name implies) and tied to the charge relaxation (Fig. 2a). It also decays at
dynamics of diffuse charge, as ions move to lower frequencies, since there is enough time to
screen the electrodes. completely screen the bulk electric field (Fig. 2c).
Perhaps the easiest way to understand ACEO Mathematical models of ACEO follow other
is to consider a pair of planar electrodes applying examples of ICEO, as described in the article on
a sudden DC voltage (which is analogous to nonlinear electrokinetic phenomena. A major
ICEO flow around a polarizable particle in simplification in the case of small voltages is to
a sudden electric field). As shown in Fig. 1, assume sinusoidal response to sinusoidal AC
charge relaxation can initially be described by forcing and solve only for the complex ampli-
an equivalent RC circuit, where the diffuse layers tudes of the potential and velocity components
act as capacitors, connected to current-tube at a single frequency o (Fourier mode) [2]. In this
resistors of varying length through the bulk solu- regime, the basic scaling of time-averaged ACEO
tion. Since the resistance is smaller (and the field flow is
larger) near the gap, the inner portions of double
layers on the electrodes charge more quickly than eV 2
the outer portions. As shown in Fig. 2, this causes <u> / h i2 (1)
ICEO flow to occur, directed outward from the 1 dL ooc ooc
gap, only when the electrodes are partially
screened, as the tangential field from the where V is the applied voltage, e and Z are the
unscreened outer portions acts on induced charge permittivity and viscosity of the liquid (both
on the inner portions. Note that the flow is inde- assumed constant), L is electrode spacing
pendent of the sign of the applied voltage: If the (roughly center to center), and d is the ratio of
polarity were reversed, then the field and induced the diffuse-layer to compact-layer capacitances
A 12 AC Electro-Osmotic Flow

AC Electro-Osmotic Flow, Fig. 2 The basic mechanism where the initial normal current is strongest and causes
of AC electroosmosis: electrochemical relaxation (top) the unscreened field lines to dip down and provide tangen-
and induced-charge electroosmotic flow (bottom) in tial components over the induced charge; the result is
response to a suddenly applied voltage across an electrode ICEO flow directed away from the electrode gap. (c)
pair. (a) At first the electric field has no tangential com- After the charging time passes, the electrodes are fully
ponent on the electrodes, since they are equipotential screened, leaving no electric field and thus no flow. An AC
surfaces, and thus there is no electroosmotic flow. (b) voltage can drive a steady time-averaged flow, similar to
Capacitive double-layer charging begins near the gap (b), if its period is comparable to the charging time

(both assumed constant). The peak frequency is [2, 8]. Although reactions surely play a role in
at the scale of the RC charging time ACEO, high-frequency flow reversal may be
attributable to other effects, such as ion crowding
D1 d in the diffuse layer [9].
oc / (2)
Fluid Pumping
where l is the Debye screening length and D is Regardless of the details of flow generation, there
a characteristic ionic diffusivity. are some useful general principles to guide the
The relevant case of a large applied voltage, development of microfluidic pumps. ACEO flow
V  kT/e = 25 mV, is not yet fully understood. over a small set of electrodes can be used for local
A variety of new effects could become important, fluid mixing or particle trapping at stagnation
such as steric effects of finite molecular sizes, points, but the flow decays quickly away from the
viscoelectric effects, discrete electrostatic corre- electrode surfaces. A symmetric, periodic array of
lations, dissociation and de-solvation kinetics, many interdigitated electrodes (of alternating
and/or Faradaic electrochemical reactions pass- polarity at each moment in time) similarly pro-
ing current through the surface. In particular, duces an array of counter-rotating convection
Faradaic charging (as opposed to capacitive rolls, but no net pumping of the fluid in one direc-
charging, described above) has been proposed tion. Instead, long-range pumping over an elec-
to explain experimentally observed flow reversal trode array requires broken symmetry within each
at high voltage [7], although this has not yet been spatial period to rectify the AC forcing.
borne out in a complete mathematical model of There are several ways to design ACEO
ACEO. Standard models of Faradaic reactions pumps by breaking symmetry in a periodic elec-
mainly predict suppression of the flow trode array. Ajdari originally suggested modulat-
(by short-circuiting of the double-layer charging ing either the electrode capacitance via
process) and flow reversal only in certain cases, a dielectric coating (Fig. 3a) or the surface height
at lower frequencies than in the experiments (Fig. 3b) with half the spatial period of the array,
AC Electro-Osmotic Flow 13 A
x x
y y

AC Electro-Osmotic Flow, Fig. 3 Sketches of local (a) nonuniform surface coatings; (b) nonuniform surface
broken symmetries in a periodic electrode array which height (Reproduced from [2])
lead to global time-averaged ACEO pumping:

so that one side of each electrode drives stronger voltage. At low frequency, a similar induced-
ACEO flow compared to the other side and charge electroosmotic mechanism, which peaks
thus wins to produce net pumping over the at the RC frequency (2), is responsible for the
array [2]. In the first implementation of an flow [12]. At high frequency (or with a thick
ACEO pump, Brown, Smith, and Rennie opted dielectric coating on the electrodes), the classical
instead to break symmetry by using planar elec- Ehrlich-Melcher effect used to pump dielectric
trodes of different widths and gaps [10], and, until liquids, which peaks at the Debye frequency,
recently, this design was the only one studied s/e = D/l2, can also be observed [13]. Although
experimentally [3] or theoretically [8]. traveling-wave ACEO seems to produce slower
The performance of ACEO pumps can be flow than standing-wave ACEO with planar elec-
greatly enhanced by designing appropriate trodes, the possibility of designing suitable
non-planar electrode geometries. As recently non-planar electrodes has not yet been
predicted by Bazant and Ben [11], various 3D considered.
ACEO designs exhibit dramatically increased
flow rate without flow reversal, due to a special
geometry in which the nonuniform slip profile on Key Research Findings
the electrodes all contributes to flow in the same
direction. The basic idea is to create a fluid con- The groups of Ramos [1] and Green [15] have
veyor belt with electrodes, each having steps of pioneered the study of ACEO by developing the-
two different heights: On each electrode, the oretical models and measuring flows around pairs
region of desired forward flow is raised up, of microelectrodes in dilute KCl. Their work has
while the region of reverse flow is recessed ruled out electrothermal effects (below 5 V in
below, so as to recirculate in a vortex aiding the dilute electrolytes) and clarified the capacitive
forward flow (rather than fighting it, as in planar charging mechanism described above. The basic
designs). This can be accomplished with elec- model successfully predicts the scaling of ACEO
trodes having electroplated metal steps, as flow at low voltage and low salt concentration,
shown in Fig. 4, although other designs are pos- although it tends to overestimate velocities.
sible, such as flat electrode steps deposited on Following Ajdari [2], there has been extensive
a grooved surface (without the vertical metal experimental [3, 4, 10, 12, 13] and theoretical
surfaces). Simulations predict that 3D ACEO [8, 9, 11] work on ACEO flows driven by periodic
pumps are faster than planar pumps by more interdigitated electrode arrays. Velocities
than an order of magnitude, at the same voltage exceeding 100 mm/s have been demonstrated in
and minimum feature size, and thus can achieve water [4, 10] and dilute KCl [3] by applying
mm/s velocities with only a few volts. This sug- several volts at 1100 kHz frequencies using
gests using 3D ACEO pumps to drive flows in asymmetric arrays of planar electrodes with
battery-powered, portable, or implantable unequal widths and gaps, at the scale of several
microfluidic devices. microns. Somewhat slower flows have also been
Fluid pumping over electrode arrays can also achieved by applying traveling-wave voltages
be achieved by applying a traveling wave of with interdigitated planar electrodes [12, 13],
A 14 AC Electro-Osmotic Flow

AC Electro-Osmotic Flow, Fig. 4 Top: Simulations of design with stepped electrodes having a symmetric foot-
ACEO microfluidic pumps, showing the time-averaged print and the same minimum feature size; the reverse slip
flow over a pair of microelectrodes (dark regions) in one now recirculates in a vortex to create a fluid conveyor belt
spatial period of an interdigitated electrode array. (a) for the raised pumping flow from left to right (reproduced
A nearly optimal planar design with different electrode from [11]). Bottom: Scanning electron microscopy
sizes and gaps; the smaller electrode has the largest local images of each design fabricated in gold on glass with
slip velocity, but the larger electrode wins in overall minimum feature size (gap) of 5 mm (Unpublished, cour-
pumping from left to right. (b) A much faster 3D ACEO tesy of J. P. Urbanski using the methods of [4, 14])

although the electric connections are more a universal decay with concentration like
complicated. u / log(cmax/c) at large voltage [9]. As shown
A puzzling feature of ACEO in many situa- in Fig. 5b, the flow effectively vanishes above
tions with planar electrodes is that the flow direc- a bulk salt concentration of only cmax  10 mM,
tion can reverse (compared to the theory) with which is an order of magnitude smaller than most
increasing voltage, typically above 2 V. For biological solutions.
example, high-voltage flow reversal has been Nevertheless, advances have been made in
observed in dilute KCl for AC forcing [3] and applying ACEO flows to pumping, mixing, and
traveling-wave forcing [13] and in tap water for trapping particles in microfluidic devices. Exper-
AC forcing with a DC offset [7]. As shown in iments on 3D ACEO pumps with non-planar
Fig. 5a, the reverse flow is eventually faster than stepped electrode arrays have demonstrated an
the forward flow and seems to grow more quickly order of magnitude increase in flow rate versus
with voltage than the expected scaling u / V2. the fastest planar design, due to the robust fluid-
Another puzzling feature of ACEO is the conveyor-belt mechanism [4]. For a wide range
strong decay of the flow with increasing salt of near-optimal step heights, there is also no
concentration. For this reason, all experiments observed flow reversal [14]. Therefore, it is now
in the literature have used either very dilute possible to drive fast, reliable mm/s flows in
solutions (mostly KCl) or water (deionized or water or dilute electrolytes with less than 3 V,
from the tap). A few groups have studied the which can be provided by a small lithium battery
concentration dependence of ACEO in aqueous (Fig. 6).
KCl solutions [3, 15, 16]. These experiments Progress has also been made in exploiting
and recent work on electrokinetic motion of ACEO flows to manipulate colloidal particles
heterogeneous particles suggest that flows and biological cells in microfluidic devices
due to induced-charge electroosmosis exhibit [57]. The basic strategy is to use ACEO flow
AC Electro-Osmotic Flow 15 A
a 6
Electrode b 100
5 damaging 0.001 mM
zone 0.003 mM
0.01 mM
Voltage (Vrms)

0.03 mM

Velocity [m/s]
60 0.10 mM
3 0.30 mM
40 1.0 mM

1 0

0 20
103 104 105 106 100 101 102
Frequency/Hz Frequency [kHz]

AC Electro-Osmotic Flow, Fig. 5 Experimental data for damaging refers to bubble formation and inability to
ACEO pumping of KCl around a microfluidic loop, observe consistent flows. (b) Velocity versus AC fre-
one-fifth covered by the asymmetric planar electrode quency at 2 V peak to peak for different concentrations
array of Fig. 4a. (a) Contour plot of mean velocity versus (Reproduced from [16]). Both show high-voltage, high-
AC frequency and RMS voltage at a bulk salt concentra- frequency flow reversal, and the latter shows the strong
tion of 0.1 mM (Reproduced from [3]), where electrode decay with salt concentration

a 500 3V 3D b 200 f [khz]

3V Planer 4
400 2V 3D 5
2V Planer
150 6
1V 3D 7
Vmax [m/s]

Umax [m/s]

1V Planer 8

0 0

100 50
1 10 100 0 2 4 6 8 10
Frequency [kHz] h [m]

AC Electro-Osmotic Flow, Fig. 6 (a) Comparison of 5 mm minimum feature size (reproduced from [4]). (b)
ACEO pumping of water around a microfluidic loop by Experimental optimization of the step height for 3D
planar and (non-optimal) 3D electrode arrays with similar ACEO pumping of dilute KCl (Reproduced from [15])

to draw particles to stagnation points on the elec- a DC bias voltage to the low-frequency AC
trodes, where they are trapped, presumably by voltage (50100 Hz) between adjacent electrodes
DEP (although the classical theory does not [7]. Particles are observed to collect only on the
seem to predict this effect). By increasing the positively biased electrode, as shown in Fig. 7b.
voltage, the ACEO flow can be reversed, and It has been suggested that opposing ACEO
particles are observed to move away from the flows are produced by the competition between
stagnation lines, overcoming any remaining trap- Faradaic charging on one electrode (positive
ping force. In this way, it is possible to write bias) and capacitive charging on the other
and erase suspended particles, bacteria, or (negative bias), but a quantitative theory remains
microalgae on microelectrodes, as shown in to be developed and confirmed systematically by
Fig. 7a. This effect can be enhanced by adding experiments.
A 16 AC Electro-Osmotic Flow

AC Electro-Osmotic Flow, Fig. 7 (a) Collection of particle trapping by asymmetric polarization ACEO on
E. coli bacteria in tap water along the stagnation lines of electrodes with positive DC bias at higher voltage
ACEO flow on Au microelectrodes at low frequency (>3 V) (Reproduced from [7])
(100 Hz) and moderate voltage (1 V). (b) Preferential

Future Directions for Research standard Butler-Volmer reaction kinetics have

failed to predict any flow reversal, except at low
ACEO is attractive to exploit in microfluidics frequencies below 100 Hz [2, 8], although this
since it offers precise local flow control without may be consistent with experiments showing
moving parts, using standard fabrication methods low-frequency flow reversal with a DC bias
from the electronics industry. Device operation [7]. In contrast, high-voltage flow reversal with
requires only the voltage (few Volts), power unbiased AC forcing in electrode-array pumps
(mW), and current (mA) of a small battery, and tends to occur at high frequency (>10 kHz), as
Faradaic reactions are reduced or eliminated. shown in Fig. 5. An alternative hypothesis, which
These features make ACEO a promising route predicts high-frequency, high-voltage flow rever-
to portable or implantable microfluidic systems. sal due to capacitive charging alone, is based on
In contrast, DC electroosmotic pumps, such as the reduction of double-layer capacitance due to
those exploiting electroosmotic flow in porous ion-crowding effects [9]. The question is far from
media, typically require several 100 V and pro- settled, however, and more theoretical and exper-
duce bubbling at the electrodes due to electroly- imental work is needed.
sis, which can require a catalytic gas management Another important open question concerns the
system. On the other hand, ACEO pumps have so dependence of ACEO flow on the solution chem-
far only produced very small pressures (<10 Pa) istry (concentration, ionic species, solvent char-
far below those of porous-plug DC pumps acteristics, etc.). The suppression of ACEO above
(>100 kPa), but new designs could close this 10 mM bulk salt concentration limits its use in
gap. In any case, the ability to tune flows and biological applications and points to possible new
particle motion by ACEO at the micron scale by physics of the double layer at large voltages, such
modulating relatively small AC voltages could as a dramatic increase in the diffuse-layer viscos-
have many uses, e.g., for mixing, flow fragmen- ity upon ion crowding [9]. Very few electrolytes
tation, or manipulation of colloids and cells. have been studied experimentally, but current
Understanding flow reversal in ACEO is an work suggests that varying the ionic species can
important open question, from both fundamental also substantially alter the flow. An interesting
and practical points of view. As noted above, direction for future research would be to develop
it has been attributed to Faradaic reactions, surface coatings, electrode materials, and buffer
using theoretical arguments suggesting a scaling solutions to mitigate or manipulate these com-
u / eV [7]. Simulations of ACEO pumps using plex interfacial phenomena.
Acoustic Streaming 17 A
Cross-References 14. Urbanski JP, Levitan JA, Burch DN, Thorsen T,
Bazant MZ (2006) The effect of step height on the
performance of AC electro-osmotic microfluidic
Dielectrophoresis pumps. J Colloid Interface Sci 309:332341 A
Dielectrophoretic Motion of Particles and Cells 15. Green NG, Ramos A, Gonzalez A, Morgan H,
Electrical Double Layers Castellanos A (2000) Fluid flow induced by
Electrokinetic Flow in Porous Media nonuniform ac electric fields in electrolytes on micro-
electrodes. I. Experimental measurements. Phys Rev
Electrokinetic Motion of Heterogeneous E 61:40114018
Particles 16. Bazant MZ, Urbanski JP, Levitan JA,
Electrokinetic Motion of Polarizable Particles Subramanian K, Kilic MS, Jones A, Thorsen T
Electroosmotic Flow (DC) (2007) Electrolyte dependence of AC electro-
osmosis. In: Proceedings of 11th international confer-
Nonlinear Electrokinetic Phenomena ence on miniaturized systems for chemistry and life
sciences (MicroTAS), pp 2852878


1. Ramos A, Morgan H, Green NG, Castellanos Acoustic Streaming

A (1999) AC electric-field-induced fluid flow in
microelectrodes. J Colloid Interface Sci 217:420422
2. Ajdari A (2000) Pumping liquids using asymmetric Yong Kweon Suh and Sangmo Kang
electrode arrays. Phys Rev E 61:R45R48 Department of Mechanical Engineering, Dong-A
3. Studer V, Pepin A, Chen Y, Ajdari A (2004) An University, Busan, South Korea
integrated AC electrokinetic pump in a microfluidic
loop for fast tunable flow control. Analyst
4. Urbanski JP, Thorsen T, Levitan JA, Bazant MZ Synonyms
(2006) Fast ac electo-osmotic pumps with
non-planar electrodes. Appl Phys Lett 89:143508
5. Green NG, Ramos A, Morgan H (2000) AC electro-
Eulerian streaming; Steady streaming
kinetics: a survey of sub-micrometre particle dynam-
ics. J Appl Phys D 33:632641
6. Wong PK, Wang TH, Deval JH, Ho CM (2004) Elec- Definition
trokinetics in micro devices for biotechnology appli-
cations. IEEE ASME Trans Mechatron 9:366376
7. Wu J (2006) Biased AC electro-osmosis for on-chip Acoustic streaming is a secondary steady flow
bioparticle processing. IEEE Trans Nanotechnol generated from the primary oscillatory flow. It
5:8488 includes not only the Eulerian streaming flow
8. Olesen LH, Bruus H, Ajdari A (2006) AC electroki-
caused by the fluid dynamical interaction but
netic micropumps: the effect of geometrical confine-
ment, Faradaic current injection, and nonlinear also the Stokes drift flow which arises from
surface capacitance. Phys Rev E 73:056313 a purely kinematic viewpoint.
9. Bazant MZ, Kilic MS, Storey B, Ajdari A (2007)
Nonlinear electrokinetics at large applied voltages.
10. Brown ABD, Smith CG, Rennie AR (2000) Pumping Overview
of water with AC electric fields applied to asymmetric
pairs of microelectrodes. Phys Rev E 63:016305 Acoustic streaming originates from the acoustics.
11. Bazant MZ, Ben Y (2006) Theoretical prediction of
When the compressible fluid experiences a high-
fast 3D ACEO pumps. Lab Chip 6:14551461
12. Cahill BP, Heyderman LJ, Gobrecht J, Stemmer frequency oscillation driven by a certain source
A (2004) Electro-osmotic streaming on application of sound, the nonlinear interaction causes
of traveling-wave electric fields. Phys Rev E a steady current in the field. This phenomenon
is usually referred to as acoustic streaming. How-
13. Ramos A, Morgan H, Green NG, Gonzalez A,
Castellanos A (2005) Pumping of liquids with ever, it was found that even an incompressible
traveling-wave electroosmosis. J Appl Phys 97:084906 fluid can produce such steady streaming when the
A 18 Acoustic Streaming

fluid oscillates adjacent to a certain obstacle or an interaction between the air and the duct wall,
interface. So, as pointed out by Riley [1], the term a steady recirculating flow takes place within
acoustic streaming is misleading. Instead of the duct. The net effect is that dust or particles
acoustic streaming, he insisted on using steady accumulate at nodes.
streaming to encompass the case of incompress- On the other hand, researchers interested in
ible fluid. Acoustic streaming in this entry means mass transport in water waves have shown that
both of these effects. a similar phenomenon occurs for incompressible
Another important issue that must be flow (see [4] for an example). The salient feature
addressed is whether or not the Stokes drift flow is when an obstacle oscillates with high fre-
should be included in the streaming flow cate- quency in the quiescent viscous fluid. In this
gory. The Stokes drift flow is purely kinematic case the steady streaming emerges from both
and so it is basically different from the Eulerian sides of the obstacle along the oscillating direc-
streaming flow, which is driven by the time aver- tion at small viscosity [5]. Increased importance
age of the Reynolds stress term. However, the of the acoustic streaming in small scales has been
mass-transport effect given by the Stokes drift addressed in [6].
flow is not weaker than the Eulerian streaming. In this article, we will describe in some detail
On the contrary, in the case of a progressive the fundamentals of the three types of streaming
wave, the Stokes drift flow plays a dominant flows associated with the microfluidic applica-
role (e.g., application of flexural plate waves to tions. First, the quartz wind which corresponds
pumping and mixing in microfluidics [2]). There- to the one-dimensional compressible flow will be
fore, the Stokes drift flow must be understood as introduced. Second, the Eulerian streaming flow
a kind of streaming flow. in two-dimensional space will be considered.
Unfortunately, little is known about the histor- Kundts dust phenomenon will also be explained.
ical background of the study of acoustic stream- Finally, we will address the flexural plate wave
ing. Therefore, instead of trying to give the and its net effect, i.e., the Stokes drift flow.
historical background of the general concept, we
will briefly survey the modeling and applications
of acoustic streaming in microfluidics. A survey Basic Methodology
of steady streaming flow has been given by
Riley [1]. One-Dimensional Compressible Flow Model:
In the early period, study mainly focused on Quartz Wind
the acoustic streaming of compressible fluids. In this and next subsections, we closely follow
The simplest and seminal phenomenon associ- the review by Riley [1] to describe the physical
ated with acoustic streaming is the quartz wind. mechanism of acoustic streaming. We first con-
When the air experiences a high-frequency sider the case when the sound is generated from
oscillation given by an ultrasound source, a source and travels along a certain direction,
a progressive wave is established in the air. Due say the x*-direction, in a space. The governing
to the attenuation of the wave, a nonzero time- equation for the motion of the compressible
averaged Reynolds stress is built in the region fluid is
close to the sound source, and this stress pushes
the air in the direction of wave propagation. The @r @r u
0 (1a)
resultant wind is called a quartz wind. Lighthill @t @x
[3] has provided a physical and simple analytical
treatment of this steady streaming flow. Another @u  @u

1 @p 4 @ 2 u
u  n (1b)
important phenomenon is the so-called Kundts @t @x r @x 3 @x2
dust pattern. When the air within a duct receives
a standing wave, a nonzero time average of the where u* is the flow velocity, p* the pressure, t*
Reynolds stress is built inside the duct. Due to the the time, r* the density of the fluid, and n the
Acoustic Streaming 19 A
kinematic viscosity of the fluid. The fluid particle We then have
is assumed to oscillate back and forth with ampli-
tude A0, frequency o, and wavelength l. We then @ 2 u0 @ 2 u0 4 @ 3 u0 A
take 1/o, l/2p, A0o, and r0A0o2(l/2p) as the  2  d 0 (7)
@t2 @x 3 @t@x2
reference quantities for the time, length, velocity,
and pressure, respectively. Further, we expand The solution of this equation is
the density as
u0 exp2dx=3 cos x  t (8)
r r0 1 e r0 e r1   
(2) Notice that the basic mode is progressive. We
need the next higher-order equation of Eq. 1b,
where r0 corresponds to the undisturbed fluid which is
density. We also expand the dimensionless veloc-
ity and pressure, respectively, as @u1 @p1 4 @ 2 u1 @u0
 d u0 (9)
@t @x 3 @x2 @x
u u0 eu1 e2 u2    (3a)
The right-hand side term can be understood as
a dimensionless force per unit mass. Substituting
p p0 ep1 e2 p2    (3b) Eq. 8 into the right-hand side gives

Then the leading-order equations of Eqs. 1a @u0

and 1b for the small value of the parameter e u0 exp4dx=3
become d d 1
cos 2x  t sin 2x  t
3 3 2
@r0 @u0
0 (4a) Taking the time average over one period (the
@t @x
operation to be denoted as hi) of this provides
@u0 @p 4 @ 2 u0  
 0 d 2 (4b) @u0 d
@t @x 3 @x  u0 exp4dx=3 (10)
@x 3
where the variables without an asterisk are
dimensionless. Here, the two dimensionless which plays a role as the driving force for the
parameters are generation of the steady flow u1 . Taking the time
average of Eq. 9 results in
A0 v  
e , d (5) @p1 4 @ 2 u1 @u0
l=2p ol=2p  d  u0 (11)
@x 3 @x2 @x
which can be considered as the dimensionless
It should be noted that the term Eq. 10 can be
amplitude of the flow motion and the inverse of
understood as a kind of body force. Lighthill [3]
the Reynolds number, respectively. As usual,
has presented solutions of this system of equa-
these parameters are assumed to be small. To
tions for the case in which the force Eq. 10 acts as
solve this system of equations, we eliminate the
a point source in an infinite space. It is seen that
perturbed density in Eq. 4a and the pressure in
when the viscosity is low enough (at high stream-
Eq. 4b by using the definition of sound
ing Reynolds numbers), the flow from the source
is like a jet as shown in Fig. 1. This flow is
c2 o2 l=2p2 (6) sometimes called the quartz wind. The term
@r acoustic streaming refers to this flow in the case
A 20 Acoustic Streaming

x1 exp2dx0 =3 sin x0  t (13)


u0  exp2dx0 =3 cos x0  t ex1
@x xx0

Here again the last term contains the steady com-

Acoustic Streaming, Fig. 1 Acoustic streaming quartz ponent, and it becomes
wind generated from a point source of ultrasound

eud eexp4dx0 =3 (15)

which is known as the Stokes drift velocity. The

steady streaming flow eu1 given by the solution of
Eq. 11 is called the Eulerian streaming flow.
Compared with the Eulerian streaming flow, the
Stokes drift flow is confined in a region close to
the sound source, i.e., within the region x = O
(1/d). Actually this region corresponds to the one
that the force Eq. 10 acts upon. That is, within this
region the steady flow velocity is composed of the
Eulerian streaming velocity and the Stokes drift
Acoustic Streaming, Fig. 2 Acoustic streaming gener- velocity
ated from an actual ultrasonic transducer [7]
uL eu1 eud (16)
of ultrasound. As can be seen from Eq. 10, the
driving force for this current vanishes when there which is known as the Lagrangian velocity.
is no attenuation, i.e., when d = 0. Therefore we Beyond this region the Stokes drift flow vanishes
can say that the acoustic streaming for the case of and only the Eulerian streaming exists. In many
ultrasound in a compressible fluid is attributed to cases the Stokes drift velocity is ignored, but
the attenuation of sound. The photograph in Fig. 2 when the location of interest is not far from the
clearly shows the streaming flow ejected by the sound source, it should be considered. In general,
transducer that creates the ultrasound [7]. the Stokes flow is expected to occur only when
On the other hand, we can also calculate the the primary wave is progressive, not when it is
Stokes drift velocity from the solution of Eq. 8. standing.
We can follow a certain fluid particles path
Eulerian Streaming
x(x0, t), where x0 is a reference point independent
of the time. From the definition of the flow veloc- Even the incompressible fluid can also give rise
ity, we can write to Eulerian streaming flow when a solid obstacle
is in contact with the oscillating fluid. Consider
dx a two-dimensional incompressible flow around
eu (12) a solid body governed by the following dimen-
sionless equations:
We substitute Eq. 8 into the right-hand side of this
and let x = x0 + ex1(t). We then have u0 (17a)
Acoustic Streaming 21 A

@u 1 2 @U @ue 1 @ 2 U @U @U
eu  u p u (17b)  e U V (21)
@t Re @t @t 2 @Y 2 @s @n
where u u, v and is the two-dimensional where ue(s) denotes the tangential component of
gradient operator. We use 1/o, L, A0o, and the potential flow velocity evaluated at the solid
rA0o2L as the reference quantities for the time, surface. We expand the velocity components as
length, velocity, and pressure, respectively. Note before
that the reference length L used here represents
a typical dimension of the obstacle. The Reynolds  p 
U, V U0 , 2=Re V 0
number Re defined as
oL2 e U 1 , 2=Re V 1 . . . (22)
The leading-order equation is
is assumed to be large. For the small value of e,
we expand @U0 @u0e 1 @ 2 U 0

u, p u0 , p0 eu1 , p1 . . . (18) @t @t 2 @Y 2

The leading-order equation of Eq. 17b is then whose solution takes the following form

@u0 U0 u0e s1  exp1 iY expit (23)

p0 (19)
The normal component V0 can also be obtained
Therefore the leading-order solution is of poten-
from the continuity equation. Note that this flow
tial flow. Moreover the solutions can take
is still time periodic. The O(e) Eq. 21 then
a separable form, e.g.,
u0 fx, yexpit (20)

@U 1 @u1e 1 @ 2 U1 @U 0 @U0
  U0 V0
where the complex functions f p must
satisfy the @t @t 2 @Y 2 @s @Y
continuity equation 17a and i 1 denotes the (24)
imaginary unit. Note that the inviscid region
governed by these equations comprises most of Here our interest is the time-independent flow.
the flow domain except the thin layer near the Taking the time average of the above equation
solid boundary. over one period of oscillation gives
Since the potential flow solution (Eq. 20) does
not satisfy the no-slip condition on the solid sur- 1 @ 2 U1 @U 0 @U0
U0 V0 (25)
face, we must consider a thin layer (called the 2 @Y 2 @s @Y
Stokes layer) adjacent to the surface n = 0, where
n refers to the local coordinate normal to the wall. The solution of this equation yields the streaming
In this thin layer, we use the strained coordinate velocity at the edge of the boundary layer,
Y defined as
" #
p 3 # du0e #
n 2=Re Y U 11  1  iu0e 1 iu0e
8 ds ds
We also use the velocity components U and (26)
V along the local coordinates s and n, respectively
(the coordinate s is along the surface). Then the where the superscript # denotes the complex con-
boundary layer equation becomes jugate. This velocity then acts as a boundary
A 22 Acoustic Streaming

condition for the exterior bulk region. The oscillation

governing equation of the steady streaming flow
takes the following form [1]:

u ud  u p 2 u (27)

where Res is the streaming Reynolds number

based on the streaming velocity at the edge of
the Stokes layer; Res = e2Re. This equation looks
very similar to the NavierStokes equation, but
here the convective velocity is replaced by the
Lagrangian velocity uL u ud .
The above formulation is effective and suit-
able when the streaming Reynolds number is
Acoustic Streaming, Fig. 3 Sketch of the flow structure
large so that the Reynolds stress action is con- of the Eulerian streaming around a circular cylinder oscil-
fined within the thin Stokes layer. In the lating at a high frequency in a viscous fluid
microfluidic application, however, Res is usually
small. So, the Reynolds stress term may be added
to the streaming flow equation (27) so that The steady flow within the Stokes layer at high
streaming Reynolds numbers shows four-cell
1 structure around the circular cylinder as shown
u ud  u p 2 u  F (28a)
Res in Fig. 3. There are two streams coming out of the
cylinder from both sides in the direction of
F hU0  U0 U0  U0 i (28b) oscillation.
Kundts dust pattern manifests another simple
where the second term within h i on the right- example of the Eulerian streaming flow given by
hand side is nonzero for the compressible fluid the two-dimensional standing wave in a duct.
case; for the compressible fluid case, Eq. 28a When an acoustic standing wave is established
itself must be modified. This means that for the in the duct with a compressible fluid, the steady
case with a low streaming Reynolds number, streaming reveals four-cell structure over a half
Eq. 28a must be solved over the whole domain wavelength (or over the space between two
including the Stokes layer. Another important neighboring nodes) as shown in Fig. 4. Near the
point for the incompressible flow is that the Reyn- duct wall, the steady streaming is toward the
olds stress vanishes when the primary oscillating nodes, and near the duct center it is coming out
flow is of the progressive-wave type. On the other of the nodes. Therefore dust within the duct
hand, it should be noted that in the microfluidic should cluster near the nodal points of the stand-
area, no literature has taken into account the ing wave. The detailed solution for this case has
Stokes drift flow in the convecting velocity (i.e., been given by Riley [1].
the velocity u ud in Eq. 28a) in the numerical
simulation of the streaming flow. Stokes Drift Flow
We can experience a typical example of the Usually, by acoustic streaming we mean Eulerian
Eulerian streaming flow around a circular cylin- streaming. However, the recent application of the
der [5]. Here, the fluid surrounding the cylinder flexural plate wave in microfluidics implies that
oscillates with high frequency; or the cylinder the Stokes drift flow should also be considered as
may oscillate in the otherwise quiescent fluid a type of steady streaming flow, because the net
without fundamental difference in the results. effect in the fluid transport is not discernable
Acoustic Streaming 23 A

Acoustic Streaming, Fig. 4 Eulerian streaming flow occurring between two parallel plates when a standing wave
takes place inside the duct. Distance between the nodes corresponds to a half wavelength of the standing wave

a Heater
Zinc oxide

b l

particle trajectory
flexural plate A0

Acoustic Streaming,
Fig. 5 Microfluidic device
utilizing the flexural plate B0
wave. (a) Fabricated h* y*
device; (b) shape of the
flexural plate and the x*
coordinates for analysis [2]

between the two. Here we derive the equations as is c = ol/2p. A particle attached on the flexural
well as the solutions for the Stokes drift flow plate describes an ellipse with major A0 and
given by the flexural plate wave. The case of minor B0 = rA0 rotating clockwise when the
one-dimensional compressible flow has already wave travels toward the right-hand side as
been treated in this article. shown in Fig. 5 [2]. The ratio r is given from
Consider a viscous fluid confined between r = pd/l, where d is the membrane thickness.
a flat wall at y = 0 and a flexural plate at A typical value of r is 0.1. The spatial coordinate
y = h. The flexural plate oscillates showing x is along the wave propagation direction and y
a traveling wave with a high frequency o normal to it.
by a well-established piezoelectric vibration As before we use 1/o as the time scale, l/2p as
system. Figure 5 illustrates the geometry of the length scale, A0o as the velocity scale, and
the flow model investigated in this entry. The rA0o2l/2p as the pressure scale, and then the
wavelength is l and the wave propagation speed dimensionless continuity and momentum
A 24 Acoustic Streaming

equations take the form given in Eqs. 17a and cosh y

u cos x  t (32a)
17b. The equations for the particle motion on the sinh h
flexural plate are
sinh y
v sin x  t (32b)
dx re sin x0  t (29a) sinh h

The bulk-flow solution given by Eqs. 32a and

dy e cos x0  t (29b) 32b does not satisfy the condition for the tangen-
tial component of the velocity on the walls. So,
As a typical example, for l = 100 mm, d = 3 mm, we must expect boundary layers near each of the
A0 = 6 nm, and the frequency o/2p = 3 MHz, we = h, we use the
walls. First, near the top wall,p
have A0o = 113 mm/s, e = 3.8104, and strained variable Z h  y Re=2 . Then we
Re = 4,760. The boundary conditions are obtain the appropriate x-momentum equation and
solve for u to obtain
u 0 at y0 (30a)
u coth h cos x  t
v0 at y0 (30b) r  coth hexpZ cos x Z  t
u r cos x  t at yh (30c)
A similar process leads to the following solu-
tion for u in the bottom boundary layer:
v sin x  t at yh (30d)
where h is the dimensionless depth of the fluid u cos x  t  expY cos x Y  t
sinh h
layer. The velocity conditions, Eqs. 30c and 30d, (34)
are derived from Eqs. 29a and 29b, respectively.
In this derivation we simply replaced x0 by x, where Y y Re=2.
which is valid for small r; later we will consider The analytical solutions obtained so far imply
the higher-order effect. that the flow field driven by the oscillating flex-
As before, we neglect the nonlinear term in ural plate is progressive. This means that the
Eq. 17b. We also assume that Re is very large. Reynolds stress should vanish and so there should
Equation 17b then becomes be no Eulerian streaming. On the other hand, it
produces the nonzero Stokes drift flow. First we
@u consider the bulk flow. Into the definition of the
p (31)
@t velocity components u and v, i.e., dx/dt = eu and
dy/dt = ev, we substitute Eqs. 32a and 32b.
This indicates that the velocity field can be Assuming the variables x and y on the right-
described by a potential function f(x,y,t) in hand side are constant, i.e., x = x0 and y = y0,
such a way that u f. As usual, this inviscid we can integrate the equations to obtain
solution must satisfy the boundary condition for
the normal component of the velocity at each of cosh y0
the walls but not the tangential component. This x x0  e sin x0  t (35a)
sinh h
is the most important part of the analysis, because
otherwise the leading-order solution is not of the sinh y0
y y0 e cos x0  t (35b)
progressive-wave type. Applying the conditions sinh h
given in Eqs. 30b and 30d to the Laplace equation
2f = 0 and solving for f, we obtain f = (cosh Now we substitute Eqs. 35a and 35b into Taylor
y/sinh h)sin(x  t) and expansions for u and v, i.e., Eqs. 32a and 32b
Acoustic Streaming 25 A
about (x, y) = (x0, y0) and take the time average 0.003
over one period to obtain the steady velocity
components v 0 and h=0.3 A
e 0.002
u 2
cosh2 y sinh2 y (36)
2sinh h u

Next, we consider the top boundary layer. 0.001

After some algebra we obtain v 0 and
( 2 h=1
e cosh y
u 1  A rA 0
2 sinh h 0 0.2 0.4 0.6 0.8 1
sinh y sinh y
1  A 1  A A r y  hA
sinh h sinh h

2 Acoustic Streaming, Fig. 6 Distributions of the Stokes
cosh y drift velocity across the fluid layer for three channel depths
B2 r 
sinh h given by the flexural plate wave
sinh y sinh y
B2  r y  h
sinh h sinh h
(37) As a typical example, for the case
where functions A and B are with l = 100 mm and h = 50 mm, we
have h = p, and for e = 3.8  104 and
A, B expZ cos Z, sin Z Re = 4,760, we evaluate u 3:8  104 as the
approximately maximum streaming velocity
Similar treatment can be done for the bottom obtained from the bulk-flow solution (Eq. 36)
boundary layer and we obtain v 0 and at y = h. When the velocity scale is
A0o = 113 mm, this becomes u 43 mm=s .
e This is not much different from the measured
u 1  2expY cos Y exp2Y 
2sinh2 h velocity (less than 100 mm/s) reported by Nguyen
(38) et al. [2], considering that the numerical result
1,000 mm/s obtained by them is more than
Now, the complete formula for the streaming 10 times the measured value for the same
velocity applicable to all the regions including parameter set.
the two boundary layers is v 0 and Figure 6 shows the distributions of the
( streaming velocity for three depths of the fluid
e cosh y layer with the parameter set presented previously.
u 1  A rA
2 sinh h The magnitude of streaming velocity increases on
  the whole as the depth decreases. Further
sinh y sinh y
1  A 1  A A ry  hA decrease of h results in the overlap of the top
sinh h sinh h

2 and bottom boundary layers, and so the analytical
cosh y
B r
2 solutions presented so far may not hold in
sinh h
  that case.
2 sinh y sinh y
B  r y  h The flow rate through the channel is given by
sinh h sinh h
1 y
exp2Y  2expY cos Y   r 2 h
sinh h h
q udy
(39) 0
A 26 Acoustic Streaming

Re = 4760
For instance, at Re = 4,760 we obtain hc = 0.041
which corresponds to hc = 0.65 mm at
0.002 l = 100 mm. This is very small compared with
the wavelength, but it is still far larger than the
0.0015 amplitude of the oscillation of the plate,
A0 = 6 nm.

Re = 400
Key Research Findings

0 Acoustic streaming has been applied to various

101 100 101
microfluidic areas. Whites group [2] have devel-
oped micropumps that utilize the flexural plate
Acoustic Streaming, Fig. 7 Effect of the channel depth wave as the actuator. They have conducted
on the flow rate at three Reynolds numbers for the flexural- a numerical simulation but the deviation was
plate-wave pumping. Dashed line corresponds to Re = 1
significant; their simulation results had to be
corrected by a factor of 7 in order to fit the
measurement data. To date, no report has been
Substituting the streaming velocity distribution given of a satisfactory agreement between the
(Eq. 39) into this yields the following formula: experimental measurements and numerical or
theoretical predictions. The numerical computa-
e tion for acoustic streaming is not as simple as it
q 2
2sinh h
 looks. There are two key factors which may
p 7 degrade the numerical results: one arises from
sinh hcosh h  2=Re 1 sinh2 h
8 the thin Stokes layer and the other from the
(40) extremely low levels of the streaming velocity
compared with the primary flow. These can be
At high Reynolds numbers, the second term overcome when we put very fine grids near the
within [] on the right-hand side of the above Stokes layer and simultaneously use high-
equation can be neglected and so the flow rate precision numerical values.
increases as the channel depth decreases, in line In order to apply acoustic streaming to fluid
with the numerical result of Nguyen et al. [2]. On mixing, Yaralioglu et al. [8] designed patterns
the other hand, when the second term is consid- of piezoelectric transducers on the bottom wall
ered, the flow rate shows a maximum value at of a microchannel. They verified experimentally
a critical depth. Figure 7 shows the dependence of that the transducers, by generating acoustic
the flow rate on the depth at three Reynolds streaming inside the channel normal to the
numbers and at the other parameter values the main flow, enhanced the mixing. The mixing
same as those presented previously. It indeed effect was, of course, affected by the configura-
shows that at each of the critical depths, the tion of the electrode patterning. They also
flow rate has a maximum value. The critical asserted that the fundamental problem of the
depth increases as the Reynolds number formation of bubbles and unintended heating of
decreases. It can be shown that the critical depth the fluid was not expected to occur in their
is given by the formula. system.
Acoustic streaming has also been applied to
r! electrochemistry. Comptons group [9] has tested
hc sinh1 (41) the use of ultrasound to study the effect of the
Re  8
various configurations of the acoustic streaming
Acoustic Streaming 27 A
on the limiting current from an electrode. It was Future Directions for Research
shown that there was a critical diffusion layer
thickness below which the layer no longer To date no numerical simulation of the acoustic A
became thinner. Recently they performed streaming flow associated with application to the
a numerical simulation on the acoustic streaming microfluidics has been successfully performed.
over an electrode within an axisymmetric space. The simulation can be performed only with the
The streaming flow was generated by the Reyn- streaming flow by using the governing equation
olds stress terms formulated by other investiga- given as Eq. 27, but we also need to solve the full
tors. They successfully verified the previous unsteady NavierStokes equations including not
experimental results that acoustic streaming was only the primary oscillatory flow component but
the principle mechanism of enhanced mass trans- also the secondary steady flow component. In this
port in sono-electrochemical cells. case locally fine grids must be adopted to deal
Ultrasonics and its universal effect (acoustic with the thin Stokes layer adjacent to the solid
streaming) have been used to manipulate parti- surface.
cles in microfluidic devices, e.g., trapping, col- We also need to analyze the case for the
lection, separation, and deposition. For instance, flexural-plate-wave flow in more detail and to
Petersson et al. [10] employed a microchannel simultaneously perform numerical simulations
system with one center and two side channels at of the resultant flows. The results of the full
the inlet and outlet, respectively. Through the numerical simulation may be used to prove that
side inlet channels they put a medium, say fluid the steady flow components are generated by the
A, containing particles, and through the center mechanism known as the Stokes drift flow.
inlet channel the fluid B containing no particles. We also need to study the dissipation effect
They then applied ultrasound normal to the chan- caused by the oscillatory flow. At present, in
nel in such a way that a standing wave was lab-on-a-chip design, we are usually interested
established in the main channel where the two in the feasibility of the functions provided by
media contacted each other. Due to the Reynolds the transducers, but sooner or later the energy
stress and recirculating streaming flows, particles consumption may be one of the most important
were collected near the channel center and this factors.
pattern persisted all the way through the main
channel. At the outlet, the particles flow with
the fluid B. They implied that such a medium- Cross-References
exchange effect could be applied to blood
washing. Acoustic-Based Biosensors
Marmottant and Hilgenfeldt [11] have consid- Piezoelectric Microdispenser
ered bubble-driven microfluidic transport for bio- Transport of Droplets by Acoustics
engineering applications. In their experiment, Ultrasonic Pumps
bubbles were adsorbed on the bottom wall of
the microchannel. When the standing ultrasound
wave was introduced within the channel, steady
streaming flow was generated. They stressed that References
the flow pattern in bubble streaming was quite
1. Riley N (2001) Steady streaming. Annu Rev Fluid
different from that induced by an oscillating solid Mech 33:4365
object. More interestingly, close contact of 2. Nguyen NT, Meng AH, Black J, White RM
a foreign solid particle to the bubble showed (2000) Integrated flow sensor for in situ measurement
and control of acoustic streaming in flexural plate
a significant effect in the overall steady flow
wave micropumps. Sens Actuators A 79:115121
field due to superposition of the bubble and par- 3. Lighthill J (1978) Waves in fluids. Cambridge
ticle streaming fields. University Press, Cambridge
A 28 Acoustic-Based Biosensors

4. Iskandarani M, Liu PLF (1991) Mass transport in three- Definitions

dimensional water waves. J Fluid Mech 231:417437
5. Schlichting H, Gersten K (2003) Boundary layer the-
ory. Springer, Berlin A biosensor is defined as an analytical device that
6. Frampton KD, Martin SE, Minor K (2003) The scal- uses a biological recognition system to target
ing of acoustic streaming for application in micro- molecules or macromolecules. Biosensors use
fluidic devices. Appl Acoust 64:681692 a physiochemical transducer to convert the signal
7. Nowicki A, Kowalewski T, Secomski W, Wojcik J
(1998) Estimation of acoustic Streaming: theoretical from the bio-recognition system into a detectable
model, Doppler measurements and optical visualisa- signal [1]. Biosensors consist of three compo-
tion. European J Ultrasound 7:7381 nents: (1) the detector, which identifies the stim-
8. Yaralioglu GG, Wygant IO, Marentis TC, Khuri- ulus; (2) the transducer, which converts this
Yakub BT (2004) Ultrasonic mixing in microfluidic
channels using integrated transducers. Anal Chem stimulus to an output; and (3) the output system,
76:36943698 which involves amplification and display of the
9. Marken F, Akkermans RP, Compton RG (1996) output in an appropriate format [1].
Voltammetry in the presence of ultrasound: the limit Piezoelectricity is a phenomenon displayed in
of acoustic streaming induced diffusion layer thinning
and the effect of solvent viscosity. J Electroanal Chem certain crystals, such as quartz and Rochelle salt,
415:5563 where mechanical stress induces voltage genera-
10. Petersson F, Nilsson A, Jonsson H, Laurell T (2005) tion and vice versa.
Carrier medium exchange through ultrasonic particles
switching in microfluidic channels. Anal Chem
11. Marmottant P, Hilgenfeldt S (2004) A bubble-driven Overview
microfluidic transport element for bioengineering.
Proc Natl Acad Sci U S A 101(26):95239527 There is an increasing demand for small, reliable,
disposable, and inexpensive sensors in industrial,
medical, and a variety of other science and engi-
Acoustic-Based Biosensors neering fields. Sensors are one of the fastest
growing markets, with annual growth of about
N. Gozde Durmus1, Richard L. Lin2, 18 %. In particular, the market for biosensors is
Mariel Kozberg3, Deniz Dermici4, increasingly promising due to their application in
Ali Khademhosseini5 and Utkan Demirci5 progressing areas of health care, biotechnology,
Department of Molecular Biology and Genetics, and medicine. For instance, the uses of biosensors
Middle East Technical University, Ankara, in personal glucose testing, HIV, and early cancer
Turkey detection are areas with great medical relevance
Department of Materials Science and [1]. In addition, the ability to detect pathogenic
Engineering, Massachusetts Institute of and physiologically relevant molecules in the
Technology, Cambridge, MA, USA body with high sensitivity and selectivity offers
Laboratory for Functional Optical Imaging, a powerful tool in early diagnosis and treatment
Department of Biomedical Engineering, of diseases [1].
Columbia University, New York, NY, USA An acoustic wave biosensor utilizes
Department of Economics, Bogazici University, acoustic or mechanical waves as a detection
Istanbul, Turkey mechanism to obtain medical, biochemical, and
Harvard-MIT Division of Health Sciences and biophysical information about the analyte of
Technology, Massachusetts Institute of interest [1, 2]. It detects changes in mass, elastic-
Technology, Cambridge, MA, USA ity, conductivity, and dielectric properties from
mechanical or electrical variations. These
devices also employ the piezoelectric effect to
Synonyms excite acoustic waves electrically at an input
transducer and to receive the waves at the output
Acoustic waves; CMUTs; Medical imaging transducer [2].
Acoustic-Based Biosensors 29 A

Acoustic-Based Biosensors, Fig. 1 Classification of acoustic waves [1]

The first scientific experiment of piezoelec- capacitive micromachined ultrasound transducers,

tricity was conducted in 1880 by Pierre and Paul a novel technology that does not require piezoelec-
Jacques Curie, who found that compression of tric substrates. For each type of these sensors, we
various crystals leads to voltage production on will present an overview of their scientific mech-
their surface. They showed the converse effect anisms and fabrication techniques in the Basic
the following year by demonstrating the elonga- Methodology section. We will then go on to
tion of certain crystals when injected with electric discuss the advantages and applications in biology
current. Although these observations intrigued and medicine in the Key Research Findings and
the scientific community, piezoelectricity Future Direction for Research sections.
remained much of a laboratory technology until
Paul Langevin used piezoelectric transducers to
construct sonar for submarine detection in 1915 Basic Methodology
[3]. Acoustic biosensors can be made with piezo-
electric crystals such as quartz, lithium niobate, Bulk Acoustic Wave (BAW) Devices
or lithium tantalate since they are robust and Bulk acoustic wave (BAW) biosensors employ
environmentally stable. In addition, such sensors either longitudinal or shear waves, although the
are versatile and can detect, in principle, various latter is often preferred to reduce acoustic radia-
biomolecules [1]. Acoustic wave sensors can be tion in the medium of interest. They are the oldest
categorized according to the waves they generate, and the simplest acoustic wave devices. BAW
such as bulk or surface acoustic waves (Fig. 1). devices consist of a parallel electrode placed on
Each of these wave mechanisms have various both sides of the thin piece of crystal. BAW
advantages and disadvantages depending on sensor can technically employ any piezoelectric
their specific applications [1]. element, and typically quartz is used, as it is an
The goal of this entry is to categorize and inexpensive material readily available in nature
describe the mainstream acoustic wave biosen- and easily synthesizable in abundant quantities.
sors. We will begin by discussing biosensors that In addition, thin disks of quartz are more stable at
utilize bulk acoustic waves and subsequently ana- high temperatures than other piezoelectric ele-
lyze devices that employ surface acoustic waves. ments. When an alternating electric field is
We will also examine sensors that employ applied, it results in a potential difference
A 30 Acoustic-Based Biosensors

between the two electrodes and the shear defor-

mation of the crystal. As a result, there is mechan-
ical oscillation of a standing wave across the bulk
of the quartz. The frequency of the vibrations is
dependent on quartz properties such as density,
size, and phase in contact with the crystal surface
[4]. Currently, thickness shear mode (TSM) res-
onator and shear horizontal acoustic plate mode
(SH-APM) sensors remain the most widespread
BAW sensors. The other two common BAW
sensors that lie beyond the scope of this entry
are the thin rod acoustic wave sensors and the
flexural plate wave devices [4].

BAW Device: Thickness Shear Mode (TSM)

TSM resonator, also known as quartz crystal
Acoustic-Based Biosensors, Fig. 2 Changes to the res-
microbalance (QCM), is the simplest and most
onance peaks due to a thin solid layer (top) and liquid
widespread acoustic wave device today. TSM (bottom). Note the decrease in Q for TSM resonator in
typically composes of a quartz plate sandwiched liquid [18]
by electrodes on opposite faces. Electric field
crosses through this plate when voltage is applied
to the electrodes, resulting in a shear mechanical The variables Df, f0, Dm, Apiezo, uq, and pq
strain or displacement in the quartz. By oscillat- represent, respectively, the crystals measured
ing the voltage frequency, a mechanical reso- frequency shift, initial resonant frequency, mass
nance can be generated, where the maximum change, surface area, shear modulus, and density
displacement of crystal occurs at the surfaces. [6]. TSM resonators immersed in liquid present
The resonant frequency, Fr, and the quality a special case because acoustic energy dissipation
factor, Q, are the two resonance parameters. through the fluid lowers Q as well. Additional
While Fr is the mechanical thickness shear reso- formulas need to be applied to account for the
nance as mentioned before, it is also defined as liquids viscosity and loading [6]. Figure 2 shows
the frequency of the maximum value of the elec- a comparison between shifts induced by solid and
trical conductance, Gel. Q is approximated liquid loading on resonators.
mathematically from the electrical conductance TSM resonators possess lower mass sensitiv-
resonance spectrum as Q = Fr/DFBW, or the ity than most acoustic wave sensors that employ
ratio of resonant frequency to the half other mechanisms since they typically operate
bandwidth [5]. between only 5 and 30 MHz. Thinner devices
Coupling of molecules in solid, liquid, or gas can be used to increase frequencies that result in
phase to the TSM resonator changes the reso- increased mass sensitivity, and they are more
nance parameters. When mass attaches onto the fragile and harder to manufacture and handle
crystal face, it raises the quartzs overall weight [5]. Recent research on this topic has been on
and reduces Fr. Saurbrey first derived the mathe- creating high-frequency TSM resonators with
matical relationship between loading mass and Fr piezoelectric films and bulk silicon
to be micromachining techniques.

2f 0 2 Dm TSM Fabrication
Df  1=2 (1) TSM resonator fabrication usually utilizes photo-
Apiezo uq pq lithography [2]. The resonance frequency of the
Acoustic-Based Biosensors 31 A
quartz crystal depends on the shear modulus,
density, and material thickness [1]. After
polishing the quartz, photoresist is deposited, A
and the device is exposed to UV light through
a mask, which determines the active areas that
have to be metalized on the devices. On a TSM
resonator, they represent the location of the elec-
trode plates. A chemical interface is then added to
the remaining crystal surface, where molecules in
the liquid can bond to and be detected [2].

BAW Device: Shear Horizontal Acoustic Plate

Acoustic-Based Biosensors, Fig. 3 SH-APM can sense
Mode (SH-APM) Sensor on either side of the device because waves travel between
First introduced in the 1980s, SH-APM sensors the upper and lower surfaces of the plate [2]
use a thin piezoelectric substrate, or a plate, to
guide the acoustic wave and to confine its energy
within the plates top and bottom surfaces. Most a similar fashion as TSM resonators. One differ-
of the production and analysis principles ence that must be noted is the use of IDTs in
employed in SH-APM sensors are used in SH-APM fabrication, which is not present in
a TSM resonator. Their most striking difference TSM resonators. Depending on the application,
is that SH-APM sensors employ interdigital the optimization of sensor performance can be
transducers (IDT) rather than electrode plates. achieved through adjustments of the length,
IDTs are deposited on opposite ends of width, position, and thickness of the IDTs.
a surface, where one IDT generates displacement These parameters can be altered by varying
waves through application of an oscillating volt- the design of the masks, amount of photoresist,
age and the other receives it (Fig. 3) [2]. The and other chemicals used in the fabrication
surface without IDT is immersed in the targeted process [2, 7].
liquid and acts as the sensor, so the device will Moreover, SH-APM devices require addi-
not suffer from corrosion problems as electrode tional processing during their fabrication of the
plates do in biological solutions [7]. thin plates with respect to the IDT periodicity.
Like TSM resonators, SH-APM sensors pre- Thin slices of piezoelectric materials such as
vent the acoustic waves from gaining a shear ver- quartz are often used. The whole procedure per-
tical component and thus lessening energy mits researchers to choose the correct crystallo-
radiation in the form of compression waves into graphic orientation for generating the desired
the liquid. They achieve this by restraining particle shear waves. Despite this improvement,
displacement within the plate plane. However, SH-APM devices remain less sensitive than
given a SH-APM and a TSM device of similar SAW devices for two reasons. First, energy of
plate thickness, the former usually attains higher the wave does not reach a maximum at the sur-
mass sensitivity than the latter because IDT is face. Second, similar to TSM, SH-APM is
much smaller and precise than the electrode plates restrained by a minimum thickness of piezoelec-
on TSM. In general, SH-APM devices generate tric plate [2, 7].
waves with a frequency ranging from 100 to
150 MHz, but they are also more complex and Surface Acoustic Wave Sensors (SAW)
expensive than their TSM relatives [2]. Surface acoustic wave (SAW) sensors have been
used for years in measuring temperature, pres-
SH-APM Fabrication sure, viscosity, acceleration, concentration, and
Although the crystals and substrates may be dif- chemical/biological entities [3]. They are also
ferent, SH-APM sensors are fabricated in used to perform signal-processing operations
A 32 Acoustic-Based Biosensors

and are sensitive to their environment [8]. This noses mostly have the ability to detect very low
type of acoustic wave sensors consists of basic vapor concentrations [8]. If there is vapor on
components such as piezoelectric substrate the piezoelectric surface of the SAW device,
(crystals such as quartz, GaAs, or LiNbO3), the chemical substances in the vapor and in the
micrometallization patterns (electrodes), acoustic wave interact and the properties of the
interdigital transducers (IDT), and active thin wave changes. Sensitivity of the SAW detector is
films. Unlike BAW, which only interacts with dependent on the temperature of the piezoelectric
the environment at the opposite surface of the crystal and the vapor-pressure of the analyte.
material by traversing through it, SAW travels Chemical concentration of analyte of interest
along or near the surface of the piezoelectric can be tested by monitoring the frequency. The
material [3]. The piezoelectric device detects SAW chemical sensor requires no special support
small mass changes at the surface of the sensor gases and no consumables other than electrical
as a frequency response, as the electrodes on the power. Many of the SAW devices have been
surface of the piezoelectric substrate transmit and produced for military and defense applications.
receive acoustic waves [8]. The acoustic wave is They are used for on-the-spot vapor detection of
confined to the surface of the piezoelectric sub- plastic explosives containing nitro groups such as
strate, and the excited wave is propagated along trinitrotoluene (TNT) and cyclotrimethylenetri-
the crystal surface. The surface wave velocity nitramine (RDX) [8]. As a result, SAW sensors
changes due to the mass, or the viscosity changes are attractive for real-time detection and moni-
due to the binding reactions on the sensor surface. toring of toxic or hazardous gases and vapors, and
The range that SAW device operates depends on they are used in biology as a sensitive method for
the acoustic velocity of the crystal substrate and detection of biochemicals and biomolecular
IDT wavelength, and it usually lies between interactions [9].
lower MHz range to the GHz range [8]. These SAW devices operate at high frequencies,
devices are commonly used because of their abil- which lead to smaller, more sensitive devices
ity to directly sense the changes in mechanical that have limits in the picogram range [8]. SAW
and mass properties, and they have a broad range devices can be miniaturized by using photolitho-
of applications such as pressure, temperature, graphic techniques, and complex circuits can be
viscosity, mass, and chemical and biological present on the substrate surface. Miniaturization
sensing [8]. SAW sensors coated with organic makes them advantageous since the smaller
and/or inorganic materials can be used to analyze dimensions increases the sensing surface area
absorption on the surface of a device [9]. For and binding properties [6]. Moreover, there is
chemical sensing, the chemoselective material a dual-delay system with SAW sensors (Fig. 4).
must be tailored to the analyte of interest. During It is a layered structure and has three fundamental
its operation, SAW sensor responds to a gas or advantages. First, it compensates the influence of
vapor exposure in air by a shift in resonant signal temperature on the differential frequency and the
frequency and then recovers to the original signal air pressure. Second, it reduces the measured fre-
frequency, when the sensor is exposed to ambient quency of the system from MHz range to kHz
air free of the responding analyte [8]. The SAW range [6]. Third, SAW devices have the ability to
velocity changes according to the absorbed directly sense mass and mechanical changes in an
chemical substance, and it results in the alterna- environment, which makes them more sensitive
tion of the mass load, the elastic stress, and the than other electrical and thermal biosensors. Fur-
conductivity of the polymer coating on the SAW thermore, they have numerous advantages in bio-
device. logical applications [6]. These devices are also
SAW devices are also considered as elec- more cost-efficient than the optical biosensors
tronic noses. An electronic nose is a device and easier to handle. They are employed to study
that is composed of a chemical-sensing system biomolecular interactions and allow easy and
and a pattern-recognition system. Electronic direct detection of analyte of interest [8].
Acoustic-Based Biosensors 33 A
f tr
Biosensors, Fig. 4 SAW
device dual-delay line fA
structure and its principal of MPc V Lw L
operation [24] VO
f tr
f LiNoCB f0
f f0


In summary, SAW technology is a cost-effective a commercial method to fabricate SAW devices,

method of label-free detection of biomolecular especially for simple pattern, one-mask-level
interactions and allows direct monitoring on the applications, where large areas can be patterned
surface of a transducer in real time [10]. with nanometer-scale critical dimensions and
high fidelity. Moreover, nanoimprint lithography
SAW Device Fabrication is becoming a popular lithographic method. SAW
Various lithographic techniques are used in the filters and SAW correlators present ideal struc-
fabrication of interdigital transducers (IDTs) for tures to be fabricated using this method.
surface acoustic wave sensors. The most widely
used and least costly photolithographic technique Capacitive Micromachined Ultrasonic
to fabricate SAW devices is conventional contact Transducers (CMUTs)
printing. It uses rigid or conformable glass plates Electrostatic transducers have been used for
bearing the photographic image, which can be sound wave excitation and detection as an alter-
negative or positive depending on the fabrication native to piezoelectric devices that are used
steps of the IDT pattern [11]. Mass production of widely for nondestructive evolution and
acoustic devices is done by optical lithography. medical-related applications. The vibration of
Excitation of the SAW harmonics depends on the a thin plate under electrostatic forces is the main
metallization ratio between the gaps and the elec- acoustic actuation and sensing mechanism of
trodes of the SAW device. A lithography method these devices. These devices are based on the
especially suited for the fabrication of small gaps electrostatic attraction force between the
is the near-field phase shift technique. In this backplate and the metallized membrane.
technique, a transparent mask with a pattern on The simplest forms of this device consist of
its surface is brought into contact with the sub- a thin metal membrane that is stretched above
strate. The phase of the exposing light is a back electrode forming a small gap [13]. Upon
modulated at the edges of the patterned relief. exposure to sound waves, the gap height is mod-
High-frequency, low-loss SAW devices can be ulated at the frequency of the incoming pressure
fabricated by depositing thin piezoelectric films field that induces a change in the device capaci-
on substrates. These include ZnO, GaN, AlN, and tance, which generates an output voltage propor-
AlxGa1  xN thin films on sapphire, SiC, and tional to the amplitude of the field. The capacitor
diamond substrates. The SiO2/ZnO/diamond is also used to generate sound waves. The advan-
structure can provide a high acoustic velocity tage of the electrostatic devices compared to
and a zero-temperature coefficient of frequency. the other types of transducers is the impedance
Extremely high velocity SAW modes with mod- matching between the transducer and the
erate coupling coefficient in AlN and GaN thin surrounding medium. Advances in the silicon
films are deposited on SiC, diamond, or sapphire micromachining techniques enabled the
substrates [12]. In addition, imprint lithography is fabrication of microelectromechanical system
A 34 Acoustic-Based Biosensors

Acoustic-Based Buried top

Biosensors, Vacuum gap
electrode Silicon nitride
Fig. 5 Schematic cross Silicon nitride
section of a CMUT isolation membrane
transducer membrane [28]


(MEMS)-based electrostatic transducers. These CMUT Fabrication

devices use the ultrasonic waves due to the mini- There are two conventional methods to fabricate
aturization capacity of the silicon machining tech- CMUT: sacrificial release process and wafer-
niques and are called capacitive micromachined bonding method. The first CMUT was created
ultrasonic transducers (CMUTs) [13]. by using a sacrificial release process, and this
CMUTs are made of small and thin transducer method has become the standard CMUT fabrica-
membranes that are suspended over a conductive tion for many years. In the sacrificial release
silicon substrate (Fig. 5). The two plates of the process, a cavity underneath the membrane is
capacitor, the substrate and the membrane, are created by depositing or growing a sacrificial
separated by a vacuum-sealed cavity. layer on the carrier substrate. The sacrificial
Two-dimensional (2D) arrays of miniaturized layer is removed with an etchant after the mem-
electrostatic cells are microscale cells that oper- brane deposition. However, there are some prob-
ate in parallel and driven in phase. Surface silicon lems associated with this technology. The surface
micromachining is used for their microfab- micromachining process causes limitation on the
rication [13]. The operation principle of these cavity and membrane size of a CMUT device. In
devices is based on the capacitance modulation addition, it is hard to batch produce surface
of a microcondenser. Between the metalized micromachined CMUTs to tight quality specifi-
membrane and the substrate, a direct current volt- cations [14]. On the other hand, wafer-bonding
age is applied. The membrane is attracted toward method is a new fabrication technique that com-
the bulk by the electrostatic force, and induced bines bulk and surface micromachining tech-
stress within the membrane resists the attraction. niques. The transducer membrane and cavity are
If the membrane is exposed to ultrasound, defined on a SOI wafer and on a prime wafer,
a current output is generated as the capacitance respectively. Two wafers are then bonded by
changes under constant bias voltage. The ampli- using silicon direct bonding in a vacuum environ-
tude of the current output is a function of the ment, together forming the transducer (Fig. 6)
frequency of the incident wave, the bias voltage, [14]. This new method improves CMUT perfor-
and the capacitance of the device. CMUTs per- mance since it decreases the number of process
form operation in the megahertz range. The lat- steps and device turn-around time and increases
eral dimensions of each cell are in the order of the overall uniformity, reliability, repeatability.
tens of microns [13]. The gap between the mem-
brane and the substrate vacuum sealed or
unsealed can be as small as 500 A . The mem- Key Research Findings
branes may be conductive or coated with
a conductive electrode. This structure results in Advantages of BAW Devices over SAW
very efficient transducers that can compete with Devices
their piezoelectric counterparts in terms of effi- Acoustic biosensors generally employ BAW
ciency and bandwidth. The small and thin mem- rather than SAW because BAW devices generate
branes that constitute the CMUT transducer are waves that mainly propagate in the shear hori-
micromachined onto a silicon substrate [13]. zontal motion. Since shear horizontal wave does
Acoustic-Based Biosensors 35 A

1. Thermal oxidation and etch 5. Ground/etch back and remove the box.

2. Silicon etch (only for deep cavitv).

6. Silicon/oxide etch

3. Thermal oxdation for electrical isolation.

7. Al deposition and pattering.

4. Silicon wafer fusion bonding 8. LTO deposition and patterning.

Acoustic-Based Biosensors, Fig. 6 Major steps of CMUT fabrication with wafer-bonding method (Reprinted with
permission from Ergun et al. [14])

not radiate significant amount of energy into liq- SH-APM sensors have also been applied in
uids, it minimizes damping in the liquid opera- immunochemical reactions. In order to observe
tion. Besides few exceptions, the SAW sensor antigen-antibody reactions through mass loading
induces considerable surface-normal displace- on the crystal surface, Dahint et al. [7] constructed
ment that radiates compression waves into the an immunosensor using a 150 MHz APM device
liquid and causes excessive damping. Since bio- on a ZX-LiNbO3 plate with a thickness of 0.5 mm.
sensors detect chemicals in liquid solutions, These sensors, which could detect mass as small as
BAW is often preferred over SAW for many to 200 pg/mm2 of crystal surface, marked a shift in
biosensing applications [2]. resonant frequency of around 1 kHz upon antigen
adsorption [7].
Biological Applications of SH-APM Sensors
SH-APM sensors have been utilized in various Biological Applications of TSM Resonators
biological applications. For example, Liren Many biosensors that employ TSM resonators are
et al. [15] presented a urease biosensor in which available on the market today. Depending on their
a 0.5 mm thick ZX-LiNbO3 substrate was used design, they can detect biological entities such as
and a resonant frequency approximately centered nucleic acids, viruses, bacteria, and cells. An
at 50 MHz. APM was generated and detected by example of TSM-based biosensor was presented
IDTs, with a period of 88 um and a center dis- by Davis et al. [16]. They used a coating of human
tance of 10 mm, on the same PZ crystal. The IgG and subsequently used a TSM resonator to
group also fabricated two pairs of IDTs on the detect the presence of sheep antihuman IgG as
substrate so that one would conduct biochemical marked by the decrease in the resonant frequency,
reactions while the other acted as a reference. signifying an increase in the mass loading from the
Using this device, they successfully measured attachment of the antibodies [16].
urea concentrations by detecting variations in TSM sensors can also be used as DNA-based
the resonant frequency [15]. sensors. Okahata et al. [17] studied the use of
A 36 Acoustic-Based Biosensors

oscillatory-driven 9 or 27 MHz quartz to quantify Advantages of CMUT Devices

DNA-DNA and DNA-protein interactions. They Piezoelectric ultrasonic transducers are used to
used quartz with one side specifically coated and generate and detect ultrasound over the last
the other side rubber cased, which was dipped 30 years. This technology gives a bandwidth as
into a stirred and thermostated solution of the high as 6080 % and a high sensitivity [20]. Nev-
target. Binding amounts were evaluated quanti- ertheless, the piezoelectric transducers have dis-
tatively as mass per unit area. Effects of probe advantages in air or fluid-coupled applications
immobilization techniques, chain lengths, ionic due to the impedance mismatch between the
strengths, temperatures and mismatches on medium and the piezoceramic material. More-
hybridization, regulatory protein binding, and over, it is an expensive technique and shows
electrostatic effects in DNA-binding properties some restrictions for the 2D arrays. On the other
of lysine-containing oligopeptides were evalu- hand, CMUT devices became an attractive alter-
ated quantitatively. In addition, Tessier et al. native to piezoelectric transducers as they have
[18] used a TSM sensor, or specifically an anti- inherently low mechanical impedance associated
body functionalized quartz sensor, to detect with a thin vibrating membrane. Thin membranes
human erythrocytes. Adsorption and desorption that are fully supported with insulating posts on
studies of antibody-conjugated phages at antigen- all sides and these devices have the advantage of
immobilized TSM resonator surfaces in a high- the good mechanical properties of silicone and
throughput setup were also performed [4]. silicon nitride [21]. As the mechanical impedance
of such thin membranes is much closer to those of
Biological Applications of SAW fluids in a wide frequency range, CMUTs are
SAW devices are useful in industrial and commer- better in immersion applications than piezoelec-
cial applications of chemical and biological sens- tric devices. Moreover, CMUTs are fabricated
ing. In biological applications, these devices can using standard micromachining processes and
be used to detect molecules by mass. A potential provide ease of fabrication, batch production,
application of SAW devices in biology is in build- and very high level of integration and repeatabil-
ing immunosensors. In a study, immunosensor ity. Taking advantage of the microfabrication
SAW devices were first coated with a parylene techniques, these devices can provide the broad
thin film to create a chemically homogeneous bandwidth and coupling efficiency required for
layer. Then, photoactive and functional groups high performance. Furthermore, this kind of
containing dextran were immobilized, and anti- ultrasonic transducers have the ability to make
bodies were bonded. This sensor consisted of lith- complex array designs using standard photolith-
ium tantalate substrate with gold transducers and ographic techniques that makes them usable for
used to monitor urease binding [10]. This sensor imaging applications [21]. A transducer with
was applied to monitor the amount of the urea in a wider band increases the resolution as well as
human urine to diagnose the onset of liver and enabling the design of the image processed. In
kidney diseases [10]. In addition, another study addition, CMUTs are an alternative for intravas-
immobilized polysaccharides to a silicon oxide cular imaging systems as they can be
surface intended for SAW. The pneumococcal microfabricated as small 2D imaging arrays and
polysaccharide (PPS) vaccine can immobilize via maintain electroacoustic performance [21].
protein A after pretreatment of the surface with
hydrochloric acid. The results show that the immo- Biological Applications of CMUTS
bilization of PPS via protein A increased the sen- Todays ultrasound technology has a wide range
sitivity of Streptococcus pneumoniae antibodies of applications in industrial and medical field,
detection in human serum when compared with from distance measurement to medical diagnostic
ELISA [19]. This study shows that SAW sensors imaging. The fields such as dermatology, oph-
can be used in detection of microbial contaminants thalmology, and cardiovascular medicine require
in the environment. ultrasonic imaging systems. CMUTs enable
Acoustic-Based Biosensors 37 A
in vivo medical imaging at high frequencies. IVUS imaging catheters [22]. Another study
They are also used in the microfluidics and acous- by Knight et al. [23] modeled and designed
tic sensing, which have applications in the bio- the CMUT membranes for performance in A
medical engineering. forward-looking IVUS. They used finite element
Intravascular ultrasound (IVUS) is a medical analysis to calculate the electromechanical
imaging technique that uses ultrasound equip- coupling coefficient, k2, and showed that the
ment with long, thin catheters and enables visu- CMUT geometry can be optimized for maximum
alization of blood vessels through the blood coupling coefficient, allowing CMUT IVUS
column and inner wall of blood vessels. This array elements to be designed with more than
gives information about cardiovascular diseases. 60 dB/V dynamic range with 10 MHz bandwidth
In addition, a pulse-echo phased array B-scan around 20 MHz by fabricating a 1.02 mm
sector, 1D linear CMUT array was produced by diameter 64-element array [22]. A recent
Demirci et al. [21] to demonstrate the viability of improvement in CMUT technology produces
CMUTs for ultrasound imaging for medical a three-dimensional ultrasound imaging system
applications. 64- and 128-element 1D CMUT using a forward-looking ring array with
arrays with 100 % yield and uniform element a diameter of 2 mm and 64 elements each 100mm
response were fabricated. For imaging experi- 100 mm in size. 3D images of several wire
ments, resolution test phantom was used to targets and Palmaz-Schatz stents indicate the
mimic the attenuation properties of soft tissue. potential of imaging with such a device. In addi-
The image quality was found in order to show the tion, further improvements point to a promising
practicality of using CMUT devices for ultra- future for forward-viewing CMUT ring arrays in
sound medical imaging due to its wide bandwidth intravascular imaging [24].
and high sensitivity. Current l-dimensional (1D) CMUT devices are used in the real-time
side-looking ultrasound imaging probes do not in vivo imaging applications with an improved
have the forward-viewing capability. Demirci performance over the existing medical ultra-
et al. [21] showed forward-viewing annular sound transducers. A wafer-bonded CMUT
CMUT arrays can be used for medical imaging probe showed a bandwidth of 130 %
applications. 64-element, forward-viewing annu- (313 MHz), which covers nearly the entire fre-
lar arrays were fabricated and designed for imag- quency range for medical ultrasound applica-
ing applications in the 20 MHz range. The tions. The carotid artery and thyroid gland were
devices operate in the 526 MHz range with imaged with surface and bulk micromachined
a fractional bandwidth of 135 %. These CMUT arrays, and real-time in vivo images
forward-viewing annular CMUT arrays are were compared with a PZT array. The bandwidth
promising tools for measuring blood flow and and resolution performance of the surface
imaging upon being mounted on the front surface micromachined arrays were slightly better than
of a cylindrical catheter probe [21]. Forward- PZT. However, depth of penetration was not as
viewing IVUS enables new procedures in cardiac good as in PZT. Bulk micromachined CMUT
medicine such as viewing blood vessels, guiding arrays showed the first real-time in vivo images
stent placement, and monitoring operations in the and had improved resolution and bandwidth but
heart [22]. In addition, researchers focused on less depth of penetration. As a result, improved
developing a forward-looking 2D imaging arrays axial resolution and extremely broad bandwidth
that can be microfabricated and integrated with are achieved by CMUTs. Improved axial resolu-
electronics. CMUTs are alternative tools for 2D tion will allow smaller targets to be resolved in
medical imaging arrays. In a study, ring annular tissue for such applications as imaging microcal-
CMUT was used for forward-looking IVUS cifications in the breast, small abnormalities in
imaging at 16 MHz. This 1 mm diameter arrayed the thyroid, or better delineation of the intimal
CMUT device was used for volumetric imaging, linings of blood vessels to search for plaque
and they were found to have potential use in buildup [18].
A 38 Acoustic-Based Biosensors

Future Directions for Research wider areas due to two main limitations.
According to the radar equation, the
Advancements made over the past decades have radiofrequency signal changes with the fourth
culminated in the initiation and development of power of the distance between the radar unit and
acoustic wave biosensors. The ability to detect SAW sensor. As a result, high signal attenuation
biological molecules by these sensors may con- is governed by this equation. Second, only a few
tribute to treatment of many diseases such as HIV echo sounds can be distinguished because all
and cancer. In addition, these devices are prom- sensors in the detection range respond immedi-
ising in medical imaging given their ability to ately. Attaining this requires differentiated sen-
generate high-resolution pictures in real time. sors that respond within different time slots, and
However, there are still limitations to some of multiple sensors can be used with a limited num-
these devices that can be surmounted through ber of reflectors and limited identity-code
future research and though the potential of the capacity [25].
application of these devices to biological situa- In addition, generation of acoustic waves in
tions has been recognized, more work is left to be microstructures requires various processes and
done in order to employ them on a wider scale. technologies, with compatibility as one of the
Below are some limitations and potentials of the main limitations in the development of smart
devices. sensors. Restrictions of manual handling and fab-
rication process limit minimum thickness of the
SAW Devices devices to be about 100 microns. Adoption of
There are a number of potential limitations with silicon fabrication opens the possibility for the
the SAW sensors as they are today. One limita- integration of the acoustic devices and electronic
tion is that SAW sensors are sensitive to a variety driving circuitry, leading to the development of
of parameters [8]. In addition, the intensity of an smart sensors. This technique involves localiza-
acoustic wave can decrease in biological solution tion of specific fabrication steps to common areas
such that the sensing system may not work effi- to minimize the total number of process steps in
ciently. Furthermore, many SAW devices that are the fabrication process as well as to maintain
used in the gas-sensing applications may not be SAW and Lamb wave device integrity and
used to detect molecules in the liquid phase. quality [26].
When a fluid contacts the SAW surface, the lon- Most of todays acoustic wave generation and
gitudinal wave component pumps a compression focusing apparatus suffer from these further lim-
wave into the fluid [2]. In air, the energy losses itations: the piezoelectric arrays are expensive,
are not too effective. However, in many liquids, and it is hard to get high-level, well-distributed
all the energy is pumped out of the acoustic wave intensities. In addition, shocks are generated at
before it reaches the devices receiving trans- a rate of one or two shocks per second, whereas
ducer. SAW devices can perform well in liquids extracorporeal shockwave treatment (ESWT)
when they rely on surface waves with no longi- typically requires thousands of shocks per treat-
tudinal components like horizontally polarized ment. Acoustic lenses are mostly fragile and
shear waves, surface transverse waves, and the noneffective for larger apertures. Moreover, elec-
guided acoustic waves called love waves [2]. trohydraulic approach has disadvantages of
Recent advances in liquid sensors and immuno- nonuniform discharges, pain, and high noise
assays improved the detection limits. In addition, level.
surface technologies like functionalization with
fluorescent probes, DNA, and enzymes increased BAW Devices: TSM Resonator
the use of SAW devices in liquid samples The limitations of TSM resonators are mostly
[7]. There are also radar-type SAW sensors for related with the experimental setup, accuracy in
high-voltage applications that detect at long dis- modeling the sensor, and physical behavior of
tances. These sensors cannot be used scans of a viscoelastic polymer. Researchers should be
Acoustic-Based Biosensors 39 A
careful in model approximations, parameter esti- medium during nucleic acid analyses. Complex-
mations, and measurement errors. For acousti- ation of DNA may limit the flexibility of the
cally thin polymers, the thickness can be molecule strands and ionic strength changes or A
calculated with a small error, but the shear moduli interaction with the surface might vary extension
can hardly be extracted from measurements due and shape of the molecule. These may change
to negligible influence on the sensor behavior. On viscoelastic behavior and produce erroneous
the other hand, the coating thickness should be results from TSM resonators [4].
known because of the existence of multiple iden-
tical solutions for acoustically thick polymers CMUTs
[27]. In addition, for detailed interpretation of CMUTs have the potential to make ultrasound
sensor response in liquid requires more sophisti- a far more versatile and important imaging
cated experimental setup [4]. In addition, modality. This technology has made advances in
mechanical, optical, and electrochemical proper- acoustic matching and the microminiaturization
ties are helpful to characterize interactions at of electronics. It enables higher-frequency imag-
biological surfaces and interfaces. However, the ing, allows clinicians to view small features in the
needed equipment are expensive. While minia- body, and is useful for ultrasound imaging espe-
turized quartz arrays seem to present a solution cially in the area of volumetric in vivo imaging,
for low-cost sensor setups, miniaturization has its intravascular, and intracardiac research.
limitations because small electrode areas can
load enough molecules to achieve measurable
frequency shifts. As a solution, combination of Cross-References
automatized pipetting systems, microfluidics,
integrated sensor arrays, and multicomponent Biosensor
data analysis may open the way toward high- Biosensors Using Atomic Force Microscopes
throughput screening with acoustic wave Biosensors Using Lasers
sensors [4]. Biosensors Using Magnetics
Two major challenges are present in the appli- Biosensors Using Surface-Enhanced Raman
cation of TSM resonator as a biosensor. First, Scattering
cross-reactions such as unspecific protein adsorp- Piezoelectric
tion affect the accuracy of the results. This limits
the resonators use to binding sites that specifi-
cally bond to the targeted biological molecules
and poses a significant disadvantage in environ- References
ments containing various compounds. Second,
1. Lec RM, Lewin PA (1999) Acoustic wave biosensors,
since the resonator only responds to mass load- engineering in medicine and biology society, 1998.
ings, the analysis cannot differentiate this signal In: Proceedings of the 20th annual international con-
from other factors that can affect the resonant ference of the IEEE, vol 6, pp 27792784
2. Drafts B (2001) Acoustic wave technology sensors.
frequency such as viscoelasticity and acoustic
IEEE Trans Microw Theory Tech 49(4):795802
coupling effect. This leads to a disagreement 3. Piezoelectricity Encyclopdia Britannica
between the experimental and theoretical values (2006) Encyclopdia Britannica online, http://
of mass uptake, an error that needs to be 25 Aug 2006
4. Kaspar M et al (2000) Thickness shear mode resona-
accounted for by researchers [4]. Immunochemi- tors (mass-sensitive devices) in bioanalysis.
cal analyses or nucleic acid hybridization results Fresenius J Anal Chem 366(67):602610
of TSM resonators should always be coupled 5. Sauerbrey G (1959) Use of quartz crystal vibrator for
with a reference analysis method. In addition, weighting thin films on a microbalance. Z Phys
dry- and wet-coated TSM resonators may not
6. Lin Z, Yip CM, Joseph IS, Ward MD (1993) Opera-
give the same results due to changes in viscoelas- tion of an ultrasensitive 30-MHz quartz crystal micro-
ticity, so calibration should be done in the sensing balance in liquids. Anal Chem 65:15461551
A 40 Active Mixer

7. Dahint R et al (1994) Acoustic plate mode sensor arrays for medical imaging. IEEE Trans Ultrason
for immunochemical reactions. Anal Chem Ferroelectr Freq Control 55:887895
66(18):28882892 22. Yeh DT, Oralkan O, Wygant IO, ODonnell M,
8. Milstein L, Das P (1977) Spread spectrum receiver Khuri-Yakub BT (2006) 3-D ultrasound imaging
using surface acoustic wave technology. IEEE Trans using a forward-looking CMUT ring array for intra-
Commun 25(8):841847 vascular/intracardiac applications. IEEE Trans
9. McGill RA, Chung R, Chrisey DB, Dorsey PC, Ultrason Ferroelectr Freq Control 53(6):12021211
Matthews P, Pique A, Mlsna TE, Stepnowski JL 23. Knight JG, Degertekin FL (2002) Capacitive
(1998) Performance optimization of surface acoustic micromachined ultrasonic transducers for forward
wave chemical sensors. IEEE Trans Ultrason looking intravascular imaging arrays. In: Ultrasonics
Ferroelectr Freq Control 45(5):13701380 symposium, 2002, proceedings, vol 2, pp 10791082
10. Lange K, Bender F, Voight A, Gao H, Rapp M (2003) 24. Jakubik W, Urbanczyk M, Opilski A (2001) Sensor
A surface acoustic wave biosensor concept with low properties of lead phthalocyanine in a surface acoustic
flow cell volumes for label-free detection. Anal Chem wave system. Ultrasonics 39(3):227232
75(20):55615566 25. Scholl G, Schmidt F, Wolff U (2001) Surface acoustic
11. Campbell CK (1984) Simplified computer-aided fab- wave devices for sensor applications. Phys Stat Sol
rication of SAW transducer pattern masks. IEEE A 185(1):4758
Trans Sonic Ultrason 31(3):185186 26. Hoummady M, Campitelli A, Wlodarski W (1997)
12. Chen Y, Emanetoglu NW, Saraf G, Wu P, Lu Y, Acoustic wave sensors: design, sensing mechanisms
Parekh A, Merai V, Udovich E, Lu D, Lee DS, and applications. Smart Mater Struct 6:647657
Armour EA, Pophristic M (2005) Analysis of SAW 27. Behling C, Lucklum R, Hauptmann P (1997) Possi-
properties in ZnO/Al/sub x/Ga/sub 1-x/N/c-Al/sub bilities and limitations in quantitative determination
2/O/sub 3/structures. IEEE Trans Ultrason Ferroelectr of polymer shear parameters by TSM resonators. Sens
Freq Control 52(7):11611169 Actuators A 61:260266
13. Ergun AS, Yaralioglu GG, Khuri-Yakub BT 28. Degertekin LF, Guldiken OR, Karaman M (2005)
(2003) Capacitive micromachined ultrasonic trans- Micromachined capacitive transducer arrays for intra-
ducers: theory and technology. J Aerosp Eng vascular ultrasound. Proc SPIE Int Soc Opt Eng
16:7684 5721:104
14. Huang Y, Ergun AS, Hggstrom E, Badi MH, Khuri-
Yakub BT (2003) Fabricating capacitive
micromachined ultrasonic transducers with wafer-
bonding technology. J Microelectromech Syst
12:128137 Active Mixer
15. Liren W, Xue W, Sixin W, Xuesong Y, Yuquan C
(1996) A two-acoustic-waveguide-channel SH-APM Chih-Chang Chang1, Lung-Ming Fu2 and
urease biosensor. Eng Med Biol Soc 1:8586
Ruey-Jen Yang1
16. Davis KA, Leary TR (1989) Continuous liquid-phase 1
piezoelectric biosensor for kinetic immunoassays. Department of Engineering Science, National
Anal Chem 61(11):12271230 Cheng Kung University, Tainan City, Taiwan
17. Okahata Y, Matsunobu Y, Ijiro K, Mukae M, 2
Department of Materials Engineering, Pingtung,
Murakami MK (1992) Hybridization of nucleic
acids immobilized on a quartz crystal microbalance.
J Am Chem Soc 114:82998300
18. Tessier L et al (1997) Potential of the thickness shear
mode acoustic immunosensors for biological analy- Synonyms
sis. Anal Chim Acta 347:207217
19. Dutra RF, Castro CMHB, Azevedo CR, Vinhas E,
Malague E, Melo EHM, Lima Filho JL, Kennedy JF Active mixing; Dynamic mixing
(2000) Immobilization of pneumococcal
polysaccharide vaccine on silicon oxide wafer for an
acoustical biosensor. Biosens Bioelectron
20. Caliano G, Carotenuto R, Cianci E, Foglietti V,
Caronti A, Iula A, Pappalardo M (2005) Design, fab- The term active mixer or active micromixer
rication and characterization of a capacitive refers to a microfluidic device in which species
micromachined ultrasonic probe for medical imaging.
mixing is enhanced by the application of some
IEEE Trans 52(12):22592269
21. Demirci U, Ergun AS, Oralkan O, Karaman M, form of external energy disturbance. Typically,
Khuri-Yakub BT (2004) Forward-viewing CMUT this disturbance is generated either by moving
Active Mixer 41 A
components within the micromixer itself, and thus turbulence does not occur. The Reynolds
e.g., magnetically actuated stirrers, or by the number of microfluidic flows in a conventional
application of an external force field, e.g., pres- straight microchannel is typically less than one, A
sure, ultrasound, acoustic, electrohydrodynamic, and the resulting Stokes flow is unidirectional.
electrokinetic, dielectrophoretic, magnetohydro- Even in curved or twisted microchannels, the
dynamic, thermal, and so forth [1, 2]. transverse component of the flow (i.e., the second-
ary flow) is very small compared to the streamwise
component since the inertial effects induced by the
Overview microchannel geometry are very weak. Therefore,
when the microchannels contain two or more fluid
Mixing occurs in many natural phenomena, streams, mixing of the different streams is domi-
including geophysical, ocean, and atmospheric nated by molecular diffusion effects. In microscale
flows, and is also an important step in industrial channels, the diffusive mixing process is very slow
processes involving chemical reactions, fermenta- compared with the rate at which the fluid is
tion reactions, combustion, and so forth. Tradition- convected along the channel. Typically,
ally, industrial mixing applications have always microfluidic flows have a high Peclet number,
been performed using large-scale apparatus. How- i.e., Pe = Ul/D, where U is the average flow
ever, in recent years, microscale devices velocity, l is the channel dimension, and D is the
(commonly referred to as microfluidic devices) molecular diffusivity. The diffusive mixing time is
have been proposed as a means of constructing given by tD ~ l2/D, and the mixing channel length
micro-total analysis systems (m-TAS) and lab-on- required to achieve an adequate species mixing
a-chip (LOC) devices. However, achieving a rapid increases linearly with the Peclet number
and efficient mixing of different reactants when (Lm
Pe l). For example, for an average flow
performing chemical and biological analyses in velocity of 500 mm/s, a channel dimension of
such microfluidic devices is highly challenging. 100 mm, and a molecular diffusivity of 10 mm2/s,
Therefore, the problem of developing enhanced the mixing time and length required to achieve
micromixing schemes suitable for microfluidic a complete species mixing are of the order of
applications has attracted significant interest 1,000 s and 0.5 m, respectively. However, many
within industrial and academic circles in recent practical biochemical applications, e.g., immuno-
years. As a result, the literature now contains assays and DNA hybridization, require the rapid
a large number of studies concerning micromixers mixing of macromolecules characterized by rela-
implemented using a variety of mixing strategies tively low molecular diffusivity. Therefore, the
[1, 2]. The micromixing schemes are categorized mixing performance obtained in typical
as either active or passive. The active mixing has microchannels is unsuitable for biochemical anal-
been defined in the above section. The passive ysis applications. Accordingly, a requirement
mixing refers to the mixing effect in systems exists for enhanced microfluidic mixing schemes
enhanced by virtue of their particular geometry to support the development of m-TAS and LOC
topologies, surface properties, or instability phe- devices. Typically, such schemes involve reducing
nomenon which occurs naturally under a static the diffusion length (i.e., l) and increasing the
(DC) electric field. The active mixing is easily interfacial contact area between the mixing
controlled using external mechanisms. Active species.
mixers are eminently suitable components for Although it is difficult to induce turbulence
reconfigurable microfluidic systems. (so-called Eulerian chaos) in microchannels, the
The characteristic size of the microchannels in mixing performance obtained in low Reynolds
microfluidic devices is very small (typically of the number flow regimes can be enhanced via the
order of several 100 mm), and hence the flow is chaotic advection mechanism (or so-called
dominated by viscous forces. As a result, the flow Lagrangian chaos). Chaotic advection occurs in
is restricted to the low Reynolds number regime, regular, smooth (from a Eulerian viewpoint)
A 42 Active Mixer

flows [3] and provides an effective increase in and by a momentum equation modified to include
both the interfacial contact area and the concen- the body force term, i.e.,
tration gradient by inducing a reduction in the
striation thickness and hence in the diffusion u 0 (1)
length. As a result, both the mixing time and the
mixing length are considerably reduced. In the and
case of a linear reduction in the striation thickness

within the flow path, the mixing time and mixing @u

rf uu p m2 u F, (2)
length exhibit a sublinear dependence on Pe. The @t
two parameters can be reduced to as much as
tm ~ ln(Pe) and Lm ~ ln(Pe), respectively, for where t, u, p, rf, and m denote the time, the fluid
chaotic flows in the limit of large Pe [3, 4]. Impor- velocity, the pressure, the fluid density, and the
tantly, chaotic advection cannot occur in steady, fluid viscosity, respectively. In Eq. 2, F represents
two-dimensional flows, but is restricted to the external driving force (i.e., the external
time-dependent two-dimensional and three- energy source) associated with the specific active
dimensional flows. Accordingly, a small cylindri- micromixer. For example, in an
cal rotor spinning about its center with a electrohydrodynamic/electrokinetic-based active
uniform angular frequency (i.e., Couette flow) mixer, F represents the Coulombic force pro-
fails to provide an efficient micromixing perfor- duced by the presence of a free space charge,
mance, whereas two such rotors turned on and reE, where re is the net free charge density in
off alternatively create a blinking vortex phe- the bulk liquid or at the solidliquid interface
nomenon which yields a powerful mixing effect (i.e., the electrical double layer) and E is
[3]. In Couette flow, the fluid streams are only the intensity of the externally applied DC or
stretched. Hence, chaotic advection cannot be AC electrical field. Meanwhile, for
induced, and thus the mixing effect is very poor. a magnetohydrodynamic mixer, F is the Lorentz
However, in the two-rotor arrangement, the force induced by the coupling between the mag-
fluid streams are repeatedly stretched around netic field B and the electrical field E (J B,
one rotor and then folded around the other. As where J is the current density). In the literature,
a result, chaotic advection is induced and a good Eqs. 1 and 2 are generally solved using well-
mixing performance is obtained. In general, known numerical methods such as the finite dif-
achieving an effective mixing of two or ference method (FDM), the finite volume method
more fluid streams always requires a repeated (FVM), the finite element method (FEM), and so
stretching and folding of the fluid elements. In forth.
recent years, various active mixing approaches In typical microfluidic applications, the Reyn-
have been proposed using chaotic advection to olds number is much lower than unity (i.e., Re =
enhance the mixing effect. The chaotic flows rf Ul/m 1), and hence the flow can be approx-
induced by such active approaches are time- imated as a creeping flow (i.e., Stokes flow).
dependent and are easily controlled using exter- Therefore, the convection term in Eq. 2 can be
nal mechanisms. As such, active mixers are emi- neglected, and the time-dependent flows in
nently suitable components for reconfigurable active micromixers are then governed by Eqs. 1
microfluidic systems. and 3, i.e.,

Basic Methodology rf p m2 u F: (3)

Flow Field Additionally, provided that the time required

Applying the continuum assumption, microfluidic to reach the viscous diffusion time scale is
flows can be modeled by the continuity equation much less than the oscillation time scale (o1),
Active Mixer 43 A
i.e., rf ol2/m 1, the time derivative term in different times can be obtained simply by inte-
Eq. 3 can be neglected, and therefore the grating Eq. 6. Many suitable integration methods
unsteady flow fields in active micromixers can are presented in the literature, ranging from the A
be regarded as quasi-steady Stokes flow simple first-order Euler method to the fourth-
described by Eqs. 1 and 4, i.e., order RungeKutta method, or even higher-
order methods. Besides simulating the dispersion
0 p m2 u F: (4) of a blob of samples, the particle tracing method
also enables the stretching and folding of fluid
Visualization of Mixing Process elements in an unsteady flow field to be clearly
Numerical studies relating to active micromixers observed.
generally visualize the mixing process and eval-
uate the mixing performance by solving the fol- Characterization of Chaotic Mixing
lowing convection-diffusion equation: In order to optimize the fluid deformation phe-
nomena in active micromixers during the mixing
@C process, it is first necessary to understand the
uC D2 C (5)
@t fluid stretching and folding mechanisms which
take place and to evaluate them in a quantitative
in which C and D represent the sample concen- manner. Over the past few decades, dynamic
tration and the diffusion coefficient, respectively. system analysis techniques such as Poincare
However, there are several drawbacks to this maps and Lyapunov exponents have revealed
approach. For example, if the deformation of that the repeated stretching and folding of fluid
the species interface is complex and the mixing elements results in a chaotic mixing effect.
channel is very long, the equation solution may (Interested readers are referred to the text edited
fail to identify the interface precisely due to by Ottino [3] for details regarding the application
numerical diffusion and natural species diffusion. of Poincare maps and Lyapunov exponents to the
Furthermore, in large Peclet number flows, con- exploration of mixing kinematics). The informa-
vective transport dominates the diffusive mecha- tion provided by Poincare maps and Lyapunov
nism and the sample dispersion in the flow field exponent analysis enables a thorough under-
can be greatly enhanced by promoting the con- standing of the characteristics of micro-active
vective transport. In such diffusion-limited cases, mixing systems. Specifically, Poincare maps can
the diffusion effects can be neglected, and the be used to explain the transitions of the mixing
sample simply convects with the fluid along the system behavior from periodic or quasi-periodic
local instantaneous streamlines. Under these to chaotic motion, while the Lyapunov exponents
conditions, the Lagrangian particle tracing tech- provide an indication of the strength of the cha-
nique, which introduces mass-less, non- otic behavior and can be used to quantify the
interacting passive particles into the flow fields, effects of the amplitude and frequency of the
provides a satisfactory means of observing the perturbations on the chaotic response. Collec-
advection phenomenon within the fluid streams tively, the data provided by Poincare maps and
without solving Eq. 5. The motion of a tracer Lyapunov exponents provide sufficient informa-
particle can be tracked by the following kine- tion to optimize the design and operation of
matic equation (dynamic system): micro-active mixing devices.

ur, t, (6)
dt Key Research Findings

where r is the location of the tracer particle and Many active mixing approaches have been pro-
u(r, t) is the Eulerian velocity field. The tracer posed in recent years. Broadly speaking these
particle position in the unsteady flow field at approaches can be classified in terms of whether
A 44 Active Mixer

Active Mixer,
Fig. 1 Pressure
perturbations induce lobe-
like distortions of the
interface and facilitate
rapid mixing in the device
with one pair of side

or not they use moving parts to generate a mixing exponent data were applied as mixing indices to
effect. In those micromixers which do use mov- evaluate the intensity of the chaotic mixing
ing parts, mixing is commonly achieved using effect. The optimal operating parameters were
piezoelectric membranes, gas bubbles, and so then determined by plotting the Lyapunov expo-
forth. However, since the integration of these nent data against the amplitude and frequency of
moving parts within the overall microfluidic sys- the perturbation.
tem is challenging, and therefore expensive,
researchers have paid increasing attention to Electrohydrodynamic (EHD) Mixer
active micromixers containing no moving ele- In electrohydrodynamic (EDH) mixers, a mixing
ments. Numerous proposals for active mixing effect is created by applying an external electrical
approaches have been presented in recent years, field to a bulk flow containing both an electrical
and space constraints prevent their detailed conductivity gradient and a permittivity gradient.
discussion here. Hence, the following provides In the micromixer proposed by Moctar et al. [6],
no more than a high-level review of some of two fluids with identical viscosity and density but
the more fundamental techniques. For a compre- different electrical properties were injected into
hensive discussion of active mixing schemes, the mixing channel by syringe pumps, and the
interested readers are directed to the review electrodes were arranged in such a way that the
presented in [1, 2]. electrical field was perpendicular to the species
interface, as shown in Fig. 2a, creating a transver-
Pressure Perturbation Mixer sal secondary flow. The mixing effects induced
In pressure perturbation mixers, the species are by both DC and AC electrical fields were
perturbed by pulsing their velocities in some way. explored in a series of experimental investiga-
For example, Niu and Lee [5] presented an active tions. From Fig. 2b, c, d, it can be seen that the
mixer comprising a main channel and multiple mixing becomes more chaotic as the intensity of
side channels, in which the mixing species were DC electric field increases. The results revealed
stirred by pulsing the velocities of the fluids that the application of an appropriate voltage and
introduced through the side channels, as shown frequency to the electrodes yielded a satisfactory
in Fig. 1. The resulting stretching and folding of mixing performance within less than 0.1 s over
the fluid elements in the main and side channels a short mixing distance, even at Reynolds num-
induced a chaotic advection mixing effect. bers as low as 0.02.
A detailed particle simulation study was
performed to examine the respective effects of Electrokinetic-Based Mixer
the amplitude and frequency of the flow pertur- Oddy et al. [7] presented a micromixer in which
bations on the species mixing performance. AC electrical fields were used to induce the elec-
Poincare maps, constructed using the simulation trokinetic instability (EKI) phenomenon in
data, showed that for a given parameter set, the microscale channels at very low Reynolds num-
fluid flow exhibited both chaotic and quasi- bers, as shown in Fig. 3. EKI is similar to
periodic responses. Additionally, Lyapunov electrohydrodynamic instability, which can be
Active Mixer 45 A
Active Mixer,
Fig. 2 From [6] (a)
Configuration of
electrohydrodynamic A
mixer. Experimental
images of mixing (b) initial
condition, (c) after the
application of a DC electric
field of intensity
E = 4 105 V/m and
(d) E = 6105 V/m

induced by either AC or DC electrical fields when

an electrical conductivity gradient exists in the
bulk flow. The results showed that for a flow rate
of 0.5 ml/min, the EKI flow field resulted in
a rapid stretching and folding of the fluid and
achieved an effective mixing performance within
3 s, corresponding to less than 1 % of the time
required to achieve mixing via diffusion effects
in a conventional straight microchannel.
Fu et al. [8] periodically switched an external
electrical field to carry out species mixing in
a T-form electrokinetic microfluidic mixer by
generating electrokinetic perturbation forces.
The authors showed that the mixing performance
of the microfluidic mixer was enhanced by
increasing the contact area and contact time of
the mixing fluids and by creating irregular flow
fields in the mixing channel. However, the results
also indicated that an adequate switching fre-
quency was required to achieve a satisfactory
mixing performance, as shown in Fig. 4.
Applications of time-wise periodic alterations
of the zeta potential along the microchannel walls
to induce an electroosmotic chaotic advection
effect can improve mixing. It was shown that
both a spatial and a temporal control of the zeta
Active Mixer, Fig. 3 (a) Configuration of electrokinetic potential could be achieved by applying a field
instability mixer and (b) time-stamped experimental effect (i.e., a capacitive effect). The effect of
images of mixing process periodically switching between several different
A 46 Active Mixer

Active Mixer,
Fig. 4 Numerical and
experimental images of
mixing for different
switching frequencies for
a constant driving electric
field of 100 V/cm

flow fields at various periods was analyzed by wire electrodes and the circular side-wall elec-
performing particle tracking simulations (i.e., trode and between the second wire electrode and
Poincare maps). The results indicated that the side-wall electrode, as shown in Fig. 5. The
a more chaotic mixing effect was achieved at particle tracing results revealed that the intensity
lower values of the switching frequency. of the chaotic mixing effect increased with
increasing T, resulting in a satisfactory mixing
Magnetohydrodynamic (MHD) Mixer result within 40 periods.
Various researchers have employed the magneto-
hydrodynamic (MHD) effect to implement
micromixers. For example, Bau et al. [9] devel- Future Directions for Research
oped an active micromixer which used either DC
or AC electrical and magnetic fields to generate It seems likely that future micromixer designs
Lorentz forces with which to stretch and fold the will continue the current trend for implementing
species fluids in a mixing chamber. The results species mixing without the use of moving
indicated that an adequate mixing result was mechanical parts. Furthermore, it seems reason-
achieved within several seconds. A more sophis- able to speculate that the use of electrokinetic
ticated MHD mixer was developed by Yi forces to induce a mixing effect will become
et al. [10] based upon the two blinking vortex increasingly common since this technique greatly
arrangement used to induce chaotic advection. simplifies the microfabrication process and
The proposed micromixer comprised a small reduces the cost and complexity involved in
cylindrical mixing chamber with an electrode embedding active mixers within microfluidic sys-
deposited on its side wall and two copper-wire tems. Electrokinetic forces have been widely
electrodes positioned eccentrically on the lower employed in active mixing schemes in recent
surface of the chamber. The chamber was orien- years, a case in point being the EKI mixer
tated such that its axis was parallel to the direc- presented by Oddy et al. in [6]. However, current
tion of an applied magnetic field, and a mixing EKI-based mixers suffer the drawback of a high
effect was generated by applying a potential dif- electrical voltage requirement. Accordingly,
ference of period T alternately between one of the low-voltage, AC electrokinetic techniques are
AFM 47 A

1. Nguyen NT, Wu Z (2005) Micromixers a review.

J Micromech Microeng 15:R1R16 A
2. Chang CC, Yang RJ (2007) Electrokinetic
mixing in microfluidic systems. Microfluid Nanofluid
3. Ottino JM (1989) The kinematics of mixing:
stretching, chaos, and transport. Cambridge Univer-
sity Press, Cambridge
4. Stroock AD, Dertinger SKW, Ajdari A, Mezic I,
Stone HA, Whitesides GM (2002) Chaotic mixer for
microchannels. Science 295:647651
5. Niu X, Lee YK (2003) Efficient spatial-temporal cha-
otic mixing in microchannels. J Micromech Microeng
6. El Moctar AO, Aubry N, Batton J (2003) Electro-
hydrodynamic micro-fluidic mixer. Lab Chip
7. Oddy MH, Santiago JG, Mikkelsen JC (2001) Electro-
kinetic instability micromixing. Anal Chem
8. Fu LM, Yang RJ, Lin CH, Chien YS (2005) A novel
microfluidic mixer utilizing electrokinetic driving
forces under low switching frequency. Electrophore-
sis 26:18141824
9. Bau HH, Zhong J, Yi M (2001) A minute magneto
hydro dynamic (MHD) mixer. Sens Actuators
B 79:207215
10. Yi M, Qian S, Bau HH (2002) A minute magneto
Active Mixer, Fig. 5 Poincare maps (left column) and hydro dynamic (MHD) chaotic stirrer. J Fluid Mech
flow visualization photographs (corotating) [10]. For dif- 468:153177
ferent periods of (a) T = 4, (b) T = 10, and (c) T = 40

expected to receive increasing attention in the

coming years. Finally, it is known that the high AFM
flow rates required to achieve species mixing can
be produced through various nonlinear electroki- Definition
netic phenomena. Therefore, the application of
nonlinear electrokinetic techniques to realize The atomic force microscope (AFM) is a very
active mixers is likely to emerge as a major high-resolution type of scanning probe micro-
research topic in the microfluidics community in scope, with demonstrated resolution of fractions
the near future. of an Angstrom, more than a 1,000 times better
than the optical diffraction limit. The AFM is
one of the foremost tools for imaging, measur-
Cross-References ing, and manipulating matter at the nanoscale.
The AFM consists of a microscale cantilever
AC Electro-Osmotic Flow with a sharp tip (probe) at its end that is used to
Chaotic Flows scan the specimen surface. When the tip is
Electrokinetic/ElectroHydrodynamic Flow brought into proximity of a sample surface,
Instability forces between the tip and the sample lead to
Electroosmotic Flow (DC) a deflection of the cantilever according to
Microfluidic Mixing Hookes law. Depending on the situation, forces
A 48 Aging

that are measured in AFM include mechanical Definition

contact force, Van der Waals forces, capillary
forces, chemical bonding, electrostatic forces, The amperometric biosensor has emerged as an
magnetic forces, etc. Typically, the deflection efficient way for the detection, monitoring, and
is measured using a laser spot reflected from reporting of biochemical analytes related to
the top of the cantilever into an array of photo- a wide range of pathologies. By measuring the
diodes. Other methods that are used include current change, the conductivity or concentration
optical interferometry, capacitive sensing or is obtained. Microfluidic lab-on-a-chip technol-
piezoresistive AFM probes. ogy is of miniaturization, low-sample consum-
ing, higher sensitivity, and automation. By
combination of the two technologies, it can pro-
Cross-References vide a way for point of care diagnostics using the
unprocessed samples such as whole blood. On the
Atomic Force Microscope (AFM) other hand, the advancement of CMOS technol-
Biosensor ogy makes the interface between microfluidics
Biosensors Using Atomic Force Microscopes and electronics more versatile. One recent
Scanning Probe Microscopy research paper even reports the success of amper-
ometric platform with 400 measurement points
for electrochemical bio-imaging and multi-point
bio-sensing. In this short entry, the achievements
and perspectives of amperometric microfluidic
biosensor are introduced.
Aging is known in material science as material
fatigue. It is a permanent structural damage that
In recent years, amperometric-based microfluidic
occurs when a material is subjected to cyclic or
sensor has been demonstrated as a highly
fluctuating strains at nominal stresses that have
sensitive method for biomolecules detection.
maximum values less than (often much less
Combination of microfluidic techniques and
than) the static yield strength of the material.
amperometric component enables dynamic mon-
These damages can although be effected by
itoring of the target molecules.
thermal stresses due to continuous heating and
Basic Methodology

Amperometry in chemistry and biochemistry is

Amperometric Detection the detection of ions in a solution based on elec-
tric current or changes in electric current.
Jinhong Guo and Yuejun Kang Electrochemistry-based amperometric sensor is
School of Chemical and Biomedical a detection technique wherein a known voltage
Engineering, Nanyang Technological University, difference is applied to the working electrodes
Singapore which responds with an initial transient current
[1]. The current is dependent on the activity of
redox species at the electrodes interface. At the
Synonyms beginning, the current will leap from the base
current I1, followed by falling to a steady-state
Amperometric; Microfluidics value I2, which is determined by the bulk
Amperometric Detection 49 A
electrode surface. Assuming that the sufficient
potential is applied to the working electrode, in
the shortly subsequent time, the responding cur- A
rent can be approximately expressed as [3]

nFAD 1
I C (2)

Consequently, the Eq. 2 can illustrate that the

responding current should have a linear relation-
ship with the analyst concentration. Based upon
this mechanism, electrochemistry-based ampero-
metric sensors are developed for the detection of
concentration, biomarker, etc.
On the other hand, poly(dimethylsiloxane)
(PDMS) has emerged as the most widely used
material in micro-/nanofluidic applications.
PDMS can be tightly bonded to glass or silicon
wafer by oxygen plasma treating on the bonding
surfaces. It enables low-cost micro-/nanofluidic
device fabrication. Since it is highly flexible, dis-
posable, and inexpensive and has a good bonding
ability with glass or wafer, PDMS is the privilege
material for rapid pattern of disposable microchips
which enables low-cost point of care diagnosis.
Recently, PDMS-based amperometric sensors are
Amperometric Detection, Fig. 1 Voltage applied to extensively studied by researchers for the various
electrodes starts with V1 in which case no chemical reac- applications. For example, Ney et al. presented
tion occurs and enhances to V2. Subsequently, the elec- a portable device in which multichannel
trode responds to the excitation with current change. After microfluidic chip is fabricated by PDMS on the
the time t, the current falls to a steady state reproduce
from [1] glass slides and on which the working electrodes
and reference electrodes are integrated for amper-
ometric detection [3]. The design is based on
chemical potential of electroactive species. The a single working electrode, which is shared by an
working principle is shown in the Fig. 1. array of five microchannels, allowing fast and
Assuming that the planar working electrode reliable analytical procedures as shown in Fig. 2.
area A, the electroactive analyst concentration N, The experimental result related to this device is
the current at the electrodes I can be expressed shown in Fig. 3.
as [2]: Figure 3a shows a current change profile when
the solution contains the electroactive species
n F A D 1 instead of the supporting electrolyte. The current
I C  C0 (1) is observed with its reduction to a stable level
determined due to the diffusion of the
where n denotes the number of electrons partici- electroactive species to the surface of the elec-
pating in the reaction, F is the Faraday constant trode. On the diffusional condition, a linear fitting
(96,485 C mol/l), D is the diffusion coefficient of is applied to obtain the approximated linear
the solution, d is the thickness of the Nernst layer, relationship between current change and
and C 0 is the solution concentration at the electroactive species concentration, as shown in
A 50 Amperometric Detection

Amperometric Detection, Fig. 2 Five independent

channels all three electrodes are positioned on-channel.
Electrode and channel widths are 50 and 100 mm, respec-
tively, and the channel height is 50 mm. WE, RE, and CE
stand for working, reference, and auxiliary electrodes,
respectively reproduce from [3]

Fig. 3b. It provides a low-cost amperometric sen-

sor for detection of electrolyte concentration with
self-calibration. One channel is injected with
Amperometric Detection, Fig. 3 (a) Representative
standard solutions for calibration, while the
current response for the oxidation of 1.0 mmol/L1 hydro-
others are for the injection of solutions with quinone (HQ) in 0.1 mol L NaNO3 using a microchip with
unknown concentration which remains to be design 1. To obtain this current profile, channels were first
detected. filled with 0.1 mol L 1 NaNO3 electrolyte. The supporting
electrolyte was then sequentially replaced by solutions
The electrochemistry-based microfluidic
containing HQ. (b) Current differences between injections
amperometric sensor provides a powerful plat- (quadruplicate), measured at the diffusion condition, pre-
form for the m-TAS for point of care diagnosis sent a linear relationship in the range of 1.04.0 mmol/L
due to its natural properties like disposability, f HQ. Other conditions, three Au electrode cell; potential,
+800 mV against Au pseudo-reference reproduce from [3]
low cost, easy fabrication, etc. The applications
relating to it have been extended to liquid chro-
matography, immunoassay, and determination of a large-scale basis by the standard CMOS tech-
biomolecules like peroxide and nucleic acids. nology. Conventional process of signal
processing module is incorporated outside of the
chip, which this methodology may introduce the
Key Research Findings extra electric noise between the sensing elec-
trodes and the amplifier in the chip. This leads
LSI-Based Amperometric Sensor to a decreasing sensitivity. In order to enhance the
With the advanced development of complemen- sensitivity of LSI amperometric sensor, one strat-
tary metal-oxide-semiconductors (CMOS) tech- egy is developed to reduce the electronic noise by
nology, large-scale integration (LSI) makes the integrating the external amplifiers with the elec-
parallel multi-point bio-sensing possible. The trodes chip. This fully system-based LSI amper-
working electrodes and reference electrodes of ometric sensor provides a powerful platform for
amperometric sensor are able to be integrated high-throughput point of care healthcare diagno-
on a tiny chip (e.g., 10.4 10.4 mm) on sis, environmental monitoring, and biochemical
Amperometric Detection 51 A

Amperometric Detection, Fig. 4 (A) Illustration of the (b) is 40 mm in diameter. (C) Photograph of Bio-LSI chip.
cross section of a unit cell. The surface of the LSI was After wire bonding, the conductive wires were insulated
modified by metal (Au/Pt/Ti) and insulated by epoxy by PDMS. (D) Photograph of Bio-LSI chip unit. The
polymer (SU-8 3005) to define the electrode area exposed acrylic wall for the sample solution is set on the substrate.
to the solution. (a) Contact hole to expose Al pad. (b) (E) The photograph showing the experimental setting of
Sensing point. (B) Microgram of unit cells. Al pads (a) the Bio-LSI chip unit, RE and CE. (F) Complete view of
can be seen under the Au layer. The size of a sensing point the settings for the Bio-LSI system reproduce from [4]

analysis. Recently, LSI-based amperometric sen- Bio-LSI is a promising platform for a wide
sor has been demonstrated for bio-imaging appli- range of analytical fields, including diagnostics,
cation. Kumi Y. Inoue et al. developed an environmental measurements, and basic bio-
LSI-based amperometric sensor with 400 mea- chemistry. This chip has a mature manufacture
surement points as a platform for electrochemical technology and is able to be mass produced in
bio-imaging and multi-point bio-sensing [4]. The industry. Consequently, the LSI-based ampero-
system consists of a 10.4 610.4 mm CMOS metric sensor is promising and has great commer-
sensor chip with 20 20 unit cells. The whole cial interests for bio-detection and biochemistry.
size of the chip is shown in Fig. 4c. A very wide
current detection range is successfully achieved FET-Based Amperometric Sensor
from 1 pA to 100 nA from the systematic The field effect transistor is a nonlinear physical
level. Figure 4 shows the detailed structure of device which is able to work as an amplifier. By
the chip and experiment setup for concentration tuning the gate voltage, the current change
imaging. between drain and source can provide an ampli-
The OsHRP-coated LSI was spotted with fied signal to characterize the changes of gate
glucose oxidase and used for bioelectrochemical voltage. By inserting an amperometric sensor
imaging of the glucose oxidase (GOx)-catalyzed between the drain and source, the information at
oxidation of glucose. Figure 5 shows the transi- the gate is recorded and analyzed. The commer-
tion of the redox current-based images of H2O2 in cial metal-oxide-semiconductor field effect tran-
the flow streams with the OsHRP LSI chip. sistor (MOSFET) has been used to design
When the H2O2 reached the chip, H2O2 was oxi- microfluidic sensor for particles counting. Xu
dized by HRP followed by the oxidation of Os3+ D. et al. has developed the MOSFET-based par-
to Os2+. The reduction current of Os3+ to Os2+ at ticle counter which can provide a current modu-
the 400 measurement points can be monitored by lation with ultrahigh sensitivity, due to its
a real-time image with a color scale at 400 points. nonlinear amplifying property [5]. The detection
Figure 5 shows clear and rapid changes of H2O2 limit of volume changes can reach to 0.006 %. In
flowing over the LSI chip. recent years, the electrical properties of graphene
A 52 Amperometric Detection

Amperometric Detection, Fig. 5 Color maps indicat- point, a reduction in current was observed according to
ing the current intensities of 400 electrodes obtained by the Os3+ generation by an HRP reaction toward H2O2. The
the 3 s injection of 10 mM H2O2 on the OsHRP modified time-dependent transition of the maps is shown in
Bio-LSI chip in PBS flow stream. An inlet and outlet of the 16 images. The operational potential was set to 0.0 V
flow were set in the upper left and lower right directions of versus Ag/AgCl/sat.KCl reproduce from [4]
the figure, respectively. When H2O2 reached a sensing

have been extensively studied, and graphene- microfluidic systems on glass or flexible PET
based sensor with good sensitivity has been substrates, which were used to detect flow veloc-
developed for many bio-detection applications. ity in the microfluidic channel as shown in Fig. 6
Due to the fact that graphene has a 2-D structure [6]. The counter-ions within the EDL will move
and every atom is exposed on its surface, the simultaneously under the pressure-driven fluids
electrical properties of graphene are very sensi- flow, which leads to a convection streaming cur-
tive to the change of the charge environment rent or streaming potentials between the two ter-
induced by the reaction or adsorption of analytes minals of the microfluidic channel. So the fluid
on the surface. Based upon the merits, graphene is flow velocity can be detected by characterizing
a promising material for highly sensitive sensors. the streaming current or streaming potentials with
The graphene-FET-based amperometric sensor is a SGGT integrated in the microfluidic channel.
able to provide even higher sensitivity due to the Figure 6a shows the schematic structures of
nonlinear properties of FET which has been a solution-gated graphene transistors on a sub-
widely used in the sensing of gas molecules, strate, where the graphene film is partially
DNA, and flow velocity. covered by a poly(dimethylsiloxane) (PDMS)
For example, Rong Xiang et al. has developed channel. One Ag/AgCl wire was injected into
solution-gated graphene transistor (SGGT) in the microchannel as the gate electrode.
Amperometric Detection 53 A

Amperometric Detection, Fig. 6 (a) Schematic dia- characteristics (IDSvs VG, VDS = 0.05 V) of a SGGT
gram of a SGGT integrated in a microfluidic chip. (b) in a flexible microchip filled with static 1 mM KCI solu-
Raman spectrum of single-layer graphene on a glass sub- tion at different bending status. (df) The photographs of
strate. The insert shows the photograph of the graphene in the flexible microchip at different bending status (d) flat,
the microfluidic chip. Scale bar is 0.5 mm. (c) Transfer (e) outcurve, and (f) incurve reproduce from [6]

The large-area single-layer graphene can be syn- with the flow velocity stepping up gradually.
thesized on copper foil with the chemical vapor The channel current increased with the increase
deposition (CVD) method. The graphene film of flow velocity. If flow direction is changed from
was then transferred onto a glass substrate or the gate to the graphene, the current profiles will
flexible PET substrate (Fig. 6df). The Raman be as shown in Fig. 3c. It is observed that the
spectrum of the graphene film indicates that the channel current IDS decreased with the increase
graphene film is approximately single layered as of flow velocity. The current changes of the
shown in Fig. 6b. The channel length of device for either flow direction were shown in
a graphene transistor between source and drain Fig. 3d. Since the current changes resulted from
electrodes is 1 mm, which is slightly wider than the modulation of the effect gate voltage by the
the width of microfluidic channel (0.8 mm), as convection steaming potential induced by
shown in Fig. 1b. pressure-driven flows, the shift of the effective
The experimental result of the device is shown gate voltage under different flow velocity was
in Fig. 3 with Ag/AgCl gate electrode set at extracted according to the current change shown
fixed gate and drain voltages (VG = 0.15 V, in Fig. 3d, which shows an approximated linear
VDS = 0.05 V) in alternate static and flowing relationships between the shift of the effective
of 10 mM KCl solution. Figure 3b shows the gate voltage, and the flow velocity for both flow
channel current profiles as a function of time directions. Consequently, by calibrating the
A 54 Amperometric Detection

Amperometric Detection, Fig. 7 Time-dependent is from graphene to Ag/AgCl gate electrode. (c) The flow
channel current of a SGGT characterized in alternate static direction is from Ag/AgCl gate electrode tographene. (d)
and flowing KCl solution with different velocities. The change of channel current and the shift of effective gate
VG = 0.15 V, VDS = 0.05 V. (a, b) The flow direction voltage as functions of flow velocity reproduce from [6]

device with standard fluids flow, the device can electrochemical properties of the material such
provide very high sensitivity for characterizing as graphene, SWNT, etc., are expected to further
the flow velocity (Fig. 7). develop to overcome the detection limits. For the
The ultrahigh detection sensitivity enables the second aspect, parallel analysis of multiple bio-
graphene promising materials for amperometric logical samples can provide rapid detection plat-
sensor in the lab-on-a-chip applications. The form which is a promising technology in the point
graphene-based sensor is now at its early stage of of care diagnosis.
invention, and lots of new graphene-based amper-
ometric sensors with stable performance are
expected in the near future. On the other hand, the Cross-References
mass production of graphene with good quality
also remains to be enhanced in future development. Amperometric Techniques

Future Directions for Research References

Future work shall be focused on two aspects: (1) 1. Christian N et al (2012) Implantable enzyme ampero-
metric biosensors. Biosens Bioelectron 35:1426
increase of the sensitivity of the amperometric
2. Bard AJ, Faulkner LR (1980) Electrochemical
microfluidic sensor and (2) multiple analysis of methods: fundamentals and applications. Wiley,
biomolecules. For the first aspect, new New York
Amperometric Techniques 55 A
3. Ney H et al (2009) Fabrication of a multichannel (or working) electrode is placed at the end of
PDMS/glass analytical microsystem with integrated the separation stream so as to permit detection of
electrodes for amperometric detection. Lab Chip
9:115121 sample components as they emerge and pass over A
4. Kumi Y et al (2012) LSI-based amperometric sensor the electrode surface. The applied potential must
for bio-imaging and multi-point biosensing. Lab Chip be sufficient to cause facile oxidation or reduc-
12:34813490 tion of the target analytes; and, in general, the
5. Xu D et al (2007) Ultrasensitive fluidic sensors with
amplifications from both fluidic circuits and magnitude of the measured current is directly
MOSFETs. Appl Phys Lett 91:013901 proportional to the analyte concentration.
6. He R et al (2012) Solution-gated graphene field effect
transistors integrated in microfluidic systems and used
for flow velocity detection. Nano Lett 12:14041409

Amperometric electrochemical (EC) detection

was initially developed in the 1970s for conven-
Amperometric Techniques tional sized high-performance liquid chromatogra-
phy (HPLC) instrumentation [1]. However, as the
Thomas J. Roussel1, Douglas J. Jackson2, principal micro-/nanofluidic applications of the
Richard P. Baldwin3 and Robert S. Keynton1 approach are most directly related to capillary
Department of Bioengineering, University of electrophoresis (CE), it is the development of
Louisville, Louisville, KY, USA these latter systems that is of greatest relevance
Department of Electrical and Computer here. Amperometric detection was first introduced
Engineering, University of Louisville, Louisville, in conventional bench-scale CE in 1987 by Wal-
KY, USA lingford and Ewing [2] for the determination of
Department of Chemistry, University of catecholamine-based neurotransmitters. The
Louisville, Louisville, KY, USA detection was carried out by carefully placing
a small carbon fiber electrode at the outlet of the
separation capillary and recording the current tran-
Synonyms sient caused by oxidation of each catecholamine as
it emerged from the capillary outlet and passed
Amperometric detection; Controlled-potential over the fiber. Subsequently, dozens of papers
instrumentation; Electrochemical detection; Oxi- have appeared both on novel instrument designs
dation/reduction in CE/EC and on a wide range of CE/EC applica-
tions [3]. During the course of this work, ampero-
metric CE/EC has been carried out on a variety of
Definition carbon and metallic electrodes and has been shown
to offer both the high sensitivity (mM-nM) required
Amperometry is one of a family of electrochem- by CE and a unique selectivity determined by the
ical methods in which the potential applied to potential applied and the specific electrode used. In
a sensing electrode is controlled instrumentally general, this detection approach is applicable to
and the current occurring as a consequence of several important families of organics. In addition
oxidation/reduction at the electrode surface is to catechols, these include carbohydrates, amino
recorded as the analytical signal. In its simplest acids, thiols, phenols, and aromatic amines, as well
form, the applied potential is stepped to and then as a variety of specific compounds of biological or
held at a constant value; and the resulting current pharmaceutical interest.
is measured as a function of time. When amper- In amperometry, quantification of species con-
ometric detection is used in conjunction with centration is possible because the magnitude of
separation techniques such as capillary electro- the current generated in any given experiment is
phoresis or liquid chromatography, the sensing determined by the number of molecules, of that
A 56 Amperometric Techniques

species, oxidized or reduced at the surface of the coated onto the inner walls) for HPLC or differ-
sensing electrode and therefore is directly pro- ential electrical migration of the charged analytes
portional to the concentration of the molecule toward the electrokinetic drive electrodes for
detected as defined by Faradays law (Eq. 1). CE. Many of these separation techniques are
being downsized using microfabrication
dQ dN methods, referred to collectively as microfluidics.
it F (1)
dt dt

Where it is the current generated at the surface Basic Methodology

of the sensing electrode at time = t, Q is the
charge at the surface of the sensing electrode, Z In a redox reaction (Eq. 2), electrons are removed
is the number of moles of electrons transferred from a solution species (oxidation) at one elec-
per mole of analyte, N is the number of moles of trode and transferred to a solute (reduction) at the
analyte oxidized or reduced, and F is the Faraday other:
constant (96,487 Cmol1). Detection limits have
been reported at the attomole and femtomole aA ! wC ne anode
levels [4, 5], roughly corresponding to micromo- bB ne ! dD cathode
lar concentrations in the carrier fluid. Because
this form of detection is coupled to the liberation where A, B, C, and D are the reactants and prod-
or consumption of electrons, the application of ucts of the reaction a, b, w, and d are their stoi-
this detection technique is limited to those spe- chiometric coefficients and n is the number of
cies that are electroactive, i. e., those that can be electrons. To carry out and monitor such
oxidized or reduced at an electrode surface. a reaction, an electrochemical cell is used.
While amperometry has been used for decades These cells typically consist of three
in bench-top scale analytical systems, the greatest electrodes the working, auxiliary, and reference
attractiveness of this particular detection electrodes placed in electrolyte solution.
approach lies with the ability to miniaturize the Amperometric sensing is controlled by applica-
necessary electrodes (working, reference, and tion of the desired potential or voltage solely to
auxiliary or counter) and incorporate them into the working electrode. This is the electrode at
miniaturized lab-on-a-chip devices. The simplic- which the redox reaction (either oxidation or
ity of the supporting electronics for amperometry reduction) of the analyte of interest occurs, and
provides advantages as well compared to other therefore it is the current flowing at this electrode
detection systems (e.g., UV-visible absorbance, that constitutes the analytical signal (Fig. 1). The
laser-induced fluorescence, mass spectrometry, function of the reference electrode is to provide
etc.) that require a considerably larger ensemble a contact whose potential remains constant dur-
of supporting equipment. ing the amperometry experiment, and its chemi-
The specific fluidic system involved, while not cal composition (saturated calomel, Ag/AgCl,
the main topic of this entry, is closely linked to etc.) is purposely selected to facilitate this. The
the detection system and warrants a brief expla- role of the counter electrode is to provide
nation. Separation techniques such as CE and a current path that completes the electrical cir-
HPLC utilize channels or columns that pass cuit. Thus, the redox reaction occurring here is
a fluidic medium carrying the analytes of interest. equal in size but opposite in direction (i.e., reduc-
Bulk flow is actuated using either gravity or pres- tion or oxidation). Normally, this electrode con-
sure, in the case of LC, or electroosmotic forces, sists of a simple metallic contact, most commonly
in the case of CE. Separation of analytes occurs platinum.
due to physical or chemical interactions with The chemical composition/structure of the
what is incorporated into the column (a packing working electrode can vary considerably, and
material that completely fills the column or is a proper selection is often critical to achieving
Amperometric Techniques 57 A
success in a given application. The minimum The function of the external electronic
requirements are reasonable degrees of chemical circuitry or the potentiostat in amperomet-
inertness and electrical conductivity. Commonly ric experiments is to supply the desired voltage A
encountered electrodes include metals, such as difference between the working and reference
Pt, Au, or Hg, and carbon. In addition, the surface electrodes and measure the resulting redox cur-
of these electrode materials can also be modified rent instantaneously. The magnitude of the work-
chemically by a number of different approaches ing electrode potential necessary to initiate the
in order to optimize performance for a specific oxidation or reduction is ideally determined by
electrochemical process [6]. For example, the the Nernst equation (shown in Eq. 3) but can also
charge, polarity, porosity, and specific chemical be affected by factors such as the conductivity of
and biochemical reactivity can be adjusted by the the carrier fluid, the electron-transfer kinetics of
addition of appropriate functional groups, self- the specific redox reaction, the composition of the
assembled mono- and bi-layers, and polymer detection electrode, and the relative distance
coatings (Fig. 2). between the working and reference electrodes.

0:059 aw ad
E E0  log C Db (3)
j nj aaA aB

where E is the potential of the working electrode,

E0 is the standard potential for the redox reaction
of interest, n is the number of electrons involved,
and the as represent the activities of the reactants
and products. It follows from the Nernst equation
that making the working electrode potential more
negative favors the reduction process and is
therefore said to be more reducing or
Amperometric Techniques, Fig. 1 Typical electrophe-
rogram for catechol and dopamine using amperometric cathodic while application of a more positive
detection on a microchip platform potential would be oxidizing or anodic.

Analyte Sample
Flow Direction

Working Reference Auxiliary


Current A

Amperometric Techniques, Fig. 2 Schematic of amperometric detection cell (general representation for lab-on-a-
chip (LOC) applications)
A 58 Amperometric Techniques

iRu Rc


Working Electrode Reference Electrode Auxiliary Electrode

Amperometric Techniques, Fig. 3 Electrical model of the auxiliary electrode; CDLA is the double-layer capaci-
three-electrode cell (ZFW is the faradaic impedance at the tance at the auxiliary electrode; RC is the resistance of the
working electrode; CDLW is the double-layer capacitance electrochemical cell; Ru is the uncompensated resistance;
at the working electrode; ZFA is the faradaic impedance at and i is the current)

Thus, in practice, amperometric detection involv- electrode is connected to an ammeter circuit,

ing analyte reduction is generally carried out by typically a sensitive transimpedance amplifier
application of negative potentials, while analyte (current-to-voltage converter). The oxidation or
oxidation requires the use of relatively positive reduction reaction occurring when an
applied potentials. Successful bench-scale electroactive analyte passes the working elec-
CE/EC in the amperometric mode requires the trode produces current flow in the ammeter that
accurate maintenance of EC potentials on the is then converted to a proportional voltage for
order of roughly 1 V at working electrodes display.
placed in CE fields on the order of 530 kV and A simple electrical model of the solution as
the measurement of EC currents typically pA in seen by the electrodes is shown in Fig. 3. Resis-
magnitude in the presence of mA-level back- tances are primarily determined by the ionic con-
ground CE currents. tent of the electrolyte solution, which is typically
A representative three-electrode amperomet- in the 0.010.10 M range. When the electrodes
ric detection configuration and a simplified are placed in an electrolyte solution, a charge
potentiostat circuit are shown in Fig. 1. The work- separation or double layer forms spontaneously
ing and reference electrodes are placed in the at the electrode-solution interface. This produces
amplifier feedback loops so that the voltage dif- a capacitance, commonly referred to as the
ference between them must match an adjustable double-layer capacitance, CDL. The size of this
external voltage source. Because the composition capacitance is a function of electrode area,
of the reference electrode is designed to maintain electrode-to-solution potential, electrode mate-
its potential at a constant value, the working rial, and the ion concentration in solution. The
electrode potential is just the selected external potentiostat compensates only for capacitance,
voltage compared to or vs. the specific refer- impedance, and resistance (CDLA, Rc, ZFA) within
ence electrode used. The reference electrode is its control loop and allows these to be largely
connected to a high input impedance buffer ignored experimentally. Therefore, the
amplifier, which serves to limit the current flow uncompensated resistor, Ru, which is outside the
through it to a negligible level, which in turn loop, causes an error in the working electrode
serves to keep its composition unchanged and potential, iRu, as a result. However, given the
its potential at its starting value. The working very small amplitude currents usually involved
Amperometric Techniques 59 A
in microfluidic systems and the relatively low Of course, one of the attractive features of lab-on-
resistivities of the aqueous electrolyte solutions a-chip CE/EC systems is that the specific loca-
normally employed, this error is seldom of any tion, size, and shape of the EC electrodes can be A
practical concern for amperometric measure- absolutely controlled during the microfabrication
ments. Of course, the product RuCDL represents process.
the cell time constant and largely determines the In amperometric detection, the placement
speed of response of the cell to any change in the of the working electrode in CE/EC detection
applied potential. For typical amperometric systems can be divided into three different cate-
detection applications where the applied poten- gories, off-channel (or column), end-channel
tial is kept constant throughout the entire exper- (or column), and in-channel detection. The pur-
iment, this also is of no concern; but, for the pose of this distinction is to indicate the location
relatively few cases in which the potential is of the EC electrodes with respect to the CE elec-
systematically scanned during the CE separation, trodes and field (Fig. 4). This is an important
limitations in accurate control of the working consideration because of the possible interactions
electrode potential due to charging of CDLW that might occur, as mentioned above, between
must be considered. the EC and much larger CE electric fields.
Application of amperometric detection to Off-channel EC detection refers to the situ-
HPLC instrumentation is usually straightforward, ation in which an electrically conducting frac-
involving the placement of conventional size ture or thin crack, opening, or hole is created in
electrodes downstream from the column exit. the capillary or channel wall a short distance
Care must be given to keep the dead volume before the exit. This fracture allows the CE volt-
entailed by the detection cell and connecting tub- age and current to be dropped before the actual
ing small enough to avoid compromising the physical end of the system and permits the EC
separation efficiency. For CE/EC systems, how- electrodes to be placed near or even inside the
ever, there are two specific issues that must be capillary exit for efficient analyte detection but
addressed experimentally: (1) physical alignment nevertheless to remain off or outside the influ-
of the electrodes with the capillary or channel ence of the CE field, commonly referred to as
outlet and (2) electrical decoupling of the EC decoupling [7, 8]. The off-channel approach,
detection from the high-voltage CE field. The which often may pose significant technical diffi-
need to optimize capillary/electrode alignment culties to implement, offers the possibility of very
arises from the very small dimensions efficient detection of the analyte while it is still
involved with the widths of both CE channels contained within the confines of the narrow sep-
and EC electrodes typically in the 10100 m aration channel and remains in the form of
range. Thus, optimum sensitivity dictates that a relatively concentrated plug. End-channel
the working electrode is placed very near the detection refers to the more convenient place-
channel outlet, while acceptable reproducibility ment of the EC detection electrode just beyond
requires that this positioning, once established, be the end of the separation channel in the CE buffer
maintained throughout extended sets of experi- reservoir. Thus, the EC cell is located within the
ments. With bench-top CE instrumentation, this high-voltage CE field, and electrical interactions
usually requires the use of microscopes and 3D between the two are still possible. However, as
micro-positioners to align the fiber or wire with long as the CE capillary or channel width is
respect to the capillary. Alternative arrangements sufficiently small, its ohmic resistance is high
to ease this problem have been suggested such as enough that the CE current is negligibly small
the use of planar wall-jet electrodes that can be and only a very small portion of the CE voltage
somewhat larger in size or the incorporation of extends beyond the exit and into the much larger
the EC electrode onto the capillary by deposition volume detection compartment. Of course, an
of an appropriate metallic film onto its outside tip. inherent problem with this technique is that the
A 60 Amperometric Techniques

Amperometric a
Fig. 4 Schematic of EC
electrode placement for the Working
different detection
electrode configurations:
(a) end-channel detection; b c
(b) in-channel detection;
and, (c) off-channel
detection Electrode Decoupler Working

well-defined analyte plugs maintained within the migrating analyte so as to maximize the delivery
capillary immediately begin to spread radially of the sample to the detection electrode surface.
upon exiting and moving into the much larger
volume detection reservoir. Therefore, care
must be taken that the detection electrode, in Key Research Findings
this case, is located close to the exit of the chan-
nel, typically 25100 mm, but not so close as to The first microfluidic lab-on-a-chip (LOC)
interact unduly with the residual CE voltage that devices were reported by Harrison and Manz in
does extend slightly beyond the separation chan- the early 1990s. These devices consisted of sim-
nel. A detailed study of the residual post-capillary ple channels, several cm in length and 10100 mm
CE field, and the effect of end-channel electrode in width and depth, etched into glass microscope
placement, has been reported by Klett slides or silica plates via conventional
et al. [9]. The in-channel approach has been micromachining techniques [10]. In this work,
employed by researchers; however, this approach samples were injected onto the chip and moved
is limited due to bubble generation in the channel, as required through the channel network by elec-
which can significantly decrease the signal-to- troosmotic flow; separations were carried out by
noise (S/N) ratio of the electrochemical signal, electrophoretic mechanisms common to bench-
as well as the high electric field can significantly scale CE instrumentation; and target analytes
effect the resting background current generated at were detected and quantitated via fluorescence
the surface of the working electrode. detection. Subsequent studies by these and other
In regard to electrode geometry, the EC groups demonstrated unique advantages for the
detector response is proportional to electron lab-on-a-chip approach: high separation efficien-
generation (or consumption) during the desired cies, short analysis times, minute sample and
redox reaction. Therefore, increasing the area of reagent consumption, high sample throughput,
the detection electrode will increase the detec- and easy automation.
tion signal. The tradeoff is that the background In 1996, the first use of amperometric detection
noise increases linearly as well. Due to the con- in microchip CE devices was reported by Wooley,
fined geometry of the microchannel, in-channel Lao, Glazer, and Mathies [11] who used off-chip
electrodes are limited to simple rectangular CE-driving electrodes together with a photolitho-
shapes determined by the channel width and graphically patterned Pt electrode placed just out-
exposed electrode area. This configuration is side the CE channel for on-chip detection of
sufficient to maximize the contact of the analyte catechol neurotransmitters by direct oxidation
with the electrode surface. For an end-channel and DNA fragments and PCR products indirectly
electrode configuration, however, a simple rect- by EC monitoring of an electroactive intercalating
angular planar electrode is not the only option. agent. In another end-channel approach, Baldwin
The conventional planar geometry can be and colleagues [12] were the first to demonstrate
altered to match the shape of the radially a lab-on-a-chip device that incorporated both CE
Amperometric Techniques 61 A
high-voltage electrodes and all amperometric EC electrode) and allow the EC-working electrode
detection electrodes directly onto an integrated, to be placed in-channel where higher efficiency
microchip platform. Key features of their device separations can be realized. Similarly, Lai A
were the formation of a shelf or ledge at the end et al. [16] integrated a Pd-film decoupler with
of the separation channel to restrict the height a series-dual electrode to significantly enhance
of the detection reservoir near the working the oxidation current when switching between
electrode and the radial pattern of the working the single electrode mode and dual-series mode.
electrodes. These features enabled optimization In another work, Wu et al. [17] electrodeposited
of EC detection by decreasing the electrical resis- Pt nanoparticles on a Au ground electrode to
tance and CE voltage drop in the detection region construct the decoupler and achieved a limit of
and increasing the residence/contact time of the detection of 0.125 mM at a S/N = 4 for dopamine
analytes with the working electrode, respectively. with a Au working electrode.
Another novel CE microchip end-column Compared to alternative detection approaches,
approach has been reported by Ertl et al. [13], amperometric methods are ideally suited to
which integrates sheath flow into the CEEC sys- microfabrication processes since the electrodes
tem to increase analyte velocity in the detection themselves have the ideal dimensions
reservoir as it exits the CE separation channel. (110 mm) for incorporation onto a microchip.
Their study further elucidated the importance of Most importantly, patterning and construction
the spacing between the working and reference of EC microelectrodes are extremely compatible
electrodes, with optimum detector performance with mainstream microfabrication technologies
achieved for a spacing of
20 mm. Using an such as photolithography and sputter-coating.
electrically isolated potentiostat with a floating As a result, electrodes (and electrochemical
working electrode, Martin and coworkers [14] cells) of different shape, size, location, and
successfully detected catechol with minimum composition can be patterned directly onto
band-broadening effects by placing the working glass, silica, and other substrates with a high
electrode just inside the separation channel but degree of accuracy using the same photolitho-
without any decoupling. Off-channel EC detec- graphic techniques employed to fabricate the
tion approaches have been also explored in micro- microchips CE channels, thereby allowing the
chip platforms. For example, the Lunte groups detection operation to be fully integrated onto
demonstrated two different methods for the chip (Fig. 5). In addition, multiple electrodes
decoupling the CE field from the EC detection in different CE channels or multiple electrodes
system. In the first, decoupling was achieved by with different response characteristics can be
laser-cutting a series of 20, 30 mm holes in the roof created to allow the device to fit the intended
of the CE channel into the upper glass cover plate application optimally (Fig. 5b). This level of
and then filling them with cellulose acetate [8]. In integration does not exist with other detection
the second, a 500-mm-wide Pd decoupler strip was approaches and has several performance and
fabricated via photolithography and sputter- application advantages. For example, with LIF,
coated onto the glass substrate that formed the multichannel detection requires either the use of
bottom of the CE channels [15]. When placed multiple fluorescence systems or a means of scan-
directly in the CE channel and used as the CE ning a single fluorometer across multiple spots on
cathode, the Pd was able to absorb the H2 gas that the chip.
is generated during normal CE operation and Chip-based amperometric EC detection tech-
leads to the formation of bubbles in the channel. niques have been utilized in several noteworthy
The key advantages of these approaches analysis situations most commonly in clinical
(and others like them) are that they allow micro- (e.g., immunoassays and protein analysis) and
chips to operate at higher field strengths (up to environmental sensing. A recent review by
1,700 V/cm) with superior limits of detection Vandaveer et al. [19] provides an excellent over-
25 nM for dopamine with a carbon fiber view of these applications.
A 62 Amperometric Techniques

Amperometric Techniques, Fig. 5 Examples of differ- electrodes (Reprinted with permission from Woolley
ent electrode geometries that can be fabricated via photo- et al. [11]) and (b) multiple radial working electrodes
lithographic techniques: (a) rectangular working (Reprinted with permission from Keynton et al. [18])

Future Directions for Research electrodes. Increased surface area electrodes

allow for a more complete interaction between
Considering the relative ease with which proto- the analyte and the detector which should be
types can be developed on the microscale, custom expected to enhance the signal-to-noise ratio
EC electrode configurations mated perfectly to and sensitivity.
the CE separation platform are possible. For A relatively new area which is likely to grow
example, devices with multiple separation chan- in importance is the construction/functiona-
nels may be outfitted with multiple EC cells, each lization of the EC electrodes with alternative
identical to the other to establish redundancy for materials to enhance the reactivity toward the
fail-safe detection. Each channel could run simul- analyte or to allow detection of analytes that do
taneously, and statistical calculations (running not oxidize or reduce ideally at simple electrodes.
averages, standard deviations, etc.) could be For example, Marwan et al. [20] reported the
used to enhance both the reliability of detection functionalization of glassy carbon electrodes
and the robustness of the detection system. The with metal-based species in order to elicit the
development of such advanced LOC instruments electrochemical reduction of nitrates and oxida-
is expected to be central to their practical utiliza- tion of hydrogen peroxide. Similarly, Sarapuu
tion in applications involving remote and et al. [21] have shown the modification of
unattended sensing. boron-doped diamond and highly oriented pyro-
Another possibility would be for each channel lytic graphite electrodes with quinones to greatly
of a multiple channel CE/EC system to be con- enhance the rate of oxygen reduction, thereby
figured with an EC cell specifically designed to allowing a highly sensitive determination of
detect a different analyte or analyte group. This hydrogen peroxide. Biological molecules (DNA,
could be accomplished, for instance, by antigens, etc.) which are often not inherently
maintaining each EC electrode at a different electroactive can be detected electrochemically
potential, fabricating each detection electrode by approaches such as Kavanaghs [22] in which
out of a different substrate, or modifying each a redox polymer and a single-stranded DNA
electrode chemically or biochemically. In this probe were attached to a gold electrode; hybrid-
case, there would be no redundancy; however, ization and formation of the double-stranded
a single experiment could yield results for probe DNA altered the electrochemistry of the
a range of analyte components. Furthermore, polymer and thereby permitted detection of the
alternative microfabrication techniques can be complementary DNA species. Continuing
developed to create electrodes with alternative research into further functionalization of tradi-
geometries, such as flow through or three- tional and modified electrode materials will fur-
dimensional, large surface area detection ther enhance sensitivity and promote selectivity
Amperometric Techniques 63 A
between analytes, especially in the areas of clin- 4. Sloss S, Ewing AG (1993) Improved method for
ical and biochemical analysis. end-column amperometric detection for capillary
electrophoresis. Anal Chem 65:577581
An alternative approach gaining widespread 5. Wang J (2000) Analytical electrochemistry, 2nd edn. A
attention is the addition of an electrochemically Wiley-VCH, New York
active tag or label to an electro-inactive biomole- 6. Bard AJ, Faulkner LR (2001) Electrochemical
cule such as DNA, SNPs, etc. Similarly to fluores- methods: fundamentals and applications, chapter
13, 2nd edn. Wiley, Hoboken, p 580
cence detection, in which a sample is treated with 7. Osbourn DM, Lunte CE (2001) Cellulose
a natively fluorescent chemical marker in order to acetate decoupler for on-column electrochemical
render it optically detectable, this technique tags detection in capillary electrophoresis. Anal Chem
the sample at a specific site with a molecule 73:59615964
8. Osbourn DM, Lunte CE (2003) On-column electro-
that undergoes oxidation/reduction at a well- chemical detection for microchip capillary electro-
characterized electrode system. Wooley et al. phoresis. Anal Chem 75:27102714
[11] first proposed this promising approach in 9. Klett O, Bjorefors F, Nyholm L (2001) Elimination of
1998. Since then, others have demonstrated the high-voltage field effects in end-column electrochem-
ical detection in capillary electrophoresis by use of
ability to carry out this technique on a microchip on-chip microband electrodes. Anal Chem
platform. Hebert and colleagues [23, 24] have 73:19091915
detected a specific SNP (single-nucleotide poly- 10. Harrison DJ, Manz A, Fan Z, Ludi H, Widmer HM
morphism) site using an electrochemically active (1992) Capillary electrophoresis and sample injection
systems integrated on a planar glass chip. Anal Chem
taggant in conjunction with sinusoidal 64:19261932
voltammetry, while Wang and coworkers [25] 11. Woolley AT, Lao K, Glazer AN, Mathies RA
labeled amino acids with electrochemically active (1998) Capillary electrophoresis chips with
1-cyanobenz(f)isoindole derivatives to permit the integrated electrochemical detection. Anal Chem
detection of arginine, lysine, phenylalanine, 12. Baldwin RP, Roussel TJ Jr, Crain MM,
glycine, and cysteine. Given the inherent sensitiv- Bathlagunda V, Jackson DJ, Gullapalli J, Conklin
ity of EC and the flexibility of microfabrication, JA, Pai R, Naber JN, Walsh KM, Keynton RS
this line of research is expected to have a major (2002) Fully-integrated on-chip electrochemical
detection for capillary electrophoresis in
impact on the detection of many important a microfabricated device. Anal Chem 74:36903697
biomolecules. 13. Ertl P, Emrich CA, Singhai P, Mathies RA
(2004) Capillary electrophoresis chips with a sheath-
flow supported electrochemical detection system.
Anal Chem 76:37493755
Cross-References 14. Martin RS, Ratzlaff KL, Huynh BH, Lunte SM
(2002) In-channel electrochemical detection for
Capillary Electrophoresis microchip capillary electrophoresis using an
Electroosmotic Flow (DC) electrically isolated potentiostat. Anal Chem
Electrophoresis 15. Lacher NA, Lunte SM, Martin RS (2004) Develop-
ment of a microfabricated palladium decoupler/
electrochemical detector for microchip capillary elec-
trophoresis using a hybrid glass/poly(dimethyl-
References siloxane) device. Anal Chem 76:24822491
16. Lai CCJ, Chen CH, Ko FH (2004) In-channel dual-
1. Kissinger PT, Ridgway TH (1996) Small-amplitude electrode amperometric detection in electrophoretic
controlled-potential techniques. In: Kissinger PT, chips with a palladium film decoupler. J Chromatogr
Heineman WR (eds) Laboratory techniques in elec- A 1023:143150
troanalytical chemistry, 2nd edn. Dekker, New York 17. Wu CC, Wu RG, Huang JG, Lin YC, Chang HC
2. Wallingford RA, Ewing AG (1987) Capillary zone (2003) Three-electrode electrochemical detector and
electrophoresis with electrochemical detection. Anal platinum film decoupler integrated with a capillary
Chem 59:17621766 electrophoresis microchip for amperometric detec-
3. Haber C (1996) Electrochemical detection in capil- tion. Anal Chem 75:947952
lary electrophoresis. In: Landers JP (ed) Handbook of 18. Keynton RS, Roussel TJ, Crain MM,
capillary electrophoresis, 2nd edn. CRC, Boca Raton Jackson DJ, Franco DB, Naber JF, Walsh KM,
A 64 Amphoteric Ion Exchange Membrane

Baldwin RP (2004) Design and development of

microfabricated capillary electrophoresis devices Analyte
with electrochemical detection. Anal Chim Acta
19. Vandaveer WR IV, Pasas-Farmer SA, Fischer DJ, Definition
Frankenfeld CN, Lunte SM (2004) Recent
developments in electrochemical detection for An analyte is the substance or chemical constitu-
microchip capillary electrophoresis. Electrophoresis
25:35283549 ent that is undergoing analysis. It is the substance
20. Marwan J, Addou T, Belanger D (2005) Functiona- being measured in an analytical procedure. For
lization of glassy carbon electrodes with metal-based instance, in an immunoassay, the analyte may be
species. Chem Mater 17:23952403 the ligand or the binder.
21. Sarapuu A, Helstein K, Schiffrin DJ,
Tammeveski K (2005) Kinetics of oxygen reduction
on quinone-modified HOPG and BDD electrodes Cross-References
in alkaline solution. Electrochem Solid State Lett
8:E30E33 Biosensor
22. Kavanagh P, Leech D (2006) Redox polymer
and probe DNA tethered to gold electrodes for Lab-on-a-Chip Devices for Chemical Analysis
enzyme-amplified amperometric detection of DNA
hybridization. Anal Chem 78:27102716
23. Hebert NE, Snyder B, McCreery RL, Kuhr WG,
Brazill SA (2003) Performance of pyrolyzed photo-
Analytical Modeling
resist carbon films in a microchip capillary electro-
phoresis device with sinusoidal voltammetric Synonyms
detection. Anal Chem 75:42654271
24. Hebert NE, Kuhr WG, Brazill SA (2003) A microchip
Analytical method
electrophoresis device with integrated electrochemi-
cal detection: a direct comparison of constant poten-
tial amperometry and sinusoidal voltammetry. Anal
Chem 75:33013307 Definition
25. Wang J, Chen G, Pumera M (2003) Microchip sepa-
ration and electrochemical detection of amino acids
and peptides following precolumn derivatization with Analytical modeling is to seek for mathematical
naphthalene-2,3-dicarboxyaldehyde. Electroanalysis functions and equations that are obtained from the
15:862865 closed-form (exact or approximate) solution to the
original physics-governing equations. The out-
come of analytical modeling is behavioral models.

Amphoteric Ion Exchange Cross-References

Behavioral Model

An ion exchange membrane which includes both Anion

a weak anion exchange and a weak cation
exchange functionality into the same membrane. Definition
This type of membrane has the ion exchange
functionalities distributed randomly throughout Negatively charged ion that moves towards the
the membrane. anode in electrophoresis.

Cross-References Cross-References

Ion Exchange Membranes Electrophoresis

Anisotropic Etching 65 A
to etch in directions governed by crystallographic
Anion Exchange Membrane orientation. Dry techniques utilize an inhibitor to
protect the sidewalls of a cavity during ion bom- A
Definition bardment to obtain high aspect ratios.

An ion exchange membrane that interacts with

anions and typically consist of amine functional- Overview
ities (NH3+, NRH2+, etc.). These functionali-
ties allow anions to pass through the membrane In contrast to isotropic etching, anisotropic etching
and prevents cations from passing through the is a fabrication technique that removes material in
membrane. The ability of the membrane to bind specific directions allowing for the production of
ions is greatly dependant on the pH of the solu- geometric characteristics such as sharp corners,
tion and the ions of interest. flat surfaces, and deep cavities. Relative to
microfabrication, selectivity can be defined as the
ratio of the etch rate of the target material to the
etch rate of other materials. The definition can also
Cross-References be applied more specifically to single crystalline
materials such as silicon where it would be con-
Ion Exchange Membranes sidered the ratio of the etch rate in the target
direction to the etch rate in other directions.

Anisotropic Etching

Dustin House1 and Dongqing Li2

Department of Mechanical Engineering,
Vanderbilt University, Nashville, TN, USA
Department of Mechanical and Mechatronics
Engineering, Faculty of Engineering, University
of Waterloo, Waterloo, ON, Canada


Bosch process; Deep reactive-ion etching

(DRIE); Etching single crystalline materials;
Physical-chemical etching; Potassium hydroxide
(KOH) etching; Silicon etching


Anisotropic etching is a subtractive microfab-

rication technique that aims to preferentially Anisotropic Etching, Fig. 1 Example of a masked sili-
remove a material in specific directions to obtain con wafer with a h111i surface orientation (a) before, (b)
intricate and often flat shapes. Wet techniques during, (c) and after exposure to a potassium
exploit the crystalline properties of a structure hydroxide etch
A 66 Anisotropic Etching

Different etchants used in anisotropic wet etch silicon due to their compatibility with
etching have specific etch rates for each crystal- metals involved in on-chip circuitry, whereas
lographic plane in the material being removed. It alkali hydroxide etchants (NaOH, CsOH) are
is the difference in etch rates between planes used for their faster etch rates [3]. A commonly
that produce shapes within the material being used alkali hydroxide etchant for silicon is
etched that appear to follow the planes within potassium hydroxide (KOH) which provides
the crystalline structure as seen in Fig. 1. For a high level of selectivity between h100i and
more information on the crystalline structure h111i planes. Materials that are highly resistant
of silicon and how miller indices are used in to these etchants, such as silicon nitride and
designating crystal planes, see The MEMS silicon dioxide, are often utilized as etch stops
Handbook [1]. It is commonly accepted that in more complex designs. Detailed specifica-
the h111i plane etches the slowest regardless of tions for each etchant including etch rates in
the etchant used. However, the crystal plane that specific crystal planes and recommended tem-
etches the fastest depends on the etchant com- peratures, concentrations, and etch stops are
position [2]. Ammonium hydroxide etchants available in the literature such as those of
(NH4OH and TMAH) are frequently used to Nguyen and Wereley [2].

a b

c d

Anisotropic Etching,
Fig. 2 Examples of
undercutting that is
possible for (a, b)
odd-shaped masks and
(c, d) misaligned masks
relative to crystal planes
Anisotropy 67 A
Careful design and alignment of the mask trench bottom, exposing the substrate to the
must be ensured when involving directional chemical etchant [4]. Several companies spe-
dependence in the etching process. Because the cialize in the production of microfabrication A
planes will continue indefinitely until they reach equipment for various reactive ion etching pro-
a surface or another plane, undercutting of the cesses. These processes, commonly referred to
masking material is possible. A shape that does as deep reactive-ion etching (DRIE), are capable
not align with the correct planes can produce an of producing channels with a depth on the order
opening that circumscribes the original shape if of 500 mm and aspect ratios on the order of
etched long enough as seen in Fig. 2. The angle 30:1 [5].
produced between the slow-etch plane and the
surface after etching is dependent on surface
orientation and the crystalline structure of the Cross-References
material. Such angles are found in the literature
and are utilized in design to achieve accurate Bulk Micromachining
depths and specific slopes (including vertical). Dry Chemical Etch
Equipment used in wet etching can be readily Fabrication of 3D Microfluidic Structures
found in most laboratories. The material being Isotropic Etching
etched is suspended within a beaker containing Plasma Etching
the etchant while a magnetic stirrer ensures con- Reactive Ion Etching (RIE)
tinuous transport of reactants to and products Silicon Micromachining
from the surface of the material. A simple Sputter Etching
temperature-controlled hotplate regulates the Wet Chemical Etch
optimum temperature.
Anisotropic dry etching is a method for cre-
ating high aspect ratio trenches with nearly ver- References
tical walls. The two common techniques utilize
a protective coating along the vertical walls of 1. Gad-el-Hak M (2002) The MEMS handbook. CRC,
Boca Raton
an etched trench such that the bottom of the
2. Nguyen N, Wereley S (2002) Fundamentals and appli-
trench etches at a faster rate allowing for large cations of microfluidics. Artech House, Norwood
aspect ratios. Each is independent of crystal 3. Lobontiu N, Ephrahim G (2005) Mechanics of
orientation. The first method forces the reactant microelectromechanical systems. Springer, New York
4. Franssila S (2004) Introduction to microfabrication.
gases to condense on the sidewalls of the trench
Wiley, Chichester
by lowering the temperature of the materials 5. Tabeling P (2006) Introduction to microfluidics.
with liquid nitrogen. Ions are then accelerated Oxford University Press, Oxford
toward the trench by means of an electric field.
This physical shower of ions results in an etch-
ing effect. After ion bombardment removes the
bottom layer, it can be etched further while Anisotropy
the sidewalls are protected by the cryogenically
frozen inhibitor. Another novel approach, Definition
commonly referred to as the Bosch process
(named after its inventor, Robert Bosch), alter- The directional dependency of a measurable
nates additive and subtractive microfabrication property of a material.
techniques. During the additive stage, chemical
vapor deposition is used to coat the sidewalls
of the trench with a polymeric passivation Cross-References
layer. Similar to the previous method, ion
bombardment is then used to sputter off the Sputtering for Film Deposition
A 68 Anode

Definition +

Positively charged electrode. Silicon

Hotplate (Anode)

Anodic Bonding, Fig. 1 Schematic setup for anodic
Electrophoresis bonding of silicon to glass

the bonding parts. The bonding process takes

Anodic Bonding only a few minutes. Figure 1 shows schematically
the setup for anodic bonding, where the conduc-
Zheng Cui tive part, in this case a silicon wafer, is connected
Rutherford Appleton Laboratory, Oxford, UK positively to the applied voltage. The essential
conditions for anodic bonding are:
1. The bonding surfaces should be flat and well
Synonyms polished to less than 5 nm in surface rough-
ness. When the bonding pair is put together,
Electrostatic bonding; Field-assisted bonding the two parts are in intimate contact at the
polished surfaces.
2. The materials of the bonding pair must have
Definition well-matched thermal expansion coefficients.
3. Anodic bonding only occurs between conduc-
Anodic bonding is a technique to produce her- tive materials and glass materials.
metic seals between metals and glass insulators. The mechanism of anodic bonding is believed
Anodic bonding is mostly used for hermetic to be both physical and chemical. The physical
sealing of silicon and glass. Different from other mechanism is electrostatic force. At the typical
sealing techniques, anodic bonding involves bonding temperature, most glass materials contain
heating and applying an electrical field to mobile positive ions compensated by almost
a bonding pair. immobile negative ions. Upon applying high volt-
age, these positive ions migrate to the negative
electrode connected to the glass and become
Overview largely neutralized, leaving a depletion layer with
negative space charge at the contact interface.
The anodic bonding technique was first reported Figure 2 shows the typical electrical current profile
in 1969, with the name of field-assisted glass- during anodic bonding [2]. The initial rise of cur-
metal sealing [1]. It has since been developed rent represents the migration and neutralization of
into a widely used hermetic sealing technique positive ions at the electrode. Then the current
for a variety of MEMS devices and microfluidic drops and a depletion layer are built up at the
devices. The equipment for anodic bonding can interface. The contact interface, although well
be simply a hot plate on which the bonding parts polished at both sides, initially has only a few
are heated to the sealing temperature between points of physical contact. Therefore, there is
300  C and 600  C and a DC power supply to a high electrostatic field generated from the nega-
apply a high voltage from 200 up to 2,000 V to tive charge layer at the gaps. The high electrostatic
Anodic Bonding 69 A

T = 400 C A

T= 350 C


Anodic Bonding, Fig. 2 Current profiles during anodic Anodic Bonding, Fig. 3 Microreactor chip assembly by
bonding at different temperatures anodic bonding of silicon and glass

attractive force then pulls the surfaces into direct cavity can occur. This may have detrimental
contact, which explains why the bonding always effects in some device applications [3].
starts at the point contact of electrode and spreads The bonding strength for silicon-glass anodic
radially. The radial spreading of the bonding front bonding is measured typically at 1020 MPa.
makes it impossible for air to become trapped Some have reported between 30 and 40 MPa at
between the bonding surfaces. The electrostatic higher bonding voltage [2]. The pull test has
force depends on a number of factors, such as the demonstrated that bonding strength is higher
total applied voltage, the width of the gap, the than the fracture strength of the glass, i.e., the
concentration, and the distribution of ions in break in the pull test occurs only in the glass and
the glass, whether under the high field the negative not at the silicon-glass interface.
ions will drift out of the glass or the positive ions Anodic bonding of glass and silicon has
will drift into the glass and possible formation of found many applications in microsystems,
surface charge on the glass (the effect of such MEMS, and microfluidics, including the fabrica-
charge would partially shield the charge in the tion of pressure sensors, accelerometers,
glass, thus reducing the field in the gap). micropumps, and other fluid handling devices.
The chemistry of anodic bonding is an oxida- A typical microfluidic device is shown in Fig. 3,
tion process at the interface. Due to the high elec- which is a microreactor chip to perform chemical
trostatic field, oxygen anions leave the glass and synthetic reactions [4]. The microreactor was
create Si-O-Si bonds, i.e., a thin layer of SiO2 is made by wet or dry etching of silicon to form
formed. This gives the strong bond between sili- microchannels and sealed by anodic bonding of
con and glass, and the bonding becomes irrevers- a glass slide on top with inlet and outlet holes.
ible. The formation of thin SiO2 layer at the Anodic bonding can provide an airtight seal to
bonding interface has been verified by Rutherford ensure proper function of fluidic flow in the
backscattering spectroscopy. It has also been dem- microchannels. Silicon is a preferred material in
onstrated that the rate of oxygen release is directly microfluidic systems because micromachining of
proportional to the average current passing silicon is the most mature fabrication technique.
through the bonding area. In areas where silicon Complex structures can be made in silicon by
and glass are in direct contact due to electrostatic either wet chemical or reactive ion etching.
attractive force, the oxygen anions will oxidize the Anodic bonding of silicon with glass can also
silicon into SiO2 and form the anodic bond. The help in the building of multistack microsystems
oxygen is therefore totally consumed. However, in to perform complicated functions. Figure 4 shows
areas where silicon is not in direct contact with an example of a micropump made by silicon-glass
glass, for example, in the case of an etched cavity micromachining and anodic bonding [5].
into the glass or silicon, oxygen evolution into the This is a glass-silicon-glass multistack system.
A 70 Anodic Bonding

Anodic Bonding,
Fig. 4 Schematic cross Pyrex
section of a micropump
assembled by anodic
bonding of glass-silicon- Actuation cavity Poly-Si

Pumping chamber



Input Microvalve

The top Pyrex glass provides a hermetically sealed The transparency of glass allows optical
cavity so that the polysilicon heater can heat up the inspection or detection of fluidic flow/reaction
air in the cavity. Expansion of the hot air in the in microfluidic devices. It also facilitates
cavity will compress the silicon membrane to per- alignment of microstructures in glass with
form a pumping action in the pumping chamber. microstructures in silicon.
The bottom Pyrex glass seals the pump chamber as Although glass-silicon sealing or assembly is
well as provides the input and output holes for the most common form of anodic bonding, the
fluid flow in and out of the pump chamber. technique can also be applied to the sealing of
Anodic bonding is the primary method for many metals with a variety of glasses. Successful
packaging silicon microstructures, for example, seals have been made with borosilicate glass,
packaging of pressure sensors, accelerometers, soda lime glass, potash soda glass, aluminosili-
and solar cell panels, because of its relatively cate glass, fused silica, and fiber optical glass
low bonding temperature. With bonding temper- [1]. Strong seals have also been made to some
ature being below the glass transition point, the ceramic materials. For those metals and alloys
glass thermal expansion coefficient remains con- whose thermal expansion coefficients are
stant, which enables stress-free packaging of sil- matched to the above glasses, successful anodic
icon microstructures (provided the glass has bonding has also been made, such as tantalum,
a good thermal expansion match with silicon titanium, Kovar, Niromet 44, and to semiconduc-
over the process temperature range). In addition, tors such as germanium and gallium arsenide, in
with the low bonding temperature, there is no addition to silicon [1].
measurable flow of glass, which ensures sealing
around silicon microstructures without any loss
of dimensional tolerances. Apart from low bond- Basic Methodology
ing temperature, there are other attractive fea-
tures of anodic bonding: Although the principle of anodic bonding is sim-
Since glass is an electrical insulator, parasitic ple, successful bonding depends on appropriate
capacitances are extremely small. selection of bonding materials, good preparation
The bonding process can be readily performed of bonding surfaces, optimization of bonding
in vacuum, allowing hermetic sealing or zero conditions, and inspection of bonded samples.
pressure reference cavity to be formed These are explained in the example of silicon-
(or sealing with special gas mixtures). glass anodic bonding.
Anodic Bonding 71 A
Material Selection to this standard. Although anodic bonding is fairly
Silicon material is normally of standard wafer tolerant to particles and contaminations on the
sizes and thicknesses. As the anode part of the bonding surfaces, they still need to be cleaned A
bonding pair, the silicon should be low in resis- prior to anodic bonding, unless the wafers are
tivity. However, some applications require an straight out of the box. A study has demonstrated
insulating layer to prevent electrical leakage in that an H2SO4 + H2O2 acidic clean of bonding
silicon. In this case, a thin layer of silicon dioxide surface can significantly reduce the bonding time
or silicon nitride should be deposited on the sili- compared to a simple cleaning with acetone
con surface. The insulating layer does not prevent [6]. Thorough cleaning is even more important
anodic bonding of silicon with glass. However, for structured wafers. Whether the bonding sur-
a thicker insulating layer can increase the bond- faces are hydrophobic or hydrophilic can also
ing time dramatically. For example, for a 0.5 mm- have an influence on bonding. At low temperature,
thick oxide layer, a bonding time of approxi- a hydrophilic surface is easier to bond than
mately five times longer is required [2]. For a hydrophobic surface. However, such difference
anodic bonding with glass, p-type silicon wafers disappears at high bonding temperature [7].
(boron doped) are preferred, as a high electro-
static field is much easier to build up, particularly Process Parameters
at low temperature, at the silicon side of a silicon- The basic process parameters for anodic bonding
glass interface for p-type silicon than for n-type are temperature, voltage, clamping pressure,
silicon (phosphor doped) [6]. However, for high- environment, and bonding time.
temperature anodic bonding (>400  C), the dop- Bonding temperature is normally in the range
ing type of silicon is no longer a factor. 300500  C. Higher temperature promotes
Borosilicate glasses such as Corning 7740 better mobility of positive ions in glass,
Pyrex glass, Schott Borofloat 33 glass, and resulting in a faster bonding process. How-
Hoya SD-2 glass are commonly used for anodic ever, the temperature must be below the soft-
bonding with silicon. They are of almost the same ening point of glass (glass transition
coefficient of thermal expansion (CTE) as that of temperature).
silicon. They also have the necessary electrical Typical bonding voltage is in the range of
conductivity at the bonding temperature. How- 2002,000 V. High voltage helps to drive the
ever, some special applications require the glass positive ions in glass toward the cathode, leav-
to be high in resistivity (Corning 7070 glass) or ing a negatively charged region at the bonding
high in spectral transmission (Pilkington CMZ interface. The electrostatic force at the inter-
glass). Glasses that have a CTE different from face pulls the silicon and glass in direct con-
silicon or a low concentration of mobile positive tact, resulting in oxygen ions in glass
ions are still able to be anodically bonded with oxidizing the silicon to form strong chemical
silicon, but with special process conditions. For bonds. Figure 5 shows the bonding time as
example, to bond Corning 7070 glass with sili- a function of bonding voltage for anodic bond-
con, much higher temperature and voltage are ing of Corning 7740 glass with silicon [6]. It is
needed. To bond a Fisher glass slide whose CTE apparent that higher voltage results in a shorter
is very different from that of silicon, the bonding bonding time. For glasses with low concentra-
temperature has to be much lower (150  C), and tion of sodium ions (Na+), both high tempera-
bonding time is much longer (60 min) [6]. ture and high voltage are necessary for
a successful bonding. Higher voltage is also
Preparation of Bonding Surfaces needed for thicker glass. High voltage also
A well-polished surface is essential for a successful helps to reduce the temperature necessary for
bonding. The average surface roughness (Ra) a good bonding.
should be less than 5 nm. Commercial silicon Standard silicon wafers are not completely flat
wafers and glass substrates are normally polished but with 2040 mm bowing and wrapping.
A 72 Anodic Bonding

loose contact at the bonding interface indepen-

dent of visual errors.
The bonding strength can be tested in different
ways, such as pressure testing, pull testing, shear
testing, and bending testing [2]. The pull test has
demonstrated bonding strength in the range
1020 MPa, which is greater than the fracture
strength of glass. The upper limit of bonding
strength is difficult to estimate, as the break in
the pull test always occurs in the glass, not at the
Anodic Bonding, Fig. 5 Bonding time as a function of silicon-glass interface. However, for imperfect
applied voltage for silicon-glass anodic bonding bonding, the bonding strength is poor, and the
glass often breaks along the boundary of voids.
In order to make sure the silicon wafer is in Although anodic bonding can be done simply
intimate contact with the glass, a pressurized with a hot plate and a DC power supply, reliable
clamping is necessary. The clamping pressure and reproducible anodic bonding can only be
is normally below 50 N. performed with purpose-built systems. There are
Anodic bonding can be carried out either in commercially available wafer bonders. With
vacuum or in a gaseous or atmospheric envi- commercial systems, the bonding process can be
ronment. However, bonding in air is much either manual or automatic. The bonding is
faster than in vacuum or in an inert gas envi- performed in a chamber of either vacuum or filled
ronment [7]. This may be due to the oxygen in with gases. One of the key issues in anodic bond-
air which helps the oxidation process during ing is the alignment when both the silicon and
anodic bonding. However, many applications glass have some fabricated microstructures. The
require anodic bonding to be carried out in alignment process in a commercial wafer bonder
vacuum. can be carried out either off-line or in situ. The in
Bonding time is typically 15 min for Corning situ alignment is advantageous because high-
7740 glass and silicon. For special glasses, the accuracy alignment ( 5 mm) occurs only when
bonding time may be longer. As mentioned the process temperature is reached, thus avoiding
before, both bonding temperature and voltage differential thermal expansion effects which can
can influence the bonding time. After bonding, compromise alignment. It enables the user to
there is a cooling time. The cooling can be make in situ adjustments and visual confirmation
a few minutes with the help of purging inert before finally committing to making the bond.
gas. However, the cooling should be slow for
materials with large difference in CTE.
Key Research Findings
Inspection and Testing
The bonding process can be visually monitored Anodic bonding is relatively simple in principle
through the glass. Initially there will be an inter- and has been well developed over the last few
ference fringe pattern when the silicon and glass decades. Research in this area has been very
are in contact because of the microscopic gaps much focused on solving practical and
between them. As the bonding progresses the application-specific issues. One of the practical
interference pattern recedes gradually, until issues is dealing with anodic bonding of silicon-
a uniform color is seen across the whole bonding glass assemblies with shallow cavities. This is
surface. Any defects in the bonding will show as often encountered in microfluidic systems where
voids at the bonding interface. Better inspection microchannels are of high lateral-to-depth aspect
of bonding quality can be achieved by scanning ratio. Anodic bonding can cause the
acoustic microscopy (SAM). SAM can detect any microchannels to collapse due to high
Anodic Bonding 73 A
electrostatic force at the bonding interface. The- been on oxygen release into sealed cavities during
oretical and experimental study indicated that the bonding or proper seals for thin metal films which
collapse can be avoided as long as the following act as lead through electrodes [3]. A
condition is satisfied [8]:

ea V 2 a=Eeff d 3 < 1 Future Directions for Research

where ea is the permittivity of air, V is the applied Anodic bonding is a fairly mature technique.
voltage for anodic bonding, Eeff is a material Many basic issues have been dealt with. Future
constant which characterizes the elastic stiffness researches will mostly focus on widening its appli-
of the materials, a is the half width of the channel, cations, particularly in microelectronic and
and d is the channel depth. A similar problem microsystem (MEMS) packaging. Packaging can
happens when packaging pressure sensors, significantly influence the performance of final
micropumps, and accelerometers by anodic devices and systems. Therefore, anodic bonding
bonding. Most of these devices have has to be integrated as part of the manufacturing
micromachined silicon membranes, diaphragms, process, and its influence should be fully taken into
or suspended beams. These micromechanical account at the design and manufacturing stages.
structures may stick to the glass when they are
too flexible and too close to the bonding interface.
Although anodic bonding is a low-temperature
process compared to fusion bonding which
requires temperatures up to 1,200  C, the bonding Wafer Bonding
process can still induce thermal residual stress in
the silicon-glass assembly, causing bowing of the
bonded wafer. The thermal expansion of Pyrex
glass is not a constant but changes with tempera-
1. Wallis G, Pomerantz DI (1969) Field assisted glass-
ture. The higher the bonding temperature, the metal sealing. J Appl Phys 40(10):39463949
larger the mismatch between silicon and glass. 2. Obermeier E (1995) Anodic wafer bonding.
For Corning 7740 Pyrex glass, the crossover Electrochem Soc Proc 957:212220
3. Roger T, Kowal J (1995) Selection of glass, anodic
point with silicon (where the thermal expansion
bonding conditions and material compatibility for
of silicon is exactly the same as that of Corning silicon-glass capacitive sensors. Sens Actuator
7740 glass) is around 300  C. Experiments A4647:113120
showed that bonding made at this temperature 4. Cao E, Gavriilidis A, Cui Z (2002) Application of
microengineered reactors in chemical engineering.
can be maintained flat within 5 mm [3]. The
MST News 3(02):1416
crossover point is also dependent on the thickness 5. Acero MC, Plaza JA, Esteve J, Carmona M, Marco S,
of glass. Experiments indicated that for 0.5 mm- Samitier J (1997) Design of a modular micropump based
thick glass (Corning 7740), the crossover temper- on anodic bonding. J Micromech Microeng 7:179182
6. Lee TMH, Lee DHY, Liaw CYN, Lao AIK, Hsing IM
ature is at ~315  C, while it becomes ~260  C for
(2000) Detailed characterisation of anodic bonding
1.5 mm-thick glass [7]. process between glass and thin-film coated silicon sub-
It is known that glass cannot be anodically strates. Sens Actuator 86:103107
bonded to glass. However, research has found 7. Cozma A, Puers B (1995) Characterization of the elec-
trostatic bonding of silicon and Pyrex glass.
that this can be realized by depositing an interme-
J Micromech Microeng 5:98102
diate layer. The intermediate layer can be 8. Shih WP, Hui CY, Tien NC (2004) Collapse of
polysilicon, amorphous silicon, silicon nitride, or microchannels during anodic bonding: theory and
silicon carbide [9]. This has opened an easy route experiments. J Appl Phys 95(5):28002808
9. Berthold A, Nicola L, Sarro PM, Vellekoop MJ
to construct glass-based microfluidic systems
(2000) Glass-to-glass anodic bonding with standard
which are widely used for capillary electrophore- IC technology thin films as intermediate layers. Sens
sis. Other investigations into anodic bonding have Actuator A82(13):224228
A 74 Anthrax

Anthrax Aperiodic Electrophoresis

Definition Synonyms

Anthrax is an acute infectious disease caused by Field-dependent electrophoretic mobility;

the spore-forming bacterium Bacillus anthracis. Nonlinear electrophoresis
Anthrax most commonly occurs in wild and
domestic lower vertebrates (cattle, sheep, goats,
camels, antelopes, and other herbivores), but it Definition
can also occur in humans when they are exposed
to infected animals or tissue from infected Aperiodic electrophoresis refers to the use of an
animals. unbalanced AC field to separate charged polar-
izable particles due to the Stotz-Wien effect of
field-dependent electrophoretic mobility.

AC Electro-Osmotic Flow
Electrokinetic Motion of Polarizable Particles
Antibodies are immune system-related proteins
Electroosmotic Flow (DC)
called immunoglobulins. An important compo-
nent of the immune system, antibodies are
Nonlinear Electrokinetic Phenomena
found in the blood of all vertebrates. The synthe-
Stotz-Wien Effect
sis, or manufacture, of antibodies is initiated
Unbalanced AC Field
when a foreign substance, referred to as an anti-
gen, enters the body. Lymphocyte cells respond
to the foreign substance by making an antibody
with a molecular arrangement that fits the shape
of molecules on the surface of the substance so
Applications of Induced-Charge
that the antibody combines with it. Common
Electrokinetic in Microfluidics
antigens are the protein components of bacteria
Yasaman Daghighi
and viruses.
Physical Sciences, Sunnybrook Research
Institute, Toronto, ON, Canada

Antigen Synonyms

Definition Microstirrer; Microswitch; Nonlinear electroki-

netic phenomena
An antigen is any substance that causes the
immune system to produce antibodies against
it. An antigen may be a foreign substance from Definition
the environment such as chemicals, bacteria,
viruses, or pollen. An antigen may also be formed ICEK Microvalve
within the body, as with bacterial toxins or The main function of a microvalve is switching
tissue cells. the direction of the flow at the desired time to
Applications of Induced-Charge Electrokinetic in Microfluidics 75 A
control and regulate the fluid stream. a conducting surface is in contact with an aque-
Microvalves are one of the most important com- ous solution. The external electric field,
ponents of integrated lab-on-a-chip (LOC) then, interacts with the electric double layer of A
devices. Using microvalves with high perfor- the induced charges. Since the induced zeta
mance, the sequential loading and washing pro- potential and local electric field vary along the
cesses could be performed accurately and fast. conducting surface, the produced

However, the existing microvalves have lots of velocity will be nonlinear, u f E . The theory
serious problems which are unsolved. Thus, and the governing equations of induced-charge
a new design is highly required to develop prac- electrokinetic (ICEK) are discussed in details in
tical and efficient microvalves and overcome the the entry Induced charge electrokinetic motion
current problems. Such a microvalve could be of particles. In this entry two applications of
a great step toward developing integrated LOC ICEK phenomena will be introduced (one
devices. In this entry a new ICEK microvalve microvalve and one micromixer) which could
using a heterogeneous particle is introduced. be used in many microdevices of biochemical,
This ICEK microvalve is easily controlled by medicine, and many other useful and critical
applying two perpendicular electric fields, stable aspects of electrokinetic devices.
and fast, has small dead volume and short
response time, has no leakage, and is easy to Induced-Charge Electrokinetic Microvalve
fabricate. Thus, it can be considered as an appli- The first microvalve was introduced by Terry [1],
cable alternative for exciting microvalves and in 1979, which was the first magnetic MEMS
an interesting application for induced-charge microvalve. After that, microvalves were
electrokinetic phenomena of a heterogeneous improved in several ways. Around the year
particle. 2000 a revolution in fabrication of microvalves
happened [24] which solved the problems of the
ICEK Micromixer various MEMS-based microvalves. However,
Mixing different solutions to obtain many problems (such as (i) using moving parts
a homogeneous mixture in a short period of that causes additional problems and difficulties,
time is a key feature for many LOC applications, (ii) external actuation means, (iii) complex fabri-
which is not easily achievable. Poor mixing pro- cation and installation processes, (iv) resistible
cess produces a heterogeneous mixture and con- flow and pressure, (v) considerable dead volume,
sequently influences the accuracy of the results. (vi) long response time, (vii) leakage, and (viii)
In this entry a new micromixer will be introduced stability) remain unsolved when classical electro-
which can be operated in low-voltage ranges to kinetic theory is used to design a microvalve.
satisfy the requirement for miniaturization. The To improve the efficiency and functionality of
efficiency of this ICEK micromixer depends on microvalve here, an ICEK microvalve is intro-
the size and the strength of the induced vortices duced. Consider a heterogeneous particle initially
around the conducting surfaces. This mixer is placed at the center of a cylindrical
easy to be fabricated and is shown to produce microchamber. This microchamber is connected
a homogeneous mixture at the downstream of to three microchannels with circular cross section
the fluid. This ICEK micromixer can be consid- of different sizes (Fig. 1a, b). The electric fields
ered as a simple high-performance and accurate are applied using six electrodes at desired loca-
solution for the problems of current micromixers. tion (Fig. 1c). Once the electric field is applied,
the vortices are induced at the conducting hemi-
sphere of the heterogeneous particle and push it
Chemical and Physical Principles forward to block the entrance of microchannel;
thus, the fluid can only flow through the other
Based on the nonlinear electrokinetic theory, the microchannel. The flow direction can be changed
electric field induces surface charge when by switching the directions of the applied electric
A 76 Applications of Induced-Charge Electrokinetic in Microfluidics

Applications of Induced-Charge Electrokinetic in represent the external electric field directions, (b) geome-
Microfluidics, Fig. 1 (a) The computational domain of try and dimensions in a non-dimensional format, and (c)
the proposed ICEK microvalve with three microchannels the location of the four electrodes
and one suspended Janus particle in it. E1, and E2,

field. The dimension of this microvalve and elec- Applications of Induced-Charge Electrokinetic in
trolyte properties are listed in Table 1. To avoid Microfluidics, Table 1 Constants used for our numeri-
cal simulation
the gravity effect, the density of the fluid and the
heterogeneous particle are assumed to be equal. Parameters Values
Using the mathematical model described in the Dielectric constant, e 80
entry Induced charge electrokinetic motion of Permittivity of vacuum, 8.854 10 12
e0 (C V 1 m 1)
particles, the motion of the heterogeneous parti-
Viscosity, m (kg m1 s1) 0.9 10 3
cle in this microvalve can be numerically Density, r (kg m3) 998
simulated [5]. Particle diameter, Dp (mm) 20
Diffusion coefficient, Di (cm2 S1) 1.52 106
Induced-Charge Electrokinetic Micromixer Molar concentration at entrance 10
Mixing two or more fluid streams in A, CA (mol/m3)
microchannels without using microstirrer or Molar concentration at entrance 0
mechanical micromixers depends on their molec- B, CB (mol/m3)
ular diffusions; since, the nature of flow in
microchannels is laminar no convective term improve the mixing results can be achieved in
assists the mixing process. Thus, disturbing flow different ways such as the following: (i) using
to change the flow regime from laminar to turbu- complicated geometries and specific controlling
lent is a practical approach to develop a reliable systems, (ii) using mechanical microstirrers to
mixing process. Changing the flow pattern to agitate the flow, and (iii) implementing
Applications of Induced-Charge Electrokinetic in Microfluidics 77 A
Applications of Induced-
Charge Electrokinetic in
Fig. 2 Schematic diagram A
of the proposed micromixer
with two microchannels
and one suspended fully
conducting particle inside
the microchamber

electrokinetic methods for mixing. The induced- particle and the induced vortices by the resultant
charge electrokinetic concept and induced micro- applied DC electric field. The main function of
vortices can be employed to design an electric the electric field E2 is controlling the position of
micromixer with high performance [6, 7]. The the particle. Using the mathematical model
simplicity of such an ICEK mixing method has described in the entry, Induced charge electro-
attracted considerable attention in the fields of kinetic motion of particles, the motion of the
microfluidics and LOC [8, 9]. heterogeneous particle in this micromixer can
To improve the efficiency of the current be numerically simulated [10].
micromixer, a new nonmechanical micromixer The concentration field of different fluid
is introduced in this entry which works based streams is described by the concentration
on the ICEK phenomena. Consider one equation
microchamber which is connected to two
microchannels. The inlet channel is placed on @Ci
the left-hand side of the chamber and delivers U Ci Di 2 Ci i 1, . . . , n (1)
two parallel streams of two different fluids into
the microchamber from entrances A and
B (Fig. 2). On the right-hand side of the where Ci is the concentration of the fluid stream
microchamber, there is a microchannel which is i and Di is its diffusion coefficient. The boundary
used as an outlet. A circular fully conducting conditions are
particle is placed in the middle of the
microchamber. The density of this particle is C Ci  at inlet
chosen to be exactly the same as the fluid which i, i 1, . . . , n entrances A and B
is filled in the micromixer. The electrolyte prop-
erties and dimension of this micromixer are listed
in Table 1. @Ci
0 at all the walls (2b)
Four electrodes are used to control this @n t 0
micromixer. E1 is responsible for generating the
electroosmotic flow to transport the fluid in and and the initial concentration in the system is
out of the mixing chamber, in addition to induc-
ing the vortices. To achieve the mixing function,
it is necessary to control the position of the Ci jt0 0 through the microchannel (3)
A 78 Applications of Induced-Charge Electrokinetic in Microfluidics

Key Research Findings microvalve requires 3.7 s to switch the flow direc-
tion without any leakage. This operation period
ICEK Microvalve can be controlled by changing the applied electric
Applying positive voltage to electrode 1 and neg- field; e.g., when applying 400 V/cm, the
ative voltage to electrode 2, the electric field E1 is microvalve requires only 2.3 s to switch the
generated from left to right. As a result of the flow direction. This switching time will be
applied electric field, vortices are induced around increased to 7.82 s when the applied electric
the conducting hemisphere of the heterogeneous field is reducing to 100 V/cm.
particle (Fig. 3a) and push the particle to move The flow rate of the fluid that is pumped out of
toward the microchannel B (Fig. 3b). Once the this microvalve is adjustable. This adjustment is
heterogeneous particle reaches the entrance of possible by using different values for E*1 or E*2 in
microchannel B and blocks it (Fig. 3c), the elec- the valid range without affecting the blocking
trodes 1 and 5 will be simultaneously activated to position of the heterogeneous particle. The
produce the electric field E1 = 200 V/cm. This numerical simulations show that while the dom-
particle does not move as long as the electric field inant electric field is E1 = 200 V/cm, the addi-
E1 is applied. The goal of blocking the desired tional electric field, E*2, has to be between
microchannel without leakage in a short period of 29 V/cm and 46 V/cm in order to have a no
time is achieved (Fig. 3c). However, since the leakage microvalve. Similarly, applying the dom-
electric field direction is from left to right, it is inant DC electric field of E2 = 200 V/cm, the
not possible to direct the fluid to flow into the additional electric field, E*1, can vary from
microchannel C (Fig. 3c). To reach this goal, 49 V/cm to73 V/cm to pump the flow toward
a second DC electric field (perpendicular to microchannel B, while the microchannel C is
E1 = 200 V/cm) is applied, E*2 = 35 V/cm. This blocked by the heterogeneous particle.
second electric field is used to drive fluid flow
through the microchannel C. This is chosen suf- ICEK Micromixer
ficiently weaker than E1 so that it will not affect Consider a microchamber described in Fig. 2
the blocking of microchannel B (Fig. 3d). The which has no particle inside. Two parallel
heterogeneous particle tends to align itself with streams of fluids enter the chamber, once the
the direction of the resultant DC electric fields DC electric field E1 is applied. One stream is
(summation of E1 and E*2) but still blocks the a buffer solution carrying a sample species with
microchannel B. a sample concentration of CA = C0. The other
To change the flow direction, E1 will be turned stream is a pure buffer solution with zero sample
off, while E2 = 200 V/cm is applied using elec- concentration, CB = 0 (mol/m3). The only mixing
trodes 3 and 4. Figure 3eg show that the hetero- mechanism (to mix these two parallel laminar
geneous particle rotates to align itself with the flows) is the diffusion at the interface region
new applied DC electric field and moves toward along the micromixer. As demonstrated in
the microchannel C and eventually blocks Fig. 4a, at the exit of the chamber, the two fluids
it. Once microchannel C is blocked, the electric are poorly mixed. The resultant mixture is not
voltage is applied to electrode 6 simultaneously acceptable to be used in microfluidic devices.
to provide the same E2 = 200 V/cm, and keeps Releasing a fully conducting particle into this
the heterogeneous particle at the entrance of microchamber, the ICEK micromixer is pro-
microchannel C (Fig. 3h). By applying voltages duced. The induced vortices will be formed
to electrodes 1 and 2, a weak DC electric field, around the conducting surface once the electric
E*1 = 50 V/cm, is provided which manipulated field is applied. These four induced vortices are
fluid flow toward the microchannel B (Fig. 3i). circulating the fluid across the streamlines of the
The simulation study showed that under the two different fluid streams and boost the mixing
applied DC electric field of 200 V/cm, this ICEK process. The fully conducting particle moves
Applications of Induced-Charge Electrokinetic in Microfluidics 79 A
Applications of Induced- a b
Charge Electrokinetic in
Fig. 3 Sequences of A
E1=200V/cm E1=200V/cm
a microvalve which is
operated by induced-charge
electrokinetics using C N/C C N/C
a heterogeneous particle
inside it. The flow pattern
and vortices are shown by t1 = 0.0 s t2 = 1.1 s
E2=0 V/cm E2=0 V/cm
plotted streamlines. The
normalized vectors show c d
the direction of the fluid.
The diameter of the
heterogeneous particle and E1=200V/cm E1=200V/cm
the microchamber are
20 and 40 mm, respectively.
The height of the C N/C C N/C
microchamber is 40 mm.
The results are presented at
the xz plane crossing the t3 = 3.3 s t4 = 3.6 s
E2=0 V/cm E*2=50 V/cm
middle of the 3D
microvalve at different time e f
steps. The zeta potential on
the nonconducting
hemisphere of the E1=0V/cm E1=0V/cm
heterogeneous particle is
set to 50 mV, while the N/C N/C
zeta potentials on the
microchannel and
microchamber walls are set t5 = 4.9 s E2=200 V/cm t6 = 5.7 s E2=200 V/cm
to 15 mV g h

E1=0V/cm E*1=50 V/cm


t7 = 6.3 s E2=200 V/cm t8 = 7.3 s E2=200 V/cm

E1*=50 V/cm

t9 = 7.7 s E2=200 V/cm

A 80 Applications of Induced-Charge Electrokinetic in Microfluidics

Applications of Induced-Charge Electrokinetic in time steps. (d) Comparison of the concentration distribu-
Microfluidics, Fig. 4 Flow pattern and concentration tion at the outlet of the micromixer at t=12s. E1 represents
distribution in the micromixer. The red lines inside the the horizontally applied electric field and is equal to
microchamber represent the streamlines of the flow field. 40V/cm. DC electric field (E1=40V/cm) is applied from
The color bar shows the concentration distribution. The left to right. C represent conductive material. Zeta
figure shows a micromixing chamber (a) without a particle potential of channel and chamber walls is 50 (mV)
inside, (b and c) with a conducting particle at different

toward the exit of the chamber driven by the conducting particle away from the right-side wall
applied electric field E1. However, the particle of the microchamber. At the same time, two
stops moving when it gets close to the right-hand vortices on the left-hand side of the fully
side wall of the microchamber. The motion of the conducting particle push the particle to the right.
rotating fluid (caused by vortices) is restricted by When these forces are balanced, the particle
the microchamber wall. Thus, the momentum of remains at an equilibrium position. Conse-
the fluid motion converts to pressure, which quently, the particle will not block the entrance
results in a repellent force between the wall and of the right microchannel. As shown in Fig. 4c the
the particle. This repellent force pushes the fully induced vortices around the fully conducting
Applications of Induced-Charge Electrokinetic in Microfluidics 81 A

Applications of Induced-Charge Electrokinetic in concentration distribution. Comparison of the concentra-

Microfluidics, Fig. 5 Control of the positions of the tion distribution at the outlet of the micromixer at
particle and corresponding induced vortices using differ- t = 12 s. E1 = 40 V/cm is horizontally applied from
ent DC electric fields perpendicular to E1. The new con- left to right, and E2 represents vertically applied
figuration of electric fields boosts the mixing process. The electric fields. C: conductive. Zeta potential of channel
red lines inside the microchamber represent the stream- walls is 50 (mV)
lines of the flow field. The color bar shows the

particle significantly enhance the mixing process. symmetrical to the center line of the
The concentration profile at the exit of the microchamber all the time (Fig. 4b, c). Each of
microchamber is plotted in Fig. 4d for these two these induced vortices covers only one-quarter
investigated cases. These numerical simulations region of the microchamber, and they are located
illustrated that using a conducting particle inside in one stream only. In order to improve the
the micromixer enhances the mixing process. mixing result, it is desirable to have each induced
Under the applied only one DC electric field, vortex be located across the boundary of the two
E1, from left to right, the induced vortices are streams. Thus, such a vortex can circulate fluid
A 82 Arnoldi-Based Model Order Reduction

from the high-concentration stream to the Induced-Charge Electrokinetic Motion of

low-concentration stream and vice versa. To Particle in a Microchannel
achieve this goal, a secondary electric field E2, Measuring Zeta Potential, Methods
perpendicular to E1, is added into the system. Microfluidic Mixing
Once the secondary electric field, E2, applies
to the system, the net electric field direction will
change. Consequently, the positions of the References
conducting particle and configuration of the
induced vortices in the chamber are affected by 1. Terry SC, Jerman JH, Angell JB (1979) A gas chro-
matographic air analyzer fabricated on a silicon
the net applied electric field. In Fig. 5a it is shown
wafer. IEEE Trans Electron Dev 26:18801886
that some part of these vortices covers both 2. Peirs J, Reynaerts D, Van Brussel H (2000) Design of
streams and boosts the mixing process. The two miniature parallel manipulators for integration in
vortices between the conducting particle and a self-propelling endoscope. Sens Actuator A Phys
microchamber walls prevent the particle to
3. Beebe DJ, Moore JS, Bauer JM, Yu Q, Liu RH,
reach the microchamber walls while helping the Devadoss C, Jo B (2000) Functional hydrogel struc-
mixing process. Figure 5d compares the concen- tures for autonomous flow control inside microfluidic
tration field at the exit of the micromixer as channels. Nature 404:588590
4. Kohl M, Dittmann D, Quandt E, Winzek B (2000)
a function of different values of E2. This figure
Thin film shape memory microvalves with adjustable
demonstrates that the best mixing result is operation temperature. Sens Actuator A Phys
achieved when the applied electric fields have 83:214219
the same magnitudes. The numerical results 5. Daghighi Y, Li D (2011) Microvalve using induced-
charge electrokinetic motion of Janus particle. Lab
show that this ICEK micromixer (E1 = E2)
Chip 11:29292940
requires small mixing length and can provide 6. Biddiss E, Erickson D, Li D (2004) Heterogeneous
a homogeneous mixture at its exit. surface charge enhanced micromixing for electroki-
netic flows. Anal Chem 76:32083213
7. Harnett CK, Templeton J, Dunphy-Guzman KA,
Senousy YM, Kanouff MP (2008) Model based
Examples and Applications design of a microfluidic mixer driven by induced
charge electroosmosis. Lab Chip 8:565572
In this entry two novel applications of ICEK 8. Olesen LH, Bruus H, Ajdari A (2006) Ac electroki-
netic micropumps: the effect of geometrical confine-
phenomena are introduced (microvalve and
ment, faradaic current injection, and nonlinear surface
micromixer). These applications could be consid- capacitance. Phys Rev E 73(5):056313_1056313_16
ered as reliable solutions to the drawbacks of 9. Sugioka H (2008) Suppression of reverse flows in
current electrokinetic microdevices. The operat- pumping by induced-charge electro-osmosis using
asymmetrically stacked elliptical metal posts. Phys
ing demonstration of an ICEK microvalve is
Rev E 78(5):057301_1057301_4
tested when a heterogeneous particle is used. 10. Daghighi Y, Li D (2013) Numerical study of a novel
Also an efficient ICEK micromixer is introduced induced-charge electrokinetic micromixer. Anal
which works based on the induced vortices Chim Acta 763:2837
around a conducting particle inside its
microchamber. These vortices will significantly
enhance the mixing of the different fluid streams
which are entering the mixing chamber. Arnoldi-Based Model Order

Cross-References Synonyms

Fabrication of 3D Microfluidic Structures Arnoldi algorithm; Arnoldi-based

Induced-Charge Electrokinetic Flow macromodeling
Atomic Force Microscope (AFM) 83 A
Aspect Ratio Dependent Etching
Arnoldi-based algorithm is a classical Krylov- A
subspace-based Model Order Reduction tech- Synonyms
nique. It reduces the dimension of the spatially
semi-discretized form of the original governing ARDE
PDEs using moment-matching of transfer

Cross-References ARDE is the phenomenon of the etching rate

related to the aspect ratio. The larger the aspect
Macromodel ratio, the lower the etching rate.
Model Order Reduction (MOR)

Anisotropic Etching
Artificial Neural Networks
Plasma Etching
Reactive Ion Etching (RIE)

An artificial neural network is an information
processing paradigm inspired by way the human Test
nervous system processes information. It consists
of an interconnected group of processing ele-
ments (neurons) which can exhibit a complex Definition
global behavior, determined by the network
topology (the scheme used to connect elements A method for the determination of the level of a
together), and the algorithm (the rules) employed given analyte in a given sample or set of samples.
to specify the values of the weights connecting
the nodes, the connections between the neurons.
Based on training rules (whereby the weights of
connections are adjusted on the basis of presented
Lab-on-a-Chip Devices for Chemical Analysis
patterns), ANNs can be trained to generate an
arbitrary number of outputs in response to an
arbitrary number of inputs.
Atomic Force Microscope (AFM)

Cross-References Definition

Neural Network Material characterization technique which uses

Model Order Reduction (MOR) a tip (typically silicon) on the end of a cantilever
A 84 Atwood Number

to determine the surface profile of a material. As

the cantilever is moved across the surface, a laser Atwood Number
is reflected off of the backside. As the cantilever is
deflected, the change in laser beam position is Definition
detected by a photodetector. This information is
then used to create a three-dimensional image of A dimensionless number that describes the den-
the surface, with atomic-scale resolution. sity difference between two adjacent fluids with
a common interface.

AFM Cross-References
Fabrication of Self-Assembled Catalytic
Nanostructures Piezoelectric Microdispenser