Eth-5934-02.pdf

Research Collection

Doctoral Thesis

Photopatternable superparamagnetic nanocomposite for the fabrication of microstructures 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 for the degree of

Doctor of Sciences presented by

MARCEL SUTER M.Sc. Micro and Nanotechnology born January 30, 1979 citizen of Rapperswil-Jona SG accepted on the recommendation of Prof. Dr. Christofer Hierold, examiner Prof. Dr. Bradley J. Nelson, co-examiner

2011

ii

Document typeset by the author using the LATEX 2ε system and the KOMA-Script document class scrbook.

c 2012 Marcel Suter, Zürich Copyright

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 additional filler materials for reinforcement, increasing wear resistance, reducing inflammability, or adding different material properties. Magnetic polymer composites 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) obtained by dispersing superparamagnetic magnetite nanoparticles in a photosensitive 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 different methods such as optical microscopy, transmission electron microscopy (TEM), and small-angle x-ray scattering (SAXS). The MPC shows a homogeneous nanoparticle 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, different dispersing agents were tested. Mechanical, magnetic, wetting, and biocompatibility properties of the nanocomposites are characterized. The nanocomposites exhibit superparamagnetic properties. The biocompatibility of the MPC is demonstrated by proliferation of human dermal fibroblasts, and show the potential 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 photolithography technique or laser writing techniques based on two-photon polymerization (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 obtained 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 composite. 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 actuated by rotating homogeneous magnetic fields. Microstructures fabricated with the developed MPC have higher nanoparticle concentration and smaller feature sizes compared to reported MPC microstructures 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-achip applications.

vi

Zusammenfassung Kunststoffe (Polymere) sind heute allgegenwärtig. Die meisten Polymere enthalten 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 Fabrikationsprozesse, mit den magnetischen Eigenschaften des Füllmaterials. In dieser Arbeit wird ein fotodefinierbarer magnetischer Polymerkomposit (MPK) vorgestellt. Für den Komposit werden superparamagnetische Magnetite Nanopartikeln in fotoempfindlichem Epoxy SU-8 dispergiert. Der Komposit kann zur Herstellung von magnetischen Mikrostrukturen verwendet werden, welche mit magnetischen Feldern aktuiert werden können. Die magnetischen Mikrostrukturen, welche in dieser Arbeit präsentiert werden, zeigen das Potential des entwickelten Komposits für Anwendungen wie Mikroresonatoren und Mikroroboter. Der erste Teil dieser Arbeit beschreibt die Evaluierung der Materalien und die Herstellung des Komposits. Die Materialeigenschaften wurden untersucht in Abhängigkeit von dem Nanopartikel-Füllgrad von 1 – 10 vol.% (4 – 32 wt.%). Weiter wurde die Dispersion von den Nanopartikeln (Durchmesser: 11.4 ± 3.4 nm, Mittelwert von TEM-Messungen) und deren Agglomeration mit verschiedenen Methoden geprüft: Mit optischer Mikroskopie, mit Transmissions-Elektronen Mikroskopie sowie mit Röntgen-Kleinwinkelstreuung-Messungen. Der MPK zeigt eine homogene Nanopartikelverteilung mit einer geringen Nanopartikelagglomeration. Dies ist eine wichtige Voraussetzung für die Fertigung von Mikrostrukturen mit kleinen Strukturabmessungen sowie uniformen magnetischen und mechanischen Eigenschaften. Verschiedene Dispergiermittel wurden ausgetested sowie die magnetischen und mechanischen Eigenschaften, Benetzbarkeit und Biokompatibilität des Komposits getestet. Der Komposit besitzt superparamagnetische Eigenschaften. Die Biokompatiblilität des Komposits für biologische Anwendungen 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 hochfrequenten magnetischen Wechselfeld aufgeheizt werden kann. Mikrostrukturen aus MPK können mittels konventioneller Fotolithografie oder

vii

Zusammenfassung der Zwei-Photonen Polymerisation (ZPP) Laserschreibtechnik gefertigt werden. Der Einfluss der Partikelkonzentration auf die Fabrikationsschritte wie Spincoating, UV-Belichtung und ZPP wurden systematisch untersucht. Die Schichtdicke für UV-Fotolithografie ist auf 2.9 µm für 5 vol.% Komposit limitiert. Mit der ZPP Technik konnten Strukturen mit Höhen bis zu 17 µm geschrieben werden und eine minimale Liniendicke von 314 nm für den 2 vol.% Komposit wurde erziehlt. Mikrokantilevers (einseitig eingespannte Mikrobalken) mit 0 – 5 vol.% Partikelkonzentrationen wurden erfolgreich fabriziert und benutzt um das E-Modul des Komposits zu bestimmen. Ein E-Modul von 4.1 GPa für ungefüllten und 5.1 GPa für 5 vol.% Komposit wurde gemessen. Weiter wurden in-plane Resonatoren hergestellt. Die Resonanzcharakteristik von beiden Resonatortypen wurden analysiert und beide erfolgreich mittels magnetischen Feldern aktuiert. Eine an der Kantileverspitze platzierte Goldmasse verursacht eine messbare Resonanzfrequenzverschiebung und bestätigt das Massendetektions-Konzept. Mit der ZPP Technik wurden spiralförmige Kompositstrukturen hergestellt. In einem rotierendem homogenen magnetischen Feld in Wasser bewegen sich die Spiralstrukturen wendelförmig vorwärts und können mit Magnetfeldern gesteuert werden. Die hergestellten magnetischen Komposit-Mikrostrukturen basierend auf dem entwickelten Komposit haben höhere Nanopartikelkonzentrationen und kleinere Strukturgrössen im Vergleich zu magnetischen Mikrostrukturen in bereits publizierten Arbeiten. Diese Arbeit zeigt die Chancen und Herausforderungen des MPK für Anwendungen in Mikrosystemen. MPK Kantilevers können als Massensensor mit magnetischer Aktuierung und optischer Auslesung gebraucht werden. Weiter können in Zukunft MPK Mikrostrukturen verwendet werden für das Transportieren von Medikamenten, für Biomanipulationen wie Zellmanipulation, für Mikroturbinen in Labor-auf-einem-Chip Anwendungen oder für Selbstanordnung von Mikrostrukturen.

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 Michael Wendlandt for his fundamental input on the project. I am also grateful to Salvador Pané who generously spent his time for scientific discussions and guidance. 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 Schöndube, 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 preparations. 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 images, Prof. Ann Marie Hirt (EPM, ETH) for the support with the magnetic characterization, 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 calculations, 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 assignment 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 constructive 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 (project no 200020-113350) is gratefully acknowledged.

x

Contents Abstract

v

Zusammenfassung

vii

Acknowledgment

ix

1

1

Introduction

1.1 1.2 1.3 1.4 2

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

Theory

2.1

2.2

2.3

3

Motivation . . . . . . . . . . . . . . . . . . . . . . . . . State-of-the-art of MPCs for microsystem applications Goals of the thesis . . . . . . . . . . . . . . . . . . . . . Approach and outline of this work . . . . . . . . . . .

11

Magnetism in nanoparticles . . . . . . . . . . . . 2.1.1 Units in magnetism . . . . . . . . . . . . . 2.1.2 Magnetic field and magnetic flux density 2.1.3 Magnetic materials . . . . . . . . . . . . . 2.1.4 Small particle magnetism . . . . . . . . . Mechanics of oscillating structures . . . . . . . . 2.2.1 Harmonic oscillator . . . . . . . . . . . . 2.2.2 Beam theory . . . . . . . . . . . . . . . . . 2.2.3 Quality factor of oscillating structures . . Young’s modulus of composites . . . . . . . . . . 2.3.1 Hashin-Shtrikman Model . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

Evaluation of materials

3.1

3.2 3.3

Polymer evaluation . . . . . . . . . . . 3.1.1 Polymer selection criteria . . . 3.1.2 Polymer selection . . . . . . . . Particle evaluation . . . . . . . . . . . Dispersant agent evaluation . . . . . . 3.3.1 Stabilization of particles . . . . 3.3.2 Selection of suitable surfactant photocurable epoxy matrix . .

1 2 8 9

11 11 12 12 13 17 17 18 22 22 23 25

. . . . . . . . . . . . on . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . magnetite particles . . . . . . . . . . . .

. . . . . . . . . . . . for . .

. . . . . .

25 25 25 28 33 33

. 34

xi

Contents 3.4 4

5

Fabrication 4.1 Composite fabrication . . . . . . . . . . . . . . . . . 4.1.1 Evaluation of mixing method for composite 4.1.2 Preparation of magnetic suspension . . . . . 4.1.3 Composite mixing . . . . . . . . . . . . . . . 4.2 Microstructure fabrication . . . . . . . . . . . . . . . 4.3 Polymer package fabrication . . . . . . . . . . . . . . 4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . .

37

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

38 38 39 39 39 43 46

Evaluation of composite properties

47

5.1

48 48 55 57 57 58 63 63 66 66 66 67 67 67 70 70 72 75 76 76 77

5.2

5.3

5.4

5.5

xii

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

Particle dispersion in composite . . . . . . . . . . . . . . . . . . . . . 5.1.1 Agglomerates and particle sizes . . . . . . . . . . . . . . . . 5.1.2 Discussion and conclusion . . . . . . . . . . . . . . . . . . . . Limitation using UV exposure . . . . . . . . . . . . . . . . . . . . . . 5.2.1 Backside exposure of composite film . . . . . . . . . . . . . . 5.2.2 UV transmittance depending on particle concentration . . . 5.2.3 UV transmittance depending on the layer thickness . . . . . 5.2.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . Minimal pattern transfer in composite using UV exposure . . . . . 5.3.1 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . 5.3.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . Magnetic characterization . . . . . . . . . . . . . . . . . . . . . . . . 5.4.1 Magnetic behavior of composite . . . . . . . . . . . . . . . . 5.4.2 Magnetic behavior of particles . . . . . . . . . . . . . . . . . 5.4.3 Material characterization of particles . . . . . . . . . . . . . . 5.4.4 Magnetic force on composite . . . . . . . . . . . . . . . . . . 5.4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . Young’s modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5.1 Measurement method . . . . . . . . . . . . . . . . . . . . . . 5.5.2 Error analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5.3 Comparison of FEM-simulation with Euler-Bernoulli approximation and evaluation of the influence of imperfect clamping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5.4 Damping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5.5 Stress gradient in cantilevers . . . . . . . . . . . . . . . . . . 5.5.6 Residuals on cantilevers . . . . . . . . . . . . . . . . . . . . . 5.5.7 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . .

79 84 85 86 86

Contents

5.6

5.7

5.8

5.9

5.5.8 Results . . . . . . . . . . . . . . . . . . 5.5.9 Discussion and conclusion . . . . . . . MPC heating with alternating magnetic field 5.6.1 Experimental . . . . . . . . . . . . . . 5.6.2 Results . . . . . . . . . . . . . . . . . . 5.6.3 Discussion and conclusion . . . . . . . Surface properties . . . . . . . . . . . . . . . . 5.7.1 Experimental . . . . . . . . . . . . . . 5.7.2 Results and discussion . . . . . . . . . 5.7.3 Conclusion . . . . . . . . . . . . . . . Biocompatibility . . . . . . . . . . . . . . . . . 5.8.1 Experimental . . . . . . . . . . . . . . 5.8.2 Results and discussion . . . . . . . . . 5.8.3 Conclusion . . . . . . . . . . . . . . . Summary and conclusion . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

87 87 90 90 91 92 95 95 96 96 96 96 97 99 99

6

Applications I: MPC cantilever resonator 101 6.1 MPC cantilever resonance characterization by thermally induced vibrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 6.1.1 Q-factor in different media . . . . . . . . . . . . . . . . . . . 103 6.1.2 Frequency shift of MPC cantilevers with additional mass . . 106 6.2 Magnetic actuation of superparamagnetic MPC cantilevers . . . . . 107 6.2.1 MPC cantilevers actuated by alternating magnetic field . . . 108 6.2.2 MPC cantilevers actuated by alternating inhomogeneous and additional uniform magnetic field . . . . . . . . . . . . . 110 6.2.3 Magnetic actuation of MPC cantilevers with Q-enhancement 114 6.2.4 Self-heating of cantilever during actuation by magnetic alternating field . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 6.2.5 Magnetic actuation of MPC cantilevers in water . . . . . . . 116 6.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

7

Application II: MPC lateral resonator 121 7.1 Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 7.2 In-plane resonance characterization by thermally induced vibrations124 7.2.1 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 7.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 124 7.4 Magnetic actuation of in-plane resonator . . . . . . . . . . . . . . . 126 7.4.1 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 7.4.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . 127 7.5 Magnetic readout of in-plane resonator . . . . . . . . . . . . . . . . 129

xiii

Contents 7.6 8

9

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

Application III: MPC 3D-microstructures by two-photon polymerization 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 State-of-the-art TPP . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3 Theory of TPP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.1 Direct laser writing tool . . . . . . . . . . . . . . . . . . . . . 8.4.2 Sample preparation . . . . . . . . . . . . . . . . . . . . . . . . 8.4.3 Magnetic actuation setup . . . . . . . . . . . . . . . . . . . . 8.5 Fabrication limitations . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5.1 Line resolution . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5.2 Fabrication of 3D structures . . . . . . . . . . . . . . . . . . . 8.6 Fabrication of helical microstructures . . . . . . . . . . . . . . . . . . 8.7 Magnetic actuation of MPC helical microstructure . . . . . . . . . . 8.8 Conclusion and Outlook . . . . . . . . . . . . . . . . . . . . . . . . .

131

Conclusion and Outlook

149

9.1 9.2

131 132 134 136 136 136 136 137 137 140 142 145 147

Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154

Bibliography

157

Publications

173

Curriculum vitae

177

xiv

List of symbols and abbreviations List of Symbols Symbol

Unit

Description

αc αSU −8 α Fe O 3 4 A AD B cth c cs χ D δ δP E E(Θ) η F F0 f f f luid f vac g Γ H Hc h hL

ppm/K ppm/K ppm/K m2 m2 T J/(kg K) Pa s m

Thermal expansion coefficient of a composite Thermal expansion coefficient of a SU-8 Thermal expansion coefficient of a magnetite Cross sectional area of a beam Damping-related effective area of a system Magnetic flux density Heat capacity Coefficient of damping Fixed beam length Susceptibility Exposure dose Length of the adsorbed molecules Penetration depth Energy Anisotropy Energy Dynamic viscosity of the fluid Force Excitation force Frequency Frequency in fluid Frequency in vacuum General mathematic function Hydrodynamic interaction function Magnetic field Coercivity Thickness of beam or layer Height of a written composite/polymer line

J/cm2 nm m J J Pa s N N Hz Hz Hz

A/m A/m m m

xv

Contents

Symbols Symbol

Unit

Description

hP hCoil I Iz I (q) K Ke f f K0 , K1 kB k L Lch LF

m2 kg/s m A m4

LB Lm λ M Mx,y,z Mr Ms m ma mL mM µ0 ν ns ω ωr ω0 Ω(ω )

m m m A/m A/m A/m A/m kg kg kg/m A m2 T m/A Hz m Hz Hz Hz

P PL

J J

Planck’s constant: 6.626 · 10−34 Height of a coil Current Moment of inertia Intensity as a function of the scattering vector q Anisotropy constant Anisotropy constant of a magnetic nanoparticle Modified Bessel functions of third kind Boltzmann constant: 1.38·10−23 Spring constant Length of beam Characteristic length of an object Length (from anchor of beam) where a force is applied Length of beam of in-plane resonator Not fixed beam length Wavelength Magnetization Magnetization in x-, y-, z-direction Remanent magnetization Saturation magnetization Mass Additional mass on beam Mass per unit length of the beam Magnetic moment of magnetic nanoparticle permeability of free space Frequency of light Overlap of a plate Angular frequency Angular resonant frequency Angular natural frequency Correction factor for beams with rectangular cross sections Total stored energy during one cycle of oscillation Power loss due to relaxation process of nanoparticles

xvi

J/m3 J/m3 J/K N/m m m m

Contents

Symbols Symbol

Unit

Description

∆P ϕ q qL Q Qclamp Qintrinsic Qlat Qmedium Qmat QSader Qsur f ace Q TED r Re r.h. rH ρc ρ Fe3 O4 ρ f luid ρSU −8 s t θ θ XRD θd τB τe f f τN τ0 T Tm u( x ) UP uL V

J

Energy lost during one cycle of oscillation Phase lag Scattering vector q Force per length Quality factor Q-factor due to clamping loss Quality factor due to intrinsic damping Quality factor of a lateral in-plane resonator Quality factor in a medium e.g. liquid, air Q-factor due to the material damping Quality factor for cantilever in water Q-factor due to surface loss Q-factor due to thermoelastic damping Radius of an object Reynolds number Relative humidity level Hydrodynamic particle radius Mass density of composite Mass density of magnetite Mass density of fluid Mass density of SU-8 Length of plate of in-plane resonator Time Scattering angle Measurement angle for XRD Dynamic water contact angle Brownian relaxation time Effective relaxation time Néel relaxation time Time constant for Néel relaxation Temperature Torque Displacement of a beam Induced voltage Static deflection of a cantilever at the tip Volume

◦

N/m

m % m kg/m3 kg/m3 kg/m3 kg/m3 m s ◦ ◦ ◦

s s s s K Nm m V m m3

xvii

Contents

Symbols Symbol

Unit

Description

v0 v wB w wL ∆x p ξ Y Yc ζ

m/s

Free-stream velocity Poisson’s ratio Deflection of a beam Width of beam Width of a written composite/polymer line Lateral deflection of microresonator Volume fraction of nanoparticles in the composite Young’s modulus Young’s modulus composite Damping ratio of a system

xviii

m m m m GPa GPa

Contents

Abbreviations Symbol

Description

ABF AFM AGM DI DMA FEM LSV MEMS MPC PDDA PDMS PMMA PSS RMS SAXS SEM SQUID SWNT TEM TPP VSM Voxel WLI XDC 3D

Artificial bacteria flagellas Atomic force microscope Alternating gradient magnetometer De-ionized Dynamic mechanical analysis Finit element method Laser-Doppler vibrometer Micro electromechanical system Magnetic polymer composite Poly(dimethyldiallylammonium chloride) Polydimethylsiloxane Polymethylmethacrylate Poly(sodium 4-styrenesulfonate) Root mean square Small angle X-ray scattering Scanning electron microscopy Superconducting quantum interference device Single wall carbon nanotube Transmission electron microscopy Two-photon polymerization Vibrating sample magnetometer Volume pixel White ligth interferrometer X-ray disc centrifuge Three dimensional

xix

1 Introduction 1.1 Motivation The development of polymers has revolutionized our world, and they have become an integral part of our everyday life. Their low weight, variety of properties, 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, flammability, and low resistance to high temperatures. Around 1900, it was recognized that nanoparticles mixed with a selective addition to polymers change the characteristics of the polymer. For example carbon black, zinc oxide or magnesium sulfate particles mixed with vulcanized rubber were used to fabricate wear resistant 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 investigated 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 reinforced 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 composites 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 incorporated silica particle [5], reduced refractive index for wave guiding applications 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 polymers with magnetic fillers. Both materials have great advantages for microsystems as described in the following. 1

For simplification „micromechanical systems“ are also included in the term MEMS in this thesis.

1

1 Introduction Polymers offer a variety of low cost fabrication processes such as photolithograpy and hot embossing. Polymers have a wide range of available surface properties and can be biocompatible [8]. These render polymers attractive for use in microsystems in contact with environment-sensitive species such as living cells [9–11] or for biomedical applications like lab-on-a-chip applications [12] or inside the human body [13]. In addition, to the high chemical selectivity, polymers benefit from the low and varied Young’s modulus. Therefore, polymers are used in applications as actuators [9–11], actuators in microfluidic systems for large movements of micro-valves [12, 14, 15] and mimicking of muscle-like behavior [16, 17]. Microstructures made from magnetic materials like Fe, Co, Ni benefit from the contactless actuation, remote control over large distances and large actuation forces when permanent magnets are involved [18]. It is shown that magnetically generated forces can be made much larger than electrostatic forces for gaps in the micrometer regime [19]. Different kinds of sensors like cochlear implants, microswitches and microactuators have been developed [20]. Magnetic robots in the micro- and millimeter range can be wirelessly controlled in the human body [21, 22]. Furthermore, magnetostrictive material can be used to generate small deflections (elongation 0.2%), and magnetostrictive microstructures can be used as resonators for remote sensing applications [23, 24]. The advantages of polymers and magnetic material can be combined within one material by incorporating a magnetic filler into the polymer matrix. A magnetic polymer composite (MPC) microsystem benefits from the exceptional variety of physical and chemical properties of the polymer and the remote actuation of the magnetic filler. MPCs are interesting for microactuators to manipulate biological materials like cells, and for free swimming microrobots that are controlled by external magnetic fields for drug transport and release at a specific target area in the human body. Furthermore, MPC microstructures can be used in microfluidic applications as pumps and mixers, or as remote biosensors like microresonators in microfluidic channels.

1.2 State-of-the-art of MPCs for microsystem applications In this section the state-of-the-art of MPCs for use in microsystem applications are discussed. There are two ways to obtain a magnetic polymer. The possibility to fabricate pure organic polymers with intrinsic magnetic characteristics has been theoretically predicted [25]. However, today’s pure organic compounds are mostly diamagnetic or show weak antiferromagnetic behavior [26], and are therefore not suitable for magnetic actuation (the terms of magnetism are explained in Chapter 2).

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 processes of polymers compared to magnetic materials deposited at high vacuum (sputtering and evaporation) or using time-consuming electrodeposition processes. 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 microstructures 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 superparamagnetic). 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 magnetic particles with diameters > 1 µm.

3

4 Photodefinable Photolithography [28] Stereolithography [31] 2-Photon polymerization [34]

Mould

Hot embossing [27] Nanoimprint [30] Casting [33] Injection molding [35]

Laser Plasma etching

Ablation

Electroplating [29]

Bottom-up

Direct-shape Inkjet printing Fiber electrospinning [32]

Table 1.1: Overview of microfabrication methods for polymers and polymer composites.

1 Introduction

NiZn, MnZn

Samariumcobalt (Sm2 Co17 )

Strontium ferrite, NiZn

Neodym (NdFeB), strontium ferrite, samariumcobalt

Ferrite

PI-2555

SU-8

PI-2555

SU-8, PDMS

SCR770

FM: Ferromagnetic

Strontium ferrite (SrFe12 O19 )

PI-2555

a

Magnetic particles

Polymer matrix material

1.3 µm

1 – 9 µm

1 – 1.4 µm

10 µm

0.8 – 1.2 µm

1.15 – 1.5 µm

Size of magnetic particles

FM

FM

FM

FM

50 wt.%

35 – 90 wt.%

45 – 80 vol.%

18 – 60 vol.%

95 wt.%

80 vol.%

FM a

FM

Particle concentration

Type of magnetism

and

Microstereolithography

Photolithography, Screen printing replica molding

Screen printing and template printing

Photolithography

Screenprinting or spincasting followed by photolithography

Screenprinting or spincasting followed by photolithography

Fabrication method

3D-structures: microscrew, microfan (Sizes: > 250 µm)

Lorentz force actuators, micromotor, microgripper

-

Rotational speed microsensor

Microinductors

Microactuator

1998

Kobayashi 2008 [31]

Feldmann 2007 [39]

Rojanapornpun 2001 [38]

Dutoit 1999 [37]

Park [36]

Lagorce 1997 [33]

Structure/Application Reference

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.

1.2 State-of-the-art of MPCs for microsystem applications

5

6

Fe3 O4

γ-Fe2 O3

Fe3 O4

Ni

Ni

γ-Fe2 O3

Fe3 O4

PDMS

PDDA b, PSS c, SWNT d

Acrylic resin

SU-8

SU-8

SU-8, 1002F

Methyl acrylate, butyl methacrylate

f

e

d

c

b

10 nm

10 nm

100 nm

8 – 150 nm

50 nm

50 nm

10 nm

200 nm

SP

SP

FM

FM

FM

5 wt.%

1 wt.%

12.5 wt.%

1.3 – 3.3 wt.%

25 wt.%

not given

40 wt.%

SP f

FM

50 wt.%

Particle concentration

FM e

Size of Type of magn. magnetparticles ism

Polydimethylsiloxane Poly(dimethyldiallylammonium chloride) Poly(sodium 4-styrenesulfonate) Single wall carbon nanotube FM: Ferromagnetic SP: Superparamagnetic

Fe3 O4

PDMS a

a

Magn. particles

Polymer matrix material

Photopatternable: Twophoton polymerization

Photopatternable: Photolithography

Photopatternable: Photolithography

Photopatternable: Photolithography

Photopatternable: Microstereolithography

Non-photopatternable: Layer-by-layer nano self-assembly

Non-photopatternable: Spin-coating and bonding techniques

Non-photopatternable: Casting

Fabrication method

Microturbine; Size: 14 µm (diameter)

Pallets for cell sorting; Size: 3 x 3 x 12 µm (aspect ratio 4:1)

Torsion actuator; Size: 430 x 130 x 15 µm

Micromirror on cantilever; Size: 1000 x 300 x 70 µm

Flow sensor; Size: 8 x 8 x 4 mm

Cantilever; Size: 200 x 50 x 0.2 µm

Membrane; Size: 4000 (diameter) x 36 µm

Microvalve, microstirrer, cell loading unit; Size: >200 µm

Structure/Application

Xia 2010 [34], Tian 2010 [46]

Gach 2010 [45]

Tsai 2011 [44]

Damean 2005 [28]

Leigh 2011 [43]

Xue 2007 [42]

Pirmoradi 2010 [41]

Yamanishi 2007 [40]

Reference

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.

1 Introduction

1.2 State-of-the-art of MPCs for microsystem applications 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 materials. In the last few years new processes have been established, enabling costeffective 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 composite 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 microvalves, microstirrers, particle separators, cell sorting tools in a microfluidic channel 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 nanoassembly with a thickness of a few hundred nanometers (200 nm), which allows the fabrication of cantilever structures is presented in [42]. However, the fabrication process for this composite is very complex. One of the most promising microfabrication processes for polymers is patterning 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 agglomerates (> 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 turing by a laser using photopolymerization or two-photon polymerization (TPP). The composites fabricated using this technique are summarized for completeness in Table 1.3, however, discussed in detail in Chapter 8 in Section 8.2. From the various applications shown in the state-of-the-art summary such as microturbines, microstirrers, microvalves, microstructures for cell sorting, microgrippers, micromotors, and flow sensors, it can be concluded that there is a high potential for the use of tailored MPC for microsystems. A challenge is the further miniaturization of these devices, which are mostly in the size range of 0.2 – 5 mm. The fabrication and use of MPC microstructures with dimensions < 5 µm is very rarely explored. Furthermore, little attention has been paid to the particle dispersion and agglomerate sizes in the developed MPCs, despite this being one of the most significant issues to be addressed in the fabrication of nanocomposites. The particle dispersion and particle agglomerate sizes limit the minimum structure size of the final microstructures, and can influence their magnetic properties. Patterning of MPCs by photopolymerization is one of the most promising fabrication processes for MPCs, allowing cost-effective batch-fabrication of microstructures and the manufacturing of small features sizes (< 1 µm).

1.3 Goals of the thesis This thesis focuses on the exploration of polymer composites filled with magnetic nanoparticles for utilization in microstructures. The goals of this thesis are: • Developing a photocurable magnetic composite which allows the fabrication microstructures with dimensions < 5 µm, • Determination of the composite’s magnetic and mechanical properties as a function of particle concentration, such as saturation magnetization and Young’s modulus, • Determination of dispersion quality and agglomerate sizes in the MPC, • Evaluation of the fabrication limits for microstructures with different nanoparticle loading, • Fabrication of suspended MPC microcantilevers, • Investigation of the magnetic actuation of MPC microcantilevers and their use for remote controlled mass sensors,

8

1.4 Approach and outline of this work • Fabrication of a suitable polymer package for the resonators, • Statement about the advantages and limitations of such a photocurable MPC for use in microfabrication for future applications.

1.4 Approach and outline of this work Firstly, the theory about magnetism in nanoparticles and fundamentals of mechanical resonators are presented in Chapter 2. Secondly, the selection of the materials (polymer, magnetic particles and dispersion agent) is discussed in Chapter 3. The research approach of this work is illustrated in Figure 1.1. Magnetite Fe3 O4 nanoparticles were mixed with photocurable epoxy polymer SU-8 (dissolved in solvent) to form a stable magnetic suspension, which was then spincoated on a substrate. The composite can be polymerized and structured by exposure of UV light. The selection of the fabrication method and the processes for the fabrication of the composite, the microstructures and a suitable package are shown in Chapter 4. Spin-coated thin films of the MPC were used for the investigation of the material properties. In Chapter 5 the magnetic properties and the nanoparticle dispersion quality of the composite are highlighted and the investigation of UV transmittance of the composite with different nanoparticle concentrations are shown. Moreover, heating results of the composite using alternating magnetic fields are discussed, and the surface properties like biocompatibility and hydrophobicity are presented. Two kinds of microstructures were fabricated with two different fabrication processes to show the use and the potential of the photocurable MPC for various microsystems. Cantilevers were fabricated by UV standard photolithography. The magnetic actuation of the MPC cantilevers, as well their performance under different magnetic actuation configurations and in different media (vacuum, air and water) are presented in Chapter 6. The cantilevers were used further to characterize the mechanical properties of the composite such as dynamic Young’s modulus, as discussed in Chapter 5. To investigate the possibility of large deflections with MPC microstructures, and for a possible remote magnetic readout by an external pick-up coil, an in-plane microresonator is developed based on the same microfabrication process (Chapter 7). The fabrication of three dimensional (3D) microstructures by two-photon polymerization with the MPC and its fabrication limits and minimal features sizes depending on the nanoparticle concentration are presented in Chapter 8. Magnetic helical microstructures were produced, and the possibility of magnetic actuation and the control of such microstructures for use as artificial bacteria flagella are investigated.

9

1 Introduction

Figure 1.1: Approach of this work.

In Chapter 9 the achievements are summarized, evaluated and the applications of the developed MPC for further applications are outlined.

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 composites are discussed.

2.1 Magnetism in nanoparticles At length scale of a nanoparticle the magnetic properties of a material deviate from the bulk properties of that material. This section gives a brief introduction in magnetism and explains the magnetic behavior in small particles.

2.1.1 Units in magnetism

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

Magn. field strength Magnetization Magn. flux density

Symbol

SI unit

CGS unit

Conversion Factor

H

A/m

Oe

1 Oe = 1000/4π A/m

M

A/m Am2 /kg Tesla [T] (kg/As2 )

emu/cm3 emu/g Gauss

1 emu/cm3 = 1000 A/m 1 emu/g = 1 Am2 /kg 1 Gauss = 10−4 T

B

11

2 Theory 2.1.2 Magnetic field and magnetic flux density 1 Magnetism

originates from the movement of electric charges. From an atomic view of matter, there are two electronic motions; the orbital motion of the electron and the spin motion of the electron, which are the source of macroscopic magnetic phenomena in materials. The magnetic moment per unit volume of a magnetic material is determined by the magnetization M [A/m]. Furthermore, a magnetic field is generated by an electric current flowing in a wire, and is quantified by the magnetic field strength H [A/m]. The field due to electric currents and magnetization is described by the magnetic flux density B [T] 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)

2.1.3 Magnetic materials

Magnetic materials can be categorized depending on their χ and µ [51]: • Ferro- and ferrimagnetism: χ and µ are large and positive, both are functions of H, examples: Fe, Ni, Co • Paramagnetism: χ is small and positive, and µ is slightly higher than 1, examples: O2 , Cr, Ti • Diamagnetism: χ is small and negative, and µ is slightly lower than 1, examples: Bi, graphite 1

The information for this chapter is primarily from [50]

12

2.1 Magnetism in nanoparticles

Figure 2.1: Schematic depiction of spin arrangements in a ferromagnet and ferrimagnet.

Ferrimagnetic materials are similar to ferromagnetic materials containing sublattices that create an antiparallel alignment that diminish the net magnetization as schematically depicted in Figure 2.1. An example is Fe3 O4 magnetite. For simplification ferrimagnetic materials are attributed to ferromagnetic materials in this work. Ferromagnetic materials show the strongest magnetic effects. They show hysteresis under an applied magnetic field H. At high applied magnetic field they reach saturation magnetization Ms . Figure 2.2 shows the magnetization of a ferromagnet under an applied magnetic field H. The remanent magnetization Mr is the remaining field at zero applied magnetic field. The coercivity Hc describes the width of the hysteresis. For comparison the magnetic behavior of paramagnetic and diamagnetic material under an applied field are also schematically depicted. Figure 2.2 shows also the magnetic behavior of a superparamagnetic material. Superparamagnetism is a small particle effect of ferro- or ferrimagnetic material and is explained in the following section. 2.1.4 Small particle magnetism

Ferromagnetic bulk materials form domains with different magnetization directions to minimize the magnetostatic energy. A domain is a region within a magnetic material which has uniform magnetization. For this reason, it is possible that a piece of iron in the absence of an applied field at room temperature has no macroscopic total moment. Two opposite energies determine the formation of domains: the exchange energy at the boundary between oppositely aligned domains, and the energy gained due to the reduction of the total magnetostatic energy. The energy balance leads to finite domain sizes [50]. Figure 2.3 illustrates the magnetic behaviors of small particles dependent on the particle size. If the particle diameter is reduced below a critical diameter, Ds , the energy required to form a domain wall is larger then the benefit of the reduction of the external magnetostatic energy. This is called the single domain state. The critical diameters for different magnetic materials are listed in Table 2.2. Magnetic particles exhibit maximal coercivity in single domain state at the critical diameter. Figure 2.4

13

2 Theory

S a tu r a tio n M a g n e tiz a tio n M

M a g n e tiz a tio n , M

R e m a n e n t M a g n e tiz a tio n M

F e r r o m a g n e t ic s

r

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

Table 2.2: Single-domain size for spherical particles. [52]

Material

Single-domain size DS [nm]

Fe Co Ni Fe3 O4 γ-Fe2 O3

14 70 55 128 166

Figure 2.3: Domain creation (for cubic crystals) in absence of external applied field. Singledomain 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 energy 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 anisotropy energy: E(Θ) = Ke f f V sin2 Θ

(2.5)

where Ke f f is the effective anisotropy constant, V is the particle volume and Θ is the angle between the moment and the easy axis. An easy axis is an energetically favorable direction of spontaneous magnetization. The energy barrier KV hinders the magnetization from flipping between two energetically identical easy directions of magnetization in a nanoparticle. If k B T > Ke f f V the magnetization can change spontaneously from one easy direction to the other [56] and the sys-

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 superparamagnetism [50]. Such a system has no hysteresis, as schematically depicted in Figure 2.2, and the data of different temperatures superimpose onto a universal 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 certain orientation is described by the Néel relaxation,τN , [56] Ke f f V

τN = τ0 e

kB T

(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 magnetometer 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].

2.2 Mechanics of oscillating structures 2.2.1 Harmonic oscillator 1

The basic model for describing oscillating mechanical systems is the harmonic oscillator model, also known as mass-spring-damper-system. It is composed of a moving mass coupled with a linear spring and a linear damper, and is represented by the second order differential equation for a externally excited oscillators m x¨ + c x˙ + 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 ω02 =

k m

(2.8)

and the damping ratio of the system is given by ζ=

c 2mω0

(2.9)

For steady vibration and slight damping, the equation (2.7) results in amplitude B of B= q

F0 /m

(ω02 − ω 2 )2 + (2ζω0 )2 ω 2

(2.10)

where the phase lag ϕ between the mass dispacement and excitation force is ! 2ζω0 ω −1 . (2.11) ϕ = tan ω02 − ω 2 A slightly damped mechanical system driven by a sinusoidal input can be described by the natural frequency w0 and the damping ratio ζ. The resonant fre1

The theory of this section is mainly based on [59] and [60].

17

2 Theory

Figure 2.5: Schematic of a linear damped harmonic oscillator with one degree of freedom.

quency ωr can be found at ∂B/∂ω = 0 and is ωr = ω0

q

1 − 2ζ 2

(2.12)

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. 2.2.2 Beam theory

A single-clamped cantilever, shown in Figure 2.6, can be described by EulerBernoulli beam theory when the transverse dimensions h and w of the beam are small in proportion to the length L (L/h > 10). In this theory shear stress, rotational inertia and damping losses are neglected and the plane sections of the cantilever must remain plane and normal to the longitudinal axis. These can be assumed for small cantilever deflections u (u < h). A cantilever can be described by Euler-Bernoulli theory for the cases of static and dynamic deflection. Resonance of a beam

By assuming only small deflections u( x, t) and linear material properties, the equation of motion can be derived from the equilibrium of forces for an infinitesimal piece of beam with no external load

18

2.2 Mechanics of oscillating structures z y

h x L

w

Figure 2.6: Ideally clamped Euler Bernoulli cantilever.

ρA

∂2 u ∂4 u + YIz 4 = 0 2 ∂t ∂x

(2.13)

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 Iz =

Ah2 12

(2.14)

where h represents the beam thickness. By solving equation 2.13 through separation with variables, the eigenfrequency of a thin beam is λ2 ω0 = 2π f 0 = n2 L

s

YIz , n = 1, 2, ... Aρ

(2.15)

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 solutions

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 λ2 h fn = n 2 2πL

s

Y , n = 1, 2, ... 12ρ

(2.17)

The Young’s modulus can then be calculated as follows  Y = 12ρ

2πL2 f n λ2n h

2 (2.18)

Static cantilever deflection

z F

LF

x

Figure 2.7: Deflection of a cantilever beam under influence of an external force.

The curvature of the beam under a small displacement can be described by a second-order differential equation:

YIz

∂2 u ( x ) = M( x) ∂x2

(2.19)

where Y represents the Young’s modulus, Iz the second moment of inertia, u( x ) the cantilever deflection, and M( x ) the bending moment at the cross section at location x. For the simple case of an end-loaded force condition, depicted in Figure 2.7, and with the two boundary conditions at the fixed end

u( x )| x=0 = 0

∂u( x ) =0 ∂x x=0

(2.20)

the equation (2.19) can be solved. The deflection of the beam for a single point

20

2.2 Mechanics of oscillating structures force at a distance L F from the fixed end can be calculated [59] as

umax =

FL3F 3YIz

(2.21)

For small deformations, the displacement and the applied force follow a linear relationship by Hooke’s law:

k( x) =

F u( x )

(2.22)

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:

k( L F ) =

3YIz L3F

(2.23)

Using the obtained equation (2.23) and inserting the moment of inertia Iz (2.14), the Young’s modulus can be expressed as

Y=

4kL3F wt3

(2.24)

Resonance of a beam with end mass

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 potential energy of the resonator the natural frequency of the first mode can be derived s ω0 =

Ywh3 33 4(m a + 140 m b ) L3

(2.25)

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 ω0 =

Yhw3 33 4(m a + 140 m b ) L3

(2.26)

2.2.3 Quality factor of oscillating structures

Mechanical resonance can be significantly damped by the presence of air or liquid media surrounding a resonant structure. The amount of damping can be described by the quality factor, Q. In physics, the Q-factor is defined as the ratio between the energy stored and the average energy loss [60]. Q = 2π

Estored Eloss

(2.27)

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] Q=

1 2ζ

(2.28)

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 Q=

f res ∆ f −3dB

with f res as resonance frequency and ∆ f −3dB as bandwidth with

(2.29) √1 2

amplitude.

2.3 Young’s modulus of composites 1 One

of the simplest theories about particle reinforcement of composites is based on Einstein’s equation for the viscosity of a suspension of rigid spherical inclusions [62] for low volume fractions (< 1 vol.%). The expression has been extended analytically to describe higher filler fractions and particle interactions [63, 64]. The macroscopic behavior of composites is affected by the size, the shape and the distribution of the particles, and the interfacial adhesion between the particles 1

This section is mainly based on [61]

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].

2.3.1 Hashin-Shtrikman Model

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 ) Gi + ξ j + ( Gj − Gi ) 5Gi (3Ki + 4Gi )   −1 1 3ξ i Ki + ξ j + (Ki − K j ) (3Ki + 4 Gi ) 

Gc

=

Kc

=

(2.30) (2.31)

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.

G=

Y 2(1 + ν )

(2.32)

K=

Y 3(1 − 2ν)

(2.33)

23

2 Theory The upper and the lower boundary for the Young’s modulus is given by [61]: Yc =

24

9Kc Gc 3Kc + Gc

(2.34)

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. Furthermore, a suitable dispersant agent is evaluated to obtain a high quality particle dispersion in the final composite.

3.1 Polymer evaluation 3.1.1 Polymer selection criteria

First of all the composite has to be processable with common microelectromechanical system (MEMS) technologies, which includes the possibility of batch fabrication. This requires good spin-coating performance without the formation of cracks. The polymer of the composite microstructures must have high chemical resistance to be able to work in different environments and to allow cleaning procedures of the microstructures. The material ideally shows high fracture strength, low creep, no plastic deformation, a high enough Young’s modulus to guarantee mechanical stability of the fabricated suspended microstructures and little chemical and physical aging. High temperature stability of the material allows the performance of chemical and biologically reactions directly on the polymer microstructures such as polymerase chain reaction (PCR) processes where the temperature reaches 100◦ C. The surface of the MPC microstructures should be easily chemically modified to enable biofunctionalization in order to use the microstructures in biological relevant applications. For this purpose the polymer must be biocompatible to allow contact to biological environments like cells and bacterias.

3.1.2 Polymer selection

The goal is to mix a composite with photodefinable property. There are a variety of non-photodefinable polymers which fulfill the required criteria and where a photoinitiator can be mixed to make them photosensitive. However, the photoinitiator 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 The photodefinable polymers can be categorized in three classes. Single component systems (no sensitizer), positive, and negative tone two-component polymer systems [66]. Single component systems need high radiation energy for structuring and are usually structured by deep UV radiation, electron beam or ion beam. These processes are rather expensive and uncommon for applications in industrial fabrication. For single component polymers, such as PMMA, it is favorable to use hot embossing for microfabrication [27]. Two-component positive tone photopolymers contain a photosensitive compound, usually a photoacid generator, which cascade a chemical transformation in the polymer and alter the solubility of the exposed regions (breaks the polymer chains) [67]. Because of the possibility of breaking chains rather than crosslinking, positive tone photodefinable polymers show often poor chemical resistance to organic solvents and the long term mechanical performance are often unsatisfying. Two-component negative tone photodefinable polymers contain a photosensitive cross-linking agent which triggers a chemical reaction to cross-link the exposed area. This cross-linking can be very strong depending on the number of the binding sites of the monomer leading to a chemically and mechanically robust polymer. Therefore, negative photodefinable polymers are preferred for the choice of the composite matrix material. Two negative tone photodefinable polymer types, epoxy and polyimide, were taken into account for the choice of matrix material of the composite for the fabrication of magnetic microstructures. Polyimides and epoxies are highly crosslinked 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 technology [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]. Table 3.1 compares two commercially available photopatternable polymer products HD-4100 (Polyimide) and SU-8 (Epoxy). Both, have similar properties and are suited for matrix material for a MPC for the fabrication of magnetic microstructures. For both materials composites with magnetic nanoparticles have been reported. Finally, SU-8 was chosen as matrix material because it exhibits better functionalization possibility due to the large amount of epoxy binding sites [70–72]. SU-8 is well known as structural material in MEMS [9–11], shows high aspect-ratios [73] and its fabrication processes are well known.

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 temperature Tg , > 100 ◦ C enables PCR reactions

330 ◦ C

210 ◦ C [75]

Functionalization possibility for biomolecules

Possible but complex [77]

Well known [70–72]

Biocompatibility (in vitro)

Not known

Cell proliferation reported [78, 79]

Processability with spin-coating

Yes

Yes

Solvent

N-Methyl-2-Pyrrolidone

Cyclopentanone, GBL

[74]

High chemical sta- Yes [74] bility

Yes [75]

Reported compos- [33, 80, 81] ites containing magnetic particles

[4, 28, 44, 45]

27

3 Evaluation of materials SU-8: Polymerization and properties

SU-8 was developed by IBM research and it is commercially available from MicroChem and Gersteltec. It is based on Epon SU-8 from Shell Chemical (glycidyl ether derivative of bisphenol-A novolac), which can be dissolved in an organic 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 (CYRACURE UVI from Union Carbide) 10 wt.%, which is dissolved in propylene carbonate [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 nearUV 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 sulfonium 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 properties 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.

3.2 Particle evaluation For the fabrication of composites for magnetic microstructures with smallest structure dimensions < 3 µm (thickness of cantilevers), the particles must have dimensions 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). Different 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 interactions 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 binding 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 suitability of the composite and are listed in Table 3.2. To investigate their magnetic properties the nanoparticles have been characterized by a vibrating sample magnetometer (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 ) nanoparticles are below the critical particle size and exhibit superparamagnetic properties. They show negligible or low remanent magnetization. Figure 3.2 shows the schematics of the magnetic characteristics of typical ferromagnetic nanoparticles (Ni particles with a diameter ∼20 nm, single domain, blocked) in comparison to superparamagnetic nanoparticles (Fe3 O4 with a diameter < 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 attraction in a liquid polymer matrix. They can easily form agglomerates with sizes bigger 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 Chemicell GmbH have a much higher saturation magnetization compared to the γFe2 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

Ferromagnetic particles F F

Ferromagnetic particles

Mr

0

Superparamagnetic particles

Superparamagnetic particles

0

Applied Field, H Figure 3.2: Schematics of the magnetic characteristics of ferromagnetic (blocked) and superparamagnetic (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, superparamagnetic particles have negligible remanent magnetization at room temperature and therefore low magnetic attraction in dispersion. The reason for the low interaction is based on the thermal energy, which flips the direction of the magnetization of c IOP.) the single domain nanoparticles. (Adapted from [87],

the particles. Therefore, the Fe3 O4 nanoparticles from Chemicell GmbH are selected as filler material for the photocurable MPC.

31

32

Superparamagnetic

Superparamagnetic

Ferromagnetic

Ferromagnetic

23 wt.% silica coated γ-Fe2 O3 [48]

MagSilica 50 (γ-Fe2 O3 ) from Evonik [48, 88]

Ni NanoAmor (Houston, USA)

Co-carbon coated from TurboBeads [89]

b

measured by TEM given by supplier

Superparamagnetic

Fe3 O4 from Chemicell GmbH

a

Magnetic behavior

Material

53.5

32

24

58

145

12.4 a

23 a

5 - 15 b

∼20 b

∼50 b

[nm]

Saturation magnetization @ ∼800 kA/m [Am2 /kg] (emu/g)

Particle diameter

32

45

8

8

<1

[kA/m]

Coercivity

21

23

4.1

6.6

<1

[Am2 /kg] (emu/g)

Remanent magnetization

Very high saturation magnetization

High saturation magnetization

Silica coating

Silica coating

Negligible remanent magnetization and high saturation magnetization

Advantages

High magnetic attraction forces between particles

High magnetic attraction forces between particles

Low saturation magnetization

Low saturation magnetization

Disadvantages

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.

3 Evaluation of materials

3.3 Dispersant agent evaluation

3.3 Dispersant agent evaluation The selected nanoparticles must be well dispersed within the polymer to achieve homogeneous mechanical and magnetic properties in the composite. Nanoparticles 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 nanoparticles and present the selection of a suitable dispersant agent. 3.3.1 Stabilization of particles

Two mechanisms exist to prevent particles from approaching. Firstly, electrical charge on the particle surface can create a repulsion. Ultra stable suspensions, so called ferrofluids, can be produced using the electric charge stabilization technique [90]. The stabilization is based on an electric-double-layer mechanism which is established on the colloid surface. However, the stabilization depends on the pH-value of the liquid and for certain pH-ranges the particles coagulate owing to insufficient surface charge. To achieve stability of nanoparticles in a polymer composite with particles stabilized by electrical charge is difficult because the polymer often contains complex chemical compounds and the used solvents are often organic liquids with undefined pH-values. Secondly, with long chain molecules adsorbed onto the particle surface, so called surfactants, steric repulsion between particles can be achieved to prevent particles from agglomerating. Figure 3.3 shows the Van der Waals attractive energy and magnetic attractive energy for magnetite nanoparticles. Coating the nanoparticle with a surfactant, steric repulsion can result in a net potential energy that is decisive for determining monodispersity of the particle suspensions. Furthermore, by choosing an appropriate end group of the surfactant the interaction between surfactants and the polymer matrix can be tuned. In this work a surfactant was used to stabilize the nanoparticles in the composite.

33

3 Evaluation of materials

Figure 3.3: Potential energy versus surface-to-surface separation of sterically protected colloidal magnetite particles with radius r = 5 nm. δ is the length of the adsorbed molecules and s is the surface-to-surface separation of the particles. Reprinted with c Dover. permission from [91],

3.3.2 Selection of suitable surfactant on magnetite particles for photocurable epoxy matrix

An adequate dispersant agent must fulfill following requirements: • strong attachment to the nanoparticle surface, • chemical affinity with the solvent (the particle must be stable in the polymer solvent up to the desired concentration), • chemical affinity with the surrounding polymer matrix. To obtain a robust functionalization a strong attachment of the surfactant to the nanoparticle surface (for example by a covalent bond) is desired. A high density surface layer of the surfactant on the particle is necessary to obtain a strong steric repulsion between the particles and enables stable dispersions in liquid with high particle concentrations. As described in Section 4.1.1 it is favorable to stabilize the nanoparticles first in the polymer solvent and mix this initial suspension with the dissolved polymer. A high particle concentration in the initial suspension is necessary to reach a high particle concentration in the composite. Furthermore, the surfactant on the particles must be compatible with the solvent and the polymer.

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: Diphosphate (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 concentration 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 dispersion 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.

Figure 3.4: Schematic of Fe3 O4 nanoparticle.

surfactant

containing

phosphate

end

group

on

a

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

Figure 3.5: Light microscope images in transmission mode of a polymerized composite containing 2 vol.% (8 wt.%) Fe3 O4 particles with a film thickness of 1.75 ± 0.15 µm. (a) Magnetite particles are coated by diphosphate and form agglomerates (mean diameter > 5 µm) in the polymer. (b) If the magnetic particles are coated with a dispersant agent containing a copolymer with a phosphate group, agglomerates are no long visible unc (2011), with permission from Elsevier). der light microscopy ([93],

3.4 Conclusion In this chapter the suitable components such as the polymer, the magnetic nanoparticles and a dispersant agent were evaluated. For the polymer matrix material 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. Superparamagnetic Fe3 O4 particles with diameters of ∼13 nm were selected for the filler material. The particle have a high saturation magnetization (53.5 Am2 /kg) and have negligible remanent magnetization (< 1 Am2 /kg). Therefore, the superparamagnetic particles have low magnetic attraction compared to ferromagnetic particles during and after mixing of a polymer composite and the formation of agglomerates can be reduced. The influence of different dispersing agents on the nanoparticles 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 composite are presented in this section. A fabrication process using spin-coating, photolithography and a sacrificial layer etch is used to fabricate suspended microstructures. The schematic of the process with the selected material components to obtain suspended microstructures (cantilevers) is shown in Figure 4.1. Microcantilevers and lateral microresonators are fabricated with the developed magnetic polymer composite (MPC). To make the handling easier and allowing measurements in water the fabricated microresonators were directly packaged into a polymer channel. The fabrication of the all-polymer package and the integration of the microresonators using batch-process is shown in the last section.

Figure 4.1: Superparamagnetic magnetite nanoparticles containing a dispersing agent with a phosphate end group are mixed with the UV-sensitive epoxy SU-8 to obtain a magnetic composite. With this magnetic photocurable composite, suspended microstructures have been fabricated using conventional microfabrication processes such as c (2011), with permission from Elsevier). spin-coating and photolithography ([93],

1

Some parts of this chapter are published in [87, 93, 94].

37

4 Fabrication

4.1 Composite fabrication 4.1.1 Evaluation of mixing method for composite

One of the main challenges in mixing a magnetic composite is to disperse the inorganic filler within the organic polymer. Clusters of particles and agglomeration have to be minimized in order to fabricate micro-sized structures. Magnetic materials in general absorb light in the UV range. Larger agglomerates in a composite can shield the light and prevent the photoreaction [95]. For smaller particles (< 1 µm) the tendency of particle agglomeration increases because of increased Van der Waals forces between particles due to the increased surface area. Conventional technologies (mixing powder directly into polymer by stirrer or extruder) cannot break agglomerates sufficiently, because of the high surface energy of the nanoparticles [81]. Hence, these technologies are mostly not suitable to obtain homogeneous nanoparticle dispersions and high filler concentrations in nanocomposites. High speed centrifugal mixing and sonication steps provide high shear forces and lead to much better dispersions and high filler concentration [96, 97]. A promising approach is to start from a stable nanoparticle suspension and mix with a dissolved polymer or monomer (solution method) [81]. When the monomer solution and the stable nanoparticle dispersion are based on the same solvent they can be mixed easier. The dissolved monomer solution and the nanoparticle suspension need a low viscosity to allow a fast mixing. The already stabilized particles in the nanoparticles suspension enable a homogeneous dispersion. This method is suitable for fabricating positive tone photodefinable composites because dissolved long-chain polymer, which break during the chemical reaction after UV exposure, can be mixed with the stabilized particle suspension. Another promising approach is to mix the stable suspension of nanoparticles with a monomer and start its polymerization around the filler using UV exposure (solution method combined with in situ polymerization). Because of the monomers (short molecules) the viscosity of the polymer solution is low, allowing an efficient mixing. With this method negative tone photodefinable composites can be produced. This method presumes the compatibility of the suspension and the polymerization reaction. In this work the solution method combined with in situ polymerization was used to mix a negative photosensitive composite combined with centrifugal mixing and ultrasonic step. For the matrix material the commercially available photocurable epoxy SU-8 (50) was taken and for the filler material superparamagnetic Fe3 O4 nanoparticles were selected.

38

4.2 Microstructure fabrication 4.1.2 Preparation of magnetic suspension

Fe3 O4 nanoparticles were fabricated and stabilized in the organic solvent γ-Butyrolacton (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, Germany). A wet-chemical synthesis from iron-II/III-salt dissolution is used to fabricate 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.%. 4.1.3 Composite mixing

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.

4.2 Microstructure fabrication The fabrication of microcantilevers with the developed superparamagnetic composite is based on two conventional photolithography steps and a sacrificial layer etch step. In Figure 4.2 the fabrication process is illustrated. The sacrificial layer contains an evaporated film stack of chrome (Cr) (5 nm), gold (Au) (50 nm) and Cr (50 nm). The first Cr layer is used for better adhesion to the silicon wafer, whereas the following two layers form an enhanced sacrificial layer [98]. These layers allow a fast release of the structures from the support wafer. Layers with different thicknesses (1.8 µm to 3.5 µm) of the mixed composite are spin-coated

39

4 Fabrication on wafers and baked for 14 min at 100◦ C. The film is patterned with photolithography 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 structure, 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 unavoidable 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 suspended 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 dissolving 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 microstructures 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 scanning electron microscope (SEM) image of a released composite cantilever with 5 vol.% (18 wt.%) of magnetite nanoparticles. Cantilevers with the following geometries and particle concentrations were fabricated Table 4.1.

Table 4.1: Geometrical dimensions and particle concentration of MPC cantilevers.

Length L

Width w

Thickness h

Concentration

30 – 400 µm

14 µm

1.6 – 3.5 µm

0, 1, 2, 3, 5 vol.%

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 dimensions 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 photosensitive 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

SU-8 Photocurable magnetic polymer composite (MPC)

Cr/Au/Cr sacrificial layer Silicon wafer

Figure 4.2: Schematic of the process flow for the fabrication of magnetic polymer cantilevers: 1) evaporation of sacrificial Cr/Au/Cr layer, 2) spin coating of photocurable magnetic polymer composite and patterning by UV exposure, 3) development of composite 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 cantilever by etching of sacrificial layer. The overlapping anchor layer ensures a defined c IOP.) anchor for the microcantilever. ([87],

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.

42

Beam length LB

Beam width w

Plate length s

Thickness h

Concentration

10 - 600 µm

3 - 25 µm

3 - 400 µm

1.9 - 3.5 µm

0, 1, 2, 3, 5 vol.%

4.3 Polymer package fabrication

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 fabricated 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: • Small distance between resonator and external coil, • Fitting into the different actuation setups, • Optical transparency for optical readout (laser Doppler vibrometer), • Batch-fabrication (to reduce fabrication effort, low-cost). 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 resonators 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 package 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 reported [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

Figure 4.4: All-polymer package for cantilever array.

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 pressure 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 actuation 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 lowviscous MPC suspension with nanoparticle concentration up to 10 vol.% was obtained which can be used for spin-coating processes. Suspended microstructures, with 5 vol.% nanoparticle concentrations, such as cantilevers and in-plane resonators 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 In this chapter, an exhaustive characterization of the composite material and its processability is carried out. The properties of the composites have to be well understood to evaluate possible applications of MPC microstructures. Since particle agglomeration is one of the most significant issues to be addressed in the fabrication of nanocomposites, a focus on agglomerates as a function of particle concentration from 1 to 10 vol.% (4 - 32 wt.%) is investigated. Despite the existence of some works dealing with the fabrication of composites of photopatternable matrices containing nanoparticles [101], there is scarce literature regarding the dispersion of nanoparticles inside such matrices [95]. Relevant physical properties of the composite such as magnetism and mechanical properties such as Young’s modulus are also characterized as a function of particle concentration. Since such properties play an important role for the use of the material in its actuation. Nanoparticles in a photocurable composite can absorb and sheald UV light [95]. Therefore, the influence of the nanoparticle concentration on the exposure doses and polymerized film thicknesses are studied for the fabrication of microcantilevers. Furthermore, the composites’ feature resolution is investigated. The influence of particle concentration on the fabrication process for microcantilevers is explored. A higher particle concentration increases the viscosity of the initial non-cross-linked composite and must be considered for the spin coating process. The incorporated magnetic nanoparticles not only provide the composites’ magnetic properties, they also provide the function for heating of the composite using a high frequency applied external magnetic field. This heating effect is also studied to evaluate its impact for novel applications. In the last part of this chapter the composites’ surface polarity is analyzed since the wettability of the surface will determine the further material functionalization. Considering the possibility to use this composite for bioapplications, the compatibility of the composite with human cells has been tested.

47

5 Evaluation of composite properties

5.1 Particle dispersion in composite 5.1.1 Agglomerates and particle sizes

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 several 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. Results and discussion Investigation of the composite dispersion with TEM images was performed. Figure 5.1 shows a TEM image of the cross-section of a composite layer containing 3 vol.% (12 wt.%) and 5 vol.% (18 wt.%) magnetite particles. Analyzing several TEM cross-sections of the composite it can be observed that at the bottom interface composite-metal and top interface composite-air particles accumulate slightly as depicted in Figure 5.1 (a). This is probably caused by diffusion and dryout effects of the nanoparticles during spin-coating. In Figure 5.2, a close up view of the composite is depicted. A clear dispersion of Fe3 O4 particles in the SU-8 matrix with a low level of agglomerates can be observed. The agglomerate (minimum two particles) diameter distribution of a composite with 5 vol.% particle concentration is measured from TEM images (366 counts) and shown in Figure 5.3 (b). The agglomerate’s average count diameter is 43.5 ± 20.4 nm. The particle diameter distribution measured from TEM images is depicted in Figure 5.3 (a). The particles’ average count diameter is 11.4 ± 3.4 nm (800 particles

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 c (2011), with permission from Elsevier). thickness ≈ 100 nm). ([93],

49

5 Evaluation of composite properties

Figure 5.2: TEM image of the particles/agglomerates in the composite c (2011), with permission from Elsevier). 5 vol.% Fe3 O4 nanoparticles ([93],

with

measured from samples with 1 vol.% and 5 vol.% particle concentrations). From a two-dimensional TEM image, it is difficult to estimate the agglomerate sizes because the particles and agglomerates can overlap within the approximately 100 nm thick slice. SAXS Analysis 1 In

order to quantitatively determine the size distribution of particles and agglomerates, small angle X-ray scattering (SAXS) measurements were performed. SAXS measurements provide information about shape and the size of incorporated 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 using 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 distribution 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 distribution from TEM images (366 counts) of a composite with 5 vol.% particle concentrac (2011), with permission from tion. The average diameter is 43.5 ± 20.4 nm. ([93], Elsevier).

to determine the agglomeration behavior depending on the fill level. The SAXS equipment consists of a S-MAX3000 instrument from Rigaku Innovative Technologies (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 mounted 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 accessible scattering vectors q is defined as q=

θ 4π sin λ 2

(5.1)

and is 0.07 < q < 4.5 nm−1 , with λ being the x-ray wavelength, and θ being the scattering 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

I(q ) [a .u .]

5 Evaluation of composite properties

1 0

-1

1 0

-2

1 0

-3

1 0

-4

P a r tic le c o n c e n tr a tio n 1 v o l% 2 v o l% 3 v o l% 5 v o l% 1 0 v o l%

0 .1

1 q [n m

-1

]

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 c (2011), with permission from Elsevier). are independent of the concentration ([93],

ical thickness between 1.54 µm and 2.23 µm were measured using a transmission geometry with a measurement time of one hour. Results and discussion Characteristic curves that provide information on the agglomeration state of the particles for different particle concentrations in the composite are shown in Figure 5.4. The SAXS curves show two distinct slopes which indicate the presence of two main filler sizes in the composite. For samples containing 1 to 10 vol.% of Fe3 O4 the SAXS curves overlap, indicating that the agglomeration states of particles in the composite are independent of the particle concentration. For increasing particle concentration the noise of the SAXS curve is lower because of the higher scattered intensity. The SAXS curves clearly show that the particles interact via an effectively repulsive interaction potential. Even at the highest particle concentration the nanoparticles are still well dispersed. The sample containing 3 vol.% (12 wt.%) of Fe3 O4 was analyzed in detail. Two different approaches to calculate the size distribution of the scattering objects

52

D iffe r e n tia l V o lu m e D is tr ib u tio n [a .u .]

5.1 Particle dispersion in composite

1 .0 f r o m m o d e l f ittin g ( b ilo g n o r m d is tr ib u tio n ) f r o m IF T ( m o d e lf r e e d is tr ib u tio n )

0 .8

1 3 n m

0 .6 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 ]

Figure 5.5: Differential volume distribution of particle and agglomerate in composite c (2011), modeled from SAXS measurements. Measurement range: 1 – 100 nm ([93], with permission from Elsevier).

(i.e. single particles or agglomerates) were used. First, an Indirect Fourier Transformation (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 distributions 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 nanoparticle dispersion (magnetite particles dispersed in the solvent γ-Butyrolacton, GBL) X-ray disc centrifuge (XDC) analysis have been performed. The suspension 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 sedimentation. 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 photosensitive polymer SU-8. XDC measures hydrodynamic diameters of agglomerates including the surfactant layer and the fluid boundary layer [103]. Therefore, the sizes are expected to be larger compared to TEM and SAXS. The measurement 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 fundamental parameter approach with the Rietveld refinement [104], crystallite sizes of Fe3 O4 (ICSD 028664) can be determined using TOPAS 3.0 (Bruker). The average 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

D iffe r e n tia l V o lu m e D is tr ib u tio n [a .u .]

5.1 Particle dispersion in composite

1 .0 In itia l p a r tic le d is p e r s io n in G B L

0 .8

1 9 n m 0 .6

3 0 - 4 5 n m 0 .4 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.

5.1.2 Discussion and conclusion

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 diameters 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 diameter measured in the composite by TEM and SAXS. Considering the hydrodynamic layer, a slightly larger diameter from the XDC measurements is expected 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 initial 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 increase significantly. The developed photosensitive MPC in this work contains agglomerates in the range of 50 nm and is suitable for the fabrication of structures with a feature size below 5 µm. Table 5.1: Comparison of particle and agglomerate diameters measured by different measurement methods.

Measurement method

Particle diameter [nm]

Agglomerate diameter [nm]

XRDCrystalsize TEMVolumeaverage SAXS c XDC Hydrodynamicdiameter d

13.1 12.4 a 13 19

52 b 40 – 50 30 - 45

a b c d

taken from 800 measurements taken from 366 measurements main peaks from Figure 5.5 main peaks from Figure 5.6

Comparison with reported magnetic photocurable composites In literature homogeneous dispersed γ-Fe2 O3 nanoparticles with minimal aggregation using oleic acid as a surfactant have been reported [45]. However, the particle concentration is very low (0.01 – 1 wt.%). A high particle concentration in the polymer is crucial to obtain sufficient forces for the actuation of magnetic composite microstructures. Damean [28] fabricated a SU-8–Ni film with 0.18 – 1.8 vol.% ferromagnetic Ni nanoparticles with particle diameters of 80 – 100 nm. Tsai [44] fabricated SU-8–Ni film with 12.5 wt.% with particles with 100 nm in diameter. They both have not investigated the agglomeration behavior. However, from the presented microscope pictures the agglomerates are deduced to be in the range of micrometers. Such agglomerates are not suitable for the fabrication of structures with structure dimension smaller than approximately 5 µm.

56

5.2 Limitation using UV exposure

5.2 Limitation using UV exposure The goal is to structure the SU-8 based nanocomposite by conventional photolithography (exposure through a mask with a UV lamp). First exposure tests of the composite have shown that the composite is less transparent to UV light compared to unfilled SU-8, and results in an unexposed area at the bottom side of the layer. The layer peels off during the developing process step. This is due to the absorption and scattering of UV light by the Fe3 O4 nanoparticles (which is also the case for other inorganic particles [95]). Therefore, the thickness of the composite layers are limited by the particle concentration in the composite and the exposure dose has to be adjusted. In this section a detailed investigation of the maximal layer thicknesses and the UV transmittance of the composite is done. The necessary exposure doses for different layer thicknesses are determined and particle concentrations are evaluated to obtain microstructures with smooth top and bottom surfaces. 5.2.1 Backside exposure of composite film

A possibility to find out the maximum polymerized thickness, which can be obtained 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

A thick film 28.7 µm ± 1 µm (1000 rpm) of 2 vol.% composite is spin-coated on a fused silica glass wafer and exposed from the back-side with different UV doses. Fused silica is used as substrate to minimize the UV light absorbance. After developing, the thickness of the exposed film is measured with a profilometer Tencore P10. The surface of the film is investigated by a white light interferometer (WLI) and by SEM observations. Results and discussion

The maximum exposed film thicknesses of a 2 vol.% composite depending on the exposure dose are shown in Table 5.2. With increasing exposure doses the

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 [J/cm2 ]

Thickness [µm]

Stdev [µm]

0.1 0.5 1 5

0 (peeled off) 1.53 3.42 6.48

0.19 0.18 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 surface 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 maximizing 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 cantilevers. 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 exposure 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.

5.2.2 UV transmittance depending on particle concentration

The fabricated composite should have a particle loading as high as possible to obtain high magnetic forces on the fabricated composite microstructure. The necessary UV doses to obtain microstructures with smooth top and bottom surfaces for different nanoparticle concentrations using a constant layer thickness are investigated. Experimental

The SU-8 based nanocomposite is spin coated and structured by conventional photolithography (exposure through a mask with a UV lamp). For the composite exposure, a mercury lamp is used. The mercury lamp has two main spectral lines: i-line 365 nm and h-line 405 nm (no filter is used). Magnetite particles in the

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 parameters 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. Results and discussion

Light transmittance measurements in a range of 280–800 nm of pure SU-8 and composites with different particle concentrations are shown in Figure 5.8. The oscillation of the transmittance signal is due to a Fabry-Perrot effect of the thin

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

1 0 0

h - lin e

T r a n s m itta n c e ( % )

8 0 0 v o 1 v o 2 v o 3 v o 5 v o 1 0 v

6 0 4 0 2 0

l.% l.% l.% l.% l.% o l.% 2

s u c c e s s f u l w ith 1 0 J /c m 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

2

8 0 0

W a v e le n g th ( n m ) Figure 5.8: UV transmittance measurements of composite with increasing Fe3 O4 particle concentration. The sinusoidal distortion of the signal is based on a Fabry-Perot effect which occurs when the thickness of the film is a multiple of the half wavelength. The c (2011), thickness and the exposure dose of the samples are listed in Table 5.3 ([93], with permission from Elsevier).

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 composite 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 exposure 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 regions 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 dispersion 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 Figure 5.9 (b) the bottom side of a fabricated cantilever tip containing 5 vol.% magnetite 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 composite 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 concentration which ensure a full polymerization are listed in Table 5.3. Using backside 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).

Table 5.3: Fabrication parameters of composite with different particle concentrations. The viscosity is measured at 5 kHz. Due to the increase of viscosity with the particle concentration the spin speed has to be increased to keep layer thicknesses in a similar range.

Concentration Dynamic viscosity @ 22.5◦ C ± 20% [vol.%] [Pa s]

Spin speed

Exposure dose

[rpm]

[J/cm2 ]

Fabricated layer thickness [µm]

0 1 2 3 5 10

4000 4200 4400 4600 5000 5000

0.2 2 2 5 10 10 b

1.54 ± 0.01 1.64 ± 0.01 1.57 ± 0.02 1.61 ± 0.01 1.79 c± 0.02 2.20 c± 0.03

a b c

0.062 0.065 0.065 0.078 0.106 n.a. a

data not available not fully polymerized bigger thickness because of spin speed limitation

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 concentration 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 comc posite cantilever is partially exposed and results in a rough bottom surface ([93], (2011), with permission from Elsevier).

62

5.2 Limitation using UV exposure 5.2.3 UV transmittance depending on the layer thickness

The necessary exposure dose to achieve a fully polymerized layer is studied depending on the layer thickness for a composite with a 5 vol.% particle loading. Experimental

Firstly, the transmittance spectra depending an the layer thickness is measured for a composite with 5 vol.% particle loading. Composite layers with different thicknesses, 1.8 µm, 2.3 µm, 2.9 µm, and 3.5 µm have been fabricated and prepared as discussed before. Secondly, fabricated microcantilevers with the same thicknesses were fabricated. The top and bottom surfaces of these microcantilevers were investigated by SEM to evaluate the necessary UV dose for microstructures with smooth top and bottom surfaces. Results and discussion

Figure 5.10 shows the transmittance measurements for composites with different 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 Figure 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 surfaces 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 exposure must be applied. The UV transmittance measurements have shown that the

63

5 Evaluation of composite properties

i- lin e

h - lin e

T r a n s m itta n c e ( %

T )

1 0 0 8 0 6 0 4 0 h = 1 h = 1 h = 2 h = 3

2 0

. 9 µm . 9 µm . 2 µm . 5 µm

0 v o 5 v o 5 v o 5 v o

l.% l.% l.% 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 ) Figure 5.10: UV VIS transmittance measurements of the composite with 5 vol.% Fe3 O4 particle loading and different layer thicknesses h = 1.9, 2.2 and 3.5 µm. The UV absorption increases with increasing composite layer thicknesses. The sinusoidal distortion of the signal is based on a Fabry-Perot effect which occurs when the thickness of the film is a multiple of half a wavelength. For comparison, the transmittance of a pure SU-8 layer (h = 1.9 µm) is measured and show that the c IOP). Fe3 O4 particles in the composite are absorbed strongly in the UV range ([87],

Table 5.4: Exposure doses of the 5 vol.% composite with different layer thicknesses h.

64

Layer thickness h ± 0.1 [µm]

Exposure dose [J/cm2 ] Not sufficient Sufficient

1.8 2.3 2.9 3.5

0.95 10 50

5 20 50 75

10 40 75 -

20 60 -

5.2 Limitation using UV exposure

Figure 5.11: SEM images of photocurable MPC microcantilever tips with 5 vol.% Fe3 O4 particle contents and different layer thicknesses h and exposure doses D. The top surface of a 1.8 µm thick cantilever with 10 J/cm2 is shown in (a), and its smooth bottom surface in (b). (c) The bottom surface of a 1.8 µm thick cantilever with 5 J/cm2 , some holes are observable. (d) The bottom surface of a 2.9 µm thick cantilever with 75 J/cm2 containing no holes, whereas the bottom surface of a c IOP). 3.5 µm thick cantilever with 75 J/cm2 shows larger holes (e) ([87],

65

5 Evaluation of composite properties Fe3 O4 particles in the composite strongly absorb light in the UV range. Therefore, 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 increased 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 thickness have been optimized for the fabrication of microcantilevers. The layer thickness 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 maximum thickness of 2.9 µm can be fabricated.

5.3 Minimal pattern transfer in composite using UV exposure To determine the resolution of the composite, test structures with different patterns 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 pattern. The parameters of the composite are: 5 vol.% Fe3 O4 , thickness: 1.8 µm, spin speed: 5000 rpm, exposure dose: 10 mJ/cm2 . 5.3.2 Results and discussion

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 transferred into the composite with a width of 1.3 ± 0.2 µm. This shows that the resolution 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 c (2011), with permission from Elsevier). increased width of 1.3 µm. ([93],

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 composite was not reached. It should be possible to fabricate even smaller feature sizes with masks containing smaller patterns.

5.4 Magnetic characterization 5.4.1 Magnetic behavior of composite

Magnetic composites can have various magnetic properties. The magnetic properties of the developed composites mainly depend on the properties of the incorporated nanoparticles. However, the matrix material, the dispersion, interparticle distances and orientation of the filler can influence the magnetic properties. 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 because it determines the maximal applied force on a microstructure by an external magnetic field.

Experimental

The magnetic characteristics of the composite structures were obtained by measuring their M-H loops. The composite structures were measured using an alternating gradient magnetometer (AGM) (Micromag 2900, Princeton Measurement Corporation). The AGM has a higher sensitivity compared to a vibrating sample magnetometer (VSM) and is therefore more suitable for this measurements. Films of composite were prepared with known dimensions for the magnetic measurements. Furthermore, the magnetic behavior of the composite at low temperature was measured by a superconducting quantum interference device (SQUID).

Results

Composites with filler concentrations from 0 to 10 vol.% were investigated. In Figure 5.13 the M-H curves for the composite films with different nanoparticle concentrations measured at room temperature are shown. The saturation magnetization depends, as expected, linearly on the particle concentrations in the composite. The measured curves show negligible hysteresis, indicating superparamagnetic characteristics of the nanocomposite. The coercivity of all composites with different particle concentrations are smaller than 0.5 kA/m at room temperature. Additionally, the superparamagnetic behavior of the composite were verified by low temperature magnetic measurements. A typical phenomenon for superparamagnetic material is an increased hysteresis at low temperature. Figure 5.14 shows M-H curve of a 1 vol.% composite at 10 K and 15 K. As expected at low temperature (10 K) the coercivity is increased. The magnetic momentum of the particles is blocked and the magnetic characteristics shows a clear hysteresis. Above 150 K the thermal energy is high enough to overcome the magnetic anisotropy energy, the magnetization of the particle is easily flipped and the composite show a negligible coercivity. The blocking temperature, which separates these two regimes must be, therefore, lower than 150 K.

68

M a g n e tiz a tio n , M

[A /m ]

5.4 Magnetic characterization

3 0

0 .1

2 0

0 .0

2 7 .9

1 0

-0 .1 -0 .1

0 .0

1 3 .8 9 .1 6 .5 2 .7

0 .1

0 -1 0

1 v o 2 v o 3 v o 5 v o 1 0 v

-2 0

×1 0

3

-3 0 -8

-6

-4 5

×1 0

-2

0

2

4

l.% l.% l.% l.% o l.%

6

8

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 c (2011), with permission from superparamagnetic behavior of the composite ([93], Elsevier).

×1 0

3

M a g n e tiz a tio n , M

[A /m ]

3

0 .2

2

0 .0

1

-0 .2 -0 .3

0 .0

0 .3

0 -1

1 5 0 K 1 0 K

-2 -3 -3

-2

×1 0

-1 5

0

1

2

3

A p p lie d F ie ld , H [A /m ]

Figure 5.14: M-H measurement of a 1 vol.% composite at 10 K and 150 K measured by a SQUID.

69

5 Evaluation of composite properties 5.4.2 Magnetic behavior of particles

The magnetic nanpoparticles were characterized to compare their magnetic properties with the mixed composites. This facilitates a predication about the influence of the matrix material on the magnetic behavior of the nanoparticles. Experimental

The magnetic characterization of the nanoparticles was done using a vibrating sample magnetometer (VSM) (Micromag 3900, Princeton Measurement Corporation). The nanoparticles were weighed using a microbalance prior to the measurement. Results and discussion

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 behavior at room temperature. This small opening of the hysteresis can be an artifact 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 magnetization 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 magnetization of bulk magnetite, 92 Am2 /kg [50], (maghemite, ∼ 80 Am2 /kg [109]). Nanoparticles have a reduced saturation magnetization compared to bulk values due to surface effects (spin surface disorders) and impurities [55, 58] (Section 2.1.4). 5.4.3 Material characterization of particles

To obtain information about the structure of the nanoparticles and possible impurities 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 structure of the particles inert gas, nitrogen, was used during solvent evaporation.

70

5.4 Magnetic characterization

3 2

5 0

0 .0 -0 .1 -0 .1

0 .0

×1 0

[e m u /g ]

2 5 1

0 .1

0

0

M

-1

F e 3O

-2 5 4

p a r tic le s

-2

-5 0

5

M a g n e tiz a tio n , M

[A /m ]

0 .1

-3 -8

-6

-4

×1 0

5

-2

0

2

4

6

8

A p p lie d F ie ld , H [A /m ]

Figure 5.15: Magnetization of Fe3 O4 nanoparticles measured by vibrating sample magnetometer at room temperature. The negligible remanent magnetization at zero applied field indicates the superparamagnetic characteristics of the nanoparticles ([93], c (2011), with permission from Elsevier).

XRD patterns were carried out from three probes with a Bruker D8 Advance diffractometer (40 kV, 40 mA, CuKα) for a range of 2θ XRD = 10 – 80◦ .

Results and discussion

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 XRDspectra other iron oxide phases such as wustite FeO (ICSD 82233) and hematite α-Fe2 O3 (ICSD 066756) can be excluded.

71

In te n s ity [a .u .]

5 Evaluation of composite properties

F

e

O 3

1 0

2 0

3 0

4 0 2

n

a

5 0

Θ

n

o

p

a

r t i c l e

s

4

6 0

7 0

8 0

[ ° ]

Figure 5.16: XRD pattern of the magnetite (Fe3 O4 ) nanoparticles. The magnetite peak poc (2011), with permission from Elsevier). sitions (ICSD 028664) are indicated ([93],

5.4.4 Magnetic force on composite 1 The forces of the composite induced by an external magnetic field are calculated

using the measured magnetization and compared to force values obtained by a micro-force sensor. These magnetic force values are important to evaluate possible applications of microdevices made from these composites. The force, F [N] acting on the magnetic composite is [112] 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 composite 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 (Institute 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 watercooled electromagnet and the current is varied to change the force on the film. A c (2009) micro-force sensor (Femtotools GmbH) is used to measure the force ([111], IEEE).

 Fz = µ0 V Mz ·

∂ H ∂z

 (5.3)

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

The magnetic force on a MPC film with dimensions of 5.25 mm x 2 mm x 2.5 µm and 3 vol.% Fe3 O4 nanoparticles is measured. The magnetic field is generated by an electrical coil. The setup is shown in Figure 5.17. The magnetization of the MPC changes as a function of the applied magnetic field H. The field of the coil at the center of the film was measured as 80 mT (64 kA/m) at I = 1 A. The details of the experiments and calculations are described in [111]. Results and discussion

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 2

Force (µN)

1.5 1 0.5 0 −1.5

−1

−0.5 0 0.5 Applied Current I (A)

1

1.5

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 c (2009) IEEE). dashed line ([111],

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 calculated. The static deflection of a beam for a distributed load is [59] q L L4 8YIz

(5.4)

F = Fv hw L

(5.5)

uL = with

qL =

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, thickness, 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 fabricated 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 behavior at room temperature. The magnetic behavior of the composite can be controlled by the selection of the incorporated particles. The magnetic characteristics 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 possible diamagnetic contribution of the sample holder during the magnetic measurement 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 current = 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 microstructures 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

5.5 Young’s modulus 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 nanoparticle 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. 5.5.1 Measurement method

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 deforms 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 measured 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 properties of a microstructure from these films. Microcantilevers are well suited to measure the Young’s modulus because their mechanical characteristics are described 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 profilometer 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 measuring the resonant frequency of the microstructures. Using Euler-Bernoulli beam approximation the Young’s modulus of a cantilever is determined by its mechanical resonance frequency, the geometrical parameters, and the density (2.18). The resonance frequency of an actuated cantilever can be measured with a laser Doppler 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 The main advantage of dynamic measurements is the simple non-contact measurement method. Using a laser Doppler vibrometer several cantilevers can be measured within minutes. To minimize the damping the measurements must be conducted in a vacuum chamber. There are several actuation methods for cantilevers. Mechanical (piezo) and magnetic actuation are the most common ones. Tests with piezo actuation showed disturbant resonance peaks in the frequency spectra of the cantilever measurements and is therefore, not suitable to determine the Young’s modulus. The MPC cantilevers can be actuated by an external magnetic field. However, it would be difficult to compare different particle concentrations due to the different magnetic forces acting on the MPC cantilevers, and the non filled polymer cantilever cannot be actuated with a magnetic actuation. Furthermore, to avoid magnetic shielding from the chamber the magnetic setup has to be placed into a vacuum chamber. This would lead to strong heating of the coils, because there is no cooling effect by air convection in the vacuum chamber. Kelvin polarization force (electrodynamic actuation) for the cantilever actuation have been reported [60]. However, Kelvin polarization forces can be influenced by the incorporated Fe3 O4 nanoparticle and complicate comparison between different particle concentrations. It is reported that the mechanical resonance behavior of a cantilever can be investigated by monitoring its vibrations due to thermal fluctuations [118, 119]. Using a laser-Doppler vibrometer the resonance frequency of the cantilever can be measured and the Young’s modulus can be determined independent on the particle concentrations of the MPC cantilever. 5.5.2 Error analysis

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 u   u ∂g 2 2 ∆y = t∑ ∆xi ∂xi i

(5.6)

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 determination using Euler-Bernoulli beam theory.

∆xi

∂g ∂xi

xi

Description

ρ

The density is calculated using (5.11). The error due to inhomogeneities in the particle dispersion, holes and pores in the MPC cantilevers are assumed to be less than 5% of the calculated value.

L

The resolution of the used microscope Leica DM4000 to measure the length and the lateral dimensions of the cantilever has a resolution of about 0.3 µm. Together with the uncertainties of the anchor position the error is assumed to be less than 1 µm.

48ρL3



2π f n λ2n h

2

fn

The uncertainty of the measured frequency due to the resolution of the laser Doppler vibrometer and the imperfect vacuum is assumed to be 15 Hz.

24ρ f n



2πL2 λ2n h

2

h

The thickness was measured with a profilo- −24 h3 meter. The difference between the thicknesses of the single cantilevers to the measured mean thickness of each chip is assumed to be less than 0.05 µm.

12



2πL2 f n λ2n h

ρ



2

2πL2 λ2n

2

0.05ρ

1 µm

15 Hz

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. For a reliable measurement further points (including systematic errors) such as damping, stress gradient in cantilevers, residuals on cantilevers, and imperfect clamping, as discussed in the next subsections, have to be taken into account.

78

5.5 Young’s modulus 5.5.3 Comparison of FEM-simulation with Euler-Bernoulli approximation and evaluation of the influence of imperfect clamping

The Euler-Bernoulli beam equation assumes a perfectly clamped beam as depicted in Figure 2.6. This cannot be achieved using micromachining layer technology due to not preventable alignment errors between the different layers. In this section the accuracy of using Euler-Bernoulli approximation compared to FEM simulations with more realistic clamping boundary conditions are studied. Further the influence of fabrication errors on the clamping are discussed. The reported fabrication process of cantilevers [60] leads to a plate-like overlap of the cantilever over the anchor structure, which has to be measured and taken into account for the calculation of the Young’s modulus. The determination of this overlap gives additional error sources. Therefore, a process was selected, where the anchor of the cantilever can be designed to overlap with the cantilever structure by 5 µm and a plate-like overlap can be avoided (see Figure 4.2). An alignment accuracy of about 2–3 µm can be achieved with the photolithography alignment machine and masks used. 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 attachment 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 section 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 (using 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 boundaries 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 microcantilevers. (a) Without a gap, (b) with a gap between cantilever and anchor.

Table 5.6: Parameters used for the FEM-simulation

Young’s modulus Poisson’s ratio Density Thickness cantilever layer h Width cantilever w Length cantilever Lm

5 0.22 1190 1.8 14 50–200

GPa [121] [121] kg/m3 [121] µm µm µm

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 cantilever 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 Figure 5.19 (c) with a cantilever with Lm + 1 µm assuming a gap of 1 µm. Results and discussion

The FEM models (a, b, c, and d) are compared to Euler-Bernoulli beam approximation (2.18). The different simulated eigenfrequencies, dependent on the cantilever lengths for the different FEM models and the Euler-Bernoulli beam approximation, are shown in Figure 5.21. Model (c) and (d) show the highest discrepancies to the Euler-Bernoulli beam approximation. All models approach the

81

5 Evaluation of composite properties

F re q u e n c y f [k H z ]

2 5 0 E u a ) b ) c ) d )

2 0 0

le r - B e r n o u S im p le c a n C a n tile v e r C a n tile v e r P la te - lik e o

lli tile v e r s tru c tu re s t r u c t u r e w i t h 1 µm v e r la p

g a p

1 5 0 1 0 0 5 0 0 6 0

8 0

1 0 0

1 2 0

1 4 0

C a n tile v e r le n g th L m

1 6 0

1 8 0

2 0 0

[ µm ]

Figure 5.21: FEM simulation with different models (Figure 5.19) and Euler-Bernoulli approximation of the resonance frequency of MPC cantilevers with different lengths.

Euler-Bernoulli approximation for long cantilevers. In Figure 5.22 the discrepancy between the obtained Young’s modulus values using different FEM models and Euler-Bernoulli beam approximation are shown. For all models the discrepancy is lower for longer cantilevers. Model (d) with a plate-like overlap gives a discrepancy of 18% for 100 µm cantilevers. In such a case the Euler-Bernoulli beam approximation cannot be used, and FEM simulations with an iterative approach as discussed by [60] has to be used. Model (b) shows that the anchor structure has to taken into account for shorter cantilevers. For cantilevers longer than 100 µm, the error between the FEM simulation of model (b) and the Euler-Bernoulli approximation is < 2%. Therefore, the Euler-Bernoulli beam approximation, with the benefit of its simplicity for the calculations of Young’s modulus from the measured frequencies of cantilevers, can be used with an accuracy of 2% for cantilevers with length longer than 100 µm. The influence of a gap at the anchor connections, which is present for few cantilevers, on the resonance frequency of the cantilever is simulated with model (c). The error due to such a gap is significant for short cantilevers. In case of an additional gap of 1 µm the error is up to 14% for a 50 µm long cantilever. However,

82

Y o u n g ’s m o d u lu s b y F E M c o m p a r e d to E u le r - B e r n o u lli [% ]

5.5 Young’s modulus

0 -1 0 -2 0 E u a ) b ) c ) d )

-3 0 -4 0

le r - B e r n o u lli S im p le c a n tile v e r C a n tile v e r s tr u c tu r e 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 P la te - lik e o v e r la p

g a p

-5 0 6 0

8 0

1 0 0

1 2 0

1 4 0

C a n tile v e r le n g th L m

1 6 0

[ µm ]

1 8 0

2 0 0

Figure 5.22: Young’s modulus simulated by different FEM models (Figure 5.19) for cantilever lengths of 50–200 µm compared to Young’s modulus calculated with EulerBernoulli 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 damping of the structure and the external damping due to the surrounding medium 1 1 1 = + Q Qmedium Qintrinsic

(5.7)

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 = + + + Qintrinsic Qclamp Q TED Qsur f ace Qmat

(5.8)

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 determined 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 minimized because the damping can lead to significant differences between the natural and the measured resonant frequency. All measurements to determine the Young’s modulus were performed at pressures below 2.2·10−2 Pa. Using equation 2.28 and 2.12 the resonant frequency of a damped resonator can be written as s ωr = ω0

1−

1 +1

2Q2

(5.9)

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 discrepancy of ωr = 0.9996 · ω0

(5.10)

This shows that the resonant frequency differs not substantially from the undamped natural frequency, therefore, it can be assumed to be equal to the eigenfrequency 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]. 5.5.5 Stress gradient in cantilevers

Polymer cantilevers are susceptible to stress gradients. Some fabricated cantilever arrays contained bended cantilevers. A reason for the bending could be based on the UV absorption of the incorporated nanoparticles in the composites. The absorption of the UV light by the magnetite nanoparticles can lead to an inhomogeneous exposure in the layer, and could result in a stress gradient. The influence of the UV dose on the mechanical performance of pure SU-8 by tensile tests is reported [126]. It is shown that the Young’s modulus and the strength for a film with an UV dose smaller than 1000 mJ/cm2 changes. When the SU-8 resist system is exposed to UV light, the photo-acid generator absorbs photons and produces a strong acid. This acid acts as the catalyst for the cross-linking reaction during the later post exposure baking step. The reaction rate of the cross-

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 exposure 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 bending 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. 5.5.6 Residuals on cantilevers

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 ultrasonic 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 measured 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 measurements and only clean ones have been .taken into account 5.5.7 Experimental

The dynamic Young’s modulus of the composites with different particle concentrations 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 domain by means of a fast Fourier transformation. The frequency spectra were averaged 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 calculated as ρc = ρ Fe3 O4 ξ + ρSU −8 (1 − ξ )

(5.11)

where ρ Fe3 O4 is the density of Fe3 O4 nanoparticles (5180 kg/m3 [127]), ρSU −8 the density of pure SU-8 (1190 kg/m3 ) [121]), and ξ is the volume fraction of the nanoparticles in the composite. 5.5.8 Results

The measured Young’s moduli of cantilevers with different particle concentrations are plotted in Figure 5.23. The cantilevers with nanoparticle concentrations varying from 0 to 3 vol.% show the same characteristics and have a mean Young’s modulus of 4.4 GPa. The cantilevers with a particle concentration of 5 vol.% show a slightly higher mean Young’s modulus of 5.1 GPa. The standard deviations of cantilevers with the same length and same nanoparticle filler concentration are between 0.1 and 0.4 GPa, and the corresponding calculated measurement errors are smaller than 0.4 GPa. Young’s modulus measurements of cantilevers with 5 vol.% particle concentrations and lengths of 160 µm from other chips (arrays) measured at different days are in the same range, shown in Table 5.7. On array 2 some residuals were observed on the cantilevers by optical microscope, therefore, these results were not taken into account for the final measurement. 5.5.9 Discussion and conclusion

The measured dynamic Young’s modulus of bare SU-8 is in good agreement with the measurements of the dynamic Young’s modulus (4.5 GPa at 25 kHz) of cantilevers actuated by the Kelvin polarization force [60]. Comparison of the measured Young’s modulus of filled polymer with the literature is difficult as it is known that particle morphology, surfactant and different baking time and

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

Array 1 Array 2 Array 3

Standard deviation GPa

Measured cantilevers

µm

Young’s Modulus GPa

160 160 160

5.08 4.88 5.06

0.10 0.13 0.37

22 19 9

Y o u n g ’s m o d u lu s ( G P a )

5 .5

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 different 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 measurement 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 indicated 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 c (2011), with permission from modulus using the model of Hashin-Shtrikman ([93], Elsevier).

88

5.5 Young’s modulus temperature for photocurable polymer affect mechanical reinforcing. It is reported that adding silica nanoparticles in photocurable epoxy slightly increases the static Young’s modulus (measured by nanoindentation) [6], while there is a significant increase in Young’s modulus with silica in PDMS [97] or silica in PMMA matrix [128]. The measurements show a slight increase in the Young’s modulus with particle concentration (5 vol.%) as expected. For comparison the theoretical data from the Hashin-Shtrikman model are shown in Figure 5.23. The values are calculated with (2.34) using a Poisson’s ratio of 0.26 [129], and 0.37 [130] and a Young’s modulus of 4.2 GPa, and 174 GPa [130], for SU-8 and Fe3 O4 , respectively. The measured Young’s moduli are close to the lower boundary. A reason for a only slight increase of the Young’s modulus with increased nanoparticle concentration can be based on a not strongly pronounced particle-polymer interaction. A slight decrease of Young’s modulus for low silica particle concentration in SU-8 has been reported [131].

89

5 Evaluation of composite properties

5.6 MPC heating with alternating magnetic field Magnetic nanoparticles can be heated by an external alternating magnetic field. In particular, magnetic fluid hyperthermia, where magnetic nanoparticles are injected in a human tumor and heated externally by an alternating magnetic field, is today an important application. The magnetic nanoparticles can increase the temperature in tumors to 41 – 46◦ C and therefore kill tumor cells locally [132]. The frequency for heating of superparamagnetic nanoparticles by alternating magnetic fields are in the range of ∼200 kHz to ∼10 MHz. At low frequencies the heating is less efficient for particles with sizes of ∼14 nm [133], while at higher frequencies organic molecules such as proteins are heated and start to denature. The heating in an alternating magnetic field could be used to heat wirelessly MPC microstructures (cantilevers, microrobots) in different environments. In this experiment the possibility and efficiency of alternating magnetic field heating with the developed composite is tested. Furthermore, this results are important to estimate in Section 6.2.4 if the MPC cantilevers already generate heat during magnetic actuation. There are three different mechanisms for heat generation of magnetic material in an alternating magnetic field [134]: • Generation of eddy current’s in bulk magnetic materials, • Hysteresis losses in bulk and multi-domain magnetic materials, • Relaxation losses in superparamagnetic particles. The developed MPC in this work contains superparamagnetic nanoparticles (singledomain). Hence, the significant mechanism that contributes to the heating is the relaxation loss mechanism. There exist two relaxation losses: Brownian relaxation loss, based on rotation of particle with fixed magnetization in direction of the external field and Néel relaxation, where the magnetic moment originally locked along the crystal easy axis, follows the external magnetic field and changes its orientation at the frequency of the applied field [134]. 5.6.1 Experimental

MPC with different Fe3 O4 nanoparticle concentration (0, 2, 4 vol.%) was spincoated 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 generator (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 alternating current of 400 A peakvalue (283 ARMS ) at a frequency of 245 kHz was

90

5.6 MPC heating with alternating magnetic field used. The temperature was measured by a fluorescent glass fiber temperature sensor „Fluotemp“ (Photon Control Inc. Burnaby, Canada) which is immune to electro-magnetic interference. The glass fiber was placed on the sample surface. The accuracy of the temperature sensor is smaller ± 1◦ C. Because the glass fiber has only a limited contact to the sample surface the average temperature of the sample is higher than the measured value. The schematic of the experimental setup is shown in Figure 5.24. The magnetic field at the composite film position was calculated by FEM simulation1 . Figure 5.25 shows the magnetic field HRMS for IRMS = 283 A at the cross-section A (Figure 5.24) of the sample in x-direction. The calculated mean magnetic field HRMS was 2947 A/m (mean of cross-section A in sample).

z y

IRMS = 283 A, 245 kHz

x

Fluorescent glassfiber temp. sensor Sample

Sample holder

Cross-section A

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

M a g n e tic fie ld n o r m , H

R M S

[A /m ]

5 Evaluation of composite properties

3 1 0 0 3 0 0 0 2 9 0 0 2 8 0 0 2 7 0 0 2 6 0 0 0

5

1 0

1 5

2 0

C r o s s - s e c tio n o f M P C film

2 5

3 0

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 (xdirection) 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. 5.6.3 Discussion and conclusion

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]: PL = πµ0 χ0 H 2 f

2π f τe f f 1 + (2π f τe f f )2

(5.12)

where µ0 is the permeability of free space (µ0 = 4π·10−7 T·m/A), χ0 is the susceptibility (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 f f

7 0

4 v o l.% 2 v o l.% 0 v o l.%

6 0 5 0 4 0

T

e

m

p

e

r a

t u

r e

a

t

s u

r f a

c e

[ ° C

]

M a g n e t ic f ie ld o n

3 0 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 temperature is measured at the surface of spin-coated MPC samples with different nanoparticle concentration: 0, 2, 4 vol.%.

1 1 1 = + τe f f τN τB

(5.13)

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 τB =

4πηr3H kB T

(5.14)

with η is the sample viscosity and r H the particle hydrodynamic radius [53]. Because the nanoparticles are assumed to be fixed in the polymer matrix the viscosity 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 nanoparticles in a stable suspension (ferrofluid) were affected by both relaxation processes. Even for an uncured polymer resist, where the nanoparticles are not fixed,

93

5 Evaluation of composite properties with a viscosity η of 0.106 Pa s (Table 5.3, 5 vol.%) and a hydrodynamic radius of the nanoparticles of 19 nm (XDC measurement in water Table 5.1) τB is a factor ∼104 higher than τN and can be neglected for the calculation of τe f f . The magnetic field heating as shown by the power loss equation (5.12) is a function of the material (χ0 ), the magnetic field H and the applied frequency f and the relaxation time τe f f . The anisotropy constant Ke f f is not known for the composite. Anisotropy constant 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 frequency 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 experiments are in the first part of the curve where the power loss has a square dependence 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.

By decreasing the sizes of structures the surface/volume ratio increases and the heat dissipation by the surrounding media increases. The next step is to run simulations to show if the surface temperature of a microstructure (cantilevers or microrobots) in water can be raised to trigger a biological reaction at the surface. This could act as a trigger to release drug molecules loaded on the surface of the microrobot as close to the target as possible. Simulations showing the possibility of heating of a hydrogel discs (diameter 4 mm and thickness 0.5 mm), with 2.5 wt.% of Fe3 O4 particle concentration in tissue have been recently published [135].

94

5.7 Surface properties

1 .2 x 1 0

8

1 .0 x 1 0

8

P o w e r l o s s P / χ0 [ W / m

3

]

K

8 .0 x 1 0

7

6 .0 x 1 0

7

4 .0 x 1 0

7

2 .0 x 1 0

7

e ff

= 1 .3 5 1 0

-3

3

J /m

0

1 0 0

2 0 0

4 .0 x 1 0

5

2 .0 x 1 0

5

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 Surface properties The surface property investigation of the composite is important for biofunctionalization of the composite surface. It gives information about the hydrophobicity of the surface, which is also important for the use of microstructures of the composite (cantilevers, micro robots) when immersed in water.

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 contact 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 5.7.2 Results and discussion

With dynamic water contact angle measurements on samples with 0, 1, 2, 3, 5, 10 vol.% no influence of the particle concentration on surface chemistry and surface roughness on the surface of the composite could be detected [136]. The advancing contact angle, which indicates the hydrophobicity of the sample is 81.1◦ ± 1.6◦ for all samples. This does not differ significantly from the unfilled samples 81.8◦ ± 1.3◦ . If the particle concentration would increase surface roughness, then the contact angle hysteresis (advancing minus receding contact angle) would also increase. However, the hysteresis was between 29◦ to 40◦ for all samples, and no distinct dependence on the particle concentration could be detected. 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] because of their variety of properties and biocompatibility. Pure SU-8 is not considered 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 proliferation 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 nanoparticle 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

Cell viability was determined by the water-soluble tetrazolium (WST-1) proliferation and cytotoxicity test. Upon reaction with various mitochondrial dehydrogenase enzymes, the light red colored tetrazolium salt is cleaved to yield a dark red water-soluble dye called formazan which can be detected using a microplate reader. Therefore, the greater the amount of formazan detected, the greater the number of viable and metabolically active cells.

96

5.8 Biocompatibility Proliferating normal dermal human fibroblasts (NDHFs) (PromoCell, Germany) were harvested from exponentially growing subconfluent monolayers. The culture was maintained in 25 cm2 culture flasks in Dulbecco’s modified eagle medium (DMEM, GibcoBRL, Canada) supplemented with 10 % heat inactivated fetal bovine serum (FBS, Sigma) and 1 % antibiotic antimycotic (ABAM, GibcoBRL, Canada) at 37 ◦ C and 5 % CO2 . Cells were used until passage 20. Prior to the transfer into the well plate, bulk MPC films were sterilized for 20 minutes under UV light and stored in 0.1 M phosphate buffered saline (PBS) solution at pH 7.4. As controls, 3000 NDHFs were incubated with tissue culture polystyrene (TCPS) under the same conditions. Mitochondrial activity was measured 24 hours after seeding by the addition of 20 µl WST-1 solution (Roche) per sample and analyzing the absorption spectrum using a microplate reader (Infite M200, Tecan GmbH, Germany). In order to visualize cell adhesion and cell viability of NDHFs cultured on surfaces of MPC films, a live and dead staining was performed using the fluorescent stains fluorescein diacetate (FDA, Fluka) and Hoechst 33342 (Invitrogen). Living cells metabolize FDA to fluorescein and Hoechst 33342 intercalates itself with the DNA present in the nuclei of cells. On each film, 3000 NDHFs were cultured for 48 hours and stained with FDA (1 mg/ml, 1:1000) and Hoechst 33342 (1:1000) in a PBS solution for 10 minutes at room temperature. A FITC filter and a DAPI filter were used to visualize fluorescein and Hoechst 33342 respectively under a fluorescence microscope (Zeiss Axiovert 200M, Carl Zeiss AG, Germany).

5.8.2 Results and discussion

Figure 5.28 shows the result of the cell viability test of NDHFs cultured on surfaces 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. Epifluorescence microscope images display living NDHFs by means of green fluorescence cultured on surfaces of MPC with different Fe3 O4 nanoparticle concentrations. 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

C e ll v ia b ility [% ]

8 0

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.% ]

Figure 5.28: Viability of NDHFs is investigated by measuring the mitochondrial activity of the living cells compared to cells cultured on control tissue culture polystyrene.

0 vol.%

1 vol.%

2 vol.%

3 vol.%

5 vol.%

10 vol.%

100 µm

Figure 5.29: Epi-fluorescence microscope image of living normal dermal human fibroblasts (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 foreskin 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 allergy tests for animals and humans are necessary [139]. Furthermore, the nanoparticle absorption by cells [45] has to be studied.

5.9 Summary and conclusion The developed photopatternable nanocomposite for the fabrication of microstructures has been characterized in detail. The investigated properties of the composite with 5 vol.% are summarized in Table 5.8 and compared to pure SU-8. The MPC has outstanding properties such as homogeneous particle dispersion with agglomerates around 50 nm, show superparamagnetic behavior, has a high Young’s modulus (important for the fabrication of suspended structures). It can be heated by an alternating magnetic field and can be used for interaction with biomaterial. In the next chapters different applications for MPC microstructures are evaluated and presented.

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.

Properties

SU-8 50 (Microchem)

MPC 5 vol.% Fe3 O4

Magnetic behavior

-

superparamagnetic

Saturation magnetization

-

13.8 kA/m @ H=800 kA/m

Coercivity

-

< 0.5 kA/m

Blocking temperature

-

< 150 K

Nanoparticle size

-

∼13 nm

Nanoparticle agglomerates

-

40 – 50 nm

Nanoparticle material

-

Fe3 O4 (small content of γ-Fe2 O3 fraction possible)

Applied magnetic force

-

45.7 kN/m3 for 3 vol.%

Density

1218 kg/m3 [140]

1416 kg/m3 a

Max. exposed thickness  100 µm with standard photolithography (75 J/cm2 )

2.9 µm

Exposure dose for full poly- ∼ 0.2 J/cm2 merization of 1.8 µm thick layer

10 J/cm2

Minimal feature size with standard Photolithography

( 1.3 µm, resol- 1.3 µm ± 0.2 µm (resoluution limit not tion limit not reached) reached)

Dynamic Young’s modulus

4.5 GPa @ ∼30 kHz

5.1 GPa @ ∼30 kHz

Hydrophobicity (dynamic contact angle)

81.8◦

81.1◦ ± 1.6◦

±

1.3◦

Heating by alternating mag- netic field

Heating rate: ∼0.5 K/s b

Biocompatibility: Cell viab- 99% ility of NDHFs c

86%

a b

c

100

calculated with (5.11) using a density of Fe3 O4 of 5180 kg/m3 [127] and pure SU-8 1218 kg/m3 [140] at the surface of a 4 vol.% MPC film with a diameter of 3 cm and a thickness of ∼250 µm at 2947 A/m and 245 kHz (Section 5.6). normal dermal human fibroblasts

6 Applications I: MPC cantilever resonator Microcantilevers are used as sensors in the field of micro electromechanical systems (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 molecules such as DNA, proteins, and antibodies by detecting a change of surface stress or change of mass [141]. Such microresonators are typically made of silicon 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 embossing, 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 chemically 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 analyte 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 magnetic actuation performed with and without a feedback-loop circuit are presented and the possibility of using such MPC cantilevers for sensing devices is dis-

101

6 Applications I: MPC cantilever resonator cussed. A magnetic actuation of MPC cantilevers with remote optical or magnetic readout setup leads to a totally passive resonator device, which can be integrated into different fluidic channels, for example low-cost polymer channels. Because no electrical contact is necessary to the device, the fabrication process can be simplified. Magnetic actuation allows measurements in electrolyte media and does not affect biological samples by electric fields.

Figure 6.1: SEM image of a MPC cantilever structure with Fe3 O4 nanoparticle concentration of 5 vol.%.

The geometrical dimensions (designed values) of the fabricated cantilevers for actuation tests are listed in Table 6.1. In this chapter only cantilevers with a particle concentration of 5 vol.% are investigated.

Table 6.1: Dimensions of the fabricated MPC cantilevers

102

Length L

Width w

Thickness h

Concentration

30 – 400 µm

14 µm

1.8 – 2.9 µm

5 vol.%

6.1 MPC cantilever resonance characterization by thermally induced vibrations

6.1 MPC cantilever resonance characterization by thermally induced vibrations As already described in Section 5.5 the mechanical resonance behavior of a cantilever can be investigated by monitoring its vibrations due to thermal fluctuations [118] by a spectrum analizer. A laser-Doppler vibrometer (LDV) (Polytec, MSA400) was used to characterize the resonance behavior of the cantilevers. This characterization method has the advantage compared to magnetic actuation that it excludes influences such as heating of the microstructure by the alternating magnetic field or sample heating due to heat transfer from the actuation coil. 6.1.1 Q-factor in different media

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 concentration 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 cantilevers Q-factors with similar geometries in water around 1 at 75 kHz were determined [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 in w a te r Q = 1 .3

D e fle c tio n [p m ]

1 0 8 6 4 2 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 pressure) 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 decreases 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 increased the resonance frequency and the Q-factor decreases. The deflection amplitude 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 cantilever with 5 vol.% nanoparticle concentration were calculated by the inviscid approximation by Chu described in [150], where the fluid is assumed to be incompressible. f f luid−Chu = q

f vac 1+

πρ f luid w 4ρc h

(6.1)

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

1 0 0

1 5 0

D e fle c tio n [p m ]

4

2

0 0

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 loading) 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 using the model of Sader [150], where viscous effects of the fluid are taken into account. 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

QSader =

4m L πρ f luid w2

+ Γr ( ω )

Γi ( ω )

(6.2)

where Γr and Γi represents the real and imaginary part of the hydrodynamic interaction 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 1+

πρ f luid w2 Γr ( ω ) 4m L

(6.3)

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 µm

f r measured kHz

f f luid−Sader kHz

f f luid−Chu khz

Qmeasured -

QSader -

60.8 ± 0.4 112.1 ± 0.4 189.0 ± 0.8

90.76 ± 0.17 22.73 ± 0.18 6.00 ± 0.09

101.2 27.4 8.6

116.8 34.4 12.1

2.7 ± 0.1 1.7 ± 0.1 1.1 ± 0.1

3.8 2.5 2.0

The hydrodynamic interaction function is given by   √ 4iK (−i iRe)  Γ ( ω ) = 1 + q 1 Ω(ω ) √ iReK0 (−i iRe)

(6.4)

where Re = ρ f luid ωw2 /4η. Ω(ω ) is a correction factor for rectangular cross sections. 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 viscosity 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. 6.1.2 Frequency shift of MPC cantilevers with additional mass

For a proof of concept to demonstrate the response of the cantilever to a mass change in air a gold layer was sputtered on the tip of a cantilever with 5 vol.% nanoparticle concentration and dimension of 116 µm x 18 µm x 1.89 µm. The cantilever was actuated by thermal noise. Figure 6.4 shows the resonance frequency with and without load. A shift to a lower frequency is observed as expected. The resonance frequency without load is 43.42 ± 0.14 kHz and decreases to 37.18 ± 0.10 kHz with the additional mass (the frequencies were measured three times and averaged). The exact amount of the additional mass could not be evaluated. With a white light interferometer (WLI) topography measurements the mass is estimated to be in the order of 1 ng distributed over the last third of the cantilever

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 agreement 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.

6 w ith o u t lo a d w ith lo a d

D e fle c tio n [p m ]

5 4 3 2 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 additional mass. A gold layer was sputtered at the cantilever tip. The cantilever was excited by thermal noise.

6.2 Magnetic actuation of superparamagnetic MPC cantilevers 1 In

this section the resonance performance of the superparamagnetic MPC cantilevers actuated by magnetic fields is evaluated. Different magnetic field actuation 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 characterization of the actuation coils are discussed in [152].

107

6 Applications I: MPC cantilever resonator methods and setups are discussed. 6.2.1 MPC cantilevers actuated by alternating magnetic field

The fabricated superparamagnetic cantilevers was actuated with an alternating inhomogeneous magnetic field, generated by an external coil (AC-setup). Experimental

To actuate the MPC cantilevers, they were placed above an electrical coil as illustrated in Figure 6.5. The alternating magnetic field H generated by the electromagnet (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 electromagnets 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 rapidly 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 approximately 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 magnetic fields and gradients are in the same range. Results

The magnetization of the MPC in this configuration is caused by the AC actuation coil. Due to superparamagnetic behavior the magnetization of the composite changes sign simultaneously with the field gradient of the coil. Hence, the force acts at double the actuation frequency. Firstly, a cantilever (5 vol.%, a thickness of 1.8 µm, a length of 200 µm, and a width of 15 µm) is excited in air at room temperature by a magnetic excitation frequency sweep from 5 to 20 kHz. A mechanical resonance of the cantilever is detected at 14 kHz with an amplitude of 3.5 nm as shown in Figure 6.7. Secondly, the cantilever is actuated from 5 to 10 kHz and the vibration is recorded at the second harmonic as shown in Figure 6.8. At the mechanical resonance of the cantilever an amplitude of 33.5 nm was

108

6.2 Magnetic actuation of superparamagnetic MPC cantilevers

laser-Doppler vibrometer

y

z

x

cantilever-chip

hCoil

A HCoil ~ ICoil

rin rout

Figure 6.5: Schematic illustration of the measurement setup. The magnetic composite cantilever 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 generation 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 superparamagnetic behavior of the MPC. The force, F, acting on the magnetic composite can be calculated using (5.2). The force on the superparamagnetic 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. The magnetization, M, of the composite depends on the applied field, magnetic characteristics of the incorporated nanoparticles, particle concentration, and the shape of the cantilever. 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 magnetization, 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]

plication under excitation with a magnetic field in alternating mode (AC-setup). The negative cycle of the excitation signal will be rectified and the actuation of the cantilever will be realized at the double frequency of the excitation signal as shown in Figure 6.8. Superparamagnetism is an effect that depends on the time scale of an experiment. If the cantilevers are excited at very high frequencies (i.e.  10 kHz) the nanoparticles may become blocked. This happens when the magnetic field changes direction faster than the relaxation time of the nanoparticles. The relaxation time (Néel relaxation) of the superparamagnetic particles of 13 nm diameter is 4.4 · 10−8 s at room temperature (calculation shown in Section 5.6). A possibility to improve the magnetic actuation of the superparamagnetic microcantilevers is explained in the next section. 6.2.2 MPC cantilevers actuated by alternating inhomogeneous and additional uniform magnetic field 2 The

actuation of superparamagnetic MPC cantilevers can be enhanced by increasing their magnetization (5.2). The magnetization can be increased by increasing the alternating magnetic field, H. However, generating strong magnetic fields at relatively high frequencies is challenging due to increased losses in the 2

Parts of this section have been published in [111].

110

Phase [degrees]

Amplitude [nm]

6.2 Magnetic actuation of superparamagnetic MPC cantilevers 4 2 0 5

10 15 Frequency [kHz]

20

10 15 Frequency [kHz]

20

180 0 -180 5

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 cantilevers. 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

Phase [degrees]

Amplitude [nm]

6 Applications I: MPC cantilever resonator

40 20 0 5

10 15 Frequency [kHz]

20

10 15 Frequency [kHz]

20

200 0 -200 5

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 generc (2009) IEEE). ated by Helmholtz coils ([111],

Table 6.3: Deflection of the cantilever in resonance with different magnetic actuation modes (at reduced pressure).

Excitation

Deflection [nm]

Thermal noise AC AC + DC

0.1 5.6 63.3

Results and discussion

The uniform magnetic field, H, generated by a DC current in the Helmholtz coils (DC-setup) increases the magnetization, M, of the MPC, approaching the saturation value for the composite as shown in the magnetization curve in Figure 5.13, resulting in an enhanced actuation (5.2). The resulting tip deflections of a 5 vol.% MPC cantilever excited with and without DC-mode are shown in Table 6.3. With the addition of the uniform DC field the tip deflection increased more than ten times. Using the additional uniform magnetic field the sign of the magnetization of the MPC cantilever does not change with the applied alternating inhomogeneous magnetic field (5.2) and the superparamagnetic cantilever is actuated mechanically in the same frequency as the applied magnetic field.

113

6 Applications I: MPC cantilever resonator 6.2.3 Magnetic actuation of MPC cantilevers with Q-enhancement 1 In

this section the influence of a positive feedback-loop (using a sweep input signal) on the actuation of a MPC cantilever is tested. A higher Q-factor increases the lowest detectable mass of a resonator sensor. Experimental

A positive feedback loop in combination with a laser-Doppler vibrometer (LDV) was used. Figure 6.10 shows the schematics of the positive feedback system. The velocity signal from the LDV of the cantilever oscillation is already phase shifted by 90◦ compared to the deflection signal of the cantilever. The phase is fine adjusted to correct the phase delays from the electronics and is amplified by a variable gain [152]. The increased signal from the second harmonics as described in the previous Section 6.2.2 cannot be used for measurements with this feedback loop circuit. The actuation must be performed at the same frequency as the deflection signal of the cantilever.

φ

Feedback

Read Out Signal

2+&3456% 789%+.

ω

Gain

Phase Shifter

Switch

-

Input

Output

+

)'*+, -./0$1$+,

!"#$%&%$'(

Feedback Loop Circuit

Figure 6.10: Schematic illustration of the feedback loop to enhance the Q-factor of the microresonators. [152]

Results and discussion

A MPC microcantilever with 5 vol.% nanoparticle concentration with dimensions of 100 µm x 14 µm and a thickness of 1.8 µm was actuated with (closed loop) and without a feedback loop (open loop) at room temperature. For comparison the frequency spectrum of the MPC cantilever excited by the ambient thermal noise was measured. Figure 6.11 shows the resonance behavior of the 5 vol.% MPC cantilever with the different excitations. The magnetic open-loop actuated cantilever 1

The feedback loop has been reported in [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 o p e n th e rm s c a le

5

d lo o p , lo o p , Q a l n o is d x 4 0 ,

Q

= 2 9 = 1 7 e , d e f le c tio n Q = 1 6

D e fle c tio n [n m ]

4 3 2 1 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 excitation, and magnetic actuation (AC-setup) with (closed loop) and without feedback loop (open loop).

6.2.4 Self-heating of cantilever during actuation by magnetic alternating field

When an alternating magnetic field is applied on the MPC with superparamagnetic particles, the particles generate heat due to relaxation losses. The effect of external heating was investigated in Section 5.6. In the presented experiment the temperature of a 4 vol.% composite samples was raised by +42 ◦ C in 90s. In this section the self-heating of the cantilever during magnetic actuation is discussed. The temperature of the cantilever in air during magnetic actuation can be es-

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 composite 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 generated 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 ratio 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.

6.2.5 Magnetic actuation of MPC cantilevers in water

MPC microcantilevers were actuated by magnetic fields in water to investigate if the MPC cantilevers can be used for possible sensor applications in water.

Experimental

A MPC cantilever with highest processable loading concentration and highest processable cantilever thickness was used for the actuation experiment in water to profit from the high magnetic volume (5 vol.% nanoparticle, dimensions of cantilevers: 112 µm x 17 µm x 2.9 µm), resulting in a higher actuation force. The microcantilevers were bonded into the PMMA package as described in Section 4.3. To enhance the reflection signal the cantilevers were coated by a thin gold layer of around 2 nm. For the magnetic actuation a coil (AC-setup) was used. The cantilever was placed in the water filled PMMA package around 2 mm above the coil. Because the measured velocity signal from the LDV was near the noise level the output signal was averaged 64 times to obtain a smoother resonance curve.

116

6.2 Magnetic actuation of superparamagnetic MPC cantilevers Results and discussion

Figure 6.12 shows the resonance behavior of the microcantilever actuated by an alternating magnetic field in water. For comparison the resonance curve of the same cantilever excited by thermal noise in water is also shown. The deflection of the cantilever when actuated with the magnetic field (∼25 pm) is enhanced compared to the thermally excited one (∼3 pm). The Q-factor of the cantilever is ∼1.7 (determined from the amplitude spectrum) for magnetic actuation and similar to the thermally excited cantilever, 1.9. (If the Q-factor is determined from the phase spectrum a slightly higher Q-factor ∼3 is obtained for the magnetic actuation). An exact determination of the Q-factor at this noise level is difficult. The resonance frequencies measured for magnetic and thermally excitations are in agreement. This magnetic actuation measurement demonstrates that the cantilever can be actuated magnetically in water. To compare, cantilevers made from CoNi with similar dimensions show a Q-factor of 4.5 measured in water with deflections up to 100 nm actuated by a remote magnetic field [152].

D e fle c tio n [p m ]

3 0 T h e r m a l a c tu a tio n , = 1 .9 , fr = 2 0 k H z Q

2 0

Q

M a g n e tic a c tu a tio n , ~ 1 .7 , fr = 2 0 k H z

1 0 d e fle c tio n m u ltip lie d b y fa c to r 3

0 5

1 0

1 5

2 0

2 5

3 0

3 5

4 0

P h a s e [d e g re e ]

1 8 0 P h a s e m a g n e tic a c tu a tio n

9 0 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 actuation (AC-setup) without feedback loop. Cantilever dimensions: 112 µm x 17 µm x 2.9 µm.

117

6 Applications I: MPC cantilever resonator The minimum detectable frequency shift, ∆ f , of a cantilever is determined by the Q-factor. Therefore, it is important to obtain a high Q-factor. The Q-factor can be enhanced by a feedback loop as demonstrated for the actuation in air. However, because of the high fluctuation of the signal (amplitude and phase) due to the low cantilever deflection it was not possible to use this feedback setup in water. The feedback-loop parameters (gain and phase shift) could not be set properly. Q-enhancement systems can be used in self-excitation mode (without an external frequency sweep signal), as reported by [148] and [147]. Cantilever Qfactors of cantilevers of ∼1 could be increased in water to an effective Q-factor of 31 using such systems. Combining such a system with magnetic actuation has the potential to enable the use of the MPC cantilevers for low-cost disposable sensing applications in liquid. However, because the magnetic actuation setup has strong mechanical resonances, this actuation is challenging. An actuation with self-excitation has not yet investigated in detail.

6.3 Conclusion The mechanical behavior of the MPC cantilever (resonance frequency and Qfactor) in different media was investigated. Measuring thermally induced vibrations of cantilevers were suitable for their resonance characterization. The resonance frequencies and Q-factors measured by thermal excitation are in agreement with those actuated by magnetic fields. The MPC cantilevers with 5 vol.% Fe3 O4 nanoparticle concentration were successfully actuated in air and in water by magnetic fields. The actuation mechanism of superparamagnetic resonators were investigated using different actuation setups. Due to the superparamagnetic behavior of the composite, the cantilevers have the highest deflection amplitude at the second harmonic of the magnetic actuation frequency when excited solely by an alternating magnetic field, generated by a single coil (AC-setup). The deflection amplitude can also be increased (x10) by the use of an additional uniform magnetic field (AC + DC-setup). The uniform magnetic field increases the magnetization of the MPC, approaching the saturation value for the composite. This results in a higher magnetic force and a higher deflection. It was shown that Q-factors of cantilevers in air can be significantly improved using a positive feedback loop circuit. The Q-factor of the cantilevers could be increased from 15 to 29. From the actuation tests it can be concluded that the actuation with the AC+DC-setup in combination with a feedback loop is most favorable. It has been shown in Section 5.6 that the MPC can be heated when an altern-

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.% magnetite particles cannot be fully excluded. However, a temperature increase smaller 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]. Polymer cantilevers are candidates for humidity sensing [60] or the detection of disease 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 miniaturized and the sensitivity and performance of the device must be tested in detail. 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 nanoparticle 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 additional 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 measurements 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 developed 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 microresonators. 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 material 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 resonance frequency of the microstructure. To detect inductively the resonance frequency 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 deflections of the resonator. A magnetic readout of a resonant magnetic microstructure (sensor) needs no electrical contact to the device. Furthermore, it has the advantage 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 frequencies can be achieved. In the following the design and the fabrication of an MPC in-plane microresonator is presented. The resonance behavior, damping and deflections of the resonators 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, listed 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 nanoparticle concentration.

Table 7.1: Variation in designs for lateral resonators

Beam length LB

Beam width w

Plate length s

Thickness h

Concentration

10 - 600 µm

3 - 25 µm

3 - 400 µm

1.9 µm

5 vol.%

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.% magnetite 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 Qfactor of a plate-like in-plane structure in air with infinite distance to the substrate can be described by [155] ωη A D 1 ≈ Qlat δP k

122

(7.1)

7.1 Design

Table 7.2: Parameters used for the in-plane resonator simulations

Density ρ kg/m3

Young’s modulus Y GPa

Poisson’s ratio v -

1416 (Table 5.8)

5.1 (Table 5.8)

0.26 [129]

where ω is the angular frequency of the resonator, η is the dynamic viscosity of the ambient fluid, A D is the damping-related effective area of the system, and k is the spring constant. δP describes the penetration depth, i.e. the distance at which the motion amplitude of a fluid decreases by e1 [155] s δP ( ω ) =

2η ωρ

(7.2)

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 substrate and the plate the additional damping effect by the substrate can be neglected. The resonance frequency of the in-plane resonator should be in the range of 1 – 30 kHz. The structure thickness is given by the fabrication limitations. In-plane resonators with thicknesses, h, of 1.9 µm were fabricated. Higher thicknesses are advantageous to obtain a high magnetic volume. However, the structures are limited to a maximum thickness of 2.9 µm by the fabrication process (Chapter 4). Structures with thicknesses smaller than 1.9 µm have a too low magnetic volume, and are susceptible to fabrication failures. 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-ofplane) 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 2 3

vertical lateral torsional

1090 1860 3065

7.2 In-plane resonance characterization by thermally induced vibrations In this section the investigation of the resonance frequency of the in-plane resonators 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.

7.3 Results and Discussion 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

M e a s u re m S im u la tio n T h e o r e tic a o f b e a m w T h e o r e tic a o f b e a m w ( u s in g d e s

1 2 1 0

ts in v a c u u m s u lts in v a c u u m a lc u la tio n e n d m a s s a lc u la tio n e n d m a s s g e o m e tr ie s )

s

L B = 8 0 µm

8

e n re l c ith l c ith ig n

s

R e s o n a n c e fre q u e n c y (k H z )

1 4

6 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 measured 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 values 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 microresonators: 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 performed. The dimensions of the resonators are individually measured by optical microscopy. For each plate size the measured mean value is taken for the simulations. 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%. Reasons for this discrepancy can be simulation uncertainty, temperature variation, residuals 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 magnetic actuation forces, with a resonant frequency of approximately 1.6 kHz were chosen for the magnetic actuation described in the next section.

7.4 Magnetic actuation of in-plane resonator In this section the magnetic actuation of superparamagnetic in-plane microresonators in air are investigated. 7.4.1 Experimental

The actuation of superparamagnetic cantilevers have shown that the best magnetic 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 radial 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 amplitude 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 LB

Beam Width w

Plate Length s

Thickness 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 actuation of the microresonators, the deflections must be recorded by the laserDoppler 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 ∂ H of 0.43·106 A/m2 from the coils at the position of the microresonator. 5 of ∂x in-plane microresonators from the same fabrication batch were tested. Their geometric dimensions, measured by an optical light microscope Leica DM4000, are listed in Table 7.4. The thickness h is determined by a contact profilometer (Tencore P10). The measurements in vacuum were performed at pressures < 10 Pa. 7.4.2 Results and Discussion

The resonance characteristic of the 5 in-plane microresonators were tested in vacuum and in air. In air, the structures were actuated by a magnetic field. Thermal ambient noise (room temperature) was used for actuation in vacuum. The results are summarized in Table 7.5. The resonant frequency in air is slightly reduced

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 deflection 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 resonators are slightly lower (14) than the ones measured with the cantilevers (16) in air. Table 7.5: Resonance behavior of MPC in-plane microresonators. Media

Actuation method

Resonance frequency

Q factor

Deflection

Measurement method

Vacuum b Air Air

Thermal Thermal Magnetic

1.67 ± 0.03 kHz 1.62 ± 0.02 kHz 1.61 ± 0.02 kHz

54 ± 17 14 ± 6 14 ± 1

223.5 ± 67.3 nm 86.8 ± 7.8 pm 28.2 ± 2.2 µm a

LDV LDV PMA

only three devices measured at 8.6 Pa.

D e fle c tio n

m a g n e tic

1 4 0

3 0

M a g n e tic a c tu a tio n Q = 1 5 , fr = 1 .6 4 k H z

1 2 0

2 5 Q

2 0

T h e r m a l a c tu a tio n = 1 5 , fr = 1 .6 4 k H z

1 0 0 8 0

1 5 6 0

1 0

4 0 5

2 0 0 1 .0

1 .5

[p m ]

[ µm ]

3 5

th e rm a l

b

D e fle c tio n

a

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

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 deflection 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 represents roughly the situation used in the calculation. The magnetization or the deflection 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 magnetic 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 microstructure 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 reproducible measurements. The Q-factor is inversely proportional to the dynamic viscosity (7.1) of the surrounding media. Therefore, a higher damping is expected 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 photoinitiator into the composite [4]. This makes the composite more sensitive to UVlight 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 modulus, density, maximal layer thicknesses depending on the nanoparticle concentrations, were used to design an in-plane resonator structure. The resonator structures 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 suggested 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 polymerization (TPP). Fabrication parameters such as writing speed and laser power, and limitations such as minimal line resolution depending on the filler concentration 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 helical 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 different 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 millimeters, which can be controlled in a human eye by applying magnetic fields and measuring the local oxygen concentration [21]. However, the movement of miniaturized objects in liquid environments faces the challenges of increased „viscous forces“. For small objects the Reynolds number decreases due to the small characteristic length, Lch , of the object [158]:

131

8 Application III: MPC 3D-microstructures by two-photon polymerization

Re =

inertial f orces v0 Lch ρ ∼ η viscous f orces

(8.1)

where, v0 , is the free-stream velocity and ρ and η the density and dynamic viscosity of the fluid, respectively. One of the swimming strategies in this scale can be adapted from bacterias like escherichia coli (E. coli). They use flagellas, helical filaments that rotate in cork-screw motion, to achieve a forward swimming movement. Such a cork-screw motion can be generated by a magnetic torque on magnetic microstructures with helical shapes. However, the fabrication of such microstructures is a challenge. The fabrication and swim performance of an artificial bacterial flagella (ABF) containing a hybrid semiconductor-metal trilayer with soft-magnetic head has been reported [22]. The reported microfabrication, involving different materials, is challenging and complex. To use such microrobots for biological relevant applications biocompatibility of the used materials and the possibility of biofunctionalization of microstructure are important. Furthermore, the research of the swimming performance in this microscales is not yet fully understood and methods which allow design variation of microobjects for swim test performance are desired. Photocurable MPC structures fabricated with two-photon polymerization technique offer solutions in this field. The properties of the MPC (investigated in Chapter 5) such as biocompatibility, high mechanical stability (high Young’s modulus) and attributes such as high thermal stability and chemical resistivity of the polymer matrix make this composite to a promising material for microrobotic applications in fluids. The fabrication of MPC microstructures using TPP in comparison to standard photolithography has much less limitations on shape design. Using UV photolithography at 350 – 410 nm the MPC has a low transmittance and the fabrication is limited to layer thickness of 2.9 µm for a composite with 5 vol.% magnetite concentration. Whereas, at longer wavelength, which is used for the laser for TPP (780 nm), the MPC is much more transparent (Figure 8.1). Therefore, the laser can penetrate deeper in the composite and higher exposure thicknesses are expected.

8.2 State-of-the-art TPP The fabrication of polymer 3D-microstructures using TPP have been reported [159–162]. One of the first work using photocurable MPC in combination with a laser to produced complex magnetic 3D-microstructures was presented by Kobayashi [31] using microstereolithography. Ferrite particles FA-700 (Toda Kogyo Corp) with a mean particle size of 1.3 µm up to 50 wt.% have been mixed in

132

8.2 State-of-the-art TPP

T P P la s e r

U V - P h o to lith o g r a p h y

M P C

T r a n s m itta n c e [% ]

1 0 0 8 0 0 v o 1 v o 2 v o 3 v o 5 v o

6 0 4 0

l.% l.% l.% l.% l.%

2 0 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 photolithography 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

photocurable polymer (SCR770, C-MEC-Ltd). Because of the strong magnetic attraction between the ferromagnetic particles, due to their remanent magnetization, agglomerates of around 50 µm are formed in the composite. Using a viscosity-increasing agent (5 wt.%) the agglomerate size could be reduced to few micrometers. 3D-micro-structures with various shapes (microscrew, microfan) with sizes of 0.25 mm to 2 mm are presented. Leigh [43] demonstrated a flow sensor (with millimeter size) fabricated by microstereolithography using magnetite nanoparticles (Sigma-Aldrich, UK) with 50 nm diameter (up to 25 wt.%) in acrylic resin formulation (Envisiontec, R11, 25 µm voxel depth). The nanoparticle agglomeration was reduced by using a more viscous composition, however, no information about the agglomerate size is given in the publication. Due to the constraint of microstereolithography and large agglomerates in the composite these two presented techniques are limited for the fabrication of microstructures with minimal feature sizes of tens of micrometers or above. Recently, Xia [34] reported the fabrication of MPC microstructures using TPP technique. Superparamagnetic Fe3 O4 nanoparticles with diameters of around 10 nm are synthesized and surface modified by 3-(trimethoxysilyl) propyl methacrylate (MPS), which imparts dispersible properties. MPS-Fe3 O4 particles were doped in methyl acrylate (monomer), pentaerythritol triacylate (cross-linker) with

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].

an additional photoinitiator and photosensitizer to obtain a stable magnetic colloidal dispersion of photopolymerizable resin. A stable suspension up to 20 wt.% could be achieved. MPC microsprings (with a filler concentration of 2.1 wt.%) with a bead at the tip were fabricated with a 790 nm Ti-sapphire laser. Approaching an external permanent magnet the spring elongates and bends towards the magnet. Furthermore, a magnetic microturbine, shown in Figure 8.2, with diameter of ∼35 µm was fabricated with the same nanoparticle concentration as the springs. Tian [46] improved the particle stability in the composite resulting in a lower surface roughness and allows the fabrication of even smaller microturbines with diameters down to 14 µm. This was obtained by surface modification of the Fe3 O4 particles using propoxylated trimethylolpropane triacrylate (PO3 -TMPTA) and the dispersion of the particles in a photosensitive resin based on butyl methacrylate. However, the nanoparticle dispersion in the composite was not investigated. The microturbines with particle concentration up to ∼5 wt.% were actuated by a rotating external ferromagnet with rotating rate of about 300 rpm in acetone. The principle of magnetic actuation is not described in the publications. It is assumed that the rotation of the symmetric superparamagnetic microturbine with an external ferromagnet piece is achieved because of an additional MPC mass on one of the wings, as it can be seen in the SEM images Figure 8.2. This additional mass breaks the microturbine’s point symmetry and generates a resulting magnetic force on the superparamagnetic structure for rotation. The discussed works are summarized in Table 1.2 and 1.3.

8.3 Theory of TPP The electrons of a UV sensitive component in a photoresist get excited if they absorb a photon with the required amount of energy and jump from their ground state to the excited state. If the frequency of the photon is less, its energy will not be enough to excite the electrons and initiate the polymerization. The energy of a photon is given by

134

8.3 Theory of TPP

Figure 8.3: Two-photon exposure of 3D structures with a tightly focused femtosecond pulsed laser beam. At the focus the threshold for photon absorption is exceeded. Reprinted from [163].

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 simultaneously (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 8.4.1 Direct laser writing tool

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 microscope, a Ti:sapphire near IR femtosecond laser (100 MHz repetition rate, sub150 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 writing 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. 8.4.2 Sample preparation

The MPC suspension was deposited by a drop or by spin-coating (layer of 10 – 30 µm) on a 170 µm thick glass substrate. The prebaking for spin-coated samples was 95◦ C for 15 minutes and for drops 95◦ C up to 0.5 – 6 hours (depending on the drop-size). Prebaking was used to obtain a lower solvent concentration in the MPC. After exposure by TPP all the samples were postbaked at 95◦ C for 3 minutes to perform the polymerization. Then, the composite was developed in resist developer MR-DEV 600 for 5 to 10 minutes, rinsed with isopropanol and dried at air. Samples with Fe3 O4 particle concentration of 0, 2, 4 vol.% were prepared to investigate the optimal writing parameters for each filler concentration. The fabrication limitations such as minimal line width, w L , and line height, h L , were explored. 8.4.3 Magnetic actuation setup

A uniform rotating magnetic field was used, to test the actuation and swimming properties of the MPC-based microstructures. The setup consists of three Helmholtz coils placed orthogonally to each other producing uniform field in any direction. 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 micostructures 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 different offsets to the substrate were written and the linewidth, w L , was measured by SEM.

8.5 Fabrication limitations The influence of the nanoparticle concentrations in the MPC on the fabrication parameters using TPP was investigated. The line resolution (thickness and width) is investigated to fabricate 3D-microstructures with minimal feature sizes. 8.5.1 Line resolution

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 material. 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 zdirection to the substrate surface as illustrated in Figure 8.4, because the focus relatively 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. Results and discussion

Figure 8.6 shows the written lines of a 2 vol.% MPC after development for different 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 scanning 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.

writing larger microstructures it is advantageous to write with a higher scanning speed to minimize the total writing time. However, in general the quality of the structures with lower scanning speed were higher than for high scanning speeds. Figure 8.7 shows the summary of the line width results of the 0, 2 and 4 vol.% MPC. When the particle concentration is increased the parameters to obtain minimal line resolution is shifted to lower scanning speeds. This is due to the scattering and absorption of the laser by the incorporated Fe3 O4 nanoparticles. For increased filler concentration either the scanning speed can be lowered or the power increased. The nanoparticle concentration is limited on one hand by the maximal acceptable fabrication time by using lowest scanning speed. On the other hand the power is limited by the heating of the polymer due to the absorbed energy resulting in total destruction of the composite microstructure. Figure 8.8 (a) shows a SEM image of a MPC (2 vol.%) fabricated woodpile structure

139

8 Application III: MPC 3D-microstructures by two-photon polymerization

L a s e r P o w e r [m W ]

2 .4 2 .0 1 .6 0 v o 0 v o 0 v o 2 v o 2 v o 2 v o 4 v o 4 v o 4 v o

1 .2 0 .8 0 .4 1

1 0

1 0 0

l% l% l% l% l% l% l% l% l%

s ta d e n o s ta d e n o s ta d e n o

b le fo rm lin e b le fo rm lin e b le fo rm lin e

lin e e d lin e lin e e d lin e lin e e d lin e

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 successful 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 microstructure 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 Figure 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. 8.5.2 Fabrication of 3D structures

To build a 3D bulk structure, the volume of an object is defined with parallel slightly overlapping lines. The line separation distance (slicing distance) between two lines depends on the line width and height. To write a defined 3D-structure this slicing distance has to be adjusted. Figure 8.9 shows a 3D-structure of pure

140

8.5 Fabrication limitations

Figure 8.8: SEM image of a MPC (2 vol.%) fabricated woodpile structure with (a) optimized 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.

5 µm/s, 1 mW 12 µm/s, 1.4 mW 25 µm/s, 1.8 mW 50 µm/s, 2 mW Average

Minimal line width

Minimal line height

350 nm 338 nm 318 nm 251 nm

1.46 µm 1.52 µm 1.11 µm

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 designed 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 optimized 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 slicing 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 exposure). For too large slicing distance (c) the surface is rippled because the lines are too far separated. (b) Shows the microstructure with optimized slicing distances. 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 windows (Figure 8.10 (b) and (c)). Hollow microstructures can be interesting for possible drug delivery applications.

8.6 Fabrication of helical microstructures 1 Magnetic

helical structures are used to mimicking natural bacteria flagellas [22]. Helical microstructures with the MPC can be written by the laser following helical trajectories. The height and width of the filament can be determined by the number of neighbor trajectories as illustrated in Figure 8.11. An overlap of trajectories of around 50% in vertical and horizontal direction was chosen. Figure 8.12 1

The helical microstructures have been fabricated and actuated in cooperation with Li Zhang (Institute 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 concentration: (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

power scanning speed trajectories (x,z) trajectory distances (x,z)

2 vol.%

4 vol.%

0.8 mW 5 µm/s 5x2 200/400

0.8 mW 1 µm/s 5x2 200/400

shows successfully fabricated helical microstructures with 2 and 4 vol.% Fe3 O4 nanoparticle 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 advantageous for biofunctionalization (drug release) because of increased surface area. Figure 8.13 (a) and (b) shows 2 vol.% MPC helical microstructures with different 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 adjusted 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 (horizontal) 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.

8.7 Magnetic actuation of MPC helical microstructure The fabricated superparamagnetic MPC helical microstructures with 2 vol.% filler concentration were successfully rotated in water along the axial direction of the helical microstructure by applying a magnetic torque with a uniform rotating magnetic field. The helical microstructure show cork-screw motion in water. The swim performance on the substrate surface is characterized by two motions, forward and drift motion. An upwards swimming against the gravitational field could not yet be proven. Figure 8.14 shows the rotation of the helical-like microstructure in water. The time domain between the images t1 , t2 and t3 are around 0.04 s. At t3 a quarter rotation is performed. After 3 s the structure displacement was around 12 µm (forward and drift motion). The cube on the bottom of the helical structure was used as anchor during fabrication, helical microstructures without cubes were also fab-

145

8 Application III: MPC 3D-microstructures by two-photon polymerization

t 1

t 3

t 2

20µm

Mot i ondi r ect i on

Magnet i cf i el dr ot at i on

Figure 8.14: Swim test in water of MPC helical structure with 2 vol.% Fe3 O4 superparamagnetic nanoparticle concentration. The images show a quarter rotation of the superparamagnetic 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 superparamagnetic MPC with only 2 vol.% nanoparticles concentration is possible. To achieve a rotation of the helical microstructure a magnetic torque perpendicular to the helical axis must act on the structure. The magnetic torque applied on the helical microstructure is given by 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 microstructure 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 magnetic 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

8.8 Conclusion and Outlook The line resolution measurements have shown that with the MPC with 2 vol.% nanoparticle 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 maximal 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 superparamagnetic 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 surface. The direction of the movement can be controlled by applied magnetic fields. Superparamagnetic MPC helical microstructures have potential for biomanipulation and drug delivery [164]. Using the developed magnetic polymer composites for magnetic remotely controlled microstructures/microrobots has the following advantages: • Human cells proliferate on composite (composite is not toxic to cells) as shown in Section 5.8 and can be used with cell tissue, • The composite has high mechanical stability (Young’s modulus: 4.2 – 5.1 GPa, see Section 5.5) , • The composite can be heated in an alternating magnetic field and eventually trigger a temperature sensitive bioreaction at the surface of a microstructure/robot, • 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 environments 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 composite (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 properties. The composite has superparamagnetic characteristics and shows good compatibility with biomaterials. It has a relatively high Young’s modulus compared to other polymers, which allows the fabrication of stable suspended microstructures. A key advantage of the composite is the homogeneous dispersion of the nanoparticles with low particle agglomerations, which enables the fabrication 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 possible 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 magnetic actuation of superparamagnetic resonant microstructures by the use of different magnetic actuation setups are presented. Furthermore, three dimensional (3D) microstructures were fabricated by two-photon polymerization (TPP). Helical microstructures show cork-screw swimming behavior in water when a rotational uniform magnetic field is applied.

9.1 Contributions Material evaluation and composite fabrication

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 the particles tend to form agglomerates. In order to maintain uniform magnetic and material properties, the distribution of particles must be homogeneous within the matrix and agglomerates must be avoided. Superparamagnetic nanoparticles, which do not retain remanent magnetization were selected as filler material. The superparamagnetic particles exhibit negligible magnetic attraction compared to ferromagnetic particles during and after the mixing of the polymer composite and this reduces particle-agglomerate formation. For the polymer matrix, the negative tone photodefinable SU-8 epoxy resist was chosen because of the high chemical stability, possibility for surface functionalization due to the epoxy binding sites, and high glass transition temperature. Additionally, a suitable surfactant for the Fe3 O4 particles was evaluated. To obtain low particle agglomerates stable nanoparticle suspensions with particle concentrations of up to 280 mg/ml were mixed with the polymer using centrifugal mixing and ultrasonic steps. Stable low viscosity MPC suspensions with nanoparticle concentration of up to 10 vol.% were obtained. These suspensions can be used for spin coating processes. Using superparamagnetic particles in combination with the evaluated surfactant, a homogeneous particle dispersion in the composite with low agglomerate sizes (i.e., ∼50 nm) was achieved. MPC microstructure fabrication with UV lithography

MPC can be structured by photolithography to fabricate magnetic microstructures. Due to the incorporated Fe3 O4 nanoparticles the UV transmittance of the MPC layer is reduced and the exposure doses must be adjusted to ensure full polymerization in the composite layer. To fabricate cantilevers the UV exposure dose was optimized for different particle concentrations and layer thicknesses. MPC microstructures can be fabricated with 5 vol.% Fe3 O4 nanoparticles with a maximum layer thickness of 2.9 µm. Microstructures with widths down to 1.3 µm (thickness of 1.8 µm) could be fabricated with a 5 vol.% MPC without reaching the resolution limit of the composite. MPC microstructure fabrication with two-photon polymerization (TPP)

Complex 3D polymer microstructures can be fabricate by TPP technique. With the developed MPC superparamagnetic microstructures with nanoparticle concentrations up to 4 vol.% were fabricated (written by a laser). Complex shapes such as hollow cubes, woodpiles, and helical microstructures were fabricated. The main advantage of the TPP fabrication technique is that it uses a laser with a wavelength at 800 nm, where the composite is much more transparent compared to the UV exposure using standard photolithography (i.e. 400 nm). The

150

9.1 Contributions optimal fabrication parameters and limitations were investigated depending on the nano-particle concentration in the MPC. It was shown that the laser power and the writing speed must be decreased if the nanoparticle concentration is increased. The nanoparticle concentration for the MPC is limited to 4 vol.%. A minimal line width of 314 nm and a line thickness of 1.36 µm were obtained for a 2 vol.% MPC. Using TPP, structures with vertical dimensions up to 6.8 µm for MPC with 4 vol.% nanoparticles can be achieved, whereas, for UV photolithography only layer thicknesses of 2.9 µm for a 5 vol.% MPC can be fabricated. Using TPP technique for the fabrication of MPC microstructures, higher vertical structure sizes, higher aspect ratios and smaller feature sizes can be obtained compared to UV photolithography. Furthermore, complex 3D microstructures can be fabricated, which cannot be obtained by other microfabrication processes. The disadvantage of TPP laser writing is the serial fabrication process and, hence, the long writing time. Nanoparticle dispersion in MPC

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 investigated 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 analysis 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 suspension and negligible agglomerate formation is occurring during the mixing of the composite. Magnetic characteristics and mechanical properties

Magnetic properties of the MPC are crucial for the microdevice performance. Saturation 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 composite and the nanoparticles exhibit consistent magnetic behavior. The magnetic properties of the composite allows remote actuation of microstructures. A saturation 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 microcantilever 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.

Heating by alternating magnetic field

There is growing interest in remote heating of polymers for various applications such as controlling bioreactions on polymer surfaces. The MPC presented in this work is heated by an alternating magnetic field at high frequencies. A rise in temperature of 42 ◦ C in 90 s was observed in a magnetic field of 3 kA/m at 245 kHz for a spin-coated MPC sample with a diameter of 3 cm and a thickness of ∼250 µm. The remote heating property of the MPC makes the composite and its microstructures interesting for localized heating applications.

Suitability for bioapplications

The biocompatibility of the MPC is crucial for bioapplications. Biocompatibility is defined based on the applications and durations of interaction with the living tissue [139]. The biocompatibility of the MPC were investigated by proliferation of human foreskin fibroblasts (NDHFs). Cell viability tests, and live and dead staining tests showed that the cells’ proliferation on the surface of MPC samples is independent of the nanoparticle concentration (0 – 10 vol.%) for 24 hours. The cell viability was over 86% for all MPC nanoparticle concentrations and MPCs show non-toxic behavior. Therefore, MPC and MPC microstructures with up to 10 vol.% nanoparticle concentration are suitable for in vitro interactions with biomaterials within 24 hours. For in vivo applications within this time span, additional allergy tests for animals and humans are necessary and the nanoparticle absorption by cells should be investigated. Furthermore, water contact angle measurements on surfaces of MPC (0 – 10 vol.%) show that the nanocomposite’s surface polarity does not differ significantly with respect to pure SU-8, and exhibits a moderate hydrophobic behavior (advancing dynamic contact angles approximately 81◦ ).

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 magnetic fields. It was found that the addition of a uniform magnetic field improves the deflection of the microresonators by increasing the magnetization of the microstructure. 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 performance have been reported recently [166]. However, a magnetic actuation of the cantilevers was not demonstrated due to the too low nanoparticle concentration (0.16 wt.%). In-plane microresonators An in-plane MPC microresonator with 5 vol.% particle concentration was designed 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). Artificial bacteria flagella Helical microstructures mimicking natural bacteria flagellas were fabricated using TPP laser writing technique. Using uniform rotating magnetic fields the 2 vol.% superparamagnetic composite helical microstructures were successfully propelled in water by cork-screw motion. It was demonstrated that they can be steered near a solid surface. Comparison of the developed MPC with literature

Reported photodefinable MPCs in literature are mainly based on ferromagnetic particles, which profit from the high remanent and saturation magnetization. However, in general they are limited to larger feature sizes (> 5 µm) due to large particles or large agglomerates. Only few photodefinable MPCs based on superparamagnetic particles for smaller feature sizes using UV photolithography or TPP laser writing are reported [45, 46]. A quantitative analysis about nanoparticle size and agglomerates in the composite, and the influence of the nano-

153

9 Conclusion and Outlook particle concentration on fabrication limitations have not been addressed so far. The achieved maximum nanoparticle concentrations in the developed MPC are much higher (≥ 3 times) than the ones reported in literature for photocurable MPCs for similar feature sizes [45, 46, 166]. A high nanoparticle concentration in the composite is crucial for the generation of magnetic forces and torques on the microstructures. The maximum nanoparticle concentrations for the developed MPC for both fabrication methods were evaluated. Microstructures with a maximum nanoparticle loading of 5 vol.% (corresponding to 18 wt.%) using UV photolithography and 4 vol.% (corresponding to 15 wt.%) using TPP laser writing could be obtained. The homogeneous dispersion of the nanoparticles with low particle agglomerations allows the fabrication of microstructures with minimum feature sizes down to 1 µm for helical filaments and ∼300 nm for single lines. Microstructures (microturbine) with sizes down to ∼14 µm using TPP have been reported for a particle concentration of 5 wt.% (< 2 vol.%) [46]. Microstructures, fabricated with the MPC presented in this work, have higher nanoparticle concentration at smaller feature sizes compared to reported MPC microstructures. The presented MPC is one of the most thoroughly characterized photodefinable MPC reported up to now. The properties of the 5 vol.% MPC are summarized in Table 5.8 and compared to unfilled SU-8.

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 characteristics, bio-compatibility, and high mechanical stability. It can be remotely heated by applied magnetic fields. A key advantage of the composite is the homogeneous dispersion of the nanoparticles with low particle agglomerations, which enables 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 actuation 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 detection 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 microturbines for lab-on-a-chip applications [46]. Friction forces often limit the efficient 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 microstructures using magnetic fields [45]. MPCs in general benefit from the availability of a variety of fabrication methods such as hot embossing [27], photolithography [28], and injection molding [35] (see Table 1.1). Fabrication processes such as inkjet printing (drops) or electro spinning (fibers) are often not considered for MPC fabrication because particle agglomerates in the composite leads to clogging of the pinhole. Due to the homogeneous 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 immunoassay tests in liquid, allowing an easy collections after the experiments by applied magnetic fields. Inkjet printed MPC spheres can be used for self assembly applications 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 Guttapercha in Verbindung mit anderen Stoffen,” Dinglers Polytechnisches Journal, vol. 139, p. 376, 1856. [2] A. C. Balazs, T. Emrick, and T. P. Russell, “Nanoparticle Polymer Composites: Where Two Small Worlds Meet,” Science, vol. 314, no. 5802, pp. 1107– 1110, 2006. [3] D. Godovsky, Device Applications of Polymer-Nanocomposites. lin / Heidelberg, 2000, vol. 153, pp. 163–205.

Springer Ber-

[4] S. Jiguet, A. Bertsch, H. Hofmann, and P. Renaud, “Conductive SU8 photoresist for microfabrication,” Advanced Functional Materials, vol. 15, no. 9, pp. 1511–1516, 2005. [5] S. Jiguet, A. Bertsch, M. Judelewicz, H. Hofmann, and P. Renaud, “SU-8 nanocomposite photoresist with low stress properties for microfabrication applications,” Microelectronic Engineering, vol. 83, no. 10, pp. 1966–1970, 2006. [6] A. Voigt, M. Heinrich, C. Martin, A. Llobera, G. Gruetzner, and F. PérezMurano, “Improved properties of epoxy nanocomposites for specific applications in the field of MEMS/NEMS,” Microelectronic Engineering, vol. 84, no. 5-8, pp. 1075–1079, 2007. [7] S. Jiguet, M. Judelewicz, S. Mischler, A. Bertch, and P. Renaud, “Effect of filler behavior on nanocomposite SU8 photoresist for moving micro-parts,” Microelectronic Engineering, vol. 83, no. 4-9, pp. 1273–1276, 2006. [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 Microengineering, vol. 14, no. 7, p. 969, 2004.

157

Bibliography [10] J. W. L. Zhou, C. Ho-Yin, T. K. H. To, K. W. C. Lai, and W. J. Li, “Polymer MEMS actuators for underwater micromanipulation,” Mechatronics, IEEE/ASME Transactions on, vol. 9, no. 2, pp. 334–342, 2004. [11] N. Chronis and L. P. Lee, “Electrothermally Activated SU-8 Microgripper for Single Cell Manipulation in Solution,” Microelectromechanical Systems, Journal of Microelectromechanical Systems, vol. 14, no. 4, pp. 857–863, 2005. [12] S. R. Quake and A. Scherer, “From Micro- to Nanofabrication with Soft Materials,” Science, vol. 290, no. 5496, pp. 1536–1540, 2000. [13] F. Umbrecht, P. Wägli, S. Dechand, F. Gattiker, J. Neuenschwander, U. Sennhauser, and C. Hierold, “Wireless implantable passive strain sensor: design, fabrication and characterization,” Journal of Micromechanics and Microengineering, vol. 20, no. 8, p. 085005, 2010. [14] V. Seidemann, S. Bütefisch, and S. Büttgenbach, “Fabrication and investigation of in-plane compliant SU8 structures for MEMS and their application to micro valves and micro grippers,” Sensors and Actuators A: Physical, vol. 97-98, pp. 457–461, 2002. [15] C. Goll, W. Bacher, B. Büstgens, D. Maas, R. Ruprecht, and W. K. Schomburg, “An electrostatically actuated polymer microvalve equipped with a movable membrane electrode,” Journal of Micromechanics and Microengineering, vol. 7, no. 3, p. 224, 1997. [16] E. Smela, “Conjugated Polymer Actuators for Biomedical Applications,” Advanced Materials, vol. 15, no. 6, pp. 481–494, 2003. [17] J. D. Madden, R. A. Cush, T. S. Kanigan, and I. W. Hunter, “Fast contracting polypyrrole actuators,” Synthetic Metals, vol. 113, no. 1-2, pp. 185–192, 2000. [18] O. Cugat, G. Reyne, J. Delamare, and H. Rostaing, “Novel magnetic microactuators and systems (MAGMAS) using permanent magnets,” Sensors and Actuators A: Physical, vol. 129, no. 1-2, pp. 265–269, 2006. [19] I. J. Busch-Vishniac, “The case for magnetically driven microactuators,” Sensors and Actuators A: Physical, vol. 33, no. 3, pp. 207–220, 1992. [20] M. R. J. Gibbs, E. W. Hill, and P. J. Wright, “Magnetic materials for MEMS applications,” Journal of Physics D-Applied Physics, vol. 37, no. 22, pp. R237– R244, 2004.

158

Bibliography [21] O. Ergeneman, G. Dogangil, M. P. Kummer, J. J. Abbott, M. K. Nazeeruddin, and B. J. Nelson, “A magnetically controlled wireless optical oxygen sensor for intraocular measurements,” Sensors Journal, IEEE, vol. 8, no. 1, pp. 29– 37, 2008. [22] L. Zhang, J. J. Abbott, L. Dong, B. E. Kratochvil, D. Bell, and B. J. Nelson, “Artificial bacterial flagella: Fabrication and magnetic control,” Applied Physics Letters, vol. 94, p. 064107, 2009. [23] A. G. Olabi and A. Grunwald, “Design and application of magnetostrictive materials,” Materials & Design, vol. 29, no. 2, pp. 469–483, 2008. [24] M. L. Johnson, J. Wan, S. Huang, Z. Cheng, V. A. Petrenko, D.-J. Kim, I. H. Chen, J. M. Barbaree, J. W. Hong, and B. A. Chin, “A wireless biosensor using microfabricated phage-interfaced magnetoelastic particles,” Sensors and Actuators A: Physical, vol. 144, no. 1, pp. 38–47, 2008. [25] J. Yakhmi, “Molecule-based magnets,” Bulletin of Materials Science, vol. 32, no. 3, pp. 217–225, 2009. [26] F. Carosio, A. Fina, and M. Coisson, “Polypropylene-based ferromagnetic composites,” Polymer Bulletin, vol. 65, no. 7, pp. 681–689, 2010. [27] 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 cantilevers,” in 16th International Solid-State Sensors, Actuators and Microsystems Conference (TRANSDUCERS), 2011, pp. 398–401. [28] N. Damean, B. A. Parviz, J. N. Lee, T. Odom, and G. M. Whitesides, “Composite ferromagnetic photoresist for the fabrication of microelectromechanical systems,” Journal of Micromechanics and Microengineering, vol. 15, no. 1, pp. 29–34, 2005. [29] P. Montoya, F. Jaramillo, J. Calderon, S. C. de Torresi, and R. Torresi, “Evidence of redox interactions between polypyrrole and Fe3 O4 in polypyrroleFe3 O4 composite films,” Electrochimica Acta, vol. 55, pp. 6116 – 6122, 2010. [30] S. S. Keller, N. Feidenhans’l, N. Fisker-Bodker, D. Soulat, A. Greve, D. V. Plackett, and A. Boisen, “Fabrication of biopolymer cantilevers using nanoimprint lithography,” Microelectronic Engineering, vol. 88, no. 8, pp. 2294–2296, 2011. [31] K. Kobayashi and K. Ikuta, “Three-dimensional magnetic microstructures fabricated by microstereolithography,” Applied Physics Letters, vol. 92, no. 26, p. 3, 2008.

159

Bibliography [32] Y. K. Sung, B. W. Ahn, and T. J. Kang, “Magnetic nanofibers with core (Fe3 O4 nanoparticle suspension)/sheath (poly ethylene terephthalate) structure fabricated by coaxial electrospinning,” Journal of Magnetism and Magnetic Materials, vol. In Press, Uncorrected Proof, 2011. [33] L. K. Lagorce and M. G. Allen, “Magnetic and mechanical properties of micromachined strontium ferrite/polyimide composites,” Microelectromechanical Systems, Journal of, vol. 6, no. 4, pp. 307–312, 1997. [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 micronanomachines by two-photon polymerization,” Advanced Materials, vol. 22, no. 29, pp. 3204–3207, 2010. [35] P. Urwyler, O. Häfeli, H. Schift, J. Gobrecht, F. Battiston, and B. Müller, “Disposable polymeric micro-cantilever arrays for sensing,” Procedia Engineering, vol. 5, pp. 347–350, 2010. [36] J. Y. Park and M. G. Allen, “Development of magnetic materials and processing techniques applicable to integrated micromagnetic devices,” Journal of Micromechanics and Microengineering, vol. 8, no. 4, pp. 307–316, 1998. [37] B. M. Dutoit, P. A. Besse, H. Blanchard, L. Guerin, and R. S. Popovic, “High performance micromachined SmCo17 polymer bonded magnets,” Sensors and Actuators A: Physical, vol. 77, no. 3, pp. 178–182, 1999. [38] O. Rojanapornpun and C. Y. Kwok, Fabrication of integrated micromachined polymer magnet, ser. Proceedings of the Society of Photo-Optical Instrumentation Engineers (Spie). Bellingham: Spie-Int Society Optical Engineering, 2001, vol. 4592, pp. 347–354. [39] M. Feldmann and S. Buttgenbach, “Novel microrobots and micromotors using Lorentz force driven linear microactuators based on polymer magnets,” IEEE Transactions on Magnetics, vol. 43, no. 10, pp. 3891–3895, 2007. [40] Y. Yamanishi, Y. C. Lin, and F. Arai, “Magnetically modified PDMS microtools for micro particle manipulation,” in 2007 IEEE/RSJ International Conference on Intelligent Robots and Systems, Vols 1-9. New York: IEEE, 2007, pp. 759–764. [41] F. Pirmoradi, L. Cheng, and M. Chiao, “A magnetic poly(dimethylesiloxane) composite membrane incorporated with uni-

160

Bibliography formly dispersed, coated iron oxide nanoparticles,” Journal of Micromechanics and Microengineering, vol. 20, no. 1, p. 015032, 2010. [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. [43] S. J. Leigh, C. P. Purssell, J. Bowen, D. A. Hutchins, J. A. Covington, and D. R. Billson, “A miniature flow sensor fabricated by microstereolithography employing a magnetite/acrylic nanocomposite resin,” Sensors and Actuators a-Physical, vol. 168, no. 1, pp. 66–71, 2011. [44] K. L. Tsai, M. Ziaei-Moayyed, R. N. Candler, H. Wei, V. Brand, N. Klejwa, S. X. Wang, and R. T. Howe, “Magnetic, mechanical, and optical characterization of a magnetic nanoparticle-embedded polymer for microactuation,” Microelectromechanical Systems, Journal of, vol. 20, no. 1, pp. 65–72, 2011. [45] P. C. Gach, C. E. Sims, and N. L. Allbritton, “Transparent magnetic photoresists for bioanalytical applications,” Biomaterials, vol. 31, no. 33, pp. 8810–8817, 2010. [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. [47] A. Camenzind, W. R. Caseri, and S. E. Pratsinis, “Flame-made nanoparticles for nanocomposites,” Nano Today, vol. 5, no. 1, pp. 48–65, 2010. [48] A. Teleki, M. Suter, P. R. Kidambi, O. Ergeneman, F. Krumeich, B. J. Nelson, and S. E. Pratsinis, “Hermetically coated superparamagnetic Fe2 O3 particles with SiO2 nanofilms,” Chemistry of Materials, vol. 21, no. 10, pp. 2094–2100, 2009. [49] Y. Yamanishi, S. Sakuma, Y. Kihara, and F. Arai, “Fabrication and application of 3-D magnetically driven microtools,” Microelectromechanical Systems, Journal of, vol. 19, no. 2, pp. 350–356, 2010. [50] C. M. Sorensen, Magnetism. New York: John Wiley & Sons, 2001, pp. 169– 221. [51] C. G. B.D. Cullity, Introduction to magnetic materials. 2008.

Wiley-IEEE Press,

[52] D. L. Leslie-Pelecky and R. D. Rieke, “Magnetic properties of nanostructured materials,” Chemistry of materials, vol. 8, no. 8, pp. 1770–1783, 1996.

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 hydrolysis,” 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: Synthesis, protection, functionalization, and application,” Angewandte ChemieInternational 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 nanoparticles,” 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. Materials & 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. 1956.

New York: Dover,

[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 media,” 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 Technology, vol. 69, no. 2, pp. 253–265, 1996.

162

Bibliography [66] M. Madou, Fundamentals of microfabrication. [67] Q. Lin, Properties of Photoresist Polymers. 965–979.

Boca Raton: CRC Press, 1997. Springer New York, 2007, pp.

[68] B. Schoeberle, “Evaluation of viscoelastic materials for MEMS by creep compliance analysis,” Ph.D. dissertation, ETH Zurich, 2008. [69] M. Hasegawa and K. Horie, “Photophysics, photochemistry, and optical properties of polyimides,” Progress in Polymer Science, vol. 26, no. 2, pp. 259–335, 2001. [70] R. Marie, S. Schmid, A. Johansson, L. Ejsing, M. Nordström, D. Häfliger, C. B. V. Christensen, A. Boisen, and M. Dufva, “Immobilisation of DNA to polymerised SU-8 photoresist,” Biosensors and Bioelectronics, vol. 21, no. 7, pp. 1327–1332, 2006. [71] G. Blagoi, S. Keller, A. Johansson, A. Boisen, and M. Dufva, “Functionalization of SU-8 photoresist surfaces with IgG proteins,” Applied Surface Science, vol. 255, no. 5, pp. 2896–2902, 2008. [72] S. Schmid, P. Wägli, and C. Hierold, “Biosensor based on all-polymer resonant microbeams,” in IEEE 22nd International Conference on Micro Electro Mechanical Systems, MEMS’09., 2009, pp. 300–3. [73] A. del Campo and C. Greiner, “SU-8: a photoresist for high-aspect-ratio and 3D submicron lithography,” Journal of Micromechanics and Microengineering, vol. 17, no. 6, pp. R81–R95, 2007. [74] HD-4100 negative tone photodefinable polyimide. Application notes., HD MicroSystems. [75] SU-8 50 permanent epoxy negative photoresist. Application notes., MicroChem Corp. [76] H. Lorenz, M. Despont, N. Fahrni, N. LaBianca, P. Renaud, and P. Vettiger, “SU-8: A low-cost negative resist for MEMS,” Journal of Micromechanics and Microengineering, vol. 7, no. 3, p. 121, 1997. [77] N. Barie, H. Sigrist, and M. Rapp, “Development of immunosensors based on commercially available surface acoustic wave (SAW) devices,” Analysis, vol. 27, no. 7, pp. 622–629, 1999. [78] G. Kotzar, M. Freas, P. Abel, A. Fleischman, S. Roy, C. Zorman, J. M. Moran, and J. Melzak, “Evaluation of MEMS materials of construction for implantable medical devices,” Biomaterials, vol. 23, no. 13, pp. 2737–2750, 2002.

163

Bibliography [79] Z.-Z. Wu, Y. Zhao, and W. S. Kisaalita, “Interfacing SH-SY5Y human neuroblastoma cells with SU-8 microstructures,” Colloids and Surfaces B: Biointerfaces, vol. 52, no. 1, pp. 14–21, 2006. [80] J. Zhan, G. Tian, L. Jiang, Z. Wu, D. Wu, X. Yang, and R. Jin, “Superparamagnetic polyimide/γ-Fe2 O3 nanocomposite films: Preparation and characterization,” Thin Solid Films, vol. 516, no. 18, pp. 6315–6320, 2008. [81] T. Banert and U. A. Peuker, “Preparation of highly filled superparamagnetic PMMA-magnetite nano composites using the solution method,” Journal of Materials Science, vol. 41, no. 10, pp. 3051–3056, 2006. [82] H. Lorenz, M. Despont, N. Fahrni, J. Brugger, P. Vettiger, and P. Renaud, “High-aspect-ratio, ultrathick, negative-tone near-UV photoresist and its applications for MEMS,” Sensors and Actuators A: Physical, vol. 64, no. 1, pp. 33–39, 1998. [83] J. M. Shaw, J. D. Gelorme, N. C. LaBianca, W. E. Conley, and S. J. Holmes, “Negative photoresists for optical lithography,” IBM Journal of Research and Development, vol. 41, no. 1-2, pp. 81–94, 1997. [84] W. H. Teh, U. Durig, U. Drechsler, C. G. Smith, and H. J. Guntherodt, “Effect of low numerical-aperture femtosecond two-photon absorption on (SU-8) resist for ultrahigh-aspect-ratio microstereolithography,” Journal of Applied Physics, vol. 97, no. 5, p. 054907, 2005. [85] S. Jiguet, “Microfabrication d’objets composites fonctionnels en 3D et à haut facteur de forme, par procédés UV-LIGA et microstéréolithographie,” Ph.D. dissertation, EPFL Lausanne, 2004. [86] J. P. Matinlinna, M. Özcan, L. V. J. Lassila, and P. K. Vallittu, “The effect of a 3-methacryloxypropyltrimethoxysilane and vinyltriisopropoxysilane blend and tris(3-trimethoxysilylpropyl)isocyanurate on the shear bond strength of composite resin to titanium metal,” Dental Materials, vol. 20, no. 9, pp. 804–813, 2004. [87] M. Suter, O. Ergeneman, J. Zürcher, S. Schmid, A. Camenzind, B. J. Nelson, and C. Hierold, “Superparamagnetic photocurable nanocomposite for the fabrication of microcantilevers,” Journal of Micromechanics and Microengineering, vol. 21, no. 2, p. 025023, 2011. [88] H. Gottfried, C. Janzen, M. Pridoehl, P. Roth, B. Trageser, and G. Zimmermann, “Superparamagnetic oxidic particles, processes for their production and their use,” Patent 10/219 267, 2003.

164

Bibliography [89] R. Grass, E. Athanassiou, and W. Stark, “Covalently Functionalized Cobalt Nanoparticles as a Platform for Magnetic Separations in Organic Synthesis,” Angewandte Chemie International Edition, vol. 46, no. 26, pp. 4909– 4912, 2007. [90] R. Massart, “Preparation of aqueous magnetic liquids in alkaline and acidic media,” in IEEE Transactions on Magnetics, vol. 17, no. 2, 1981, pp. 1247– 1248. [91] R. Rosensweig, Ferrohydrodynamics.

Courier Dover Publications, 1997.

[92] T. J. Daou, S. Begin-Colin, J. M. Grenèche, F. Thomas, A. Derory, P. Bernhardt, P. Legaré, and G. Pourroy, “Phosphate adsorption properties of magnetite-based nanoparticles,” Chemistry of Materials, vol. 19, no. 18, pp. 4494–4505, 2007. [93] M. Suter, O. Ergeneman, J. Zürcher, C. Moitzi, S. Pané, T. Rudin, S. E. Pratsinis, B. J. Nelson, and C. Hierold, “A photopatternable superparamagnetic nanocomposite: Material characterization and fabrication of microstructures,” Sensors and Actuators B: Chemical, vol. 156, no. 1, pp. 433–443, 2011. [94] M. Suter, S. Graf, O. Ergeneman, S. Schmid, A. Camenzind, B. J. Nelson, and C. Hierold, “Superparamagnetic photosensitive polymer nanocomposite for microactuators,” in 15th International Solid-State Sensors, Actuators and Microsystems Conference (TRANSDUCERS), 2009, pp. 869–72. [95] S. Jiguet, A. Bertsch, H. Hofmann, and P. Renaud, “SU8-silver photosensitive nanocomposite,” Advanced Engineering Materials, vol. 6, no. 9, pp. 719– 724, 2004. [96] H. Schulz, B. Schimmoeller, S. E. Pratsinis, U. Salz, and T. Bock, “Radiopaque dental adhesives: Dispersion of flame-made Ta2 O5 /SiO2 nanoparticles in methacrylic matrices,” Journal of Dentistry, vol. 36, no. 8, pp. 579–587, 2008. [97] A. Camenzind, T. Schweizer, M. Sztucki, and S. E. Pratsinis, “Structure & strength of silica-PDMS nanocomposites,” Polymer, vol. 51, no. 8, pp. 1796– 1804, 2010. [98] G. Genolet, “New photoplastic fabrication techniques and devices based on high aspect ratio photoresist,” Ph.D. dissertation, EPFL Lausanne, 2001.

165

Bibliography [99] F. Umbrecht, “Wireless implantable passive strain sensor for strain monitoring of orthopedic implants,” Ph.D. dissertation, ETH Zurich, 2010. [100] F. Umbrecht, D. Müller, F. Gattiker, C. M. Boutry, J. Neuenschwander, U. Sennhauser, and C. Hierold, “Solvent assisted bonding of polymethylmethacrylate: Characterization using the response surface methodology,” Sensors and Actuators A: Physical, vol. 156, no. 1, pp. 121–128, 2009. [101] H. C. Chiamori, J. W. Brown, E. V. Adhiprakasha, E. T. Hantsoo, J. B. Straalsund, N. A. Melosh, and B. L. Pruitt, “Suspension of nanoparticles in SU-8: Processing and characterization of nanocomposite polymers,” Microelectronics Journal, vol. 39, no. 2, pp. 228–236, 2008. [102] O. Glatter, “A new method for the evaluation of small-angle scattering data,” J. Appl. Crystallogr., vol. 10, pp. 415–421, 1977. [103] S. P. Jang and S. U. S. Choi, “Role of Brownian motion in the enhanced thermal conductivity of nanofluids,” Applied Physics Letters, vol. 84, pp. 4316–4318, 2004. [104] R. W. Cheary and A. A. Coelho, “An experimental investigation of the effects of axial divergence on diffraction line profiles,” Powder Diffraction, vol. 13, no. 2, pp. 100–106, 1998. [105] H. Borchert, E. V. Shevchenko, A. Robert, I. Mekis, A. Kornowski, G. Grübel, and H. Weller, “Determination of nanocrystal sizes: A comparison of TEM, SAXS, and XRD studies of highly monodisperse CoPt3 particles,” Langmuir, vol. 21, no. 5, pp. 1931–1936, 2005. [106] N. E. Antolino, G. Hayes, R. Kirkpatrick, C. L. Muhlstein, M. I. Frecker, E. M. Mockensturm, and J. H. Adair, “Lost mold rapid infiltration forming of mesoscale ceramics: Part 1, Fabrication,” Journal of the American Ceramic Society, vol. 92, pp. S70–S78, 2009. [107] D. M. Spori, T. Drobek, S. Zürcher, J.rcher, M. Ochsner, C. Sprecher, A. Mühlebach, and N. D. Spencer, “Beyond the lotus effect: roughness influences on wetting over a wide surface-energy range,” Langmuir, vol. 24, no. 10, pp. 5411–5417, 2008. [108] G. F. Goya, T. S. Berquo, F. C. Fonseca, and M. P. Morales, “Static and dynamic magnetic properties of spherical magnetite nanoparticles,” Journal of Applied Physics, vol. 94, pp. 3520–3528, 2003.

166

Bibliography [109] P. Dutta, A. Manivannan, M. S. Seehra, N. Shah, and G. P. Huffman, “Magnetic properties of nearly defect-free maghemite nanocrystals,” Physical Review B, vol. 70, no. 17, p. 174428, 2004. [110] W. Feitknecht and K. J. Gallagher, “Mechanisms for the Oxidation of Fe3O4,” Nature, vol. 228, no. 5271, pp. 548–549, 1970. [111] O. Ergeneman, M. Suter, G. Chatzipirpiridis, J. Zürcher, S. Graf, S. Pané, C. Hierold, and B. J. Nelson, “Characterization and actuation of a magnetic photosensitive polymer cantilever,” in Int. Symposium on Optomechatronic Technologies. ISOT ’09, 2009, pp. 266–70. [112] J. J. Abbott, O. Ergeneman, M. P. Kummer, A. M. Hirt, and B. J. Nelson, “Modeling magnetic torque and force for controlled manipulation of softmagnetic bodies,” IEEE Transactions on Robotics, vol. 23, pp. 1247–1252, 2007. [113] Y. Taechung and K. Chang-Jin, “Measurement of mechanical properties for MEMS materials,” Measurement Science and Technology, vol. 10, no. 8, p. 706, 1999. [114] M. Hopcroft, T. Kramer, G. Kim, K. Takashima, Y. Higo, D. Moore, and J. Brugger, “Micromechanical testing of SU-8 cantilevers,” Fatigue & Fracture of Engineering Materials & Structures, vol. 28, no. 8, pp. 735–742, 2005. [115] U. Lang, N. Naujoks, and J. Dual, “Mechanical characterization of PEDOT:PSS thin films,” Synthetic Metals, vol. 159, no. 5-6, pp. 473–479, 2009. [116] C. Serre, P. Gorostiza, A. Pérez-Murano, F.rez-Rodriguez, F. Sanz, and J. R. Morante, “Measurement of micromechanical properties of polysilicon microstructures with an atomic force microscope,” Sensors and Actuators A: Physical, vol. 67, no. 1-3, pp. 215–219, 1998. [117] C. Serre, A. Pérez-Murano, F.rez-Rodriguez, J. R. Morante, P. Gorostiza, and J. Esteve, “Determination of micromechanical properties of thin films by beam bending measurements with an atomic force microscope,” Sensors and Actuators A: Physical, vol. 74, no. 1-3, pp. 134–138, 1999. [118] J. E. Sader, I. Larson, P. Mulvaney, and L. R. White, “Method for the calibration of atomic force microscope cantilevers,” Review of Scientific Instruments, vol. 66, no. 7, pp. 3789–3798, 1995. [119] B. Ilic, D. Czaplewski, H. G. Craighead, P. Neuzil, C. Campagnolo, and C. Batt, “Mechanical resonant immunospecific biological detector,” Applied Physics Letters, vol. 77, no. 3, pp. 450–452, 2000.

167

Bibliography [120] J. R. Taylor, An Introduction to Error Analysis: The study of uncertainties in physical measurements, 2nd ed. Sausalito CA: University Science Books, 1997. [121] I. Roch, P. Bidaud, D. Collard, and L. Buchaillot, “Fabrication and characterization of an SU-8 gripper actuated by a shape memory alloy thin film,” Journal of Micromechanics and Microengineering, vol. 13, no. 2, p. 330, 2003. [122] D. Osmont, “Computation of the dynamic response of structures with unilateral constraints (contact)–comparison with experimental results,” Computer Methods in Applied Mechanics and Engineering, vol. 34, no. 1-3, pp. 847– 859, 1982. [123] F. Pourroy and P. Trompette, “Vibration of beams on visco-elastic foundations using an explicit time integration scheme,” Computers & Structures, vol. 48, no. 5, pp. 757–762, 1993. [124] S. Schmid and C. Hierold, “Damping mechanisms of single-clamped and prestressed double-clamped resonant polymer microbeams,” Journal of Applied Physics, vol. 104, no. 9, p. 093516, 2008. [125] R. S. Lakes, Viscoelastic solids.

CRC Press, 1999.

[126] R. Feng and R. J. Farris, “Influence of processing conditions on the thermal and mechanical properties of SU8 negative photoresist coatings,” Journal of Micromechanics and Microengineering, vol. 13, no. 1, p. 80, 2003. [127] S. Yang and H. Liu, “A novel approach to hollow superparamagnetic magnetite/polystyrene nanocomposite microspheres via interfacial polymerization,” Journal of Materials Chemistry, vol. 16, no. 46, pp. 4480–4487, 2006. [128] Y. H. Hu, C. Y. Chen, and C. C. Wang, “Viscoelastic properties and thermal degradation kinetics of silica/PMMA nanocomposites,” Polymer Degradation and Stability, vol. 84, no. 3, pp. 545–553, 2004. [129] C. Luo, T. W. Schneider, R. C. White, J. Currie, and M. Paranjape, “A simple deflection-testing method to determine Poisson’s ratio for MEMS applications,” Journal of Micromechanics and Microengineering, vol. 13, no. 1, p. 129, 2003. [130] D. Chicot, F. Roudet, A. Zaoui, G. Louis, and V. Lepingle, “Influence of visco-elasto-plastic properties of magnetite on the elastic modulus: Multicyclic indentation and theoretical studies,” Materials Chemistry and Physics, vol. 119, no. 1-2, pp. 75–81, 2010.

168

Bibliography [131] S. Jiguet, M. Judelewicz, S. Mischler, H. Hofmann, A. Bertsch, and P. Renaud, “SU-8 nanocomposite coatings with improved tribological performance for MEMS,” Surface and Coatings Technology, vol. 201, no. 6, pp. 2289–2295, 2006. [132] A. Jordan, R. Scholz, P. Wust, H. Schirra, S. Thomas, H. Schmidt, and R. Felix, “Endocytosis of dextran and silan-coated magnetite nanoparticles and the effect of intracellular hyperthermia on human mammary carcinoma cells in vitro,” Journal of Magnetism and Magnetic Materials, vol. 194, no. 1-3, pp. 185–196, 1999. [133] R. Rosensweig, “Heating magnetic fluid with alternating magnetic field,” Journal of Magnetism and Magnetic Materials, vol. 252, pp. pp. 370–374(5), 2002. [134] V. S. Kalambur, B. Han, B. E. Hammer, T. W. Shield, and J. C. Bischof, “In vitro characterization of movement, heating and visualization of magnetic nanoparticles for biomedical applications,” Nanotechnology, vol. 16, no. 8, pp. 1221–1233, 2005. [135] N. S. Satarkar, S. A. Meenach, K. W. Anderson, and J. Z. Hilt, “Remote actuation of hydrogel nanocomposites: Heating analysis, modeling, and simulations,” AIChE Journal, vol. 57, no. 4, pp. 852–860, 2011. [136] D. M. Spori, T. Drobek, S. Zürcher, and N. D. Spencer, “Cassie-state wetting investigated by means of a hole-to-pillar density gradient,” Langmuir, vol. 26, no. 12, pp. 9465–9473, 2010. [137] S. Ramakrishna, J. Mayer, E. Wintermantel, and K. W. Leong, “Biomedical applications of polymer-composite materials: a review,” Composites Science and Technology, vol. 61, no. 9, pp. 1189–1224, 2001. [138] M. Hennemeyer, F. Walther, S. Kerstan, K. Schürzinger, A. M. Gigler, and R. W. Stark, “Cell proliferation assays on plasma activated SU-8,” Microelectronic Engineering, vol. 85, no. 5-6, pp. 1298–1301, 2008. [139] D. F. Williams, “On the mechanisms of biocompatibility,” Biomaterials, vol. 29, pp. 2941 – 2953, 2008. [140] R. Feng and R. J. Farris, “The characterization of thermal and elastic constants for an epoxy photoresist SU8 coating,” Journal of Materials Science, vol. 37, no. 22, pp. 4793–4799, 2002.

169

Bibliography [141] H. P. Lang, M. Hegner, and C. Gerber, Nanomechanical Cantilever Array Sensors, 3rd ed. Berlin Heidelberg: Springer-Verlag, 2010, pp. 427 – 452. [142] F. M. Battiston, J. P. Ramseyer, H. P. Lang, M. K. Baller, C. Gerber, J. K. Gimzewski, E. Meyer, and H. J. Güntherodt, “A chemical sensor based on a microfabricated cantilever array with simultaneous resonance-frequency and bending readout,” Sensors and Actuators B: Chemical, vol. 77, no. 1-2, pp. 122–131, 2001. [143] A. Johansson, G. Blagoi, and A. Boisen, “Polymeric cantilever-based biosensors with integrated readout,” Applied Physics Letters, vol. 89, no. 17, p. 173505, 2006. [144] M. Calleja, M. Nordstrom, M. Alvarez, J. Tamayo, L. M. Lechuga, and A. Boisen, “Highly sensitive polymer-based cantilever-sensors for DNA detection,” 6th Int. Conf. on Scanning Probe Microscopy, Sensors and Nanostructures, pp. 215–222, 2004. [145] J. Tamayo, D. Ramos, J. Mertens, and M. Calleja, “Effect of the adsorbate stiffness on the resonance response of microcantilever sensors,” Applied Physics Letters, vol. 89, no. 22, p. 224104, 2006. [146] N. V. Lavrik, M. J. Sepaniak, and P. G. Datskos, “Cantilever transducers as a platform for chemical and biological sensors,” Review of Scientific Instruments, vol. 75, no. 7, pp. 2229–2253, 2004. [147] S. Schmid, P. Senn, and C. Hierold, “Electrostatically actuated nonconductive polymer microresonators in gaseous and aqueous environment,” Sensors and Actuators a-Physical, vol. 145, pp. 442–448, 2008. [148] A. Mehta, S. Cherian, D. Hedden, and T. Thundat, “Manipulation and controlled amplification of Brownian motion of microcantilever sensors,” Applied Physics Letters, vol. 78, no. 11, pp. 1637–1639, 2001. [149] Y. Li, C. Vancura, C. Hagleitner, J. Lichtenberg, O. Brand, and H. Baltes, “Very high Q-factor in water achieved by monolithic, resonant cantilever sensor with fully integrated feedback,” in Sensors, 2003. Proceedings of IEEE, vol. 2, 2003, pp. 809–813 Vol.2. [150] J. E. Sader, “Frequency response of cantilever beams immersed in viscous fluids with applications to the atomic force microscope,” Journal of Applied Physics, vol. 84, pp. 64–76, 1998.

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. [152] O. Ergeneman, “Wireless actuation and readout of magnetic biomicrosystems,” Ph.D. dissertation, ETH Zurich, 2011. [153] S. Truax, K. S. Demirci, A. Hierlemann, and O. Brand, “Exploring the resolution of different disk-type chemical sensors,” in 15th International SolidState Sensors, Actuators and Microsystems Conference (TRANSDUCERS), 2009, pp. 1838–1841. [154] S. B. Truax, K. S. Demirci, L. A. Beardslee, Y. Luzinova, A. Hierlemann, B. Mizaikoff, and O. Brand, “Mass-sensitive detection of gas-phase volatile organics using disk microresonators,” Analytical Chemistry, vol. 83, no. 9, pp. 3305–3311, 2011. [155] X. Zhang and W. C. Tang, “Viscous air damping in laterally driven microresonators,” in Micro Electro Mechanical Systems, MEMS ’94, Proceedings, IEEE Workshop, 1994, pp. 199–204. [156] O. Ergeneman, P. Eberle, M. Suter, G. Chatzipirpiridis, K. M. Sivaraman, S. Pané, C. Hierold, and B. J. Nelson, “An in-plane cobalt-nickel microresonator sensor with magnetic actuation and readout,” in 16th International Solid-State Sensors, Actuators and Microsystems Conference (TRANSDUCERS), 2011, pp. 1068–1071. [157] B. J. Nelson, I. K. Kaliakatsos, and J. J. Abbott, “Microrobots for minimally invasive medicine,” Annual Review of Biomedical Engineering, vol. 12, no. 1, pp. 55–85, 2010. [158] E. M. Purcell, “Life at low Reynolds number,” American Journal of Physics, vol. 45, pp. 3–11, 1977. [159] Y.-L. Zhang, Q.-D. Chen, H. Xia, and H.-B. Sun, “Designable 3D nanofabrication by femtosecond laser direct writing,” Nano Today, vol. 5, no. 5, pp. 435–448, 2010. [160] S. H. Park, D. Y. Yang, and K. S. Lee, “Two-photon stereolithography for realizing ultraprecise three-dimensional nano/microdevices,” Laser & Photonics Reviews, vol. 3, no. 1-2, pp. 1–11, 2009.

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 microfabrications,” Polymers for Advanced Technologies, vol. 17, no. 2, pp. 72–82, 2006. [162] B. Jia, J. Li, and M. Gu, “Two-photon polymerization for three-dimensional photonic devices in polymers and nanocomposites,” Australian Journal of Chemistry, vol. 60, no. 7, pp. 484–495, 2007. [163] M. Thiel, “Design, fabrication and characterization of three-dimensional chiral photonic crystals,” Ph.D. dissertation, KIT Karlsruhe, 2010. [164] J. P. Desai, A. Pillarisetti, and A. D. Brooks, “Engineering approaches to biomanipulation,” Annual Review of Biomedical Engineering, vol. 9, no. 1, pp. 35–53, 2007. [165] S. Taccola, A. Desii, V. Pensabene, T. Fujie, A. Saito, S. Takeoka, P. Dario, A. Menciassi, and V. Mattoli, “Free-standing poly(l-lactic acid) nanofilms loaded with superparamagnetic nanoparticles,” Langmuir, vol. 27, no. 9, pp. 5589–5595, 2011. [166] C. Ingrosso, C. Martin-Olmos, A. Llobera, C. Innocenti, C. Sangregorio, M. Striccoli, A. Agostiano, A. Voigt, G. Gruetzner, J. Brugger, F. PerezMurano, and M. L. Curri, “Oxide nanocrystal based nanocomposites for fabricating photoplastic AFM probes,” Nanoscale, 2011. [167] M. Hagiwara, T. Kawahara, Y. Yamanishi, T. Masuda, L. Feng, and F. Arai, “On-chip magnetically actuated robot with ultrasonic vibration for single cell manipulations,” Lab on a Chip, vol. 11, no. 12, pp. 2049–2054, 2011.

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. A2 O. Ergeneman, P. Eberle, M. Suter, G. Chatzipirpiridis, K. M. Sivaraman, S. Pane, C. Hierold, and B. J. Nelson, „An in-plane cobalt-nickel microresonator sensor with magnetic actuation and readout", accepted for publication in Sensor and Actuators A, 2011. A3 M. Suter, O. Ergeneman, J. Zürcher, C. Moitzi, S. Pané, T. Rudin, S. E. Pratsinis, B. J. Nelson, and C. Hierold, „A photopatternable superparamagnetic nanocomposite: Material characterization and fabrication of microstructures", Sensors and Actuators B: Chemical, vol. 156, pp. 433-443, 2011. A4 M. Suter, O. Ergeneman, J. Zürcher, S. Schmid, A. Camenzind, B. J. Nelson, and C. Hierold, „Superparamagnetic photocurable nanocomposite for the fabrication of microcantilevers", Journal of Micromechanics and Microengineering, vol. 21, p. 025023, 2011. A5 A. Teleki, M. Suter, P. R. Kidambi, O. Ergeneman, F. Krumeich, B. J. Nelson, and S. E. Pratsinis, „Hermetically coated superparamagnetic Fe2 O3 particles with SiO2 nanofilms", Chemistry of Materials, vol. 21, pp. 2094-2100, 2009. 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 cantilevers", in 16th International Solid-State Sensors, Actuators and Microsystems Conference (TRANSDUCERS), 2011, pp. 398-401. C2 O. Ergeneman, P. Eberle, M. Suter, G. Chatzipirpiridis, K. M. Sivaraman, S. Pané, C. Hierold, and B. J. Nelson, „An in-plane cobalt-nickel microresonator sensor with magnetic actuation and readout", in 16th International

173

Bibliography Solid-State Sensors, Actuators and Microsystems Conference (TRANSDUCERS), 2011, pp. 1068-1071. C3 M. Suter, S. Graf, O. Ergeneman, S. Schmid, A. Camenzind, B. J. Nelson, and C. Hierold, „Superparamagnetic photosensitive polymer nanocomposite for microactuators", in 15th International Solid-State Sensors, Actuators and Microsystems Conference (TRANSDUCERS), 2009, pp. 869-72. C4 O. Ergeneman, M. Suter, G. Chatzipirpiridis, J. Zürcher, S. Graf, S. Pané, C. Hierold, and B. J. Nelson, „Characterization and actuation of a magnetic photosensitive polymer cantilever", in Int. Symposium on Optomechatronic Technologies. ISOT 2009, 2009, pp. 266-70. C5 T. Kawano, M. Suter, C. Y. Cho, H. Chiamori, and L. Lin, „Single Carbon Nanotube Pirani Gauge By Local Synthesis", in 14th International Solid-State Sensors, Actuators and Microsystems Conference (TRANSDUCERS), 2007, pp. 1015-1018. C6 D. Briand, M. Vincent, M. Suter, G. Schurmann, N. F. de Rooij, D. C. T. Doan, and M. C. Dang, „Direct Integration of Carbon Nanotubes on Micro Gas Sensing Platforms", in 5th IEEE Conference on Sensors, 2006, pp. 671-674. Conference Talks and Invited Talks

T1 "M. Suter, „Magnetic photosensitive nanocomposite for the fabrication of microcantilevers", MRC Graduate Symposium, ETH Zürich, 2010. T2 M. Suter, S. Graf, O. Ergeneman, S. Schmid, A. Camenzind, B. J. Nelson, and C. Hierold, „Superparamagnetic photosensitive polymer nanocomposite for microactuators", in 15th International Conference on Solid-State Sensors, Actuators and Microsystems (TRANSDUCERS), 2009, pp. 869-72. 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 cantilevers", in 16th International Solid-State Sensors, Actuators and Microsystems Conference (TRANSDUCERS), 2011, pp. 398-401. P2 M. Suter, O. Ergeneman, D. Grob, D. Kraus, J. Zürcher, P. Eberle, G. Chatzipirpiridis, B.J. Nelson and C. Hierold, „Magnetic polymer microstructures", Nanoconvention 2011, Baden, 2011.

174

Bibliography P3 M. Suter, O. Ergeneman, D. Grob, D. Kraus, J. Zürcher, P. Eberle, G. Chatzipirpiridis, B.J. Nelson and C. Hierold, „Magnetic polymer microstructures", MNSP Industry Day, ETH Zürich, 2010. P4 O. Ergeneman, M. Suter, P. Eberle, G. Chatzipirpiridis, K. Sivaraman, S. Pané, B.J. Nelson, „Magnetic Cantilevers as Sensors", MRC Graduate Symposium, ETH Zürich, 2010. P5 M. Suter, O. Ergeneman, J. Zürcher, S. Graf, S. Schmid, A. Camenzind, B.J. Nelson and C. Hierold, „Superparamagnetic Photosensitive Polymer Nanocomposite for Microactuators", MRC Graduate Symposium, ETH Zürich, 2009.

175

Curriculum vitae Personal Details

Name Birth Citizenship

Marcel Suter 30 January 1979, Switzerland Switzerland

Education

05/2007 – 12/2011

10/2004 – 02/2007 10/2006 – 02/2007

10/2000 – 12/2003

08/1995 – 08/1999

ETH Zurich, Micro and Nanosystems, Switzerland Dissertation: Photopatternable superparamagnetic nanocomposite for the fabrication of microstructures University of Neuchâtel, Micro and Nanotechnology Graduation with Bachelor and Master Degree University of California at Berkeley (UCB), Group of Prof. Lin (BSAC), USA Visiting Scholar, Master Thesis: Single carbon nanotube pirani gauge by local synthesis Interstate University of Applied Sciences Buchs NTB Dipl. Ingenieur FH in Systemtechnik (Bachelor in Systems Engineering), major: MEMS, Diploma thesis: Anemometer 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-detectionsystem

Languages

German English French Spanish

mother tongue Level C2 (CEFR) Level B2 Level A2

177