You are on page 1of 14

Sensors and Actuators B 117 (2006) 495–508

Convective mixing and chemical reactions in microchannels


with high flow rates
Norbert Kockmann ∗ , Thomas Kiefer, Michael Engler, Peter Woias
Laboratory for Design of Microsystems, Department of Microsystems Engineering (IMTEK), University of Freiburg,
Georges-Köhler-Allee 102, D-79110 Freiburg, Germany
Received 9 June 2005; received in revised form 20 December 2005; accepted 3 January 2006
Available online 9 February 2006

Abstract
This work presents a theoretical and experimental investigation of convective micromixing in various mixer structures and combinations with
the aim of high mixing intensity and a high throughput. Different mixing elements are integrated on a silicon chip to achieve a device for high flow
rates above 20 kg/h. Test structures are fabricated and characterized according to their flow behavior and mixing performance. Flow measurements
with pH neutralization and indication by bromothymol blue confirm the numerical simulations of the flow characteristics and concentration fields.
The integral mixing quality in the micromixer is measured with the iodide–iodate reaction (Villermaux–Dushman) and shows excellent values for
high Re numbers. This offers the potential to use microstructures for new applications in the production of chemicals. With the help of the obtained
experimental and theoretical results, a new class of dimensionless numbers is proposed which characterizes the effectiveness of a mixing device
and of the mixing process and compares different mixing times.
© 2006 Elsevier B.V. All rights reserved.

Keywords: Mixing quality; Mixing time; Vortex flow; Micromixer; Mixing effectiveness; Microreactor

1. Introduction with stretching, splitting, recombination, and at last diffusive


mixing (e.g. [7]). In these mixers, laminar flow is predominant
The mixing process is fundamental in many unit opera- and diffusion limits the mixing process. A successful microre-
tions and forms the basic operation in other processes like actor, i.e. providing a complete reaction with a high selectivity
heat exchange or chemical reactions. The short diffusion length and product yield, is based on an efficient mixing process. More
between the components in microstructures results in a short specific, the time scale of the mixing process in relation to the
mixing time, which has been shown by various applications time scale of the reaction plays the major role, as the mixing
in analytical equipment [1] and for laboratory applications [2]. process must be finished before the reaction is completed. A
Meanwhile, the application of microstructured equipment has detailed description of the processes and relevant time scales
entered the production of chemicals and the corresponding pro- can be found in Bourne [8]. Microstructured T-mixers have been
cess development [3]. often investigated in the past, however, mainly in the straight
Microreactors belong to the most predominant equipment in laminar flow regime [9,10]. Not only the mixing characteris-
the developing area of micro process engineering. Most of the tics are important, but high throughput and mass flow rates at
microreactors studied today are running in continuous operation tolerable pressure losses are also essential factors. To achieve
under laminar flow conditions (see [2]). Microreactors are mix- a high mass flow rate, either the number of channels could be
ing two (or more) components using various principles (see e.g. increased, the channel cross-section could be enlarged, or the
[4]). One can distinguish between active [5] and passive mixing flow rate in the channels could be increased. The first mea-
[6]. In passive mixers the flow itself induces various streams sure is practiced very often and is connected with the concept
of numbering-up [2]. The main difficulty with numbering-up
is the homogeneous flow distribution in the inlet and outlet
∗ Corresponding author. Tel.: +49 761 203 7499; fax: +49 761 203 7492. manifolds. Initially, larger channel cross-sections seem contra-
E-mail address: kockmann@imtek.de (N. Kockmann). dictory to miniaturization benefits, however, together with the

0925-4005/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.snb.2006.01.004
496 N. Kockmann et al. / Sensors and Actuators B 117 (2006) 495–508

increased fluid velocity the larger channels lead to higher Re


Nomenclature numbers and the occurrence of vortices. The diffusive mix-
ing is assisted by these vortices and convective effects, which
AM cross-section of mixing channel (m2 )
were first studied in conventional T-shaped micromixers and
b channel width (m)
reported by Kockmann et al. [11]. The flow process and mix-
c concentration (kmol/kmol)
ing characteristics are reported in succeeding papers [12,13].
D diffusion coefficient (m2 /s)
Recently, Wong et al. [14] presented a comprehensive study on
Dax axial diffusion coefficient (m2 /s)
the flow phenomena and mixing characteristics in a T-mixer,
DaI first Damköhler number (−)
but their simulations do not match with their experimental find-
DaII second Damköhler number (−)
ings. Deerberg et al. [10] employ the T-shaped micromixer to
De Dean number (−)
formulate their cell–grid-model and give some general mixing
dh hydraulic diameter (m)
results.
Eu Euler number (−)
This paper presents the design and characterization of single
F force (N)
mixing elements and their suitable combination for high mass
k kinetic constant (unit depends on reaction order
throughput. The flow of a liquid (water) is numerically simu-
m, here M4 /s)
lated with the software package CFD-ACE+ from ESI Group
l length (m)
and characterized by the mixing quality. Single mixing struc-
lm mixing length
tures are systematically combined in series to give optimized
lM length of mixing channel (m)
mixing elements and arranged in parallel (up to 16 mixing chan-
MEI mixer effectiveness (−)
nels) on a single mixer chip. The microchannels are fabricated
MEII mixing effectiveness (−)
on two silicon wafers by deep reactive ion etching (DRIE) using
m reaction order (−)
the ASE process, bonded, and covered by a Pyrex glass lid to
n mole number (mol)
observe the flow structures under a microscope. The mixing
p pressure (N/m2 )
of two fluids with different pH values combined with the indi-
p pressure difference (N/m2 )
cator bromothymol blue gives an evaluation of the simulation
Re Reynolds number (−)
results. The characteristic time of the chemical reaction, a result
r pipe radius (m)
of the reaction kinetics, is given for the species concentration
R bend radius (m)
and the reaction order. Characteristic time scales of the T-shaped
Sc Schmidt number (−)
micromixer are determined to describe the micromixer with the
tD diffusion time (s)
Damköhler number. The overall mixing quality is determined
tm mixing time (s)
further by the iodide–iodate reaction (Villermaux–Dushman), a
tP process time (s)
competitive-parallel reaction, which is well documented in lit-
tR characteristic reaction time (s)
erature [15–17]. To describe the mixing process quantitatively,
ū mean velocity (m/s)
the iodide–iodate reaction gives an indication of the mixing time
Xs segregation index (−)
in the investigated structures.
Y yield of iodine (−)
A new concept for evaluating the mixing performance is
Greek symbols presented to compare different mixer types and mixing pro-
α mixing quality (−) cesses. These dimensionless numbers can easily be determined
ε specific energy dissipation ((N m)/(kg s)) from primary data of the mixing process and equipment and
ν kinematic viscosity (m2 /s) present many opportunities for use in other mixing application
νi stoichiometric coefficient (−) fields.
ρ density (kg/m3 )
σ mean square deviation (−) 2. Convective flow regimes and mass transfer
␨vortex friction factor originating from vortex creation
(−) The channels in micromixers can be divided into two parts.
␨friction friction factor originating from wall friction (−) The first part, e.g. a T-shaped or Y-shaped junction, serves as the
␨total total friction factor (−) contacting unit for two or more fluids. The second part is used for
further in-line mixing, mixing enhancement or provides the resi-
Subscripts dence time needed for chemical reaction or other transformation.
in inlet channel These two parts fulfill different duties and are shaped accord-
M mixing channel ing to their tasks. The contact device can be designed in various
ST total segregation forms for its mixing characteristics, which has been shown for
T-shaped and similar devices in [13]. The mixing channel can
be designed in various forms with surface elements like the her-
ringbone mixer [7] or with a meandering structure ([2], Chapter
2). The mixing process in bent channels is described in the fol-
lowing.
N. Kockmann et al. / Sensors and Actuators B 117 (2006) 495–508 497

2.1. Bent channels bend radius R:



r
2.1.1. Simulations of 90◦ bends De = Re . (1)
R
The flow in microchannels is mainly regarded as stratified
laminar flow. This is correct for straight channels with a low If the De number is larger than 10, the first symmetrical double
flow velocity and, therefore, a low Reynolds number Re. The vortex forms in a curved channel. The comparison of the two
occurrence of turbulence is related to a critical Re number, which major forces in bends, the centrifugal force FC and the viscous
is Recrit = 2300 for the flow in a straight pipe or channel. Heat friction FV at the wall indicates the damping of vortices in narrow
and mass transfer in the laminar flow regime is controlled by and flat channels. A high ratio indicates vortex development in
conduction and diffusion, respectively. the bend which results from high flow velocities and channels
The straight laminar flow breaks up when the fluid flows with nearly rectangular cross-sections [19].
through curves and bends. Centrifugal forces push the fluid from
the center of the channel, where the bulk fluid flows with high 2.1.2. Mixing and mass transfer in bends
velocity, to the outer side. The fluid at the outer wall is pressed The flow in the microchannels is numerically simulated by
either up or downwards producing a double vortex filling the CFD-ACE+ from ESI Group and graphically analyzed [20]. In
entire mixing channel. The centrifugal force depends mainly on the following examples, the mixture of two fluids (water, 20 ◦ C)
the mean flow velocity and the bend radius. The viscous wall is treated with a flow ratio of 1:1. The fluid dynamics are fun-
friction acts against the centrifugal force and therefore dampens damental for the transport processes in microchannels like heat
the vortex flow. Along the outlet channel, the vortex flow dis- and mass transfer. The concentration distribution in a channel
appears and straight laminar flow readjusts again. According to with a 90◦ bend is shown in Fig. 1.
the work of Dean [18], dealing with the laminar flow in curved The fluid stream with the two unmixed components enters
pipes with low Re numbers, the flow regimes and generation of the bend (data of rhodamine B in water, D = 2.8 × 10−10 m2 /s,
vortices depend on a dimensionless number De including the Sc = 3588). Due to the centrifugal forces the interface between
Reynolds number Re and the ratio of the pipe radius r to the the two parts forms a bell-shaped curve directly at the beginning

Fig. 1. Streamlines and concentration distribution in the cross-sections of a 90◦ bend (100 ␮m × 100 ␮m).
498 N. Kockmann et al. / Sensors and Actuators B 117 (2006) 495–508

sections between 100 ␮m × 100 ␮m and 500 ␮m × 500 ␮m to


flat channels.
For small Re numbers, the mixing quality is relatively high
due to the diffusion between the two slowly flowing compo-
nents. The velocity profile is straight laminar and the molecules
have sufficient time to diffuse to the opposite side. In small
channels (100 ␮m × 100 ␮m), the diffusion distance is short,
which results in a higher mixing quality than in larger chan-
nels (500 ␮m × 500 ␮m). With increasing fluid velocity and Re
number, the residence time of the fluid and the mixing quality
decrease. Wide and flat channels (e.g. 500 ␮m × 50 ␮m) show
the worst mixing quality, as a result of the large diffusion dis-
tance. For Re numbers larger than 10, the vortex pair is induced
in the bend, which increases the mixing quality in the outlet
channel. Convection effects and the enlargement of the interface
Fig. 2. Mixing quality behind a 90◦ bend for various geometries (width × height, between the components play the dominant role. This observa-
dimensions in micrometers) and Re numbers, diffusion of rhodamine B dye in
tion is in accordance with the statement in Kraume ([22], Chapter
water.
9), that in static mixers the flow departs from laminar behavior
of the outlet channel (0 ␮m). At a distance of 100 ␮m around for Re numbers larger than 20. In channels with square cross-
the bend, the interface forms already a mushroom shape. The sections, the mixing quality increases in the same way. In flat
length of the interface is enlarged by the vortex flow, and the channels, the channel dimensions show no effect on the mix-
potential for an exchange of the components is increased. The ing quality, if only the Re number is considered. The vortex
mushroom shape of the interface marginally changes between formation is dampened by the viscous wall friction, hence, the
the outlet length of 200 and 500 ␮m. Between 500 and 1000 ␮m, enhancement of the mixing quality is weaker with increasing Re
only the width of the interface grows due to the diffusion of the number than in square ducts.
components. The concentration distribution in a cross-section
has to be corrected according to the numerical diffusion, which 2.2. T-shaped micromixers
depends on the numerical grid distance, and may lead to higher
2.2.1. Simulations of T-shaped micromixers
diffusion rates compared to real systems. A similar picture of the
The geometrical setup of the channel structure in a T-shaped
concentration distribution is given by Schönfeld and Hardt [21]
joint of rectangular microchannels is displayed in Fig. 3, left
for a helical flow in a curved microchannel with a bend radius
side. The mixing of two fluids (water, 20 ◦ C) is characterized for
of 1000 ␮m (Dean flow).
a constant mixing ratio of 1:1 and a diffusivity of rhodamine B
To describe the mixing process, the mixing quality α is
in water (D = 2.8 × 10−10 m2 /s, Sc = 3588). For low flow veloci-
defined for a cross-section as:
 ties, Re < 10, the flow is laminar and the streamlines are straight
σ2 (see Fig. 3, right side).
α=1− 2
(2) With increasing flow velocity the centrifugal forces of the
σmax
bent flow induce symmetrical vortex pairs at the entrance of
This value describes the relative variance of the concentration the mixing channel [19] (see Fig. 3, middle right side). Vis-
in a cross-section of the channel (for more details see [22], cous forces dampen the vortices and the flow becomes laminar
Chapter 1). In Fig. 2 the mixing quality is displayed for vari- again after a certain distance. The mass transfer between the two
ous Re numbers at a constant length of three times the channel fluids is controlled by diffusion as can be seen in the concentra-
width downstream in the mixing channel, where the laminar tion profiles in the cross-section of the mixing channel (Fig. 3,
flow readjusts again. The geometry varies from square cross- middle row). The symmetry of the vortex pairs is broken for

Fig. 3. Numerical simulation of the fluid flow and mixing in a T-shaped micro mixer; left: geometry of the numerical model; middle: concentration distribution in
the mixing channel for 1:1-mixing; right: streamlines in a T-shaped micro mixer for three flow situations: straight laminar, laminar vortex, and engulfment flow.
N. Kockmann et al. / Sensors and Actuators B 117 (2006) 495–508 499

velocity stays on the inlet side and forms a large vortex. The fluid
located at the sharp, outer part of the swapping area streams next
to the stagnation region at the backside of the mixing channel
and takes part in the small vortex, which is displayed in Fig. 4.
The streamlines at the sharp tip of the grey-tone area form an
intensely mixed zone in the mixing channel, which can be seen
in the multiple intertwining of the streamlines. This is also the
vortex center, which can be seen in Fig. 3, right side.
This pair of small vortices starts at the corner region of the
stagnation area and shows a strong laminar swirling without
any turbulence, i.e. no chaotic transient behavior. With higher
Re numbers, the starting point of transient effects is located in
Fig. 4. Engulfment flow in a T-shaped micro mixer (200 × 100 × 100) at a mass this area. The fluid forming this small vortex pair comes from
flow of 73 g/h and a corresponding Re number of approx. 135; streamlines for the sharp edge of the swapping area of the entrance flow.
the small vortex pair which indicating the area with a high mixing intensity; The high concentration gradients could not be simulated due
encircled area in the small images (inlet cross-section) indicated the range of to the limited numerical resolution of the numerical grid. The
fluid which swaps into the opposite side in the mixing channel.
numerical diffusion prevents the visualization of the small flu-
idic lamellae coming from the strongly swirling flow in the
higher Re numbers at the entrance of the mixing channel and T-junction. To avoid this, a view into the direction of the mix-
fluid swaps to the opposite side. The critical Re number, when ing channel containing streamlines is given in Fig. 5. Each inlet
symmetry is broken, depends on the geometry and flow condi- channel is represented by 3500 streamlines (black from the right
tions, and ranges from 120 to 300. This asymmetrical flow gives side), which intermingle in the mixing channel. The fine lamel-
an additional interfacial area for mass transfer and dramatically lae structure as well as the influence of the Re number is clearly
increases the mixing quality. A more-detailed analysis is given visible.
by Kockmann et al. in [20].
A closer look into the flow conditions of this so-called engulf-
2.2.2. Experimental characterization of T-shaped mixing
ment flow gives a deeper insight into the convective-mixing
elements
process in a T-shaped micromixer (see Fig. 4). The flow field
The typical geometrical shape of a T-shaped mixer with
at the entrance of the mixing channel can be divided into three
homogeneous inlet flows is shown in Fig. 6, left side. Above
major areas:
a certain Re number (120–300, depending on the geometry and
aspect ratio) the symmetry breaks up and fluid segments enter
• stagnation point flow at the backside of the mixing channel; the opposite side of the mixing channel producing small flu-
• small vortex pair with strong rotation and fast engulfment; idic filaments. Two major vortices are established in the mixing
• large vortex region with slow rotation. channel which relaminarize downstream in the mixing channel.
A top view of this engulfment situation is shown in Fig. 6, a
The stagnation region and the small vortex pair are characterized simulation of the concentration profile in the left image, and
by a strong stretching and folding of the fluidic elements and are, an experimental image with bromothymol blue in the middle.
therefore, the key regions for the mixing enhancement. A 3D image of the flow and concentration profile with water
The grey areas in the side views A and B of Fig. 4 indicate and a rhodamine B solution is given in Fig. 6, right side, which
the portion of the inlet flow, which goes to the opposite side in was produced with the help of a micro-LIF process from Hoff-
the mixing channel. This swapping area includes the middle of mann et al. [24]. Here, the small fluid filaments and the high
the entrance flow with the highest velocity. The fluid with lower concentration gradients are clearly visible.

Fig. 5. Numerical simulation of the streamlines at two different Re numbers, illustrated by 7000 streamlines each at two different Reynolds numbers, ReM = 179
(left) and ReM = 249 (right), black lines from the right side.
500 N. Kockmann et al. / Sensors and Actuators B 117 (2006) 495–508

Fig. 6. Qualitative comparison between simulation (left side) and experimental images (middle) in a micro mixer with rectangular cross-section inlet channels
(300 ␮m × 300 ␮m width × depth) and mixing channel (600 ␮m × 300 ␮m width × depth) with a mass flow of 260 g/h and a Re number of about 180; right side:
Engulfment flow in a T-shaped micro mixer (200 × 100 × 100), 3D image of rhodamine B-dyed solution and clearly visible fluid filaments at Re number of 180 [21].

The simulation results in Fig. 6, left side, give a good of the formed iodine:
indication of the flow profile and concentration distribution. 6[IO3 − ]0
To obtain meaningful mixing characteristics from experimen- YST = − . (5)
6[IO3 ]0 + [H2 BO3 − ]0
tal investigations, many methods have been proposed, which
often deal with chemical reactions [8,17]. For the presented A small value of XS denotes a good mixing process since a
mixing geometries, the competitive-parallel iodide–iodate reac- low segregation index indicates either fast mixing or homo-
tion (Villermaux–Dushman reaction) was implemented, which geneous mixing or a combination of both. Unfortunately, the
is described in more detail in [23]. It consists of an almost instan- influence of the different effects cannot be separated. The seg-
taneous reaction (neutralization of H2 BO3 − ) and a reaction that regation index is displayed over the Re number for the T-shaped
is nearly as fast as the mixing process (creation of iodine I2 and micromixer with three different geometries in Fig. 7. For com-
I− 3 ). Details can be found in [15,16], while the implementation parison, the segregation index in a stirred vessel achieves values
for micromixers is shown in [25]. The initial concentrations of of 0.1–0.003 depending on the feed point [26]. Mixing processes
the different species are crucial for the results. The initial con- in microchannels give a segregation index of higher than 0.01
centrations used in this work are shown in Table 1. [9], but a direct comparison of the different devices is difficult
The concentrations are given for one inlet only. Since the [23], due to the various mixing effects.
mass flow ratio of the two inlet streams is adjusted to 1:1, the The decreasing segregation index with increasing Re num-
overall concentrations in the mixed fluid would reduce to half ber indicates the improving mixing process in the T-shaped
of the initial concentrations, if no chemical reaction occur. The micromixer. All mixers are operating in the engulfment flow
segregation index XS is calculated by regime. The segregation index reaches a plateau or is somewhat
increased for higher Re numbers which can be explained by the
Y formation of transient, fluctuating vortices at the entrance of the
XS = , (3)
YST mixing channel. These vortices transport unmixed fluid through
the mixing channel and do not participate in the axial mixing
where Y is the ratio of the acid’s mole number consumed by the process. The miniaturization results in a fast mixing process,
slow reaction (the creation of iodine I2 and I− 3 ) and the total indicated by a lower XS value for the smaller geometries.
acid mole number: The complex situation of the engulfment flow in the mixing
     channel provides good mixing results with high mass flow rates.
2 nI2 + nI3 − 4 [I2 ] + I3 −
Y= =   , (4)
n H0 + H+ 0

where the value of [H+ ]0 is given in Table 1. The concentra-


tions of I2 and I3 − are spectroscopically determined. Finally,
YST is the value of Y in a total segregation case when mixing
is considered to be infinitely slow. With this, the two reactions
appear as quasi-instantaneous compared to mixing and acid is
consumed in proportion to the local concentration of both borate
and iodide/iodate ions. YST is then related to the concentration

Table 1
Initial concentrations of the involved species for the iodide–iodate reaction
(H2 BO3 − )0 (mol/l) (H+ )0 (mol/l) (I− )0 (mol/l) (IO3 − )0 (mol/l)

Inlet 1 0.0455 0 0.0159 0.0033 Fig. 7. Segregation index XS over the Re number in the mixing channel for
Inlet 2 0 0.028 0 0 three scaled T-mixers (ratios 3:2:1). Nomenclature: width of the mixing chan-
nel × width of the entrance channel x depth of the channels in micrometers.
N. Kockmann et al. / Sensors and Actuators B 117 (2006) 495–508 501

Due to the relaminarization of the flow in the mixing channel, (a) Asymmetric T-mixer: consists of a simple T-junction where
some parts of the flow remain unmixed. To obtain a complete one inlet channel is bent in order to create asymmetric flow
mixing of the two fluids for Re numbers lower than approx. 500, conditions at the junction. This leads to a stronger distur-
a single contact element is not sufficient and additional mixing bance of the flow symmetry and an enhanced mixing quality.
elements are needed. Single mixing elements have been varied in (b) T-tree mixer: the mixing channels of two equal T-junctions
different combinations to find an optimum geometry, such as the merge into a third T-junction. With this configuration, addi-
channel length between 90◦ bends. The next section describes tional alternating fluid streams are created.
the design and setup of optimized mixing elements. (c) Tangential mixer: two liquid streams enter a cylindrical mix-
ing chamber tangentially and leave through a transition hole
located in the center of the chamber, inducing a swirling
3. Design and fabrication of high throughput mixers
movement of the fluids.
(d) T-cascade mixer: two equal T-junctions are connected in
Considering the simulation results as well as the necessity for
series, leading to a two-fold impingement of the fluid
a high throughput and a low pressure loss, optimized micromixer
streams. At the second junction, the inlets are interchanged
structures were designed. For this, several single mixing ele-
in order to create alternating fluid streams.
ments, each consisting of the combination of several elementary
structures, were arranged in parallel. For topological reasons, i.e.
to prevent intersections between the channels, the parallelization Besides the possibility of a parallel arrangement of the mixing
requires a two-layer design. The numerical simulations give the elements, the two layers facilitate short distances between bends
first geometrical indications for an optimal combination of mix- in different directions without the need for much lateral space.
ing structures. A short mixing time can be achieved due to the proximity of the
contact elements.
The four mixing configurations were fabricated on 4 in. sili-
3.1. Structural design and fabrication of the con wafers with a thickness of 525 ␮m each. The fluidic channels
convective-mixing elements on the front sides of the wafers and the connection holes on the
backsides are structured by the DRIE process (ASE process)
A total of four different combinations of mixing elements to obtain perpendicular walls. Rectangular channels with struc-
were designed, fabricated and tested. The schematic layouts of ture depths of 300 ␮m and widths of several 100 ␮m are easily
their superimposed layers are displayed in Fig. 8. The contact attained. Using this dry-etching method, many planar shapes in
elements on the first (bottom) layer are followed by a subsequent, the mixing structures are possible. With silicon fusion bonding
meandering channel shape located on the second (top) layer to (SFB), the two wafers are permanently connected. The struc-
provide further mixing. tures are sealed with a Pyrex lid that is attached to the silicon

Fig. 8. Schematic layouts of the single convective-mixing elements (dark grey: inlet channels with contact element on bottom layer/dark squares: transition holes/light
grey: meander shaped mixing channel on top layer).
502 N. Kockmann et al. / Sensors and Actuators B 117 (2006) 495–508

Fig. 9. Tangential mixer; top left: CFD simulation of the concentration profile within the mixer; top right: SEM image of the tangential mixing channel; bottom left:
SEM image of the meandering mixing channel, bottom right: monolithic array of 16 tangential mixers with inlet manifold (light grey) and outlet manifold (dark
grey).

using anodic bonding, which enable the visual observation of Fig. 10. The respective widths and lengths have been analytically
the flow inside the top layer. estimated by the pressure loss of laminar flow inside a channel
A total of 12 chips (20 mm × 20 mm) can be fabricated on a due to fluid friction:
standard 4-in. wafer. As previously mentioned, the mixing con-
ζfriction l ρ 2
figurations are designed for a multiple parallel arrangement of pfriction = ū . (6)
several single mixing elements on one chip. For the asymmetric Re dh 2
T- and the tangential mixers, 16 mixing elements are possible, Here, the pressure loss from the changing cross-section of the
whereas the T-tree and the T-cascade mixers allow only eight ele- channels has been neglected. Further details about the design of
ments. As a result, the total throughput can be strongly increased. the optimized micromixers and its result can be found in [27].
Fig. 9 shows a mask layout including 16 parallel tangential mix- A completely assembled tangential mixer chip,
ers together with SEM photos of the mixing element. 20 mm × 20 mm × 1.6 mm, is shown in Fig. 11. The out-
let hole is clearly visible in the middle of the chip. The mixer
3.2. Arrangement and dimensions of the micromixer chip is mounted on a fluidic connector block and either glued
elements onto the block or sealed with a silicone membrane. Visual
inspection is done by an optical microscope. The fluidic setup
A concentric design is chosen for the arrangement of the is described in [5,12].
single mixing elements in order to provide both a maximum
number of elements and equal pressure losses at each side of
the chip. Each of the two fluid streams enters the chip through
three inlet holes and is guided into the inlet manifold of the
mixing elements (see Fig. 9, bottom right). From here, the fluid
is distributed among the single mixing elements and led to the
second layer, where the fluid streams through the meandering
mixing channel into the outlet manifold. The outlet hole is placed
in the center of the chip.
To provide symmetrical flow conditions at each single mixing
Fig. 10. Schematic layout of the adjusted inlet channels (light grey) of two
element, the pressure loss inside the individual inlet and outlet mixing elements (white squares). The thin arrows indicate the flow direction
channels has to be equal. This has been achieved by broadening (“fluidic path”) from the inlet holes (dark grey) towards the mixing elements.
the inlet and outlet channels to a certain extent, as shown in Pressure is equal at each side of each mixing element, i.e. pL = pR .
N. Kockmann et al. / Sensors and Actuators B 117 (2006) 495–508 503

Fig. 11. Left: image of a complete mixer chip with 16 tangential mixing units, consisting of two microstructured silicon wafers covered with a Pyrex lid. Right:
fluidic connector block with silicon chip glued onto the block.

4. Experimental results The first result is the quadratic dependency of the pressure
loss on the mean flow velocity, according to the following equa-
As shown by the parallelization of the single mixing ele- tion (adopted from [28]):
ments above, the mixing chips have been designed for liquid  
flow with a high throughput. Besides the fabrication of the ρ ζfriction I ρ 2
p (ū) = ζtotal ū2 = ζvortex + ū , (7)
high throughput micromixers with 16 mixing elements, called 2 Re dh 2
“batch mixers”, mixer chips with a single mixing element of
each configuration have also been fabricated. This so-called where ζ friction is the constant multiplication factor originating
“single mixer” allows the examination of the mixing behavior from wall friction and ζ vortex the multiplication factor originat-
of each single convective element. For the measurements, the ing from both flow bending and vortex creation. Since the mass
overall mass flow rate, pressure loss, mixing times and integral flow rate is proportional to the mean flow velocity, the quadratic
mixing quality have been determined using the iodide–iodate dependency correlates to the data in Fig. 12 to a high degree, indi-
reaction. cated by the values of the correlation coefficient R2 , which lie in
the range between 0.99671 and 0.99993 for the eight chips. The
4.1. Pressure loss measurements quadratic dependency originates from the laminar operation of
the mixing devices combined with induced vortices in the mix-
The pressure has been measured by piezo-resistive pressure ing elements. The second main result is the effect of the higher
sensors using deionized water. Fig. 12 shows the pressure loss mass flow rate by parallelization on the overall pressure loss.
of the mixing chips in relation to the mass flow rate for the four This is clearly visible in the comparison of the mass flow rates
batch mixers (left side) and the four single mixers (right side). for the batch and the single mixing chips at any fixed pressure
Two important results can be extracted. loss using Eq. (7). Although the mass flow rates of the batch mix-

Fig. 12. Pressure loss over mass flow for the four batch (left) and the four single (right) mixing chips.
504 N. Kockmann et al. / Sensors and Actuators B 117 (2006) 495–508

Table 2
Comparison of the mass flow rates at equal pressure loss for the four mixing
chips in batch and single configuration
Chip type # Mixing elements Mass flow rate at equal
batch: single pressure loss batch: single

Asymmetric T-mixer 16:1 16.44:1


Tangential mixer 16:1 11.92:1
T-cascade mixer 8:1 6.56:1
T-tree mixer 8:1 7.68:1

ers are larger than the values of the single mixers, the pressure
loss changes only marginally (see Table 2). For the asymmetric
T-mixer, the relative pressure loss is even better in the batch than
in the single mixing chip. This originates from the broadening of
Fig. 13. Mixing times for the single mixers with respect to the Re number.
the channels in the batch mixer chips, as described in the design
and fabrication section. It indicates that the parallelization of
the single elements has been carried out in a very successful The mixing inside the tangential mixer occurs at lower mass
manner. Furthermore, as pressure loss indicates the amount of flow rates than inside the other mixers, as the simulations have
energy required by the mixing process, the batch mixers show indicated. Shorter mixing times in the tangential mixer have not
a much more effective mixing process for the same amount of been measured since partial mixing has no longer been observed
fluids than the single mixers. at mass flow rates larger than 360 g/h. The mixing time decreases
with increasing Re number and falls below 1 ms for all investi-
gated mixing configurations for Re > 700–900. The mixing times
4.2. Mixing times by optical measurements
for each mixer are very short, down to 300 ␮s for the asymmetric
T-mixer. The curves follow the same trend, indicating the same
For optical investigations of the mixing quality, measure-
mixing characteristics of convective mixing.
ments using the pH-induced color change of bromothymol blue
(BTB) have been performed. One inlet fluid consists of a 1.5 mM
solution of BTB with a pH value of 7, where BTB appears green. 4.3. Mixing quality by the iodide–iodate reaction
The other inlet fluid consists of a 67 mM solution of Na2 HPO4
with a pH value of 8, which is colorless. When the fluids are Results of measurements of the iodide–iodate reaction are
mixed together, a pH value of approximately 7.5 is the result. shown in Fig. 14 for the four designed mixers, together with
At this value, BTB appears blue. It is thus easily possible to measurements of a simple T-mixer with similar dimensions. The
optically distinguish the two unmixed inlet zones (green and col- decrease of the segregation index XS with increasing Reynolds
orless) and the mixed zone (blue). The optical investigations of number is evident, since the rising mass flow rate leads to a
the mixing quality show an almost homogeneous flow distribu- stronger vortex creation, thinner fluid filaments and therefore
tion in the mixing channels, indicating similar mixing conditions shorter diffusion lengths. The local oversupply of H+ -ions is
at each single mixing element. equalized faster, as less time is available for the creation of
For the determination of the mixing times, the mixing length iodine. Altogether, the segregation index of the mixers is very
lm can be determined by image processing. It has to be men- small (<0.001), indicating a very fast and effective mixing.
tioned that these measurements are a first estimation and they
will be improved in later investigations. The order of magnitude
is certainly reliable, but the actual values vary with a relative
uncertainty up to 20%. The mean velocity was determined by
measuring the mass flow rate:

ū = , (8)
ρS A M
where ρS is the density of the mixed solution (equivalent to
water) and AM is the area of the cross-section of the mixing
channel. The mixing time is calculated by the length where the
mixture is fully homogeneous, the mixing length lm , divided by
the mean velocity:
lm
tm = . (9)

Fig. 14. Segregation index for the four different single mixers compared with a
The results are shown in Fig. 13. simple T-shaped mixer.
N. Kockmann et al. / Sensors and Actuators B 117 (2006) 495–508 505

Finally, the improved mixing characteristics of the high through- The ratio can be expanded with the dimensionless groups of the
put mixers compared to the T-mixer can also be seen very clearly. Re number and the Schmidt number and Sc number. A ratio with
Unfortunately, there are some points to be considered when a value of 1 indicates a good mixing at the end of the mixing
analyzing the results of the iodide–iodate reaction. On the one channel; a higher ratio indicates an incomplete mixing. Hence,
hand, the relation of the segregation index to the mixing time the mixing quality is inversely proportional to the Re and Sc
is highly non-linear, since the creation of iodine is only accom- number. This is, strictly speaking, only valid for straight laminar
plished by a local oversupply of H+ -ions, for example, two fluids flow. With convective effects, the mixing quality is increased and
from inlets 1 and 2 flowing parallel in a long straight mixing almost proportional to the Re number, since the arising fluid
channel. As the mixing is determined by lateral diffusion, there filaments decrease the time for diffusion [19]. This is displayed
would be no creation of iodine, independent of the diffusion in Fig. 14.
length, since nearly all H+ -ions would immediately be consumed A chemical reaction within the mixing channel brings another
by H2 BO3 − . In the case of the investigated micromixers, the cre- time scale into the process, which depends on the reaction kinet-
ation of iodine is only accomplished in the fluid filaments with ics. The time variation of a species concentration is described
different thickness, which leads to a local oversupply of H+ . by the reaction rate:
On the other hand, the segregation indices measured here can-
dci
not be compared directly with values measured by other groups = νi r = νi kcim . (14)
using different mixers, since there is a strong dependency of the dt
H+ -concentration from the pH value; commonly, H2 SO4 is used For a first order reaction (m = 1) with a constant volume the
as H+ -ion donor. Depending on the pH value, sulphuric acid half-life time is calculated by:
is capable of providing two H+ -ions, so the initial concentra-
ci 1
tion of H+ -ions depends on the concentration of sulphuric acid. tR = = . (15)
Even worse is the fact that the concentration of H+ -ions changes νi r k
dynamically during the reaction due to the change of the pH The characteristic time of the reaction is inversely proportional
value, which makes it very difficult to determine the “correct” to the kinetic constant k. The ratio of the characteristic pro-
initial concentrations for the H+ -ions. Here, the concentration cess times to the reaction time are formulated in the so-called
[H+ ]0 is obtained from pH-measurements. Generally, there is a Damköhler numbers:
need for more research concerning this fact.
tP lm k
DaI = = ; (16)
5. Discussion tR ū
tD b2 k
5.1. Characteristic times for microreactors DaII = = . (17)
tR Dax
The mean residence time of the fluid in the mixing channel For high DaI numbers (DaI  1) the components have enough
is given by the channel length divided by the mean velocity: time to react in the mixing channel and will perform a com-
plete reaction. This requires a fast reaction, long mixing chan-
lm
tP = . (10) nels or a very slow mean velocity. The above mentioned
ū Villermaux–Dushman reaction consist of a very fast neutral-
The mass transfer in radial direction, perpendicular to the ization and iodide formation, which has a reaction rate of about
flow direction, is limited by diffusion in laminar flow. The 7.29 × 107 m−4 s−1 . Depending on the inlet concentration of the
axial diffusion length of a molecule can be calculated by the components, the typical reaction time of the second reaction in
Einstein–Smoluchowski equation: the actual setup was about 40 ms. With a mixing channel length
in the T-mixer of about 16 mm, the mean velocity should not be
x2 = 2Dax t, (11)
higher than 0.4 m/s. For low DaII numbers, the components are
where Dax is the diffusion coefficient of the binary mixture in mixed very quickly, which leads to a complete reaction within
axial direction, also called axial dispersion. The mixing is fin- small microchannels. During convective mixing, the axial dis-
ished, when all molecules are homogeneously distributed over persion is determined by the distance between fluid lamellae
the entire cross-section, which denotes a mean diffusion length with different concentration instead of the channel width. For a
of half the channel width. The time for this diffusion process is reaction time of 40 ms, the fluid element distances have to be
calculated by smaller than 6.3 ␮m. This value has been confirmed by exper-
imental measurements [24]. A miniaturization of the mixing
(b/2)2 b2 channel leads to a very fast mixing by diffusion (DaII ∝ b2 ),
tD = ∝ . (12)
2Dax Dax which can be assisted by convective effects as shown above.
The ratio of the diffusion time to the residence time indicates
the progress of diffusion at the end of the mixing channel: 5.2. Performance evaluation of mixing devices

tD b2 ū b2 There is no simple technique to compare the performance


= = ReScax . (13)
tP Dax lm lm dh of different micromixers, but there are many ways to mix and
506 N. Kockmann et al. / Sensors and Actuators B 117 (2006) 495–508

homogenize two fluids in different devices. It is necessary to


bring a certain effort into the device (such as pressure loss)
to achieve a desired effect (fast mixing within a short range).
To characterize the achievement of the effect, the ratio of the
performance to the effort gives a reasonable value to compare
different devices. For this reason, two identification numbers
for micromixers, the mixing effectiveness MEI and MEII , are
introduced, which serve as a simple tool to identify the quality
of any arbitrary flow mixing process or the performance of any
arbitrary flow mixing apparatus.

5.2.1. The mixing effectiveness MEI


The mixing effectiveness MEI is constituted by the perfor-
mance of the mixing process divided by the effort that has to been
put into it. The performance of the mixing process is determined
by the ratio of the hydraulic diameter to the mixing length dh /lm . Fig. 15. Mixing effectiveness of various mixing principles.
The higher this number is, the better the performance of the
mixer, since a high ratio indicates a short mixing length (com-
pared to the hydraulic diameter) and therefore a good mixing utilizes the energy dissipation ε = pū/ρlm the mixing process,
performance. With the mixing time tm , this number can further determined from the pressure loss related to the mixing length.
be written as: With the energy dissipation, Eq. (22) gives:
dh dh ν
= . (18) MEI = Re 2
. (23)
lm ūtm εtm

The effort of the mixing process can be characterized by the The ratio of the kinematic viscosity to the energy dissipation is
Euler number, which is a measure of the pressure loss in relation a time scale measure in fluid dynamics which is often also con-
to the flow velocity. When doing so, the mixing effectiveness nected with the Kolmogorov time scale tE of convective mixing
calculates to in isotropic turbulence [29]. Introducing the time scale and rear-
ranging gives the following form for the mixing effectiveness:
dh 1 dh ρū2 
MEI = = . (19)
lm Eu ūtm p tE MEI
= . (24)
The definition can also be regarded as the ratio of the gain of tm Re
the mixing process, a short mixing length, and the effort of the The square root of the ratio of the mixing effectiveness MEI to the
process, the pressure loss, which have to be overcome. Within Re number gives a measure or a comparison of two mixing times.
a nearly square cross-section, the mixing effectiveness can be With this comparison, different mixing processes can be bench-
written as: marked to a convective-mixing time in isotropic turbulence (see
ṁ Fig. 16). The nearly horizontal curves of the 3D-T-mixer and
MEI = . (20) the T-shaped mixer indicate a nearly constant energy conversion
dh ptm
to mixing effectiveness. The increasing curve of the tangential
In this form, the mixing effectiveness depends only on macro-
scopic values of the mixing process. The mixing effective-
ness inside various micromixers from this work is displayed
in Fig. 15.
To illustrate the meaning of the mixing effectiveness MEI for
various cases, Eq. (20) can be written in the following form:
ṁ η
MEI = = Re . (21)
dh ptm ptm
If a mixing device can be described simply as pressure loss by
convective effects and using Eq. (7), the mixing effectiveness
can be written as:
2 dh
MEI = (22)
ςvortex lm lm
Using the friction coefficient and the mixing length, the mixing
effectiveness can be determined and compared with other mixing
mechanisms. A second way to treat the mixing effectiveness Fig. 16. Comparison of the time scales in micromixers.
N. Kockmann et al. / Sensors and Actuators B 117 (2006) 495–508 507

mixer results from the short mixing time due to an efficient tangential mixer is approx. one-fifth of the asymmetric T-mixer.
energy conversion. The interdigital mixer has a value of 0.55, which is not surpris-
ing as it was one of the first micromixers ever investigated. The
5.2.2. The mixer effectiveness MEII SuperFocus Mixer from IMM also has very good value of 3000.
For the effectiveness of the mixing apparatus, the mixing Overall, the mixers designed in this study seem to have a very
effectiveness MEI is extended by means of the mass flow rate. high performance, even compared to commercially available
The reason for this is that a mixing apparatus needs not only to micromixers. The mixing effectiveness introduced here makes
use a proper mixing process, but also to have a high throughput. it possible to compare different micromixers with each other.
Therefore, MEI is extended by the Reynolds number: Furthermore, it would be interesting to compare the values cal-
culated here with theses of micromixers already implemented
dh ρū2 ūdh dh2 ṁ2
MEI Re = = (25) by other groups.
ūtm p ν ηA2M ptm
6. Conclusion
Allowing that AM is equivalent to dh2 in a good approximation,
the number is further simplified and gives the mixer effectiveness This paper presents the mixing characteristics of single con-
MEII , as first introduced by Engler et al. [23]: tacting elements such as the T-shaped micromixer. Based on
ṁ2 these results the design, fabrication, and mixing characteristics
MEII := . (26) of different mixer configurations are given for a high throughput
ηdh2 ptm
of aqueous solutions. Relatively large channels with character-
This approach consists only of primary mixer attributes, which istic dimensions from 300 to 600 ␮m and a parallel arrangement
are the mass flow rate, the pressure loss and the mixing time, of 16 mixer elements allow a high mass flow rate up to 20 kg/h
plus the geometry of the channels and the viscosity of the fluid with relatively low pressure losses of about 1 bar on one sili-
used. It is dimensionless and therefore invariant with respect to con chip with the dimensions of 20 mm × 20 mm × 1.6 mm. The
the scale. Furthermore, the characteristic performance data is convective flow structures inside the micromixers provide a suc-
implemented in a proper manner, since a larger MEII number is cessful way to enhance the mixing performance. The flow and
achieved by mixing characteristics have been investigated with the help of
two chemical reactions, a pH-indication with bromothymol blue
• high mass flow rate, for flow characteristics and mixing time, and the competitive-
• low pressure loss, parallel iodide–iodate reaction (Villermaux–Dushman) for the
• short mixing time. mixing quality. For moderate Re numbers between 400 and 800,
mixing times below 1 ms have been achieved. The mixing qual-
Therefore, a high MEII number indicates a high performance ity, or the segregation index, is similar to prior investigation with
mixing apparatus. As a first evaluation, the mixer effectiveness a T-shaped micromixer and better when compared to stirred ves-
MEII is applied to two mixers presented in this paper and to other sels. With this work, the use of micromixers as microreactors for
commercially available micromixers, namely the well-known the production of chemicals can be considered. From the various
interdigital mixer SSIMM, the CaterPillar Split-and-Recombine geometries investigated and characterized in this work, design
Mixer CPMM-R1200-V1.1, and the SuperFocus Mixer, all from criteria can be derived for the implementation of other mixing
the Institut Mikrotechnik Mainz IMM. The values for these mix- tasks in microsystems engineering. For this purpose, two new
ers have been taken from IMM 2003 and IMM 2004 [30,31]. identification numbers, the mixing and the mixer effectiveness
Results are shown in Fig. 17. The 3D-T-mixer is the most effi- MEI and MEII have been introduced, which are calculated by
cient one with a MEII value of about 12,000. The value for the primary mixer attributes and serve as a tool for the comparison
of different mixing principles and devices.

Acknowledgement

We gratefully acknowledge the DFG, “Deutsche Forschungs-


gemeinschaft” for their financial support in the priority program
1141 “Strömungsmischer”.

References

[1] N. Nguyen, Z. Wu, J. Micromech. Microeng. 15 (2005) R1–R16.


[2] V. Hessel, S. Hardt, H. Löwe, Chemical Micro Process Engineering,
Wiley-VCH, Weinheim, 2004.
[3] T. Bayer, M. Matlosz, J. Jenck, Chem. Ing. Technol. 76 (2004) 528–533.
[4] V. Hessel, H. Löwe, Chem. Ing. Technol. 74 (2002) 17–30.
[5] P. Woias, K. Hauser, E. Yacoub-George, B. Hillerich, A silicon-based
microreaction system for analytical applications, in: W. Ehrfeld (Ed.),
Fig. 17. Values of the mixer effectiveness MEII for five different micromixers. Proceedings of the IMRET3, Springer, Berlin, 2000.
508 N. Kockmann et al. / Sensors and Actuators B 117 (2006) 495–508

[6] V. Hessel, H. Löwe, F. Schönfeld, Chem. Eng. Sci. 60 (2005) [31] IMM Institut für Mikrotechnik Mainz GmbH, Hochdurchsatz bis in den
2479–2501. m3/h-Bereich, IMM Institut für Mikrotechnik Mainz GmbH, Mainz,
[7] A.D. Stroock, S.K.W. Dertinger, A. Ajdari, G.M. Whitesides, Science Germany, 2004.
295 (2002) 647–651.
[8] J.R. Bourne, Org. Proc. Res. Dev. 7 (2003) 471. Biographies
[9] D. Gobby, P. Angeli, A. Gavriilidis, J. Micromech. Microeng. 11 (2001)
126–132.
[10] G. Deerberg, J. Grän-Heedfeld, T. Hennig, E. Weidner, Chem. Ing. Tech- Norbert Kockmann is research assistant in the Lab-
nol. 77 (2005) 1501–1511. oratory “Design of Microsystems”, Department of
[11] N. Kockmann, C. Föll, P. Woias, Flow regimes and mass transfer char- Microsystems Engineering (IMTEK) at the University
acteristics in static micro mixers, SPIE Photonics West, Micromachining of Freiburg, Germany, where his research activities
and Microfabrication [4982-38], San Jose, USA, 2003. include micro process engineering and heat and mass
[12] M. Engler, N. Kockmann, T. Kiefer, P. Woias, Chem. Eng. J. 101 (2004) transfer in micro systems. He received his diploma
315–322. degree from the Technical University of Munich
[13] M. Engler, N. Kockmann, T. Kiefer, P. Woias, Convective mixing and (Aeronautical Engineering) in 1991 and finished his
its application to micro reactors, in: Proceedings of the 2nd Interna- Ph.D. in 1996 at the University of Bremen, Chair for
tional Conference on Micro and Minichannels [ICMM-2412], ASME, Technical Thermodynamics, Heat and Mass Transfer.
Rochester, NY, USA, 2004, 781–788. After almost 5 years industrial experience as a project
[14] S.H. Wong, M.C.L. Ward, C.W. Wharton, Sens. Act. B 100 (2004) manager for design and construction of air separation units and chemical
359–379. plants, he joined the University in Freiburg in 2001 and works on several
[15] P. Guichardon, L. Falk, Chem. Eng. Sci. 55 (2000) 4233–4243. research projects on heat and mass transfer, mixing and chemical reactions
[16] P. Guichardon, L. Falk, J. Villermaux, Chem. Eng. Sci. 55 (2000) in micro channels.
4245–4253.
[17] G. Trippa, R.J.J. Jachuk, Characterization of mixing efficiency in narrow Thomas Kiefer is a scientific assistant in the Lab-
channels by using the iodide–iodate reaction system, 2nd Int. Conf. oratory “Design of Microsystems”, Department of
Micro and Minichannels, [ICMM2004-2413], ASME, Rochester, USA, Microsystems Engineering (IMTEK) at the Univer-
2004, 789–793. sity of Freiburg, Germany. From 1999 to 2005 he
[18] W.R. Dean, Note on the motion of a fluid in a curved pipe, Philos. Mag. studied Micro Systems Technology at the University
4 (1927) 208–223. of Freiburg and received his diploma degree with a
[19] N. Kockmann, M. Engler, D. Haller, P. Woias, Heat Trans. Eng. 26 focus on the design and fabrication of highly effi-
(2005) 71–78. cient micro mixers. He is currently Ph.D. student at
[20] N. Kockmann, M. Engler, C. Föll, P. Woias, Liquid mixing in static the EPF Lausanne.
micro mixers with various cross-sections, in: S.G. Kandlikar (Ed.), Pro-
ceedings of the 1st International Conference on Micro- and Minichannels
Michael Engler is Ph.D. student at the Laboratory
ASME, Rochester, 2003 (ICMM2003-1121).
“Design of Microsystems”, Department of Microsys-
[21] F. Schönfeld, S. Hardt, AIChE J. 50 (2004) 771–778.
tems Engineering (IMTEK) at the University of
[22] M. Kraume (Ed.), Mischen und Rühren — Grundlagen und moderne
Freiburg, Germany, since 2002. He received his
Verfahren, Wiley-VCH, Weinheim, 2003.
diploma degree in Micro System Engineering and his
[23] M. Engler, T. Kiefer, N. Kockmann, P. Woias effective mixing by the
Bacc.-Math. from the Mathematical Institute of the
use of convective micro mixers, in: Proceedings of the AIChE Spring
University of Freiburg. Currently he works on the
Meeting, IMRET8, April 2005 (Atlanta, USA, plenary talk, TK001-
simulation, design, fabrication, and characterization
128d).
of micro mixers and micro reactors with complex
[24] M. Hoffmann, M. Schlüter, N. Räbiger, Experimental investigation of
chemical reactions. His main research interests are
mixing in rectangular cross-section micromixers, in: Proceedings of the
the design and characterization of highly efficient
AIChE Annual Meeting, Austin, TX, paper 330b, 2004.
micro mixers.
[25] S. Panić, S. Löbbecke, T. Türcke, J. Antes, D. Bošković, Chem. Eng.
J. 101 (2004) 409–419.
[26] M.C. Fournier, L. Falk, J. Villermaux, Chem. Eng. Sci. 51 (1996) Peter Woias is full professor at the Laboratory
5053–5064. “Design of Microsystems”, Department of Microsys-
[27] T. Kiefer, Design and fabrication of highly efficient passive micro mix- tems Engineering (IMTEK) at the University of
ers, Diploma Thesis University of Freiburg, IMTEK, Laboratory for Freiburg, Germany. After studying electrical engi-
Design of Microsystems, Germany, 2005. neering from 1982 to 1988, he received his Ph.D. in
[28] E. Truckenbrodt, Fluidmechanik, Springer, Heidelberg, 1996. 1995 from the Technical University in Munich, Insti-
[29] M. Zlokarnik, Ruhrtechnik Theorie und Praxis, Springer, Berlin, 1999; tute of Integrated Circuits. Later, he worked at the
M. Zlokarnik, English translation: Stirring Theory and Practice, Wiley- Fraunhofer Research Institute in Munich as a group
VCH, Weinheim, 2001. leader on microfluidics. In 2000, he applied as full
[30] IMM Institut für Mikrotechnik Mainz GmbH, Process Technology of professor in Freiburg, where he enlarged his research
Tomorrow — The Catalogue, IMM Institut für Mikrotechnik Mainz interest to micro rapid prototyping, mechanical actu-
GmbH, Mainz, Germany, 2003. ators, CAD and simulation in micro system design.

You might also like