This project is the direct continuation of the MINAST-project
"micropipetting device with integrated sensors". The motivation is to
supplement a pipetting device for medical analysis with a possibility for
contact free release of drops in the nanoliter range. Additionaly a new method
of liquid level detection (LLD) will be developed as well
The contact free release of drops shall be achieved by using the wave
propagating properties of a pipetting needle. Simulations of the wave
propagation are performed to obtain this aim. The actuator that is used to for
LLD will be developed. This new method of LLD, which is based on resonant
vibrations, must also work reliably in present problem cases, where the fluid
surface is covered with foam and the fluid container is closed with a cap.
My research interests focus on microsystems (or: Micro-Electromechanical
Systems, MEMS). Recently such microsystems have seen widespread applications in
automobile industry and medicine. Examples include accelaration sensors for
airbags and drug delivery systems.
During a two years stay at the MEMS Lab (Prof. Böhringer) of the University of
Washington, Seattle, I participated in a project on BioMEMS. The goal was to
design intracellular neuronal recording systems. These devices allow for
implantation of sensors into live, freely behaving animals. First results were
presented at the Transducers 01 in Munich .
Another field of interest are polymers in microsystems technology. In the
last couple of years huge progress has been made towards integrated circuits
made out of conducting and semiconducting polymers. This approach offers the
advantage of cheap roll-to-roll fabrication
and therefore cheap electronics on flexible and even transparent substrates.
Potential applications include RF tags and electronic papers (by using OLEDs).
Our idea is to combine plastic electronics and mechanical parts made out of
polymers to a new kind of MEMS devices.
: Y. Hanein, U. Lang, J. Theobald, R. Wyeth, T. Daniel, A.O.D. Willows,
D.D. Denton, K. Böhringer, Intralcellular Neuronal Recording with High Aspect
Ratio MEMS Probes, in:
Proceedings of the 11th International Conference on Solid-State Sensors and
Actuators, 2001, p.386
with Double Clamped Si Beam as Vibrating Mass
Author: Stefan Blunier
Research group: Jürg Dual
For online screening of the position of a moving object, satellite based
systems (GPS) in combination with "on board" navigation is mandatory.
GPS allows position detection in space while on board sensors have to detect the
orientation of the object. Traditionally rotations are measured with
gyro-compasses. Because of their big size and high energy consumption, a strong
interest in miniaturized inertial sensors exists. The smallest are built in MEMS
(Micro Electro Mechanical Systems) technology and are based on the influence of
Coriolis force on a vibrating system. The mechanical part is built as
microstructure in single crystalline silicon and the electronics can be
integrated in CMOS technology on the same silicon chip.
In our project we make use of a special patented electronic circuit
(PCT/EP95/00761) to overcome the problem of electrical cross talk in
microstructures. The electronics allows to use the same transducers as actuators
The intention of the project is to realize a low cost gyroscope which allows
insertion into mass products. Therefore the design and the manufacturing process
are kept as simple as possible. The mechanical structure is built in SOI
(Silicon On Insulator) technology. The use of three photolithographies only and
an anodic bonding step keeps manufacturing costs low.
Laser-based Ultrasonics at the Nanometer Scale
Authors: Dieter M.
Profunser, Jacqueline Vollmann
Research group: Jürg Dual
The goal of this project is the characterization of microstructures, MEMS
devices, and thin films geometrically and mechanically in a rapid, non-contact,
and non-destructive manner at high lateral resolution. Furthermore the
quantitative measurement of structural properties, like residual stresses,
bonding, and discontinuities of the acoustic impedance lie within the scope of
The smaller the structures to be investigated are, the higher is the demand
for very short acoustic wavelength in order to achieve reasonable resolution.
Short Laser pulses represent an excellent - not to say the only - way to excite
and detect mechanical waves of several 10 nm wave length thus providing access
to 'small scale ultrasonics': Nanosonics. The laser pulse width amounts to 70
The basic principle of the technique consists of a thermoelastically induced
acoustic pulse, which propagates perpendicular to the surface of a planar
structure until it encounters a discontinuity of the acoustic impedance causing
the pulse to be partly reflected and partly transmitted (see Figure 1).
Since the optical reflectivity coefficient is strain dependent, the reflected
and/or the transmitted acoustic pulse cause a slight change of the optical
reflectivity at the surface of the specimen, which is detected optically. Thus
the time of flight of an acoustic pulse can be measured and once the speed of
sound is known the thickness of the film can be determined. For metallic layers
the achievable resolution amounts to few nanometers.
The technique requires a precise handling of the time shift between the
excitational pulse and the measuring pulse. Therefore the pulsed laser beam is
split into a pump beam and a weaker probe beam. Both beams follow different
paths and the pump beam path has a variable length, thus the time shift between
two corresponding pulses is controlled by the variable optical path and the
optical reflectivity at the surface is scanned relative to the time of the
arrival of the pump pulse.
Figure 2 shows the experimental set-up and the measurements of the optical
reflectivity of a three aluminum films on a sapphire substrate having three
different thicknesses. A typical reflectivity curve is dominantly governed by
the thermal reflectivity change with the superimposed effects of the arriving
echoes at the surface.
Beyond the application in the field of Nondestructive Evaluation several
engineering applications involving wave propagation phenomena in the order of 10
nm wavelength are currently developed.
More Information: http://www.zfm.ethz.ch/e/nanosonic/
Properties of Microstructures
Authors: Jürg Dual, Edoardo
Mazza, Gilbert Schiltges, Dirk Schlums
(from the 1998 brochure)
The purpose of the present studies is to determine the mechanical properties
of the materials used for the fabrication of micrometer-sized structures. The
knowledge of design parameters, such as strength and elastic constants, allows
to define the critical loading conditions for microstructures. Moreover, the
variation of the mechanical behaviour with respect to macroscopic structures as
well as the influence of manufacturing processes and conditions is investigated.
New experimental techniques have been developed, and conventional models
based on continuum mechanics have been extended for the evaluation of the test
results. The same material parameters are measured using different experimental
techniques. The good agreement between these different techniques ensures their
reliability as well as the reliability of the evaluation models.
Samples made of low doped single crystal silicon (SCSi), nickel or
nickel-iron alloys have been investigated. In figure 1 a SCSi and a metallic
sample are shown. They consist of two plates (approximate dimensions in the
plane 5x5 mm2) connected by a microbridge, which is the actual testing region.
In the specimen design several requirements had to be satisfied: feasibility of
the microfabrication, avoidance of microbar breakage before testing, possibility
of specimen clamping into the testing machine, detection of internal stresses
due to the fabrication process.
The silicon samples have been fabricated using photolithography and wet
etching in the clean room facilities of the Institute of Mechanics. The testing
region has approximately the following dimensions: 50x50x300 mm3.
The metallic samples were produced in a cooperation with the IMM (Institute
of Microtechnology, Mainz, Germany). With the LIGA (for Lithographie, Galvanik,
Abformung) technique it has been possible to fabricate metallic specimens with
the same design as for the silicon samples. Different sizes of testing regions
have been realized, with the following dimensions: width 20...40 mm, thickness
100...200 mm, length 300 to 1000 mm.
The samples were analysed with the light optical microscope and with SEM to
detect possible defects and with a scanning optical system (UBM) to precisely
define the geometry. Microstructures for internal stress detection are present
in the sample. Their deformation is not detectable with optical microscope
measurements. Therefore, the internal stresses in the specimens tested are
considered to be negligible.
A special procedure with appropriate devices has been created in order to
handle the samples during the test preparation, avoiding damages of the testing
region. In particular, the protective frame which connects the two plates must
be removed before performing the experiments. This is carried out with the so
Figure 1: Silicon sample (left) and LIGA sample (right)
As has been shown in torsional tests, the metallic specimens may show an
anisotropic behaviour. To study this behaviour in further detail, nondestructive
resonance experiments were carried out, during which the resonance frequencies
of the samples as well as the corresponding mode shape were determined (see
Figure 1 ).
Figure 1: Resonance modes with dominating elastic
constants experiment and simulation
To extract the intrinsic properties of the material, i.e. elastic moduli,
from this data a numerical simulation has to be carried out. As can easily be
imagined, it is of utmost importance to know the geometry of the samples. An
error calculus shows that errors in dimensions have the biggest influence on the
determined moduli. Mixed numerical, experimental techniques (MNET) which involve
an iterative process allow the determination of four out of five elastic
constants, if the material is assumed to be transversely isotropic. The working
principle of MNET is shown in Figure 2 .
Figure 2: Mixed numerical experimental techniques to
determine optimal parameters from resonance frequencies mi
When both the elongation and the force are measured in the tensile test,
Young's modulus as well as yield stress and strain, and strength can be
determined. A microsample tensile test apparatus has been developed, which is
schematically represented in Figure 1 .
The two plates are pulled apart so that the microbar is deformed in the axial
direction until failure. The deformation is induced using a piezoceramical
element (maximal displacement 30 mm). In order to realize large elongations (as
occurring while testing metallic samples, due to the plastic deformation) a
micrometer screw is implemented. The microbar axis is parallel to the
gravitational field in the tensile test set-up. Therefore, the force can be
measured with a very precise balance (resolution 10-5 N, maximum force measured:
Figure 1: the microsample tensile test apparatus
During the experiment, the sample is observed with a microscope (Olympus SZH,
Zoom Stereo Microscope, numerical aperture = 0.162, working distance ª10 cm,
typical magnification: 128X). With a CCD camera, digital images of the testing
region in the initial state (before the pulling process begins) and in the
deformed state (during the tensile process) are recorded. By comparing these
images the bar elongation at every stage of the tensile process can be measured.
For this purpose the least square template matching (LSM) algorithm is used. In
spite of the resolution limit of the optical microscope, the applied algorithm
can resolve elongations down to 10 nm.
Figure 2: Stress-strain curves of a SCSi sample
In particular, no artificial target points on the sample are required and the
results are not affected by any rigid body motion of the sample (which are
separately estimated). This method has been tested to verify its reliability and
Figure 3: Stress-strain curves of a LIGA nickel sample
In Figure 2 and Figure 3 stress-strain curves for a SCSi and for a LIGA
sample are shown. These are the so-called engineering curves, in which the
stress and strain values are calculated from the original dimensions of the
Elastic deformation and plastic deformation are present in the stress-strain
curve of the metallic samples. In the following table, Young's modulus, yield
stress, and strength of the LIGA samples are reported (the values correspond to
the average of the results of four tests with each material).
The values of Young's modulus from the vibration tests are in excellent
agreement with those from the tensile test (differences of 1.5% and 3.5% for
nickel and nickel-iron respectively). This is a very good confirmation of the
presented testing methods.
The values for yield stress and for the strength are very high. For
metallurgical pure nickel ("macroscopic"), the two values are found in
handbooks to be 110 MPa and 345 MPa respectively. In the tests presented here,
yield stress and strength are 3.7 and 2.3 times larger than these values,
respectively. The nickel-iron alloy shows a strength which is comparable to that
of the best construction steels (stainless, "maraging" steels) for
macroscopic structures (strength ª2 GPa).
The SCSi samples behave elastically until failure. In the following table the
average values of maximum force measured in the tensile test are reported. The
stress is evaluated in the middle of the testing region.
Maximum force (N)
Maximum stress (GPa)
The standard deviation is very low. Considering the uncertainties in the
specimen dimensions (2.2%), the standard deviation due to other errors appears
almost negligible. This result was expected, since the tested material has a
very regular lattice structure.
The maximum stress reported in the above table is not the failure stress for
single crystal silicon: It is the ultimate axial stress in the middle of the
testing region, whereas the maximum stress in the sample is located at the tip
of the sharp notches, which are present at the extremities of the microbar (see
Figure 4 ). In all experiments, failure originates at one corner and the crack
Figure 4: Fractured sample after the tensile test. The
starting point of the crack and the (111) fracture surfaces can be seen.
This type of notch is very common in silicon microstructures and it is due to
the anisotropic wet etching process. Therefore, the analysis of the stress
concentration at the notch (which is omitted in most previous publications) is
The radius of curvature at the corner has an order of magnitude of 1 nm.
Local plasticity effects in single crystal silicon can be excluded. Thus, the
calculation of the stress field around such a corner is a particular problem of
fracture mechanics. The Stroh formalism has been applied for the solution of the
anisotropic elasticity problem. Thus, the stress and strain fields at the notch
tip (near field solution) has been determined. With the values of critical
tensile load of several experiments (force=2.47 N 3%) and a suitable FEM model,
the critical stress intensity factor for the present notch was calculated.
The critical stress intensity factor can be interpreted as a failure
criterion: failure occurs when the stress intensity factor exceeds the critical
value. However, this design criterion suffers from severe limitations. In fact,
it is only valid for a specific wedge angle and for a specific loading mode. In
this work a new failure criterion based on energy considerations is presented.
This new criterion represents an attempt to overcome the limitations of the
critical stress intensity factor design rule.
The strain energy in a surface layer at the notch is calculated for critical
loading conditions. The fracture surfaces observed in the experiments correspond
to (111) crystal planes. The comparison of the total strain energy with the
energy required to form 2 new (111) crystal planes leads to the following
empirical design rule: failure occurs if the amount of strain energy in a
circular region of radius r=0.8 nm with centre at the notch tip exceeds the
quantity of energy which is required to form 2 (111) crystal planes in the same
The proposed design criterion has the advantage that it can be applied for
every notch (with arbitrary wedge angle) in SCSi and that it is not dependent on
the loading case. The proposed design rule can be applied for the microgripper
(see See Tactile Microgripper with Eye. ) and therefore leads to the critical
values for the emergency stop of the Nanorobot.
As in the macroscopic world, several kinds of tests have to be performed to
be able to establish reliable failure criteria. An important experiment in this
context is the torsional test, as it allows to validate or expand the criteria
gained from tensile tests.
The experimental setup is a major challenge. The actuator part has to be able
to apply pure torsion by minimizing errors due to bending. The sensor part is
able to measure the resulting angle and torque, with a resolution of 0.03o and
0.3 mNm, respectively. A balance records the tensile forces occurring in the
specimens during the experiments. Experiments have been performed on both
silicon and metallic specimens. The resulting curves are shown in Figure 1 .
Figure 1: Torque/rotation diagrams for silicon and Ni
A numerical simulation (see Figure 2 ), using finite element techniques, of
the experiments in combination with an analytical analysis allows the
determination of the governing elastic moduli, which in this case consist of two
shear-moduli, as the material behaviour is considered to be transversely
Figure 2: FE-model of LIGA specimen
In addition to the determination of elastic moduli, it is important to know
the relevant failure criterion in order to have engineering design rules. For Ni
specimens, which show an isotropic behaviour, it is possible to show that the
von Mises yield criterion is in good agreement with the experiments. The
anisotropic Ni-Fe alloys do not show the same behaviour. Their yielding point in
torsion lies beyond the point predicted by von Mises, using the data from the
tensile test and assuming an isotropic yield criterion.
Micromechanical structures often reveal defects or stress concentrations
which can cause cracks to nucleate and propagate. Very small crack growth rates
will lead to fatigue failure of dynamically loaded small-scale structures, thus
a high sensitivity in crack depth measurements is necessary to monitor the crack
Fatigue crack growth measurements are performed by continuously determining
the resonant frequency in vibrating microstructures, which is stabilized through
a phase-locked feedback control loop. The precisely controlled resonant
frequency is related to the crack depth by a nonlinear model based on fracture
mechanics, hence crack growth can be monitored with respect to time with very
high accuracy (see also chapter Impact and Fracture Mechanics).
Figure 1: Dimensions of LIGA specimens used for fatigue
The nickel-iron alloy samples (dimensions according to Figure 1 ) for the
fatigue tests were manufactured with the LIGA-technique. Elastic parameters such
as Young's modulus, the elastic limit, and shear moduli were previously
determined with both, vibration tests and tensile tests.
Figure 2: Resonant frequency of the unnotched LIGA
specimen (NiFe, 13% Fe) during a fatigue test
The structure is excited to bending vibrations utilizing a piezoelectric
transducer. Considering the high stability in frequency the resolution of crack
depth measurements is in the range of 10 nm. An example of the measured resonant
frequency with respect to time for a LIGA-sample without a notch is given in
Figure 2 . The crack nucleated and propagated in a cross-section about 20 mm
below the transition from the upper plate into the microbeam (as indicated in
Figure 1 ), i.e. the crack started in a region where the bending moment was
expected to be largest
Figure 3: SEM photograph of the fracture surface of the
notched specimen (500 x)
The SEM photograph of the fracture surface of the notched microbeam is shown
in Figure . The dark rectangle represents the smooth side wall of the notch
whereas the remaining portion of the crack surface can be divided into a region
of fatigue crack growth and final rupture. The four horizontal lines within the
region of fatigue crack growth mark off regions of different plane heights.
The fatigue testing technique and crack model originally developed for
slender beams could easily be applied to microstructures after suitable
transducers and sensors were selected.
03/28/08 | Stephan Kaufmann |
ZfM | ETH