Beruflich Dokumente
Kultur Dokumente
Doctoral Thesis
Author(s):
Suter, Marcel
Publication Date:
2012
Permanent Link:
https://doi.org/10.3929/ethz-a-7355930
Rights / License:
In Copyright - Non-Commercial Use Permitted
This page was generated automatically upon download from the ETH Zurich Research Collection. For more
information please consult the Terms of use.
ETH Library
Diss. ETH No. 20104
Photopatternable superparamagnetic
nanocomposite for the fabrication of
microstructures
A dissertation submitted to
ETH ZURICH
2011
ii
Document typeset by the author using the LATEX 2ε system and the KOMA-Script
document class scrbook.
Copyright
c 2012 Marcel Suter, Zürich
Viel Grosses wurde schon entdeckt,
nun wird es Zeit an das Kleine zu denken!
Walter Suter
to my family
iv
Abstract
Today polymer materials are present everywhere. Often these materials are not
pristine polymers; rather, they contain chemical additives or are doped by addi-
tional filler materials for reinforcement, increasing wear resistance, reducing in-
flammability, or adding different material properties. Magnetic polymer compos-
ites combine the advantages of the magnetic characteristics of the filler material
and the properties of the polymer matrix, such as low cost fabrication processes.
This work presents a photodefinable magnetic polymer composite (MPC) ob-
tained by dispersing superparamagnetic magnetite nanoparticles in a photosensit-
ive epoxy, SU-8. The composite is used to fabricate magnetic microstructures that
can be actuated by magnetic fields. The microstructures presented in this work
show the potential of the MPC in different applications such as microresonators
and microrobots.
The first part of this work presents the evaluation of the materials and the
fabrication of the MPC. The material properties of the obtained composite are
characterized in detail depending on the particle concentration 1 – 10 vol.% (4 –
32 wt.%). The dispersion of the nanoparticles (diameters of 11.4 ± 3.4 nm, count
average diameter by TEM) and the level of agglomeration are analyzed by differ-
ent methods such as optical microscopy, transmission electron microscopy (TEM),
and small-angle x-ray scattering (SAXS). The MPC shows a homogeneous nano-
particle distribution with low agglomeration, which is a key requirement for the
fabrication of microstructures with small feature sizes and uniform magnetic and
mechanical properties. In order to obtain a high-quality nanocomposite, differ-
ent dispersing agents were tested. Mechanical, magnetic, wetting, and biocom-
patibility properties of the nanocomposites are characterized. The nanocompos-
ites exhibit superparamagnetic properties. The biocompatibility of the MPC is
demonstrated by proliferation of human dermal fibroblasts, and show the po-
tential of the composite for bioapplications. Furthermore, it is shown that the
MPC with 4 vol.% particle concentration can be remotely heated by alternating
magnetic fields at high frequencies due to the incorporated superparamagnetic
nanoparticles.
The MPCs can be used to fabricate microstructures using conventional photo-
lithography technique or laser writing techniques based on two-photon polymer-
ization (TPP). The influence of particle concentration on composite fabrication
v
Abstract
steps such as spin coating, UV exposure, and TPP are systematically analyzed.
For UV lithography, the polymerized layer thickness of the MPC is limited to
2.9 µm for a 5 vol.% MPC, whereas, for TPP, 4 vol.% MPC microstructure with
heights up to 17 µm can be fabricated. Minimal line widths of 314 nm are ob-
tained with 2 vol.% MPC using the TPP technique.
Microcantilevers with particle concentrations of 0 – 5 vol.% are successfully
fabricated and are used to determine the dynamic Young’s modulus of the com-
posite. An increase of the Young’s modulus with increased particle concentration
from 4.1 GPa, for pure SU-8, up to 5.1 GPa, for 5 vol.%, is observed. Furthermore,
MPC in-plane resonators are fabricated. The resonance characteristics of both
microcantilevers and in-plane resonators are investigated and they are actuated
successfully by magnetic fields. A proof of concept for mass sensing is performed
for cantilevers in air by observing the resonance frequency shift upon placing a
small gold mass on the cantilever tip. Helical MPC microstructures are fabricated
with TPP technique which show corkscrew swimming motion in water when ac-
tuated by rotating homogeneous magnetic fields.
Microstructures fabricated with the developed MPC have higher nanoparticle
concentration and smaller feature sizes compared to reported MPC microstruc-
tures in literature. This work shows the opportunities and challenges of using
MPC in various microsystem applications. MPC cantilevers were used as passive
remote mass sensors with magnetic actuation and optical readout. Furthermore,
MPC microstructures can be used for drug delivery, biomanipulations such as
cell manipulations, self assembly applications, and microturbines for lab-on-a-
chip applications.
vi
Zusammenfassung
Kunststoffe (Polymere) sind heute allgegenwärtig. Die meisten Polymere enthal-
ten chemische Zusatzstoffe oder Füllmaterialien zur mechanischen Verstärkung,
um die Verschleissfestigkeit zu erhöhen, die Entflammbarkeit zu reduzieren oder
Materialeigenschaften zu erweitern. Magnetische Polymerkomposite verbinden
die Eigenschaften von Polymeren, wie vielseitige und günstige Fabrikationspro-
zesse, mit den magnetischen Eigenschaften des Füllmaterials.
In dieser Arbeit wird ein fotodefinierbarer magnetischer Polymerkomposit
(MPK) vorgestellt. Für den Komposit werden superparamagnetische Magneti-
te Nanopartikeln in fotoempfindlichem Epoxy SU-8 dispergiert. Der Komposit
kann zur Herstellung von magnetischen Mikrostrukturen verwendet werden, wel-
che mit magnetischen Feldern aktuiert werden können. Die magnetischen Mi-
krostrukturen, welche in dieser Arbeit präsentiert werden, zeigen das Potential
des entwickelten Komposits für Anwendungen wie Mikroresonatoren und Mi-
kroroboter.
Der erste Teil dieser Arbeit beschreibt die Evaluierung der Materalien und die
Herstellung des Komposits. Die Materialeigenschaften wurden untersucht in Ab-
hängigkeit von dem Nanopartikel-Füllgrad von 1 – 10 vol.% (4 – 32 wt.%). Weiter
wurde die Dispersion von den Nanopartikeln (Durchmesser: 11.4 ± 3.4 nm, Mit-
telwert von TEM-Messungen) und deren Agglomeration mit verschiedenen Me-
thoden geprüft: Mit optischer Mikroskopie, mit Transmissions-Elektronen Mikro-
skopie sowie mit Röntgen-Kleinwinkelstreuung-Messungen. Der MPK zeigt eine
homogene Nanopartikelverteilung mit einer geringen Nanopartikelagglomerati-
on. Dies ist eine wichtige Voraussetzung für die Fertigung von Mikrostrukturen
mit kleinen Strukturabmessungen sowie uniformen magnetischen und mechani-
schen Eigenschaften. Verschiedene Dispergiermittel wurden ausgetested sowie
die magnetischen und mechanischen Eigenschaften, Benetzbarkeit und Biokom-
patibilität des Komposits getestet. Der Komposit besitzt superparamagnetische
Eigenschaften. Die Biokompatiblilität des Komposits für biologische Anwendun-
gen wurde mittels Proliferation von Haut-Fibroblasten untersucht und zeigt ein
nicht toxischen Verhalten gegenüber Zellen. Zusätzlich wurde gezeigt, dass der
Komposit mit einer Partikelkonzentration von 4 vol.% mit einem externen hoch-
frequenten magnetischen Wechselfeld aufgeheizt werden kann.
Mikrostrukturen aus MPK können mittels konventioneller Fotolithografie oder
vii
Zusammenfassung
viii
Acknowledgment
Numerous people have helped in many ways to complete this thesis. First, I
would like to thank Prof. Christofer Hierold for giving me the opportunity to
pursue my PhD thesis in his research group. I enjoyed working in the Micro and
Nanosystems group and would like to thank all group members for the good
team spirit. Also, I would like to thank my co-supervisor Prof. Bradley Nelson
for his support and advice throughout this work.
I’m very grateful to Olgaç Ergeneman who was my collaborator of this project,
for his generous support, conducted measurements, and enjoyable discussions.
In particulary, I am indebted to Prof. Pratsinis (PTL, ETH) for granting access
to the group facilities, and his group members Heiko Schulz, Alexandra Teleki,
Adrian Camenzind, Thomas Rudin for the collaboration and partner projects. I
would like to thank Silvan Schmid for his useful suggestions and tips, aswell Mi-
chael Wendlandt for his fundamental input on the project. I am also grateful to
Salvador Pané who generously spent his time for scientific discussions and guid-
ance. A special thank goes to Christian Bergemann from Chemicell GmbH for
the cooperation for the fabrication of the magnetic paricles.
During my project I had the pleasure to collaborate with Master and Bachelor
students, who I wish to thank for their significant contributions to my thesis.
These are Jonas Zürcher, Li Yunjia, Dominic Kraus, David Grob, Jonas Schön-
dube, Sarah Fried, Silvio Graf, and Philipp Bächtold. Furthermore, I would like
to thank to all other students who helped to evolve this thesis: Patric Eberle for
the electronic circuits, George Chatzipirpiridis for the coil setups and image pre-
parations.
The work presented in this dissertation was realized with the support of many
people. With their unique skills they contributed to many different aspects of this
work. I thank to Elisabeth Müller and Eszter Barthazy (EMEZ) for the TEM im-
ages, Prof. Ann Marie Hirt (EPM, ETH) for the support with the magnetic char-
acterization, Christian Moitzi for the SAXS measurements, Tessa Lühmann for
the biocompatibility measurements, Alberto Sánchez Cebrián for the magnetic
heating experiment and simulations, Philipp Rüst for cantilever damping calcu-
lations, Doris Spori for surface contact angle tests, Prof. Batlogg for the SQUID
facilities, and Loic Jacot-Descombes for the composite inkjet printing.
For the two-photon polymerization project I would specially thank to Li Zhang
ix
Acknowledgment
for his collaboration and taking SEM images, Erdem Siringil for his work assign-
ment to fabricate composite microstructures, and Kathrin Peyer for the flagella
actuation.
Thanks to Christian Peters for his support and ideas, Valentin Döring for the help
with FEM simulations, and Emine Cagin, Stuart Truax, Florian Umbrecht, Nina
Wojtas, Bernau Vianney for fruitful discussions about my work and a construct-
ive feedback. Special thanks goes to Eeva Köpilä for her support.
My parents, my sister and brothers, and friends thank you for being my patient
companions during the last four years.
The financial support for this project by ETH Zürich (TH-28 06-3) and SNSF (pro-
ject no 200020-113350) is gratefully acknowledged.
x
Contents
Abstract v
Zusammenfassung vii
Acknowledgment ix
1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 State-of-the-art of MPCs for microsystem applications . . . . . . . . 2
1.3 Goals of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 Approach and outline of this work . . . . . . . . . . . . . . . . . . . 9
2 Theory 11
2.1 Magnetism in nanoparticles . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.1 Units in magnetism . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.2 Magnetic field and magnetic flux density . . . . . . . . . . . 12
2.1.3 Magnetic materials . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.4 Small particle magnetism . . . . . . . . . . . . . . . . . . . . 13
2.2 Mechanics of oscillating structures . . . . . . . . . . . . . . . . . . . 17
2.2.1 Harmonic oscillator . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.2 Beam theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.3 Quality factor of oscillating structures . . . . . . . . . . . . . 22
2.3 Young’s modulus of composites . . . . . . . . . . . . . . . . . . . . . 22
2.3.1 Hashin-Shtrikman Model . . . . . . . . . . . . . . . . . . . . 23
3 Evaluation of materials 25
3.1 Polymer evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.1.1 Polymer selection criteria . . . . . . . . . . . . . . . . . . . . 25
3.1.2 Polymer selection . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2 Particle evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.3 Dispersant agent evaluation . . . . . . . . . . . . . . . . . . . . . . . 33
3.3.1 Stabilization of particles . . . . . . . . . . . . . . . . . . . . . 33
3.3.2 Selection of suitable surfactant on magnetite particles for
photocurable epoxy matrix . . . . . . . . . . . . . . . . . . . 34
xi
Contents
3.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4 Fabrication 37
4.1 Composite fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.1.1 Evaluation of mixing method for composite . . . . . . . . . 38
4.1.2 Preparation of magnetic suspension . . . . . . . . . . . . . . 39
4.1.3 Composite mixing . . . . . . . . . . . . . . . . . . . . . . . . 39
4.2 Microstructure fabrication . . . . . . . . . . . . . . . . . . . . . . . . 39
4.3 Polymer package fabrication . . . . . . . . . . . . . . . . . . . . . . . 43
4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
xii
Contents
5.5.8 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
5.5.9 Discussion and conclusion . . . . . . . . . . . . . . . . . . . . 87
5.6 MPC heating with alternating magnetic field . . . . . . . . . . . . . 90
5.6.1 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
5.6.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.6.3 Discussion and conclusion . . . . . . . . . . . . . . . . . . . . 92
5.7 Surface properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.7.1 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.7.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . 96
5.7.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.8 Biocompatibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.8.1 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.8.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . 97
5.8.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
5.9 Summary and conclusion . . . . . . . . . . . . . . . . . . . . . . . . 99
xiii
Contents
Bibliography 157
Publications 173
xiv
List of symbols and
abbreviations
List of Symbols
Symbol Unit Description
αc ppm/K Thermal expansion coefficient of a composite
αSU −8 ppm/K Thermal expansion coefficient of a SU-8
α Fe O ppm/K Thermal expansion coefficient of a magnetite
3 4
A m2 Cross sectional area of a beam
AD m2 Damping-related effective area of a system
B T Magnetic flux density
cth J/(kg K) Heat capacity
c Pa s Coefficient of damping
cs m Fixed beam length
χ Susceptibility
D J/cm2 Exposure dose
δ nm Length of the adsorbed molecules
δP m Penetration depth
E J Energy
E(Θ) J Anisotropy Energy
η Pa s Dynamic viscosity of the fluid
F N Force
F0 N Excitation force
f Hz Frequency
f f luid Hz Frequency in fluid
f vac Hz Frequency in vacuum
g General mathematic function
Γ Hydrodynamic interaction function
H A/m Magnetic field
Hc A/m Coercivity
h m Thickness of beam or layer
hL m Height of a written composite/polymer line
xv
Contents
Symbols
xvi
Contents
Symbols
xvii
Contents
Symbols
xviii
Contents
Abbreviations
Symbol Description
ABF Artificial bacteria flagellas
AFM Atomic force microscope
AGM Alternating gradient magnetometer
DI De-ionized
DMA Dynamic mechanical analysis
FEM Finit element method
LSV Laser-Doppler vibrometer
MEMS Micro electromechanical system
MPC Magnetic polymer composite
PDDA Poly(dimethyldiallylammonium chloride)
PDMS Polydimethylsiloxane
PMMA Polymethylmethacrylate
PSS Poly(sodium 4-styrenesulfonate)
RMS Root mean square
SAXS Small angle X-ray scattering
SEM Scanning electron microscopy
SQUID Superconducting quantum interference device
SWNT Single wall carbon nanotube
TEM Transmission electron microscopy
TPP Two-photon polymerization
VSM Vibrating sample magnetometer
Voxel Volume pixel
WLI White ligth interferrometer
XDC X-ray disc centrifuge
3D Three dimensional
xix
1 Introduction
1.1 Motivation
The development of polymers has revolutionized our world, and they have be-
come an integral part of our everyday life. Their low weight, variety of proper-
ties, efficient processing, and low cost makes them attractive for various objects
such as car bodies, furniture, computer casings and ball pens. Pristine polymers;
however, often show insufficient strength, low long-term durability, flammabil-
ity, and low resistance to high temperatures. Around 1900, it was recognized
that nanoparticles mixed with a selective addition to polymers change the char-
acteristics of the polymer. For example carbon black, zinc oxide or magnesium
sulfate particles mixed with vulcanized rubber were used to fabricate wear resist-
ant automobile tires [1]. Later it was discovered that glass fiber combined with a
polymer creates an incredibly strong structure with lightweight properties. The
incorporation of different kinds of fibers, platelets and particles were investig-
ated to improve the properties of the pure polymers. In the mean time, a large
number of composites with tailored inorganic fillers have been developed for all
kinds of applications such as flexible magnetic strips, lightweight carbon fiber re-
inforced polymers for bicycles and airplanes [2]. The composite materials benefit
from both the advantages of the selected polymer and the properties of the filler.
These composite materials also show advantages for structures in small scale.
In the recent years composites materials were developed for the use in micro
electromechanical systems (MEMS)1 . Microsystems based on polymer compos-
ites with tailored material properties are interesting for new sensors, actuators
and other devices [3]. Composites with improved properties for microsystems
have been presented in several works. Photocurable polymers with electrical
conductivity using silver nanoparticles [4] and lower required UV dose with in-
corporated silica particle [5], reduced refractive index for wave guiding applic-
ations using silica nanoparticles in epoxy resin [6] and lower internal stress in
photoresist using silica nanoparticles [7] have been published.
A very promising material combination for a composite is the mixing of poly-
mers with magnetic fillers. Both materials have great advantages for microsys-
tems as described in the following.
1 For simplification „micromechanical systems“ are also included in the term MEMS in this thesis.
1
1 Introduction
2
1.2 State-of-the-art of MPCs for microsystem applications
The relevant magnetic materials used in present-day technology are all inorganic
(e.g. Fe, Co, Mn, Fe3 O4 , γ-Fe2 O3 , SmCo5 ). The most promising method to bring
magnetic behavior into polymers is the incorporation of a magnetic filler material
into the polymer to obtain a magnetic polymer composite.
MPCs in general benefit from the simplicity and the variety of fabrication pro-
cesses of polymers compared to magnetic materials deposited at high vacuum
(sputtering and evaporation) or using time-consuming electrodeposition proces-
ses. Table 1.1 shows an overview of fabrication processes for polymers and MPC.
The fabrication process depends on the choice of the selected polymer/composite
and the kind of microstructure to be fabricated. The filler material often limits the
process parameters. Therefore, the process parameters must be investigated for
each composite material.
MPC for microsystems have been investigated by several groups and MPC mi-
crostructures for a variety of applications have been fabricated. The reported
MPCs can be mainly categorized by the size of the used filler particles (diameter
smaller or bigger than 1 µm), by the fabrication method (photostructurable or
non-photostructurable composites), and by the magnetic behavior (ferro- or su-
perparamagnetic). The terms of magnetism are explained in Chapter 2.
For MPCs containing particle with diameter > 1 µm various applications have
been reported such as microgrippers, micromotors, microinductors, pumps, and
rotational speed microsensors. The details are summarized in Table 1.2. Feature
sizes from 100 µm to 5 mm have been achieved using composites containing mag-
netic particles with diameters > 1 µm.
3
4
1 Introduction
Table 1.1: Overview of microfabrication methods for polymers and polymer composites.
Mould Photodefinable Ablation Bottom-up Direct-shape
Hot embossing [27] Photolithography [28] Laser Electroplating [29] Inkjet printing
Nanoimprint [30] Stereolithography [31] Plasma etching Fiber electrospinning [32]
Casting [33] 2-Photon polymerization [34]
Injection molding [35]
Table 1.2: Summary of state-of-the-art of MPC materials developed for microsystems containing magnetic particles > 1 µm.
The works are sorted by the publication year.
Polymer Magnetic Size of Type Particle con- Fabrication method Structure/Application Reference
matrix particles magnetic of mag- centration
material particles netism
PI-2555 Strontium 1.15 – FM a 80 vol.% Screenprinting or spin- Microactuator Lagorce
ferrite 1.5 µm casting followed by pho- 1997 [33]
(SrFe12 O19 ) tolithography
PI-2555 NiZn, MnZn 0.8 – FM 95 wt.% Screenprinting or spin- Microinductors Park 1998
1.2 µm casting followed by pho- [36]
tolithography
SU-8 Samarium- 10 µm FM 18 – Photolithography Rotational speed mi- Dutoit 1999
cobalt 60 vol.% crosensor [37]
(Sm2 Co17 )
PI-2555 Strontium fer- 1 – 1.4 µm FM 45 – Screen printing and tem- - Rojanapornpun
rite, NiZn 80 vol.% plate printing 2001 [38]
SU-8, Neodym (Nd- 1 – 9 µm FM 35 – 90 wt.% Photolithography, Lorentz force actuat- Feldmann
PDMS FeB), stron- Screen printing and ors, micromotor, mi- 2007 [39]
tium ferrite, replica molding crogripper
samarium-
cobalt
SCR770 Ferrite 1.3 µm FM 50 wt.% Microstereolithography 3D-structures: mi- Kobayashi
croscrew, microfan 2008 [31]
(Sizes: > 250 µm)
a
FM: Ferromagnetic
5
1.2 State-of-the-art of MPCs for microsystem applications
6
Table 1.3: Summary of state-of-the-art of MPC materials developed for microsystems with magnetic particles < 1 µm. The
contributions are sorted by photopatternable/non-photopatternable and on the type of magnetism of the nanoparticles.
Polymer mat- Magn. Size of Type of Particle Fabrication method Structure/Application Reference
rix material particles magn. magnet- concentra-
1 Introduction
The use of particles with diameters < 1 µm allows shrinking of the feature sizes
of the MPC structures to a size where they benefit more from the advantages
of microsystems, such as efficient batch fabrication, enhanced heating processes
and chemical reactions, small probe volumes, and interaction with biological ma-
terials. In the last few years new processes have been established, enabling cost-
effective fabrication (e.g. flame spray pyrolysis) of various magnetic particles in
the nanometer size range [47, 48]. To achieve uniform mechanical and magnetic
properties in the microstructures, the particles must be much smaller than the
structure itself and well dispersed, with agglomerate sizes as small as possible.
MPCs with particle sizes < 1 µm are summarized in Table 1.3, and discussed in
the following.
The fabrication and the integration of various magnetic PDMS based compos-
ite microfluidic tools (Fe3 O4 particles with diameters of 200 nm mixed in PDMS
using a casting process) have been presented by Yamanishi [40, 49] such as micro-
valves, microstirrers, particle separators, cell sorting tools in a microfluidic chan-
nel for performing nonintrusive and contamination-free experiments on chips. A
further PDMS composite is developed by Pirmoradi [41]. Superparamagnetic
fatty-acid coated Fe3 O4 nanoparticles with 10 nm diameter have been mixed
in PDMS for the fabrication of a magnetic MPC membrane for possible use in
micropumps. The fabrication of a composite made with layer-by-layer nano-
assembly with a thickness of a few hundred nanometers (200 nm), which allows
the fabrication of cantilever structures is presented in [42]. However, the fabrica-
tion process for this composite is very complex.
One of the most promising microfabrication processes for polymers is pattern-
ing by photopolymerization. It allows batch-fabrication of microstructures with
small feature sizes. Nano-sized ferromagnetic particles have been incorporated
in the photocurable polymer SU-8 to create a ferromagnetic composite for use
of a micromirror on a cantilever [28] or a torsion actuator [44]. Structures with
sizes down to 5 µm have been presented [28]. However, there is no investigation
in these two works about the dispersion of the particles in the composite. The
presented optical images in these publications indicate large particle agglomer-
ates (> 1 µm).
During the time of this work other groups reported achievements in the same
field. Superparamagnetic γ-Fe2 O3 maghemite nanoparticles with diameters of
∼10 nm have been incorporated in SU-8 and 1002F with minimal agglomeration
(mean agglomerate diameter estimated from presented TEM images is ∼25 nm)
using oleic acid as a particle surfactant [45]. Fabricated micropallets with aspect
ratios of 4:1 are used for cell sorting. The particle concentration in the composite
is (0.01 – 1 wt.%).
A further advantage of photopatternable composites is the possibility of struc-
7
1 Introduction
This thesis focuses on the exploration of polymer composites filled with magnetic
nanoparticles for utilization in microstructures.
The goals of this thesis are:
8
1.4 Approach and outline of this work
9
1 Introduction
10
2 Theory
In this chapter the theory about magnetism in nanoparticles, the mechanics and
damping mechanism of oscillating structures, and the Young’s modulus of com-
posites are discussed.
There are several different unit systems used in magnetism, and it is important
that they be differentiated. The two most commonly used unit systems are the
SI system and the CGS system (centimeter, gram, second system). The equations
of magnetism varies depending on which unit system one uses. Table 2.1 shows
the main units in these two systems and the respective conversion factors. In this
work the SI unit system is used.
Table 2.1: Units and conversion factors between SI and CGS unit systems
Symbol SI unit CGS unit Conversion Factor
Magn. field H A/m Oe 1 Oe = 1000/4π A/m
strength
Magnetization M A/m emu/cm3 1 emu/cm3 = 1000 A/m
Am2 /kg emu/g 1 emu/g = 1 Am2 /kg
Magn. flux B Tesla [T] Gauss 1 Gauss = 10−4 T
density (kg/As2 )
11
2 Theory
B = µ0 ( H + M ) (2.1)
where µ0 = 4π· 10−7 T·m/A is the permeability of free space. The magnetization
of a material is related to the magnetic field by the susceptibility of the material
χ, which describes how a material reacts on an applied field.
M = χH (2.2)
The permeability, µ, describes the enhancement of a field H [A/m] generated by
an electric current when applied to a material to obtain a larger magnetic flux
density B (e.g. large B can be obtained when iron is inserted in a coil, µ Fe = 920).
B = µH (2.3)
Permeability and susceptibility are related as follows:
µ = µ0 (1 + χ ) (2.4)
• Ferro- and ferrimagnetism: χ and µ are large and positive, both are func-
tions of H, examples: Fe, Ni, Co
12
2.1 Magnetism in nanoparticles
13
2 Theory
R e m a n e n t M a g n e tiz a tio n M r
M a g n e tiz a tio n , M
S u p e r p a r a m a g n e tic
P a r a m a g n e tic
D ia m a g n e tic
C o e r c iv ity H c
A p p lie d F ie ld , H
Figure 2.2: Schematic of the M-H characteristics of different magnetic materials at room
temperature: Ferromagnetism, Paramagnetism, Diamagnetism, Superparamagnetism.
14
2.1 Magnetism in nanoparticles
Figure 2.3: Domain creation (for cubic crystals) in absence of external applied field. Single-
domain forms below the critical diameter Ds . Superparamagnetism occurs at particle
sizes Dsuperparamagnetism , where k B T > Ke f f V. Parts of the schematic are adapted from
[50].
shows the schematics of the particle coercivity versus the particle size.
As the particle size is further decreased, we reach a second critical diameter
called the superparamagnetic limit. The superparamagnetic limit for Fe3 O4 is
reported to be ∼20 nm [53–55]. Below the superparamagnetic limit, the energy
required to switch the magnetic moment is in the same range as the thermal en-
ergy k B T, where k B is the Boltzmann constant and T the temperature. The energy
required to hold the magnetic moment in a certain direction is called the aniso-
tropy energy:
15
2 Theory
Figure 2.4: Schematic of the particle coercivity versus the particle size (diameter). Ds is
the single-domain size, Dsuperparmagnetic the critical diameter for superparamagnetism.
Schematic adapted from [50].
tem behaves like a paramagnet. The moment is free to move and respond to an
applied field independent from the particle. The energy kT try to disorder the
alignment as it does in a paramagnet. Thus, this phenomenon is called super-
paramagnetism [50]. Such a system has no hysteresis, as schematically depicted
in Figure 2.2, and the data of different temperatures superimpose onto a univer-
sal curve of M versus H/T [56].
The phenomenon of superparamagnetism is timescale-dependent because of
the stochastic nature of the thermal energy [50]. The relaxation time from a cer-
tain orientation is described by the Néel relaxation,τN , [56]
Ke f f V
kB T
τN = τ0 e (2.6)
with τ0 the time constant (τ0 ∼ 10−9 s). If the experimental time scales is larger
than this Néel relaxation time the particle magnetic moment flips. As a result,
the overall ferromagnetic moment of the particle is randomized to zero and the
system is in a superparamagnetic state (a typical experiment with a magneto-
meter takes 10 to 100 s) [57]. If the experimental time is shorter, the particle is in
the so-called blocked state. The temperature which separates these two regimes
is called blocking temperature. Therefore, superparamgnetism depends on the
particle size, the temperature and the timescale of the measurement.
With decreasing particle size the saturation magnetization decreases [55, 58].
The ratio of surface atoms to bulk atoms in the nanoparticle increases with de-
16
2.2 Mechanics of oscillating structures
creasing particle size. The reduction has been assigned to spin canting, magnetic
dead layers on the particle’s surface or the existence of spin-glass-like behavior
of the surface spins [56]. For example about 60 % of the total number of spins for
a 1.6 nm cobalt nanoparticle are surface spins [56].
m ẍ + c ẋ + kx = F0 sin ωt (2.7)
with the mass m, the coefficient of damping force c, the spring constant k. F0
and ω are the excitation force and frequency, respectively, assuming sinusoidal
excitation.
ω0 is the natural frequency of the system without damping
k
ω02 = (2.8)
m
and the damping ratio of the system is given by
c
ζ= (2.9)
2mω0
For steady vibration and slight damping, the equation (2.7) results in amplitude
B of
F0 /m
B= q (2.10)
(ω02 − ω 2 )2 + (2ζω0 )2 ω 2
where the phase lag ϕ between the mass dispacement and excitation force is
!
−1 2ζω0 ω
ϕ = tan . (2.11)
ω02 − ω 2
A slightly damped mechanical system driven by a sinusoidal input can be de-
scribed by the natural frequency w0 and the damping ratio ζ. The resonant fre-
17
2 Theory
Figure 2.5: Schematic of a linear damped harmonic oscillator with one degree of freedom.
For ζ < 0.707 the system is slightly damped and has a resonance peak. If ζ ≥ 0.707
the system is overdamped and the resonance disappears. The damping ratio ζ =
0.707 is called critical damping.
Resonance of a beam
18
2.2 Mechanics of oscillating structures
L w
∂2 u ∂4 u
ρA 2
+ YIz 4 = 0 (2.13)
∂t ∂x
with ρ as mass density, A as cross sectional area, Y as Young’s modulus and Iz as
geometrical moment of inertia. For a rectangular beam the moment of inertia is
Ah2
Iz = (2.14)
12
where h represents the beam thickness. By solving equation 2.13 through separa-
tion with variables, the eigenfrequency of a thin beam is
s
λ2 YIz
ω0 = 2π f 0 = n2 , n = 1, 2, ... (2.15)
L Aρ
where L is the length of the cantilever and λn is the solution of the frequency
equation 1 + cos λ cosh λ = 0 for single-clamped beams with the following solu-
tions
λ1 = 1.8751 (2.16)
λ2 = 4.6941
λ3 = 7.8548
π
λn = (2n − 1)
2
19
2 Theory
Using the moment of inertia Iz (Equation 2.14) the n-th eigenfrequency can be
written as
s
λ2 h Y
fn = n 2 , n = 1, 2, ... (2.17)
2πL 12ρ
F
LF
∂2 u ( x )
YIz = M( x) (2.19)
∂x2
∂u( x )
u( x )| x=0 = 0 =0 (2.20)
∂x x=0
the equation (2.19) can be solved. The deflection of the beam for a single point
20
2.2 Mechanics of oscillating structures
FL3F
umax = (2.21)
3YIz
For small deformations, the displacement and the applied force follow a linear
relationship by Hooke’s law:
F
k( x) = (2.22)
u( x )
where k is the spring constant and F is the applied force. The mechanical spring
constant is the ratio of the applied force and the resulting displacement where
the force is applied.
If Equation 2.21 and Equation 2.22 are combined, the spring constant k of a
beam with load position L F can be described as:
3YIz
k( L F ) = (2.23)
L3F
Using the obtained equation (2.23) and inserting the moment of inertia Iz (2.14),
the Young’s modulus can be expressed as
4kL3F
Y= (2.24)
wt3
For the calculation of the resonance frequency of a beam with end mass the
Rayleigh-Ritz method can be used [59]. Comparing the maximal kinetic and po-
tential energy of the resonator the natural frequency of the first mode can be
derived
s
Ywh3
ω0 = 33
(2.25)
4(m a + 140 m b ) L3
for vertical movement (cantilever), where m a is the additional mass at the tip and
mb the beam mass. For lateral movement (in-plane resonator) the equation is
written as
21
2 Theory
s
Yhw3
ω0 = 33
(2.26)
4(m a + 140 m b ) L3
Estored
Q = 2π (2.27)
Eloss
where Estored is the total stored energy, Eloss is the energy loss per oscillation cycle,
and ζ is the damping ratio. In case of slight damping the quality factor can be
expressed as [59]
1
Q= (2.28)
2ζ
For resonating systems the quality factor Q is an important parameter. The higher
the Q-factor, the lower the energy loss per cycle and the sharper the resonance
peak. It is possible to determine the Q-factor experimentally by measuring the
half-power bandwidth around resonance. This is the frequency range where the
displacement response is at √1 times its value at resonance, or on a logarithmic
2
scale, -3dB. Q is than defined as
f res
Q= (2.29)
∆ f −3dB
with f res as resonance frequency and ∆ f −3dB as bandwidth with √1 amplitude.
2
22
2.3 Young’s modulus of composites
and the matrix. The combination of these different influences produces a system
with high complexity. A non-bonded particle could in principle act as a hole
and decrease the Young’s modulus of the composite [61]. Furthermore, for small
particles and high particle fractions the surfactant layer also must be considered
for the volume fraction calculation [65]. Hashin-Shtrikman suggested a model for
the calculation of Young’s modulus with an upper and a lower boundary, which
describes most of the experimental data for composites [61].
The Hashin-Shtrikman Model gives an approximation for the upper and the
lower boundaries of the Young’s modulus of a composite dependent upon the
ratio of moduli of the two phases with arbitrary interface geometry. The elastic
modulus of the filler must be considerably higher than the glassy modulus, and
the filler must be evenly dispersed [61]. In case of rigid polymeric-filled system
(large ratio of moduli) the boundaries are more spaced.
For the case of a material containing only two different specimens, the static
shear and bulk modulus for the composite are given by:
−1
1 6ξ i (Ki + 2Gi )
Gc = Gi + ξ j + (2.30)
( Gj − Gi ) 5Gi (3Ki + 4Gi )
−1
1 3ξ i
Kc = Ki + ξ j + (2.31)
(Ki − K j ) (3Ki + 4 Gi )
where Gi and Gj are the shear moduli, Ki and K j are the bulk moduli and ξ i and
ξ j are the volume fractions of the two specimens. The two parameters Gc and Kc
describe the shear and the bulk moduli for the composite. Interchanging the two
indices i and j in (2.30) and (2.31), an upper and a lower boundary of the shear
and the bulk moduli is obtained.
The shear and the bulk moduli can be calculated from the Young’s modulus
and the Poisson’s ratio ν of the two materials.
Y
G= (2.32)
2(1 + ν )
Y
K= (2.33)
3(1 − 2ν)
23
2 Theory
The upper and the lower boundary for the Young’s modulus is given by [61]:
9Kc Gc
Yc = (2.34)
3Kc + Gc
24
3 Evaluation of materials
In this chapter the selection of the polymer material and the magnetic filler for
the composite for the fabrication of magnetic microstructures are presented. Fur-
thermore, a suitable dispersant agent is evaluated to obtain a high quality particle
dispersion in the final composite.
The goal is to mix a composite with photodefinable property. There are a vari-
ety of non-photodefinable polymers which fulfill the required criteria and where
a photoinitiator can be mixed to make them photosensitive. However, the pho-
toinitiator has to be well adjusted to the polymer chemistry. Here, only polymers
are considered where the photosensitive chemistry is already tuned.
25
3 Evaluation of materials
Two negative tone photodefinable polymer types, epoxy and polyimide, were
taken into account for the choice of matrix material of the composite for the fab-
rication of magnetic microstructures. Polyimides and epoxies are highly cross-
linked amorphous polymers and have a high glass transition temperature, and
usually low water uptake. They show little physical aging in their glass state and
the chemical inertness is usually high, making them suitable for application in
different environments [68]. It has been shown that polyimides and epoxies have
relatively low creep and are suited for mechanical applications in MEMS techno-
logy [68]. Both polymers are known for their high wear resistance, high strength
and Young’s modulus and high temperature resistance. Furthermore, they show
excellent resistance to chemicals and cracking [69].
26
3.1 Polymer evaluation
Table 3.1: Considered photodefinable polymers for the fabrication of MPC for
microstructures.
Polymer Polyimide Epoxy
Product HD-4100 (PI-2737) SU-8
Company HD MicroSystems MicroChem Corp.
Tone Negative resist Negative resist
Component Two-component Two-component
Fabrication Photolithography Photolitography
Fracture strength 200 MPa [74] 60 MPa [75]
Young’s modulus 3.4 GPa [74] 4.0 GPa [76]
Glass transition 330 ◦ C [74] 210 ◦ C [75]
temperature Tg ,
> 100 ◦ C enables
PCR reactions
Functionalization Possible but complex [77] Well known [70–72]
possibility for
biomolecules
Biocompatibility Not known Cell proliferation repor-
(in vitro) ted [78, 79]
Processability with Yes Yes
spin-coating
Solvent N-Methyl-2-Pyrrolidone Cyclopentanone, GBL
High chemical sta- Yes [74] Yes [75]
bility
Reported compos- [33, 80, 81] [4, 28, 44, 45]
ites containing
magnetic particles
27
3 Evaluation of materials
SU-8 was developed by IBM research and it is commercially available from Mi-
croChem and Gersteltec. It is based on Epon SU-8 from Shell Chemical (gly-
cidyl ether derivative of bisphenol-A novolac), which can be dissolved in an or-
ganic solvent like γ-Butyrolacton (GBL), propylene glycol methyl ether acetate
(PGMEA), methyl iso-butyl ketone (MIBK) and cyclopentanone. The dissolved
monomer is mixed with the photoacid generator triaryl sulfonium salt (CYRA-
CURE UVI from Union Carbide) 10 wt.%, which is dissolved in propylene car-
bonate [82]. The monomer is characterized by very high epoxy functionality and
low molecular weight at the same time [83]. The very low absorption in the near-
UV range makes SU-8 attractive for ultra thick resist applications. The chemical
structure of SU-8 monomer and the crosslinked polymer is depicted in Figure 3.1
(a) and (b), respectively. Upon irradiation, the photoacid generator triaryl sulf-
onium salt decomposes to form hexafluoroantimonic acid, Figure 3.1(c), which
initiates the cationic ring-opening polymerization of the epoxy groups (d). This
starts the chain reaction of the crosslinking process during heating (e) [84]. Due
to the eight epoxy sites (which gives the resin its name) the polymerization yields
in a very dense, stable polymer after full curing with a degradation temperature
of ∼380 ◦ C. SU-8 is optically transparent and highly functional. Further proper-
ties are listed in Table 3.1. Aspect ratios of around 20 using UV exposure and
over 100 using X-ray lithography have been reported with straight sidewalls [84].
Layer thicknesses from 2 to 300 µm in a single coating process can be obtained.
SU-8 exhibits a high Young’s modulus which guarantees mechanical stability for
microstructures and is well suited as material for functional microdevices.
For the fabrication of composites for magnetic microstructures with smallest struc-
ture dimensions < 3 µm (thickness of cantilevers), the particles must have dimen-
sions much smaller than this minimal structure size (< 100 nm). Furthermore, the
magnetic particles need to be well dispersed to obtain uniform mechanical and
magnetic properties. For the fabrication of thin layers (< 3 µm) by spin-coating,
the viscosity of the composite must be low (static viscosity < 4·10−4 m2 /s). Dif-
ferent magnetic particles with sizes < 100 nm, have been investigated for their
suitability as filler material for the polymer composite. The dispersability of the
nanoparticles depends on forces on the particle surfaces (Van der Waal attraction)
and on magnetic forces (magnetic attraction between particles). The surface inter-
actions can be controlled by immobilization of molecules on the particle surface.
The investigation of a suitable surfactant is discussed in detail in Section 5.1 and
28
3.2 Particle evaluation
Figure 3.1: Chemical structure of the SU-8 monomer (a) and polymer (b). The eight bind-
ing sites per monomer lead to a highly crosslinked polymer after polymerization. The
photoacid generator triaryl sulfonium salt decomposes during UV exposure to form
hexafluoroantimonic acid (c). This initiates the cationic ring-opening polymerization
of the epoxy group (d) and starts the chain reaction of the crosslinking process during
heating (e). Chemical formulas adapted from [85].
29
3 Evaluation of materials
depends on the selected particle material. Therefore, magnetic properties are the
determining factors for the particle selection.
Different available magnetic nanoparticles were investigated for the suitabil-
ity of the composite and are listed in Table 3.2. To investigate their magnetic
properties the nanoparticles have been characterized by a vibrating sample mag-
netometer (VSM) (MicroMag, Model 3900). The obtained magnetic properties
and advantages and disadvantages of the particles are listed in Table 3.2. All
particles are single domain (Table 2.2). The nickel (Ni) and cobalt (Co) particles
show ferromagnetic characteristics and have an elevated coercivity and higher
saturation magnetization. The magnetite (Fe3 O4 ) and maghemite (γ-Fe2 O3 ) nan-
oparticles are below the critical particle size and exhibit superparamagnetic prop-
erties. They show negligible or low remanent magnetization.
Figure 3.2 shows the schematics of the magnetic characteristics of typical fer-
romagnetic nanoparticles (Ni particles with a diameter ∼20 nm, single domain,
blocked) in comparison to superparamagnetic nanoparticles (Fe3 O4 with a dia-
meter < 20 nm, single domain, not blocked). Ferromagnetic particles exhibit a
high remanent magnetization, Mr , which can be a benefit for the actuation of
magnetic microstructures. However, due to the high remanent magnetization,
ferromagnetic (blocked) nanoparticles tend to agglomerate by magnetic attrac-
tion in a liquid polymer matrix. They can easily form agglomerates with sizes big-
ger than the desired microstructures (∼5 µm) [31]. A viscosity-increasing agent
can be used to reduce this particle agglomeration after dispersion [31]. However,
for applications by thin layer spin coating with thicknesses smaller than 5 µm the
polymer needs a low viscosity (< 4·10−4 m2 /s).
Ideal superparamagnetic particles do not retain any remanent magnetization.
Therefore, the particles have low magnetic attraction during and after mixing
of the low-viscosity polymer composite. Particles can be dispersed more easily
and the dispersion remains more stable. Therefore, superparamagnetic particles
are more suitable to fabricate a magnetic composite which can be used for the
fabrication of microstructures. However, even without any magnetic interactions
nanoparticles tend to form agglomerations to reduce the energy associated with
the high surface area to volume ratio of the nanosized particles [56]. Additionally,
a surfactant on the superparamagnetic particles must be evaluated (described in
Section 5.1).
Superparamagnetic MagSilica and γ-Fe2 O3 silica coated particles both have
the advantage of a silica surface, which is well studied for functionalization and
for compatibility with polymers [86]. However, Fe3 O4 nanoparticles from Chemi-
cell GmbH have a much higher saturation magnetization compared to the γ-
Fe2 O3 particles. Furthermore, the Fe3 O4 particles have negligible coercivity and
remanent magnetization, which is ideal to have low magnetic attraction between
30
3.2 Particle evaluation
Magnetization, M
Superparamagnetic particles
0 Superparamagnetic
particles
0
Applied Field, H
Figure 3.2: Schematics of the magnetic characteristics of ferromagnetic (blocked) and su-
perparamagnetic (not blocked) nanoparticles. Ferromagnetic nanoparticles like Co or
Ni particles have a high remanent magnetization Mr and agglomerate in a dispersion
with a low-viscosity due to strong magnetic forces between particles. In contrast, su-
perparamagnetic particles have negligible remanent magnetization at room temperat-
ure and therefore low magnetic attraction in dispersion. The reason for the low inter-
action is based on the thermal energy, which flips the direction of the magnetization of
the single domain nanoparticles. (Adapted from [87],
c IOP.)
the particles. Therefore, the Fe3 O4 nanoparticles from Chemicell GmbH are selec-
ted as filler material for the photocurable MPC.
31
32
Table 3.2: Evaluation of filler particles. The nanoparticles have been characterized by a vibrating sample magnetometer (VSM) at
room temperature to obtain the magnetic properties.
Material Magnetic Particle Saturation Coercivity Remanent Advantages Disadvantages
behavior diameter magnetization magnetization
@ ∼800 kA/m
[nm] [Am2 /kg] [kA/m] [Am2 /kg]
(emu/g) (emu/g)
Fe3 O4 from Superpara- 12.4 a 53.5 <1 <1 Negligible reman-
3 Evaluation of materials
The selected nanoparticles must be well dispersed within the polymer to achieve
homogeneous mechanical and magnetic properties in the composite. Nanopar-
ticles tend to form agglomerates in a suspension due to inter-particle attractive
forces, like magnetic forces or Van der Waals forces (electrical dipole-dipole force).
A surfactant (dispersant agent) on the nanoparticle can reduce the formation of
agglomerates. The selection of a suitable surfactant is a crucial part to obtain a
high quality MPC.
This section gives a brief overview about the possibility of stabilizing nano-
particles and present the selection of a suitable dispersant agent.
33
3 Evaluation of materials
• chemical affinity with the solvent (the particle must be stable in the polymer
solvent up to the desired concentration),
34
3.3 Dispersant agent evaluation
Surfactants with a phosphate end group have been selected. Phosphate groups
are known to adsorb strongly onto magnetite surfaces [92]. Figure 3.4 shows the
schematic how the surfactant connect to the magnetite particle. Two phosphate
based surfactants were investigated in detail for the use in the composite: Diphos-
phate (agent 1) and a linear copolymer containing a phosphate group (agent 2).
With both surfactants suspensions with very high particle concentration were
achieved. The suspension with agent 1 coated Fe3 O4 particles have a concentra-
tion of 115 mg/ml. The second suspension, particles coated with agent 2, have
a particle concentration of 280 mg/ml. Such a high concentration in the disper-
sion is crucial for the fabrication of composites with particle concentrations up
to 10 vol.%. Both suspension were still stable after 10 months and no obvious
sedimentation could be observed.
However, the stability in the solvent does not guarantee the stability and well
dispersion in the polymer matrix. Using agent 1 as a dispersant, relatively large
magnetite agglomerates bigger than five microns in size can be observed under
an optical microscope after mixing with the polymer (Figure 3.5 (a)). Such a
composite would result in microstructures with rough surfaces and non-uniform
mechanical properties. When magnetite nanoparticles are functionalized with
agent 2, agglomerates are too small to be recognized (Figure 3.5(b)) by optical
microscopy. Agent 2 is selected for the fabrication of the composite.
35
3 Evaluation of materials
3.4 Conclusion
In this chapter the suitable components such as the polymer, the magnetic nan-
oparticles and a dispersant agent were evaluated. For the polymer matrix ma-
terial the negative tone photodefinable SU-8 epoxy resist was selected because
of the high chemical stability, possibility for surface functionalization due to the
epoxy binding sites, and the high glass transition temperature. Superparamag-
netic Fe3 O4 particles with diameters of ∼13 nm were selected for the filler mater-
ial. The particle have a high saturation magnetization (53.5 Am2 /kg) and have
negligible remanent magnetization (< 1 Am2 /kg). Therefore, the superparamag-
netic particles have low magnetic attraction compared to ferromagnetic particles
during and after mixing of a polymer composite and the formation of agglom-
erates can be reduced. The influence of different dispersing agents on the nano-
particles agglomerations was investigated. A surfactant with a linear copolymer
containing a phosphate group was evaluated, which leads in combination with
the superparamagnetic nanoparticles to a homogeneous particle dispersion in
the composite with low agglomeration (Figure 3.5 (b)).
36
4 Fabrication
1 The fabrication of the stabilized nanoparticle suspension and the mixing of the
selected magnetic particles with the photocurable polymer to a magnetic compos-
ite are presented in this section. A fabrication process using spin-coating, photo-
lithography and a sacrificial layer etch is used to fabricate suspended microstruc-
tures. The schematic of the process with the selected material components to ob-
tain suspended microstructures (cantilevers) is shown in Figure 4.1. Microcanti-
levers and lateral microresonators are fabricated with the developed magnetic
polymer composite (MPC). To make the handling easier and allowing measure-
ments in water the fabricated microresonators were directly packaged into a poly-
mer channel. The fabrication of the all-polymer package and the integration of
the microresonators using batch-process is shown in the last section.
37
4 Fabrication
38
4.2 Microstructure fabrication
Fe3 O4 nanoparticles were fabricated and stabilized in the organic solvent γ-Buty-
rolacton (GBL) to obtain a stable nanoparticle dispersion which can be mixed
with the SU-8 resin. GBL is used because it is a solvent for SU-8. The nanoparticle
dispersions were developed in cooperation with Chemicell GmbH (Berlin, Ger-
many). A wet-chemical synthesis from iron-II/III-salt dissolution is used to fab-
ricate the magnetite particles. The particles are washed with deionized water,
coated with different dispersion agents and transferred into the organic solvent
γ-Butyrolacton (GBL). A stable particle suspension could be obtained for two
different phosphate-containing dispersing agents: Diphosphate (agent 1) and a
dispersant agent consisting of a linear copolymer containing a phosphate group
(agent 2). The dispersions are very stable and even after 10 months no obvious
sedimentation is observed. Nanoparticle concentrations of up to 280 mg/ml are
achieved. Such a high concentration in the dispersion is crucial for the fabrication
of composites with particle concentrations up to 10 vol.%.
For the composite mixing the obtained nanoparticle dispersion is sonicated for
10 min in a conventional ultrasonic-bath. The nanoparticle dispersion is then
mixed (ratio 2:3) with the dissolved SU-8 (with a low solvent content, 48% GBL)
from MicroChem Corp. (Newton, USA) to set the viscosity to a range smaller
than 4·10−4 m2 /s, where the desired thickness of 1 – 3.5 µm by spin coating can
be achieved. Additionally, the composite is mixed for 10 minutes in a planetary
mixer (dual asymmetric centrifugal mixer, DAC 150 FVZ, Hausschild) with an
initial 2 min speed ramp (0 to 3000 rpm). The composite is treated by ultrasonic
sound (Vibracell VCX 600 Sonics and Materials Inc.) for 20 minutes [96] and then
spin-coated onto a support glass wafer for further processing.
39
4 Fabrication
on wafers and baked for 14 min at 100◦ C. The film is patterned with photolitho-
graphy using a dose of 5 J/cm2 to 75 J/cm2 depending on the thickness (Needed
exposure doses are discussed in detail in Section 5.2). The exposed composite is
then developed with MR-Dev 600 (MicroChem Corp.). A hard-bake at 150◦ C for
30 min ensures full polymerization of the composite. To form the anchor struc-
ture, a 100 µm thick pure SU-8 layer is spin coated on top of the composite layer
and is exposed with 0.4 J/cm2 . A suspended cantilever base area given by an un-
avoidable misalignment of the cantilever-layer to the anchor layer could change
the mechanical resonant frequency characteristics of the cantilever. Therefore,
the anchor layer is designed to overlap with the cantilever structures by 5 µm.
Even with a misalignment (< 5 µm) the cantilever lies on the anchor and the sus-
pended cantilever length can be determined by optical measurements. After the
anchor layer development the wafer is diced into 4 x 4 mm chips containing
cantilever arrays. The chip is placed into chromium etchant (500 ml H2 O, 100 g
Ce(NH4 )2 (NO3 )6 , 17.5 ml acetic acid) to release the cantilever (1 hour) by dissolv-
ing the sacrificial layer. After the release the resonator arrays were rinsed with
DI-water. Due to the high surface tension of the water, the suspended micro-
structures bend during drying and stick to the substrate, even if the anchor has a
height of 90 µm. To avoid bending and sticking of the resonators to the substrate
after drying process the resonators were placed facing a clean room paper (TX
5811) and dried in vacuum to avoid any water film which would increase the
adhesion forces between the microstructure and the paper. The porosity of the
paper allows a drying of the microstructures without damage and an easy release
due to low contact areas.
Microcantilever arrays were successful fabricated. Figure 4.3(a) shows a scan-
ning electron microscope (SEM) image of a released composite cantilever with
5 vol.% (18 wt.%) of magnetite nanoparticles. Cantilevers with the following geo-
metries and particle concentrations were fabricated Table 4.1.
With the same process in-plane microresonators with different geometries (Table
4.2) were successful fabricated. Figure 4.3 (b) shows a released MPC in-plane
microresonator with 5 vol.% of magnetite nanoparticles. The geometrical dimen-
sions of the fabricated in-plane structures are listed in Table 4.2.
40
4.2 Microstructure fabrication
1 Evaporation of Cr/Au/Cr
(5 nm / 50 nm / 50 nm)
on silicon support wafer
2 Spin-coating of photo-
sensitive MPC (1.5 µm)
and UV-exposure
3
Development of
photosensitive MPC
4
Spin-coating of
SU-8 (100 µm)
and UV-exposure
5
Development of SU-8
6
Release from
sacrificial layer
Figure 4.2: Schematic of the process flow for the fabrication of magnetic polymer canti-
levers: 1) evaporation of sacrificial Cr/Au/Cr layer, 2) spin coating of photocurable
magnetic polymer composite and patterning by UV exposure, 3) development of com-
posite to achieve microstructures, 4) spin coating of 100 µm thick SU-8 anchor layer
and patterning by UV exposure, 5) development of anchor layer, and 6) release of can-
tilever by etching of sacrificial layer. The overlapping anchor layer ensures a defined
anchor for the microcantilever. ([87],
c IOP.)
41
4 Fabrication
Figure 4.3: SEM image of MPC microstructures with 5 vol.% of magnetite nanoparticles
concentrations: (a) micro cantilever, (b) in-plane resonator.
Table 4.2: Geometrical dimensions and particle concentration of MPC in-plane resonators.
Beam length Beam width Plate length Thickness Concentration
LB w s h
10 - 600 µm 3 - 25 µm 3 - 400 µm 1.9 - 3.5 µm 0, 1, 2, 3, 5 vol.%
42
4.3 Polymer package fabrication
The fabricated microresonators have to be packaged into a fluid chamber for the
measurements in water. An all-polymer low-cost transparent package was fab-
ricated which enables an easy handling and the testing of the resonators in water
and is suitable for disposable applications. Figure 4.4 shows cantilever arrays
bonded into the developed polymer package. The requirements for the package
are:
The MPC cantilevers will be actuated by the magnetic field of an external coil.
The magnetic field decreases fast with increasing distance from the coil surface
(Figure 6.6). Therefore, it is important that the distance between MPC resonat-
ors and the actuation coil is as small as possible. However, the coil should not
touch the package to avoid mechanical disturbance from the coil vibrations. The
bottom layer of the package has to be as thin as possible (< 500 µm). The pack-
age must fit into the different actuation setups (thickness of packaging < 3 mm,
width < 15 mm, one of the used actuation setup is shown in Figure 7.3). PMMA
was chosen as package material, because it is transparent and can be bonded
by various bonding technique such as fusion bonding and bonding by solvents
[99]. Hot embossing of PMMA for the fabrication of fluid channel is well known.
Channels with aspect ratios of 0.03 and channel depths of 3 µm have been repor-
ted [100]. Another possibility to fabricate microchannels is to cut fluid channels
into a polymer sheet by a laser and bonding a top and a bottom sheet to close the
package. Laser cutting technique has the advantages compared to hot embossing
technique, that the design of the fluid channels can be easily changed. For small
batch-fabrications the fabrication time of the laser cutting is similar compared to
hot embossing at high temperature. Additionally, for hot embossing a stamp is
needed. In this work the package fabrication by laser technique was used.
The fabrication process of the package is shown in Figure 4.5. A PMMA sheet
with a thickness of 250 µm was selected for the bottom plate to ensure a short
distance of the resonator to the external coil and to guarantee enough mechanical
stability. Firstly, solvent GBL is spin-coated onto the PMMA bottom plate with
1000 rpm for 30 s to partially dissolve the PMMA surface. The released resonator
arrays can be placed (Figure 4.5 a) onto a bottom plate and the SU-8 anchors bond
43
4 Fabrication
with the bottom plate. The anchors of the cantilever arrays were designed to
allow a handling by tweezers. Secondly, a PMMA channel plate with a thickness
of 1 mm is treated by laser cutting to obtain the fluid channels (Figure 4.5 b).
The channel plate is bonded with the bottom plate by solvent bonding technique
using GBL and a pressure of 6 N/cm2 at 25 ◦ C (Figure 4.5 c). The batch was
closed by a polycarbonate (PC) top plate (with thickness of 1 mm). The top plate
was prior treated by 1,2-dichloroethane to partially dissolve the surface and was
bonded with a pressure of 47 N/cm2 at 25 ◦ C (Figure 4.5 d). Finally, the package
batch is diced by a laser mill. The final package (Figure 4.5 e) have dimensions of
2 x 10 x 30 mm. A final package is presented in Figure 4.4.
A critical process parameter is the bonding pressure. A too high bonding pres-
sure can result in cracks in the middle PMMA plate. A too low pressure can lead
to peeling of the top or bottom sheet. Furthermore, it is important that the GBL
is spin coated with 1000 rpm to obtain a thin film. Lower spin coating speeds
give thicker solvent films, which can make contact to the cantilevers, and result
in sticking of the cantilevers to the bottom sheet due to surface tensions of the
solvent during drying.
The presented fabrication process is easy adjustable for other channel designs
and package sizes. Packages with larger dimensions depending on the actu-
ation setup were fabricated. In principle the package can be miniaturized by
taking thinner plates and fabricating finer channel structures. However, during
laser cutting the polymer at the cutting edge melts and forms a bump of around
15 µm height. This elevation can raise adhesion problems of the bonded plates
and may limit the process for smaller feature sizes.
44
4.3 Polymer package fabrication
Figure 4.5: Schematic of the process flow for the fabrication of the all-polymer package
and the integration of the resonator array. (a) The resonator arrays are bonded onto
the PMMA bottom plate. (b) The channel plate is shaped by a laser mill and (c) bonded
to the bottom plate. (c) A top plate is bonded onto the channel plate. (d) The batch is
diced into single packages.
45
4 Fabrication
4.4 Conclusion
To fabricate the MPC the stable initial magnetite nanoparticle suspension was
mixed with the polymer by centrifugal mixing and ultrasonic steps. A stable low-
viscous MPC suspension with nanoparticle concentration up to 10 vol.% was ob-
tained which can be used for spin-coating processes. Suspended microstructures,
with 5 vol.% nanoparticle concentrations, such as cantilevers and in-plane reson-
ators were successfully fabricated using photolithography and a sacrificial layer
etch process. An all-polymer low-cost transparent package was manufactured
where the fabricated microresonators can be integrated. The package allows an
easy handling and the testing of the resonators in media such as water and air
using different magnetic actuation setups.
46
5 Evaluation of composite
properties
47
5 Evaluation of composite properties
In this section the agglomerations of the particles and the particle sizes in the
composite are investigated. First, slices of the composite have been prepared to
analyze the agglomerate and particle sizes by transmission electron microscopy
(TEM). Scanning electron microspobe (SEM) images are not shown here because
they do only represent the particles very close to the surface and do not have an
enough high resolution. TEM images represent only a very small section from the
composite. To obtain information from larger areas small angle X-ray scattering
(SAXS) measurements have been performed from samples with different particle
concentrations. With X-ray disc centrifuge (XDC) measurements the agglomerate
and particle sizes in the initial particle suspension were investigated to evaluate
if the mixing procedure enhance the formation of agglomerates.
TEM Analysis
Experimental
The particle diameter and the agglomerate diameter are determined from sev-
eral TEM images (Philips CM12 with a tungsten filament at 100 kV) using the
software ImageJ 1.42q. Grinding and dimple-grinding (Gatan 656) are used to
prepare thin film slices of the composite. A hole is etched in the sample by two
focused argon ion beams with the precision ion polishing system (PIPS) Gatan
691. The probe is thin enough for TEM observation at the edge of the hole.
48
5.1 Particle dispersion in composite
Figure 5.1: TEM image of a cross-section of the polymer composite film with (a)
3 vol.% (12 wt.%) and (b) 5 vol.% (18 wt.%) Fe3 O4 particle concentration. (Sample
thickness ≈ 100 nm). ([93],
c (2011), with permission from Elsevier).
49
5 Evaluation of composite properties
SAXS Analysis
1 Inorder to quantitatively determine the size distribution of particles and ag-
glomerates, small angle X-ray scattering (SAXS) measurements were performed.
SAXS measurements provide information about shape and the size of incorpor-
ated particles in a composite. The particle distribution in the composite can be
modeled assuming the particle shape and by using the measured intensity I (q) of
the scattering vector q. The detailed characterization of the SAXS measurements
can be found in [93].
Experimental
The agglomeration states of the particles and particle sizes have been analyzed us-
ing SAXS. Samples with different particle concentrations have been investigated
1 The SAXS analysis were performed in cooperation with Christian Moitzi (Adolphe Merkle Institute,
University of Fribourg, Switzerland)
50
5.1 Particle dispersion in composite
Figure 5.3: (a) Particle diameter distribution measured from TEM images (count distribu-
tion from sample with 1 vol.% and 5 vol.%particle concentrations). The particles have
a mean count diameter of 11.4 ± 3.4 nm (800 counts). (b) Agglomerate diameter distri-
bution from TEM images (366 counts) of a composite with 5 vol.% particle concentra-
tion. The average diameter is 43.5 ± 20.4 nm. ([93],
c (2011), with permission from
Elsevier).
to determine the agglomeration behavior depending on the fill level. The SAXS
equipment consists of a S-MAX3000 instrument from Rigaku Innovative Techno-
logies (Auburn Hills, MI, USA). The radiation is produced by a high-intensity
micro-focus sealed tube X-ray generator (Rigaku MicroMax-002+) with a copper
anode (wavelength 0.154 nm). A pair of multilayer reflectors is used to produce a
monochromatic beam which is collimated by a three pinhole system. In the fully
evacuated camera the sample is placed at a distance of 1525 mm from the detector.
The diameter of the incident x-ray beam at the sample position is approximately
0.8 mm. The sample absorption is measured by a photo diode which is moun-
ted on the beam-stop. The scattered intensity is detected by a fully integrated
two-dimensional multi-wire proportional counter (Rigaku Triton) which makes
highly sensitive measurements from isotropic and anisotropic materials possible.
The diameter of the sensitive area of the detector is 200 mm. The range of access-
ible scattering vectors q is defined as
4π θ
q= sin (5.1)
λ 2
and is 0.07 < q < 4.5 nm−1 , with λ being the x-ray wavelength, and θ being the scat-
tering angle. Program SASfit (by Joachim Kohlbrecher and Ingo Bressler, Paul
Scherrer Institute, CH) was used for the model fitting approach to calculate the
bi-lognormal distribution of homogeneous spheres. Composite films with a typ-
51
5 Evaluation of composite properties
-1
1 0
I(q ) [a .u .]
-2
1 0 P a r tic le c o n c e n tr a tio n
1 v o l%
2 v o l%
-3 3 v o l%
1 0
5 v o l%
1 0 v o l%
-4
1 0
0 .1 1
-1
q [n m ]
Figure 5.4: SAXS measurements: 0 vol.% probe is subtracted, the thickness of the sample
was considered and the data are normalized with respect to particle concentration.
The overlap of the curves confirms the expected particle concentrations. The similar
shape of the curves shows that the agglomeration states of particles in the composite
are independent of the concentration ([93],
c (2011), with permission from Elsevier).
ical thickness between 1.54 µm and 2.23 µm were measured using a transmission
geometry with a measurement time of one hour.
52
5.1 Particle dispersion in composite
0 .6 1 3 n m
0 .4
4 0 - 5 0 n m
0 .2
0 .0
0 2 0 4 0 6 0 8 0 1 0 0
A g g lo m e r a tio n d ia m e te r [n m ]
(i.e. single particles or agglomerates) were used. First, an Indirect Fourier Trans-
formation (IFT) [102] was applied to calculate the size distribution by assuming
homogeneous spherical particles. Second, a model fitting approach was used to
calculate the bi-lognormal distribution of homogeneous spheres that gives the
best fit to the experimental data. In Figure 5.5, both volume weighted distribu-
tions are shown. Within the experimental error of the method, the two results
overlap. The main fraction of particles is between 10 and 20 nm in diameter.
However, there is a significant portion of particles or agglomerates which are
larger. The distribution of diameters extends up to 80 nm. The overlap of the
scattering curves measured with different particle fillings shows that the same
size distribution is present at all Fe3 O4 concentrations being investigated. The
distribution originates either from the size distribution of individual particles or
from agglomerates that cannot be broken up by the homogenization procedure
and which are present at all particle fillings. Compared with the size distribution
of the particles measured by TEM, the first peak fits very well to the individual
particles size distribution, whereas the second hump at 40–50 nm corresponds to
existing agglomerates.
53
5 Evaluation of composite properties
XDC Analysis
To investigate if the particles already have an agglomerated state in the initial nan-
oparticle dispersion (magnetite particles dispersed in the solvent γ-Butyrolacton,
GBL) X-ray disc centrifuge (XDC) analysis have been performed. The suspen-
sion is filled into a hollow disc and rotated at high speed to force centrifugal
sedimentation of the particles. An X-ray is used to detect the particle sediment-
ation. Larger particles have fast sedimentation, whereas smaller particles have
slow sedimentation. Particle/agglomerate sizes > 10 nm can be detected with
this method.
Experimental
The hydrodynamic diameter of the initial particle dispersion is measured with
a Brookhaven Instrument X-ray disc centrifuge (3000 rpm, 300 min, 22◦ C). 1.2 ml
particle dispersion is diluted with 23.8 ml GBL and sonicated in a conventional
ultrasonic-bath for 10 minutes. To ensure solvent compatibility a polycarbonat
homolite H-911 disc is used for the XDC measurements.
Results
The measurement of hydrodynamic particle/agglomerate sizes from the X-ray
disc centrifuge (XDC) depicted in Figure 5.6 shows the diameter distribution of
the particles in the initial nanoparticle dispersion before mixing with the photo-
sensitive polymer SU-8. XDC measures hydrodynamic diameters of agglomer-
ates including the surfactant layer and the fluid boundary layer [103]. Therefore,
the sizes are expected to be larger compared to TEM and SAXS. The measure-
ment shows two peaks: One at 19 nm, which corresponds to the single particle
diameter or agglomerates with low primary particle number; and the second
around 38 nm (30–45 nm), which matches with the agglomerate size measured
from SAXS. This indicates that agglomerates that cannot be broken by the used
mixing methods are present in the initial magnetic suspension.
XRD Analysis
The XRD measurements described before are used to determine the crystallite
sizes of the particles which have been mixed into the polymer. By the funda-
mental parameter approach with the Rietveld refinement [104], crystallite sizes
of Fe3 O4 (ICSD 028664) can be determined using TOPAS 3.0 (Bruker). The aver-
age crystallite size of Fe3 O4 (ICSD 028664) measured from the main peak (311)
(34.6◦ - 36.6◦ ) results in 13.1 ± 0.5 nm.
54
5.1 Particle dispersion in composite
0 .2
0 .0
1 0 3 0 6 0 9 0 1 2 0 1 5 0 1 8 0
E q u iv a le n t s p h e r ic a l d ia m e te r [n m ]
Figure 5.6: Differential volume distribution of particle and agglomerate in initial particle
dispersion (solvent: GBL) measured with XDC (surfactant layer contributes to the
sizes). Measurement range: > 10 nm.
Particle and agglomerate size measurements by SAXS, XRD and XDC provide
information on the diameter in terms of a volume or mass distribution, whereas
TEM measurements result in a count size distribution. For comparison the count
average diameter of the TEM measured particles (11.4 ± 3.4 nm) was translated
by calculation into the volume average diameter resulting in 12.4 nm. The volume
average diameter of the agglomerates is 52 nm. The comparison of particle dia-
meters measured by SAXS, XRD and TEM methods are discussed in detail in
[105]. The different particle and agglomerate sizes are summarized in Table 5.1.
The crystal size measured by XRD is in good agreement with the particle dia-
meter measured in the composite by TEM and SAXS. Considering the hydro-
dynamic layer, a slightly larger diameter from the XDC measurements is expec-
ted and corresponds to the particle diameters determined by the other methods
as well. The volume average agglomerate sizes measured by TEM match the
main agglomerate sizes of the SAXS measurements. The comparison between
SAXS measurements of the composite and the XDC measurements from the ini-
tial particle dispersion shows that some agglomerates are already present in the
initial dispersion. The main agglomerate size during composite mixing does not
55
5 Evaluation of composite properties
56
5.2 Limitation using UV exposure
A possibility to find out the maximum polymerized thickness, which can be ob-
tained by UV exposure, is the backside exposure of a thick spin coated MPC layer
through a UV transparent substrate. Due to the UV absorption of the particles in
the composite only the first part of the composite material is sufficiently exposed
for polymerization. During development, the not sufficient exposed composite
material is dissolved. The final layer thickness depends on the exposure dose
used. Here, the maximal layer thickness of a 2 vol.% composite dependent on
different exposure doses is studied. Furthermore, the layers’ surface roughness
is investigated.
Experimental
57
5 Evaluation of composite properties
Table 5.2: Film thicknesses from a 2 vol.% composite with different exposure dose using
backside exposure.
Exposure dose Thickness Stdev
[J/cm2 ] [µm] [µm]
0.1 0 (peeled off)
0.5 1.53 0.19
1 3.42 0.18
5 6.48 0.21
exposed film thickness increases. Figure 5.7 shows the top surface of a film
with bottom-side exposure with 5 J/cm2 after the developing step. From the
28.7 µm thick spin coated layer a fully exposed layer of 6.5 µm remains. The sur-
face of this layer has a roughness of 294 ± 10 nm (RMS) (This is much rougher
compared to a surface of spin coated MPC with top-side exposure: 1.4 ± 0.9 nm
(RMS)). A rough surface can be an advantage for several applications like max-
imizing surface area for drug delivery, for obtaining hydrophobic surfaces (low
contact area), or for increasing cell attachment of muscle cells. However, such
a rough surface is not desired for reliable mechanical microstructures like canti-
levers. To fabricate microstructures with a smooth top and bottom surfaces, a
desired layer thickness has to be spin coated and exposed from the top. The ex-
posure dose has to be adjusted to achieve a full polymerization of the spin coated
layer. In the next section the UV absorption in the MPC and the necessary UV
dose for a determined layer thickness using top-exposure are investigated.
Experimental
58
5.2 Limitation using UV exposure
Figure 5.7: SEM image of MPC layer with 2 vol.% particle concentration exposed with
5 J/cm2 through a fused silica substrate. The exposed 6.5 µm thick layer has a rough
top surface.
composite absorb in this area and can hinder polymerization, creating a limit in
particle concentration. The spectra of the transmittance of composite layers with
different particle concentrations (1, 2, 3, 5 and 10 vol.%) have been measured for
samples with approximately the same layer thickness of 1.6 µm. The thickness of
the samples considered was 1.65 ± 0.15 µm, except for the 10 vol.% case, which
was 2.2 µm, see Table 5.3. The light transmittance was carried out with a UV /VIS
spectrometer (Cary 500, Varian). Exposed composite areas have been released
from the substrate and placed on fused silica glass supports. Fused silica has
a transmittance of 90% above 200 nm and, therefore, it is suitable as a substrate
for transmittance measurements. The wavelength dependent absorption of the
fused silica substrate was eliminated by a background measurement.
Microcantilevers with different particle concentrations, using the same para-
meters as for the UV transmittance test, were fabricated. The top and bottom
surfaces of these microcantilevers were observed by SEM (FEI Quanta 200 FEG)
to investigate the influence of the UV dose on the fabrication of microstructures.
59
5 Evaluation of composite properties
U V - T r a n s m itta n c e o f c o m p o s ite
i- lin e h - lin e
1 0 0
T r a n s m itta n c e ( % ) 8 0
0 v o l.%
6 0
1 v o l.%
2 v o l.%
4 0 3 v o l.%
5 v o l.%
1 0 v o l.%
2 0
2
s u c c e s s f u l w ith 1 0 J /c m
2
n o t s u c c e s s f u l w ith 1 0 J /c m
0
3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0
W a v e le n g th ( n m )
layers when the thickness of the film is a multiple of half a wavelength and the
layer thickness is on the order of the wavelength. For the exposure of the compos-
ite a mercury lamp with 300 W (no filter) is used. The mercury lamp possesses
two main spectral peaks at 365 and 405 nm which are mainly responsible for
the polymerization reaction in the epoxy. Higher nanoparticle content leads to
a higher absorption in the UV region. Therefore, for a given layer thickness ex-
posure doses for full polymerization must be increased with increasing particle
concentration. The strong absorption in the UV region shows the necessity of
higher exposure doses for full polymerization of the composite layers. The UV
absorption of the Fe3 O4 particles inhibits the crosslinking reactions in lower re-
gions of the composite layer during exposure. Insufficiently exposed composite
is removed during the developing process of the structures, and leads to holes
and porosity at the bottom surface. This results in inhomogeneous mechanical
properties of the cantilevers. Larger agglomerates in the composite can shield
the UV light [95]. Therefore, a low agglomerate size and homogeneous disper-
sion of particles in the composite are important.
Figure 5.9 shows SEM images of the tip of fabricated magnetic composite canti-
60
5.2 Limitation using UV exposure
levers with different particle concentrations. Figure 5.9 (a) shows the top and Fig-
ure 5.9 (b) the bottom side of a fabricated cantilever tip containing 5 vol.% mag-
netite particles, an exposure dose, D, of 10 J/cm2 , and a thickness, h, of 1.8 µm.
The UV transmittance of the composite layer is 8% at 365 nm and 21% at 405 nm
as depicted in Figure 5.9 (b). Changing the particle concentration of the compos-
ite to 10 vol.% leads to a smooth top layer but to a rough and porous bottom
layer (Figure 5.9 (c) and (d)). The rough bottom layer is caused by insufficient
exposure dose because of UV absorption by the filler particles (transmittance of
the composite layer is 0.3% at 365 nm and 2.4% at 405 nm). A partially exposed
bottom layer results in peeling of the structure during the baking process. Also,
with increasing the exposure dose to 20 J/cm2 it was not possible to ensure a full
polymerization of the cantilevers with 10 vol.% nanoparticles. An exposure dose
of 10 J/cm2 corresponds to an approximately 20 min exposure time.
The evaluated exposure doses for the composite with different particle concen-
tration which ensure a full polymerization are listed in Table 5.3. Using back-
side exposure for a 2 vol.% composite an exposure dose of 0.5 J/cm2 leads to a
thickness of around 1.5 µm (Table 5.2). To achieve a smooth bottom surface of a
cantilever for a similar thickness (1.8 µm) the exposure dose must be increased to
2 J/cm2 (Table 5.3).
61
5 Evaluation of composite properties
Figure 5.9: SEM image of a composite microcantilever tip exposed with a dose of 10 J/cm2 .
(a) Shows the top surface and (b) the bottom surface of a tip with 5 vol.% Fe3 O4 particle
content. Both surfaces are smooth and fully exposed. A cantilever tip with a concentra-
tion of 10 vol.% Fe3 O4 is shown in (c) presenting the top surface and in (d) the bottom
surface. Due to the high absorption of the Fe3 O4 particles the lower part of the com-
posite cantilever is partially exposed and results in a rough bottom surface ([93],
c
(2011), with permission from Elsevier).
62
5.2 Limitation using UV exposure
The necessary exposure dose to achieve a fully polymerized layer is studied de-
pending on the layer thickness for a composite with a 5 vol.% particle loading.
Experimental
Figure 5.10 shows the transmittance measurements for composites with differ-
ent layer thicknesses. With increasing composite layer thicknesses (h = 2.2 and
3.5 µm), the transmittance of the composite at the two exposure wavelengths
(365 nm and 405 nm) decreases further (Figure 5.10) and the exposure dose for
a full polymerization has to be increased.
Figure 5.11 shows the tips of the fabricated cantilevers. For all thicknesses,
the top surfaces (exposure side) of the cantilevers are smooth, as shown in Fig-
ure 5.11 (a). The bottom surface of cantilevers with a thickness of 1.8 µm and an
exposure dose of 10 J/cm2 is also smooth (Figure 5.11 (b)). On the other hand,
at the bottom surface of cantilevers exposed with an UV dose of only 5 J/cm2
some holes can be observed (Figure 5.11 (c)). For thicker cantilevers the exposure
dose has to be increased. The highest feasible exposure dose applied is 75 J/cm2 ,
which corresponds to an exposure time of around 2h. Cantilever bottom sur-
faces without holes can be fabricated up to a thickness of 2.9 µm (Figure 5.11 (d)).
For thicker structures the bottom surfaces become porous (Figure 5.11 (e)). Some
nanoparticle residues from the dissolved composite remain on the cantilevers as
seen in Figure 5.11 (d) and (e). The sufficient exposure doses for the fabrication
of composite cantilevers with different layer thicknesses are listed in Table 5.4.
5.2.4 Conclusion
To obtain microstructures with a smooth top and bottom surfaces topside expos-
ure must be applied. The UV transmittance measurements have shown that the
63
5 Evaluation of composite properties
i- lin e h - lin e
1 0 0
8 0
T )
T r a n s m itta n c e ( %
6 0
4 0
h = 1 . 9 µm 0 v o l.%
h = 1 . 9 µm 5 v o l.%
2 0 h = 2 . 2 µm 5 v o l.%
h = 3 . 5 µm 5 v o l.%
0
3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0
W a v e le n g th ( n m )
Table 5.4: Exposure doses of the 5 vol.% composite with different layer thicknesses h.
Layer thickness Exposure dose
h ± 0.1 [µm] [J/cm2 ]
Not sufficient Sufficient
1.8 0.95 5 10 20
2.3 10 20 40 60
2.9 50 75 -
3.5 50 75 - -
64
5.2 Limitation using UV exposure
65
5 Evaluation of composite properties
Fe3 O4 particles in the composite strongly absorb light in the UV range. There-
fore, for higher particle concentration in the composite and thicker layers the
exposure dose has to be drastically increased to ensure full polymerization of the
composite layers. To obtain high magnetic forces on the microstructures either
the particle concentration in the composite or the layer thickness should be in-
creased to increase the magnetic active volume. However, there is a trade of
between thickness and particle concentration due to the UV absorbance of the
magnetite nanoparticles. In this work the particle concentration and the thick-
ness have been optimized for the fabrication of microcantilevers. The layer thick-
ness range for cantilevers is limited. Too thin cantilevers lose their static stability.
For too thick cantilevers the mechanical resonance frequency increases and the
mechanical deflection of the cantilevers decreases. A particle concentration of
5 vol.% Fe3 O4 was selected. A fully exposed cantilever can be fabricated with
exposure doses of 10 J/cm2 for 1.8 µm thick layers allowing small variation in
thicknesses. Increasing the exposure dose to 75 J/cm2 , cantilevers with a max-
imum thickness of 2.9 µm can be fabricated.
To determine the resolution of the composite, test structures with different pat-
terns on the mask have been designed.
5.3.1 Experimental
The widths of the fabricated structures and mask patterns have been measured
on three different probes using a Leica DM4000 optical microscope calibrated
with a 70 µm circular standard sample. Figure 5.12, top, shows the mask pat-
tern. The parameters of the composite are: 5 vol.% Fe3 O4 , thickness: 1.8 µm, spin
speed: 5000 rpm, exposure dose: 10 mJ/cm2 .
Figure 5.12, bottom, shows the pattern transfer in the composite. The smallest
structure size on the mask, having a width of 0.8 ± 0.1 µm, is successfully trans-
ferred into the composite with a width of 1.3 ± 0.2 µm. This shows that the resolu-
tion limit of the composite was not reached. The fabricated composite structures
are in general 0.6 ± 0.2 µm wider, most likely resulting from the light dispersion
in the composite by the incorporated nanoparticles. The slight v-shape of the
composite structures can result from the absorption of UV light in the upper part
of the composite layer by the nanoparticles. It is known that bare SU-8 structures
66
5.4 Magnetic characterization
Figure 5.12: The upper picture shows an optical microscope image of the resolution test
pattern on the mask with different slit openings. On the lower picture a SEM view
of the fabricated composite (5 vol.%) pattern is shown. To see the side wall of the
composite structures the fabricated composite sample is tilted by 45◦ . The narrowest
pattern on the mask is 0.8 µm and is successfully transferred to the composite with an
increased width of 1.3 µm. ([93],
c (2011), with permission from Elsevier).
can exhibit a negative slope as well, because of the absorption of deep UV light
in the upper layer. Using a UV filter to remove the sub-365 nm light could help
reduce the effect of the negative slope [106, 107].
5.3.3 Conclusion
Feature sizes with widths of 1.3 µm for a 5 vol.% MPC with a thickness of 1.8 µm
were obtained by conventional photolithography. The resolution limit of the com-
posite was not reached. It should be possible to fabricate even smaller feature
sizes with masks containing smaller patterns.
Magnetic composites can have various magnetic properties. The magnetic prop-
erties of the developed composites mainly depend on the properties of the in-
corporated nanoparticles. However, the matrix material, the dispersion, inter-
particle distances and orientation of the filler can influence the magnetic prop-
erties. Here, we investigate the magnetic characteristics of the developed com-
67
5 Evaluation of composite properties
posite such as saturation magnetization, coercivity and the influence of the filler
concentration on these properties. For the actuation of microstructures made out
of the composite it is important to characterize carefully the magnetic behavior
of the composite. The saturation magnetization of the composite is of interest be-
cause it determines the maximal applied force on a microstructure by an external
magnetic field.
Experimental
Results
68
5.4 Magnetic characterization
3 0 0 .1
[A /m ]
2 7 .9
2 0 0 .0
1 3 .8
M a g n e tiz a tio n , M 1 0 9 .1
-0 .1 6 .5
-0 .1 0 .0 0 .1
2 .7
0
-1 0
1 v o l.%
2 v o l.%
-2 0 3 v o l.%
5 v o l.%
-3 0 1 0 v o l.%
3
×1 0
-8 -6 -4 -2 0 2 4 6 8
5
×1 0 A p p lie d F ie ld , H [A /m ]
Figure 5.13: Magnetization measurements of the composite with different Fe3 O4 particles
concentration at room temperature. The negligible remanent magnetization indicates
superparamagnetic behavior of the composite ([93],
c (2011), with permission from
Elsevier).
3
[A /m ]
0 .2
2 0 .0
-0 .2
1
M a g n e tiz a tio n , M
-0 .3 0 .0 0 .3
0
-1
1 5 0 K
-2 1 0 K
-3
3
×1 0
-3 -2 -1 0 1 2 3
5
×1 0 A p p lie d F ie ld , H [A /m ]
69
5 Evaluation of composite properties
Experimental
Figure 5.15 shows the M-H curve for Fe3 O4 nanoparticles. The particles show
negligible hysteresis (0.5 kA/m), as expected, indicating superparamagnetic be-
havior at room temperature. This small opening of the hysteresis can be an arti-
fact from the measurements or a few blocked larger particles can be present. At
the maximum applied field (800 kA/m) the nanoparticles reach saturation and
magnetization was measured as 277 kA/m (53.5 Am2 /kg). When the magnetiza-
tion values of the different composites are scaled to 100 %, the average saturation
magnetization is found as 291 kA/m. These two values are in good agreement.
Reported saturation magnetization of 60.1 Am2 /kg for Fe3 O4 nanoparticles with
sizes of 11.5 nm are in good agreement with the measurement in this work [108].
The magnetization value of the nanoparticles is significantly lower than the mag-
netization of bulk magnetite, 92 Am2 /kg [50], (maghemite, ∼ 80 Am2 /kg [109]).
Nanoparticles have a reduced saturation magnetization compared to bulk val-
ues due to surface effects (spin surface disorders) and impurities [55, 58] (Section
2.1.4).
To obtain information about the structure of the nanoparticles and possible im-
purities x-ray diffraction (XRD) measurements were taken.
Experimental
For the XRD measurements particles, which have been dried from the initial
particle dispersion in a vacuum oven, were used. To avoid changes in the struc-
ture of the particles inert gas, nitrogen, was used during solvent evaporation.
70
5.4 Magnetic characterization
3
0 .1
5 0
[A /m ]
2
0 .0
2 5
1
M a g n e tiz a tio n , M
-0 .1
[e m u /g ]
-0 .1 0 .0 0 .1
0 0
-1
M
F e 3O 4
-2 5
-2 p a r tic le s
-5 0
5
×1 0
-3
-8 -6 -4 -2 0 2 4 6 8
5
×1 0 A p p lie d F ie ld , H [A /m ]
XRD patterns were carried out from three probes with a Bruker D8 Advance dif-
fractometer (40 kV, 40 mA, CuKα) for a range of 2θ XRD = 10 – 80◦ .
The results of the XRD measurements are shown in Figure 5.16. The XRD-spectra
fits to magnetite Fe3 O4 (ICSD 028664). The main diffraction peaks are indicated
in the graph. The spectra of magnetite Fe3 O4 and maghemite γ-Fe2 O3 (ICSD
87119) are very similar. Despite the lack of the additional peaks of maghemite
(210) at 23.7◦ and (211) at 26.1◦ , the presence of maghemite cannot be completely
excluded. Fe3 O4 particles (magnetite) can oxidize to γ-Fe2 O3 (maghemite) [110].
Therefore, it is possible that the particle (the outer layer) contains some γ-Fe2 O3 .
This would lead to a lower saturation magnetization. However, a high content of
maghemite can be ruled out because pure maghemite nanoparticles with similar
diameters (23 nm) show a magnetization of only 34 Am2 /kg [48]. From the XRD-
spectra other iron oxide phases such as wustite FeO (ICSD 82233) and hematite
α-Fe2 O3 (ICSD 066756) can be excluded.
71
5 Evaluation of composite properties
In te n s ity [a .u .]
F e O n a n o p a r t i c l e s
3 4
1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0
2 Θ [ ° ]
Figure 5.16: XRD pattern of the magnetite (Fe3 O4 ) nanoparticles. The magnetite peak po-
sitions (ICSD 028664) are indicated ([93],
c (2011), with permission from Elsevier).
1 The forces of the composite induced by an external magnetic field are calculated
F = µ0 V ( M · ∇) H (5.2)
where µ0 [N/A2 ] is the magnetic permeability of free space, V [m3 ] is the volume
of the composite, M [A/m] is the magnetization of the composite, H [A/m] is
the magnetic field, and ∇ is the gradient operator. The magnetization of the com-
posite depends on the applied field, the magnetic characteristics of nanoparticles,
and the particle-loading level. The coil is placed underneath the composite to
have the magnetic field parallel to the axis of the coil (in the z direction). The
z-component of the magnetic force is
1 The measurements in this section have been performed in cooperation with Olgaç Ergeneman (In-
stitute of Robotics and Intelligent Systems, ETH Zurich, Switzerland) and are published in [111].
72
5.4 Magnetic characterization
Figure 5.17: A film of MPC with 3 vol.% particle loading is placed 200 µm above a water-
cooled electromagnet and the current is varied to change the force on the film. A
micro-force sensor (Femtotools GmbH) is used to measure the force ([111],
c (2009)
IEEE).
∂
Fz = µ0 V Mz · H (5.3)
∂z
The force on the composite is induced by the magnetic gradient ∇ H and the
magnetization M of the composite, which in turn is related to the magnetic field
H.
Experimental
Figure 5.18 shows the measured force of a film with 3 vol.% concentration (dots)
and the calculated force using (5.3) (dashed line). The measured force is in agree-
73
5 Evaluation of composite properties
0.5
0
−1.5 −1 −0.5 0 0.5 1 1.5
Applied Current I (A)
Figure 5.18: The force of a MPC film with 3 vol.% particle loading as a function of applied
current. Experimental data is shown with dots and calculated values is shown with a
dashed line ([111],
c (2009) IEEE).
ment with the calculated force values. For a current of 1 A a force of 1.2 µN is
measured, this results in a force per volume, Fv , of 45.7 · 103 N/m3 .
Using the obtained force value the static deflection of a cantilever can be calcu-
lated. The static deflection of a beam for a distributed load is [59]
q L L4
uL = (5.4)
8YIz
with
F
qL = = Fv hw (5.5)
L
where q L is the force per length, Y the Young’s modulus, 4.4 GPa (value taken
from Young’s modulus measurements Figure 5.23), Iz the moment of inertia (2.14).
For a cantilever with dimensions, length, L, of 200 µm, width, w, of 14 µm, thick-
ness, h, of 3 µm, the static deflection at the tip of the cantilever, u L , for a 3 vol.% MPC
with the described experimental conditions results in 2.8·10−15 m. This result
shows that it is not possible to achieve a useful static deflection of the fabric-
ated cantilevers with the applied magnetic field with a particle concentration of
3 vol.%.
74
5.4 Magnetic characterization
5.4.5 Conclusion
The magnetic measurements show that the composite has superparamagnetic be-
havior at room temperature. The magnetic behavior of the composite can be
controlled by the selection of the incorporated particles. The magnetic character-
istics of the composite (Figure 5.13) and the particles (Figure 5.15) are consistent.
The polymer matrix does not considerably change the magnetic properties. The
slight lower saturation magnetization in the composite can be explained by a pos-
sible diamagnetic contribution of the sample holder during the magnetic meas-
urement by the AGM. The superparamagnetic behavior has to be considered for
the actuation of microstructures fabricated with this composite (Section 6).
The magnetic force which can be generated on a 3 vol.% MPC by an electrical
coil was investigated. A force per volume of 45.7 · 103 N/m3 can be achieved. For
a 3 vol.% MPC microcantilever, the static deflection with the used setup (coil cur-
rent = 1 A) is < 1 nm. To achieve a useful static deformation for microstructures
a polymer with a low Young’s modulus must be selected and a high magnetic
particle loading of the microstructures must be achieved (∼10 – 50 vol.%). MPC
can be favorable compared to bulk magnetic material for static deflections of mi-
crostructures because the Young’s modulus of the composite does not increase
linearly with the nanoparticle concentration [61]. For the MPC discussed in this
work, a polymer for cantilever applications was chosen, which has a relative
high Young’s modulus to obtain stable suspended microstructures, and has a low
creep which is important for resonant applications. The particle concentration is
limited to 5 vol.% for a thickness of 2.9 µm due to fabrication limitation (low
UV transmittance of the nanoparticles, Section 5.2). The microstructures made
from this composite are suitable for applications where low magnetic forces are
sufficient for an actuation. For example for applications like microstructures in
resonant mode or a magnetic control of MPC objects in liquid.
75
5 Evaluation of composite properties
The static deflection and the dynamic behavior of microstructures are defined by
the Young’s modulus of the structure material. For the design and performance
of novel microstructures the Young’s modulus must be known. In this section the
measurement of the Young’s modulus of the MPC composite with different nan-
oparticle concentrations are presented. Microcantilevers were fabricated and the
dynamic Young’s modulus was determined measuring the resonance frequency
of the microcantilevers and using Euler beam theory.
The most common method to evaluate the Young’s modulus for materials is the
tensile test where an increasing force is applied onto a macro sample until it de-
forms or breaks. The mechanical properties of thin films can differ from those of
bulk material [113]. Therefore, different mechanical testing methods have been
developed for the characterization of thin films [114]. Polymer thin films down
to thicknesses of > 15 µm can be measured with a thin film tensile test setup [115].
Thin films with smaller thicknesses as used in this work (3 µm) cannot be meas-
ured with this setup due to handling issues. Another approach to determine the
Young’s modulus of a thin film material is to characterize the mechanical prop-
erties of a microstructure from these films. Microcantilevers are well suited to
measure the Young’s modulus because their mechanical characteristics are de-
scribed by only few parameters, as discussed below. The Young’s modulus of
microcantilevers have been determined by static [114, 116, 117] and dynamic [60]
methods.
For the static method, a force (by an atomic force microscope (AFM) or a pro-
filometer tip) is applied to the cantilever and with the slope of the deflection
versus displacement the spring constant of the cantilever can be determined and
the Young’s modulus (2.24) calculated. The dynamic methods are based on meas-
uring the resonant frequency of the microstructures. Using Euler-Bernoulli beam
approximation the Young’s modulus of a cantilever is determined by its mechan-
ical resonance frequency, the geometrical parameters, and the density (2.18). The
resonance frequency of an actuated cantilever can be measured with a laser Dop-
pler vibrometer, which does not cause any physical damage to soft cantilevers.
It is reported that the Young’s modulus increases with increasing the resonance
frequency of the microstructure and the value differs to static Young’s modulus
values [60]. The microstructures designed in this work will be used in resonant
mode. Therefore, the resonant method for the characterization of the Young’s
modulus of the MPC is selected.
76
5.5 Young’s modulus
In this subsection the influence of the different error sources for the dynamic
Young’s modulus measurements are analyzed and discussed. The error analysis
is important to evaluate which part of the measurement is critical and to see if
the measurement method is reliable.
The boundary conditions for the calculation of the Young’s modulus using
Euler Bernoulli beam theory (Section 2.2.2) must be fulfilled. The amplitudes
of the cantilever tips are < 100 nm. The ratios L/h are > 50, and L/w are > 6 for
all measured cantilevers.
The propagation of the random errors can be calculated by [120]:
v
∂g 2 2
u
∆y = t∑
u
∆xi (5.6)
i
∂xi
Where ∆y is the error of the output, g the function to calculate the output y
77
5 Evaluation of composite properties
Table 5.5: Discussion of error sources in equation 5.6 for the Young’s modulus determina-
tion using Euler-Bernoulli beam theory.
∆xi
∂g
xi Description ∂xi
2
2πL2 f n
ρ The density is calculated using (5.11). The 12 λ2n h
0.05ρ
error due to inhomogeneities in the particle
dispersion, holes and pores in the MPC canti-
levers are assumed to be less than 5% of the
calculated value.
2
2π f n
L The resolution of the used microscope Leica 48ρL3 λ2n h
1 µm
DM4000 to measure the length and the lat-
eral dimensions of the cantilever has a resol-
ution of about 0.3 µm. Together with the un-
certainties of the anchor position the error is
assumed to be less than 1 µm.
2
2πL2
fn The uncertainty of the measured frequency 24ρ f n λ2n h
15 Hz
due to the resolution of the laser Doppler
vibrometer and the imperfect vacuum is as-
sumed to be 15 Hz.
2
ρ 2πL2
h The thickness was measured with a profilo- −24 h3 λ2n
0.05 µm
meter. The difference between the thick-
nesses of the single cantilevers to the meas-
ured mean thickness of each chip is assumed
to be less than 0.05 µm.
from the inputs xi and ∆xi the errors of the single inputs. With this equation
the error sources for the Young’s modulus measurements using Euler-Bernoulli
(2.18) are summarized in Table 5.5.
For a typical cantilever with a length of 50 µm, the Young’s modulus results
in a relative error of 11%. Whereas, for a cantilever with a length of 200 µm the
relative error is 7.7%. It is favorable to measure longer cantilevers.
78
5.5 Young’s modulus
Experimental
The eigenmode analysis of the FEM Simulation Comsol 4.1 was used to solve the
different structure models shown in Figure 5.19. Firstly, a single beam with a
fixed end was simulated Figure 5.19 (a). Secondly, the whole cantilever structure
layer is modeled using a fixed bottom boundary condition simulating the attach-
ment of the cantilever to the anchor structure (Figure 5.19 (b)). In case of a perfect
alignment of the cantilever structure layer to the anchor, the cantilever has a sec-
tion cs of 5 µm, with a fixed bottom side, simulating the anchor attachment. In
case of a misalignment of more than 5 µm, the structure can result in a plate-like
overlap. The case of an overlap of ns = 5 µm is shown in Figure 5.19 (d). The
parameters used for the simulations are given in Table 5.6.
Detailed investigations with SEM images of the cantilever attachment have
shown that some cantilevers do not properly attach to the anchor structure. A
small slit or crack has been observed at the anchor attachment connection for
some cantilevers. Figure 5.20 (a) shows a cantilever with a solid clamping of the
anchor and the cantilever. In contrast, in 5.20 (b) a gap between cantilever and
anchor is present. The gap length of these structures could not be measured (us-
ing non-destructive measurement methods). However, several anchor areas of
cantilevers from neigbouring cantilever arrays have been analyzed by SEM. The
length of the gap are estimated to be smaller than 1 µm. The gap size varies
on the cantilever array and no correlation of the appearance of these gaps were
79
5 Evaluation of composite properties
Figure 5.19: Schematics of the different models of the FEM simulations. The fixed bound-
aries in the images (a, b, c, and d) are marked yellow (bright).
80
5.5 Young’s modulus
Figure 5.20: SEM images of the anchor area from the bottom side of 2 vol.% MPC mi-
crocantilevers. (a) Without a gap, (b) with a gap between cantilever and anchor.
found. Cantilevers made from unfilled SU-8 do not show these defects. The crack
or slit is most probably caused by the different coefficients of thermal expansion
of the composite and the pure SU-8 anchor layer or the gaps are induced after
the drying process, when the cantilevers were picked from the drying tissue. An
opening between the cantilever and the anchor results in a longer actual canti-
lever length and gives a measurement result with a too low Young’s modulus.
The discrepancy in such a case is simulated using the model as depicted in Fig-
ure 5.19 (c) with a cantilever with Lm + 1 µm assuming a gap of 1 µm.
The FEM models (a, b, c, and d) are compared to Euler-Bernoulli beam approx-
imation (2.18). The different simulated eigenfrequencies, dependent on the can-
tilever lengths for the different FEM models and the Euler-Bernoulli beam ap-
proximation, are shown in Figure 5.21. Model (c) and (d) show the highest dis-
crepancies to the Euler-Bernoulli beam approximation. All models approach the
81
5 Evaluation of composite properties
2 5 0
E u le r - B e r n o u lli
a ) S im p le c a n tile v e r
2 0 0 b ) C a n tile v e r s tru c tu re
F re q u e n c y f [k H z ]
c ) C a n tile v e r s t r u c t u r e w i t h 1 µm g a p
d ) P la te - lik e o v e r la p
1 5 0
1 0 0
5 0
0
6 0 8 0 1 0 0 1 2 0 1 4 0 1 6 0 1 8 0 2 0 0
C a n tile v e r le n g th L m
[ µm ]
Figure 5.21: FEM simulation with different models (Figure 5.19) and Euler-Bernoulli ap-
proximation of the resonance frequency of MPC cantilevers with different lengths.
82
5.5 Young’s modulus
c o m p a r e d to E u le r - B e r n o u lli [% ]
0
Y o u n g ’s m o d u lu s b y F E M
-1 0
-2 0
-3 0 E u le r - B e r n o u lli
a ) S im p le c a n tile v e r
b ) C a n tile v e r s tr u c tu r e
-4 0 c ) C a n t i l e v e r s t r u c t u r e w i t h 1 µm g a p
d ) P la te - lik e o v e r la p
-5 0
6 0 8 0 1 0 0 1 2 0 1 4 0 1 6 0 1 8 0 2 0 0
C a n tile v e r le n g th L m
[ µm ]
Figure 5.22: Young’s modulus simulated by different FEM models (Figure 5.19) for can-
tilever lengths of 50–200 µm compared to Young’s modulus calculated with Euler-
Bernoulli approximation.
83
5 Evaluation of composite properties
for cantilever longer than 100 µm the error decreases to < 8%. Furthermore, due
to this gap, there is the possibility that the cantilever can hit the anchor structure
edge during vibration. This would lead to a cantilever with unilateral contact and
the resonance behavior of the beam would change (introducing non-linearities)
[122, 123]. However, the cantilevers excited by thermal noise have deflections in
the range of 1 nm at the tip and the deflections at the cantilever anchor are much
smaller. The measured resonance curves show no manifestation of vibrations
with unilateral contact (distortions, hysteresis, bending of resonance curve). The
influence of the effect of the cantilever hitting the anchor edge seems to be much
smaller than the error due to an additional length caused by a gap (model (c)).
The effect of cantilever hitting the anchor edge is neglected in this work. The
error based on the gap is not taken into account for the error calculation because
only few cantilevers (10%) have a gap, as depicted in Figure 5.20 (b), and for the
Young’s modulus determination several cantilevers (9–25) with the same length
were measured.
5.5.4 Damping
The damping of an oscillating cantilever system is the sum of the intrinsic damp-
ing of the structure and the external damping due to the surrounding medium
1 1 1
= + (5.7)
Q Qmedium Qintrinsic
with Qmedium being the quality factor due to losses from the medium, and Qintrinsic
the sum of the internal damping mechanisms. In air the Q of the total system is
slightly influenced by the air damping, whereas for cantilevers in water the Q of
the total system is mainly determined by the damping of the liquid. The intrinsic
damping itself is the sum of internal damping mechanisms [124],
1 1 1 1 1
= + + + (5.8)
Qintrinsic Qclamp Q TED Qsur f ace Qmat
whereas Qclamp is the Q-factor due to clamping loss, Q TED is the Q-factor due to
thermoelastic damping, Qsur f ace is the Q-factor due to surface loss, and Qmat is
the Q-factor due to the material damping. It has been shown that the Q-factors of
SU-8 cantilevers measured in vacuum, at pressures below 10 Pa, reach a constant
value [124]. The damping by air can then be neglected, and the damping is de-
termined by the intrinsic damping of the cantilever. It is reported that damping
mechanisms such as clamping loss, thermoelastic damping and surface loss have
a minor influence in vacuum for SU-8 cantilevers and the main damping is based
on the material loss [60].
84
5.5 Young’s modulus
For a reliable Young’s modulus measurement the air damping has to be min-
imized because the damping can lead to significant differences between the nat-
ural and the measured resonant frequency. All measurements to determine the
Young’s modulus were performed at pressures below 2.2·10−2 Pa. Using equa-
tion 2.28 and 2.12 the resonant frequency of a damped resonator can be written
as
s
1
ωr = ω0 1− (5.9)
2Q2+1
The cantilever used for the Young’s modulus measurements have a Q-factor > 25
in vacuum. Inserting this minimal Q value in (5.9) results in a maximum discrep-
ancy of
ωr = 0.9996 · ω0 (5.10)
This shows that the resonant frequency differs not substantially from the un-
damped natural frequency, therefore, it can be assumed to be equal to the ei-
genfrequency of the microcantilevers.
Polymers show an intermediate mechanical behavior between elastic solids
and viscous liquids, so called viscoelastic material. The Young’s modulus of a
viscoelastic material can be described by the complex Young’s modulus, where
the real part is the storage Young’s modulus which is in phase with an applied
strain and the imaginary part the loss Young’s modulus, which is out of phase with
an applied strain [125]. Cantilevers made from solid polymers used below their
glass transition temperature posses only a low viscoelastic behavior and can be
approximated by the storage Young’s modulus [60].
85
5 Evaluation of composite properties
link reaction depends on the concentration of the catalyst, which depends on the
UV dosage. Above a certain UV dose a saturation of the catalyst concentration
may occur [126]. For a composite containing light absorbent particles the expos-
ure leads to different UV exposure dose in the composite layer and to different
cross-linking densities in the polymer. When the cantilevers are released from
the substrate, the stress releases and the cantilevers bend. This effect is enhanced
for composites with higher particle concentrations. It may be minimized with
a high UV dose to reach saturation of the catalyst in the whole composite or by
using an exposure from both side (leads to alignment problems). Cantilevers
immersed in water show slightly stronger bending than cantilevers stored in air.
This is probably due to an enhanced relaxation of the internal stress. The bend-
ing and the cause for the stress gradient was not further investigated in this work.
For the Young’s Modulus measurements only cantilevers with minimal bending
were taken, to avoid a frequency change induced by the bending.
Some cantilevers contain residuals of nanoparticles on the surface (Figure 5.11 (d)).
During the development of the composite the unexposed and not crosslinked
material is dissolved. The nanoparticles are released from the matrix and can
contaminate the cross-linked structures. An enhanced cleaning step was used for
the fabrication of the photopatternable composite microcantilevers. In most cases
the residual layers could be dissolved completely by a cleaning step in an ultra-
sonic bath with low power before the release step of the cantilevers. A residual
layer on the surface of the cantilevers can change the stiffness of the cantilever
and thereby increase the resonance frequency. This would lead to a higher meas-
ured Young’s modulus value. Measurements of cantilevers with a residual layer
showed an increased Young’s modulus of up to 15%. Therefore, the cantilevers
have been checked by an optical microscope before the Young’s modulus meas-
urements and only clean ones have been .taken into account
5.5.7 Experimental
The dynamic Young’s modulus of the composites with different particle concen-
trations can be determined by measuring the resonant frequency of the fabricated
composite cantilevers and the Euler-Bernoulli beam (2.17). For each fill grade
9 to 25 cantilevers with the same length from the same device and batch were
measured. Cantilevers with lengths between 100 – 160 µm were tested. Longer
and shorter cantilevers have not taken into account because it is shown that the
Young’s modulus of polymer cantilevers slightly increase with increasing fre-
86
5.5 Young’s modulus
quency [60]. To reduce air damping, the Young’s modulus measurements were
performed under vacuum (4.6 – 22·10−3 Pa). The cantilevers were excited by
thermal noise. The vibrations were measured by a laser Doppler vibrometer
(Polytec GmbH, MSA-500). The signal was transformed into the frequency do-
main by means of a fast Fourier transformation. The frequency spectra were av-
eraged over 300 measurements in order to minimize noise. Additionally, white
noise and 1/f noise were filtered and the square amplitude frequency spectra are
fitted by a Lorentzian curve shape to determine the resonance frequency. The
length, L, of the cantilevers is measured with an optical light microscope Leica
DM4000. The thickness, h, is determined by a profilometer (Tencore P10). The
nanoparticles are assumed to be uniform dispersed and the density can be calcu-
lated as
5.5.8 Results
87
5 Evaluation of composite properties
Table 5.7: Young’s modulus of cantilevers with lengths of 160 µm from different chips
(arrays).
Lengths L Young’s Standard Measured
Modulus deviation cantilevers
µm GPa GPa
Array 1 160 5.08 0.10 22
Array 2 160 4.88 0.13 19
Array 3 160 5.06 0.37 9
5 .5
Y o u n g ’s m o d u lu s ( G P a )
5 .0
4 .5
4 .0
3 .5
0 1 2 3 4 5
P a r tic le c o n c e n tr a tio n (v o l.% )
Figure 5.23: Dynamic Young’s modulus measurements of composite cantilevers with dif-
ferent particle concentration 0, 1, 2, 3, 5 vol.% dependent on the frequency (lengths) of
the cantilevers. The Young’s modulus is extracted from resonant frequency measure-
ment of the cantilevers by a laser-Doppler vibrometer and using Euler-Bernoulli beam
theory. For each fill grade 9 to 25 cantilevers with the same length from the same device
and batch, respectively, are measured. All cantilevers have frequencies between 23 and
47 kHz (lengths: 100 – 160 µm). The standard deviations of the measurements are in-
dicated with the small error bars and the calculated ”measurement error” with the
large error bars. The dashed lines show the upper and the lower boundary of Young’s
modulus using the model of Hashin-Shtrikman ([93],
c (2011), with permission from
Elsevier).
88
5.5 Young’s modulus
89
5 Evaluation of composite properties
5.6.1 Experimental
MPC with different Fe3 O4 nanoparticle concentration (0, 2, 4 vol.%) was spin-
coated with a thickness of ∼250 µm on a glass substrate with a diameter of 3 cm
and baked at 110◦ C for 2 hours. For the heating an induction heating gener-
ator (Cheltenham induction heating LTD) with an output power of 2.0 kW was
used. The hollow one-turn coil was cooled by a liquid cooling system. An al-
ternating current of 400 A peakvalue (283 ARMS ) at a frequency of 245 kHz was
90
5.6 MPC heating with alternating magnetic field
z
y IRMS = 283 A, 245 kHz
x
Cross-section A
Sample holder
Cooled single-turn coil
Figure 5.24: Schematic of the heating setup: A cooled single-turn coil with alternating
magnetic field at 245 kHz (400 A) is used to heat the MPC film. The temperature is
measured at the surface of the spin-coated MPC samples.
5.6.2 Results
Figure 5.26 shows the measured temperature at the surface of the spin-coated
MPC sample with different particle concentration (0, 2, 4 vol.%) in the alternating
magnetic field. The temperature of the 4 vol.% sample raised from 28◦ C up to
1 The simulations have been performed in cooperation with Alberto Sánchez Cebrián, Centre of
Structure Technologies, ETH Zürich.
91
5 Evaluation of composite properties
[A /m ]
3 1 0 0
R M S
3 0 0 0
M a g n e tic fie ld n o r m , H
2 9 0 0
2 8 0 0
2 7 0 0
2 6 0 0
0 5 1 0 1 5 2 0 2 5 3 0
C r o s s - s e c tio n o f M P C film in x - d ir e c tio n [m m ]
Figure 5.25: Magnetic field H at the cross-section A (Figure 5.24) in the MPC sample (x-
direction) simulated by a FEM model.
70◦ C in 90 s (heating rate: 0.5◦ /s). Due to heat convection and radiation the
temperature approaches the thermal equilibrium at around 70◦ C. After switching
the magnetic field off the temperature decays. The temperature increase of the
0 vol.% sample is caused by background heating of the coil.
The experiment shows that the MPC can be remotely heated by an alternating
magnetic field and the generated heat depends on the nanoparticle concentration
in the composite. The heating of a superparamagnetic nanoparticle is described
by the power loss equation [133]:
2π f τe f f
PL = πµ0 χ0 H 2 f (5.12)
1 + (2π f τe f f )2
where µ0 is the permeability of free space (µ0 = 4π·10−7 T·m/A), χ0 is the sus-
ceptibility (dependent on the magnetic field), H the applied magnetic field, f the
field frequency, and τe f f the effective relaxation time of the particle. τe f f depends
on both Néel τN and Brownian τB relaxation losses which can be described by
92
5.6 MPC heating with alternating magnetic field
M a g n e t ic f ie ld o n M a g n e t ic f ie ld o f f
]
7 0
[ ° C
4 v o l.%
2 v o l.%
c e
0 v o l.%
6 0
r f a
s u
5 0
t
a
r e
t u
4 0
r a
e
p
m
3 0
e
T
0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0
T im e [s ]
Figure 5.26: Heating of MPC films with alternating magnetic field at 245 kHz. The tem-
perature is measured at the surface of spin-coated MPC samples with different nano-
particle concentration: 0, 2, 4 vol.%.
1 1 1
= + (5.13)
τe f f τN τB
The time scale of Néel relaxation is described by (2.6) with τ0 the time constant
(τ0 ∼ 10−9 s), Ke f f the anisotropy constant of the particle, V the nanoparticle
volume, k B the Boltzmann constant (1.38·10−23 J/K) and T the temperature. The
timescale of Brownian relaxation loss is
4πηr3H
τB = (5.14)
kB T
with η is the sample viscosity and r H the particle hydrodynamic radius [53]. Be-
cause the nanoparticles are assumed to be fixed in the polymer matrix the vis-
cosity is infinite and the relaxation time of Brownian motion, which is based on
the rotation of the magnetic particles is infinite. Therefore, τe f f depends only on
Néel relaxation in a composite, in contrast, superparamagnetic magnetite nano-
particles in a stable suspension (ferrofluid) were affected by both relaxation pro-
cesses. Even for an uncured polymer resist, where the nanoparticles are not fixed,
93
5 Evaluation of composite properties
The anisotropy constant Ke f f is not known for the composite. Anisotropy con-
stant K (first-order) for bulk Fe3 O4 is 1.35·104 J/m3 [108] and this value has been
used to calculate the heat loss of magnetite particles in a ferrofluid [53]. However,
the anisotropy constant, Ke f f , highly depends on the type of crystal structure, the
degree of crystallinity and size of the nanoparticles. Furthermore, the anisotropy
constant of nanoparticles is influenced by the particle-particle interactions in a
composite (dependent on the particle dispersion and the particle concentration).
The Néel relaxation time (2.6) for the particles used in this work (with a particle
diameter of 13 nm, a temperature of 25◦ C) is calculated using the values Ke f f
1.35·104 J/m3 [108]. A Néel relaxation time of 4.4·10−8 s results. To obtain an
exact Néel relaxation time of the particles in the composite the Ke f f value has to
be determined experimentally. Using equation (5.13) and (5.12) the power loss
per particle depending on the frequency can be calculated (mean magnetic field
of 3000 A/m). Figure 5.27 shows the power loss depending on the applied fre-
quency for a particle diameter of 13 nm using the reported Ke f f value. With
increased frequency the power loss increases. The frequency used for the exper-
iments are in the first part of the curve where the power loss has a square de-
pendence on the frequency. As it can be seen in the inlet of Figure 5.27 increasing
the frequency by factor 2 from 100 kHz to 200 kHz the generated power in the
composite can be increased by factor of 4.
94
5.7 Surface properties
8
1 .2 x 1 0
-3 3
8
K e ff
= 1 .3 5 1 0 J /m
]
1 .0 x 1 0
3
P o w e r l o s s P / χ0 [ W / m
8 .0 x 1 0
7
7
6 .0 x 1 0 4 .0 x 1 0
5
7
4 .0 x 1 0 2 .0 x 1 0
5
7
2 .0 x 1 0 0 .0
0 1 0 0 2 0 0
0 .0
0 2 0 0 0 4 0 0 0 6 0 0 0 8 0 0 0 1 0 0 0 0
F re q u e n c y f [k H z ]
Figure 5.27: Calculated power loss of a magnetite nanoparticle in the MPC dependent on
the applied frequency (particle diameter: 13 nm, magnetic field: 3 kA/m, temperature:
298 K, Ke f f : 1.35·104 J/m3 ).
5.7.1 Experimental
Dynamic contact angle measurements were performed with a Krüss DSA 100
(Krüss, Germany) on samples with 0, 1, 2, 3, 5, 10 vol.% Fe3 O4 . For dynamic con-
tact angle (advancing (θda ) and receding (θdr )) measurements the drop volume
was increased and decreased with a speed of 15 µL/min. This leads to low-rate
contact-angle measurements with advancing contact-line speeds below 0.012 mm/s.
Receding contact-line speeds are slightly higher on strongly pinning surfaces at
around 0.03 mm/s. For the advancing drop, one video with 100 frames and, for
the receding drop, one video with 250 frames was recorded.
95
5 Evaluation of composite properties
5.7.3 Conclusion
The hydrophobicity of all filled MPC samples is very similar to the unfilled MPC
sample. The composite can be used for the same applications and purposes as
normal SU-8.
5.8 Biocompatibility
Polymer composite materials are often used for medical applications [137] be-
cause of their variety of properties and biocompatibility. Pure SU-8 is not con-
sidered fully biocompatible according to ISO 10993, however, investigations show
that toxicity derived from its degradation products is very low [78]. Furthermore,
surface treatments such as oxygen plasma activation exist that enhance cell pro-
liferation on SU-8 [138]. Pure SU-8 polymer is considered suitable for a large
number of BioMEMS applications, such as cell culturing and biosensors [72, 79].
In this section the biocompatibility of the surface of MPC with different nano-
particle concentration is studied by cell proliferation assay with human foreskin
fibroblasts. The investigation of the biocompatibility of the MPC is important to
evaluate the potential of MPC microstructures interfacing biological systems.
5.8.1 Experimental
96
5.8 Biocompatibility
Figure 5.28 shows the result of the cell viability test of NDHFs cultured on sur-
faces of MPC with different nanoparticle concentrations, compared to NDHFs
cultured on control tissue culture polystyrene (TCPS). Cell viability was found
to be over 86% for all conditions tested. This result indicates that an increase in
nanoparticle concentration up to 10 vol.% in the composite does not significantly
influence the viability of human foreskin fibroblasts.
Figure 5.29 shows the results of the live and dead staining experiment. Epi-
fluorescence microscope images display living NDHFs by means of green fluor-
escence cultured on surfaces of MPC with different Fe3 O4 nanoparticle concen-
trations. The images indicates that on all samples NDHFs adhere and proliferate
on the samples which is in agreement with the previous cell viability test.
97
5 Evaluation of composite properties
1 0 0
8 0
C e ll v ia b ility [% ]
6 0
4 0
2 0
0
0 1 2 3 4 5 6 7 8 9 1 0
N a n o p a r tic le c o n c e n tr a tio n in c o m p o s ite [v o l.% ]
100 µm
Figure 5.29: Epi-fluorescence microscope image of living normal dermal human fibro-
blasts (NDHFs) on MPC surfaces with different Fe3 O4 nanoparticle concentrations.
The living cells are green colored and the cell nuclei is blue colored.
98
5.9 Summary and conclusion
5.8.3 Conclusion
The cell viability test and the live and dead staining test show that human fore-
skin fibroblasts proliferate on MPC samples containing 0 to 10 vol.% nanoparticles.
On all samples cell viability of over 86% was monitored. It can be conclud
that MPC and MPC microstructures with nanoparticle concentration of up to
10 vol.% are not toxic to cells (within 24 hours) and can be used for interactions
with biomaterials. For in vivo applications inwith this time span, additional al-
lergy tests for animals and humans are necessary [139]. Furthermore, the nano-
particle absorption by cells [45] has to be studied.
99
5 Evaluation of composite properties
Table 5.8: The most important properties of the composite compared to pure SU-8 are
summarized in this table. All values without a reference are based on measurements
from chapter 5.
100
6 Applications I: MPC
cantilever resonator
Microcantilevers are used as sensors in the field of micro electromechanical sys-
tems (MEMS). They can be operated in static mode or dynamic mode for the
label-free detection of a variety of species like gas molecules and biological mo-
lecules such as DNA, proteins, and antibodies by detecting a change of surface
stress or change of mass [141]. Such microresonators are typically made of sil-
icon covered with a functional polymer layer [142]. Polymer cantilevers can
have several advantages compared to silicon or metal cantilevers. They have
lower Young’s moduli (∼1 – 4 GPa) compared to those of silicon or metals (∼100
– 200 GPa). Therefore, polymer cantilevers can achieve larger deflections for an
applied force. Polymers allow simple fabrication methods such as hot emboss-
ing, injection molding or photolithography. These process methods can make
the final devices more cost-effective than conventional Si-based cantilevers [143].
Furthermore, the surface properties of polymer cantilevers can be easily chemic-
ally engineered if functionalization is required. Polymer cantilevers as sensors
have been reported to detect biomolecules [72, 143, 144]. Due to their softness,
polymer microcantilevers are very sensitive to change in surface stress when ana-
lyte molecules attach to the surfaces [145]. The main disadvantage of polymer
based microcantilevers is their low quality factors (Q-factor) in vacuum (Q = 20
to 100) and in air (Q = 10 to 30) [124] compared to silicon and metal cantilevers
(in air Q = 10 to 1000) [146]. However, in liquids the Q-factors of microcantilevers
of all types are dominated by the surrounding media and rarely show Q-factors
above 10 [146].
In this chapter the performance of the fabricated MPC cantilevers in dynamic
mode is shown. A typical MPC cantilever used for the tests is shown in Figure 6.1.
Firstly, the results of Q-factor investigation by thermally induced vibrations in
different media of MPC cantilevers are presented. Secondly, the investigation
of magnetic actuation of the cantilevers tested by different magnetic actuation
setups and the resonant characteristic of the superparamagnetic microdevices
are highlighted. Furthermore, the results of actuation in air and water using mag-
netic actuation performed with and without a feedback-loop circuit are presen-
ted and the possibility of using such MPC cantilevers for sensing devices is dis-
101
6 Applications I: MPC cantilever resonator
Figure 6.1: SEM image of a MPC cantilever structure with Fe3 O4 nanoparticle concentra-
tion of 5 vol.%.
102
6.1 MPC cantilever resonance characterization by thermally induced vibrations
In this section the quality factor of the MPC cantilevers in vacuum, air and water
are discussed. The frequency spectrum was averaged over 100 measurements in
order to minimize noise. The amplitude was fitted by a Lorentzian curve shape
to determine the resonance frequency and the Q-factor.
In the first measurement 11 MPC cantilevers with 5 vol.% nanoparticle concen-
tration and dimension of 116.2 ± 0.3 µm x 18.1 ± 0.3 µm x 1.89 ± 0.05 µm were
measured in vacuum and in air. 4 cantilevers with the same dimensions were
measured in water.
Figure 6.2 shows typical resonance curves for a cantilever measured in vacuum,
air and water. The resonance frequency of the cantilevers measured in vacuum
is 42.0 ± 0.86 kHz and very close to the calculated value, 42.9 kHz, using (2.17)
with a density of 1416 kg/m3 and a Young’s modulus of 5.1 GPa. Due to air
damping the resonance frequency is shifted in air towards 41.5 ± 0.87 kHz. In
water the resonance frequency is shifted down to 9.28 ± 0.15 kHz. The Q-factor
in vacuum has a mean value of 35.3 ± 0.7 (standard deviation) and 15.1 ± 0.2
in air. The Q-factor in air is similar to the Q-factors of ∼16 and 15.6 measured
for pure SU-8 cantilevers with similar geometries published by [124] and [144],
respectively. In water the Q-factor is reduced to 1.27 ± 0.05. For pure SU-8 can-
tilevers Q-factors with similar geometries in water around 1 at 75 kHz were de-
termined [147]. These reported Q-factors are lower due to the low distance to the
substrate, 3 µm, resulting in an additional squeeze film damping. In comparison
the cantilevers investigated here have a distance of 90 µm to the substrate. For
silicon nitride cantilevers a Q-factor in water of 1.8 at 13 kHz [148] and for silicon
cantilevers a Q-factor of 23 at 220 kHz [149] are reported.
In a second experiment, the resonance behavior of cantilevers with mean lengths
of 61 µm, 112 µm, and 189 µm in water excited by thermal noise are investigated.
Figure 6.3 presents the resonance behavior of typical cantilevers with thicknesses
103
6 Applications I: MPC cantilever resonator
1 2
in v a c u u m Q = 3 5
in a ir Q = 1 6
1 0 in w a te r Q = 1 .3
D e fle c tio n [p m ]
0
1 0 2 0 3 0 4 0 5 0 6 0 7 0
F re q u e n c y f [k H z ]
Figure 6.2: Resonance curves of a MPC cantilever (with 5 vol.% nanoparticle loading) with
dimensions of 116 µm x 18 µm x 1.89 µm in vacuum (10−3 Pa), in air (atmospheric pres-
sure) and in water actuated by thermal noise. The resonance frequency shifts slightly
from 41.93 kHz (vacuum) to 41.27 kHz (air) to 9.1 kHz (water) and the Q-factor de-
creases from 35 to 16 and to 1.3 due to air and water damping, respectively.
of 2.90 ± 0.03 µm and widths of 17 ± 0.4 µm. When the cantilever length is in-
creased the resonance frequency and the Q-factor decreases. The deflection amp-
litude at the tip of the cantilevers for longer cantilevers are higher due to the
lower spring constants.
Firstly, the resonance frequencies in water, f f luid , of the measured MPC can-
tilever with 5 vol.% nanoparticle concentration were calculated by the inviscid
approximation by Chu described in [150], where the fluid is assumed to be in-
compressible.
f vac
f f luid−Chu = q πρ f luid w
(6.1)
1+ 4ρc h
where ρ f luid is the fluid density of water (997 kg/m3 ) and ρc is the composite
density (1416 kg/m3 , Table 5.8). The frequency in vacuum f vac can be obtained
from the calculation with (2.17) using a Young’s modulus of 5.1 GPa (Table 5.8).
( f vac could not be measured with the same samples because the cantilevers were
104
6.1 MPC cantilever resonance characterization by thermally induced vibrations
6
1 9 0 µm , Q = 1 .2 , fr = 6 .1 k H z
1 1 2 µm , Q = 1 .7 , fr = 2 2 .9 k H z
6 1 µm , Q = 2 .6 , fr = 9 0 .8 k H z
4
D e fle c tio n [p m ]
0
0 5 0 1 0 0 1 5 0
F re q u e n c y f [k H z ]
Figure 6.3: Resonance curves of typical MPC cantilevers (with 5 vol.% nanoparticle load-
ing) thicknesses of 2.90 ± 0.03 µm and widths of 17 ± 0.4 µm and with different lengths
of 61 µm, 112 µm and 190 µm in water excited by thermally induced vibrations. Longer
cantilevers show lower Q-factors and resonance frequencies.
directly released from the sacrificial layer in the fluid chip during fabrication).
Secondly, the resonance frequency and Q-factors in water were calculated us-
ing the model of Sader [150], where viscous effects of the fluid are taken into ac-
count. The restrictions for the model are that L w, the amplitude of vibration
is small, and the fluid is incompressible, which is the case for the used samples.
The Q-factor for the first mode can be determined by
4m L
πρ f luid w2
+ Γr ( ω )
QSader = (6.2)
Γi ( ω )
where Γr and Γi represents the real and imaginary part of the hydrodynamic in-
teraction function and m L is the mass per unit length of the beam. The resonance
frequency is described by
f vac
f f luid−Sader = r (6.3)
πρ f luid w2
1+ 4m L Γr ( ω )
105
6 Applications I: MPC cantilever resonator
Table 6.2: Comparison of measured and calculated resonance frequencies in water for
61 µm, 112 µm, and 189 µm long 5 vol.% MPC cantilevers. The measurements of three
individual cantilevers are shown in Figure 6.3
Length f r measured f f luid−Sader f f luid−Chu Qmeasured QSader
µm kHz kHz khz - -
60.8 ± 0.4 90.76 ± 0.17 101.2 116.8 2.7 ± 0.1 3.8
112.1 ± 0.4 22.73 ± 0.18 27.4 34.4 1.7 ± 0.1 2.5
189.0 ± 0.8 6.00 ± 0.09 8.6 12.1 1.1 ± 0.1 2.0
where Re = ρ f luid ωw2 /4η. Ω(ω ) is a correction factor for rectangular cross sec-
tions. K0 and K1 are modified Bessel functions of the third kind. η is the dynamic
viscosity of water (0.93 10−3 Pa s). The detailed calculation is shown in [150, 151].
The same parameters as for the calculation of f f luid−Chu where chosen.
The results of the calculations and measurements are compared in Table 6.2.
The calculated resonance frequencies in water using the model of Sader (6.3)
are close to the measured resonance frequencies. Due to the neglected viscos-
ity effects the resonance frequencies using the model of Chu (6.1) gives higher
resonance frequency values. The Q-factors in water calculated with (6.2) give
slightly higher values. However, they are in the expected range of the measured
Q-factors. A decrease in Q-factor for longer cantilevers is expected with higher
damping because of larger cantilever surfaces.
106
6.2 Magnetic actuation of superparamagnetic MPC cantilevers
top surface. Rayleigh-Ritz equation (2.25) can be used to calculate the frequency
shift for a beam with an additional end mass. Using composite parameters from
Table 5.8 the resonance cantilever in the unloaded case is 43.7 kHz and is in agree-
ment with the measured value 43.42 ± 0.14 kHz. For a frequency shift of 6.2 kHz
the additional end mass can be calculated and is 0.5 ng. The calculated value is
reasonable and is lower than the measured gold mass (1 ng) as expected because
it is distributed over the end of the tip, which corresponds to a lower tip end
mass.
w ith o u t lo a d
5 w ith lo a d
D e fle c tio n [p m ]
1
2 5 3 0 3 5 4 0 4 5 5 0 5 5 6 0
F re q u e n c y f [k H z ]
Figure 6.4: Frequency shift of resonance curve of a MPC cantilever measured in air (with
5 vol.% nanoparticle loading and dimensions of 116 µm x 18 µm x 1.89 µm) due to ad-
ditional mass. A gold layer was sputtered at the cantilever tip. The cantilever was
excited by thermal noise.
1 Parts of this section have been published in [87] and [152]. The magnetic actuation of the MPC
cantilevers were accomplished in cooperation with Olgaç Ergeneman (Institute of Robotics and
Intelligent Systems, ETH Zurich, Switzerland). The simulation, calculation, fabrication and charac-
terization of the actuation coils are discussed in [152].
107
6 Applications I: MPC cantilever resonator
Experimental
To actuate the MPC cantilevers, they were placed above an electrical coil as illus-
trated in Figure 6.5. The alternating magnetic field H generated by the electro-
magnet (coil) is proportional to the sinusoidal current in the coil. The deflections
of the cantilevers were measured with a laser-Doppler vibrometer. To generate
high magnetic fields and field gradients between 10 kHz and 100 kHz, the electro-
magnets were optimized by finite element simulations and analytical calculations
([152] Chapter 5). Losses in the coil such as skin effect and inductance of the coil
were considered.
The magnetic field of a typical actuation coil was investigated. Figure 6.6
shows the magnetic field Hz of an excitation coil at a current density of J =
3.6·106 A/m2 (RMS) with coil dimension of rin : 2.75 mm, rout : 12.75 mm, hCoil :
10 mm. The measurement shows that the generated magnetic field decreases rap-
idly with distance from the coil. Therefore, cantilevers must be placed as close as
possible to the coil while avoiding contact. Contact would transfer mechanical
vibration from the coil to the microstructure. The cantilever was placed approx-
imately 2 mm above the coil. A z-direction magnetic field Hz of 10 kA/m and a
z-direction field gradient of -4.05·106 A/m2 were measured at this position.
For the cantilever actuation different coil designs were used, however, the mag-
netic fields and gradients are in the same range.
Results
108
6.2 Magnetic actuation of superparamagnetic MPC cantilevers
z
y laser-Doppler
x vibrometer
cantilever-chip
hCoil
A
rin HCoil ~ ICoil
rout
Figure 6.5: Schematic illustration of the measurement setup. The magnetic composite can-
tilever is actuated by the alternating magnetic field, HCoil , of an external coil. The
resonant frequency of the cantilever is measured by a laser-Doppler vibrometer ([87],
c IOP).
obtained. The cantilever is excited more significantly at two times the magnetic
excitation frequency.
The cantilevers used in this measurements are from an earlier fabrication gener-
ation where the full polymerization of the cantilever is not ensured. The Q-factor
of this cantilevers show lower values (around 7 in air) than for cantilevers used
in all the other experiments (around 16 in air).
Discussion
The reason for the higher amplitude at the double frequency is based on the su-
perparamagnetic behavior of the MPC. The force, F, acting on the magnetic com-
posite can be calculated using (5.2). The force on the superparamagnetic compos-
ite is induced by the magnetic gradient, ∇ H, and the magnetization, M, of the
composite which in turn is related to the magnetic field, H. The magnetization,
M, of the composite depends on the applied field, magnetic characteristics of
the incorporated nanoparticles, particle concentration, and the shape of the can-
tilever. From the equation (5.3) we can see that the magnetic forces experienced
by the composite can only be attractive since magnetic field, H, and magnetiza-
tion, M, change sign simultaneously. H and M thus make the equation always
positive (positive is used for attractive force). This behavior has an important im-
109
6 Applications I: MPC cantilever resonator
Figure 6.6: Magnetic field of excitation coil at r = 0 along z-direction. The current density
is J = 3.6 · 106 A/m2 (RMS) and the coil dimension are rin = 2.75 mm, rout = 12.75 mm,
hCoil = 10 mm. The resonator structure is located at z = 0 mm. The excitation coil is
located at z = -12 mm to -2 mm. [152]
110
6.2 Magnetic actuation of superparamagnetic MPC cantilevers
Amplitude [nm]
2
0
5 10 15 20
Frequency [kHz]
180
Phase [degrees]
-180
5 10 15 20
Frequency [kHz]
Figure 6.7: Laser-Doppler vibrometer measurements of the deflection amplitude at the tip
of the composite cantilever in air. The cantilever is actuated by alternating magnetic
field with a sweep from 5 to 20 kHz and the mechanical resonance is measured at the
same frequency as the actuation frequency. [152]
coil (skin effect). With an additional uniform magnetic field the magnetization of
the cantilever can be enhanced and the force and the deflection of cantilevers is
increased.
Experimental
Figure 6.9 shows the excitation setup used for this experiment. The cantilevers
were placed in a custom made vacuum chamber and excited at the mechanical
resonance using a custom electromagnet. In addition to the AC electromagnet a
Helmholtz pair was placed around the vacuum chamber to apply an additional
DC field. The Helmholtz coil generates a uniform magnetic field with almost
negligible field gradient, so it does not induce a static magnetic force on the canti-
levers. The applied magnetic field at the cantilever was measured for the AC and
DC coils (AC + DC-setup) as 5.18 kA/m (RMS), 27.85 kA/m, respectively. The
tip deflection measurements were performed by a LDV at reduced pressure. The
pressure could not be determined in the chamber because of lack of space for a
pressure gauge.
111
6 Applications I: MPC cantilever resonator
40
Amplitude [nm]
20
0
5 10 15 20
Frequency [kHz]
200
Phase [degrees]
-200
5 10 15 20
Frequency [kHz]
Figure 6.8: Laser-Doppler vibrometer measurements of the deflection amplitude at the tip
of the composite cantilever in air. The cantilever is actuated by alternating magnetic
field with a sweep from 5 to 20 kHz and the mechanical frequency of the cantilever is
recorded at the 2nd harmonic (double frequency). The actuation frequency is depicted
in the x-axis. The cantilever resonates mechanically at 14 kHz when the magnetic
excitation frequency is at half (7 kHz). [87]
112
6.2 Magnetic actuation of superparamagnetic MPC cantilevers
Figure 6.9: Schematic of the setup for the enhanced actuation of MPC cantilevers actuated
by a alternating magnetic field with an additional homogeneous magnetic field gener-
ated by Helmholtz coils ([111],
c (2009) IEEE).
Table 6.3: Deflection of the cantilever in resonance with different magnetic actuation
modes (at reduced pressure).
Excitation Deflection [nm]
Thermal noise 0.1
AC 5.6
AC + DC 63.3
113
6 Applications I: MPC cantilever resonator
Experimental
φ
Feedback Read Out Signal 2+&3456%
789%+.
ω
Switch
-
Input Output
!"#$%&%$'(
+ )'*+,
-./0$1$+,
Figure 6.10: Schematic illustration of the feedback loop to enhance the Q-factor of the mi-
croresonators. [152]
114
6.2 Magnetic actuation of superparamagnetic MPC cantilevers
shows higher amplitude than the cantilever excited by thermal noise because of
the higher actuation forces. However, the Q-factors are similar, as expected. In
closed loop condition the Q-factor increases to 29. This measurement shows that
the Q-factors of MPC cantilevers can be significantly improved by a feedback
system. The Q-factor of the MPC cantilevers is lower than the Q-factor of metal
cantilevers. Q-factors of CoNi cantilevers (205 µm x 15 µm x 4.5 µm) measured
with the same setup results in Q-factors of 550 (open loop) and 1300 (close loop)
[152]. However, the Q-factor cannot be directly compared because of the different
cantilever geometries.
6
c lo s e d lo o p , Q = 2 9
5 o p e n lo o p , Q = 1 7
th e rm a l n o is e , d e f le c tio n
s c a le d x 4 0 , Q = 1 6
4
D e fle c tio n [n m ]
0
1 0 1 5 2 0 2 5 3 0
F re q u e n c y f [k H z ]
Figure 6.11: Resonance behavior of a 5 vol.% MPC cantilever in air with thermal noise ex-
citation, and magnetic actuation (AC-setup) with (closed loop) and without feedback
loop (open loop).
115
6 Applications I: MPC cantilever resonator
timated by taking the results from Section 5.6. The heating of the MPC depends
on the frequency and the magnetic field (5.12). For the magnetic actuation of
cantilevers, a magnetic field of ∼10 kA/m and frequencies < 50 kHz were used.
The magnetic field of a 4 vol.% composite in the heating experiments is 3 kA/m
(RMS) at a frequency of 245 kHz. The power generated in the composite is a
quadratic function of the frequency. The 4.9 times lower frequency results in a 24
times lower power (see inlet in Figure 5.27). The heating power in the compos-
ite is proportional to the magnetic field and the particle volume concentration.
The magnetic field used for the cantilever actuation is maximum 3 times higher
(< 10 kA/m) and the volume concentration is a factor 1.25 higher. The final gen-
erated power in the composite in the cantilever actuation setup is at least a factor
6.4 lower. A maximum temperature increase of < 10◦ C in 90s is estimated. The
sweep time as performed in the measurements discussed here is approximately
60 s. Additionally, decreasing the structure size increases the surface/volume ra-
tio and the heat dissipation due to heat convection and radiation increases. For
cantilever measurements in water the temperature increase is attenuated because
of the high heat capacity of water. However, for a more exact evaluation of the
generated temperature, heat simulations with the cantilever geometries must be
performed.
Experimental
116
6.2 Magnetic actuation of superparamagnetic MPC cantilevers
3 0
T h e r m a l a c tu a tio n , M a g n e tic a c tu a tio n ,
D e fle c tio n [p m ]
Q = 1 .9 , Q ~ 1 .7 , fr = 2 0 k H z
2 0 fr = 2 0 k H z
1 0
9 0
-9 0
-1 8 0
1 0 1 5 2 0 2 5 3 0 3 5 4 0
F re q u e n c y f [k H z ]
Figure 6.12: Resonance behavior of a 5 vol.% MPC cantilever in water with magnetic ac-
tuation (AC-setup) without feedback loop. Cantilever dimensions: 112 µm x 17 µm x
2.9 µm.
117
6 Applications I: MPC cantilever resonator
6.3 Conclusion
118
6.3 Conclusion
ating magnetic field with high frequency (245 kHz) is applied. Heat generation
during actuation with the AC setup in MPC cantilevers containing 5 vol.% mag-
netite particles cannot be fully excluded. However, a temperature increase smal-
ler 10◦ C is estimated.
For a proof of concept, gold is sputtered at the tip of MPC cantilevers (∼1 ng).
A resonant frequency shift in air of 6.2 kHz to lower frequency was observed.
This measurement shows that the fabricated MPC microcantilevers can be used
as mass sensors or micro balances. Furthermore, polymer cantilevers are known
to be sensitive to the absorption of gas molecules and surface stress [143]. Poly-
mer cantilevers are candidates for humidity sensing [60] or the detection of dis-
ease markers in human breath for early diagnosis. However, to use the MPC
cantilevers for such a device the readout and the actuation system has to be mini-
aturized and the sensitivity and performance of the device must be tested in de-
tail.
The MPC cantilevers were successfully actuated in water by an alternating
magnetic field. However, the obtained deflection is low and near the noise level,
preventing the use of the feedback loop circuit for Q-enhancement. A possible
solution is to increase the nanoparticle concentration in the composite to obtain
higher magnetic forces. However, because of fabrication limitations a higher nan-
oparticle loading than 5 vol.% could not be achieved in this work. The absorption
and scattering of the UV light by the nanoparticles limits the full polymerization
of the composite. Another possibility is to increase the cantilever dimensions to
increase the magnetic active volume of the cantilever. However, in this case an ad-
ditional mass results in lower frequency shifts. The measured Q-factor in water
is approximately 2 and not sufficient for a sensor with high sensitivity. The use
of a Q-enhancement system with self-actuation as reported for cantilever meas-
urements in water [147, 148] have to be investigated more in detail. This could
improve the Q-factor to enable the sensing in water of biomolecule interactions
or the detection of disease markers in human blood. In combination with the de-
veloped all-polymer packaging, this could lead to a low-cost disposable sensor.
119
7 Application II: MPC lateral
resonator
This chapter describes the design and the performance of a MPC in-plane mi-
croresonators. In-plane resonators have the advantage of low damping in viscous
environments because they shear the media compared to out-of-plane movement
of cantilevers [153, 154]. Polymer in-plane resonators benefit from the low mater-
ial density enabling a big resonator surface area (high mass detection area) with
a low total mass of the resonator. With a plate-like MPC resonator, as shown
in Figure 7.1, the magnetic volume at the end of the cantilever can be increased.
This leads to a higher magnetic force on the resonator and higher deflections can
be generated compared to out-of plane cantilevers.
Large deflections of microresonators can enable a magnetic readout of the res-
onance frequency of the microstructure. To detect inductively the resonance fre-
quency of a magnetic resonator by a conventional macroscopic pick-up coil, from
the outside of the sensor package (distance around 2 mm), a high magnetic flux
change must be generated by the resonator. This can be achieved by large deflec-
tions of the resonator. A magnetic readout of a resonant magnetic microstructure
(sensor) needs no electrical contact to the device. Furthermore, it has the ad-
vantage compared to optical readout of detecting the resonance frequency of a
microresonator in an optically opaque media or package. Additionally, a remote
multiplexed readout from resonators with slightly different resonance frequen-
cies can be achieved.
In the following the design and the fabrication of an MPC in-plane microreson-
ator is presented. The resonance behavior, damping and deflections of the reson-
ators are characterized and the use of such structures for a magnetic read-out are
discussed. Furthermore, an optimized magnetic actuation setup is presented.
7.1 Design
Lateral plate resonators were designed and fabricated with various shapes, lis-
ted in Table 7.1. Large plates are favorable for the magnetic actuation and readout
(larger magnetic volume), however, smaller plates reducing the risk of fabrication
121
7 Application II: MPC lateral resonator
Figure 7.1: SEM image of a MPC lateral resonator structure with 5 vol.% Fe3 O4 nano-
particle concentration.
failure due to bending, sticking and breaking of the plate-like structures. In this
chapter only MPC microstructures with a particle concentration of 5 vol.% mag-
netite nanoparticle are investigated.
To investigate the modal performance and design the microstructures, FEM
simulations were performed. To obtain a high magnetic force on the resonator
the resonance frequencies must be lower than 30 kHz. At high frequencies it
is difficult to achieve high magnetic fields because of the induction of the coil.
On the other hand, a higher frequency is favorable to minimize damping. The Q-
factor of a plate-like in-plane structure in air with infinite distance to the substrate
can be described by [155]
1 ωη A D
≈ (7.1)
Qlat δP k
122
7.1 Design
s
2η
δP ( ω ) = (7.2)
ωρ
where ρ is the density of the fluid. A plate oscillating slowly in a viscous medium
is expected to drag substantially more of the fluid ambient compared to a fast
moving one. Therefore,
√
Qlat ∝ ω (7.3)
A small distance (< 15 µm) between a resonating plate and a parallel wall leads
to additional damping. Because of the large distance of 100 µm between the sub-
strate and the plate the additional damping effect by the substrate can be neg-
lected.
The parameters used for the simulations are shown in Table 7.2. The simulation
results are shown in Table 7.3. The first three modes of the microresonator are
listed. The second mode is the desired lateral mode. The first mode (out-of-
plane) and the third mode (torsional) are assumed to be damped in a viscous
environment.
123
7 Application II: MPC lateral resonator
Table 7.3: Simulation results of an in-plane resonator with the parameters from Table 7.2
for dimension of L B = 80.4 µm, w = 3.4 µm, s = 200 µm, and h = 1.88 µm.
Mode number Mode Frequency
Hz
1 vertical 1090
2 lateral 1860
3 torsional 3065
In this section the investigation of the resonance frequency of the in-plane reson-
ators with different plate sizes in vacuum and air by monitoring the resonator
vibrations due to thermal fluctuations are presented.
7.2.1 Experimental
Microresonators with beam length, L B , of 80 µm, beam width, w, of 3.35 µm, and
a structure thickness, h, of 1.88 µm were investigated. The plate length, s, was
varied from 60 to 200 µm. The in-plane MPC microresonators resonate due to
the ambient thermal noise and the resonance frequency can be measured with
a laser-Doppler vibrometer Polytec MSA-400 (LSV). The samples were placed
into a vacuum chamber with a sapphire glass window. The laser of the LSV was
pointed at the sidewall of the vertically placed microresonator plates. At least 4
measurements were taken for each plate size.
The measured frequencies of the lateral mode of MPC in-plane resonators with
different plate sizes in vacuum (at a pressure of < 8.6 Pa) are shown in Figure 7.2.
With increasing plate size, the frequency drops because of the increased mass at
the end of the beam and the overall increased length of the resonant structure.
The damping of polymer microcantilevers by air molecules at pressures < 10 Pa
can be neglected [124]. Therefore, it is assumed that the damping of the measured
in-plane resonators at 8.6 Pa has a minor influence on the resonance frequency.
The measured resonance frequency is compared with the frequency calculation
of an in-plane resonator using the approximation of a beam with loaded end mass
(2.26), where m a is the plate mass. The balance point of the plate is assumed to be
124
7.3 Results and Discussion
1 4
M e a s u re m e n ts in v a c u u m
R e s o n a n c e fre q u e n c y (k H z )
1 2 S im u la tio n re s u lts in v a c u u m
T h e o r e tic a l c a lc u la tio n
o f b e a m w ith e n d m a s s s
1 0 T h e o r e tic a l c a lc u la tio n
o f b e a m w ith e n d m a s s
( u s in g d e s ig n g e o m e tr ie s )
8
s
6
L B = 8 0 µm
4
w = 3 . 3 5 µm
2
0
5 0 1 0 0 1 5 0 2 0 0 2 5 0
P l a t e l e n g t h s ( µm )
Figure 7.2: Comparison of the measured resonance frequencies (triangle filled) of in-plane
MPC microresonators with the calculation of a beam with end mass (2.26) (triangle
empty) and simulations (dots) for various plate lengths. For each plate size the meas-
ured mean dimensions of the microstructures are taken. For the measurements the
standard deviation of at least four measurements are indicated by error bars. For the
simulation results a Young’s modulus of 5.1 ± 0.4 GPa is used (standard deviation val-
ues indicated with bars). The dashed curve shows the calculation for different plate
sizes of a beam with end mass (2.26) for the designed geometries of the microresonat-
ors: beam length, L B , of 80 µm, beam width, w, of 3.35 µm, and a structure thickness,
h, of 1.88 µm.
right in the middle of the plate. L is in this case the distance from the anchor to the
balance point (L B + s/2). For the calculations, the dimensions of the individual
resonators were measured by an optical microscope and a density of 1416 kg/m3
was used. The simplified model leads to calculations that show slightly higher
values than the measured resonance frequencies.
For comparison, simulations with the parameters shown in Table 7.2 were per-
formed. The dimensions of the resonators are individually measured by optical
microscopy. For each plate size the measured mean value is taken for the sim-
ulations. A Young’s modulus of 5.1 ± 0.4 GPa was used. The upper and lower
values from the standard deviation were also simulated. The simulation results
are shown in Figure 7.2. The simulation results show slightly higher values than
125
7 Application II: MPC lateral resonator
the measured ones. The discrepancy between the simulations, using the mean
Young’s modulus 5.1 GPa, and the measured values is between 6 and 11%. Reas-
ons for this discrepancy can be simulation uncertainty, temperature variation, re-
siduals from fabrication on the plate, higher density than expected (influenced
by humidity and temperature), and slightly lower Young’s modulus of resonant
polymer microstructures for lower frequencies as shown by [60].
The resonance frequencies measured in air are only slightly lower than the
ones measured in vacuum. The difference is below the standard deviation of
the measurements. For the largest plate size, s = 200 µm, a Q-factor of 14 was
measured in air as discussed in the next section (Table 7.4). The relationship
between the air and vacuum resonance frequencies is ωr = 0.9987ω0 using the
approximation of (2.28) and (2.12).
In-plane resonators with a large plate size, s = 200 µm, to obtain high mag-
netic actuation forces, with a resonant frequency of approximately 1.6 kHz were
chosen for the magnetic actuation described in the next section.
7.4.1 Experimental
The actuation of superparamagnetic cantilevers have shown that the best mag-
netic actuation performance is obtained when the alternating magnetic field is
combined with a static uniform magnetic field (see Section 6.2.2). The actuation
setup is depicted in Figure 7.3. The microresonator plate was positioned so that
the AC coils are perpendicular to the plate surface. Two permanent magnets
were placed left and right of the resonator to generate a uniform magnetic field
to magnetize the microresonator plate in the in-plane direction as shown in the
inset in Figure 7.3. The two electromagnets generate a gradient in axial and ra-
dial directions of the coil when the magnetic fields of the coils show in opposite
directions. The magnetic force on the microresonator depends on the magnetic
field gradient, ∇ H, generated by the actuation coil, and the magnetization of the
plate, M (5.2). An in-plane force acts on the microresonator due to the in-plane
magnetization of the plate forced by the permanent magnets.
For the magnetically actuated microresonators, the deflection was measured
with a planar motion analyzer from Polytec MSA-400 (PMA). A deflection amp-
litude of > 200 nm is necessary to evaluate the resonance frequency with the PMA.
126
7.4 Magnetic actuation of in-plane resonator
Figure 7.3: Setup for the magnetic actuation of the MPC in-plane microresonators. The
inset shows the magnetization of the superparamagnetic in-plane microresonators.
Table 7.4: Measured dimensions for in-plane resonators actuated by magnetic field.
Beam Length Beam Width Plate Length Thickness
LB w s h
80.4 ± 0.2 µm 3.4 ± 0.1 µm 200.0 ± 0.4 µm 1.88 ± 0.03 µm
Because of the small deflection amplitude (picometer range) for the thermal ac-
tuation of the microresonators, the deflections must be recorded by the laser-
Doppler vibrometer Polytec MSA-400 (LSV) by focusing the laser on the structure
sidewall. The coils were actuated with an alternating current of 0.5 A. This lead
to a magnetic field Hx of 215 kA/m from the permanent magnets and a gradient
∂
of ∂x H of 0.43·106 A/m2 from the coils at the position of the microresonator. 5
in-plane microresonators from the same fabrication batch were tested. Their geo-
metric dimensions, measured by an optical light microscope Leica DM4000, are
listed in Table 7.4. The thickness h is determined by a contact profilometer (Ten-
core P10). The measurements in vacuum were performed at pressures < 10 Pa.
127
7 Application II: MPC lateral resonator
compared to that in vacuum due to air damping. The Q-factor in air is around 14
compared to the Q-factor in vacuum of 54. With magnetic actuation, a lateral de-
flection amplitude of 28 µm and a Q-factor of 14 was measured. The Q-factor and
the resonance frequency are consistent for both magnetic and thermal excitation
mechanisms. Figure 7.4 shows the measurement of one of the tested devices in
air with thermal and magnetic actuation. The Q-factors in air of the in-plane res-
onators are slightly lower (14) than the ones measured with the cantilevers (16)
in air.
3 5
1 4 0
3 0 M a g n e tic a c tu a tio n
[ µm ]
Q = 1 5 , fr = 1 .6 4 k H z 1 2 0
2 5
T h e r m a l a c tu a tio n
[p m ]
1 0 0
m a g n e tic
Q = 1 5 , fr = 1 .6 4 k H z
th e rm a l
2 0
8 0
1 5
D e fle c tio n
D e fle c tio n
6 0
1 0
4 0
5
2 0
0
1 .0 1 .5 2 .0
M e a s u r e d m e c h a n ic a l fr e q u e n c y f [k H z ]
Figure 7.4: Deflection of a superparamagnetic MPC in-plane microresonator with
5 vol.% Fe3 O4 nanoparticle loading excited with ambient thermal noise and magnetic
field. The dimensions of the microstructure are listed in Table 7.4.
128
7.5 Magnetic readout of in-plane resonator
This section shows an evaluation of whether the MPC resonator can be detected
by a magnetic readout system as reported in [152] using an external pick-up coil.
The induced voltage, U p , in an external pick-up coil is proportional to the deflec-
tion of the microresonator, ∆x p , and its magnetization, M. M depends on the
nanoparticle concentration and on the volume
U p ∝ ∆x p M (7.4)
The resonance of CoNi lateral plate resonator with geometrical plate dimensions
of 400 µm x 400 µm x 12.5 µm was successfully detected in air and water [156].
Calculations show that CoNi resonators with dimensions of 300 µm x 300 µm x
2.5 µm can produce a readout amplitude of 67.5 nV. A deflection amplitude of
10 µm, optimal alignment and a distance of the magnetic body to the pick-up coil
of 300 µm were assumed. A minimum distance of 300 µm is necessary because
of the package. For a MPC resonator with 5 vol.% particle concentration with the
same geometrical dimensions, a readout signal of ∼4 nV is calculated. However,
a readout signal of at least 20 nV is necessary for a reliable magnetic readout. The
detailed calculation and simulations are presented in [156].
With the MPC resonators with 5 vol.% particle concentration a plate size of
200 µm x 200 µm x 1.88 µm, a deflection of 28 µm can be achieved. This repres-
ents roughly the situation used in the calculation. The magnetization or the de-
flection of the MPC resonator must be increased by at least a factor 4 to allow
a feasible readout. The magnetization can be increased by increasing the active
magnetic volume. The layer thickness of the MPC resonator can be increased up
to 2.9 µm for a 5 vol.% MPC, as investigated in Section 5.2.3. A plate size increase
up to 400 µm x 400 µm is possible. However, the fabrication yield of structures
of this size decreases due to fabrication limitations (sticking of the structures on
the substrate during drying process and bending of plate due to internal stress
in the composite layer). An increase in particle concentration would be the most
promising option. It would improve as well the deflection due to higher mag-
netic actuation force. However, an increase in the particle concentration results
in a lower processable layer thickness because of the higher UV absorption by
the Fe3 O4 nanoparticles during exposure as described in Section 5.2.2. Moreover,
to increase the deflection of the resonator, the spring constant can be decreased
(increasing the beam length L B and decreasing the beam width w). However,
the chosen dimensions of L B = 80 µm and w = 3.35 µm for plate sizes with s =
200 µm are already at the limit of guaranteeing mechanical stability of the micro-
structure and avoiding fabrication failure. When all the parameters are optim-
129
7 Application II: MPC lateral resonator
ized a magnetic readout could be possible with the in-plane resonator structures
fabricated by the MPC. However, it will be a challenge to achieve stable and re-
producible measurements. The Q-factor is inversely proportional to the dynamic
viscosity (7.1) of the surrounding media. Therefore, a higher damping is expec-
ted in water for the fabricated structures, decreasing the amplitude and making
a readout in water with the reported setup [152] is not feasible.
To make a magnetic readout in air easier, it is suggested to mix additional pho-
toinitiator into the composite [4]. This makes the composite more sensitive to UV-
light during exposure and higher nanoparticle loadings and layer thicknesses can
be obtained.
7.6 Conclusion
The evaluated parameters of the composite (Chapter 5), such as Young’s modu-
lus, density, maximal layer thicknesses depending on the nanoparticle concentra-
tions, were used to design an in-plane resonator structure. The resonator struc-
tures with a 5 vol.% MPC were fabricated and successfully actuated remotely in
air in lateral mode by magnetic fields. A deflection of up 28 µm was obtained and
showed that high deflections can be achieved with microstructures fabricated by
the MPC even with layer thickness of 1.9 µm and nanoparticle concentrations of
5 vol.%. The Q-factors of the in-plane resonators (14) are similar to Q-factors of
cantilevers with out-of-plane motions (16) in air. A magnetic readout could be
possible for MPC resonators with plate sizes of 400 x 400 µm and an increased
thicknesses of 2.9 µm. However, the fabrication of such devices is a fabrication
challenge. To make a magnetic readout measurement more feasible, it is sugges-
ted to fabricate thicker layer structures or higher particle concentrations. This
can possibly be achieved by adding additional photoinitiator into the composite
to make it more sensitive to UV light during exposure.
130
8 Application III: MPC
3D-microstructures by
two-photon polymerization
This chapter presents the fabrication of magnetic polymer three dimensional (3D)
microstructures with the superparamagentic MPC using two-photon polymer-
ization (TPP). Fabrication parameters such as writing speed and laser power,
and limitations such as minimal line resolution depending on the filler concen-
tration were investigated. Different microstructures such as pillars and hollow
cubes were fabricated and demonstrate the fabrication possibilities. Finally, MPC
microspirals were manufactured which can be used as microrobots mimicking
nature bacteria flagellas. Using a uniform rotating magnetic field the MPC hel-
ical microstructures were successfully rotated in water with a cork-screw motion
and can be steered near a solid surface. The direction of the movement can be
controlled by applied magnetic fields.
8.1 Introduction
Due to the fast research progress in the biomedical field, the interest to use remote
control microsystems which enables manipulation and sensing in the microscale
for in vivo and in vitro bioapplications has increased. The combination of differ-
ent disciplines such as material science and microfabrication technology directed
to a great progress in the field of micro- and nanorobots in the recent years [157].
Magnetic actuation allows a remote and wireless control of a magnetic object over
a large distance (several centimeters) in different media. A typical example is the
opthalmic luminescence oxygen sensor on a microrobot, with a size of few milli-
meters, which can be controlled in a human eye by applying magnetic fields and
measuring the local oxygen concentration [21]. However, the movement of mini-
aturized objects in liquid environments faces the challenges of increased „viscous
forces“. For small objects the Reynolds number decreases due to the small char-
acteristic length, Lch , of the object [158]:
131
8 Application III: MPC 3D-microstructures by two-photon polymerization
132
8.2 State-of-the-art TPP
U V - P h o to lith o g r a p h y T P P la s e r
1 0 0
T r a n s m itta n c e [% ]
8 0
0 v o l.%
6 0 1 v o l.%
2 v o l.%
3 v o l.%
4 0 5 v o l.%
2 0
M P C
0
3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0
W a v e le n g th [n m ]
Figure 8.1: Transmittance of MPC at different wavelength. At larger wavelength, (780 nm)
of the laser for TPP, the MPC has a higher transmittance compared to UV photolitho-
graphy at wavelength of 350 – 410 nm. (The transmittance curve are the same as in
Figure 5.8). The sample thicknesses are around 1.6 µm and listed in Table 5.3
133
8 Application III: MPC 3D-microstructures by two-photon polymerization
Figure 8.2: Microturbine fabricated with TPP using a 2.1 wt.% superparamagnetic MPC.
a) model of the micro-turbine, b,c) SEM images of the micro-turbine. Reprinted from
[34].
134
8.3 Theory of TPP
E = hP ν (8.2)
where, h P , is the Planck’s constant and, ν, is the frequency of the light. For TPP
a pulsed near-IR laser is focused into a near-UV light sensitive photoresist which
posses high optical transparency at the laser wavelength. Polymerization based
on a single photon is improbable. However, in the focal volume of the laser,
the photon density is high enough for an absorption of two photons simultan-
eously (within the time of the absorption process). This provides the required
total (double) excitation energy to start the polymerization (Figure 8.3) [162, 163].
A volume pixel with ellipsoidal shape (voxel) is exposed and can be used as
writing tip for the fabrication of various shapes. After a dip into a developer
bath 3D structures are obtained. The size of the exposed volume is given by
the laser intensity distribution, which depends on the applied laser power. The
combination of optical, chemical and material non-linearities makes it possible to
achieve reproducible fabrication resolution down to a level of λ/10 to λ/50 [159].
Thus, voxel volumes much beyond the optical diffraction limit can be obtained.
For the two or multi photons absorption a femtosecond laser is needed (typical
wavelength is 780 nm) to generate an extremely large transient power density, e.g.
1013 W/µm2 [159]. Positive and negative tone photoresist can be written with the
two-photon absorption technique.
135
8 Application III: MPC 3D-microstructures by two-photon polymerization
8.4 Experimental
A direct laser writing tool from the company Nanoscribe GmbH is used for the
fabrication of the MPC microstructures. The setup consists of an inverted mi-
croscope, a Ti:sapphire near IR femtosecond laser (100 MHz repetition rate, sub-
150 fs pulses, central wavelength of 780 nm) and an optical setup to process and
direct the laser light through an objective into the resist.
For rough position of the sample a motorized stage was used and for the writ-
ing space a piezo stage of 300 µm x 300 µm x 300 µm volume relative to the fixed
laser was used. In this work, an oil immersion objective with 100x magnification
(NA = 1.4) was used to achieve MPC microstructures with smallest feature sizes.
A uniform rotating magnetic field was used, to test the actuation and swimming
properties of the MPC-based microstructures. The setup consists of three Helm-
holtz coils placed orthogonally to each other producing uniform field in any direc-
tion. The field can be rotated by varying the currents through the coils. The MPC
microstructure was immersed in a water tank. The structure was observed by an
optical microscope. To investigate the swimming properties the microstructure
was mechanically detached from the glass substrate. The motions of micostruc-
tures were recorded using a CCD camera placed on top of the microscope. The
setup is described in detail in [22].
136
8.5 Fabrication limitations
Figure 8.4: Schematic of experiment to determine the smallest line width. Lines with dif-
ferent offsets to the substrate were written and the linewidth, w L , was measured by
SEM.
The highest line resolution (smallest thickness and width) using TPP will be
achieved at the dose where the exposure threshold is exceeded slightly. The line
resolution depends on the scanning speed and the laser power and the used mater-
ial.
Experimental
Firstly, the minimal line width of the MPC with 0, 2, 4 vol.% Fe3 O4 nanoparticle
concentration was investigated. Lines were written with different offsets in z-
direction to the substrate surface as illustrated in Figure 8.4, because the focus rel-
atively to the substrate surface cannot be adjusted exactly and varies from sample
to sample. The line width, w L , is measured at the highest or second highest not
detached line for a parameter field varying scanning speed and laser power.
Secondly, the minimal line height, h L , for the different scanning speed and laser
power were investigated. Inclined lines were written with different scanning
speed and laser power from the substrate as illustrated in Figure 8.5. During
air drying after developing process the freestanding lines tumble onto the side
137
8 Application III: MPC 3D-microstructures by two-photon polymerization
Figure 8.5: Inclined MPC lines were written with an angle of 5◦ with different scanning
speed and laser power into the spin coated MPC composite layer on a glass substrate.
During air drying process of the composite after development the polymerized line
tumbles onto the substrate. The height of the line, h L , can be measured with an SEM
as shown in the inlet image.
towards the substrate and the line height can be determined by scanning electron
microscopy (SEM) measurements.
Figure 8.6 shows the written lines of a 2 vol.% MPC after development for dif-
ferent scanning speed and laser power. The parameter field to obtain stable line
structures is limited. For too high scanning speed and too low power the line
cannot polymerize. The smallest line thicknesses are obtained for a certain scan-
ning speed-power area, which is marked in Figure 8.6 with square frames. The
inset image in the same figure shows an enlargement of the line set written with
1.8 mW and 25 µm/s. Increasing power and decreasing scanning speed at the
same time the exposure dose increases and the polymerized lines thicken. When
138
8.5 Fabrication limitations
Figure 8.6: SEM of 2 vol.% MPC lines written with TPP technique with different offsets
towards the glass substrate: A,B,C,D,E (Figure 8.4). The scan speed and laser power
was varied. The thinnest linewidth for stable lines were determined (marked with a
square frame). The inlet shows an enlargement of a line-set using 1.8 mW and 25 µm/s.
139
8 Application III: MPC 3D-microstructures by two-photon polymerization
2 .4
2 .0
L a s e r P o w e r [m W ]
1 .6
0 v o l% s ta b le lin e
0 v o l% d e fo rm e d lin e
1 .2 0 v o l% n o lin e
2 v o l% s ta b le lin e
2 v o l% d e fo rm e d lin e
0 .8 2 v o l% n o lin e
4 v o l% s ta b le lin e
4 v o l% d e fo rm e d lin e
0 .4 4 v o l% n o lin e
1 1 0 1 0 0 1 0 0 0
S c a n n i n g s p e e d [ µm / s ]
Figure 8.7: Summary of the fabrication results for written MPC lines with 0, 2, and
4 vol.% nanoparticle concentration with different scanning speed and laser power. The
inserted lines in the graph mark roughly the estimated fabrication border between suc-
cessful written straight MPC lines with minimal line widths and deformed lines. The
region left of the marked lines are the parameter areas for successful microstructure
fabrication.
with optimized laser power 1.2 mW and 25 µm/s, and (b) the same microstruc-
ture with an increased laser power of 2.9 mW and 25 µm/s, where the composite
structure is destroyed.
The measured minimal line widths for a 2 vol.% sample are summarized in
Table 8.1. The corresponding heights measured by the method explained in Fig-
ure 8.5 are also listed. An aspect ration of ∼4 is obtained.
The average of the minimal line width of stable lines for 0, 2, and 4 vol.% MPC
are 325 ± 14 nm, 314 ± 44 nm and 280 ± 13 nm, respectively. The minimal line
widths depend only slightly on the nanoparticle concentrations.
140
8.5 Fabrication limitations
Figure 8.8: SEM image of a MPC (2 vol.%) fabricated woodpile structure with (a) optim-
ized laser power 1.2 mW (25 µm/s) and (b) the same microstructure with an increased
laser power of 2.9 mW (25 µm/s). where the composite decompose due to a too high
laser power.
Table 8.1: Measured minimal line widths and heights for 2 vol.% MPC.
Minimal line width Minimal line height
5 µm/s, 1 mW 350 nm 1.46 µm
12 µm/s, 1.4 mW 338 nm -
25 µm/s, 1.8 mW 318 nm 1.52 µm
50 µm/s, 2 mW 251 nm 1.11 µm
Average 314 ± 44 nm 1.36 ± 0.22 µm
141
8 Application III: MPC 3D-microstructures by two-photon polymerization
Figure 8.9: SEM images of 3D-microstructures fabricated by TPP (cantilevers with de-
signed dimensions of 15 x 10 x 1 µm). Three different slicing distances (distances
between written lines to obtain a polymerized volume) were tested. (a) 200 nm, (b)
400 nm, (c) 600 nm in vertical and horizontal direction. (a) is overexposed, (b) has op-
timized parameters, (c) slicing distance too far separated, results in a rough surface.
The images are taken under an angle of 60◦ .
SU-8 fabricated with different slicing distances: (a) 200 nm, (b) 400 nm, and (c)
600 nm in vertical and horizontal direction and all written with 25 µm/s and
2 mW (design cantilever volume: L = 15 µm, w = 10 µm, h = 1 µm). When the sli-
cing distance is too short (vertically and horizontally), as depicted in Figure 8.9 (a),
the polymer is overexposed due to the overlap of the polymerized lines (multi ex-
posure). For too large slicing distance (c) the surface is rippled because the lines
are too far separated. (b) Shows the microstructure with optimized slicing dis-
tances.
Larger objects like a MPC cuboid with side length of 10 µm and heights of
25 µm were successfully fabricated using TPP with a 2 vol.% MPC (Figure 8.10
(a)). For comparison, for a 2 vol.% MPC using conventional UV photolithography
the maximum polymerized thickness (using 5 J/cm2 ) is 6.5 µm (Table 5.2). With
TPP technique objects with hollow shapes were fabricated such as hollow cubes
with dimensions of 20 x 20 x 20 µm, with a 2 µm thick wall and four 10 x 10 µm win-
dows (Figure 8.10 (b) and (c)). Hollow microstructures can be interesting for pos-
sible drug delivery applications.
1 The helical microstructures have been fabricated and actuated in cooperation with Li Zhang (Insti-
tute of Robotics and Intelligent Systems, ETH Zurich, Switzerland)
142
8.6 Fabrication of helical microstructures
Figure 8.10: SEM images from MPC objects with 2 vol.% Fe3 O4 nanoparticles concentra-
tion: (a) A cuboid (10 x 10 x 25 µm), a hollow cube (b) (20 x 20 x 20 µm with 2 µm wall
thickness and four 10 x 10 µm windows), (c) shows a closer view of the cube.
Table 8.2: Parameters for the fabrication of helical microstructures as shown in Figure 8.12
2 vol.% 4 vol.%
power 0.8 mW 0.8 mW
scanning speed 5 µm/s 1 µm/s
trajectories (x,z) 5x2 5x2
trajectory distances (x,z) 200/400 200/400
shows successfully fabricated helical microstructures with 2 and 4 vol.% Fe3 O4 nan-
oparticle concentration. Based on the parameters obtained by the minimal line
resolution the scanning speed and power parameters were optimized (power
slightly reduced) because of the writing with overlaping lines. The optimized
parameters for the fabricated helical microstructures are given in Table 8.2.
If the particle concentration is increased, the surface roughness of the helical
microstructures increases due to the radiation absorption and scattering by the
nanoparticles and their agglomerates. A rough surface of the structure can be ad-
vantageous for biofunctionalization (drug release) because of increased surface
area.
Figure 8.13 (a) and (b) shows 2 vol.% MPC helical microstructures with differ-
ent filament widths. The filament in Figure 8.13 (a) has a width of 1.85 µm and
a height of 1.3 µm with 9:2 (horizontal : vertical) trajectories and (b) a width of
1.0 µm and a height of 1.3 µm with 5:2 trajectories. The filament width can be ad-
justed by the selection of the number of horizontal trajectory lines. The width, w L ,
and height, h L , for a 2 vol.% MPC single line using 0.8 mW and 5 µm/s are 245 nm,
and 971 nm, respectively. Distances between the trajectory lines of 200 nm (hori-
zontal) and 400 nm (vertical) were chosen.
For high particle loading (4 vol.%) and low exposure doses (0.8mW, 3 µm/s)
143
8 Application III: MPC 3D-microstructures by two-photon polymerization
Figure 8.11: Schematic of the fabrication of helical microstructure with line trajectories.
The cross-section of the filament shows the group of trajectories used to build the spiral.
The voxels of the laser trajectories overlap each other.
Figure 8.12: SEM images of helical microstructures made from MPC with different particle
concentrations. Parameters are given in Table 8.2
144
8.7 Magnetic actuation of MPC helical microstructure
Figure 8.13: SEM images of helical microstructures made with MPC. (a) shows a
2 vol.% helical structure with thick filament (9:2 trajectories); filament height: 1.3 µm,
width: 1.85 µm (0.8 mW, 5 µm/s). (b) shows a 2 vol.% helical microstructure with thin
filament (5:2 trajectories); filament height: 1.3 µm, width: 1 µm (0.8 mW, 5 µm/s). (c)
A 4 vol.% structure with low exposure dose (0.8 mW, 3 µm/s) shows a clear thinning
of the filament thickness towards the top.
a thinning of the width and the height of the filaments towards the top of the
spiral is observed (Figure 8.13 (c)). This is due to the absorption and scattering of
the laser by the nanoparticles in the composite. Helical structures (4 vol.%) with
heights of 16.8 µm were fabricated.
145
8 Application III: MPC 3D-microstructures by two-photon polymerization
t
1 t
2
t
3
20µm
Mot
iondi
rect
ion
Magnet
icf
iel
drot
ati
on
Figure 8.14: Swim test in water of MPC helical structure with 2 vol.% Fe3 O4 superpara-
magnetic nanoparticle concentration. The images show a quarter rotation of the super-
paramagnetic helical microstructure around its helical axis. The time domain between
the microscope pictures t1 , t2 and t3 are ∼0.04 s. The helical structure turns with ∼
3 Hz. After 3 s the helical microstructure has moved a distance of around 12 µm (not
shown).
ricated and showed similar swimming behavior. This experiment shows that an
actuation and magnetic control of microstructures made from the superparamag-
netic MPC with only 2 vol.% nanoparticles concentration is possible.
Tm = µ0 V M × H (8.3)
where V and M are the volume and magnetization of the body, respectively, and
µ0 =4π·107 T·m/A is the permeability of free space [112]. There must be a fixed
component of magnetization in the non-axial plane to result a torque. It seems
that the magnetization is based on the shape anisotropy of the helical microstruc-
ture and it is assumed that the superparamagnetic particles interact between each
other to establish a preferred and anisotropic magnetization in a certain direction.
The swim performance like speed, wobbling, performance under different mag-
netic fields, maximal rotation speed, different shape geometries and the direction
of the magnetization (easy axis) due to shape anisotropy and particle interaction
in the superparamagnetic helical microstructure must be investigated more in
detail.
146
8.8 Conclusion and Outlook
The line resolution measurements have shown that with the MPC with 2 vol.% nan-
oparticle concentration lines with widths of 314 nm and heights of 1.36 µm can
be fabricated. The minimal line resolution determines the minimal feature size
for this composite using TPP technique.
Increasing the nanoparticle concentration lower scanning speeds and a lower
laser power are required. This narrows the fabrication parameter field, which
is constricted by the minimal speed (long fabrication process) and by the max-
imal power value where the microstructure is destroyed, due to absorbed energy.
For a 5 vol.% MPC parameters for a successful fabrication could not be found.
For the fabrication of microstructures with TPP technique the Fe3 O4 nanoparticle
loading in the MPC is limited to 4 vol.%.
With TPP technique MPC microstructures with greater vertical dimensions,
higher aspect ratios and smaller feature sizes can be obtained compared to UV
photolithography technique.
This work has shown that the developed MPC can be used to fabricate super-
paramagnetic MPC microstructures with two-photon polymerization. Helical
microstructures were fabricated which mimic natural bacteria flagellas. Using
a uniform rotating magnetic the MPC helical microstructures were successfully
rotated in water with a cork-screw motion and can be steered near a solid sur-
face. The direction of the movement can be controlled by applied magnetic fields.
Superparamagnetic MPC helical microstructures have potential for biomanipula-
tion and drug delivery [164]. Using the developed magnetic polymer composites
for magnetic remotely controlled microstructures/microrobots has the following
advantages:
• The composite has high mechanical stability (Young’s modulus: 4.2 – 5.1 GPa,
see Section 5.5) ,
• The high thermal stability of the SU-8 allows chemical reactions at the MPC
microstructure surface at elevated temperature,
• The high chemical resistance of SU-8 allows the operation in different envir-
onments and cleaning of the microstructure surface e.g. by acetone.
147
8 Application III: MPC 3D-microstructures by two-photon polymerization
MPC are auspicious candidates for the fabrication of microrobots and can help
to push the research in the field of drug delivery and micromanipulation. An
improvement for a next generation of MPC artificial bacteria flagellas could be
the use of a bioerodable polymer matrix [165] allowing the decomposition of
microstructures after drug delivery in human body.
148
9 Conclusion and Outlook
This work focuses on fabrication and characterization of a magnetic polymer com-
posite (MPC) to fabricate magnetic microstructures. The composite consists of a
photopatternable polymer and superparamagnetic Fe3 O4 nanoparticles with a
mean diameter of ∼13 nm (size depending on the measurement method). The
combination of these two materials leads to a composite with outstanding prop-
erties. The composite has superparamagnetic characteristics and shows good
compatibility with biomaterials. It has a relatively high Young’s modulus com-
pared to other polymers, which allows the fabrication of stable suspended mi-
crostructures. A key advantage of the composite is the homogeneous dispersion
of the nanoparticles with low particle agglomerations, which enables the fabric-
ation of microstructures with feature sizes down to 300 nm. MPC can also be
heated remotely by external magnetic fields.
In the first part of this work material combinations were evaluated. Epoxy
SU-8 and Fe3 O4 nanoparticles were found as the most promising materials. Then,
the particle suspension and mixing were optimized and the properties of the
MPC were thoroughly characterized. In the second part of this work three pos-
sible applications of MPC microstructures were investigated. Microcantilevers
and in-plane resonators were fabricated by standard photolithography, and their
resonant behavior were characterized. The improvement and successful mag-
netic actuation of superparamagnetic resonant microstructures by the use of dif-
ferent magnetic actuation setups are presented. Furthermore, three dimensional
(3D) microstructures were fabricated by two-photon polymerization (TPP). Hel-
ical microstructures show cork-screw swimming behavior in water when a rota-
tional uniform magnetic field is applied.
9.1 Contributions
MPCs for the fabrication of composite structures with feature sizes > 5 µm have
been reported in literature [31, 33, 36–40]. When further device miniaturization is
desired, the processing of composite materials with magnetic particles becomes
a challenge. Due to magnetic and Van der Waals forces between nanoparticles,
149
9 Conclusion and Outlook
150
9.1 Contributions
A dispersion with low agglomerate sizes is a key requirement for the fabrication
of microstructures with small feature sizes and uniform magnetic and mechanical
properties. Spin coated MPC films with up to 10 vol.% (32 wt.%) were investig-
ated by TEM analysis and show a homogeneous nanoparticle distribution with a
low amount of agglomerates (agglomerate mean sizes are around 50 nm). Small
angle X-ray scattering (SAXS) measurements are in agreement with the TEM ana-
lysis and show that agglomerates are independent of the amount of embedded
particles for the investigated range. X-ray disc centrifuge (XDC) measurements
indicate that the agglomerates were already present in the initial magnetic sus-
pension and negligible agglomerate formation is occurring during the mixing of
the composite.
Magnetic properties of the MPC are crucial for the microdevice performance. Sat-
uration magnetization and coercivity are important parameters for the magnetic
actuation of the microstructure. It was shown that the developed composite
exhibits superparamagnetic behavior, and the magnetic characteristics depend
mainly on the incorporated particles. Comparison of the magnetic characteristics
of the composite (Figure 5.13) and the particles (Figure 5.15) show that the com-
posite and the nanoparticles exhibit consistent magnetic behavior. The magnetic
properties of the composite allows remote actuation of microstructures. A sat-
uration magnetization of 13.8 kA/m was measured for the composite film with
151
9 Conclusion and Outlook
5 vol.% particles. The measured magnetic force per volume, Fv , of 45.7 · 103 N/m3
for a 3 vol.% composite is too small to achieve static deflection of a MPC mi-
crocantilever with a Young’s modulus of 4.4 GPa. The composite is therefore
not suitable for applications where static deflections are desired by magnetic
actuation. However, the composite is suitable for microstructures where low
magnetic forces are sufficient, such as resonant structures or the propulsion of
microdevices in liquids. The dynamic Young’s modulus of the composite was
characterized using resonant cantilevers and it varies between 4.1 GPa for pure
SU-8 to 5.1 GPa for 5 vol.% particle concentration.
152
9.1 Contributions
Applications
Cantilevers
Fabricated superparamagnetic MPC cantilevers with 5 vol.% Fe3 O4 nanoparticle
concentration were successfully actuated in air and in water by alternating mag-
netic fields. It was found that the addition of a uniform magnetic field improves
the deflection of the microresonators by increasing the magnetization of the mi-
crostructure. Using a feedback system, a Q-factor enhancement in air of a factor
two (from 15 to 29) was achieved. As a proof of concept for mass sensing, gold
was sputtered at the tip of a cantilever and a clear resonance frequency shift
was measured. Photodefinable MPC cantilevers with γ-Fe2 O3 nanoparticles for
atomic force microscopy (AFM) applications with high resolution imaging per-
formance have been reported recently [166]. However, a magnetic actuation of
the cantilevers was not demonstrated due to the too low nanoparticle concentra-
tion (0.16 wt.%).
In-plane microresonators
An in-plane MPC microresonator with 5 vol.% particle concentration was design-
ed and fabricated. The in-plane resonant behavior was investigated depending
on the plate size. The microresonators were successfully actuated in air. Large
deflections (28 µm) can be achieved with MPC microstructures despite the low
magnetic volume of only 5 vol.% compared to a full magnetic metal structure
(e.g. CoNi).
153
9 Conclusion and Outlook
9.2 Outlook
This work has highlighted the various characteristics of the developed MPC. The
MPC has a variety of interesting properties such as superparamagnetic character-
istics, bio-compatibility, and high mechanical stability. It can be remotely heated
by applied magnetic fields. A key advantage of the composite is the homogen-
eous dispersion of the nanoparticles with low particle agglomerations, which en-
ables the fabrication of microstructures with small feature sizes. In the following
possible future applications with the composite are presented.
MPC cantilevers can be used as remote mass sensors in air with magnetic actu-
ation and optical readout as presented in Section 6.1.2. The self-actuation mode
(without an external frequency sweep signal) for the Q-enhancement system as
reported [147, 148] has to be investigated more in detail. This could improve the
Q-factor to enable the sensing of biomolecule interactions in water or the detec-
tion of disease markers in human blood. In combination with the all-polymer
packaging, a low-cost disposable sensor can be developed.
The MPC can be used for microtools such as a micro cell manipulators or micro-
turbines for lab-on-a-chip applications [46]. Friction forces often limit the effi-
cient rotation and movement of objects in microchannels [167]. Due to the small
154
9.2 Outlook
particles, the homogeneous dispersion, and low aggregate sizes, the developed
MPC allows the fabrication of small structures with low surface roughnesses as
demonstrated for the microcantilevers Figure 6.1. Based on the cell proliferation
studies the MPC can be used for cell collection applications on MPC microstruc-
tures using magnetic fields [45].
MPCs in general benefit from the availability of a variety of fabrication meth-
ods such as hot embossing [27], photolithography [28], and injection molding
[35] (see Table 1.1). Fabrication processes such as inkjet printing (drops) or elec-
tro spinning (fibers) are often not considered for MPC fabrication because particle
agglomerates in the composite leads to clogging of the pinhole. Due to the homo-
geneous nanoparticle dispersion and small aggregate sizes the developed MPC
can be used for such fabrication processes. Because of the good functionalization
properties of SU-8, magnetic MPC microstructures are interesting for immunoas-
say tests in liquid, allowing an easy collections after the experiments by applied
magnetic fields. Inkjet printed MPC spheres can be used for self assembly applic-
ations using magnetic fields.
It has been shown that superparamagnetic MPC helical microstructures can be
controlled by external magnetic fields. MPC microstructures are non-toxic to cells
and have potential for in vitro biomanipulation [164]. For in vivo applications
in animals or humans over long durations, further biocompatibility tests such
as allergy tests and experiments with nanoparticle absorption by cells have to
be conducted. MPC microstructures can be heated in an alternating magnetic
field to eventually trigger a temperature sensitive bioreaction at the surface of
the microstructure/robot. The high thermal stability and chemical resistance of
the SU-8 matrix allows chemical reactions at the MPC microstructure surface at
elevated temperatures. MPCs are auspicious candidates for the fabrication of
microrobots and can help to push the research in the field of micromanipulation,
drug delivery, and self assembly.
155
Bibliography
[1] C. Goodyear, “Verfahren zur Fabrikation von Gegenständen welche mehr
oder weniger hart, biegsam oder elastisch sind, aus Kautschuk und Gutta-
percha in Verbindung mit anderen Stoffen,” Dinglers Polytechnisches Journal,
vol. 139, p. 376, 1856.
[8] J.B.Park and J.D.Bronzino, Biomaterials. Boca Raton, FL: CRC Press, 2003.
[9] N.-T. Nguyen, S.-S. Ho, and C. L.-N. Low, “A polymeric microgripper with
integrated thermal actuators,” Journal of Micromechanics and Microengineer-
ing, vol. 14, no. 7, p. 969, 2004.
157
Bibliography
158
Bibliography
159
Bibliography
[34] H. Xia, J. Wang, Y. Tian, Q.-D. Chen, X.-B. Du, Y.-L. Zhang, Y. He, and
H.-B. Sun, “Ferrofluids for fabrication of remotely controllable micro-
nanomachines by two-photon polymerization,” Advanced Materials, vol. 22,
no. 29, pp. 3204–3207, 2010.
160
Bibliography
[42] W. Xue and T. Cui, “Carbon nanotube micropatterns and cantilever arrays
fabricated with layer-by-layer nano self-assembly,” Sensors and Actuators A:
Physical, vol. 136, no. 2, pp. 510–517, 2007.
[46] Y. Tian, Y.-L. Zhang, J.-F. Ku, Y. He, B.-B. Xu, Q.-D. Chen, H. Xia, and H.-B.
Sun, “High performance magnetically controllable microturbines,” Lab on
a Chip, vol. 10, no. 21, pp. 2902–2905, 2010.
[50] C. M. Sorensen, Magnetism. New York: John Wiley & Sons, 2001, pp. 169–
221.
161
Bibliography
[53] M. Ma, Y. Wu, J. Zhou, Y. Sun, Y. Zhang, and N. Gu, “Size dependence of
specific power absorption of Fe3 O4 particles in AC magnetic field,” Journal
of Magnetism and Magnetic Materials, vol. 268, no. 1-2, pp. 33–39, 2004.
[54] Y.-h. Zheng, Y. Cheng, F. Bao, and Y.-s. Wang, “Synthesis and magnetic
properties of Fe3 O4 nanoparticles,” Materials Research Bulletin, vol. 41, no. 3,
pp. 525–529, 2006.
[55] H. Iida, K. Takayanagi, T. Nakanishi, and T. Osaka, “Synthesis of Fe3 O4
nanoparticles with various sizes and magnetic properties by controlled hy-
drolysis,” Journal of Colloid and Interface Science, vol. 314, no. 1, pp. 274–280,
2007.
[56] A. H. Lu, E. L. Salabas, and F. Schuth, “Magnetic nanoparticles: Syn-
thesis, protection, functionalization, and application,” Angewandte Chemie-
International Edition, vol. 46, no. 8, pp. 1222–1244, 2007.
[57] J. Wang, H. Xia, B.-B. Xu, L.-G. Niu, D. Wu, Q.-D. Chen, and H.-B. Sun,
“Remote manipulation of micronanomachines containing magnetic nano-
particles,” Opt. Lett., vol. 34, no. 5, pp. 581–583, 2009.
[58] R. H. Kodama, “Magnetic nanoparticles,” Journal of Magnetism and Magnetic
Materials, vol. 200, no. 1-3, pp. 359–372, 1999.
[59] M. Bao, Analysis and design principles of MEMS devices, Chapter 2.5, ser. Ma-
terials & Mechanical. Elsevier Science Ltd, 2005.
[60] S. Schmid, “Electrostatically actuated all-polymer microbeam resonators -
characterization and application,” Ph.D. dissertation, ETH Zurich, 2009.
[61] S. Ahmed and F. R. Jones, “A review of particulate reinforcement theories
for polymer composites,” Journal of Materials Science, vol. 25, no. 12, pp.
4933–4942, 1990.
[62] A. Einstein, Investigation on theory of brownian motion. New York: Dover,
1956.
[63] M. Mooney, “The viscosity of a concentrated suspension of spherical
particles,” Journal of Colloid Science, vol. 6, no. 2, pp. 162–170, 1951.
[64] E. H. Kerner, “The elastic and thermo-elastic properties of composite me-
dia,” Proceedings of the Physical Society. Section B, vol. 69, no. 8, p. 808, 1956.
[65] H. Eggers and P. Schummer, “Reinforcement mechanisms in carbon black
and silica loaded rubber melts at low stresses,” Rubber Chemistry and Tech-
nology, vol. 69, no. 2, pp. 253–265, 1996.
162
Bibliography
163
Bibliography
[79] Z.-Z. Wu, Y. Zhao, and W. S. Kisaalita, “Interfacing SH-SY5Y human neuro-
blastoma cells with SU-8 microstructures,” Colloids and Surfaces B: Biointer-
faces, vol. 52, no. 1, pp. 14–21, 2006.
[80] J. Zhan, G. Tian, L. Jiang, Z. Wu, D. Wu, X. Yang, and R. Jin, “Superpara-
magnetic polyimide/γ-Fe2 O3 nanocomposite films: Preparation and char-
acterization,” Thin Solid Films, vol. 516, no. 18, pp. 6315–6320, 2008.
164
Bibliography
165
Bibliography
[99] F. Umbrecht, “Wireless implantable passive strain sensor for strain monit-
oring of orthopedic implants,” Ph.D. dissertation, ETH Zurich, 2010.
166
Bibliography
167
Bibliography
168
Bibliography
[140] R. Feng and R. J. Farris, “The characterization of thermal and elastic con-
stants for an epoxy photoresist SU8 coating,” Journal of Materials Science,
vol. 37, no. 22, pp. 4793–4799, 2002.
169
Bibliography
170
Bibliography
[151] P. Rust and J. Dual, “Novel method for gated inductive readout for highly
sensitive and low cost viscosity and density sensors,” in 16th International
Solid-State Sensors, Actuators and Microsystems Conference (TRANSDUCERS),
2011, pp. 1088–1091.
[155] X. Zhang and W. C. Tang, “Viscous air damping in laterally driven microres-
onators,” in Micro Electro Mechanical Systems, MEMS ’94, Proceedings, IEEE
Workshop, 1994, pp. 199–204.
[159] Y.-L. Zhang, Q.-D. Chen, H. Xia, and H.-B. Sun, “Designable 3D nanofab-
rication by femtosecond laser direct writing,” Nano Today, vol. 5, no. 5, pp.
435–448, 2010.
171
Bibliography
[161] K.-S. Lee, D.-Y. Yang, S. H. Park, and R. H. Kim, “Recent developments in
the use of two-photon polymerization in precise 2D and 3D microfabrica-
tions,” Polymers for Advanced Technologies, vol. 17, no. 2, pp. 72–82, 2006.
172
Publications
Reviewed Articles
A1 O. Ergeneman, M. Suter, K. Sivaraman, B. Özkale, S. Pané, T. Lühmann,
H. Hall, C. Hierold, B. J. Nelson, „Drug release from SU-8 nanocomposite
surface", In preparation for submission to Small, 2011.
Conference Proceedings
C1 M. Suter, Y. Li, G. A. Sotiriou, A. Teleki, S. E. Pratsinis, and C. Hierold,
„Low-cost fabrication of PMMA and PMMA based magnetic composite can-
tilevers", in 16th International Solid-State Sensors, Actuators and Microsystems
Conference (TRANSDUCERS), 2011, pp. 398-401.
173
Bibliography
Poster Presentations
P1 M. Suter, Y. Li, G. A. Sotiriou, A. Teleki, S. E. Pratsinis, and C. Hierold,
„Low-cost fabrication of PMMA and PMMA based magnetic composite can-
tilevers", in 16th International Solid-State Sensors, Actuators and Microsystems
Conference (TRANSDUCERS), 2011, pp. 398-401.
174
Bibliography
175
Curriculum vitae
Personal Details
Name Marcel Suter
Birth 30 January 1979, Switzerland
Citizenship Switzerland
Education
05/2007 – 12/2011 ETH Zurich, Micro and Nanosystems, Switzerland
Dissertation: Photopatternable superparamagnetic nanocomposite for
the fabrication of microstructures
10/2004 – 02/2007 University of Neuchâtel, Micro and Nanotechnology
Graduation with Bachelor and Master Degree
10/2006 – 02/2007 University of California at Berkeley (UCB), Group of Prof. Lin
(BSAC), USA
Visiting Scholar, Master Thesis: Single carbon nanotube pirani
gauge by local synthesis
10/2000 – 12/2003 Interstate University of Applied Sciences Buchs NTB
Dipl. Ingenieur FH in Systemtechnik (Bachelor in Systems Engin-
eering), major: MEMS, Diploma thesis: Anemometer
08/1995 – 08/1999 Siemens Building Technologies Cerberus Division AG,
Männedorf, Switzerland
Vocational training as physical laboratory technician (Lehre als
Physiklaborant), major: Electronics / IT
Work Experience
01/2004 – 07/2004 ETH Zurich, Institute for mechanical systems, Switzerland
Process engineering: Development of a plasma etch process
08/1999 – 10/2000 Siemens Building Technologies Cerberus Division AG,
Männedorf, Switzerland
Application development: Development of optical smoke-detection-
system
Languages
German mother tongue
English Level C2 (CEFR)
French Level B2
Spanish Level A2
177