Research Collection

Doctoral Thesis

Thin Film Cathodes for Micro Solid Oxide Fuel Cells

Author(s): Beckel, Daniel

Publication Date: 2007

Permanent Link: https://doi.org/10.3929/ethz-a-005415820

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. 17206

Thin Film Cathodes for Micro Solid Oxide Fuel Cells

A dissertation submitted to

ETH Zurich

for the degree of

Doctor of Sciences

presented by

Daniel Beckel

Dipl. Ing. University of Karlsruhe (TH) born 05.05.1977, citizen of Germany

accepted on the recommendation of

Prof. Dr. Ludwig J. Gauckler, examiner Prof. Dr. Paul Muralt, co-examiner Dr. Danick Briand, co-examiner

2007

Acknowledgements

First of all I would like to thank Prof. Ludwig J. Gauckler for giving me the opportunity to conduct my PhD thesis in his institute. I am especially grateful that he dared to explore really new and exciting topics where a successful outcome was not clear beforehand. I also appreciate that he let us PhD students fly around the globe to present our work at many conferences. This was always a good opportunity to get to know other researchers in the field and to exchange valuable ideas.

I would like to thank Prof. Paul Muralt from Swiss Federal Institute of Technology Lausanne, for being co-examiner of this thesis and participating in our OneBat project to build a micro solid oxide fuel cell (µSOFC). I also thank Dr. Danick Briand from University of Neuchâtel, for joint work within the General Olfaction and Sensing Projects on a European Level (GOSPEL), for fabricating excellent high temperature micro-hotplates and for being co-examiner of this thesis. The collaboration of Jérôme Courbat, Dr. Markus Graf and Dr. Anja Bieberle-Hütter within the GOSPEL project is also gratefully acknowledged.

Thanks go to all colleagues who participate in the OneBat project, especially Patrik Müller from University of Applied Science Buchs for his expertise on Foturan® processing. I am indebted to the staff of FIRST, especially Dr. Otte Homan, for providing a well organized cleanroom, where it was a pleasure to work. I want to thank Bernd Schöberle and Thomas Helbling for their support of the differential pressure stability measurements of our membranes. Dr. Karl Vollmers and Dr. Karim Alchalabi are gratefully acknowledged for their support during exploration of alternative processes. I thank Dr. Seunghwan Lee very much for help with the atomic force microscopy.

I am also indebted to all colleagues at our institute for making it a great place to work. Special thanks go to Dr. André Studart for many ideas that helped to understand the spray pyrolysis process, Dr. A. Nicholas Grundy for helpful discussions and defragmentation of manuscripts and of course to all present and former members of the solid oxide fuel cell group. I also thank Dr. Anna Infortuna for help with the pulsed laser deposition and Dr. Ashley Harvey for proof reading of many manuscripts. Especially I want to thank the members of the µSOFC team, who became my friends. In particular I would like to thank my officemate Ulrich Mücke, for introducing me to 2 the voodoo of electrochemical impedance spectroscopy, for never ending knowledge of chemistry and electrochemistry, for endless discussions what is happening during spray pyrolysis, why the µSOFC did not work, why it finally did work and high-end outdoor equipment and for sharing, creating and forcing enthusiasm for jogging, hiking, snowshoeing, airboarding, climbing, and canyoning.

The work of my semester students Alban Dubach, Guillaume Florey, Dorothee Grieshaber, Thomas Gyger and Aurèle Mariaux who contributed many experiments and ideas to this thesis is gratefully acknowledged as is the work of the helping assistants that either worked on the cathode or within the µSOFC project: Alban Dubach, Silvio Graf, Thomas Gyger, Knut Makowski, Lukas Schlagenhauf and Julian Schneider.

Special thanks go to my family and friends outside the institute for their support throughout the thesis, for many enjoyable bike rides and for lots of fun. Finally I would like to thank Moni for all her support, advice, patience and love. Contents

Acknowledgements ...... 1 Summary...... 7 Zusammenfassung ...... 9 1. General Introduction...... 11 1.1 Fuel Cells ...... 11 1.2 Solid Oxide Fuel Cells...... 12 1.2.1 Design and Function of SOFCs ...... 12 1.2.2 Miniaturization of SOFCs...... 14 1.3 Structure of the Thesis...... 14 1.4 References...... 15 2. State-of-the-Art...... 17 2.1 Introduction ...... 17 2.2. Cathodes...... 19 2.2.1 Oxygen Reduction ...... 19 2.2.2 Material Selection ...... 21 2.2.3 Geometrically Well-Defined Cathodes...... 31 2.2.4 Cathode Performance: Thin Films vs. Porous Thick Films...... 33 2.2.5 Porous Thin Film Cathodes ...... 34 2.2.6 New Issues Raised by Thin Films...... 35 2.3 Cells...... 38 2.3.1 Modeling ...... 39 2.3.2 Free-Standing Electrolytes...... 41 2.3.3 Si-Based Cells ...... 43 2.3.4 Ni-Based Cells ...... 44 2.3.5 Glass-Ceramic Based Cells...... 44 2.4 Summary and Conclusion ...... 44 2.5 References...... 46 3 Aim of the Thesis ...... 63

4 Spray Pyrolysis of La0.6Sr0.4Co0.2Fe0.8O3±δ Thin Film Cathodes...... 65 4.1 Introduction ...... 65 4.2 Experimental ...... 67 4.2.1 Film Preparation...... 67 4.2.2 Film Characterization ...... 69 4.3 Results and Discussion ...... 70 4.3.1 Ratio of Deposition Temperature to Solvent Boiling Point...... 70 4.3.2 Salt Concentration ...... 72 4.3.3 Solution Flow Rate...... 73 4.3.4 Air Pressure ...... 74 4.3.5 Model for Film Formation ...... 75 4.4 Summary...... 78 4.5 References...... 79

5 Solid-State Dewetting of La0.6Sr0.4Co0.2Fe0.8O3±δ Thin Films during Annealing ...... 85 5.1 Introduction ...... 85 5.2. Experimental ...... 86 5.2.1 Thin Film Deposition ...... 86 5.2.2 Porosity and Grain Size Characterization...... 89 5.2.3 X-Ray Diffraction...... 89 5.2.4 Electrical Conductivity Measurements...... 89 5.2.5 Differential Thermal Analysis, Thermogravimetry, Mass Spectroscopy ...... 90 5.3 Results and Discussion...... 90 5.3.1 XRD...... 90 5.3.2 DTA / TG / MS ...... 92 5.3.3 Microstructural Evolution during Annealing ...... 93 5.3.4 Porosity Formation due to Solid-State Dewetting...... 103 5.3.5 Influence of the Precursor Salt ...... 104 5.3.6 Electrical Conductivity...... 106 5.4 Summary and Conclusion ...... 108 5.5 References...... 109 6 Electrochemical Performance of LSCF Based Thin Film Cathodes Prepared by Spray Pyrolysis...... 115 6.1 Introduction ...... 115 6.2 Experimental ...... 116 6.2.1 Electrolyte Preparation...... 117 6.2.2 Thin Film Cathode Deposition ...... 117 6.2.3 Annealing ...... 120 6.2.4 Contacting...... 121 6.2.5 Measurements Conditions ...... 122 6.2.6 Microstructure Characterization...... 123 6.2.7 X-Ray Characterization...... 123 6.3 Results and Discussion...... 124 6.3.1 Verification of the ASR Measurements...... 124 6.3.2 Influence of the Annealing Temperature...... 125 6.3.3 Influence of the Cathode Material...... 126 6.3.4 Influence of the Preparation Process...... 129 6.4 Conclusions ...... 131 6.5 References...... 131 7 Topography Mediated Patterning of Inorganic Materials by Spray Pyrolysis ...... 141 7.1 Introduction ...... 141 7.2 Experimental ...... 143 7.2.1 Substrate Structuring ...... 143 7.2.2 Film Deposition ...... 143 7.2.3 Topography Characterization ...... 144 7.3. Results...... 144 7.4. Summary...... 149 7.5 References...... 150 8 Micro-Hotplates – a Platform for Micro-Solid Oxide Fuel Cells ...... 155 8.1 Introduction ...... 155 8.2 Experimental ...... 157 8.2.1 Micro-Hotplate Design and Fabrication ...... 157 8.2.2 Thin Film Deposition ...... 158 8.2.3 Thin Film Characterization...... 159 8.3 Results and Discussion...... 160 8.3.1 Microstructure ...... 160 8.3.2 Mechanical Stability ...... 162 8.3.3 Conductivity ...... 164 8.4 Summary and Conclusion ...... 167 8.5 References...... 168 9 Stability of NiO Membranes on Photostructurable Glass Substrates for Micro Solid Oxide Fuel Cells ...... 171 9.1 Introduction ...... 172 9.2 Experimental ...... 174 9.2.1 Thin Film Deposition ...... 174 9.2.2 Micromachining of the Substrate ...... 174 9.2.3 Thin Film Characterization...... 177 9.3 Results and Discussion...... 179 9.3.1 Differential Pressure Stability...... 179 9.3.2 Thermal Stability ...... 182 9.4 Summary...... 184 9.5 References...... 184 10 Conclusion and Outlook...... 191 10.1 Cathodes...... 191 10.2 Support Structures ...... 194 10.3 References...... 195 11 Appendix...... 197 11.1 Process for Fabrication of µSOFCs Based on Foturan® ...... 197 11.1.1 Preparation of the Substrate...... 197 11.1.2 Thin Film Deposition ...... 199 11.1.3 Membrane Release...... 202 11.1.4 Contacting...... 203 11.2 Polycrystalline Si Etch Stop ...... 204 11.2.1 Process Details ...... 204 11.2.2 Process Evaluation ...... 205 11.3 Microfabrication using a Sacrificial Layer...... 206 11.3.1 Copper as Sacrificial Layer...... 206 11.3.2 Polymer as Sacrificial Layer ...... 211 11.3.3 Conclusion Sacrificial Layer...... 212 11.4 Effect of Thickness on Crack Formation and Wrinkling...... 212 11.5 Unexpected Crystallization of Foturan®...... 213 11.6 References...... 213 Publications...... 215 Curriculum Vitae ...... 221

Summary

Thin film cathodes were fabricated and characterized for use in micro solid oxide fuel cells (µSOFCs). La0.6Sr0.4Co0.2Fe0.8O3±δ (LSCF) was used as base material for the cathodes and spray pyrolysis for fabrication of most thin films. The drying kinetics during film deposition were very critical to obtain crack free films. The ratio of deposition temperature to solvent boiling point proved to be a decisive parameter to determine these kinetics. Proper adjustment led to crack free films, the surface roughness could be adjusted from smooth to random ridge formation by manipulating the amount of precipitates in the droplets impacting on the substrate. More precipitates led to smoother films. The amount of precipitates could be varied by the salt concentration in the spray solution, the air pressure used for atomization of the spray solution and also by the ratio of deposition temperature to solvent boiling point. The random ridges could be aligned to the topography of the substrate, thereby allowing simple microstructuring of ceramic thin films.

The as deposited films are amorphous and crystallize in a subsequent annealing step. During annealing LSCF develops porosity which is desirable for application as a cathode. The higher the annealing temperature and the longer the annealing time, the more porosity is formed. The porosity development is attributed to solid state dewetting; the pores nucleate at defects in the thin films. Spray pyrolysis produces more defects than pulsed laser deposition (PLD) and consequently higher porosity is found in films prepared by spray pyrolysis than in films prepared by PLD under otherwise identical conditions.

The performance of the thin film cathodes was evaluated by measuring the area specific resistance in the temperature range where µSOFCs will be operated (500 – 600°C). The microstructure and the choice of material play a crucial role for the performance of a cathode. The finer the grain size the better the performance of the cathode. Small grain sizes can be achieved by low annealing temperatures. Composite cathodes containing electrolyte and cathode material, or double layer cathodes containing a thin, dense bottom layer and a porous top layer showed improvement in the performance over standard cathode layers. However, the best performance was achieved with the new material composition

Ba0.25La0.25Sr0.5Co0.8Fe0.2O3±δ which only exhibited half the grain size of LSCF at otherwise identical preparation conditions. With an area specific resistance of 8

0.30 Ω·cm2 at 650°C, the performance was more than twice as good as the performance of LSCF.

Compatibility of the LSCF thin film cathodes with a possible platform for µSOFCs was investigated by depositing these films onto high temperature micro-hotplates. The functionality of the thin film cathode was proven by obtaining the same electrical conductivity as for reference samples on bulk substrates. The integrated heater of the micro-hotplate enabled very fast heating and cooling rates which led to the possibility of accelerated aging tests.

Besides the cathode development, part of the work was also dedicated to microstructuring of glass ceramic substrates (Foturan®) for a µSOFC. This is exemplarily shown with the fabrication of free-standing membranes for µSOFC applications on a photostructurable glass substrate and tests of their differential pressure and thermal stability. The differential pressure stability of more than 100 mbar seems to be sufficient for application in a µSOFC, where one side is exposed to fuel from pressurized cartridges and the other side is exposed to air. The thermal stability even exceeds the requirements for application in a µSOFC. The good thermal stability is attributed to a softening of the glass ceramic substrate, which enables stress reduction in the thin film.

Together with theses which focused on development of thin film anodes and electrolyte, this work finally led to successful fabrication of µSOFCs. An open circuit voltage of 1.05 V and power density of 150 mW/cm2 at 550°C was achieved. Zusammenfassung

Im Rahmen dieser Dissertation wurden Dünnfilm Kathoden für Mikro Festoxid

Brennstoffzellen (μSOFCs) hergestellt und charakterisiert. La0.6Sr0.4Co0.2Fe0.8O3±δ (LSCF) war das Basis Material für diese Kathoden und Sprüh Pyrolyse diente als Herstellmethode für die meisten Dünnfilme. Die Trocknungskinetik während der Filmabscheidung war kritisch für die Herstellung rissfreier Filme. Das Verhältnis von Abscheide Temperatur zum Siedepunkt des Lösungsmittels stellte sich als entscheidender Parameter heraus um die Trocknungskinetik zu bestimmen. Die richtige Einstellung dieses Parameters ermöglichte die Herstellung rissfreier Filme. Die Oberflächenrauheit konnte von glatt, bis zum Auftreten von zufällig angeordneten Stegen eingestellt werden, in Abhängigkeit der Menge an Ausfällungen in den auf dem Substrat auftreffenden Tropfen. Eine grosse Menge an Ausfällungen führte zu ebenmässigeren Filmen. Die Menge der Ausfällungen konnte durch die Salzkonzentration der Sprühlösung, den zur Zerstäubung der Sprühlösung eingesetzten Luftdruck und auch durch das Verhältnis von Abscheide Temperatur zum Lösungsmittelsiedepunkt eingestellt werden. Die zufällig angeordneten Stege konnten durch die Topografie des Substrats ausgerichtet werden, was eine einfache Möglichkeit der Mikrostrukturierung von keramischen Dünnfilmen darstellt.

Die abgeschiedenen Filme sind amorph und können in einem anschliessenden Glühprozess kristallisiert werden. LSCF entwickelt während dieses Glühprozesses Porosität, was für die Anwendung als Kathode wünschenswert ist. Je höher die Glühtemperatur oder je länger die Glühzeit ist, desto mehr Porosität entsteht. Die Entwicklung von Porosität hängt mit einer Festkörper Entnetzung zusammen, die Poren entstehen an Fehlstellen der Dünnfilme. Sprüh Pyrolyse produziert mehr Fehlstellen als gepulste Laser Abscheidung (PLD), daher sind Filme die mit Sprüh Pyrolyse hergestellt wurden poröser als solche die mit PLD unter ansonsten gleichen Bedingungen hergestellt wurden.

Die Leistung der Dünnfilm Kathoden wurde über den flächenspezifischen Widerstand im für μSOFCs interessanten Temperaturbereich (500 – 600°C) bestimmt. Die Mikrostruktur und das Kathodenmaterial sind für die Leistung der Kathode entscheidend. Je feiner das Gefüge desto besser die Leistung der Kathode. Feine Gefüge lassen sich durch niedrige Glühtemperaturen realisieren. Verbundkathoden aus Elektrolyt- und Kathodenmaterial, oder Zweilagenkathoden die aus einer dünnen, 10 dichten Unterschicht und einer porösen darüber liegenden Schicht bestehen, zeigten eine Leistungsverbesserung. Die beste Leistung wurde jedoch mit einer Kathode erreicht die aus der neuen Material Kombination Ba0.25La0.25Sr0.5Co0.8Fe0.2O3±δ bestand und bei ansonsten gleicher Herstellung nur halb so grosse Körner wie LSCF besass. Mit einem flächenspezifischen Widerstand von 0.30 Ω·cm2 bei 650°C wurde, verglichen mit LSCF, eine mehr als zweimal bessere Leistung erreicht.

Die Kompatibilität von LSCF Dünnfilmkathoden mit einer möglichen Plattform für μSOFCs wurde untersucht indem diese Dünnfilme auf Hochtemperatur Mikroheizplatten abgeschieden wurden. Die Funktionsfähigkeit der Dünnfilmkathoden wurde durch Messungen der elektrischen Leitfähigkeit gezeigt, welche die gleichen Werte ergab wie für Referenzproben welche auf grossen Substraten abgeschieden wurden. Das integrierte Heizelement der Mikroheizplatten erlaubte sehr schnelle Aufheiz- und Abkühlraten welche beschleunigte Alterungstests ermöglichen.

Neben der Kathodenentwicklung war ein Teil der Arbeit der Mikrostrukturierung von Glaskeramischen Substraten (Foturan®) für μSOFCs gewidmet. Exemplarisch dafür wurden freistehende Membranen für μSOFC Anwendungen auf fotostrukturierbarem Glas hergestellt und auf ihre Differenzdruck- und Temperaturbeständigkeit hin untersucht. Die Differenzdruckbeständigkeit von mehr als 100 mbar scheint ausreichend für die Anwendung in μSOFCs zu sein, wo eine Seite der Membran Brenngas aus Druckbehältern ausgesetzt ist und die andere Seite Luft ausgesetzt ist. Die Temperaturbeständigkeit übersteigt die Anforderungen für μSOFCs sogar. Die gute Temperaturbeständigkeit wird der Erweichung des Glaskeramik Substrats zugeschrieben, die einen Stress Abbau in den Dünnfilmen ermöglicht.

Zusammen mit Doktorarbeiten zur Entwicklung von Dünnfilm Anoden und Elektrolyten, hat diese Arbeit dazugeführt, dass schliesslich funktionierende μSOFCs hergestellt werden konnten. Eine Leerlaufspannung von 1.05 V und eine Leistungsdichte von 150 mW/cm2 bei 550°C wurden erreicht. 1 General Introduction

This chapter explains the basic working principle of solid oxide fuel cells (SOFCs) and the interest in miniaturizing SOFCs. At the end of this chapter the structure of the thesis outlined.

1.1 Fuel Cells

Fuel cells are electrochemical devices that convert chemical energy directly into electricity and heat. Therefore they are more efficient than internal combustion engines, which first convert the chemical energy stored in the fuel into heat and then in movement and electricity [1]. The need for efficient energy conversion led to strong development of fuel cells starting from about 1960 when fuel cells were developed for vehicles where space and/or weight is limited such as spacecraft or submarines [2]. A renewed interest in fuel cells started in the early 1990ies when the interest in efficient energy conversion was stimulated by the wish to save energy, reduce pollution and the emission of greenhouse gases.

Different types of fuel cells have been developed up to now they are basically different in the materials of their electrolytes [3]: polymer based fuel cells known as polymer electrolyte membrane fuel cells (PEM) and direct methanol fuel cells (DMFC). The charge is carried through the polymer electrolyte by H+ in both types of cells. Both offer the advantage of low operating temperatures (80 – 120°C), which simplifies the surrounding technology. However, the low operating temperatures also hinder the commercialization of these fuel cells, since they require precious metal catalysts that raise the costs of such fuel cells. The PEM is also limited by the need for very pure hydrogen as fuel, especially the low tolerance for CO (a few ppm) causes problems since CO is the main by product of hydrogen when hydrogen is produced by steam reforming of natural gas. Storage and efficiency of hydrogen as energy carrier is also questionable. Phosphoric-acid fuel cells (PAFC) involve phosphoric acid constrained in a polymer as electrolyte, the same operating temperature and restrictions for fuel purity apply as for PEM. Other acids such as sulfuric acids have been used as electrolyte as well, allowing the use of a liquid fuel 12 C H A P T E R 1

(formic acid), which simplifies energy storage [4]. Alkaline fuel cells (AFC), which use - KOH as electrolyte and OH as charge carrier, are very efficient but need CO2 free gas, which limits the use to special applications such as spacecraft. Molten carbonate fuel cells (MCFC) work at rather high temperatures around 600°C and 2- involve a liquid Li/Na/KCo3 electrolyte, where Co3 is the charge carrier. The corrosive liquid electrolyte reacts with the electrodes and the container material and therefore complicates fabrication of MCFCs. Solid oxide fuel cells (SOFCs) traditionally operate at high temperatures (800 – 1000°C) which simplifies fuel processing and removes the need for precious metal catalysts. Furthermore, they offer high power densities. The electrolyte is yttria-stabilized zirconia (YSZ) and oxygen ions are the charge carriers. However, the high operating temperature in conjunction with ceramic materials is the challenge one faces when working on SOFCs. Therefore research aims towards lowering of the operating temperature towards 500 – 600°C where fuel processing is still convenient and material degradation is less pronounced. Since lowering of the operating temperature is associated with a loss in performance, thin film components that reduce the ohmic losses are employed to achieve sufficient power at lower operating temperatures.

1.2 Solid Oxide Fuel Cells

1.2.1 Design and Function of SOFCs

As schematically shown in Fig. 1.1, an SOFC consists of three active layers: an air electrode (cathode), an electrolyte and a fuel electrode (anode). In an SOFC, all three layers consist of ceramic materials, except for the anode which is a two phase material consisting of ceramic and metal (cermet).

Air enters at the cathode side, where the oxygen molecules are reduced to oxygen ions:

- 2- O2 + 2·e → ½·O (1.1)

Usually the cathode is porous to allow gas access and is catalytically active towards reduction of oxygen; furthermore it is capable of conducting electrons and preferably also ions. The most common materials are LaxSr1-xMnO3 (LSM) and I N T R O D U C T I O N 13

LaxSr1-xCoyFe1-yO3 (LSCF), more cathode materials will be discussed in chapter 2.2.2.

Fig. 1.1: Schematic cut through a SOFC element.

The oxygen ions cross the gas tight electrolyte, which is a predominately ionic conductor. Usually YSZ or ceria gadolinia oxide (CGO) are used as materials. In the anode compartment, the oxygen ions recombine with the fuel and release the electrons to an external circuit, which conducts the electrons again to the cathode. In the simplest case, the fuel is hydrogen, but hydrocarbons can be used as well. They are then partially reformed to CO and H2, which both maintain a low oxygen partial pressure at the anode due to their low enthalpies of formation of CO2 and H2O. For hydrogen, the reaction at the anode is given by:

2- - ½·O + H2 → H2O + 2·e (1.2)

The metal phase, usually Ni, of the porous cermet anode is catalytically active towards fuel oxidation and conducts electrons whereas the ceramic phase is a predominantly ionic conductor and often identical with the electrolyte material. Several of these cells are normally connected by ceramic or metal interconnectors to form a stack of cells with higher voltage and therefore also higher power output. The open circuit voltage (OCV) i.e. when no current is drawn is typically 1 V for one cell. Cells are usually operated at the working conditions of maximum power output, e.g. at 0.6 to 0.7 V at temperatures between 900 and 1000°C. 14 C H A P T E R 1

1.2.2 Miniaturization of SOFCs

Miniaturization of SOFCs is desirable because portable devices such as mobile phones and laptops have an increasing demand for energy due to the integration of many features into one device, which makes it difficult for batteries to supply enough energy without taking too much volume and weight. Since SOFCs offer more energy per volume and weight than batteries [5] they become attractive for portable applications. Furthermore, SOFCs can be refueled instantly without delay when using propane/butane pressurized liquid fuel cartridges, which are used for lighters all over the world [6]. If SOFCs offer additional advantages compared to batteries, even higher costs and prices than Li ion batteries maybe tolerated. Thus miniaturization of fuel cells stimulates commercialization of fuel cells.

1.3 Structure of the thesis

The state-of-the-art in SOFC cathodes and the research dedicated to micro SOFC (µSOFCs) is reviewed in chapter 2 to give an overview of the field of research that is relevant for this thesis. After the overview, the aim of this thesis is described in chapter 3. Then, in chapter 4, the spray pyrolysis process, which was used for preparation of most of the thin film cathodes is described in detail. A model for thin film formation and the relation between various process parameters and the resulting thin film is given. Chapter 5 describes the thermal stability and changes in microstructure that occur upon annealing of the LSCF thin films at the operating temperature of µSOFCs and above. Chapter 6 reports about the electrochemical performance achieved with these thin film cathodes. Some special applications of the LSCF thin films follow: In chapter 7 a possibility is outlined how microstructuring of ceramic thin films can be realized by spray pyrolysis. In chapter 8 results are reported of micro-hotplates used as a platform for µSOFCs. Chapter 9 reports about the fabrication of free standing membranes on a photostructurable substrate. The membrane consists of the first layer on which the µSOFCs were built later on. Conclusions and outlook are given in chapter 10. In the appendix (chapter 11) the process for successful fabrication of the µSOFC and several process alternatives are summarized.

I N T R O D U C T I O N 15

1.4 References

[1] N. Q. Minh, "Ceramic Fuel Cells", Journal of the American Ceramic Society, 76, [3] 563-88 (1993).

[2] Fuel Cells, www.bpa.gov/energy/n/tech/fuel_cell/pem_fuel_cells.cfm, (2007).

[3] B. C. H. Steele, A. Heinzel, "Materials for Fuel-Cell Technologies", Nature, 414, [6861] 345-52 (2001).

[4] K.-L. Chu, M. A. Shannon, R. I. Masel, "An Improved Miniature Direct Formic Acid Fuel Cell Based on Nanoporous Silicon for Portable Power Generation", Journal of the Electrochemical Society, 153, [8] A1562-A7 (2006).

[5] C. K. Dyer, "Fuel Cells for Portable Applications", Journal of Power Sources, 106, [1-2] 31-4 (2002).

[6] www.bicworld.com/inter_en/lighters/product_history/index.asp#, (2007). 16 C H A P T E R 1

2 State-of-the-Art

This chapter reviews the state-of-the-art of solid oxide fuel cell (SOFC) cathodes and micro SOFCs (µSOFCs). Cathodes are reviewed with regard to the cathode reaction, the material selection and with special focus on thin film cathodes, since they are important for application in µSOFCs. In the last section of this chapter the research towards modeling and fabrication of µSOFCs published so far is reviewed.

2.1 Introduction

Solid oxide fuel cells (SOFCs) convert chemical energy with high efficiency directly into electricity and heat and can operate on a variety of fuels such as natural gas or hydrogen. As depicted in Fig. 2.1, the fuel supplying H2 is fed into the anode compartment where it is oxidized, and the electrons released as a result are conducted to an external circuit [1]. The reaction products on the anode side of an

SOFC are mainly water and CO2. Air enters on the cathode side and oxygen is reduced here to O2- by reaction with electrons from the external circuit. The O2- ions can travel through the ion-conducting and gas-tight electrolyte, which separates the anode compartment from the cathode compartment. Once on the anode side the O2- joins with hydrogen to form water. The driving force for an SOFC is the difference in electrochemical potential between both electrodes. In equilibrium, this is represented by the oxygen partial pressure between the anode (low pO2) and cathode (high pO2). Open circuit voltage (OCV) is the voltage obtained at zero current. According to Nernst the OCV at equilibrium conditions is given by:

This chapter is contained in parts in the review articles „Thin Films for Micro Solid Oxide Fuel Cells“, D. Beckel; A. Bieberle-Hütter; A. Harvey; A. Infortuna; U. P. Muecke; M. Prestat; J. L. M. Rupp, L. J. Gauckler, Journal of Power Sources doi:10.1016/j.jpowsour.2007.04.070 and „Solid Oxide Fuel Cells: Systems and Materials”, L.J. Gauckler, D. Beckel, B.E. Buergler, E. Jud, U. P. Muecke, M. Prestat, J.L.M. Rupp; J. Richter, Chimia 58, [12] 837-50 (2004). 18 C H A P T E R 2

high RT pO OCV = ·ln 2 (2.1). 4F plow O2

R is the universal gas constant (8.31 J/ (mol·K)), T the temperature in K and F the Faraday constant (9.65·104 C/mol). In practice the OCV ranges from about 1.1 to less than 0.8 V due to gas leakage and current by passes. A low OCV reduces the SOFC power output.

Fig. 2.1: Sketch of an SOFC illustrating also the working principle.

The choice of materials for each component is given by the requirements resulting from the functions discussed above: The anode should be porous to allow gas access, it should act as a catalyst for fuel oxidation and should have electronic and ionic conductivity. Nickel satisfies the first two requirements, while yttria-stabilized zircona (YSZ) and, at low temperatures, cerium gadolinium oxide (CGO) fulfill the last. Therefore, two-phase cermets (ceramic-metal composites) that combine all three properties are used. The electrolyte should be dense and predominantly an ionic conductor like YSZ or CGO while the cathode should be porous to enhance gas access, catalytically active towards oxygen reduction, and a good ionic and electronic (mixed) conductor. For cathodes perovskites are commonly chosen, for example

LaxSr1-xMnO3±δ (LSM), LaxSr1-xCoyFe1-yO3±δ (LSCF) or Ba0.5Sr0.5Co0.8Fe0.2O3±δ (BSCF) [2] or precious metals such as Pt [3]. S T A T E – O F – T H E A R T 19

Differing from this two-gas-chamber concept is the single-chamber SOFC [4], where the anode and cathode are exposed to the same gas atmosphere, a mixture of fuel and air in a safe ratio. Here the driving force is the locally different oxygen partial pressure at the electrodes, which is generated by the different selectivity of the anode and cathode towards fuel oxidation [5-7]. For a recent review see [8].

Typical operating temperatures for current thick-film-based two-chamber SOFCs are 800 – 1000°C, placing heavy demands on the materials and complicating the sealing mechanism. Therefore research trends towards lowering the operating temperature down to 500 – 600°C. To compensate for the performance losses associated with a lower operating temperature, thin film components with lower ohmic resistance have been developed. Thin film components facilitate the fabrication of µSOFCs, leading to new applications for SOFCs, namely portable electronic devices such as laptops, personal digital assistants (PDAs) and scanners [9, 10].

This chapter will review cathode materials and studies performed in the field of cathodes as well as giving an overview on the research performed towards modeling and fabrication of µSOFCs.

2.2 Cathodes

First, the oxygen reduction at the cathode side will be examined, then an overview of cathode materials is given and finally cathode studies focusing on thin films are highlighted.

2.2.1 Oxygen Reduction

Oxygen reduction at SOFC cathodes follows a sophisticated reaction path that has been drawing much attention [11]. Today's cathodes are viewed preferably for intermediate and low operating temperatures (< 600°C). They are often made of mixed ionic-electronic conducting perovskites (ABO3) where both the A sites (a rare earth element, such as lanthanum) and the B sites (a transition metal) can be substituted in order to tailor the material properties. Compositions with lanthanum partially substituted by strontium (LaxSr1-xBO3±δ) have been extensively used ([12] and references therein). In those compounds, the Sr2+ cations sit on the La3+ sites because of similar ionic radii. The substitution of La3+ by Sr2+ creates a fully ionized 20 C H A P T E R 2

' • acceptor level (SrLa ) whose charge is compensated by holes ( BB ) in the valence

•• band as well as oxygen vacancies ( VO ). LaxSr1-xBO3±δ perovskites are therefore highly doped p-type semiconductors [13-15]. Typical materials are the well-known LSM, LSCF and BSCF [2]. Oxygen can be transported through the bulk of the perovskite by a hopping mechanism via the oxygen vacancies. The concentrations of the defects in the perovskite structure are dependant on the chemical composition, the temperature, the oxygen partial pressure and the electrochemical potential.

Fig. 2.2: Schematic representation of the possible oxygen reduction pathways for a mixed ionic-electronic cathode. Bulk and surface pathways are parallel, i.e. in competition, after [12].

Fig. 2.2 schematically describes the possible reaction pathways of oxygen reduction. Bulk and surface pathways are parallel, i.e. in competition. The fastest path determines the overall kinetics of the reaction. One pathway may be dominant or both may have similar importance, depending on the rate constants of the reaction steps. The nature of the rate-determining pathway determines the optimal microstructure of the cathode. The mechanism of oxygen reduction involves many reaction steps, such as adsorption; dissociation; gas-phase, surface and bulk diffusion; and charge transfer. When the bulk pathway plays a significant role, the defect chemistry of the electrode has to be taken into account as well. The need to simplify the investigated systems and the attempt to gain a better understanding of oxygen reduction has stimulated the development of thin, dense geometrically well- S T A T E – O F – T H E A R T 21 defined (gwd) films instead of porous electrodes. The complex description of gas- phase diffusion through the electrode pores can be discarded. Most important, key- parameters such as the triple phase boundary (tpb) length and bulk diffusion length (i.e. film thickness) can be controlled when using gwd electrodes.

2.2.2 Material Selection

For proper function as a cathode in a solid oxide fuel cell, the material should have a high electrocatalytic activity towards oxygen reduction and a high chemical stability in an oxidizing environment without forming highly resistive reaction products with the electrolyte and current collector [16, 17]. The material should exhibit similar thermomechanical properties as the electrolyte to avoid stresses to develop upon heating and cooling [18] and it should have high electrical conductivity.

Most reviews on SOFCs deal with state-of-the-art cathode materials such as

LaxSr1-xMnO3±δ (LSM) and LaxSr1-xCoyFe1-yO3±δ (LSCF) [19-31]. A few of these reviews also include emerging materials [22, 24, 29, 32]. The following will be limited to cathode material aspects and exclude most processing related techniques which can be found elsewhere [33]. The performance of cathodes is judged by the area specific resistance (ASR), which is not only influenced by the choice of the material, but also by the microstructure and the processing. To provide an overview of cathode performance Table 2.1 contains the ASRs for many different cathodes. Different performances of the same material are included on purpose, to illustrate the broad range of performance that can be achieved by nominally the same material. 22 C H A P T E R 2

Table 2.1: Area specific resistance (ASR) of different cathode materials at different temperatures, stoichiometry and composition of multiphase cathodes are given if they are detailed in the original publication.

Material ASR/(Ω·cm2) Ref.

400°C 650°C 800°C

La0.72Sr0.18MnO3 + Ce0.9Gd0.1O1.95 + - 0.4 0.08 [34] La0.6Sr0.4Co0.2Fe0.8O3

(La0.75Sr0.25)0.95MnO3 + 50 vol % YSZ - 5.6 0.63 [35]

(La0.75Sr0.25)0.95MnO3 + 50 vol % YSZ - 3.2 0.25 [35]

La0.8Sr0.2MnO3 - 223 1.4 [36]

La0.8Sr0.2MnO3 - 0.3 [37]

La0.8Sr0.2MnO3 - 1100 60 [38]

La0.8Sr0.2MnO3 + YSZ - - 7.6 [39]

(La0.8Sr0.2)0.98MnO3 + 50 vol % YSZ - 0.5 - [40]

La0.8Sr0.2MnO3 + 50 wt % Sm0.2Ce0.8O1.9 - 10 - [41]

La0.8Sr0.2MnO3 + 10 wt % Ce0.7Bi0.3O2 - 8.4 - [42]

La0.8Sr0.2MnO3 + 50 wt % Ce0.7Bi0.3O2 - 3.2 - [42]

La0.82Sr0.18MnO3 + (Y2O3)0.15(CeO2)0.85 + YSZ - - 2.6 [39]

La0.8Sr0.2MnO3 + La0.5Sr0.5CoO3 - - 1.3 [37]

La0.8Sr0.2MnO3 + LaCoO3 - - 0.2 [37]

La0.8Sr0.2MnO3 + LaNi0.6Fe0.4O3 - - 0.4 [37]

La0.85Sr0.15MnO3 + YSZ - - 0.08 [43]

La0.85Sr0.15MnO3 + YSZ - - 0.58 [44] S T A T E – O F – T H E A R T 23

Material ASR/(Ω·cm2) Ref.

400°C 650°C 800°C

La0.85Sr0.15MnO3 + YSZ - - 1.2 [44]

La0.85Sr0.15MnO3 + YSZ - - 2.4 [44]

La0.85Sr0.15MnO3 + YSZ + La0.8Sr0.2CoO3 - 7.5* 1.2 [45]

La0.85Sr0.15MnO3 + YSZ + La0.84Sr0.16CoO3 - 34* 3.7 [46]

La0.85Sr0.15MnO3 + Ce0.9Gd0.1O1.95 + - 3.7* 0.47 [46] La0.84Sr0.16CoO3

La0.85Sr0.15MnO3 + Sm0.2Ce0.8O1.9 - - 0.63 [43]

LSM - - 22 [47]

LSM + Ce0.9Gd0.1O1.95 - 0.6 0.04 [47]

La0.7Sr0.25FeO3 - 5.5 0.13 [38]

La0.8Sr0.2FeO3 - 120 9 [38]

La0.6Ca0.4MnO3 - 10 0.48 [36]

Pr0.8Sr0.2FeO3 - 65 3.5 [38]

La0.2Sr0.8Co0.8Fe0.2O3 - 14 1.6 [48]

La0.6Sr0.4Co0.2Fe0.8O3 - 0.41 0.027* [49]

La0.6Sr0.4Co0.2Fe0.8O3 - 1.2 - [50]

La0.6Sr0.4Co0.2Fe0.8O3 - 6.5 - [51]

La0.6Sr0.4Co0.2Fe0.8O3 - 1.9 0.52 [48]

La0.6Sr0.4Co0.2Fe0.8O3 - 0.65 0.05 [47] 24 C H A P T E R 2

Material ASR/(Ω·cm2) Ref.

400°C 650°C 800°C

La0.6Sr0.4Co0.2Fe0.8O3 - 0.8 - [38]

La0.6Sr0.4Co0.2Fe0.8O3 - 2.9 - [52]

La0.6Sr0.4Co0.2Fe0.8O3 76 0.30 0.03* [53]

La0.6Sr0.4Co0.2Fe0.8O3 17 0.09 0.01* [53]

La0.6Sr0.4Co0.2Fe0.8O3 - 0.67 - [54]

La0.6Sr0.4Co0.2Fe0.8O3 104* 0.43 - [54]

La0.6Sr0.4Co0.2Fe0.8O3 + 25 vol % Ce0.9Gd0.1O1.95 - 0.95 - [50]

La0.6Sr0.4Co0.2Fe0.8O3 + 30 wt % Ce0.8Gd0.2O1.9 - 0.14 0.009 [55]

La0.6Sr0.4Co0.2Fe0.8O3 + 30 wt % Ce0.8Gd0.2O1.9 - 0.6 0.05 [55]

La0.6Sr0.4Co0.2Fe0.8O3 + 35 vol % Ce0.9Gd0.1O1.95 - 0.35 - [50]

La0.6Sr0.4Co0.2Fe0.8O3 + 40 vol % Ce0.9Gd0.1O1.95 - 51 - [50]

La0.6Sr0.4Co0.2Fe0.8O3 + 50 wt % Ce0.9Gd0.1O1.95 - - 0.01 [56]

La0.6Sr0.4Co0.2Fe0.8O3 + 35 vol % Ce0.7Bi0.3O2 - 1.05 - [51]

La0.6Sr0.4Co0.2Fe0.8O3 + Pd - 0.11 0.007* [49]

(La0.6Sr0.4)0.99Co0.2Fe0.8O3 - - 1.9 [57]

La0.8Sr0.2Co0.1Fe0.9O3 - 66 4.8 [48]

La0.8Sr0.2Co0.2Fe0.8O3 - 11 0.85 [38]

La0.8Sr0.2Co0.2Fe0.8O3 - 6 - [38]

La0.6Sr0.4Co0.6Ga0.4O3 - - 0.12 [58] S T A T E – O F – T H E A R T 25

Material ASR/(Ω·cm2) Ref.

400°C 650°C 800°C

La0.4Sr0.6Ni0.2Fe0.8O3 - 1.4 0.12 [59]

La0.4Sr0.6Ni0.2Fe0.8O3 + 45 vol % Sm0.2Ce0.8O1.9 - 0.4 0.06 [59]

La0.6Ba0.4Co0.2Fe0.8O3 - 0.18 0.025 [60]

Ba0.5Sr0.5Co0.8Fe0.2O3 - 0.07 0.015 [60]

Ba0.5Sr0.5Co0.8Fe0.2O3 8 0.03 - [2]

Sm0.5Sr0.5Co0.8Fe0.2O3 - 1.6 0.89 [61]

(Sm0.6Sr0.4)0.99Fe0.8Co0.2O3 - - 0.1 [57]

(Pr0.6Sr0.4)0.99Co0.2Fe0.8O3 - - 0.23 [57]

(Gd0.6Sr0.4)0.99Fe0.8Co0.2O3 - - 0.04 [57]

Ba0.25La0.25Sr0.5Co0.8Fe0.2O3 122* 0.30 - [54]

La0.6Sr0.4CoO3 - 1.9 0.52 [48]

La0.6Sr0.4CoO3 - 1.2 0.23 [62]

La0.6Sr0.4CoO3 + 45 wt % La0.45Ce0.55O2 - 0.61 0.16 [62]

La0.8Sr0.2CoO3 - 1.5 - [38]

La0.8Sr0.2CoO3 + Ce0.8Gd0.2O1.9 + YSZ - - 0.044 [39]

La0.8Sr0.2CoO3 + (Y2O3)0.15(CeO2)0.85 + YSZ - - 0.038 [39]

+ + La0.84Sr0.16CoO3 - 0.49 0.07 [63]

La0.75Sr0.25CuO2.5 - 1.95 0.65 [64]

Sm0.5Sr0.5CoO3 - 0.7 0.10 [61] 26 C H A P T E R 2

Material ASR/(Ω·cm2) Ref.

400°C 650°C 800°C

Sm0.5Sr0.5CoO3 + 30 wt % Sm0.2Ce0.9O1.9 - 0.018* - [65]

Sm0.5Sr0.5CoO3 + 10 wt % Sm0.2Ce0.9O1.9 - 0.05* - [65]

Pr0.5Sr0.5CoO3 - 0.12 - [38]

Pr0.8Sr0.2CoO3 - 0.4 - [38]

Pr0.8Sr0.2CoO3 - 10 0.8 [38]

Gd0.5Sr0.5CoO3 - 0.2 - [38]

Gd0.8Sr0.2CoO3 - 0.55 - [38]

SrFeCo0.5O3.25 - 0.17 - [38]

SrCoFe0.5O3.25 - 0.1 - [38]

Sr0.9Ce0.1FeO3 - 13 0.95 [38]

Sr0.9Ce0.1CoO3 - 4.5 0.65 [38]

Sr0.9Ce0.1CoO3 - 0.3 - [38]

Sr0.8Ce0.1Fe0.7Co0.3O3 - 2.3 0.14* [66]

Nd2NiO4 - 3.2 0.23 [36]

Y0.25Bi0.75O1.5 + 50 wt % Ag - 2 - [41]

Pt 30# - - [3]

Pt - 28 4.2 [67]

Pt - 32 1 [48]

Pt - 4.8+ 0.5+ [63] S T A T E – O F – T H E A R T 27

Material ASR/(Ω·cm2) Ref.

400°C 650°C 800°C

Pt + YSZ 333 - - [68]

Au - 50+ 11+ [63]

Ag - 1.8+ 0.33+ [63]

* Values extrapolated for this temperature.

+ Values extracted from overpotential plots starting around 0.1 A/cm2 possibly leading to underestimation of the resistance.

# Value extracted from cell impedance spectrum, the authors stated that the electrode losses are mainly due to the cathode.

2.2.2.1 LaxSr1-xMnO3±δ (LSM) and LaxSr1-xCoyFe1-yO3±δ (LSCF) Cathodes

The choice of cathode materials is rather limited: Noble metals such as Pt and Ag are suitable, but exhibit prohibitive costs for SOFC application at higher temperatures due to high Pt suboxide vapor pressure at 900 to 1000°C. LaxSr1-xMnO3±δ (LSM), as the state-of-the-art electronic conducting material, is widely used since it fulfills most of the requirements listed above. Its properties are given in Table 2.2 with the data taken from references [18, 69-71]. Usually LSM is used for the cathode when YSZ is used as the electrolyte, because the thermal expansion coefficients match well [45]. However, the rather high operating temperatures of the SOFC around 900 to 1000°C promote degradation of the cathode and the formation of undesired resistive reaction products, such as La2Zr2O7, especially during manufacturing of LSM on YSZ [17, 38, 72-74]. 28 C H A P T E R 2

Table 2.2: Coefficient of thermal expansion (TEC) (30 – 1000°C), electronic (σe) and ionic (σi) conductivities and bulk diffusion D and surface exchange coefficient k at 800°C of

La0.65Sr0.35MnO3-δ (LSM) and La0.6Sr0.4Co0.2Fe0.8O3 (LSCF).

σe/ σi/ D/ k/ Material TEC/10-6 K-1 (S/cm) (S/cm) (cm2/s) (cm/s)

12.3 [18] 4·10-14 [69] 5·10-8 [69] LSM 102 [18] 1.7·10-4 [18] (YSZ: 11.0·[69]) (900°C) (900°C)

17.5 [18] LSCF 302 [18] 8·10-3 [18] 2.5·10-8 [71] 5.6·10-6 [71] (CGO: 10.5 [70])

Increased triple phase boundary length, better adhesion to the electrolyte and lower thermal expansion mismatch is achieved when using a LSM-YSZ composite material [40, 73, 75] or even composites with graded compositions [45]. Besides YSZ, CGO

[46], Sm0.2Ce0.8O2 (SDC) [43] and Ce0.7Bi0.3O2 [42] are also used for fabrication of composite cathodes with LSM with improved performance.

As for most perovskite materials, the properties of LSM can be tailored by partially substituting the A and B sites of the ABO3 perovskite. The thermal expansion coefficient (TEC) can be further adjusted to that of the YSZ electrolyte by using

(La1-xYx)0.7Sr0.3MnO3 [76] or Sr1-xCexMnO3-δ [77]. Compositions which are compatible with CGO as regards TEC and chemical stability are Gd1-xSrxMnO3, Nd1-xSrxMnO3-δ

[78] and Pr1-xSrxMnO3 [79]. The formation of reaction products between the YSZ electrolyte and the cathode can be suppressed for Ln1-xSrxMnO3 (Ln = Pr, Nd) [80] and Pr1-xCaxMnO3 [81], whereas for La1-xCaxMnO3 on a CaO-stabilized ZrO2 electrolyte no stable composition was found [82]. The conductivity can be increased by using Pr0.6-xSr0.4MnO3 [83, 84], but for substitution of Mn with Co in

Y0.6Sr0.4Mn1-yCoyO3 (0 ≤ y ≤ 0.4) mixtures, increasing y resulted in lower conductivity [85], the same is observed for adding Al to LSM [86].

The La1-xSrxCoO3-δ (LSC) based cathodes [87-89] are typical mixed conductors offering the advantage of higher electronic and, more important, higher ionic conductivity (see Table 2.2). By providing this second pathway for oxygen ions, S T A T E – O F – T H E A R T 29 activity of the cathode is increased and lower operating temperatures are feasible. The disadvantage is that those materials react with YSZ [38, 89], thus either ceria based electrolytes or protective layers of ceria [89] or LSGM [88, 90, 91] on YSZ electrolytes should be used. In order to adjust the TEC of LSC based cathodes to the one of CGO, Fe was introduced to obtain lower TEC [92]. Depending on the composition, the conductivities of LaxSr1-xCoyFe1-yO3±δ can vary about one order of magnitude [13, 18, 92, 93]. One strategy to improve performance of LSCF cathodes is the fabrication of composite electrodes with CGO [50, 94], CGO/Ag [95] or SDC [96] or to obtain higher surface exchange coefficient k by impregnating LSCF with Pd [49].

Cathode performance can also be improved by substituting one or more of the elements in LnxSr1-xCoyFe1-yO3±δ. Enhanced performance at low temperatures (~600°C) is obtained for Ln = Ce, Dy [97], whereas TEC is lowered for Ln = Nd [98]. Reaction products with YSZ are less pronounced for Ln = Pr, Nd, Gd [99]. On CGO, no reaction products are found for Ln = La, Gd, Sm, Nd [100, 101], although no distinct reaction products with LSGM are found, Co diffusion into the electrolyte is detected [102]. Sr doped lanthanum ferrites have also been investigated, since they have a lower TEC than LSCF [103], but they also form Sr or La-zirconates with YSZ

[104], which can be reduced by adding Al to LaFe1-xAlxO3 systems without Sr doping

[105] or using Ce0.8Sm0.2O1.9 protection layers [104]. The conductivity is comparable to that of LSCF, and is enhanced by adding Ni [106, 107], or replacing Sr with Ni [108], but is decreased by adding Al [109, 110].

Another material that is investigated for cathodes is Sm1-xSrxCoO3 (SSC) [38, 111, 112], showing lower overpotential than LSC [112]. Fabricating composites with the electrolyte material (Ce0.8Sm0.2O1.9), the interfacial resistances are reduced [65]. SSC is also used for single chamber SOFC applications [113, 114].

Barium cobaltates Ba1-xLnxCoO3, Ln = La, Pr are studied on either BaCeO3 [111, 115] or LSGM [116] based electrolytes and found to have less polarization losses than SSC for Ln = Pr [111], but higher overpotentials than SSC for Ln = La [116].

30 C H A P T E R 2

2.2.2.2 New Cathode Materials

Pyrochlore ruthenates have been investigated with compositions of Bi2Ru2O7.3,

Pb2Ru2O6.5 and Y2Ru2O7. Only the latter was found to be stable on CGO electrolytes, but additional doping with SrO is necessary in order to reach reasonable conductivity [117].

The search for new cathode materials for intermediate temperatures led to the discovery of La1-xSrxCuO2.5-δ. This material is a possible cathode candidate because it shows no reaction with YSZ, it exhibits high conductivity and gives reasonably low overpotential [118]. La2Ni1-xCuxO4+δ on the other hand shows high diffusion and surface exchange coefficients, but rather low conductivity, comparable to LSM [119]. Composite cathodes of Ag and yttrium doped bismuth oxide show comparable performance to LSCF [94]. For Y1Ba2Cu3O7 an additional layer of Pt or Ag is needed to promote oxygen adsorption [120]. Nd2NiO4+δ cathodes show lower polarization resistance than LSM but long term stability test have not be performed [36].

Another material that has recently drawn much attention is Ba0.5Sr0.5Co0.8Fe0.2O3-δ, which was initially developed for air separation membranes [121] and was used for fabrication of high performance cathodes [2], although the long term stability has not been proven.

2.2.2.3 Metal Cathodes

With the exception of Ag, metal electrodes exhibit higher ASR than mixed conducting perovskites at intermediate to high temperatures (600 to 1000°C) due to their poor oxygen conductivity [122]. However, since the ionic conductivity of the mixed conducting perovskites decreases strongly with decreasing temperature, metal cathodes become interesting at lower temperatures (300 to 400°C), since they retain there catalytic activity also at low temperatures (see also Table 2.1). Therefore they are interesting for µSOFCs, were they have been successfully used [3] giving high power densities.

S T A T E – O F – T H E A R T 31

2.2.3 Geometrically Well-Defined Cathodes

Several studies have been published on gwd cathode thin films. Few are devoted to fabrication and processing-related issues, and instead focus on the investigation of the reaction mechanism.

2.2.3.1 Studies of the Reaction Mechanism

The oxygen reduction pathway at thin La0.9Sr0.1MnO3-δ electrodes has been visualized by Horita et al. [123-125]. After deposition on YSZ with radio frequency (RF) sputtering, the film of 490 nm thickness was microstructured as a mesh with photolithography and atomic etching techniques. Oxygen reduction was performed at 18 and out of equilibrium using the stable O2 isotope as a marker. The samples were quenched and analyzed with secondary ion mass spectroscopy (SIMS) at various points of the electrode, notably at the LSM surface and close to the LSM/YSZ interface. This qualitative approach to oxygen reduction showed that thin LSM films promote the bulk pathway especially at high overpotentials.

Brichzin et al. used gwd, dense circular (La0.8Sr0.2)0.92MnO3-δ microelectrodes and measured their impedance as a function of the diameter [126]. The films were prepared by PLD followed by microstructuring with photolithography and etching. The resulting electrodes had a thickness of 100 nm and a diameter, d, ranging between 10 and 200 µm. The polarization resistance was found to be dependant on d-2. In a continuing work a linear dependence between the electrode thickness and the polarization resistance was established [127]. The authors concluded that the bulk pathway is dominant, in contrast to the common view that LSM, whose oxygen transport properties are known to be poor [128], favors the surface pathway. Similarly, Baumann et al. investigated the impedance of thin microstructured

La0.6Sr0.4Co0.8Fe0.2O3-δ electrodes and concluded the same as Brichzin with regard to the preferred reaction bulk pathway [129].

Koep et al. investigated oxygen reduction using arrays of La0.8Sr0.2MnO3 stripe electrodes [130]. The LSM patterns with 260 nm thickness were prepared using RF sputtering and photolithographic techniques. To determine the contribution of the bulk pathway, the oxygen reduction impedance of the LSM electrodes was measured with and without a thin insulating TiO2 layer on top. The role of the TiO2 was to block 32 C H A P T E R 2 the oxygen access to the LSM/gas interface while leaving the triple phase boundary active. The significantly larger polarization resistance obtained for the TiO2-covered electrode confirmed the findings of Brichzin [126, 127], i.e. a contribution of the bulk pathway and a large ASR.

Fig. 2.3: Polarization resistance Rp of oxygen reduction at thin dense La0.52Sr0.48Co0.18Fe0.82O3-δ electrodes as a function of the thickness d at 500°C and 600°C at equilibrium [131-133]. The full symbols (■, ▲) represent the experimental data. The dashed lines represent the simulated data from the model obtained after numerical optimization. Oxygen bulk diffusion is evidenced to be one of the rate determining step of the reaction.

The use of gwd electrodes allows to study the influence of geometry on the kinetics, both from the experimental and theoretical point of view [134]. Prestat et al. investigated oxygen reduction at thin dense gwd La0.52Sr0.48Co0.18Fe0.82O3-δ electrodes by combining state-space modeling of the Faradaic impedance and experimental measurements [131-133]. The thickness of the square-shaped films was varied between 16 nm and 800 nm while the surface area remained unchanged (3 x 3 mm2). Simulations showed the contribution of the surface pathways is negligible for electrodes with such geometry. Bulk diffusion of oxygen vacancies was determined to be one of the rate-determining steps by the increase of the polarization resistance upon increasing the film thickness (Fig. 2.3). The other rate-limiting step (adsorption and incorporation) were identified using a numerical optimization between modeling and experimental data. Rate constants of the rate-determining S T A T E – O F – T H E A R T 33 step were assessed and found to depend on the thickness of the cathode. Besides these physical models that aim for understanding of the reaction mechanism, other theoretical work is based on providing equivalent circuits to simulate the behavior of a mixed conducting electrode [135].

2.2.3.2 Fabrication Related Studies

Most related to fabrication is the work by Bieberle-Hütter and Tuller [136]. They investigated the microstructuring of LSC into interdigitated electrodes on different substrates. It was found that the etching quality was strongly dependent on the substrate material (decreasing quality from Si to CGO to YSZ) meaning that even at a temperature of less than 500°C, LSC interacts with the substrate and forms some interface. Hence, for feature sizes of less than 50 µm, a microfabrication process alternative to etching has to be used. The electrical and electrochemical characterization of the interdigitated LSC electrodes showed a strong influence of the thin film deposition parameters (sputtering) on the conductivity [137]. The substrate selection was found to be important not only with respect to fabrication of the pattern [136] but also with respect to electrical disturbance. Detailed electrochemical analysis of the LSC electrodes was not possible due to the geometrical constraints.

2.2.4 Cathode Performance: Thin Films vs. Porous Thick Films

Though thin films were first considered as model systems to investigate the fundamentals of oxygen reduction, they can also be utilized to increase the performance of SOFC cathodes. This is notably true for miniaturized systems where traditional ceramic techniques (like screen-printing) cannot be applied for practical reasons. Discrepancies about the performance of thin cathode films still exist in the literature. Sirman et al. compared the impedance of porous and dense

La0.6Sr0.4Co0.2Fe0.8O3-δ electrodes [138]. The dense electrode with 1 µm thickness showed a significantly larger polarization resistance than the porous one. Those findings contradict the results reported by Steele et al. for the same perovskite [53]. In this work, the authors measured the influence of a dense LSCF layer (~1 µm thick) placed between a porous LSCF film and CGO electrolyte. The area specific resistance of the electrode with the dense film was 2 to 3 times lower. More recently, 34 C H A P T E R 2

Prestat et al. compared the current-potential plots of porous LSCF electrodes with and without an LSCF thin dense film adjoining the electrolyte [131, 133]. The performance was improved when the film was sufficiently thin (< ca. 60 nm) and poorer when the dense layer was too thick (> ca. 340 nm). This suggests that high performance could be achieved by coating the electrolyte with an ultra-thin cathode film (see Fig. 2.4). The loss of the tpb (contribution of the surface pathway) would be compensated for by a complete coverage of the electrolyte by the active material. The kinetic losses due to bulk diffusion would be reduced significantly by minimizing the film thickness as much as possible. Materials with facile charge transfer at the electrode-electrolyte interface and with fast incorporation of oxygen should be sought. Promising advances in this field have been recently reported in the literature [2].

Fig. 2.4: Cross-section micrograph of an ultrathin dense La0.52Sr0.48Co0.18Fe0.82O3-δ layer prepared by PLD. The substrate is a polished cerium-gadolinium oxide pellet.

Moreover a recent paper of Baumann et al. [139] suggests that electrochemical performance could be “activated” by applying a large voltage bias to thin film cathodes. In the case of La0.6Sr0.4Co0.8Fe0.2O3-δ, the effect is attributed to the change of cation stoichiometry close to the surface. However, at the time of writing, this type of electrochemical activation had not been proven to be stable with time.

2.2.5 Porous Thin Film Cathodes

Few studies are available on porous thin film cathodes. Most studies are limited to the preparation of such layers [140-143], and only a very limited number of studies are available where typical cathode performance data such as ASR is given [52, 54]. S T A T E – O F – T H E A R T 35

Both publications showed that the ASR of porous thin film cathodes is not significantly different from that of porous thick film cathodes.

2.2.6 New Issues Raised by Thin Films

Thin film cathodes helped to gain better understanding of the complicated reaction at the cathode. However, also new questions emerged when using thin films. These shall be addressed in the following sections.

2.2.6.1 Geometric Considerations

Reaction kinetics depend not only intrinsically on the nature of the material but also on the electrode geometry. Thus, one has to keep in mind that the data obtained on thin layers (in the sub-micrometer range) with large surface area cannot be directly transposed to porous electrodes with a typical grain size of ca. 1 µm. For instance, the thin LSM electrodes of Brichzin et al. [126, 127] exhibit a specific resistivity of ca. 400 Ω·cm2 for LSM at 800°C, which is much greater than other data reported in literature [144]. This can be attributed to the fact that the geometry of the electrodes (the diameter was much greater than the thickness) forced the reaction to occur through the bulk of the LSM, which does not transport oxygen ions efficiently.

2.2.6.2 Current Collection

Electrical resistance occurs also in-plane of a cathode. Therefore current collectors serving also as interconnects between two cells are used. When working with thick porous electrodes (typically > 10 µm) prepared by conventional ceramic processing methods, such as screen-printing, current collectors can be made from a metallic mesh. For laboratory purposes a metallic gauze (made of platinum, for instance) is pressed on the layer in its green state. Co-firing the paste and the current collector yields a good contact and reliable mechanical stability between the components. Homogeneous potential distribution is achieved throughout the bulk of high conductivity materials like LSCF and LSM. Current collection is more arduous when dealing with thin electrodes. Besides the fact that the films do not have any green state, two contradictory requirements have to be fulfilled: On one hand, molecular 36 C H A P T E R 2 oxygen should have access to the film, which implies a sufficiently open porosity. On the other hand, in order to ensure a homogeneous potential distribution when working out of equilibrium, the current collector should fully cover the thin layer in the ideal case. Different approaches to meet these constraints have been reported in the literature, each of them having its own advantages and drawbacks.

Brichzin et al. [126, 127], Baumann et al. [129, 139] and Yildiz et al. [134] collected the current by means of a tungsten carbide tip when using LSM and LSCF microelectrodes, enabling a homogeneous access of oxygen to the gas/electrode interface. The contribution of the tip to the oxygen reduction can be neglected. Yet, when working with thin films, this kind of point contact may lead to heterogeneous potential and current distribution within the electrode and current constrictions which add unknown resistance to the electrode. This may explain the huge area specific resistivity of LSCF reported by Baumann [129] (10 Ω·cm2 at 750°C), i.e. around two orders of magnitudes greater than that of porous electrodes [145].

A thin dense silver layer may also be employed since oxygen can be transported through it and a homogeneous potential distribution at the electrode/current collector interface is ensured. Conversely, it does not enable one to investigate the adsorption and incorporation of oxygen at the perovskite/gas interface. Application as such of a silver layer could be suitable for studies focusing particularly on the bulk properties of the electrode, such as diffusion coefficient. Ringuedé et al. utilized this kind of current collector to investigate oxygen reduction of La0.7Sr0.3CoO3-δ electrodes [146].

Platinum is a standard current collector in SOFC experiments. Several authors utilized it as paste or gauze to contact thin cathode films [138, 147, 148]. Although this kind of current collector is rather straightforward to set up, it may also significantly affect the kinetics of the thin film/gas interface. In theory, the incorporation of oxygen from the gas phase in the perovskite electrode is chemical in nature and involves no charge transfer. The oxygen is incorporated on oxygen vacancies and holes are created within the perovskite. That means no charged species crosses the gas/perovskite interface. The charge transfer occurs at the perovskite/electrolyte interface, where the oxygen vacancies (and the O2- ions, respectively) cross the interface. By introducing a metallic current collector that is active towards oxygen reduction, it is then no longer clear whether oxygen is reduced directly by the perovskite or by the platinum. In other words, oxygen reduction can S T A T E – O F – T H E A R T 37 occur at the platinum/perovskite/gas interface as well. Hence the experimental kinetic data, such as adsorption and incorporation rate constants, cannot be unambiguously assigned to the investigated thin film only.

Prestat et al. used a porous LSCF film as current collector to investigate oxygen reduction at a thin dense LSCF film [131, 133] constituting a compromise between the constraints of homogeneous potential distribution and access of molecular oxygen to the film. The porous layer was prepared by screen-printing. This approach is possible when working at macroscopic scales (in [131, 133], the surface area of the electrode was 9 mm2). The advantage is that no supplementary undesired interface is created (compared to the case of platinum current collectors described above) however, oxygen cannot uniformly adsorb on the electrode surface and it is difficult to determine to what extent the current collector may contribute to the overall kinetics of the reaction.

2.2.6.3 Defect Chemistry

Another essential aspect of oxygen reduction at thin films is the defect chemistry. Solid-state electrochemistry is a peculiar type of electrochemistry, since the electro- active species (oxygen in this case) are incorporated in the bulk of the electrode and modify its composition. A defect chemistry model is absolutely essential in the case of oxygen reduction at mixed ionic-electronic conducting perovskites. The B-site atoms are involved in the incorporation of oxygen into the mixed conductor and tied to the oxygen vacancy concentration if electro-neutrality is assumed. To the best of our knowledge, the defect chemistry of thin SOFC-related perovskite films out of electrochemical equilibrium (i.e. when a Faradaic current flows, under realistic SOFC operating conditions) has not been described yet. Facile charge transfer at the electrode/electrolyte interface and rate-limiting oxygen incorporation at the electrode/gas interface have been reported in the literature for perovskite electrodes [149]. For thin films, which “geometrically” promote the bulk pathway (large surface, thickness in the sub-micrometer range), those kinetic features favor a depletion of oxygen that may destroy the initial crystalline structure out of equilibrium. What is the maximum oxygen nonstoichiometry that the perovskite can withstand without undergoing decomposition? At high oxygen vacancy concentrations, what model can describe defect interactions? In the literature, defect chemistry models at low oxygen 38 C H A P T E R 2

x • ''−−•• ' − •• partial pressures involving clusters such as {BVBBoB} [150], {SrVLa o } [151,

• ' − •• 152] and, {BBoV } [153] have been proposed for various perovskites at equilibrium.

Do those models apply when an ionic current flows through the mixed conductor? Elucidating these open questions and more generally the defect chemistry of thin perovskite layers constitutes challenging and essential topics for future investigations.

2.3 Cells

µSOFCs are currently an important issue in research [9, 10], because they offer more geographical independence, higher energy densities than batteries and the possibility of fast recharging by refueling. Furthermore, the use of thin films allows lower operating temperatures. Thin film deposition is combined with micromachining techniques in order to realize fuel cells with micron dimensions. Typically such a µSOFC would look like that depicted in Fig. 2.5. The gas channels are micromachined in the supporting substrate. The thin film membrane, comprising the cell components, is deposited on the substrate. Although single-chamber SOFCs have been fabricated by micromachining [154, 155], to our knowledge there are no single-chamber SOFCs based on thin films so far.

Fig. 2.5: Sketch of a µSOFC. S T A T E – O F – T H E A R T 39

2.3.1 Modeling

To evaluate the feasibility of µSOFCs and to propose reasonable designs, calculations were done that took into consideration mechanical stability [156, 157], electrochemical performance [157, 158] as well as issues of heat losses [157].

Assuming thermal stress only and no residual stress from the fabrication processes, the stability of flat versus corrugated membranes was calculated [156]. An example of a corrugated membrane showing the geometry is shown in Fig. 2.6. Under compressive stress, buckling was defined as failure mode. Assuming no changes in the membrane material (YSZ in this case) with temperature, calculations showed that a corrugated membrane with a ratio of H/t = 10 can be more than one order of magnitude larger (500 µm vs. 16 µm) than a flat membrane with the same thickness (1 µm) when exposed to a temperature difference between membrane and substrate of 600°C. If tensile stress and, thus, fracture was considered as failure mode, the corrugated membrane could still take roughly twice the stress of a flat membrane. For practical use, a flat membrane with corrugated support on the edges might be the most useful, since the highest stress always occurs at the edge of the membrane.

Fig. 2.6: Example of a corrugated membrane, after [156].

In case the residual stresses of fabrication processes are known or can be tuned, they have to be added to the thermal stress [157]. In any case, from the point of mechanical stability, a thick membrane with a small radius is favorable, and a low mismatch in thermal expansion coefficient between the membrane and substrate material also improves stability. Rectangular membranes were identified to be slightly more resistant to buckling than circular or square membranes.

From the point of view maintaining a low heat loss, circular membranes were the most favorable [157], since they have the smallest perimeter for a given area. Assuming a radial temperature gradient for the membrane, with room temperature at the edge of the membrane and considering heat losses through conduction only, the 40 C H A P T E R 2 heat loss of the structure was calculated and found to exceed the assumed heat production of a µSOFC.

Design considerations concerning the electrochemical performance [157] were calculated using

P = i·(OCV-ηOhmic-ηActivation-ηConcentration). (2.2)

With P being the power density, i the current density and η the different polarization losses. For ohmic polarization, the losses at the electrolyte were considered as follows:

t η e Ohmic = i· σ (2.3) e

where te and σe are the thickness and ionic conductivity of the electrolyte, respectively. The activation polarization was calculated by:

RT i 2RT i ηActivation = · + ·ln (2.4) nF i1 nF i2 where R = 8.314 J·mol-1·K-1, F = 96485 C·mol-1, T is the temperature, n the number of electrons exchanged per oxygen ion. The exchange current densities i1 and i2, which are treated as free parameters, were obtained by fits to measurements from literature [159]. Concentration polarization was neglected. Based on these calculations, it was found that already for an electrolyte thickness of 2 µm, the activation losses already exceed the ohmic losses by far, thus when choosing an electrolyte thickness for µSOFCs, it can be chosen with respect to mechanical stability without spoiling the electrochemical performance.

A very detailed calculation concerning the electrochemical performance of µSOFCs and single-chamber µSOFCs is given by Fleig et al. [158]. In these calculations, it was shown that the resistance of the electrolyte in a real fuel cell situation does not decrease linearly with thickness as might be expected. The reason for this behavior is the current constriction in the electrolyte. The electrodes were considered as particles sitting on the electrolyte. Depending on the reaction kinetics and conduction mechanism of the electrode material, the current is either constricted to the electrode particle/electrolyte interface, or to the tpb at the edges of the electrode particles. For S T A T E – O F – T H E A R T 41 both cases the electrolyte resistance decreases less than linearly with electrolyte thickness, if the electrolyte thickness is below the particle distance of the electrodes. Only by using electrodes with finer and closer packed particles a meaningful reduction in electrolyte resistance was achieved. However, the ohmic losses of electrolytes with 500 nm thickness at 550°C are smaller than 0.25 Ω·cm2 for both CGO and YSZ. Thus, the performance of a µSOFC is limited by the electrodes and not by the electrolyte.

In case of single-chamber µSOFCs, geometries with interdigitated electrodes placed on one side of the electrolyte were discussed [158]. The necessary geometric constraints in order to achieve a total resistance in the cell of 1.5 Ω·cm2 were calculated. For rather thick electrodes (> 1 µm), an electrode stripe width of around 10 µm would be required, while the distance between the electrodes would have to be as small as a few µm. To facilitate the micro-fabrication of such a cell, thin film technology might be considered for fabrication of the electrodes. If this were to involve a dense thin film cathode, the necessary tracer exchange coefficient of the cathode material should be 5·10-6 cm/s at 600°C to match the target performance. However, this is not fulfilled by standard cathode materials, such as LSC or LSCF. On the anode side, if a thin dense Ni film would be used, further nanostructuring of the film would be needed in order to increase the tpb. These severe geometrical constraints do not only complicate the fabrication process, but the small distances between the cathode and anode may cause further problems if reaction products of the anode modify the cathode potential. The rapid transport of highly reactive products of partial oxidation (CO + H2) to the cathode, as occurs for such a cell geometry, limits the performance of single-chamber SOFCs, since these products might also be catalyzed at the cathode [160].

2.3.2 Free-Standing Electrolytes

Free-standing YSZ [161, 162] and CGO [161] membranes have both been fabricated on silicon substrates. For fabrication of YSZ membranes [162], SiO2 was used as masking layer. After removing the oxide from the opposite side, YSZ was deposited by RF sputtering, and the membranes were released with wet etching. In another process, YSZ was deposited in the cavities after wet etching and the oxide was removed from the opposite side afterwards. For both processes, an annealing step at 42 C H A P T E R 2

400°C in air was carried out after YSZ deposition in order to reduce the compressive stress in the membrane. The second process resulted in a better yield and larger membranes. Square membranes with dimensions up to 170 µm were obtained, but no tests concerning thermal stability are reported.

Baertsch et al. deposited silicon nitride (SiN) on both sides of a Si wafer [161], which served as a masking layer for wet etching of Si on one side and as electrical insulation on the other. The electrolyte was deposited by either electron beam (e-beam) evaporation (YSZ, CGO) or RF sputtering (YSZ). Afterwards the SiN below the electrolyte was removed using dry etching techniques. The residual stress in the as-deposited thin films on the substrate was evaluated and all e-beam evaporated films showed tensile stresses around 200 MPa, independent of the film thickness, for films < 1 µm. The RF sputtered YSZ films showed compressive residual stress from 850 MPa for thin (100 nm) to about 200 MPa for thick (1 µm) films. Using the same film thickness and deposition method, the yield of YSZ membranes was higher than that of CGO membranes. The maximum membrane width was roughly twice as high for YSZ (1025 µm) than for CGO (525 µm) for membranes with the same thickness. The membranes fabricated by e-beam evaporation failed during heating through membrane cracking from the center of the membrane, which was attributed to brittle fracture caused by tension. Generally YSZ membranes were able to withstand higher temperatures than CGO membranes; thicker and smaller membranes were also more robust. A high yield of more than 90 % was achieved for sputtered YSZ films. Contrary to the calculated results [156, 157], buckling was not considered as a failure, since buckling without fracture does not harm SOFC operation. Upon annealing at 500°C buckling of the membranes disappeared completely and fracture via tensile stresses occurred with further annealing at 650°C. Since the TEC of YSZ is roughly three times higher than that of the substrate material, Si, one would expect the contrary, i.e. evolution of compressive stress upon heating. Although the underlying mechanism is not fully understood, microstructural and chemical changes in the material might contribute to this result. This phenomenon also questions the use of stability calculations, where these changes in the material cannot be included easily. The work on free-standing membranes, in this case electrolytes, gives a good insight into what can be expected in terms of mechanical stability of such a ceramic thin film membrane, which is essential to know when fabricating µSOFCs. S T A T E – O F – T H E A R T 43

Other studies also showed that material behaves more complex than expected. Free- standing CeO2 membranes were fabricated by RF sputtering on Si substrates and wet etching [163]. Depending on the preparation conditions the membranes either showed buckling irrespective if the stress after deposition was tensile or compressive or, when sputtered with less oxygen partial pressure, the membranes were flat. However, the buckled membranes proved to be mechanically more stable surviving higher temperatures (> 220°C) than the flat membranes (150°C). This phenomenon was attributed to ordering of disordered vacancies, associated with lateral expansion. The material can change its dimensions reversibly and thereby respond to stress [164].

2.3.3 Si-Based Cells

Tests of µSOFCs based on Si substrates were reported in [3, 159, 165]. Electrodes with 500 to 850 nm thickness [165] were either sputtered through masks with porous microstructure, or, when sputtering was done that resulted in a dense microstructure, pores were etched in the electrodes after deposition. Ni was used for the anode, silver for the cathode. The electrolyte was dense sputtered YSZ with a thickness of 2.5 µm. A free-standing membrane of this tri-layer was obtained by wet etching a Si wafer with SiN masking and SiN etch stop which was subsequently removed by dry etching. A power output of 3.8 mW/cm2 was obtained at 316°C and the measured open circuit voltage (OCV) was 0.8 V. The geometries of the free-standing membranes were not given and for dense electrodes no output could be measured [165]. At 600°C a maximum power of 145 mW/cm2 and an OCV of 1.0 V was reported [159]. In this case the micromachining was modified to obtain membranes of 5 µm size, with the thickness of each layer between 500 nm and 10 µm.

Another very recent publication [3] shows very high power densities of 200 and 400 mW/cm2 at 350 and 400°C, respectively. A very thin electrolyte consisting of 50 nm YSZ and 50 nm CGO was sputtered on a SiN/Si substrate, wet and dry etching was done to structure the substrate and obtain membranes up to 240 by 240 µm in size. Afterwards, porous Pt (80 nm) was sputtered for use as an anode and cathode. With only YSZ as electrolyte the power output was less, 130, 85 and 60 mW/cm2 at 350°C for 50, 100 and 150 nm electrolyte thickness. This would indicate a strong relation between electrolyte thickness and cell performance 44 C H A P T E R 2 although the authors report the ASR of the electrodes to be one order of magnitude larger than the contribution of the electrolyte. Further studies are necessary to fully understand the behavior of µSOFCs.

The area of the free-standing membrane can be enlarged by mechanically supporting the membrane with a Ni-grid, which can also be used as a current collector [166].

2.3.4 Ni-Based Cells

Ni was also used as substrate for a µSOFC. Kang et al. [167] used sputtered Pt as an anode catalyst on a nano-porous Ni support. On top, a YSZ electrolyte 200 nm thick was sputtered, followed by porous sputtered Pt as a cathode layer. The maximum power density obtained at 400°C was 7 mW/cm2. Chen et al. used Ni foils as substrates [168]. PLD was used to deposit YSZ films 2 µm thick on Ni foils 6 µm thick. The Ni foil was then patterned with pores of about 70 µm. Afterwards a porous LSC cathode layer with 6 µm thickness was deposited also using PLD and, to increase the triple phase boundary at the anode side, a 6-µm-thick NiO-YSZ anode was added to the Ni foil. A power output of 110 mW/cm2 at 570°C was achieved and an OCV of 0.8 V. The total thickness of the cell was only about 20 µm and thus one order of magnitude thinner than thin film µSOFCs based on silicon substrates.

2.3.5 Glass-Ceramic Based Cells

Glass-ceramics capable of being photostructured can also be used as micromachinable substrates for µSOFCs [9]. The advantages are a TEC close to that of SOFC materials and the possibility of rapid wet etching during fabrication of the free standing membrane.

2.4 Summary and Conclusion

Mostly perovskites are used for cathodes since the desired properties can be adjusted by manipulating the material composition. LSM and LSCF are the best investigated materials while BSCF seems to be one of the most promising materials today. While metal cathodes, namely Pt cathodes, were discarded for high temperature application they become again interesting for application at lower S T A T E – O F – T H E A R T 45 temperatures (300 – 400°C) due to their high catalytic activity. This makes them interesting as cathodes for µSOFCs. Research on thin film cathodes has helped to better understand the cathode reaction. However, new issues have also appeared during these investigations. Only a few studies were carried out on porous thin film cathodes, but the emergence of µSOFCs will stimulate more research in this field.

When it comes to entire cells, calculations of geometrical constraints of µSOFCs are useful particularly for evaluating their electrochemical performance. From these results it is clear that a reduction of electrolyte thickness below 500 nm is not meaningful (depending on the electrode grain size) and that the electrodes limit the performance of µSOFCs. Geometrical restrictions for single-chamber µSOFCs are severe. Calculations of the mechanical stability seem to be only of limited use, since the complex behavior of the material is not well represented by those calculations. Modeling of the expected heat losses indicates that thermal insulation is an important issue in fabrication of µSOFCs. Stability tests of free-standing electrolytes and complete µSOFCs suggest that rather small membranes of about 100 µm feature size are feasible. Cells with high power output at very low temperatures (350 – 400°C) were fabricated and showed a strong influence of the electrolyte thickness on the cell performance. This was not expected from modeling work and needs further work to understand µSOFCs, especially since the electrodes were identified to govern the cell performance. The use of Ni foils instead of Si as the substrate allows a further reduction of cell thickness; glass-ceramic substrates have better thermal properties than Si.

More work will be carried out on µSOFCs to analyze the complex mechanical and electrochemical behavior of these nano- and micro-sized materials in SOFCs. The next challenges in terms of µSOFCs production will be fabrication and contacting of an array of µSOFCs, since it also requires strong process control to obtain a reasonable yield. Stacking and housing, start-up operation and thermally self- sustained operation of a portable device are also big challenges for microfabrication and process engineering. However, the research carried out so far already proves that the concept of a µSOFC is realizable and that high power density at low temperatures can be expected. Hence, new applications of SOFCs are foreseen for the future.

46 C H A P T E R 2

2.5 References

[1] S. C. Singhal, K. Kendall, "High Temperature Solid Oxide Fuel Cells: Fundamentals, Design and Applications" Elsevier Advanced Technology, Oxford, UK p. 1 (2003).

[2] Z. P. Shao, S. M. Haile, "A High-Performance Cathode for the Next Generation of Solid- Oxide Fuel Cells", Nature, 431, [7005] 170-3 (2004).

[3] H. Huang, M. Nakamura, P. Su, R. Fasching, Y. Saito, F. B. Prinz, "High- Performance Ultrathin Solid Oxide Fuel Cells for Low-Temperature Operation", Journal of the Electrochemical Society, 154, [1] B20-B4 (2007).

[4] T. Hibino, S. Q. Wang, S. Kakimoto, M. Sano, "Single Chamber Solid Oxide Fuel Cell Constructed From an Yttria-Stabilized Zirconia Electrolyte", Electrochemical and Solid State Letters, 2, [7] 317-9 (1999).

[5] T. Hibino, H. Iwahara, "Simplification of Solid Oxide Fuel-Cell System Using Partial Oxidation of Methane", Chemistry Letters, [7] 1131-4 (1993).

[6] K. Asano, T. Hibino, H. Iwahara, "A Novel Solid Oxide Fuel-Cell System Using the Partial Oxidation of Methane", Journal of the Electrochemical Society, 142, [10] 3241-5 (1995).

[7] T. Hibino, A. Hashimoto, M. Yano, M. Suzuki, S. Yoshida, M. Sano, "High Performance Anodes for SOFCs Operating in Methane-Air Mixture at Reduced Temperatures", Journal of the Electrochemical Society, 149, [2] A133-A6 (2002).

[8] M. Yano, A. Tomita, M. Sano, T. Hibino, "Recent Advances in Single-Chamber Solid Oxide Fuel Cells: A Review", Solid State Ionics, 177, [39-40] 3351-9 (2007).

[9] A. Bieberle-Hütter, D. Beckel, U. P. Muecke, J. L. M. Rupp, A. Infortuna, L. J. Gauckler, "Micro-Solid Oxide Fuel Cells as Battery Replacements", Mstnews, 04/05, 12-5 (2005).

[10] D. Nikbin, "Micro SOFCs: Why Small is Beautiful", The Fuel Cell Review, April/May, 21-4 (2006).

[11] S. B. Adler, "Factors Governing Oxygen Reduction in Solid Oxide Fuel Cell Cathodes", Chemical Reviews, 104, [10] 4791-843 (2004). S T A T E – O F – T H E A R T 47

[12] L. J. Gauckler, D. Beckel, B. Buergler, E. Jud, U. P. Muecke, M. Prestat, J. Rupp, J. Richter, "Solid Oxide Fuel Cells: Systems and Materials", Chimia, 58, [12] 837-50 (2004).

[13] S. Wang, M. Katsuki, M. Dokiya, T. Hashimoto, "High Temperature Properties

of La0.6Sr0.4Co0.8Fe0.2O3-[delta] Phase Structure and Electrical Conductivity", Solid State Ionics, 159, [1-2] 71-8 (2003).

[14] H. U. Anderson, "Review of P-Type Doped Perovskite Materials for SOFC and Other Applications", Solid State Ionics, 52, [1-3] 33-41 (1992).

[15] J. W. Stevenson, T. R. Armstrong, R. D. Carneim, L. R. Pederson, W. J. Weber, "Electrochemical Properties of Mixed Conducting Perovskites

La1-xMxCo1-yFeyO3-delta (M = Sr, Ba, Ca)", Journal of the Electrochemical Society, 143, [9] 2722-9 (1996).

[16] T. Nakamura, G. Petzow, L. J. Gauckler, "Stability of the Perosvkite Phase

LaBO3 (B = V, Cr, Mn, Fe, Co, Ni) in Reducing Atmosphere. 1. Experimental Results", Materials Research Bulletin, 14, [5] 649 (1979).

[17] A. Mitterdorfer, L. J. Gauckler, "La2Zr2O7 Formation and Oxygen Reduction

Kinetics of the La0.85Sr0.15MnyO3, O2(g)|YSZ System", Solid State Ionics, 111, [3-4] 185-218 (1998).

[18] H. Ullmann, N. Trofimenko, F. Tietz, D. Stover, A. Ahmad-Khanlou, "Correlation Between Thermal Expansion and Oxide Ion Transport in Mixed Conducting Perovskite-Type Oxides for SOFC Cathodes", Solid State Ionics, 138, [1-2] 79-90 (2000).

[19] N. Q. Minh, "Ceramic Fuel Cells", Journal of the American Ceramic Society, 76, [3] 563-88 (1993).

[20] H. Yokokawa, N. Sakai, T. Horita, K. Yamaji, "Recent Developments in Solid Oxide Fuel Cell Materials", Fuel Cells, 1, [2] 117 - 31 (2001).

[21] B. C. H. Steele, A. Heinzel, "Materials for Fuel-Cell Technologies", Nature, 414, [6861] 345-52 (2001).

[22] N. P. Brandon, S. Skinner, B. C. H. Steele, "Recent Advances in Materials for Fuel Cells", Annual Review of Materials Research, 33, 183-213 (2003). 48 C H A P T E R 2

[23] S. P. S. Badwal, K. Foger, "Solid Oxide Electrolyte Fuel Cell Review", Ceramics International, 22, [3] 257-65 (1996).

[24] V. Kozhukharov, N. Brashkova, M. Ivanova, J. Carda, M. Machkova, "Ceramic Materials for SOFCs: Current Status", Boletin De La Sociedad Espanola De Ceramica Y Vidrio, 41, [5] 471-80 (2002).

[25] J. P. P. Huijsmans, "Ceramics in Solid Oxide Fuel Cells", Current Opinion in Solid State & Materials Science, 5, [4] 317-23 (2001).

[26] B. C. H. Steele, "Materials for IT-SOFC Stacks: 35 Years R&D: the Inevitability of Gradualness?" Solid State Ionics, 134, [1-2] 3-20 (2000).

[27] S. M. Haile, "Fuel Cell Materials and Components", Acta Materialia, 51, 5981- 6000 (2003).

[28] A. J. McEvoy, "Materials for High-Temperature Oxygen Reduction in Solid Oxide Fuel Cells", Journal of Materials Science, 36, [5] 1087-91 (2001).

[29] J. M. Ralph, A. C. Schoeler, M. Krumpelt, "Materials for Lower Temperature Solid Oxide Fuel Cells", Journal of Materials Science, 36, [5] 1161-72 (2001).

[30] R. M. Ormerod, "Solid Oxide Fuel Cells", Chemical Society Reviews, 32, [1] 17-28 (2003).

[31] A. Weber, E. Ivers-Tiffee, "Materials and Concepts for Solid Oxide Fuel Cells (SOFCs) in Stationary and Mobile Applications", Journal of Power Sources, 127, [1-2] 273-83 (2004).

[32] S. J. Skinner, J. A. Kilner, "Oxygen Ion Conductors", Materials Today, 6, [3] 30-7 (2003).

[33] J. Will, R. Stadler, M. K. M. Hruschka, L. J. Gauckler, "Fabrication Processes for Electroceramic Components", in Oxygen Ion and Mixed Conductors and Their Technological Applications, Edited by H.L. Tuller et al., Kluwer Academic Press. p. 165-243 (2000).

[34] S. W. Zha, Y. L. Zhang, M. L. Liu, "Functionally Graded Cathodes Fabricated by Sol-Gel/Slurry Coating for Honeycomb SOFCs", Solid State Ionics, 176, [1- 2] 25-31 (2005). S T A T E – O F – T H E A R T 49

[35] A. Barbucci, M. Viviani, P. Carpanese, D. Vladikova, Z. Stoynov, "Impedance Analysis of Oxygen Reduction in SOFC Composite Electrodes", Electrochimica Acta, 51, [8-9] 1641-50 (2006).

[36] F. Mauvy, J.-M. Bassat, E. Boehm, J.-P. Manaud, P. Dordor, J.-C. Grenier,

"Oxygen Electrode Reaction on Nd2NiO4+[delta] Cathode Materials: Impedance Spectroscopy Study", Solid State Ionics, 158, [1-2] 17-28 (2003).

[37] R. Chiba, F. Yoshimura, Y. Sakurai, Y. Tabata, M. Arakawa, "A Study of Cathode Materials for Intermediate Temperature SOFCs Prepared by the Sol- Gel Method", Solid State Ionics, 175, [1-4] 23-7 (2004).

[38] J. M. Ralph, C. Rossignol, R. Kumar, "Cathode Materials for Reduced- Temperature SOFCs", Journal of the Electrochemical Society, 150, [11] A1518-A22 (2003).

[39] K. Choy, W. Bai, S. Charojrochkul, B. C. H. Steele, "The Development of Intermediate-Temperature Solid Oxide Fuel Cells for the Next Millennium", Journal of Power Sources, 71, [1-2] 361-9 (1998).

[40] K. Barthel, S. Rambert, S. Siegmann, "Microstructure and Polarization Resistance of Thermally Sprayed Composite Cathodes for Solid Oxide Fuel Cell Use", Journal of Thermal Spray Technology, 9, [3] 343-7 (2000).

[41] X. Y. Xu, C. R. Xia, S. G. Huang, G. Y. Meng, "Intermediate-Temperature

Solid Oxide Fuel Cells with Y0.25Bi0.75O1.5-Ag Cathodes", in Pricm 5: The Fifth Pacific Rim International Conference on Advanced Materials and Processing, Pts 1-5, Trans Tech Publication LTD, Zurich-Uetikon. p. 1157-60 (2005).

[42] H. Zhao, L. Huo, S. Gao, "Electrochemical Properties of LSM-CBO Composite Cathode", Journal of Power Sources, 125, [2] 149-54 (2004).

[43] S. P. Yoon, J. Han, S. W. Nam, T.-H. Lim, I.-H. Oh, S.-A. Hong, Y.-S. Yoo, H. C. Lim, "Performance of Anode-Supported Solid Oxide Fuel Cell with

La0.85Sr0.15MnO3 Cathode Modified by Sol-Gel Coating Technique", Journal of Power Sources, 106, [1-2] 160-6 (2002).

[44] H. S. Song, S. H. Hyun, J. Moon, R. H. Song, "Electrochemical and Microstructural Characterization of Polymeric Resin-Derived Multilayered Composite Cathode for SOFC", Journal of Power Sources, 145, [2] 272-7 (2005). 50 C H A P T E R 2

[45] N. T. Hart, N. P. Brandon, M. J. Day, J. E. Shemilt, "Functionally Graded Cathodes for Solid Oxide Fuel Cells", Journal of Materials Science, 36, [5] 1077-85 (2001).

[46] N. T. Hart, N. P. Brandon, M. J. Day, N. Lapena-Rey, "Functionally Graded Composite Cathodes for Solid Oxide Fuel Cells", Journal of Power Sources, 106, [1-2] 42-50 (2002).

[47] S. P. Jiang, "A Review of Wet Impregnation - An Alternative Method for the Fabrication of High Performance and Nano-Structured Electrodes of Solid Oxide Fuel Cells", Materials Science and Engineering A, 418, [1-2] 199-210 (2006).

[48] M. Hori, K. Nagasaka, M. Miyayama, G. Trunfio, E. Traversa, "Evaluation of Electrode Performances of Single-Chamber Solid Oxide Fuel Cells", in Electroceramics in Japan Viii, p. 155-8 (2005).

[49] M. Sahibzada, S. J. Benson, R. A. Rudkin, J. A. Kilner, "Pd-Promoted

La0.6Sr0.4Co0.2Fe0.8O3 Cathodes", Solid State Ionics, 113-115, 285-90 (1998).

[50] V. Dusastre, J. A. Kilner, "Optimisation of Composite Cathodes for Intermediate Temperature SOFC Applications", Solid State Ionics, 126, [1-2] 163-74 (1999).

[51] H. Zhao, L. H. Huo, L. P. Sun, L. J. Yu, S. Gao, J. G. Zhao, "Preparation, Chemical Stability and Electrochemical Properties of LSCF-CBO Composite Cathodes", Materials Chemistry and Physics, 88, [1] 160-6 (2004).

[52] C. S. Hsu, B. H. Hwang, "Microstructure and Properties of the

La0.6Sr0.4Co0.2Fe0.8O3 Cathodes Prepared by Electrostatic-Assisted Ultrasonic Spray Pyrolysis Method", Journal of the Electrochemical Society, 153, [8] A1478-A83 (2006).

[53] B. C. H. Steele, J.-M. Bae, "Properties of La0.6Sr0.4Co0.2Fe0.8O3-x (LSCF) Double Layer Cathodes on Gadolinium-Doped Cerium Oxide (CGO) Electrolytes: II. Role of Oxygen Exchange and Diffusion", Solid State Ionics, 106, [3-4] 255-61 (1998).

[54] D. Beckel, U. P. Muecke, T. Gyger, G. Florey, A. Infortuna, L. J. Gauckler, "Electrochemical Performance of LSCF Based Thin Film Cathodes Prepared by Spray Pyrolysis", Solid State Ionics, 178, 407-15 (2007). S T A T E – O F – T H E A R T 51

[55] A. Esquirol, J. Kilner, N. Brandon, "Oxygen Transport in

La0.6Sr0.4Co0.2Fe0.8O3-[delta]/Ce0.8Ge0.2O2-x Composite Cathode for IT-SOFCs", Solid State Ionics, 175, [1-4] 63-7 (2004).

[56] W. G. Wang, M. Mogensen, "High-Performance Lanthanum-Ferrite-Based Cathode for SOFC", Solid State Ionics, 176, [5-6] 457-62 (2005).

[57] K. Kammer, "Studies of Fe-Co Based Perovskite Cathodes with Different A- Site Cations", Solid State Ionics, 177, [11-12] 1047-51 (2006).

[58] S. Wang, R. Zheng, A. Suzuki, T. Hashimoto, "Preparation, Thermal

Expansion and Electrical Conductivity of La0.6Sr0.4Co1-xGaxO3-delta (x=0.0-0.4) as a New Cathode Material of SOFC", Solid State Ionics, 174, [1-4] 157-62 (2004).

[59] G. Y. Zhu, X. H. Fang, C. R. Xia, X. Q. Liu, "Preparation and Electrical

Properties of La0.4Sr0.6Ni0.2Fe0.8O3 Using a Glycine Nitrate Process", Ceramics International, 31, [1] 115-9 (2005).

[60] S. Lee, Y. Lim, E. A. Lee, H. J. Hwang, J. W. Moon, "Ba0.5Sr0.5Co0.8Fe0.2O3-delta

(BSCF) and La0.6Ba0.4Co0.2Fe0.8O3-delta (LBCF) Cathodes Prepared by Combined Citrate-EDTA Method for IT-SOFCs", Journal of Power Sources, 157, [2] 848-54 (2006).

[61] H. Lv, Y. J. Wu, B. Huang, B. Y. Zhao, K. A. Hu, "Structure and

Electrochemical Properties of Sm0.5Sr0.5Co1-xFexO3(-delta) Cathodes for Solid Oxide Fuel Cells", Solid State Ionics, 177, [9-10] 901-6 (2006).

[62] Z. H. Bi, M. J. Cheng, Y. L. Dong, H. J. Wu, Y. C. She, B. L. Yi,

"Electrochemical Evaluation of La0.6Sr0.4CoO3-La0.45Ce0.55O2 Composite

Cathodes for Anode-Supported La0.45Ce0.55O2-La0.9Sr0.1Ga0.8Mg0.2O2.85 Bilayer Electrolyte Solid Oxide Fuel Cells", Solid State Ionics, 176, [7-8] 655-61 (2005).

[63] M. Gödickemeier, K. Sasaki, L. J. Gauckler, "Electrochemical Characteristics of cathodes in Solid Oxide Fuel Cells Based on Ceria Electrolytes", Journal of the Electrochemical Society, 144, [5] 1635-45 (1997).

[64] H. C. Yu, F. Zhao, A. V. Virkar, K. Z. Fung, "Electrochemical Characterization and Performance Evaluation of Intermediate Temperature Solid Oxide Fuel

Cell with La0.75Sr0.25CuO2.5-delta Cathode", Journal of Power Sources, 152, [1] 22-6 (2005). 52 C H A P T E R 2

[65] C. Xia, W. Rauch, F. Chen, M. Liu, "Sm0.5Sr0.5CoO3 Cathodes for Low- Temperature SOFCs", Solid State Ionics, 149, [1-2] 11-9 (2002).

[66] M. T. Colomer, B. C. H. Steele, J. A. Kilner, "Structural and Electrochemical

Properties of the Sr0.8Ce0.1Fe0.7Co0.3O3-[delta] Perovskite as Cathode Material for ITSOFCs", Solid State Ionics, 147, [1-2] 41-8 (2002).

[67] R. Radhakrishnan, A. V. Virkar, S. C. Singhal, "Estimation of Charge-Transfer Resistivity of Pt Cathode on YSZ Electrolyte Using Patterned Electrodes", Journal of the Electrochemical Society, 152, [5] A927-A36 (2005).

[68] J. L. Hertz, H. L. Tuller, "Nanocomposite Platimun-Yttria Stanilized Zirconia Electrode and Implications for Micro-SOFC Operation", Journal of the Electrochemical Society, 154, [4] B413-B8 (2007).

[69] S. Carter, A. Selcuk, R. J. Chater, J. Kajda, J. A. Kilner, B. C. H. Steele, "Oxygen Transport in Selected Nonstoichiometric Perovskite-Structure Oxides", Solid State Ionics, 53-56, [Part 1] 597-605 (1992).

[70] H. Hayashi, M. Kanoh, C. J. Quan, H. Inaba, S. Wang, M. Dokiya, H. Tagawa, "Thermal Expansion of Gd-Doped Ceria and Reduced Ceria", Solid State Ionics, 132, [3-4] 227-33 (2000).

[71] S. J. Benson, R. J. Chater, J. A. Kilner, "Oxygen Diffusion and Surface

Exchange in the Mixed Conducting Perovskite La0.6Sr0.4Fe0.8Co0.2O3-d"; pp. 596 - 609 in Proceedings of Ionic and Mixed Conducting Caramics III Edited by T.A. Ramanarayanan, PV 97-24 Electrochemical Society, (1997).

[72] M. C. Brant, T. Matencio, L. Dessemond, R. Z. Domingues, "Electrical and Microstructural Aging of Porous Lanthanum Strontium Manganite/Yttria-Doped Cubic Zirconia Electrodes", Chemistry of Materials, 13, [11] 3954-61 (2001).

[73] M. J. L. Ostergard, C. Clausen, C. Bagger, M. Mogensen, "Manganite-Zirconia Composite Cathodes for SOFC: Influence of Structure and Composition", Electrochimica Acta, 40, [12] 1971 - 81 (1994).

[74] H. Kamata, A. Hosaka, J. Mizusaki, H. Tagawa, "High Temperature

Electrocatalytic Properties of the SOFC Air Electrode La0.8Sr0.2MnO3/YSZ", Solid State Ionics, 106, [3-4] 237-45 (1998). S T A T E – O F – T H E A R T 53

[75] K. Hayashi, M. Hosokawa, T. Yoshida, Y. Ohya, Y. Takahashi, O. Yamamoto,

H. Minoura, "La1-xSrxMnO3-YSZ Composite Film Electrodes Prepared by Metal-Organic Decomposition for Solid Oxide Fuel Cells", Materials Science and Engineering B, 49, [3] 239-42 (1997).

[76] K. Murata, M. Shimotsu, "Yttrium-Substituted (La, Y, Sr) MnO3 as a SOFC Cathode Material", Journal of the Ceramic Society of Japan, 110, [7] 618-21 (2002).

[77] S. Hashimoto, H. Iwahara, "Structural, Thermal and Electrical Properties of

Ce-Doped SrMnO3", Journal of Electroceramics, 4, [1] 225-31 (2000).

[78] G. C. Kostogloudis, C. Ftikos, "Characterization of Nd1-xSrxMnO3 +/-delta SOFC Cathode Materials", Journal of the European Ceramic Society, 19, [4] 497-505 (1999).

[79] G. C. Kostogloudis, N. Vasilakos, C. Ftikos, "Preparation and Characterization

of Pr1-xSrxMnO3+/-[delta] (x = 0, 0.15, 0.3, 0.4, 0.5) as a Potential SOFC Cathode Material Operating at Intermediate Temperatures (500-700[deg]C)", Journal of the European Ceramic Society, 17, [12] 1513-21 (1997).

[80] Y. Sakaki, Y. Takeda, A. Kato, N. Imanishi, O. Yamamoto, M. Hattori, M. Iio, Y.

Esaki, "Ln1-xSrxMnO3 (Ln=Pr, Nd, Sm and Gd) as the Cathode Material for Solid Oxide Fuel Cells", Solid State Ionics, 118, [3-4] 187-94 (1999).

[81] H.-R. Rim, S.-K. Jeung, E. Jung, J.-S. Lee, "Characteristics of Pr1-xMxMnO3 (M = Ca, Sr) as Cathode Material in Solid Oxide Fuel Cells", Materials Chemistry and Physics, 52, [1] 54-9 (1998).

[82] S. Faaland, M. A. Einarsrud, K. Wiik, T. Grande, R. Hoier, "Reactions Between

La1-xCaxMnO3 and CaO-Stabilized ZrO2 - Part II - Diffusion Couples", Journal of Materials Science, 34, [23] 5811-9 (1999).

[83] X. Huang, J. Liu, Z. Lu, W. Liu, L. Pei, T. He, Z. Liu, W. Su, "Properties of

Nonstoichiometric Pr0.6-xSr0.4MnO3 as the Cathodes of SOFCs", Solid State Ionics, 130, [3-4] 195-201 (2000).

[84] X. Huang, L. Pei, Z. Liu, Z. Lu, Y. Sui, Z. Qian, W. Su, "A Study on PrMnO3- Based Perovskite Oxides used in SOFC Cathodes", Journal of Alloys and Compounds, 345, [1-2] 265-70 (2002). 54 C H A P T E R 2

[85] C. Y. Huang, T. J. Huang, "Effect of co Substitution for Mn on Y1-xSrxMnO3 Properties for SOFC Cathode Material", Journal of Materials Science, 37, [21] 4581-7 (2002).

[86] D. Kuscer, M. Hrovat, J. Holc, S. Bernik, D. Kolar, "Electrical and

Microstructural Characteristics of Materials in the LaMnO3+/-[delta]-LaAlO3-

SrMnO3-[delta] System", Journal of Power Sources, 71, [1-2] 195-8 (1998).

[87] I. Riess, M. Godickemeier, L. J. Gauckler, "Characterization of Solid Oxide Fuel Cells based on Solid Electrolytes or Mixed Ionic Electronic Conductors", Solid State Ionics, 90, [1-4] 91-104 (1996).

[88] T. Horita, K. Yamaji, N. Sakai, H. Yokokawa, A. Weber, E. Ivers-Tiffee,

"Stability at La0.6Sr0.4CoO3-d Cathode/La0.8Sr0.2Ga0.8Mg0.2O2.8 Electrolyte Interface under Current Flow for Solid Oxide Fuel Cells", Solid State Ionics, 133, [3-4] 143-52 (2000).

[89] H. Uchida, S. Arisaka, M. Watanabe, "High Performance Electrodes for

Medium-Temperature Solid Oxide Fuel Cells: Activation of La(Sr)CoO3 Cathode with Highly Dispersed Pt Metal Electrocatalysts", Solid State Ionics, 135, [1-4] 347-51 (2000).

[90] T. Inagaki, K. Miura, H. Yoshida, R. Maric, S. Ohara, X. Zhang, K. Mukai, T. Fukui, "High-Performance Electrodes for Reduced Temperature Solid Oxide

Fuel Cells with Doped Lanthanum Gallate Electrolyte: II. La(Sr)CoO3 Cathode", Journal of Power Sources, 86, [1-2] 347-51 (2000).

[91] R. Maric, S. Ohara, T. Fukui, H. Yoshida, M. Nishimura, T. Inagaki, K. Miura, "Solid Oxide Fuel Cells with Doped Lanthanum Gallate Electrolyte and

LaSrCoO3 Cathode, and Ni-Samaria-Doped Ceria Cermet Anode", Journal of the Electrochemical Society, 146, [6] 2006-10 (1999).

[92] A. Petric, P. Huang, F. Tietz, "Evaluation of La-Sr-Co-Fe-O Perovskites for Solid Oxide Fuel Cells and Gas Separation Membranes", Solid State Ionics, 135, [1-4] 719-25 (2000).

[93] L.-W. Tai, M. M. Nasrallah, H. U. Anderson, D. M. Sparlin, S. R. Sehlin,

"Structure and Electrical Properties of La1-xSrxCo1-yFeyO3. Part 1. The System

La0.8Sr0.2Co1-yFeyO3", Solid State Ionics, 76, [3-4] 259-71 (1995). S T A T E – O F – T H E A R T 55

[94] R. Doshi, V. L. Richards, J. D. Carter, X. P. Wang, M. Krumpelt, "Development of Solid-Oxide Fuel Cells that Operate at 500 Degrees C", Journal of the Electrochemical Society, 146, [4] 1273-8 (1999).

[95] S. Wang, T. Kato, S. Nagata, T. Honda, T. Kaneko, N. Iwashita, M. Dokiya,

"Performance of a La0.6Sr0.4Co0.8Fe0.2O3-Ce0.8Gd0.2O1.9-Ag Cathode for Ceria Electrolyte SOFCs", Solid State Ionics, 146, [3-4] 203-10 (2002).

[96] Y. Matsuzaki, I. Yasuda, "Electrochemical Properties of Reduced-Temperature SOFCs with Mixed Ionic-Electronic Conductors in Electrodes and/or Interlayers", Solid State Ionics, 152, 463-8 (2002).

[97] J. Gao, X. Liu, D. Peng, G. Meng, "Electrochemical Behavior of

Ln0.6Sr0.4Co0.2Fe0.8O3-[delta] (Ln=Ce, Gd, Sm, Dy) Materials used as Cathode of IT-SOFC", Catalysis Today, 82, [1-4] 207-11 (2003).

[98] N. Dasgupta, R. Krishnamoorthy, K. T. Jacob, "Crystal Structure, Thermal

Expansion and Electrical Conductivity of Nd0.7Sr0.3Fe1-xCoxO3 (0 <= x <= 0.8)", Materials Science and Engineering B, 90, [3] 278-86 (2002).

[99] L. Qiu, T. Ichikawa, A. Hirano, N. Imanishi, Y. Takeda,

"Ln(1-x)Sr(x)Co(1-y)Fe(y)O(3-delta) (Ln=Pr, Nd, Gd; x=0.2, 0.3) for the Electrodes of Solid Oxide Fuel Cells", Solid State Ionics, 158, [1-2] 55-65 (2003).

[100] W. Chen, T. Wen, H. Nie, R. Zheng, "Study of Ln0.6Sr0.4Co0.8Mn0.2O3-[delta] (Ln=La, Gd, Sm or Nd) as the Cathode Materials for Intermediate Temperature SOFC", Materials Research Bulletin, 38, [8] 1319-28 (2003).

[101] W. X. Chen, H. W. Nie, W. H. Huang, R. Zheng, H. Y. Tu, Z. Y. Lu, T. L. Wen,

"La0.6Sr0.4Co0.8Mn0.2O3-delta Cathode for an Intermediate Temperature SOFC", Journal of Materials Science Letters, 22, [9] 651-3 (2003).

[102] G. C. Kostogloudis, C. Ftikos, A. Ahmad-Khanlou, A. Naoumidis, D. Stover, "Chemical Compatibility of Alternative Perovskite Oxide SOFC Cathodes with Doped Lanthanum Gallate Solid Electrolyte", Solid State Ionics, 134, [1-2] 127-38 (2000).

[103] S. P. Simner, J. F. Bonnett, N. L. Canfield, K. D. Meinhardt, J. P. Shelton, V. L. Sprenkle, J. W. Stevenson, "Development of Lanthanum Ferrite SOFC Cathodes", Journal of Power Sources, 113, [1] 1-10 (2003). 56 C H A P T E R 2

[104] S. P. Simner, J. P. Shelton, M. D. Anderson, J. W. Stevenson, "Interaction

Between La(Sr)FeO3 SOFC Cathode and YSZ Electrolyte", Solid State Ionics, 161, [1-2] 11-8 (2003).

[105] D. Kuscer, J. Holc, M. Hrovat, D. Kolar, "Correlation Between the Defect

Structure, Conductivity and Chemical Stability of La1-ySryFe1-xAlxO3-[delta] Cathodes for SOFC", Journal of the European Ceramic Society, 21, [10-11] 1817-20 (2001).

[106] R. Chiba, F. Yoshimura, Y. Sakurai, "Properties of La1-ySryNi1-xFexO3 as a Cathode Material for a Low-Temperature Operating SOFC", Solid State Ionics, 152-153, 575-82 (2002).

[107] S. P. Simner, J. F. Bonnett, N. L. Canfield, K. D. Meinhardt, V. L. Sprenkle, J. W. Stevenson, "Optimized Lanthanum Ferrite-Based Cathodes for Anode- Supported SOFCs", Electrochemical and Solid-State Letters, 5, [7] A173-A5 (2002).

[108] R. Chiba, F. Yoshimura, Y. Sakurai, "An Investigation of LaNi1-xFexO3 as a Cathode Material for Solid Oxide Fuel Cells", Solid State Ionics, 124, [3-4] 281-8 (1999).

[109] G. W. Coffey, J. Hardy, L. R. Pedersen, P. C. Rieke, E. C. Thomsen, M. Walpole, "Electrochemical Properties of Lanthanum Strontium Aluminum Ferrites for the Oxygen Reduction Reaction", Solid State Ionics, 158, [1-2] 1-9 (2003).

[110] J. Holc, D. Kuscer, M. Hrovat, S. Bernik, D. Kolar, "Electrical and

Microstructural Characterisation of (La0.8Sr0.2)(Fe1-xAlx)O3 and

(La0.8Sr0.2)(Mn1-xAlx)O3 as Possible SOFC Cathode Materials", Solid State Ionics, 95, [3-4] 259-68 (1997).

[111] T. Hibino, A. Hashimoto, M. Suzuki, M. Sano, "A Solid Oxide Fuel Cell Using

Y-Doped BaCeO3 with Pd-Loaded FeO Anode and Ba0.5Pr0.5CoO3 Cathode at Low Temperatures", Journal of the Electrochemical Society, 149, [11] A1503- A8 (2002).

[112] H. Fukunaga, M. Koyama, N. Takahashi, C. Wen, K. Yamada, "Reaction

Model of Dense Sm0.5Sr0.5CoO3 as SOFC Cathode", Solid State Ionics, 132, [3-4] 279-85 (2000). S T A T E – O F – T H E A R T 57

[113] T. Hibino, A. Hashimoto, T. Inoue, J.-i. Tokuno, S.-i. Yoshida, M. Sano, "A Low-Operating-Temperature Solid Oxide Fuel Cell in Hydrocarbon-Air Mixtures", Science, 288, [5473] 2031-3 (2000).

[114] B. E. Bürgler, L. J. Gauckler, "Single Chamber Solid Oxide Fuel Cell with Integrated Current Collectors"; pp. 1405-13 in Proceedings of 6th European Fuel Cell Forum Edited by M. Mogensen, (2004).

[115] M. Koyama, C.-j. Wen, K. Yamada, "La0.6Ba0.4CoO3 as a Cathode Material for

Solid Oxide Fuel Cells Using a BaCeO3 Electrolyte", Journal of the Electrochemical Society, 147, [1] 87-91 (2000).

[116] T. Ishihara, S. Fukui, H. Nishiguchi, Y. Takita, "Mixed Electronic-Oxide Ionic

Conductor of BaCoO3 Doped with La for Cathode of Intermediate- Temperature-Operating Solid Oxide Fuel Cell", Solid State Ionics, 152-153, 609-13 (2002).

[117] J. M. Bae, B. C. H. Steele, "Properties of Pyrochlore Ruthenate Cathodes for Intermediate Temperature Solid Oxide Fuel Cells", Journal of Electroceramics, 3, [1] 37-46 (1999).

[118] H.-C. Yu, K.-Z. Fung, "La1-xSrxCuO2.5-[delta] as New Cathode Materials for Intermediate Temperature Solid Oxide Fuel Cells", Materials Research Bulletin, 38, [2] 231-9 (2003).

[119] E. Boehm, J.-M. Bassat, M. C. Steil, P. Dordor, F. Mauvy, J.-C. Grenier,

"Oxygen Transport Properties of La2Ni1-xCuxO4+[delta] Mixed Conducting Oxides", Solid State Sciences, 5, [7] 973-81 (2003).

[120] C. L. Chang, T. C. Lee, T. J. Huang, "Oxygen Reduction Mechanism and

Performance of Y1Ba2Cu3O7-delta as a Cathode Material in a High-Temperature Solid-Oxide Fuel Cell", Journal of Solid State Electrochemistry, 2, [5] 291-8 (1998).

[121] Z. P. Shao, W. S. Yang, Y. Cong, H. Dong, J. H. Tong, G. X. Xiong, "Investigation of the Permeation Behavior and Stability of a

Ba0.5Sr0.5Co0.8Fe0.2O3-delta Oxygen Membrane", Journal of Membrane Science, 172, [1-2] 177-88 (2000). 58 C H A P T E R 2

[122] M. Gödickemeier, K. Sasaki, L. J. Gauckler, I. Riess, "Electrochemical Characteristics of Cathodes in Solid Oxide Fuel Cells based on Ceria Electrolytes", Journal of the Electrochemical Society, 144, [5] 1635-46 (1997).

[123] T. Horita, K. Yamaji, M. Ishikawa, N. Sakai, H. Yokokawa, T. Kawada, T. Kato,

"Active Sites Imaging for Oxygen Reduction at the La0.9Sr0.1MnO3-x/Yttria- Stabilized Zirconia Interface by Secondary-Ion Mass Spectrometry", Journal of the Electrochemical Society, 145, [9] 3196-202 (1998).

[124] T. Horita, K. Yamaji, N. Sakai, H. Yokokawa, T. Kawada, T. Kato, "Oxygen

Reduction Sites and Diffusion Paths at La0.9Sr0.1MnO3-x/Yttria-Stabilized Zirconia Interface for Different Cathodic Overvoltages by Secondary-Ion Mass Spectrometry", Solid State Ionics, 127, [1-2] 55-65 (2000).

[125] T. Horita, K. Yamaji, N. Sakai, H. Yokokawa, T. Kato, "Oxygen Transport at

the LaMnO3 Film/Yttria-Stabilized Zirconia Interface under Different Cathodic Overpotentials by Secondary Ion Mass Spectrometry", Journal of the Electrochemical Society, 148, [5] J25-J30 (2001).

[126] V. Brichzin, J. Fleig, H.-U. Habermeier, J. Maier, "Geometry Dependence of Cathode Polarization in Solid Oxide Fuel Cells Investigated by Defined Sr-

Doped LaMnO3 Microelectrodes", Electrochemical and Solid-State Letters, 3, [9] 403-6 (2000).

[127] V. Brichzin, J. Fleig, H.-U. Habermeier, G. Cristiani, J. Maier, "The Geometry

Dependence of the Polarization Resistance of Sr-doped LaMnO3 Microelectrodes on Yttria-Stabilized Zirconia", Solid State Ionics, 152-153, 499-507 (2002).

[128] R. A. de Souza, J. A. Kilner, J. F. Walker, "A SIMS Study of Oxygen Tracer

Diffusion and Surface Exchange in La0.8Sr0.2MnO3+[delta]", Materials Letters, 43, [1-2] 43-52 (2000).

[129] F. S. Baumann, J. Fleig, J. Maier, "Microelectrode Impedance Study of SOFC

Cathode Materials: La0.6Sr0.4Co0.8Fe0.2O3-d and YBa2Cu3O7-d"; pp. 1241 in Proceedings of 6th European Solid Oxide Forum Edited by M. Mogensen, (2004). S T A T E – O F – T H E A R T 59

[130] E. Koep, C. Compson, M. Liu, Z. Zhou, "A Photolithographic Process for Investigation of Electrode Reaction Sites in Solid Oxide Fuel Cells", Solid State Ionics, 176, [1-2] 1-8 (2005).

[131] M. Prestat, "Oxygen Reduction in Solid Oxide Fuel Cells at Intermediate Temperatures: State-Space Modelling and Experimental Validation", PhD Thesis ETH Zurich, (2006).

[132] M. Prestat, J.-F. Koenig, L. J. Gauckler, "Oxygen Reduction at Thin Dense

La0.52Sr0.48Co0.18Fe0.8O3-d Electrodes. Part I: reaction Model and Faradaic Impedance", Journal of Electroceramics 18, [1-2] 87-101 (2007).

[133] M. Prestat, S. Korrodi, A. Infortuna, S. Rey-Mermet, P. Muralt, L. J. Gauckler,

"Oxygen Reduction at Thin Dense La0.52Sr0.48Co0.18Fe0.8O3-d Electrodes. Part II: Experimental Assessment of the Reaction kinetics", Journal of Electroceramics 18, [1-2] 111-20 (2007).

[134] B. Yildiz, G. J. La O, Y. Shao-Horn, "Oxygen Reduction Kinetics at Sr-doped

LaMnO3 supported on Yttria Stabilized Zirconia: An Electrochemical Kinetics Modeling Study", submitted to Journal of the Electrochemical Society,

[135] J. Jamnik, J. Maier, "Treatment of the Impedance of Mixed Conductors Equivalent Circuit Model and Explicit Approximate Solutions", Journal of the Electrochemical Society, 146, [11] 4183-8 (1999).

[136] A. Bieberle-Hütter, H. L. Tuller, "Fabrication and Structural Characterization of

Interdigitated Thin Film La1-xSrxCoO3(LSCO) Electrodes", Journal of Electroceramics, 16, 151-7 (2006).

[137] A. Bieberle-Hütter, M. Soegaard, H. L. Tuller, "Electrical and Electrochemical

Characterization of Microstructured Thin Film La1-xSrxCoO3 Electrodes", Solid State Ionics, 177, 1969-75 (2006).

[138] J. D. Sirman, J. A. Lane, J. A. Kilner, "Comparison of Dense and Porous Perovskite cathodes for Solid Oxide Fuel Cells and Ceramic Oxygen Generators"; pp. 57-72 in Proceedings of Ionic and Mixed Conducting Ceramics III Edited by T.A. Ramanarayanan, W.L. Worrel, T.H. L., A.C. Khandkar, M. Mogensen, W. Gopel, PV 97-24 The Electrochemical Society, Pennington NJ, (1997). 60 C H A P T E R 2

[139] F. S. Baumann, J. Fleig, M. Konuma, U. Starke, H.-U. Habermeier, J. Maier,

"Strong Performance Improvement of La0.6Sr0.4Co0.8Fe0.2O3-delta SOFC Cathodes by Electrochemical Activation", Journal of the Electrochemical Society, 152, [10] A2074-A9 (2005).

[140] I. Taniguchi, R. C. van Landschoot, J. Schoonman, "Fabrication of

La1-xSrxCo1-yFeyO3 Thin Films by Electrostatic Spray Deposition", Solid State Ionics, 156, [1-2] 1-13 (2003).

[141] M. L. Liu, D. S. Wang, "Preparation of La1-zSrzCo1-yFeyO3-x Thin-Films, Membranes, and Coatings on Dense and Porous Substrates", Journal of Materials Research, 10, [12] 3210-21 (1995).

[142] C.-Y. Fu, C.-L. Chang, C.-S. Hsu, B.-H. Hwang, "Electrostatic Spray

Deposition of La0.8Sr0.2Co0.2Fe0.8O3 Films", Materials Chemistry and Physics, 91, [1] 28-35 (2005).

[143] C. Argirusis, T. Damjanovic, G. Borchardt, "Preparation of SOFC Cells by Means of Electrophoretic Deposition", Key Engineering Materials, 314, 101-6 (2006).

[144] S. P. Jiang, "A comparison of O2 reduction reactions on porous (La,Sr)MnO3

and (La,Sr)(Co,Fe)O3 electrodes", Solid State Ionics, 146, [1-2] 1-22 (2002).

[145] A. Esquirol, N. P. Brandon, J. A. Kilner, M. Mogensen, "Electrochemical

Characterization of La0.6Sr0.4Co0.2Fe0.8O3 Cathodes for Intermediate- Temperature SOFCs", Journal of the Electrochemical Society, 151, [11] A1847-A55 (2004).

[146] A. Ringuede, J. Fouletier, "Oxygen Reaction on Strontium-Doped Lanthanum Cobaltite Dense Electrodes at Intermediate Temperatures", Solid State Ionics, 139, [3-4] 167-77 (2001).

[147] Y. L. Yang, C. L. Chen, S. Y. Chen, C. W. Chu, A. J. Jacobson, "Impedance Studies of Oxygen Exchange on Dense Thin Film Electrodes of

La0.5Sr0.5CoO3-delta", Journal of the Electrochemical Society, 147, [11] 4001-7 (2000).

[148] A. Endo, H. Fukunaga, C. Wen, K. Yamada, "Cathodic Reaction Mechanism

of Dense La0.6Sr0.4CoO3 and La0.81Sr0.09MnO3 Electrodes for Solid Oxide Fuel Cells", Solid State Ionics, 135, [1-4] 353-8 (2000). S T A T E – O F – T H E A R T 61

[149] T. Kawada, J. Suzuki, M. Sase, A. Kaimai, K. Yashiro, Y. Nigara, J. Mizusaki, K. Kawamura, H. Yugami, "Determination of Oxygen Vacancy Concentration

in a Thin Film of La0.6Sr0.4CoO3-delta by an Electrochemical Method", Journal of the Electrochemical Society, 149, [7] E252-E9 (2002).

[150] S. Diethelm, A. Closset, J. Van Herle, K. Nisancioglu, "Oxygen Transport and

Nonstoichiometry in SrFeO3-delta", Electrochemistry, 68, 444 (2000).

[151] M. S. Islam, "Computer Modelling of Defects and Transport in Perovskite Oxides", Solid State Ionics, 154-155, 75-85 (2002).

[152] M. Cherry, M. S. Islam, C. R. A. Catlow, "Oxygen Ion Migration in Perovskite- Type Oxides", Journal of Solid State Chemistry, 118, [1] 125-32 (1995).

[153] L. Gavrilova, V. A. Cherepanov, "Oxygen Nonstoichiometry and Defect

Structure of La1-xMexCoO3-d"; pp. 404-14 in Proceedings of 6th Internationla Conference on Solid Oxide Fuel Cells (SOFC VI) Edited by S.C. Singhal, M. Dokiya, PV 1999-19 The Electrochemical Society, Pennington, NJ, (1999).

[154] S. J. Ahn, J. H. Lee, J. Kim, J. Moon, "Single-Chamber Solid Oxide Fuel Cell with Micropatterned Interdigitated Electrodes", Electrochemical and Solid State Letters, 9, [5] A228-A31 (2006).

[155] B. E. Buergler, M. Ochsner, S. Vuillemin, L. J. Gauckler, "From Macro- to Micro-Single Chamber Solid Oxide Fuel Cells", submitted to Journal of Power Sources, (2006).

[156] Y. Tang, K. Stanley, Q. M. J. Wu, D. Ghosh, J. J. Zhang, "Design Consideration of the Micro Thin Film Solid-Oxide Fuel Cells", Journal of Micromechanics and Microengineering, 15, [9] S185-S92 (2005).

[157] V. T. Srikar, K. T. Turner, T. Y. Andrew Ie, S. M. Spearing, "Structural Design Considerations for Micromachined Solid-Oxide Fuel Cells", Journal of Power Sources, 125, [1] 62-9 (2004).

[158] J. Fleig, H. L. Tuller, J. Maier, "Electrodes and Electrolytes in Micro-SOFCs: a Discussion of Geometrical Constraints", Solid State Ionics, 174, [1-4] 261-70 (2004). 62 C H A P T E R 2

[159] Jankowski, Jeffrey P. Hayes, R. Tim Graff, J. D. Morse, "Micro-Fabricated Thin-Film Fuel Cells for Portable Power Requirements"; pp. 730 in Proceedings of Materials Research Society Symposium Proceedings (2002).

[160] Z. P. Shao, S. M. Haile, J. Ahn, P. D. Ronney, Z. L. Zhan, S. A. Barnett, "A Thermally Self-Sustained Micro Solid-Oxide Fuel-Cell Stack with High Power Density", Nature, 435, [7043] 795-8 (2005).

[161] C. D. Baertsch, K. F. Jensen, J. L. Hertz, H. L. Tuller, S. T. Vengallatore, S. M. Spearing, M. A. Schmidt, "Fabrication and Structural Characterization of Self- Supporting Electrolyte Membranes for a Micro Solid-Oxide Fuel Cell", Journal of Materials Research, 19, [9] 2604-15 (2004).

[162] P. Bruschi, A. Diligenti, A. Nannini, M. Piotto, "Technology of Integrable Free- Standing Yttria-Stabilized Zirconia Membranes", Thin Solid Films, 346, [1-2] 251-4 (1999).

[163] J. P. Nair, E. Wachtel, I. Lubomirsky, J. Fleig, J. Maier, "Anomalous Expansion

of CeO2 Nanocrystalline Membranes", Advanced Materials, 15, [24] 2077 - 81 (2003).

[164] M. Greenberg, E. Wachtel, I. Lubomirsky, J. Fleig, J. Maier, "Elasticity of Solids with a Large Concentration of Point Defects", Advanced Functional Materials, 16, 48-52 (2006).

[165] A. F. Jankowski, R. T. Graff, J. P. Wayes, J. D. Morse, "Testing of Solid-Oxide Fuel Cells for Micro to Macro Power Generation"; pp. 932-7 in Proceedings of 6th International on Symposium Solid Oxide Fuel Cells (SOFC VI) Edited by S.C. Singhal, M. Dokiya, (1999).

[166] S. Rey-Mermet, P. Muralt, "Metallic Supporting Grid for Thin Electrolyte Membrane in Solid Oxide Fuel Cell", PCT/EP2006/069688, (2006).

[167] S. Kang, P. C. Su, Y. I. Park, Y. Salto, F. B. Prinz, "Thin-Film Solid Oxide Fuel Cells on Porous Nickel Substrates with Multistage Nanohole Array", Journal of the Electrochemical Society, 153, [3] A554-A9 (2006).

[168] X. Chen, N. J. Wu, L. Smith, A. Ignatiev, "Thin-Film Heterostructure Solid Oxide Fuel Cells", Applied Physics Letters, 84, [14] 2700-2 (2004).

3 Aim of the thesis

The aim of this thesis is the realization of thin film cathodes for micro solid oxide fuel cells (µSOFCs) and microstructuring of the supporting substrate. This includes identification and understanding of a suitable process for fabrication of thin film cathodes that is compatible with fabrication of a µSOFC. The cathodes need to be characterized in terms of microstructure and performance in the temperature range applicable for µSOFCs (500 – 600°C). For the support structure suitable materials need to be identified and processed in order to realize a µSOFC. 64

4 Spray Pyrolysis of La0.6Sr0.4Co0.2Fe0.8O3±δ Thin Film

Cathodes

Spray pyrolysis has been used to prepare

La0.6Sr0.4Co0.2Fe0.8O3±δ thin film cathodes for solid oxide fuel cell applications. The films are polycrystalline with nano-meter sized grains and less than 1 µm in thickness. Deposition parameters for film deposition have been established. The ratio of deposition temperature to solvent boiling point is found to be the most important processing parameter that determines whether a crack free homogeneous and coherent film is obtained. The morphology can be tailored by the deposition parameters. Annealing at 650°C for four hours in air results in coherent films of the desired perovskite phase. The films are potential cathodes for thin film micro solid oxide fuel cells.

4.1 Introduction

Spray pyrolysis is a very versatile technique to obtain thin films of various materials and morphologies. Easy control of stoichiometry and simple experimental setup are the main advantages [1-11]. Different processes are available, which are distinguished by the method of atomizing the precursor, namely air pressurized [3, 6, 7, 11-20], electrostatic [1, 2, 4, 6, 7, 9, 18, 21-24] and ultrasonic spray pyrolysis [5, 10, 25, 26]. The method of atomization mainly determines the droplet size of the generated aerosol, which in turn determines the film quality. The film formation is also influenced by the atomization method, electrostatic atomization leads to preferential landing of droplets due to their charge, a phenomenon which is not present in other atomization techniques and thus leads to unique formation mechanisms and

D. Beckel, A. Dubach, A. R. Studart L. J. Gauckler, Journal of Electroceramics, 16, [3] 221-8 (2006). 66 C H A P T E R 4 morphologies [1, 4, 21, 23]. Even though the atomization method plays a crucial role in the spraying process, general trends are expected for some of the processing parameters such as the deposition temperature, regardless of the method of atomization. Thus for some preparation parameters comparison can be done even if the films are prepared by different atomization techniques.

The applications for thin films fabricated by spray pyrolysis are very broad, they are used as barrier layers [10, 26], for semiconductor devices such as solar cells [3, 11, 14-17, 22], sensors [3, 11, 14, 22], or photoactive layers [22]. Electrochromic materials [13, 27], catalytically active thin films [14, 16, 21] and battery components [14, 16, 21, 28] are also fabricated by means of spray pyrolysis. Solid oxide fuel cell (SOFC) electrolytes [6, 7, 18-20, 25], interconnectors [5] and cathodes [4, 9, 12, 23] have also been prepared.

SOFCs are of great interest, because of their potential to convert chemical energy into electrical energy with high efficiency. However, high operating temperatures of 800 – 1000°C are required in state-of-the-art SOFCs, placing heavy demands on the materials used and leading to degradation. Reliability can be increased by lowering the operating temperature, which, however, results in loss of performance. One way to improve performance at low temperatures is to use thin films as components for the SOFC [23, 25, 29], which leads to lower ohmic resistances of the single components. Indeed, SOFCs using very thin electrolytes fabricated by spray pyrolysis showed excellent performance with power densities up to 760 mW/cm2 at 770°C [18]. The different microstructures needed for SOFC electrolytes (dense) and SOFC electrodes (porous) can be obtained using spray pyrolysis, because the large number of parameters involved in the spray process allows to fabricate thin films with very different microstructures. Some of the basic parameters are reported in recent work of Perednis et al. [7, 18, 30, 31]. On the other hand the amount of parameters is the challenge one faces during optimization of the process.

Most authors [1, 3, 5-8, 14, 16, 18-20, 22, 25, 28, 32-34] consider the deposition temperature to be one of the most important parameters, but no general rule for an optimum temperature has been found yet, because it also depends on the investigated materials’, solvents’ systems and spray pyrolysis setup. Furthermore, the literature on spray pyrolysis is also sometimes contradictive. Only limited literature is available on the preparation of La0.6Sr0.4Co0.2Fe0.8O3-δ (LSCF) thin films S P R A Y P Y R O L Y S I S 67 by spray pyrolysis. Taniguchi et al. and Fu et al. [4, 23] used electrostatic spray deposition to prepare LSCF thin films. They observed different morphologies depending on the deposition temperature, composition and liquid-flow rate. The highest porosity was found for the lowest investigated deposition temperature (250°C) [23]. However, porosity was mainly restricted to the surface of the film, the lower part was rather dense. Fu et al. [4] on the other hand observed an increase of porosity with increasing deposition temperature. Very porous layers with fractal structure were obtained. However, it is not reported if they are still coherently enough to exhibit reasonable conductivity. Both authors used different solvent systems, which may explain the contradicting results. To our knowledge, no literature exists on air-pressurized spray pyrolysis of LSCF thin films, which will be the subject of this work. This paper reports on the influence of the preparation parameters on LSCF thin films fabricated with an air-pressurized spray pyrolysis process and suggests a model of the formation of these films. The aim is also to establish guidelines to prepare these films which may be valid also for other systems in order to facilitate fabrication of thin films by spray pyrolysis.

4.2 Experimental

4.2.1 Film Preparation

All films are prepared by air-pressurized spray pyrolysis. The metal salts are dissolved in a solvent mixture. The solution is pumped (peristaltic pump: Ismatec MS Reglo or syringe pump: Razell Scientific Instruments A 99) using a viton tube (Masterflex/Cole-Parmer) through a nozzle (Badger Air-brush Model 150) and atomized by adjustable (EAR 2000 F 02, SMC, and Norgren/IMI) air pressure. The formed droplets are sprayed onto a heated substrate (custom made heating plate with temperature precision of ± 1°C in the deposition area) where an amorphous metal-oxide film is formed, as shown schematically in Fig. 4.1. In an additional annealing step the films can be crystallized (Nabertherm L 3 oven). 68 C H A P T E R 4

sprayspray solutionsolution

pumppump sprayspray gungun pressurizedpressurized air air

substratesubstrate filmfilm

heatingheating plate plate

Fig. 4.1: Sketch of the air pressurized spray pyrolysis setup.

Silicon, sapphire, and glass are used as substrates. The thermal expansion coefficient (TEC) varies from 3.6·10-6/°C for Silicon [35], 8.1·10-6/°C for sapphire [35] to 8.6·10-6/°C for the glass [36]. Despite this difference in TEC, no influence of the substrate material on the film morphology is found. The standard conditions for sample preparations are: At a deposition temperature of the substrate of 255 ± 5°C (measured with a handheld surface probe of type K, Omega Model 88108), the droplets are deposited at an air pressure of 1 bar with a solution flow rate of 30 ml/h, using a nozzle to substrate distance of 20 cm. In general, a spraying time of 60 min is sufficient for a layer thickness of ~700 nm. For the spray solution, a mixture of salts with a molar ratio of LaNO3·6H2O : SrCl2·6H2O : Co(NO3)2·6H2O : Fe(NO3)3·9H2O = 3:2:1:4 (all from Fluka with purity > 98 or 99 %) and a total salt concentration of 0.02 mol/l was dissolved in a solvent composition of 1/3 (volume fraction) ethanol (Scharlau and Merck, purity > 99.5 %, boiling point 78°C) and 2/3 diethylene glycol monobutyl ether (from Fluka and Acros Organic, purity 98 and 99 %, boiling point 231°C). The post deposition annealing conditions are chosen to be 4 hrs at 650°C in air with a heating ramp of 3°C/min. In the results section, preparation details are only given if they deviate from the standard parameters given above. S P R A Y P Y R O L Y S I S 69

4.2.2 Film Characterization

The droplet volume distribution during spray pyrolysis is measured near the nozzle using a laser deflection spectrometer (Sympatec Helos KF). The adhesion of the films to the substrate is checked by sticking a scotch tape to the film and removing it. The films show very good adhesion, because no trace of the film is found on the tape after removing. Morphology is investigated using scanning electron microscopy (SEM, Leo 1530). The roughness of the films is measured with a profilometer (Hommel Tester T1000 LV15).

Fig. 4.2: XRD pattern of an LSCF thin film deposited on Si showing the desired rhombohedral perovskite. The film was annealed at 650°C for 4 hours. References for the same composition

(La0.6Sr0.4Co0.2Fe0.8O3-δ) from Kostogloudis [37] and ten Elshof [38] as well as for the substrate [39] are also shown.

Determination of the crystal phase is done by X-ray diffraction (XRD, Bruker AXS D8 Advance) on the thin films deposited on silicon substrates. The substrate can still be detected and is used as a calibration standard. After annealing of 4 hrs at 650°C in air, crystalline films of the desired rhombohedral perovskite are obtained, as shown in 70 C H A P T E R 4

Fig. 4.2. The hexagonal lattice parameters of a = 5.51 Å and c = 13.51 Å are in good agreement with literature data obtained for a powder of the same stoichiometry: a = 5.51 Å, c = 13.39 Å [38].

4.3 Results and Discussion

4.3.1 Ratio of Deposition Temperature to Solvent Boiling Point

Depositing an LSCF film, using the standard parameters described in the experimental section, results in a coherent crack-free film as shown in Fig. 4.3 A). In this case the deposition temperature (Tdep) is 255°C and the solvent boiling point

(Tsbp) is 180°C, thus the ratio of Tdep/Tsbp equals 1.17 (in K). For these single phase solvent mixtures we simply take a solvent boiling point reflecting the composition of the solvent by Tsbp = 1/3 TbpEthanol (78°C) + 2/3 TbpDiethyleneGlycolMonobutylEther (231°C) = 180°C. Even if equilibrium evaporation data, such as lower or upper boiling point would be available, it would not apply here, due to the fast evaporation during the spray pyrolysis process. Thus we take this easy accessible temperature to establish this empirical guideline which correctly describes our experimental data.

Fig. 4.3 B) shows a film deposited at 225°C. To maintain the same ratio of Tdep/Tsbp, the solvent boiling point is lowered by increasing the ethanol content in the solvent from 1/3 to 1/2. The film is still coherent and crack-free and shows a smoother surface. Fig. 4.3 C) shows a film which is again deposited at the same Tdep/Tsbp, but the absolute deposition temperature is only 195°C. The ethanol content of the solvent is increased even further to 2/3 to keep the ratio Tdep/Tsbp constant by lowering the solvent boiling point to 129°C. Also in this case, the film is coherent and crack free with an even smoother surface. As shown in Fig. 4.3 A) to C), when the ratio Tdep/Tsbp is constant, crack-free coherent films are obtained independently of the absolute deposition temperature. However, the roughness of the films changes with the solvent composition. S P R A Y P Y R O L Y S I S 71

Fig. 4.3: A) to C): LSCF films deposited at TDeposition/TSolventBoilingPoint = 1.16 ± 0.01 (in K), i.e. A) deposition at 255°C, solvent composition 1/3 ethanol, 2/3 diethylene glycol monobutyl ether (DGME); B) 225°C, 1/2 ethanol, 1/2 DGME; C) 195°C, 2/3 ethanol, 1/3 DGME. D) deposition temperature 195°C, i.e. Tdep/Tsbp = 1.02, E) deposition at 320°C i.e. Tdep/Tsbp = 1.31. F) Schematic representation for the correlation between different morphologies, deposition temperature and solvent boiling point.

For a lower Tdep/Tsbp ratio of 1.02, the film shows cracks developing perpendicular to the surface, as exemplified in Fig. 4.3 D) for a deposition temperature of 190°C and the standard solvent composition with 1/3 ethanol. Fig. 4.3 E) on the other hand shows a film deposited at 320°C with the same solvent composition, corresponding to a ratio of Tdep/Tsbp = 1.31. In this case no continuous film is obtained, but irregular deposits which do not cover the substrate completely. In the present work, coherent crack-free films are obtained for 1.15 < Tdep/Tsbp < 1.25, as schematically shown Fig.

4.3 F). Furthermore, faster film deposition is observed for smaller Tdep/Tsbp ratios.

The observation of faster film deposition at lower deposition temperature is widely reported in literature on spray pyrolysis [5, 14, 16]. This observation indicates that at this condition the droplets are still wet [6, 8, 17, 20, 25] when hitting the substrate. This means that at higher temperatures some of the small droplets are already dry and blown away from the film surface and do not contribute to the film formation. Instead of liquid droplets reaching the substrate a chemical vapor deposition (CVD) like process [22] where vapor reactants are responsible for film growth has also been 72 C H A P T E R 4 proposed to explain film formation during spray pyrolysis. However because of the low temperatures (< 300°C) and the reported small evaporation rates during droplets transport [20, 30] this cannot be the case in our experimental setup.

In most papers reporting on spray pyrolysis [1, 3, 5-8, 14, 16, 18-20, 22, 25, 28, 32- 34], the substrate temperature during deposition is considered to be amongst the most critical parameters. However, the reported optimum temperature for crack-free films varies over a very wide range from 80°C [19] to about 500°C [3]. This is because the substrate temperature alone is not sufficient to determine the film quality. The decisive parameters are the drying and decomposition kinetics of the flying droplets and the growing film, which are in turn also determined by the solvent, the liquid and the gas flow rate, the material to be deposited, and the setup geometry. As shown above, by just considering the ratio of deposition temperature

(Tdep) to the solvent boiling point (Tsbp), one already gets an easy and effective “rule” for fabricating good films.

4.3.2 Salt Concentration

Using a spray solution with a high salt concentration (0.04 mol/l) near the solubility limit, results in a rather smooth film as shown in Fig. 4.4 A). If the salt concentration is lower (0.02 mol/l), the film shows a rougher surface, see Fig. 4.4 B). In general, the roughness (Ra) measured for the case of low-salt concentration (0.02 mol/l) increases with deposition time from about Ra = 0.15 μm for a 20 min deposition time to about Ra = 0.25 μm for 90 min. S P R A Y P Y R O L Y S I S 73

A B

10 μm 10 μm

1 μm 1 μm

Fig. 4.4: Effect of different salt concentration. A) Higher salt concentration (0.04 mol/l) and shorter deposition times (30 min) lead to smoother films than in case B) with low salt concentration (0.02 mol/l) and longer deposition times (60 min). Both: deposition temperature 275°C, annealing: 10 h at 1000°C, 2°C/min.

4.3.3 Solution Flow Rate

Spraying with the standard solution flow rate of 30 ml/h results in crack-free coherent films as already shown in Fig. 4.3 A) to C). Increasing the flow rate by a factor of two, results in cracked films. In this case again the residual solvent content in the growing film is too high and leads to differential shrinkage and ultimately to crack formation. But when not varied drastically, the flow rate has a minor importance in the spray process.

Perednis et al. [7] also identified the solution flow rate to have a minor influence on the film morphology, especially as long as a certain limit is not exceeded, which then 74 C H A P T E R 4 would result in crack formation. Only one reference is found attributing a major role to the solution flow rate [12]. The authors report a significant change in microstructure from particle aggregates obtained at low solution flow rates to a more layered structure obtained at high solution flow rates. Unfortunately, no values are given for the two solution flow rates, thus the significance of this observation cannot be quantified.

4.3.4 Air Pressure

A 0.5 bar 1 bar 1.5 bar 2 bar 3 bar B 10 μm 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Droplet Vol. Distribution 0 5 10 15 20 25 Droplet Diameter / Micrometer

C 10 μm D 10 μm )

Fig. 4.5: A) Change of droplet volume distribution with pressure. As-deposited LSCF films deposited with different air pressure: B) 0.5 bar, C) 1.5 bar, D) 3 bar. Total salt concentration in the solution: 0.04 mol/l, spraying time 30 min.

Varying the air pressure results in a change of the droplet volume distribution as shown in Fig. 4.5 A). Raising the pressure leads to a more uniform droplet volume distribution. The films shown so far (e.g. Fig. 4.3) are deposited at the standard pressure of 1 bar and do not show cracks. Fig. 4.5 B) to D) shows a series of films deposited at the standard temperature but at different pressures: The film deposited at a lower pressure of 0.5 bar exhibits many cracks as shown in Fig. 4.5 B). For an S P R A Y P Y R O L Y S I S 75 elevated pressure of 1.5 and 3 bar, as shown in Fig. 4.5 C) and D), no cracks are found. On the contrary, with higher pressure, the film surface looks even smoother.

The change in droplet volume distribution with pressure is also confirmed by literature [6]. The air pressure influences also the drying kinetics during film deposition. Choosing low pressure leads to slower evaporation of the solvent in the droplet and results in a solvent rich film. Upon drying of the film, the differential shrinkage is too high thus cracks are formed. At elevated pressures, evaporation of the droplets is faster, hence the solvent content of the film is lower and cracks are avoided during drying of the film.

4.3.5 Model for Film Formation

Based on the results obtained, we suggest a model to describe the mechanisms and processes that take place during spray pyrolysis of the thin films. As shown schematically in Fig. 4.6, one can distinguish three basic situations. In the first case, shown in Fig. 4.6 A), the ratio of Tdep/Tsbp is low, i.e. in our case < 1.15. Starting from a clear solution, notable evaporation of the solvents occurs only in the last millimeters before the droplets hit the substrate [6, 20]. Due to the slow drying and decomposition kinetics obtained at low Tdep/Tsbp, only few particles precipitate in the droplet. The deposited film is thus a dilute suspension with a high solvent content. Thus the differential shrinkage is so high that many cracks develop during drying as shown in Fig. 4.3 D). 76 C H A P T E R 4

Fig. 4.6: Sketch of film formation, A) low Tdep/Tsbp results in little precipitations, thus a diluted suspension is deposited, cracks form upon drying because of the high solvent content. B)

Medium Tdep/Tsbp results in a concentrated liquid with many precipitations. Depending on the conditions, B1) leads to rough films, B2) to smooth films. C) High Tdep/Tsbp results in dry droplets, which lead to non continuous coverage of the substrate and powdery deposits that do not stick to the substrate.

In the other extreme, shown in Fig. 4.6 C) for Tdep/Tsbp > 1.25, evaporation of the solvent is very fast, most droplets are already dry when reaching the substrate. Thus they do not stick to the substrate surface but are blown away and consequently they do not contribute to the film formation. Only few big droplets that are statistically formed at the nozzle will still reach the substrate in a wet state. However, since there S P R A Y P Y R O L Y S I S 77 are only few of these, they form isolated deposits, which do not cover the substrate continuously as shown in Fig. 4.3 E).

In the case, shown in Fig. 4.6 B), a coherent crack-free film is obtained for

1.15 < Tdep/Tsbp < 1.25. Depending on the exact conditions, the morphology and roughness varies. However, these details will be discussed later after the conditions to obtain coherent films are clarified. In case B, upon evaporation near the substrate, more particles precipitate than in case A. We consider formation of some precipitates in the droplets directly before or during impact on the substrate for the investigated system, because some of the solutions are at the solubility limit. A concentrated liquid containing many precipitates is obtained on the substrate. The residual solvent content is in this case sufficient to provide enough wetting of the substrate. On the other hand the solvent content is small enough to reduce differential shrinkage compared to case A. Thus no cracks develop upon drying or annealing as shown in Fig. 4.3 A) to C).

Now the intermediate range 1.15 < Tdep/Tsbp < 1.25 (case B) is discussed more closely. The first question is how the rough ridges on the film surface are formed (Fig. 4.3 A)), the second question is, why the morphology depends on the solvent composition. One possibility to form these ridges are capillary forces, which drag the solids load of a drying droplet to the edge of the droplet [40]. When comparing the amount of material carried by one droplet with the amount of material incorporated in one of these ridges, it can be calculated that due to the low solubility of the used salts, the material transported by one droplet is simply insufficient to build one of these ridges. Even a droplet which is more than twice as big (10 µm diameter) as the average droplet, carries 100 times less material than is required to build one of the smaller structures with a width and height of 1 µm and 10 µm diameter.

Therefore, we assume a different model: As schematically shown in Fig. 4.6 B1) lateral movement of the droplet’s content occurs upon impact on the substrate. The movement of the precipitates may be caused by the constant impact of new droplets on the substrate or by capillary forces. The important point is that the precipitates can move in the spreading and still wet droplets, until they hit an obstacle during their horizontal movement on the film surface. By this means, the precipitates are constantly accumulated at rough obstacles, leading to the rougher films shown in Fig. 4.3 A). If the concentration of the precipitates is higher as depicted in Fig. 4.6 B2), 78 C H A P T E R 4 the precipitates will obstruct each other and can not move over long distances. Thus, they stay where they randomly hit the substrate, forming smoother films as shown in Fig. 4.3 C).

There are five possibilities to increase the number of precipitates: 1) using more of the worse solvent (ethanol) and less of the better solvent (diethylene glycol monobutylether). Thereby the saturation of salts in the solution is increased, leading to more precipitates. 2) Increasing the salt concentration also leads to smoother films due to more precipitates in the droplet. 3) Increasing the ratio Tdep/Tsbp within the range for continuous crack-free film formation. 4) Decreasing the solution flow rate, thus less droplets share the same thermal energy for evaporation. 5) Increasing the air pressure, leads to faster evaporation of the solvent and thus to more precipitates in the droplets.

We test the rule that coherent crack free films are obtained for 1.15 < Tdep/Tsbp < 1.25 by also applying it to other systems: We can exchange some of the used nitrate salts by chlorides and vice versa and even replace the diethylene glycol monobutyl ether in the solvent by water and still obtain crack-free films. Although in this case the absolute substrate temperature is only 170°C. Furthermore we can also exchange some of the cations, e.g. replacing La by Ba, or spraying LSC only, or

Sm0.5Sr0.5CoO3 without a loss in film quality. For another setup geometry or very different liquid or gas flow rates, one may need to change the absolute numbers of the ratio Tdep/Tsbp, but knowing that this is the most important parameter, already simplifies the process of optimizing the spray pyrolysis parameters.

4.4 Summary

The most critical parameter when preparing thin films by spray pyrolysis is the ratio of deposition temperature to mean solvent boiling point, because it determines the drying and decomposition kinetics of the droplets and the growing film. For the investigated system, a ratio in the range of 1.15 to 1.25 (in K) proved to be useful. By keeping this ratio constant, the absolute deposition temperature could be varied by about 100°C, while still keeping a coherent crack-free film. Furthermore, the solvent, the salts, and some of the cations could also be exchanged without loss of film quality. Higher salt concentrations and the use of poor solvents during preparation of S P R A Y P Y R O L Y S I S 79 the spray solution lead to smoother films. Both parameters lead to stronger precipitation during evaporation of the droplets, which reach the substrate in a wet state. The more precipitates are present in the spreading droplets, the lower is their mobility because they obstruct each other and get stuck. Thus they stay where they randomly hit the substrate and form smooth films. The solution flow rate plays a minor role as long as it does not exceed a certain limit. The air pressure influences the drying kinetics. Higher pressure leads to faster drying and, furthermore, to a more uniform droplet distribution and smoother films. In the post deposition annealing step in air, a crystalline film of the desired perovskite phase is already obtained at 650°C having nano-sized grains.

4.5 References

[1] C. H. Chen, E. M. Kelder, P. J. J. M. van der Put, J. Schoonman, "Morphology

Control of Thin LiCoO2 Films Fabricated using the Electrostatic Spray Deposition (ESD) Technique", Journal of Materials Chemistry, 6, [5] 765-71 (1996).

[2] K. Choy, W. Bai, S. Charojrochkul, B. C. H. Steele, "The Development of Intermediate-Temperature Solid Oxide Fuel Cells for the Next Millennium", Journal of Power Sources, 71, [1-2] 361-9 (1998).

[3] T. Dedova, M. Krunks, O. Volobujeva, I. Oja, "ZnS Thin Films Deposited by Spray Pyrolysis Technique", Physicy Status Solidi C, 2, [3] 1161-6 (2005).

[4] C.-Y. Fu, C.-L. Chang, C.-S. Hsu, B.-H. Hwang, "Electrostatic Spray

Deposition of La0.8Sr0.2Co0.2Fe0.8O3 Films", Materials Chemistry and Physics, 91, [1] 28-35 (2005).

[5] A. Furusaki, H. Konno, R. Furuichi, "Perovskite-Type Lanthanum Chromium- Based Oxide-Films Prepared by Ultrasonic Spray-Pyrolysis", Journal of Materials Science, 30, [11] 2829-34 (1995).

[6] D. Perednis, "Thin Film Deposition by Spray Pyrolysis and the Application in Solid Oxide fuel Cells", PhD Thesis Swiss Federal Institute of Technology, (2003). 80 C H A P T E R 4

[7] D. Perednis, O. Wilhelm, S. E. Pratsinis, L. J. Gauckler, "Morphology and Deposition of Thin Yttria-Stabilized Zirconia Films using Spray Pyrolysis", Thin Solid Films, 474, [1-2] 84-95 (2005).

[8] D. Perednis, L. J. Gauckler, "Thin Film Deposition Using Spray Pyrolysis", Journal of Electroceramics, 14, [2] 103-11 (2005).

[9] A. Princivalle, D. Perednis, R. Neagu, E. Djurado, "Microstructural

Investigations of Nanostructured La(Sr)MnO3-delta Films Deposited by Electrostatic Spray Deposition", Chemistry of Materials, 16, [19] 3733-9 (2004).

[10] O. Stryckmans, T. Segato, P. H. Duvigneaud, "Formation of MgO Films by Ultrasonic Spray Pyrolysis from Beta-Diketonate", Thin Solid Films, 283, [1-2] 17-25 (1996).

[11] M. S. Tomar, F. J. Garcia, "Spray Pyrolysis in Solar Cells and Gas Sensors", Progress in Crystal Growth and Characterization, 4, [3] 221-48 (1981).

[12] P. Charpentier, P. Fragnaud, D. M. Schleich, C. Lunot, E. Gehain, "Preparation of Cathodes for Thin Film SOFCs", Ionics, 3, [1-2] 155-60 (1997).

[13] L. D. Kadam, P. S. Patil, "Studies on Electrochromic Properties of Nickel Oxide Thin Films Prepared by Spray Pyrolysis Technique", Solar Energy Materials and Solar Cells, 69, [4] 361-9 (2001).

[14] J. Morales, L. Sanchez, F. Martin, J. Ramos-Barrado, M. Sanchez, "Use of Low-Temperature Nanostructured CuO Thin Films Deposited by Spray- Pyrolysis in Lithium Cells", Thin Solid Films, 474, [1-2] 133-40 (2005).

[15] B. R. Pamplin, "Spray Pyrolysis of Ternary and Quaternary Solar Cell Materials", Progress in Crystal Growth and Characterization, 1, [4] 395-403 (1979).

[16] P. S. Patil, A. R. Patil, S. H. Mujawar, S. B. Sadale, "Properties of Spray Deposited Niobium Oxide Thin Films", Journal of Materials Science: Materials in Electronics, 16, [1] 35-41 (2005).

[17] S. H. Pawar, P. S. Patil, R. D. Madhale, C. D. Lokhande, "Preparative

Parameters and Dependent Properties of Fe2O3 Films Formed by Spray S P R A Y P Y R O L Y S I S 81

Pyrolysis Technique", Indian Journal of Pure and Applied Physics, 27, [5] 227- 30 (1989).

[18] D. Perednis, L. J. Gauckler, "Solid Oxide Fuel Cells with Electrolytes Prepared via Spray Pyrolysis", Solid State Ionics, 166, [3-4] 229-39 (2004).

[19] T. Setoguchi, M. Sawano, K. Eguchi, H. Arai, "Application of the Stabilized Zirconia Thin-Film Prepared by Spray Pyrolysis Method to SOFC", Solid State Ionics, 40-1, 502-5 (1990).

[20] O. Wilhelm, S. E. Pratsinis, D. Perednis, L. J. Gauckler, "Electrospray and Pressurized Spray Deposition of Yttria-Stabilized Zirconia Films", Thin Solid Films, 479, [1-2] 121-9 (2005).

[21] C. H. Chen, E. M. Kelder, J. Schoonman, "Unique Porous LiCoO2 Thin Layers Prepared by Electrostatic Spray Deposition", Journal of Materials Science, 31, [20] 5437-42 (1996).

[22] K. L. Choy, B. Su, "Growth Behavior and Microstructure of CdS Thin Films Deposited by an Electrostatic Spray Assisted Vapor Deposition (ESAVD) Process", Thin Solid Films, 388, [1-2] 9-14 (2001).

[23] I. Taniguchi, R. C. van Landschoot, J. Schoonman, "Fabrication of

La1-xSrxCo1-yFeyO3 Thin Films by Electrostatic Spray Deposition", Solid State Ionics, 156, [1-2] 1-13 (2003).

[24] I. Taniguchi, R. C. van Landschoot, J. Schoonman, "Electrostatic Spray

Deposition of Gd0.1Ce0.9O1.95 and La0.9Sr0.1Ga0.8Mg0.2O2.87 Thin Films", Solid State Ionics, 160, [3-4] 271-9 (2003).

[25] P. Bohac, L. J. Gauckler, "Chemical Spray Deposition of YSZ and GCO Solid Electrolyte Films", Solid State Ionics, 119, [1-4] 317-21 (1999).

[26] W. M. Sears, M. A. Gee, "Mechanics of Film Formation during the Spray Pyrolysis of Tin Oxide", Thin Solid Films, 165, [1] 265-77 (1988).

[27] L. D. Kadam, S. H. Pawar, P. S. Patil, "Studies on Ionic Intercalation Properties of Cobalt Oxide Thin Films Prepared by Spray Pyrolysis Technique", Materials Chemistry and Physics, 68, [1-3] 280-2 (2001). 82 C H A P T E R 4

[28] P. Fragnaud, R. Nagarajan, D. M. Schleich, D. Vujic, "Thin-Film Cathodes for Secondary Lithium Batteries", Journal of Power Sources, 54, [2] 362-6 (1995).

[29] C. H. Chen, H. J. M. Bouwmeester, H. Kruidhof, J. E. ten Elshof, A. J.

Burggraaf, "Fabrication of La1-xSrxCoO3-delta Thin Layers on Porous Supports by a Polymeric Sol-Gel Process", Journal of Materials Chemistry, 6, [5] 815-9 (1996).

[30] D. Perednis, L. J. Gauckler, "Solid Oxide Fuel Cells with YSZ Films Prepared Using Spray Pyrolysis"; pp. 970-5 in Proceedings of 8th International Symposium on Solid Oxide fuel Cells Edited by S.C. Singhal, M. Dokiya, The Electrochemical Society, (2003).

[31] D. Perednis, M. B. Joerger, K. Honegger, L. J. Gauckler, "Fabrication of Thin YSZ Electrolyte Films Using Spray Pyrolysis Technique"; pp. 989-94 in Proceedings of 7th International Symposium on Solid Oxide Fuel Cells 16 Edited by H. Yokokawa, S.C. Singhal, The Electrochemical Society, (2001).

[32] C. M. Lampkin, "Aerodynamics of Nozzles used in Spray Pyrolysis", Progress in Crystal Growth and Characterization, 1, [4] 405-16 (1979).

[33] G. L. Messing, S. C. Zhang, G. V. Jayanthi, "Ceramic Powder Synthesis by Spray-Pyrolysis", Journal of the American Ceramic Society, 76, [11] 2707-26 (1993).

[34] J. B. Mooney, S. B. Radding, "Spray Pyrolysis Processing", Annual Review of Materials Science, 81 - 101 (1982).

[35] W. M. Yim, R. J. Paff, "Thermal-Expansion of AlN, Sapphire, and Silicon", Journal of Applied Physics, 45, [3] 1456-7 (1974).

[36] Mikroglas, Material Properties of Foturan, www.mikroglas.de, 2005.

[37] G. C. Kostogloudis, C. Ftikos, "Properties of A-site-Deficient

La0.6Sr0.4Co0.2Fe0.8O3-[delta]-Based Perovskite Oxides", Solid State Ionics, 126, [1-2] 143-51 (1999).

[38] J. ten Elshof, J. Boeijsma, "Influence of Iron Content on Cell Parameters of

Rhombohedral La0.6Sr0.4Co1-yFeyO3", Powder Diffraction, 11, [3] 240-5 (1996).

[39] Si, PDF 05-0565, JCPDS Database. S P R A Y P Y R O L Y S I S 83

[40] R. D. Deegan, O. Bakajin, T. F. Dupont, G. Huber, S. R. Nagel, T. A. Witten, "Capillary Flow as the Cause of Ring Stains from Dried Liquid Drops", Nature, 389, [6653] 827-9 (1997).

84 C H A P T E R 4

5 Solid-State Dewetting of La0.6Sr0.4Co0.2Fe0.8O3±δ Thin

Films during Annealing

Porous La0.6Sr0.4Co0.2Fe0.8O3±δ (LSCF) thin film cathodes were fabricated by spray pyrolysis and pulsed laser deposition (PLD). These films show nanometer sized grains and a thickness < 1 µm. They are dense after deposition and develop porosity upon annealing at 600 – 900°C. Sprayed films develop roughly twice the porosity of films deposited by PLD. Solid state dewetting similar to metallic films is proposed as mechanism for the pore formation. The amount of porosity can be tailored by the deposition process, the annealing conditions and the chemistry of the precursor. The connection between porosity and electrical conductivity was also investigated.

5.1 Introduction

Porous thin films are of great importance for many chemical and electrochemical applications such as electronic devices, sensors, catalysts and electrodes. The material investigated in the current work, La0.6Sr0.4Co0.2Fe0.8O3±δ (LSCF), is used as a cathode in solid oxide fuel cells (SOFCs). SOFCs are used for efficient conversion of chemical into electrical energy at high (800 – 1000°C) temperatures. The cathode has to provide a large solid-gas interface for oxygen reduction, thus a porous microstructure is desirable. For large, thick-film-based state-of-the-art fuel cells, cathodes (20 – 50 µm thick) are fabricated by spraying suspensions and pastes, or by screen printing or tape casting [1-5]. The required microstructure is achieved by the use of pore formers that are burned out to create porosity.

D. Beckel, A. Dubach, A. N. Grundy, A. Infortuna L. J. Gauckler, accepted by Journal of the European Ceramic Society, (2007). 86 C H A P T E R 5

In order to reduce material degradation, SOFC research aims to reduce the operating temperature to 500 – 700°C. The increase of resistivity associated with the lowered operating temperature is compensated for by the use of thin films (some 100 nm thick). This way the ohmic resistance of the cell is kept low by geometrical means [6]. Therefore tape casting or screen printing have to be replaced by thin film deposition techniques [7] e.g. spray pyrolysis or pulsed laser deposition (PLD). The use of thin films also enables the fabrication of micro SOFCs [8-10], which leads to new applications for SOFCs.

In this paper we investigate the preparation of porous LSCF thin film cathodes, which were prepared without the aid of a pore former, by spray pyrolysis and PLD and discuss their microstructural evolution during post-deposition annealing. Different means to tailor the porosity are also presented.

5.2 Experimental

The thin films were deposited onto different substrates. Afterwards, they were annealed in air for specific times at specific temperatures and the microstructure and the crystalline phase were analyzed. The details of each step are given in the following sections.

5.2.1 Thin Film Deposition

Air-pressurized spray pyrolysis [11-17] and PLD [18, 19] were used for deposition of the thin films.

5.2.1.1 Spray Pyrolysis

Details of the experimental setup can be found elsewhere [11]. Nitrates and chlorides were used as metal salts to prepare the spray solution. Specifically, La(NO3)3·6H2O

(ABCR purity > 99 %), LaCl3·7H2O (Alfa Aesar), SrCl2·6H2O (Fluka > 99 %), Sr(NO3)2

(Fluka > 99 %), Co(NO3)2·6H2O (Alfa Aesar and Fluka > 98 %), Fe(NO3)3·9H2O

(Fluka > 98 %), FeCl3·6H2O (Fluka > 98 %) were used. As solvent a mixture of 1/3 (vol. fraction) ethanol (Scharlau and Merck, > 99.5 %) and 2/3 diethylene glycol S O L I D – S T A T E D E W E T T I N G 87 monobutyl ether (Acros Organic > 99 %) was used. Unless otherwise noted, nitrates were used for preparation of the spray solutions except for Sr, where the chloride was used. The total salt concentration was 0.02 or 0.04 mol/l. The deposition time was 45 min for 0.04 mol/l salt concentration and twice the time for half the salt concentration. The flow rate of the spray solution was 30 ml/h. These conditions resulted in a film thickness of about 600 nm. For atomization of the solution, an air pressure of 1 bar was used. The obtained droplets were sprayed onto the heated substrate where an amorphous metal oxide film was formed.

The substrate surface temperature was 280°C for sapphire single crystal (Stettler, Lyss, Switzerland ( 1120 ) orientation parallel to the surface) and polycrystalline aluminum oxide (alox) substrates and 255°C for Si substrates ((100) single crystal, Prolog Semicor, Kiev, Ukraine). A thin native oxide layer due to the processing at elevated temperatures is expected. The substrate surface temperature was measured with a type K surface probe (Omega Model 88108). The difference in substrate temperature was to account for the difference in heat transfer from the substrate to the arriving droplet [13]. Si and sapphire substrates were used for X-ray diffraction (XRD) characterization; for conductivity measurements sapphire was used as substrate. Alox substrates were used for microstructure characterization of the sprayed films.

For the conductivity measurements, the thin films were deposited through a shadow mask (200 µm thick laser-cut molybdenum) to achieve a defined film geometry as illustrated in Fig. 5.1 A) – C).

Crystallization was carried out in a separate annealing step in air; details of annealing temperature and time are given in the results and discussion section. 88 C H A P T E R 5

Fig. 5.1: Sketch of sample contacting: A): sapphire substrate, B): spray mask to define film geometry, C): thin film on sapphire, D): mask for sputtering the Pt contacts E). F): Picture of a fully contacted sample with Pt wires attached with Pt paste and ceramic glue to the sputtered Pt and the substrate respectively.

5.2.1.2 Pulsed Laser Deposition

The target for PLD (PLD workstation, Surface, Hueckelhoven, Germany) was uniaxially (7 MPa 100 s) and isostatically (256 MPa, 3 min) pressed from LSCF powder (Praxair, Danbury, CT, USA > 99.9 %) and sintered for 4 hrs at 1250°C in air with a heating rate of 3 °C/min and a cooling rate of 5°C/min. More details on the pressing procedure can be found in [6]. The Si substrate ((100) single crystal with native oxide layer, University Wafers) on which the films were deposited was placed at a distance of 6.5 cm to the target. A 248 nm excimer laser with 4 J/cm2 fluence was used with 0.26 mbar oxygen pressure in the deposition chamber. The substrate temperature was 400°C. Fifty thousand pulses at a rate of 10 Hz were used to form 400 nm thick films.

S O L I D – S T A T E D E W E T T I N G 89

5.2.2 Porosity and Grain Size Characterization

Scanning electron microscope (SEM) images were obtained with a Leo 1530 (Carl Zeiss SMT). Images were taken with the in-lens detector using 5 kV acceleration voltage. The grain size and the porosity were determined from SEM images by quantitative image analysis using the software Lince [20].

5.2.3 X-Ray Diffraction

The crystal phases were analyzed using XRD (Siemens Diffraktometer D5000

Kristalloflex, Cu Kα radiation) on the actual thin films on silicon and sapphire substrates. The reflections of the substrates were used as calibration standard. The step size was 0.01 and the time per step 10 s.

5.2.4 Electrical Conductivity Measurements

The electrical conductivity of the thin films was measured in air using four-point measurement, similar to [21-23]. Generally, the samples were annealed prior to contacting. As electrodes, four Pt (target from Baltec) stripes 1 mm wide each were sputtered through a shadow mask (200 μm thick laser-cut molybdenum, see Fig. 5.1 D) and E) at room temperature for 4 min with 60 mA at 5·10-2 mbar Ar pressure and a working distance of 6 cm using a Baltec SCD 050 sputter coater. Fritless Pt paste (Heraeus C 3605 P) was painted on flat-pressed Pt wires (0.25 mm diameter, Johnson Matthey, thermocouple quality) which were then put on the four sputtered Pt stripes. As depicted in Fig. 5.1 F), the wires were glued to the substrate outside the film using non-conductive ceramic glue (Firag, Ebmatingen, Switzerland, type WH 1500) mixed with some Pt powder (> 99.9 %, Johnson Matthey, < 10 μm) to improve adhesion to the sapphire. The amount of Pt powder added to the ceramic glue was not enough to make the resulting glue conductive. After a drying step (1 h at 120°C in air), the electrodes were spot-welded (Resistronic 3201) to Pt wires that led from the inside of a furnace to the outside where they were connected to a Multimeter (Keithley 2000). Then the samples were heated with this furnace to the target temperature (3°C/min) and cooled again (3°C/min). The data were recorded during the cooling ramp. 90 C H A P T E R 5

5.2.5 Differential Thermal Analysis, Thermogravimetry, Mass Spectroscopy

Experiments to follow the decomposition reactions were carried out on material from scratched off films after spray deposition using differential thermal analysis (DTA) combined with thermogravimetry (TG, both Netzsch STA 449 C Jupiter) coupled to a mass spectrometer (MS, Balzers Thermostar). The atmosphere during investigation was 21 % oxygen and 79 % argon, the heating rate was 10°C/min.

5.3 Results and Discussion

5.3.1 XRD

All films investigated by XRD were deposited by spray pyrolysis onto sapphire and Si. The reason to use two different substrates was to exclude substrate-specific reactions. In Fig. 5.2 A), the XRD scan of an as-deposited LSCF film is shown, no distinct peaks except for the substrate (sapphire) can be found. Fig. 5.2 B) shows the XRD scan for a film deposited on Si and annealed at 650°C for 4 hrs in air, which already results in the formation of only the desired rhombohedral perovskite. The hexagonal lattice parameters of a = 5.51 Å and c = 13.51 Å were calculated according to [24] and are in good agreement with literature data obtained for powder of the same stoichiometry: a= 5.51 Å, c = 13.39 Å [25]. Fig. 5.2 C) shows a film deposited on sapphire and annealed for 10 hrs at 800°C in air. Also this film shows single phase perovskite. Here no peaks from the substrate (sapphire) were found, because the appearance of peaks from single crystals depends on the orientation of the sample relative to the X-ray beam and on the film thickness, which could vary by about ± 30 %. The film shown in Fig. 5.2 D) is deposited on Si and annealed at 1000°C for 10 hrs. In this sample the formation of unidentified secondary phases was detected. The reason for the formation of those secondary phases is not fully understood, since bulk LSCF perovskites are still stable at these temperatures [26, 27]. Reactions with the substrate are possible, but no Si-containing phase could be identified with these peaks. Evaporation of Co and phase separation of the remaining oxides is also possible as it is suggested for LSC thin films at 1000°C [28]. However, in the temperature range for the application of these films (< 800°C) single perovskite phase is obtained and these additional phases are of no consequence. S O L I D – S T A T E D E W E T T I N G 91

Fig. 5.2: XRD patterns of sprayed LSCF thin films. A): as deposited on sapphire, B): deposited on Si, annealed at 650°C for 4 hrs. C): 10 hrs at 800°C on sapphire, D): 10 hrs at 1000°C on Si. At 1000°C secondary phases appear. A reference for the same composition

(La0.6Sr0.4Co0.2Fe0.8O3 ten Elshof [25]) is also shown. 92 C H A P T E R 5

5.3.2 DTA / TG / MS

Mass loss, heat flux and outgassing species were detected using DTA/TG/MS during heating of the film material after spray deposition. The result is shown in Fig. 5.3. From the deposition temperature to about 600°C, the TG curve indicates a mass loss of 18 wt %. Additionally, all significant peaks in the DTA signal as well as the outgassing hydrocarbons and nitrate decomposition products were detected below 600°C. Chlorides could not be detected and possibly remained in the film. Between 600°C and 1000°C the mass loss is only 0.5 wt %.

During investigation of single salts by DTA/TG/MS, no general trend, such that chloride salts generally decompose at lower or higher temperatures than nitrate salts was found.

Fig. 5.3: DTA, TG and MS results of a film scratched off for analysis after deposition by spray pyrolysis.

S O L I D – S T A T E D E W E T T I N G 93

5.3.3 Microstructural Evolution during Annealing

Fig. 5.4: Evolution of grain size A) and porosity B) with annealing time for different temperatures (sprayed films).

To characterize the microstructural evolution, the thin films were heated to the target temperature (600, 700, 800 or 900°C) with a heating rate of 3°C/min. The annealing 94 C H A P T E R 5 time was counted after reaching the target temperature. The samples were removed from the furnace after annealing for the specified time. The microstructure was then analyzed to determine the grain size, the porosity and the pore size. For films fabricated by spray pyrolysis, the result is shown graphically in Fig. 5.4 and numerically in Table 5.1. SEM images showing the microstructure of some samples are shown in Fig. 5.5. For films prepared by PLD the numerical data is given in Table 5.2 and representative microstructures are shown in Fig. 5.6. Fig. 5.7 shows cross section images of the films and reveals that the pores are not restricted to the film surface they are distributed over the entire cross section.

Grain sizes and porosity increases with annealing time for all annealing temperatures. Increasing annealing temperature proves to be more effective than increasing the annealing time to obtain larger grains and higher porosity. Linear fits for the grain size and the porosity vs. logarithmic annealing time were calculated according to:

GS = AGS + BGS·log(t) and Por = APor + BPor·log(t) ( 5.1 )

In these equations, GS is the grain size in nm, AGS, APor, BGS, BPor are constants, t is the annealing time in min and Por is the porosity in %. These formulae were chosen since they give a good fit to the data. It is well known [29] that grain growth laws derived for dense material ([30, 31] and references therein) do not give a reasonable fit for the grain growth in porous material. S O L I D – S T A T E D E W E T T I N G 95

Table 5.1: Numerical data from the graphs in Fig. 5.4. GS = grain size, Por = porosity. The formulae for the fits are GS/nm = AGS + BGS · log(t/min) and Por/% = APor + BPor · log(t/min), respectively with the constants A and B given in the table and t = annealing time. PS = pore size.

Annealing Annealing Por GS/nm A B A B PS/nm temperature time/min /% GS GS Por Por

1 54±17 0 -

50 54±11 1.4 30±10

155 56±16 0 50.50 3.54 -0.33 0.77 - 600°C 350 60±14 1.0 22±2.6

1550 65±14 3.3 31±8.0

601782 79±21 7.8 - - - - 29±10

1 62±41 1.7 33±8.3

40 68±23 2.4 23±3.0 700°C 61.91 5.33 0.17 2.96 120 77±19 5.2 28±10

1240 77±23 11.4 30±11

1 66±52 0 -

20 73±20 6.6 33±17

60 85±18 9.4 31±17 800°C 61.27 12.57 -0.80 6.15 130 102±32 11.7 33±17

270 92±23 13.3 34±14

550 92±22 18.0 43±26

1 80±23 5.0 29±9.3

5 97±32 8.7 36±13

15 109±32 10.0 28±11 900°C 80.70 22.68 4.71 5.31 35 117±37 13.4 41±21

75 122±22 13.2 35± 7

155 132±36 17.7 62±30 96 C H A P T E R 5

The slope (B) of the linear fit for the grain sizes increases from 3.5 to 22.7 when raising the annealing temperature from 600 to 900°C, for the sprayed films and from 12.1 (600°C) to 23.8 (800°C) for the films deposited by PLD. The steeper slopes for the films prepared by PLD are mainly due to the very small grain sizes at the beginning of the annealing. At 600°C annealing temperature, the grain size of the samples prepared by PLD was always smaller than the grain size of the sprayed samples for comparable annealing times. However, at an annealing temperature of 800°C this was only the case for shorter annealing times below ~4 hrs. For longer annealing times the grains in a film prepared by PLD grow faster. The smaller initial grain size of films prepared by PLD might be correlated with the columnar structure of the grains as shown in Fig. 5.7 L) to O). The grain size analysis was done with top view images, so the widths and not the lengths of the columns are represented in the grain size plots. The reason for the stronger grain growth in PLD films than in sprayed films for long annealing times at 800°C is less obvious. It can be noted that it correlates very roughly with the situation when the porosity becomes noticeable in films prepared by PLD. However, since the sprayed films always show higher porosity than the films prepared by PLD, the reason for this behavior is not yet fully clarified. S O L I D – S T A T E D E W E T T I N G 97

Fig. 5.5: SEM images of the evolving microstructure with annealing time for different temperatures (sprayed films): First row: 600°C; A) 1min, B) 1550 min, C) 601782 min. Second row: 700°C; D) 1min, E) 120 min, F) 1240 min. Third row: 800°C; G) 1 min, H) 60 min, I) 550 min. Fourth row: 900°C; J) 1 min, K) 35 min, L) 155 min. 98 C H A P T E R 5

For the porosity of the sprayed films, the slope of the linear fit increases from 0.8 to 6.1 when raising the annealing temperature from 600 to 800°C. The slope for the samples annealed at 900°C is 5.3. For the films prepared by PLD the slope for the linear fit to the porosity evolution increases from 0.8 (600°C) to 1.2 (800°C). Thus the porosity increase (and also the amount of porosity) is comparable for low temperatures for both types of films. However, at higher temperatures sprayed films develop porosity faster and show always higher porosity than films deposited by PLD.

Table 5.2: Numerical data for the microstructural evolution of thin films deposited by PLD.

GS = grain size, Por = porosity. The formulae for the fits are GS/nm = AGS + BGS · log(t/min) and

Por/% = APor + BPor · log(t/min), respectively with the constants A and B given in the table and t = annealing time. PS = pore size.

Annealing Annealing Por GS/nm AGS BGS APor BPor PS/nm temperature time/min /%

1 15±5 0 -

60 35±13 0 - 600°C 14.71 12.12 -0.52 0.82 240 48±17 1.3 29±6.0

2880 55±16 2.9 15±5.0

1 20±5 2.1 9.2±2.6

60 39±11 4.0 15±4.8

240 119±38 4.4 58±17 800°C 18.33 23.83 2.10 1.17 480 144±38 6.3 62±26

1440 147±38 7.6 67±24

2880 156±64 4.7 58±20

With the linear fits, the grain size and porosity can be calculated for fixed annealing times and varying temperatures as shown in Fig. 5.8 for the sprayed films. Arrhenius- type activation energies for grain growth show two activation energies: for temperatures between 600 and 800°C activation energies from 0.08 eV (1 min S O L I D – S T A T E D E W E T T I N G 99 annealing) to 0.19 eV (1000 min annealing) are found. For the temperature range of 800 to 900°C, activation energies from 0.30 eV (10 min annealing) to 0.44 eV (1000 min annealing) are found. The two different activation energies indicate that two different material transport mechanisms are dominant in the two temperature regimes. Activation energies for the increase in porosity decrease from 0.89 eV (10 min annealing) to 0.70 eV (1000 min annealing). Here no difference between the two temperatures regimes is found. All activation energies are detailed in Table 5.3.

Fig. 5.6: SEM images of the evolving microstructure with annealing time for different temperatures (films prepared by PLD): First row: 600°C; A) 1 min, B) 60 min, C) 2880 min. Second row: 800°C; D) 1 min, E) 480 min, F) 2880 min. 100 C H A P T E R 5

Fig. 5.7: SEM cross sections of films deposited by spray pyrolysis A) to K) and PLD L) to O). Films were annealed at different temperatures for varying times. First row: 600°C; A) 1 min, B) 155 min, C) 601782 min. Second row: 700°C; D) 1 min, E) 40 min, F) 120 min. Third row: 800°C; G) 20 min, H) 60 min, I) 550 min. Fourth row: 900°C; J) 1 min, K) 35 min. Fifth row: 600°C; L) 60min, M) 1440 min. Sixth row: 800°C; N) 240 min, O) 480 min. S O L I D – S T A T E D E W E T T I N G 101

In order to check if grain growth and porosity are saturated with long annealing times a sprayed sample was annealed for more than one year at 600°C. Grain growth was still observed and the grain size and porosity are close to what is expected from the linear fit for the first 1550 min, so no saturation is observed. The microstructure of this sample is shown in Fig. 5.5 C) and the data is also included in Table 5.1.

Table 5.3: Activation energies (AE) for grain growth (GS) and porosity (Por) increase. Samples prepared by spray pyrolysis. t = annealing time.

AE, GS/eV AE, Por/eV

600°C ≤ T ≤ 800°C 0.08 0.13 0.16 0.19 0.89 0.75 0.70 800°C ≤ T ≤ 900°C 0.30 0.37 0.41 0.44

t/min 1 10 100 1000 10 100 1000

Generally, the pore size also increases with annealing time and temperature. However the scattering of the data is more pronounced than for the grain growth. This can be attributed to the larger pore size distribution, which is caused by the irregular shape of the pores, e.g. the length of some pore is more than five times their width.

In a previous study for Ce0.8Gd0.2O1.9 (CGO) thin films that were also produced by spray pyrolysis [15] we found limiting grain growth for long annealing times and no porosity being formed. Here, in contrast to these findings, we do not observe a limiting grain growth for LSCF although our study was carried out for the same annealing times and temperatures. However, in contrast to CGO, the LSCF grains are 2 to 5 times larger and, more important, while the CGO films were found to be dense and remained dense during annealing, the originally dense LSCF films developed porosity during annealing parallel to grain growth. 102 C H A P T E R 5

Fig. 5.8: Calculated grain sizes A) and porosity B) for different annealing times and temperatures (sprayed films).

S O L I D – S T A T E D E W E T T I N G 103

5.3.4 Porosity Formation due to Solid-State Dewetting

Annealing of granular materials is usually associated with densification of the microstructure. In contrast to this behavior, formation of porosity during annealing of solid films has been reported. It was found that thin (< 30 nm) Si films on SiO2 substrates show dewetting driven by surface energy reduction, when these films were annealed at 850°C [32]. A similar observation was made for metal thin films (3 –

10 nm) such as Pt on SiO2 substrates [33]. Here, thermally activated dewetting occurred at 620°C due to the high energies of metal surfaces and interfaces between

Si and SiO2 ([33] and references herein).

For dense solid films, dewetting starts with the nucleation of voids on defects. Once holes are nucleated, the growth is driven by capillary forces arising from the balance of surface and interface energy [34-38]. It was demonstrated that also metal oxides such as Fe2O3 [39] and NiO [34] in form of thin films of 80 to 100 nm thickness on Si substrates show dewetting. Nucleation occurred on defects created by ion bombardment.

Similar to these observations, we assume that the developing porosity in the LSCF thin films is also a solid state dewetting phenomenon which is related to the film and substrate materials, the preparation process and the geometry of the film. We assume the film material to be the most important parameter for this phenomenon, since we observed the porosity increase for LSCF films on every substrate tested so far (Si, alox, sapphire, stainless steel) and we observed it for different deposition methods (spray pyrolysis and PLD).

The preparation process however does have an influence on the amount of porosity found in the films. Films grown by PLD developed fewer pores than films grown by spray pyrolysis. The film preparation process is responsible for the amount of defects created in the film, which serve as nucleation sites for the holes. Spray pyrolysis might produce a higher density of such defects than PLD due to the large mass loss during pyrolysis of the precursor as described in section 5.3.2. PLD on the other hand is a solid state process, so no evaporation of solid ceramic material at these temperatures can be expected, thus only the grain boundaries serve as defects.

The geometry might also play a role. The films are constrained on rigid substrates, thus they cannot shrink laterally. Literature describes this phenomenon as 104 C H A P T E R 5 constrained sintering of thin films [40-45]. During sintering the thin film cannot densify isotropically due to the constraint, so the initial pores grow further and densification is hindered. Since not all films of the same geometry show pore development (LSCF shows porosity, CGO does not), the fixture to the substrate during annealing can only support and possibly enhance pore formation but it cannot be the main cause.

We also briefly want to discuss other mechanisms that can lead to porosity in thin films; however, we do not think that they apply here. One possibility to create porous films is through outgassing material. This can be excluded here, since outgassing is only found below the annealing temperatures and times where we find porosity as shown in Fig. 5.3. Similarly we can exclude a notable density change in the films. The density of the films cannot be measured by the Archimedes method since they are only about 500 nm thick and the supporting substrate is at least 600 times thicker. This means the mass of the sample is always dominated by the mass of the substrate. Consequently, the density of the amorphous films is not accessible. However, the crystallographic density can be determined for the annealed samples. Here we do not find a change in density, between dense samples (annealed for 1 min at 600°C) and porous samples (annealed at 900°C). If a distinct change in density resulting in porosity formation had occurred in the transition from amorphous to crystalline we would expect to see the porosity already in the samples annealed for short time at 600°C (where we see first peaks in the XRD), which is not the case. Porosity due to phase separation or reaction with the substrate material can also be ruled out since we find single-phase material in the temperature regime where we observe the porosity and no reaction with the substrate could be observed, see Fig. 5.2. Dewetting in the liquid state during processing of the films by spray pyrolysis is also not relevant, since porosity should then be found in the as deposited films prior to annealing, which was not the case. Furthermore, we observe the porosity also in the films fabricated by PLD where no liquid processing step is included.

5.3.5 Influence of the Precursor Salt

The amount of porosity can be tailored to some extent by the type of salt used for the precursor. The samples shown in Fig. 5.5 and the sample shown in Fig. 5.9 A) and C) were prepared as defined in the experimental section using nitrate salts except for S O L I D – S T A T E D E W E T T I N G 105

Sr where chloride was used (“nitrate” sample). As shown in Fig. 5.9 A) and C) a porosity of 19 % is observed after annealing for 10 hrs at 1000°C. The sample shown in Fig. 5.9 B) and D) of comparable thickness was prepared using different salts, i.e. chloride salts except for Co where nitrate was used (“chloride” sample). The annealing conditions were the same as for both samples. However, in the case of the “chloride” sample, the film is denser; the porosity as determined by SEM is only around 7 %.

Fig. 5.9: LSCF films prepared using A) La, Co, Fe nitrates and Sr chloride, B) La, Sr, Fe chlorides and Co nitrate. C) cross section of A; D) cross section of B).

We attribute the appearance of denser films with the use of more chloride salts to an enhanced densification in the presence of chlorides, as it is known for sintering of calcium hydroxyapatites [46] in the temperature range of our study, although for temperature higher than 900°C, chlorides lowered the densification [46, 47].

Indications that the type of salt influences the porosity are also found for LiCoO2 films. In this case acetates led to higher porosity than nitrates [48].

We also address the possibility that different decomposition temperatures of the different types of salts might cause the observed porosity as reported by Princivalle et al. [49], who observed that salts with higher decomposition temperatures led to 106 C H A P T E R 5 denser films than those with lower decomposition temperatures. During decomposition of single salts in DTA/TG/MS experiments, we did not find any convincing connection between decomposition temperatures of single salts and film microstructures. During spray pyrolysis the salts will form complexes in the solution and also during the early stage of film formation. The decomposition of these complexes is more important than the decomposition of the single salts as was already shown for CdS [50] and MgO [51].

5.3.6 Electrical Conductivity

The microstructure of the films also affected their electrical conductivity. The electrical conductivity of samples annealed at different temperatures and prepared from different precursors is shown in Fig. 5.10. The activation energy for the electrical conductivity calculated between 350 to 600°C and the porosity of the samples are also shown in this figure. The activation energies of 0.05 eV (sample annealed at 600°C) to 0.15 eV (samples annealed at 1000°C) are close to the 0.10 eV found in literature [26] for dense bulk samples between 100 and 500°C.

The sample annealed at the lowest temperature (600°C, 4 hrs) showed the least porosity (2 %) and consequently the highest conductivity. When the porosity increased from 2 to 14 % (sample annealed at 600 and 800°C for 4 hrs, respectively), the conductivity measured at 600°C decreased from 3170 S/m to 2400 S/m. This is in agreement with equations previously derived for the relation between porosity and conductivity. For a two phase material, consisting of an electrically conductive (LSCF) and a non conductive phase (pores), the effective electrical conductivity σeff is given by [52]:

σeff = 3/2·σLSCF·(1-por-1/dim) ( 5.2 )

Where por is the fraction of porosity in the sample and dim the number of dimensions covered by the conductive phase. Since the film thickness is about five to ten times the grain size, we would consider the conductive phase to be three dimensional. For two samples with different porosity the ratio of the effective electrical conductivity according to equation ( 5.2 ) is given by: S O L I D – S T A T E D E W E T T I N G 107

σeff,1 / σeff,2 = [1-por1-(1/dim)] / [1-por2-(1/dim)] ( 5.3 )

According to equation ( 5.3 ), the sample with 14 % porosity should have 82% of the conductivity achieved by the sample with 2 % porosity. Comparing the values given above, we find that it exhibits 76 % of the conductivity. Considering the error bars shown in Fig. 5.10 the actual results are represented by equation ( 5.3 ). The error in determining the conductivity is mainly due to a non homogenous thickness (see also Fig. 5.7) of the samples in the measured area (5 by 14 mm).

Fig. 5.10: Electrical conductivity of LSCF films deposited on sapphire annealed for 4 hrs at 600°C and 800°C and for 10 hrs at 1000°C, the less porous film annealed at 1000°C was prepared using mostly chloride salts. The activation energy for the electrical conductivity calculated between 350 and 600°C and the porosity of the specimens is also shown.

The specimens annealed for 10 hrs at 1000°C showed the lowest conductivity. One of these samples was prepared according to the standard parameters given in the experimental section (“nitrate” sample) and one (“chloride” sample) was prepared using mainly chloride salts as precursors except for Co where nitrate was used. The “chloride” sample had fewer pores (7 % porosity) and consequently shows a higher conductivity than the “nitrate” sample (19 % porosity) when both were annealed at the same temperature. The conductivity of the samples annealed at 1000°C 108 C H A P T E R 5 additionally suffered from the formation of the secondary phases that are detected by XRD analysis. Consequently the conductivity can not be predicted by equation ( 5.3 ) which predicts that the samples should show 92 % (“chloride” sample) and 74 % (“nitrate” sample) of the conductivity of the sample annealed at 600°C. However, their conductivity is only 17 and 10 % of the sample annealed at 600°C.

The fact that even for the films exhibiting low porosity the conductivity is lower than for dense bulk samples [53, 54] is attributed to the smaller grains and thus more grain boundaries in the thin films and is also confirmed by earlier results [55, 56] for similar perovskites with small grain sized microstructures.

5.4 Summary and Conclusion

The preparation of dense LSCF thin films of 100 to 500 nm thickness with nanosized grains by spray pyrolysis and PLD on different substrates has been shown. These films show solid-state dewetting with pore formation at temperatures higher than 600°C. At 1000°C decomposition of this perovskite occurred. The amount of porosity and pore size can be tailored by the annealing temperature and time. The amount of defects in the films probably controls the nucleation of pores. The number of defects is determined by the thin film deposition process. Spray pyrolysis leads to more porosity than PLD for the same annealing conditions. The constraint of the thin film by the rigid substrate during annealing supports pore growth and hinders densification. Further tailoring of the porosity can be achieved by choosing precursors that either support densification (chlorides) or not (e.g. nitrates). The films show semiconductivity in the temperature range from room temperature to 900°C with a transition to metallic behavior for higher temperatures. The conductivities are one order of magnitude lower than for micron grain sized samples due to the small grain size. Porosity further reduces the conductivity.

The presented results allow engineering of the porosity of LSCF thin films, which allows fabrication of porous thin film cathodes suitable for micro SOFC.

S O L I D – S T A T E D E W E T T I N G 109

5.5 References

[1] K. Barthel, S. Rambert, S. Siegmann, "Microstructure and Polarization Resistance of Thermally Sprayed Composite Cathodes for Solid Oxide Fuel Cell Use", Journal of Thermal Spray Technology, 9, [3] 343-7 (2000).

[2] S. P. S. Badwal, K. Foger, "Solid Oxide Electrolyte Fuel Cell Review", Ceramics International, 22, [3] 257-65 (1996).

[3] L. J. Gauckler, D. Beckel, B. Buergler, E. Jud, U. P. Muecke, M. Prestat, J. Rupp, J. Richter, "Solid Oxide Fuel Cells: Systems and Materials", Chimia, 58, [12] 837-50 (2004).

[4] N. Q. Minh, "Ceramic Fuel-Cells", Journal of the American Ceramic Society, 76, [3] 563-88 (1993).

[5] J. van Herle, R. Ihringer, R. V. Cavieres, L. Constantin, O. Bucheli, "Anode Supported Solid Oxide Fuel Cells with Screen-Printed Cathodes", Journal of the European Ceramic Society, 21, [10-11] 1855-9 (2001).

[6] D. Beckel, U. P. Muecke, T. Gyger, G. Florey, A. Infortuna, L. J. Gauckler, "Electrochemical Performance of LSCF Based Thin Film Cathodes Prepared by Spray Pyrolysis", Solid State Ionics, 178, 407-15 (2007).

[7] D. Beckel, A. Bieberle-Hütter, A. Harvey, A. Infortuna, U. P. Muecke, M. Prestat, J. L. M. Rupp, L. J. Gauckler, "Thin Films for Micro Solid Oxide Fuel Cells", Journal of Power Sources doi:10.1016/j.jpowsour.2007.04.070, (2007).

[8] H. Huang, M. Nakamura, P. Su, R. Fasching, Y. Saito, F. B. Prinz, "High- Performance Ultrathin Solid Oxide Fuel Cells for Low-Temperature Operation", Journal of the Electrochemical Society, 154, [1] B20-B4 (2007).

[9] A. Bieberle-Hütter, D. Beckel, U. P. Muecke, J. L. M. Rupp, A. Infortuna, L. J. Gauckler, "Micro-Solid Oxide Fuel Cells as Battery Replacements", Mstnews, 04/05, 12-5 (2005).

[10] D. Nikbin, "Micro SOFCs: Why Small is Beautiful", The Fuel Cell Review, April/May, 21-4 (2006). 110 C H A P T E R 5

[11] D. Beckel, A. Dubach, A. R. Studart, L. J. Gauckler, "Spray Pyrolysis of

La0.6Sr0.4Co0.2Fe0.8O3-d Thin Film Cathodes", Journal of Electroceramics, 16, [3] 221-8 (2006).

[12] D. Beckel, D. Briand, A. R. Studart, N. F. de Rooij, L. J. Gauckler, "Topography Mediated Patterning of Inorganic Materials by Spray Pyrolysis", Advanced Materials, 18, [22] 3015-8 (2006).

[13] U. P. Muecke, G. L. Messing, L. J. Gauckler, "The Leidenfrost Effect During

Spray Pyrolysis of Dense NiO-Ce0.8Gd0.2O1.9-x Thin Films", Submitted to Thin Solid Films, (2006).

[14] U. P. Muecke, N. Luechinger, L. J. Gauckler, "Initial Status of Deposition and Film Formation During Spray Pyrolysis of Nickel Oxide, Cerium Gadolinium Oxide and NiO-CGO Thin Films", Submitted to Thin Solid Films, (2006).

[15] J. L. M. Rupp, A. Infortuna, L. J. Gauckler, "Microstrain and Self-Limited Grain Growth in Nanocrystalline Ceria Ceramics", Acta Materialia, 54, [7] 1721-30 (2006).

[16] D. Perednis, L. J. Gauckler, "Thin Film Deposition Using Spray Pyrolysis", Journal of Electroceramics, 14, [2] 103-11 (2005).

[17] D. Perednis, L. J. Gauckler, "Solid Oxide Fuel Cells with Electrolytes Prepared via Spray Pyrolysis", Solid State Ionics, 166, [3-4] 229-39 (2004).

[18] P. R. Willmott, "Deposition of Complex Multielemental Thin Films", Progress in Surface Science, 76, [6-8] 163-217 (2004).

[19] P. R. Willmott, J. R. Huber, "Pulsed Laser Vaporization and Deposition", Review of Modern Physics, 72, [1] 315-28 (2000).

[20] S. L. dos Santos e Lucato, Lince - Linear Intercept v2.4, Department of Material Science, Darmstadt University of Technology, Darmstadt, Germany (1999).

[21] D. Beckel, D. Briand, A. Bieberle-Hütter, J. Courbat, N. F. De Rooij, L. J. Gauckler, "Micro-Hotplates - a Platform for Micro-Solid Oxide Fuel Cells", Journal of Power Sources, 166, [1] 143-8 (2007). S O L I D – S T A T E D E W E T T I N G 111

[22] U. P. Muecke, S. Graf, U. Rhyner, L. J. Gauckler, "Microstructure and Electrical Conductivity of Nanocrystalline Ni-and NiO-CGO Thin Films", Submitted to Acta Materialia, (2007).

[23] J. L. M. Rupp, L. J. Gauckler, "Microstructures and Electrical Conductivity of Nanocrystalline Ceria based Thin Films", Solid State Ionics, 177, [26-32] 2513- 8 (2006).

[24] L. V. Azároff, M. J. Buerger, "The Powder Method in X-ray Crystallography" McGraw-Hill Book Company, New York, Toronto, London p. 46 (1958).

[25] J. ten Elshof, J. Boeijsma, "Influence of Iron Content on Cell Parameters of

Rhombohedral La0.6Sr0.4Co1-yFeyO3", Powder Diffraction, 11, [3] 240-5 (1996).

[26] L.-W. Tai, M. M. Nasrallah, H. U. Anderson, D. M. Sparlin, S. R. Sehlin,

"Structure and Electrical Properties of La1-xSrxCo1-yFeyO3. Part 2. The System

La1-xSrxCo0.2Fe0.8O3", Solid State Ionics, 76, [3-4] 273-83 (1995).

[27] L.-W. Tai, M. M. Nasrallah, H. U. Anderson, D. M. Sparlin, S. R. Sehlin,

"Structure and Electrical Properties of La1-xSrxCo1-yFeyO3. Part 1. The System

La0.8Sr0.2Co1-yFeyO3", Solid State Ionics, 76, [3-4] 259-71 (1995).

[28] H. D. Bhatt, R. Vedula, S. B. Desu, G. C. Fralick, "La(1-x)SrxCoO3 for Thin Film Thermocouple Applications", Thin Solid Films, 350, 249-57 (1999).

[29] F. F. Lange, B. J. Kellett, "Thermodynamics of Densification: II, Grain Growth in Porous Compacts and Relation to Densification", Journal of the American Ceramic Society, 72, [5] 735-41 (1989).

[30] E. Jud, C. B. Huwiler, L. J. Gauckler, "Grain Growth of Micron-Sized Grains in

undoped and CoO-doped Ce0.8Gd0.2O1.9", Journal of the Ceramic Society of Japan, 114, [1335] 963-9 (2006).

[31] P.-L. Chen, I.-W. Chen, "Grain Growth in CeO2: Dopant Effects, Defect Mechanism, and Solute Drag", Journal of the American Ceramic Society, 79, [7] 1793-800 (1996).

[32] D. T. Danielson, D. K. Sparacin, J. Michel, L. C. Kimerling, "Surface-Energy- Driven Dewetting Theory of Silicon-on-Insulator Agglomeration", Journal of Applied Physics, 100, [8] 083507-10 (2006). 112 C H A P T E R 5

[33] X. Hu, D. G. Cahill, R. S. Averback, "Dewetting and Nanopattern Formation of

Thin Pt Films on SiO2 Induced by Ion Beam Irradiation", Journal of Applied Physics, 89, [12] 7777-83 (2001).

[34] T. Bolse, A. Elsanousi, H. Paulus, W. Bolse, "Dewetting of Nickel Oxide-Films on Silicon Under Swift Heavy Ion Irradiation", Nuclear Instruments and Methods in Physics Research Section B, 244, [1] 115-9 (2006).

[35] M. Bouville, S. Hu, L.-Q. Chen, D. Chi, D. J. Srolovitz, "Phase-Field Model for Grain Boundary Grooving in Multi-Component Thin Films", Modelling and Simulation in Materials Science and Engineering, 14, 433-43 (2006).

[36] M. A. Grinfeld, P. M. Hazzledine, "Rearrangement at Coherent Interfaces in Heterogeneous Solids", Philosophical Magazine Letters, 74, [1] 17-23 (1996).

[37] E. Jiran, C. V. Thompson, "Capillary Instabilities in Thin, Continuous Films", Thin Solid Films, 208, 23-8 (1992).

[38] D. J. Srolovitz, S. A. Safran, "Capillary Instabilities in Thin Films. I. Energetics", Journal of Applied Physics, 60, [1] 247-54 (1986).

[39] T. Bolse, H. Paulus, W. Bolse, "Swift Heavy Ion Induced Dewetting of Metal Oxide Thin Films on Silicon", Nuclear Instruments and Methods in Physics Research Section B, 245, [1] 264-8 (2006).

[40] R. K. Bordia, R. Raj, "Sintering Behavior of Ceramic Films Constrained by a Rigid Substrate", Journal of the American Ceramic Society, 68, [6] 287-92 (1985).

[41] J. N. Calata, A. Matthys, G. Q. Lu, "Constrained-Film Sintering of Cordierite Glass-Ceramic on Silicon Substrate", Journal of Materials Research, 13, [8] 2334-41 (1998).

[42] T. J. Garino, H. K. Bowen, "Deposition and Sintering of Particle Films on a Rigid Substrate", Journal of the American Ceramic Society, 70, [11] C315-C7 (1987).

[43] T. J. Garino, H. K. Bowen, "Kinetics of Constrained-Film Sintering", Journal of the American Ceramic Society, 73, [2] 251-7 (1990). S O L I D – S T A T E D E W E T T I N G 113

[44] P. Letullier, J. M. Heintz, "Elaboration and Sintering Behavior of a Laminar Ceramic-Ceramic Material", Journal De Physique Iv, 3, [C7] 1471-5 (1993).

[45] M. Stech, P. Reynders, J. Rodel, "Constrained Film Sintering of

Nanocrystalline TiO2", Journal of the American Ceramic Society, 83, [8] 1889- 96 (2000).

[46] A. Nzihou, B. Adhikari, R. Pfeffer, "Effect of Metal Chlorides on the Sintering and Densification of Hydroxyapatite Adsorbent", Industrial & Engineering Chemistry Research, 44, [6] 1787-94 (2005).

[47] R. M. Smith, X. D. Zhou, W. Huebner, H. U. Anderson, "Novel Yttrium- Stabilized Zirconia Polymeric Precursor for the Fabrication of Thin Films", Journal of Materials Research, 19, [9] 2708-13 (2004).

[48] C. H. Chen, E. M. Kelder, J. Schoonman, "Unique Porous LiCoO2 Thin Layers Prepared by Electrostatic Spray Deposition", Journal of Materials Science, 31, [20] 5437-42 (1996).

[49] A. Princivalle, D. Perednis, R. Neagu, E. Djurado, "Microstructural

Investigations of Nanostructured La(Sr)MnO3-delta Films Deposited by Electrostatic Spray Deposition", Chemistry of Materials, 16, [19] 3733-9 (2004).

[50] M. Krunks, J. Madarasz, L. Hiltunen, R. Mannonen, E. Mellikov, L. Niinisto, "Structure and Thermal Behaviour of Dichlorobis(thiourea)cadmium(II), a Single-Source Precursor for CdS thin films", Acta Chemica Scandinavica, 51, [3] 294-301 (1997).

[51] O. Stryckmans, T. Segato, P. H. Duvigneaud, "Formation of MgO Films by Ultrasonic Spray Pyrolysis from Beta-Diketonate", Thin Solid Films, 283, [1-2] 17-25 (1996).

[52] D. Stroud, "The Effective Medium Approximation: Some Recent Developments", Superlattices and Microstructures, 23, [3-4] 567-73 (1998).

[53] G. C. Kostogloudis, C. Ftikos, "Properties of A-site-Deficient

La0.6Sr0.4Co0.2Fe0.8O3-[delta]-Based Perovskite Oxides", Solid State Ionics, 126, [1-2] 143-51 (1999).

[54] H. Ullmann, N. Trofimenko, F. Tietz, D. Stover, A. Ahmad-Khanlou, "Correlation Between Thermal Expansion and Oxide Ion Transport in Mixed 114 C H A P T E R 5

Conducting Perovskite-Type Oxides for SOFC Cathodes", Solid State Ionics, 138, [1-2] 79-90 (2000).

[55] X. Chen, N. J. Wu, D. L. Ritums, A. Ignatiev, "Pulsed Laser Deposition of Conducting Porous La-Sr-Co-O Films", Thin Solid Films, 342, [1-2] 61-6 (1999).

[56] A. Furusaki, H. Konno, R. Furuichi, "Perovskite-Type Lanthanum Chromium- Based Oxide-Films Prepared by Ultrasonic Spray-Pyrolysis", Journal of Materials Science, 30, [11] 2829-34 (1995).

6 Electrochemical Performance of LSCF Based Thin Film

Cathodes Prepared by Spray Pyrolysis

La0.6Sr0.4Co0.2Fe0.8O3 (LSCF) based thin film cathodes were fabricated by spray pyrolysis and their performance was evaluated with area specific resistance (ASR). With a maximum processing temperature of 650°C, these cathodes are suitable for application in micro solid oxide fuel cells (μSOFC). The ~500 nm thick cathodes showed performance similar to traditional thick film LSCF cathodes (10 – 100 µm). However, by

using a new material composition Ba0.25La0.25Sr0.5Co0.8Fe0.2O3-δ (BLSCF) or by modifying the microstructure, a significant improvement in performance was achieved. Reducing the grain size or introducing a thin dense cathode layer between the porous cathode and the electrolyte were beneficial modifications of the microstructure.

6.1 Introduction

In the search for efficient generation of electricity, solid oxide fuel cells (SOFCs) have become very interesting due to their ability to convert chemical energy directly into electrical energy and due to their ability to use a variety of fuels. However, high operating temperatures of 800 – 1000°C place heavy demands on the material components and accelerate fuel cell degradation. To increase SOFC lifetime and to reduce system costs, the reduction of operating temperature has been proposed [1]. However, strategies need to be found to compensate for the loss in performance associated with a lowering of the operating temperature. One way is to use thin films as SOFC components [2-5], which is effective because ohmic resistances of the

D. Beckel, U. P. Muecke, T. Gyger, G. Florey, A. Infortuna L. J. Gauckler, Solid State Ionics, 178, 407-15 (2007). 116 C H A P T E R 6 single components are reduced. Thin film based components can easily be integrated into μSOFCs [6, 7].

In the field of cathodes, research on thin films has been directed mostly towards model electrodes with defined geometry and/or defined triple phase boundaries, thus dense thin films were investigated [8-27]. These investigations were necessary to reveal the mechanisms of cathode reactions, however, no performance data can be evaluated that apply to a cathode in SOFC operation conditions, since “real” cathodes are porous in contrast to the model electrodes. This study is directed towards the evaluation of porous LSCF thin film cathodes for the application in μSOFCs in order to fill this gap.

Spray pyrolysis was chosen for the preparation of thin film cathodes since it is a versatile technique for thin film fabrication of various materials and morphologies. The main advantages of the technique are its easy control of stoichiometry and simple experimental setup [2, 28-32].

LSCF was chosen as the cathode material since it shows good performance at intermediate temperatures [1, 33, 34] due to its mixed ionic and electronic conductivity. Although several authors report the preparation of thin film LSCF intended for use as a cathode [4, 29, 35, 36], only one very recent publication was found to report on the cathode performance of thin film LSCF [37]. The authors investigated the influence of one preparation parameter (voltage of the electrostatic- assisted ultrasonic spray pyrolysis) on the cathode performance. The present work will investigate the performance of LSCF thin film cathodes prepared by a simple air- pressurized spray pyrolysis technique. Material variations as well as variations in the cathode preparation and microstructure are investigated for their influence on the cathode performance.

6.2 Experimental

The overall samples preparation can be described as follows: Thin film cathodes were deposited onto thick electrolyte pellets and then annealed to 650°C. Next, the cathodes were contacted using Pt paste and grids and placed in a test rig. The test rig is a furnace equipped with wires to contact the cathode to an impedance E L E C T R O C H E M I C A L P E R F O R M A N C E 117 spectrometer to determine the area specific resistance (ASR). A detailed description of each process step is given in the following sections.

6.2.1 Electrolyte Preparation

Ce0.8Gd0.2O1.9 (CGO) powder (Praxair, purity >99.99%, d50 = 0.5 μm) was used as starting material. Eight grams were uniaxially pressed with 7 MPa for 100 s in a mold with 30 mm diameter. Afterwards the pellet was protected against oil contamination by placing it in a rubber tube and isostatically pressed (Paul Weber Maschinen und Apparatebau) with 256 MPa for 3 min. Sintering was done for 4 h at 1600°C in air with a heating rate of 3°C/min and a cooling rate of 5°C/min. The pellets were ground and polished on both sides to obtain parallel surfaces. The finest diamond polishing suspension (Struers DiaPro Nap B) had 1 μm particle size. For alignment purposes an edge of 1 cm length was ground in the pellet, as shown in Fig. 6.1 A). The final electrolyte pellet had a thickness of 2 mm and a diameter of 23.3 mm.

6.2.2 Thin Film Cathode Deposition

Air-pressurized spray pyrolysis [31, 38-43] and pulsed laser deposition (PLD) [44, 45] were used for deposition of the thin film cathodes. 118 C H A P T E R 6

Fig. 6.1: Sample preparation A): Electrolyte pellet with alignment edge, cathode and reference electrode on top. B): Alignment of shadow mask during spray pyrolysis. C): Sketch and SEM images of a cross section of a symmetrical cell with reference electrodes, showing the electrolyte pellet with the cathode and the contacts (sputtered Pt, Pt paste and Pt mesh) on top. D): Example of impedance spectra for single electrode characterization and for symmetrical cell measurements, respectively. E L E C T R O C H E M I C A L P E R F O R M A N C E 119

6.2.2.1 Spray Pyrolysis

Nitrates and chlorides were used as metal salts to prepare the spray solution.

Specifically, La(NO3)3·6H2O (ABCR > 99 %), LaCl3·7H2O (Alfa Aesar), SrCl2·6H2O

(Fluka > 99 %), Sr(NO3)2 (Fluka > 99 %), Co(NO3)2·6H2O (Alfa Aesar and Fluka

> 98 %), Fe(NO3)3·9H2O (Fluka > 98 %), FeCl3·6H2O (Fluka > 98 %), Ba(NO3)2

(Riedel de Haen), Ce(NO3)3·6H2O (Alfa Aesar > 99.5 %), and Gd(NO3)3·6H2O (Alfa Aesar > 99.9 %) were used. As solvents a mixture of 1/3 (vol. fraction) ethanol (Scharlau and Merk, > 99.5 %) and 2/3 diethylene glycol monobutyl ether (Acros Organic > 99 %) was used for LSCF and CGO. For the Ba-containing films, deionized H2O was used instead of diethylene glycol monobutyl ether due to the poor solubility of the Ba salt in the latter. Unless otherwise noted, nitrates were used for preparation of the spray solutions except for Sr, where the chloride was used. The total salt concentration was 0.02 or 0.04 mol/l. For some experiments 30.6 mg carbon black (Degussa Printex 90 14 nm) were dispersed in 1 liter spray solution using ultrasound. The carbon black served as a pore former in the resulting suspension.

The air-pressurized spray pyrolysis setup is described in detail elsewhere [38]. The flow rate of the spray solution or the suspension containing the pore former was 30 ml/h. In order to keep the suspension dispersed, it was continuously stirred during the deposition. For atomization of the solution or suspension, an air pressure of 1 bar was used. The droplets were sprayed onto each side of the heated electrolyte pellet in turn through a shadow mask (laser cut from 0.1 mm-thick Mo sheet) to form an amorphous metal oxide film, shown in Fig. 6.1 A) and B). The surface temperature of the electrolyte pellet was measured with a type K surface probe (Omega Model 88108) and adjusted to 270°C. For a water-containing solvent, the deposition temperature was 170°C to compensate for the lower boiling point of water compared to diethylene glycol monobutyl ether [38]. After 45 min, a film thickness of ~600 nm was deposited for 0.04 mol/l salt concentration of the spray solution. For a 0.02 mol/l salt concentration, the deposition time was doubled, to achieve the same thickness. The shadow mask was aligned by placing the pellet to a right angle tool in a way that the edge touches one side of the right angle tool, (see Fig. 6.1 B)). The second side of the pellet was prepared in the same way, turning the shadow mask to obtain a symmetric electrode configuration with an approximate maximum alignment 120 C H A P T E R 6 mismatch of 250 μm. The geometry of the cathode and reference electrode obtained by the shadow mask can be seen in Fig. 6.1 A). The geometry and especially the alignment mismatch of the electrodes are important since they can influence the obtained results [46, 47]. Generally correct results are obtained for an alignment mismatch of electrode displacement/electrolyte thickness ≤ 0.1 [47] and a distance of reference electrode to active electrode > 3 times the electrolyte thickness [46]. Both requirements are fulfilled with the geometry used in this work.

The LSCF/CGO composite cathode was prepared by spraying CGO and LSCF layers consecutively. CGO layers were sprayed for 20, 15, 10 and 5 min, with deposition of LSCF in between for 5, 10, 15 and 40 min. For deposition times < 15 min complete coverage was not achieved, thus intertwining CGO and LSCF material is expected.

6.2.2.2 Pulsed Laser Deposition

The target for PLD (PLD workstation, Surface) was prepared the same way as the electrolyte pellet (section 6.2.1) except that LSCF powder (Praxair > 99.9 %) was used instead of CGO and the sintering conditions were 4 h at 1250°C in air with a heating rate of 3 °C/min and a cooling rate of 5°C/min. The electrolyte pellet on which the films were deposited was placed behind a shadow mask (as for spray pyrolysis). The target-to-substrate distance was 6.5 cm. A 248 nm excimer laser with 4 J/cm2 fluence was used with 200 mtorr oxygen partial pressure in the deposition chamber. The substrate temperature was 400°C to form dense films and room temperature for porous films. Five thousand pulses at a rate of 10 Hz were used for ~50 nm thick films and 20000 pulses for ~200 nm thick films.

6.2.3 Annealing

If not noted otherwise, annealing of the thin film cathodes was done for 4 hrs in air at 650°C with a heating and cooling rate of 3°C/min prior to contacting.

E L E C T R O C H E M I C A L P E R F O R M A N C E 121

6.2.4 Contacting

Fifty to 100 nm of Pt was sputtered on top of the cathodes and reference electrodes using a shadow mask of the same geometry and thickness as that used for film deposition. The Pt (Baltec) was sputtered at room temperature for 4 min with 60 mA at 5·10-2 mbar Ar pressure at a working distance of 6 cm using a Baltec SCD 050 sputter coater. Approximately 4 μm fritless Pt paste (Heraeus C 3605 P) was painted with a brush on top of the sputtered Pt. The sputtered Pt improved the adhesion between the Pt paste and the cathode. Uniaxial pressing was used to flatten Pt wires (0.25 mm diameter Johnson Matthey thermocouple quality) which were afterwards spot-welded (Resistronic 3201) to flattened Pt meshes (52 mesh, woven from 0.1 mm diameter wire, Johnson Matthey > 99.9 %). These meshes were placed in the Pt paste and the wires were attached to the electrolyte pellet outside the cathode area using non-conductive ceramic glue (Firag WH 1500). The reference electrode was contacted the same way, except that no Pt mesh was used. The contacting is depicted in Fig. 6.1 C).

To measure the temperature of the pellet, a self-made type S thermocouple was glued to the pellet using the ceramic glue. To fabricate this thermocouple, a Pt wire and a Pt/Rh (90/10 wt %) wire (0.35 mm diameter Johnson Matthey) were welded together in an acetylene/oxygen flame.

Before connecting the cell to the test rig, the cell was placed in a heating chamber for one hour at 120°C to dry the Pt paste and ceramic glue. Afterwards, the wires of the sample were spot-welded to the corresponding wires of the test rig, which consist of Pt wires except for one side of the type S thermocouple which is a Pt/Rh wire. The current and voltage leads to the main electrodes were twisted in pairs to reduce inductive coupling. An additional type K thermocouple (MTS Messtechnik Schaffhausen) was used to monitor the air temperature inside the furnace. The cell was heated with 3°C/min to 650°C in air and remained there for one hour in order to burn in the Pt paste. Then the first measurement was taken at the same temperature. The cell was then cooled with 3°C/min to the next measurement temperature (~580°C). Further measurements were done at ~520°C and ~450°C using the same cooling rate between the measurements.

122 C H A P T E R 6

6.2.5 Measurement Conditions

The cathode impedances were measured at open circuit voltage (OCV) by applying a sinusoidal voltage with an amplitude of ± 10 mV, and measuring the corresponding current using a Solartron SI 1260 impedance analyzer together with a Solartron SI 1287 Electrochemical Interface. The measurements were started at a frequency of 1 MHz and measured to about 0.01 Hz.

Both cathodes on one electrolyte pellet were always prepared identically, thus half of the polarization resistance corresponded to one cathode (see symmetrical cell configuration in Fig. 6.1 D)). As indicated by the arrow in Fig. 6.1 D), the low frequency arc corresponds to the cathode polarization resistance, which is given by the section of the real axis between the intercepts of the cathode arc [37]. When comparing samples with identically prepared electrolyte pellets and different cathodes, the high frequency arc does not change, thus it can be attributed to the electrolyte. Furthermore, the resistance obtained from the high frequency arc corresponds well to the properties (conductivity and activation energy of the conductivity) of the electrolyte material. In addition, the summit frequencies of the electrode arc (Fig. 6.1 D)) correspond well to the ones reported in the literature [48] for electrodes. The assignment of low frequency arc to electrode and high frequency arc to electrolyte is also reported in literature [49]. The ASR was obtained from the impedance spectra by multiplying the cathode polarization resistance by the cathode area.

For all ASR values the symmetrical cell configuration was used, only to check if both cathodes performed similarly, single cathodes were characterized using the reference electrodes (see single electrode characterization in Fig. 6.1 D)). To visualize the evolution of the impedance spectra with temperature Fig. 6.2 displays impedance spectra measured at different temperatures in the symmetrical cell configuration for a sprayed LSCF cathode. E L E C T R O C H E M I C A L P E R F O R M A N C E 123

Fig. 6.2: Impedance spectra at different temperatures for a sprayed LSCF cathode measured in the symmetrical cell configuration.

6.2.6 Microstructure Characterization

Scanning electron microscope (SEM) images were obtained with a Leo 1530 (Carl Zeiss SMT). Images were taken with the in-lens detector using 5 kV acceleration voltage. The grain size was determined by quantitative image analysis using the software Lince [50].

6.2.7 X-Ray Characterization

The crystal phases were analyzed using X-ray diffraction (XRD, Siemens

Diffraktometer D5000 Kristalloflex, CuKα radiation) on the actual thin films on silicon substrates ((100) single crystal, Prolog Semicor). The reflection of the substrate was used as calibration standard. The step size was 0.01 and the time/step 10 s. 124 C H A P T E R 6

6.3 Results and Discussion

6.3.1 Verification of the ASR Measurements

To check the reproducibility of the preparation, contacting and measurement process, three cells were prepared identically and the results were compared. The ASR of all three cells differed by ± 10 %, thus this was taken as the relative error between the samples. In Figs. 6.3 – 6.7 the vertical size of the symbols corresponds to this error.

Fig. 6.3: ASR of a cell with thin film LSCF cathode and without cathode, i.e. only the Pt current collector acts as cathode. For comparison thin [37] and thick film LSCF [34] cathodes from literature are also shown.

To verify if the thin film cathode is actually characterized or if the Pt current collector acts as a cathode, a cell was prepared as described in the experimental section except that no cathode was applied – only the current collector. In Fig. 6.3, the result is compared to the sprayed LSCF thin film cathode, which always serves as a standard (labeled STD in all figures) to which the results obtained by changes in the material or in the preparation process are compared. As can be seen, the ASR of the E L E C T R O C H E M I C A L P E R F O R M A N C E 125 current collector is more than three orders of magnitude higher than the ASR of the cathode. This might be associated with the lower diffusivity of oxygen in Pt than in perovskites [51].

As already noted in literature [37, 52], a comparison of the obtained ASR to previously found literature results is difficult, since there is a huge scatter of data. If one compares the ASR measured for LSCF at 650°C, values from ~0.47 Ω·cm2 [53] ~0.8 Ω·cm2 [34], ~1.1 Ω·cm2 [49], ~1.2 Ω·cm2 [54], ~2.9 Ω·cm2 [37], to ~6.2 Ω·cm2 [55] are reported. The reason for this variation obtained with cathodes of the same stoichiometry is related to the complexity of preparation procedures. Cathodes are prepared by different methods, annealing is done at different temperatures, and consequently the microstructure is different. Furthermore, it has been suggested that foreign phases related to impurities in the precursors and laboratory environment are responsible for varying results [56]. These effects become more important with smaller sample sizes. Nevertheless, we compare our data in Fig. 6.3 to a thick film [34] as well as to the only available data of a thin film cathode [37] made of the same material. The ASR of the thick film cathode is close to the ASR of our thin film cathode, whereas the thin film cathode of Hsu and Hwang shows a higher ASR especially at high temperatures. The difference might be attributed to the coarser microstructure of their thin film cathode.

6.3.2 Influence of the Annealing Temperature

The annealing temperature of the cathodes proved to have a strong influence on the cathode performance. As shown in Fig. 6.4, when the cathode is annealed at 800°C for 4 hrs the ASR is almost twice as high compared to the standard cathode annealed at 650°C for 4 hrs. The reason for this decrease in performance is the coarser microstructure obtained at the higher annealing temperature. Samples annealed at 800°C show almost twice the grain size (124 ± 21 nm) of those annealed at 650°C (65 ± 15 nm), thus the available surface area and the triple phase boundary at the electrolyte interface decreases. 126 C H A P T E R 6

Fig. 6.4: Comparison of two cathodes, which were annealed for 4 hrs at 650°C and 800°C, respectively.

6.3.3 Influence of the Cathode Material

Besides LSCF, an LSCF/CGO composite, Ba0.5Sr0.5Co0.8Fe0.2O3-δ (BSCF) and

Ba0.25La0.25Sr0.5Co0.8Fe0.2O3-δ (BLSCF) were deposited by spray pyrolysis and tested as cathode materials.

The LSCF/CGO composite showed better performances than the standard LSCF especially at lower temperatures (see Fig. 6.5). A change in activation energy from 1.55 eV (LSCF) to 1.27 eV (LSCF/CGO) is also observed for the thick film composite cathode. Dusastre and Kilner [49] observed a change in activation energy from 1.40 – 1.54 eV for LSCF to 0.95 – 1.31 eV for LSCF/CGO composites depending on the LSCF to CGO ratio. A similar trend is also reported for composite cathode of

LSCF/Ce0.7Bi0.3O2 [55]. The reason for this improved performance at lower temperature is given by the oxygen transport properties of LSCF and CGO. At high temperatures (800°C), the oxygen self-diffusion coefficient D of LSCF/CGO and CGO are the same [57] and close to the one of LSCF. However, the ionic conductivity in LSCF decreases strongly with decreasing temperature due to the high activation E L E C T R O C H E M I C A L P E R F O R M A N C E 127 energy [49]. At lower temperatures (T < 600°C), the contribution of the ionic conductivity to the total conductivity of LSCF is low [52], thus improvement of the cathode performance by CGO addition is achieved due to a higher diffusion coefficient of the composite material [57]. Furthermore, the oxygen surface exchange coefficient k of CGO is enhanced by two orders of magnitude in the presence of LSCF on the CGO surface [58, 59], thus faster oxygen transfer and consequently lower ASR are obtained when using composite cathodes.

Fig. 6.5: ASR of cathodes, which consist of different materials: LSCF, an LSCF/CGO composite, BSCF and BLSCF.

Results from a BSCF thin film cathode are also shown in Fig. 6.5, and it performs better than the standard LSCF cathode. However, the performance still does not reach what is reported in the literature for thick film BSCF [60]. X-ray diffraction (see Fig. 6.6 A)) and compare with the reference therein [61]) of the thin film BSCF showed that the correct crystalline phase did not form under the annealing conditions (4 hrs at 650°C). Indeed, annealing at temperatures exceeding 950°C is reported in literature for thick film BSCF cathodes [60, 62-64]. Since the radius of the Ba2+ cation (1.61 nm [65]) is larger than that of other A-site cations (La3+ 1.36 nm [65], Sr2+ 1.44 nm [65]) greater distortion of the lattice is expected and the perovskite formation 128 C H A P T E R 6 might be more difficult than in the case of LSCF, where the perovskite phase is obtained under the same annealing conditions and confirmed by XRD (Fig. 6.6 B)) which also contains a reference for the same stoichiometry [66]). To decrease the lattice distortion, a cathode was prepared where only half the La was replaced by Ba. With the resulting BLSCF, a significant improvement in cathode performance was achieved in comparison to LSCF and the thin film BSCF. By introducing the Ba into the LSCF system the microstructure also changes drastically: The grain size of BLSCF is only about half the grain size of LSCF although they are processed and annealed the same way. The reason for the smaller grains might be associated with the number of different cations. More different sized cations will lead to less dense packaging and thus show more resistance towards crystallization and grain growth. An XRD scan of a BLSCF thin film is also shown in Fig. 6.6 C), however the phase is not yet clear.

Fig. 6.6: XRD scans of different thin film cathodes: A) BSCF, B) LSCF and C) BLSCF prepared by spray pyrolysis and D) LSCF prepared by PLD). References for the same stoichiometry of BSCF (Shao [61]) and LSCF (ten Elshof [66]) are also given. E L E C T R O C H E M I C A L P E R F O R M A N C E 129

6.3.4 Influence of the Preparation Process

Different modifications of the preparation process were evaluated. Changes in the precursor of the sprayed films, by either using more chloride salts instead of nitrates or replacing all chlorides by nitrates showed only a minor effect on the cathode performance, barely exceeding the experimental error.

An LSCF thin film cathode fabricated by PLD instead of spray pyrolysis shows a four times higher ASR, as depicted in Fig. 6.7. Furthermore, the microstructure (Fig. 6.8 A)) is different compared to the sprayed cathode. The columnar microstructure of the PLD film hinders in-plane conduction in the cathode and thus increases the resistance. It was confirmed by XRD (see Fig. 6.6 D)) that the thin film prepared by PLD has the same phase as the one prepared by spray pyrolysis.

Fig. 6.7: Effect of different process modifications on ASR: cathode preparation by PLD, by spray pyrolysis, with the addition of pore former and a double layer cathode with a dense bottom layer.

The addition of carbon black as a pore former to the spray solution resulted in a better performing cathode (Fig. 6.7). Compared to a sprayed cathode without pore former (Fig. 6.8 B)), the porosity is increased (Fig. 6.8 C)), leading to a larger surface 130 C H A P T E R 6 area. However, in contrast to the cathode prepared by PLD the pores are not aligned to columns, thus the in-plane conductivity is not hindered. Thus not only the porosity itself but also the pore shape is important for the cathode performance. Irregular shaped and random distributed pores (as in the sprayed films) are better than columnar pores (as in the PLD films). Although the amount of LSCF deposited was the same for cathodes with and without pore former, the cathode deposited with pore former is thicker, due to the larger volume of the pores. This will also contribute to a reduced polarization resistance, since more area is available for cathode reaction. It can be expected that the entire cathode takes part in the reaction for the thicknesses investigated.

Fig. 6.8: Microstructures of the differently prepared cathodes: A) cathode prepared by PLD, B) cathode prepared by spray pyrolysis, C) sprayed cathode with the addition of pore former, D) double layer cathode comprising a thin dense cathode layer prepared by PLD and a sprayed cathode layer on top. For an easier distinction between cathode and contacts, a sketch of the individual layers is added beside every SEM image.

A possibility to modify the cathode–electrolyte interface is the introduction of a dense bottom layer of the cathode material (double layer cathode [67, 68]). For the thin film cathode presented here, a thin (~50 nm Fig. 6.8 D)) dense layer was introduced by PLD prior to spray pyrolysis deposition. For LSCF thick film cathodes (several µm E L E C T R O C H E M I C A L P E R F O R M A N C E 131 thickness) an improvement of the cathode performance by a factor of 2 – 3 is reported upon introduction of a thinner (~1 μm), dense LSCF layer on the electrolyte [67, 68]. In our case, the total thickness (~600 nm) of the cathode is already below the thickness of the dense layer in the thick film cathode, but the ratio of the grain size of the porous cathode layer to the thickness of the dense layer is the same (~1) and we also observe a reduction in ASR by a factor of 2 – 3 (Fig. 6.7). This reduction is associated with increasing the effective contact area at the cathode/electrolyte interface [68]. Additionally, rapid diffusion of O2- ions via grain boundaries or A-site deficiency in the thin dense LSCF layer has also been proposed [68] to account for this improvement.

6.4 Conclusions

Thin film cathodes suitable for application in µSOFCs have been fabricated by a simple spray technique and evaluated with ASR. It was possible to fabricate these cathodes with a maximum processing temperature of 650°C, which is desirable for the processes and materials involved in µSOFCs. It was found that the microstructure in combination with the choice of material is the most critical point for performance of the thin film cathode. Smaller grain sizes led to better performance and can be achieved by low annealing temperature.

The best performance was achieved with the new material composition BLSCF which exhibits a very fine microstructure (grain size 35 ± 7 nm). Introduction of a thin dense cathode layer between the porous cathode and the electrolyte was also very effective. Increase in the porosity or fabrication of a composite electrode could also improve the performance but not to the same extent. The microstructure achieved by spray pyrolysis proved to be better for the cathode performance than the one achieved by PLD.

6.5 References

[1] J. P. P. Huijsmans, F. P. F. van Berkel, G. M. Christie, "Intermediate Temperature SOFC - a Promise for the 21st Century", Journal of Power Sources, 71, [1-2] 107-10 (1998). 132 C H A P T E R 6

[2] C. H. Chen, H. J. M. Bouwmeester, H. Kruidhof, J. E. ten Elshof, A. J.

Burggraaf, "Fabrication of La1-xSrxCoO3-delta Thin Layers on Porous Supports by a Polymeric Sol-Gel Process", Journal of Materials Chemistry, 6, [5] 815-9 (1996).

[3] P. Bohac, L. J. Gauckler, "Chemical Spray Deposition of YSZ and GCO Solid Electrolyte Films", Solid State Ionics, 119, [1-4] 317-21 (1999).

[4] I. Taniguchi, R. C. van Landschoot, J. Schoonman, "Fabrication of

La1-xSrxCo1-yFeyO3 Thin Films by Electrostatic Spray Deposition", Solid State Ionics, 156, [1-2] 1-13 (2003).

[5] A. Weber, E. Ivers-Tiffee, "Materials and Concepts for Solid Oxide Fuel Cells (SOFCs) in Stationary and Mobile Applications", Journal of Power Sources, 127, [1-2] 273-83 (2004).

[6] X. Chen, N. J. Wu, L. Smith, A. Ignatiev, "Thin-Film Heterostructure Solid Oxide Fuel Cells", Applied Physics Letters, 84, [14] 2700-2 (2004).

[7] A. Bieberle-Hütter, D. Beckel, U. P. Muecke, J. L. M. Rupp, A. Infortuna, L. J. Gauckler, "Micro-Solid Oxide Fuel Cells as Battery Replacements", Mstnews, 04/05, 12-5 (2005).

[8] T. Ioroi, T. Hara, Y. Uchimoto, Z. Ogumi, Z.-I. Takehara, "Preparation of

Perovskite-Type La1-xSrxMnO3 Films by Vapor-Phase Processes and Their Electrochemical Properties", Journal of the Electrochemical Society, 145, [6] 1999-2004 (1998).

[9] T. Horita, K. Yamaji, M. Ishikawa, N. Sakai, H. Yokokawa, T. Kawada, T. Kato,

"Active Sites Imaging for Oxygen Reduction at the La0.9Sr0.1MnO3-x/Yttria- Stabilized Zirconia Interface by Secondary-Ion Mass Spectrometry", Journal of the Electrochemical Society, 145, [9] 3196-202 (1998).

[10] H. Fukunaga, M. Koyama, N. Takahashi, C. Wen, K. Yamada, "Reaction

Model of Dense Sm0.5Sr0.5CoO3 as SOFC Cathode", Solid State Ionics, 132, [3-4] 279-85 (2000).

[11] T. Horita, K. Yamaji, N. Sakai, H. Yokokawa, T. Kawada, T. Kato, "Oxygen

Reduction Sites and Diffusion Paths at La0.9Sr0.1MnO3-x/Yttria-Stabilized E L E C T R O C H E M I C A L P E R F O R M A N C E 133

Zirconia Interface for Different Cathodic Overvoltages by Secondary-Ion Mass Spectrometry", Solid State Ionics, 127, [1-2] 55-65 (2000).

[12] V. Brichzin, J. Fleig, H.-U. Habermeier, J. Maier, "Geometry Dependence of Cathode Polarization in Solid Oxide Fuel Cells Investigated by Defined Sr-

Doped LaMnO3 Microelectrodes", Electrochemical and Solid-State Letters, 3, [9] 403-6 (2000).

[13] Y. L. Yang, C. L. Chen, S. Y. Chen, C. W. Chu, A. J. Jacobson, "Impedance Studies of Oxygen Exchange on Dense Thin Film Electrodes of

La0.5Sr0.5CoO3-delta", Journal of the Electrochemical Society, 147, [11] 4001-7 (2000).

[14] A. Endo, H. Fukunaga, C. Wen, K. Yamada, "Cathodic Reaction Mechanism

of Dense La0.6Sr0.4CoO3 and La0.81Sr0.09MnO3 Electrodes for Solid Oxide Fuel Cells", Solid State Ionics, 135, [1-4] 353-8 (2000).

[15] A. Ringuede, J. Fouletier, "Oxygen Reaction on Strontium-Doped Lanthanum Cobaltite Dense Electrodes at Intermediate Temperatures", Solid State Ionics, 139, [3-4] 167-77 (2001).

[16] Y. L. Yang, A. J. Jacobson, C. L. Chen, G. P. Luo, K. D. Ross, C. W. Chu,

"Oxygen Exchange Kinetics on a Highly Oriented La0.5Sr0.5CoO3-delta Thin Film Prepared by Pulsed-Laser Deposition", Applied Physics Letters, 79, [6] 776-8 (2001).

[17] T. Horita, K. Yamaji, N. Sakai, H. Yokokawa, T. Kato, "Oxygen Transport at

the LaMnO3 Film/Yttria-Stabilized Zirconia Interface under Different Cathodic Overpotentials by Secondary Ion Mass Spectrometry", Journal of the Electrochemical Society, 148, [5] J25-J30 (2001).

[18] T. Kawada, J. Suzuki, M. Sase, A. Kaimai, K. Yashiro, Y. Nigara, J. Mizusaki, K. Kawamura, H. Yugami, "Determination of Oxygen Vacancy Concentration

in a Thin Film of La0.6Sr 0.4CoO3-delta by an Electrochemical Method", Journal of the Electrochemical Society, 149, [7] E252-E9 (2002).

[19] V. Brichzin, J. Fleig, H.-U. Habermeier, G. Cristiani, J. Maier, "The Geometry

Dependence of the Polarization Resistance of Sr-doped LaMnO3 134 C H A P T E R 6

Microelectrodes on Yttria-Stabilized Zirconia", Solid State Ionics, 152-153, 499-507 (2002).

[20] T. Horita, K. Yamaji, N. Sakai, Y. Xiong, T. Kato, H. Yokokawa, T. Kawada, "Imaging of Oxygen Transport at SOFC Cathode/Electrolyte Interfaces by a Novel Technique", Journal of Power Sources, 106, [1-2] 224-30 (2002).

[21] E. Koep, D. S. Mebane, R. Das, C. Compson, M. Liu, "Characteristic

Thickness for a Dense La0.8Sr0.2MnO3 Electrode", Electrochemical and Solid- State Letters, 8, [11] A592-A5 (2005).

[22] M. Sase, D. Ueno, K. Yashiro, A. Kaimai, T. Kawada, J. Mizusaki, "Interfacial

Reaction and Electrochemical Properties of Dense (La,Sr) CoO3-delta Cathode on YSZ (1 0 0)", Journal of Physics and Chemistry of Solids, 66, [2-4] 343-8 (2005).

[23] F. S. Baumann, J. Fleig, M. Konuma, U. Starke, H.-U. Habermeier, J. Maier,

"Strong Performance Improvement of La0.6Sr0.4Co0.8Fe0.2O3-delta SOFC Cathodes by Electrochemical Activation", Journal of the Electrochemical Society, 152, [10] A2074-A9 (2005).

[24] M. Prestat, J.-F. Koenig, L. J. Gauckler, "Oxygen Reduction at Thin Dense

La0.52Sr0.48Co0.18Fe0.8O3-d Electrodes. Part I: reaction Model and Faradaic Impedance", Journal of Electroceramics 18, 87-101 (2007).

[25] M. Prestat, S. Korrodi, A. Infortuna, S. Rey-Mermet, P. Muralt, L. J. Gauckler,

"Oxygen Reduction at Thin Dense La0.52Sr0.48Co0.18Fe0.8O3-d Electrodes. Part II: Experimental Assessment of the Reaction kinetics", Journal of Electroceramics 18, 111-20 (2006).

[26] F. S. Baumann, J. Fleig, H. U. Habermeier, J. Maier, "Impedance

Spectroscopic Study on Well-Defined (La,Sr)(Co,Fe)O3-delta Model Electrodes", Solid State Ionics, 177, [11-12] 1071-81 (2006).

[27] J. Fleig, F. S. Baumann, V. Brichzin, H. R. Kim, J. Jamnik, G. Cristiani, H. U. Habermeier, J. Maier, "Thin Film Microelectrodes in SOFC Electrode Research", Fuel Cells, 6, [3-4] 284-92 (2006). E L E C T R O C H E M I C A L P E R F O R M A N C E 135

[28] K. Choy, W. Bai, S. Charojrochkul, B. C. H. Steele, "The Development of Intermediate-Temperature Solid Oxide Fuel Cells for the Next Millennium", Journal of Power Sources, 71, [1-2] 361-9 (1998).

[29] C.-Y. Fu, C.-L. Chang, C.-S. Hsu, B.-H. Hwang, "Electrostatic Spray

Deposition of La0.8Sr0.2Co0.2Fe0.8O3 Films", Materials Chemistry and Physics, 91, [1] 28-35 (2005).

[30] D. Perednis, O. Wilhelm, S. E. Pratsinis, L. J. Gauckler, "Morphology and Deposition of Thin Yttria-Stabilized Zirconia Films using Spray Pyrolysis", Thin Solid Films, 474, [1-2] 84-95 (2005).

[31] D. Perednis, L. J. Gauckler, "Thin Film Deposition Using Spray Pyrolysis", Journal of Electroceramics, 14, [2] 103-11 (2005).

[32] A. Princivalle, D. Perednis, R. Neagu, E. Djurado, "Microstructural

Investigations of Nanostructured La(Sr)MnO3-delta Films Deposited by Electrostatic Spray Deposition", Chemistry of Materials, 16, [19] 3733-9 (2004).

[33] L. J. Gauckler, D. Beckel, B. Buergler, E. Jud, U. P. Muecke, M. Prestat, J. Rupp, J. Richter, "Solid Oxide Fuel Cells: Systems and Materials", Chimia, 58, [12] 837-50 (2004).

[34] J. M. Ralph, C. Rossignol, R. Kumar, "Cathode Materials for Reduced- Temperature SOFCs", Journal of the Electrochemical Society, 150, [11] A1518-A22 (2003).

[35] C. Argirusis, T. Damjanovic, G. Borchardt, "Preparation of SOFC Cells by Means of Electrophoretic Deposition", Key Engineering Materials, 314, 101-6 (2006).

[36] M. L. Liu, D. S. Wang, "Preparation of La1-zSrzCo1-yFeyO3-x Thin-Films, Membranes, and Coatings on Dense and Porous Substrates", Journal of Materials Research, 10, [12] 3210-21 (1995).

[37] C. S. Hsu, B. H. Hwang, "Microstructure and Properties of the

La0.6Sr0.4Co0.2Fe0.8O3 Cathodes Prepared by Electrostatic-Assisted Ultrasonic Spray Pyrolysis Method", Journal of the Electrochemical Society, 153, [8] A1478-A83 (2006). 136 C H A P T E R 6

[38] D. Beckel, A. Dubach, A. R. Studart, L. J. Gauckler, "Spray Pyrolysis of

La0.6Sr0.4Co0.2Fe0.8O3-d Thin Film Cathodes", Journal of Electroceramics, 16, [3] 221-8 (2006).

[39] D. Beckel, D. Briand, A. R. Studart, N. F. de Rooij, L. J. Gauckler, "Topography Mediated Patterning of Inorganic Materials by Spray Pyrolysis", Advanced Materials, 18, [22] 3015-8 (2006).

[40] U. P. Muecke, G. L. Messing, L. J. Gauckler, "The Leidenfrost Effect During

Spray Pyrolysis of Dense NiO-Ce0.8Gd0.2O1.9-x Thin Films", Submitted to Thin Solid Films, (2006).

[41] U. P. Muecke, N. Luechinger, L. J. Gauckler, "Initial Status of Deposition and Film Formation During Spray Pyrolysis of Nickel Oxide, Cerium Gadolinium Oxide and NiO-CGO Thin Films", Submitted to Thin Solid Films, (2006).

[42] J. L. M. Rupp, A. Infortuna, L. J. Gauckler, "Microstrain and Self-Limited Grain Growth in Nanocrystalline Ceria Ceramics", Acta Materialia, 54, [7] 1721-30 (2006).

[43] D. Perednis, L. J. Gauckler, "Solid Oxide Fuel Cells with Electrolytes Prepared via Spray Pyrolysis", Solid State Ionics, 166, [3-4] 229-39 (2004).

[44] P. R. Willmott, "Deposition of Complex Multielemental Thin Films", Progress in Surface Science, 76, [6-8] 163-217 (2004).

[45] P. R. Willmott, J. R. Huber, "Pulsed Laser Vaporization and Deposition", Review of Modern Physics, 72, [1] 315-28 (2000).

[46] S. B. Adler, B. T. Henderson, M. A. Wilson, D. M. Taylor, R. E. Richards, "Reference Electrode Placement and Seals in Electrochemical Oxygen Generators", Solid State Ionics, 134, [1-2] 35-42 (2000).

[47] J. Winkler, P. V. Hendriksen, N. Bonanos, M. Mogensen, "Geometric Requirements of Solid Electrolyte Cells with a Reference Electrode", Journal of the Electrochemical Society, 145, [4] 1184-92 (1998).

[48] W. G. Wang, M. Mogensen, "High-Performance Lanthanum-Ferrite-Based Cathode for SOFC", Solid State Ionics, 176, [5-6] 457-62 (2005). E L E C T R O C H E M I C A L P E R F O R M A N C E 137

[49] V. Dusastre, J. A. Kilner, "Optimisation of Composite Cathodes for Intermediate Temperature SOFC Applications", Solid State Ionics, 126, [1-2] 163-74 (1999).

[50] S. L. dos Santos e Lucato, Lince - Linear Intercept v2.4, Department of Material Science, Darmstadt University of Technology, Darmstadt, Germany (1999).

[51] M. Gödickemeier, K. Sasaki, L. J. Gauckler, I. Riess, "Electrochemical Characteristics of Cathodes in Solid Oxide Fuel Cells based on Ceria Electrolytes", Journal of the Electrochemical Society, 144, [5] 1635-46 (1997).

[52] A. Esquirol, N. P. Brandon, J. A. Kilner, M. Mogensen, "Electrochemical

Characterization of La0.6Sr0.4Co0.2Fe0.8O3 Cathodes for Intermediate- Temperature SOFCs", Journal of the Electrochemical Society, 151, [11] A1847-A55 (2004).

[53] F. van Berkel, S. Brussel, M. van Tuel, G. Schoemakers, B. Rietveld, P. V. Aravind, "Development of Low Temperature Cathode Materials"; P0626 in Proceedings of 7th European Solid Oxide Fuel Cell Forum Edited by U. Bossel, (2006).

[54] J. A. Lane, P. H. Middleton, H. Fox, B. C. H. Steele, J. A. Kilner, "Study of Selected perovskite Cathode Materials on Doped Ceria Electrolyte by AC impedance Spectroscopy"; pp. 489-504 in Proceedings of 2nd International Symposium on Ionic and Mixed Conducting Ceramics PV 94-12 Edited by T.A. Ramanarayanan, W.L. Worrel, H.L. Tuller, The Electrochemical Society, (1994).

[55] H. Zhao, L. H. Huo, L. P. Sun, L. J. Yu, S. Gao, J. G. Zhao, "Preparation, Chemical Stability and Electrochemical Properties of LSCF-CBO Composite Cathodes", Materials Chemistry and Physics, 88, [1] 160-6 (2004).

[56] M. Mogensen, K. V. Jensen, M. J. Jorgensen, S. Primdahl, "Progress in Understanding SOFC Electrodes", Solid State Ionics, 150, [1-2] 123-9 (2002).

[57] A. Esquirol, J. Kilner, N. Brandon, "Oxygen Transport in

La0.6Sr0.4Co0.2Fe0.8O3-delta/Ce0.8Ge0.2O2-x Composite Cathode for IT-SOFCs", Solid State Ionics, 175, [1-4] 63-7 (2004). 138 C H A P T E R 6

[58] J. D. Sirman, J. A. Kilner, "Surface Exchange Properties of Ce0.9Gd0.1O2-x

Coated with La1-xSrxFeyCo1-yO3-delta", Journal of the Electrochemical Society, 143, [10] L229-L31 (1996).

[59] B. C. H. Steele, K. M. Hori, S. Uchino, "Kinetic Parameters Influencing the Performance of IT-SOFC Composite Electrodes", Solid State Ionics, 135, [1-4] 445-50 (2000).

[60] Z. P. Shao, S. M. Haile, "A High-Performance Cathode for the Next Generation of Solid- Oxide Fuel Cells", Nature, 431, [7005] 170-3 (2004).

[61] Z. P. Shao, W. S. Yang, Y. Cong, H. Dong, J. H. Tong, G. X. Xiong, "Investigation of the Permeation Behavior and Stability of a

Ba0.5Sr0.5Co0.8Fe0.2O3-delta Oxygen Membrane", Journal of Membrane Science, 172, [1-2] 177-88 (2000).

[62] S. Lee, Y. Lim, E. A. Lee, H. J. Hwang, J. W. Moon, "Ba0.5Sr0.5Co0.8Fe0.2O3-delta

(BSCF) and La0.6Ba0.4Co0.2Fe0.8O3-delta (LBCF) Cathodes Prepared by Combined Citrate-EDTA Method for IT-SOFCs", Journal of Power Sources, 157, [2] 848-54 (2006).

[63] A. Yan, M. Cheng, Y. L. Dong, W. S. Yang, V. Maragou, S. Q. Song, P.

Tsiakaras, "Investigation of a Ba(0.5)Sr(0.5)Co(0.8)Fe(0.2)O3-delta based Cathode IT-

SOFC - I. The Effect of CO2 on the Cell Performance", Applied Catalysis B, 66, [1-2] 64-71 (2006).

[64] Q. S. Zhu, T. A. Jin, Y. Wang, "Thermal Expansion Behavior and Chemical

Compatibility of BaxSr1-xCo1-yFeyO3-delta with 8YSZ and 20GDC", Solid State Ionics, 177, [13-14] 1199-204 (2006).

[65] R. D. Shannon, "Revised Effective Ionic Radii and Systematic Studies of Interatomic Distances in Halides and Chalcogenides", Acta Crystallographica Section A, 32, 751 (1976).

[66] J. ten Elshof, J. Boeijsma, "Influence of Iron Content on Cell Parameters of

Rhombohedral La0.6Sr0.4Co1-yFeyO3", Powder Diffraction, 11, [3] 240-5 (1996).

[67] J.-M. Bae, B. C. H. Steele, "Properties of La0.6Sr0.4Co0.2Fe0.8O3-[delta] (LSCF) Double Layer Cathodes on Gadolinium-doped Cerium Oxide (CGO) Electrolytes: I. Role of SiO2", Solid State Ionics, 106, [3-4] 247-53 (1998). E L E C T R O C H E M I C A L P E R F O R M A N C E 139

[68] B. C. H. Steele, J.-M. Bae, "Properties of La0.6Sr0.4Co0.2Fe0.8O3-x (LSCF) Double Layer Cathodes on Gadolinium-Doped Cerium Oxide (CGO) Electrolytes: II. Role of Oxygen Exchange and Diffusion", Solid State Ionics, 106, [3-4] 255-61 (1998).

140 C H A P T E R 6

7 Topography Mediated Patterning of Inorganic Materials

by Spray Pyrolysis

Microstructured ceramic thin films were produced by spray pyrolysis deposition of metal salt solutions onto micromachined substrates. The ceramic structures built on the substrate result from the preferential assembly of particles on the edge of the initially micromachined structures, leading to smaller lateral dimensions (1 – 2 µm in width) than the initial structures on the substrate. This assembly process was observed to amplify the height of the initial structures by a factor of ~30, resulting in features with an aspect ratio of 3. This novel simple method is expected to aid the fabrication of miniaturized devices for many microsystem technologies.

7.1 Introduction

Ceramic materials are of interest in miniaturized systems not only due to their high thermal and chemical resistance, but also because of their hardness and possible functional properties such as high dielectric constant, ionic conductivity and catalytic activity. In micro-electromechanical systems (MEMS), most of the ceramics materials are used in the form of thin films. These materials can be structured with a variety of techniques, including molding, direct writing, photolithography, self assembly and substrate surface modification.

Molding techniques [1] use ceramic powders as starting material and can be a convenient simple approach when casting the suspension on a mold [2-4]. Infiltration of a particle containing suspension into capillaries of a mold is a modification of this

D. Beckel, D. Briand, A. R. Studart, N. F. de Rooij L. J. Gauckler, Advanced Materials, 18, [22] 3015-8 (2006). 142 C H A P T E R 7 process referred to as micro-molding in capillaries (MIMIC) [5-13]. This method was recently applied for the fabrication of miniaturized gas sensors with structures of the sensing material of less than 5 µm [8]. Another development of the molding technique is the combination with thin film deposition techniques. [14, 15]

Direct writing techniques are usually carried out by either a selective deposition of material using ink-jet printing [16-19] or selective removal using a laser [20-23]. In case of selective material deposition the minimum lateral feature size is typically 100 μm, whereas features of about 10 µm can be achieved with selective removal approach.

Photolithography in combination with etching of ceramics is difficult due to the chemical inertness of most ceramic materials. Structuring of the precursor film itself by photolithography was however demonstrated for Ti based alkoxide precursor films modified by ethanol amines [24].

Self assembling of the precursor typically leads to films with ordered micro- or nano- structures, e.g. three dimensional porous structures with nano-sized pores and struts were realized with silicon oxycarbide [25], but without deliberately formed shapes.

Substrate surface modification [1, 26, 27] encompasses another large group of microstructuring techniques which also includes most of the bio inspired approaches [28]. In these techniques the resolution limit is generally given by the photolithography technique employed.

Here we present a novel and simple technique for micro-structuring ceramic thin films. In this method, structures obtained by micromachining substrates using standard photolithographic techniques are used to guide the assembly of ceramic particles into deliberately designed structures of various shapes. Using this bottom up approach on previously structured substrates, features sizes close to the theoretical resolution limit of standard photolithography [29] can be achieved. The ceramic perovskite La0.6Sr0.4Co0.2Fe0.8O3 (LSCF) is used to illustrate the method, due to its importance for the fabrication of electrodes for solid oxide fuel cell [30-32] and oxygen separation membranes [33, 34].

T O P O G R A P H Y M E D I A T E D P A T T E R N I N G 143

7.2 Experimental

7.2.1 Substrate Structuring

The processing of the structures was performed on 390 µm-thick silicon wafers with a diameter of 100 mm. The silicon wafers were cleaned in hot sulfuric acid and coated with a 0.5 µm-thick low-stress silicon nitride layer by low pressure chemical vapor deposition (LPCVD). Metal lines were patterned on the top of the silicon nitride layer by e-beam evaporation of platinum / tantalum thin films (225 nm / 15 nm) using a lift- off process. The metal lines were finally passivated with a 0.5 µm-thick low-stress silicon nitride layer deposited by LPCVD. This process was developed for the fabrication of high temperature micro-hotplates [35].

7.2.2 Film Deposition

LSCF films are deposited by air-pressurized spray pyrolysis. The metal salts were dissolved in a solvent mixture and the resulting solution was pumped (peristaltic pump: Ismatec MS Reglo or syringe pump: Razell Scientific Instruments A 99) through a nozzle (Badger Air-brush Model 150) and atomized by air pressure. The resulting droplets were sprayed onto a heated substrate where an amorphous metal oxide film is formed. The films were subsequently crystallized in an additional annealing step. For a substrate temperature of about 270°C, the droplets were deposited at an air pressure of 1 bar with a solution flow rate of 30 ml/h, using a nozzle to substrate distance of 20 cm. At these conditions, a spraying time of 60 min is sufficient to obtain a layer thickness of ~500 nm. The spray solution consists of a mixture of salts dissolved in organic solvent at a molar ratio of LaNO3·6H2O :

SrCl2·6H2O : Co(NO3)2·6H2O : Fe(NO3)3·9H2O = 3:2:1:4 (all from Fluka with purity > 98 %). The solution had a total salt concentration of 0.02 mol/l and the organic solvent consists of a mixture of 1/3 (volume fraction) ethanol (Scharlau and Merk, purity > 99.5 %) and 2/3 diethylene glycol monobutyl ether (from Fluka and Acros Organic, purity ≥ 98 %). The post deposition annealing conditions were chosen to be 4 hrs at 600°C in air with a heating ramp of 3°C/min. More details about the experimental setup can be found elsewhere [30]. 144 C H A P T E R 7

7.2.3 Topography Characterization

Microstructures were investigated using scanning electron microscopy (SEM, Leo 1530). The structures on the substrate were characterized using atomic force microscopy (AFM Dimension 3000, Veeco) in contact mode with a SiN tip (Veeco) at a scan rate of 1 Hz. All AFM images were corrected using Plane Fit Manual of the instrument’s software (Ver. 4.23r3). The height of the structures on the substrate was determined by cross sectional analysis with AFM.

7.3 Results

LSCF thin films were deposited by spray pyrolysis on flat Si substrates and on substrates with thin structures on top as schematically shown in Fig. 7.1. The spraying parameters used are described in the experimental section. On the non- structured substrates, rough films with random distributed ridges on top of the thin film are obtained as shown in Fig. 7.2 A). In the case of structured substrates, the formerly irregular ridges can be aligned according to the original structures as shown in Fig. 7.2 B). The ridges follow the topography of the substrate irrespective of the geometry of the structures. Straight lines, curves and even sharp edges can be realized as shown in Fig. 7.3 A).

The appearance of random shaped ridges in between the intended structures depends on the distance between these structures. Fig. 7.3 B) shows that for distances larger than about 50 µm, random ridge formation starts between the intended structures. The relation between the topography of the substrate and the geometry of the ridge on the thin film is shown in the cross section micrograph in Fig. 7.3 C). The ridges assemble only at the edge of the initial structure on the substrate and are only 1 – 2 µm wide, which is also the minimum feature size. In case the initial structure is a flat line, two ridges will follow the line in parallel, separated by the width of the line. Thereby the structure density is doubled. Interestingly, the height of the substrate’s topography is remarkably amplified from about 100 to 200 nm to nearly 3 μm. By further thin film deposition, the thickness of the film and the ridges are both increased, thus the contrast is not improved by longer deposition. Furthermore, the ridges grow in width during longer thin film deposition.

T O P O G R A P H Y M E D I A T E D P A T T E R N I N G 145

Fig. 7.1: Sketch of a typical topography of the substrate prior to thin film deposition. The height is exaggerated by a factor of 100 for clarity. Inset: AFM image of such a structured substrate.

The creation of oriented ridges on the thin film is related to the mechanism of film formation during spray pyrolysis [30]. In this process, fine droplets (~5 µm) of a metal salt solution are sprayed onto a heated substrate. Notable evaporation of the droplets occurs near the substrate surface [36, 37], leading to the precipitation of metal salt complexes inside the droplet prior to deposition. Upon impact of the droplet on the substrate, the precipitates move laterally with the spreading droplet and collide into neighboring obstacles, as depicted in Fig. 7.4. Van-der-Waals and electrostatic 146 C H A P T E R 7 interactions between precipitates and the defect-rich edges might also contribute to ridge formation. However, these interactions would only lead to a smoothening of the edge rather than building a high-aspect-ratio structure as observed in Fig. 7.3. Therefore, we expect the lateral transport of precipitates upon drop impact to play a major role in the process of ridge formation. On a non-structured substrate, the first dried droplets form randomly distributed obstacles, which subsequently trap the precipitates of other spreading droplets (Fig. 7.2 A)). In case of a structured substrate, the obstacles are the previously micro-machined structures which guide the assembly of the precipitates during deposition. This model explains why random ridges are formed as soon as the structures on the substrate end (Fig. 7.3 B) or when the distance between two structures is too large. In principle this distance can be extended when using larger droplets, since they spread over larger areas.

A B

40 μm 40 μm

Fig. 7.2: A) Film deposited onto an unstructured substrate showing random distribution of the ridges. B) Film deposited onto a structured substrate resulting in aligned ridges. When spacing of the structures is too large, some random ridges form in between.

T O P O G R A P H Y M E D I A T E D P A T T E R N I N G 147

A B

100 μm 40 μm

C

1 μm

Ridge on thin film

Height of Thin film structure Substrate

Fig. 7.3: A) Top view of the structures achieved by spray pyrolysis, illustrating the different geometries that can be obtained (straight lines, curves and edges). B) Depending on the spacing between the structures, random ridge formation occurs. C) Cross section through a ridge formed on the structured substrate.

The tendency of the thin film to form ridges can be tailored by the spray pyrolysis conditions. The amount of precipitates formed in the droplet determines whether ridges are built, or a smooth surface is achieved [30]. In case many precipitates are present in the spreading droplet, no extensive lateral displacement of precipitates is possible, because the precipitates impede themselves. Thus they remain where they statistically hit the substrate and form smoother films. If there are only few precipitates in the spreading droplet, these precipitates can move laterally before the 148 C H A P T E R 7 droplet is completely dried. The lateral movement of precipitates is then impeded by the neighboring obstacles, favoring ridge formation. The concentration of precipitates in the droplets can be tailored by the metal salt concentration in the spray solution, the type of solvent, the deposition temperature, the air pressure and the droplet size, since larger droplets will evaporate slower, leading to less precipitates in the droplet.

Besides LSCF, we also observed ridge formation for La0.6Sr0.4CoO3, Sm0.5Sr0.5CoO3,

La0.75Sr0.25Cr0.5Mn0.5O3 and NiO, as shown in Fig. 7.5.

Pressurized air

Spray solution Evaporation of solvent

Controlled temperature Substrate

Fig. 7.4: Sketch illustrating the formation of the ridges. Precipitates are formed inside the droplets upon evaporation of the solvent close to the substrate. Upon impact of the droplet on the substrate, the precipitates move laterally and collide against the original structure on the substrate.

T O P O G R A P H Y M E D I A T E D P A T T E R N I N G 149

A B

C D

Fig. 7.5: Ridge formation is also observed for A) La0.6Sr0.4CoO3, B) Sm0.5Sr0.5CoO3, C)

La0.75Sr0.25Cr0.5Mn0.5O3 and D) NiO.

7.4 Summary

In summary, LSCF thin films exhibiting random shaped ridges can be produced by choosing the appropriate parameters during spray pyrolysis of metal salt solutions. When these solutions are sprayed onto substrates with structured surfaces, the 150 C H A P T E R 7 ridges follow the deliberately formed initial pattern. The lateral dimension of the resulting patterned ridges (1 – 2 µm) is at least one order of magnitude smaller than that of the initial structure, whereas the height is amplified by a factor of about 30. High feature densities are achieved, because the number of the initial structures is doubled. Using this combined approach of standard photolithography for structuring the substrate with an inexpensive spray technique for the deposition of ceramic thin films on large substrates, we produced microstructured ceramics with feature sizes in the range of the theoretical resolution limit of standard photolithography, exhibiting an aspect ratio of ~3. Other well-established techniques can be used for the initial structuring of the substrate, which imparts great flexibility to this novel method. Due to the versatility of the spray technique, micropatterned films with a wide range of materials can also be prepared using the approach outlined here. Therefore, we expect this simple combined method to aid the fabrication of a wide number of functional miniaturized devices for sensing, energy conversion, MEMS and related applications.

7.5 References

[1] C. R. Martin, I. A. Aksay, "Submicrometer-Scale Patterning of Ceramic Thin Films", Journal of Electroceramics, 12, [1-2] 53-68 (2004).

[2] U. P. Schonholzer, N. Stutzmann, T. A. Tervoort, P. Smith, L. J. Gauckler, "Micropatterned Ceramics by Casting into Polymer Molds", Journal of the American Ceramic Society, 85, [7] 1885-7 (2002).

[3] U. P. Schonholzer, L. J. Gauckler, "Ceramic Parts Patterned in the Micrometer Range", Advanced Materials, 11, [8] 630-2 (1999).

[4] U. P. Schonholzer, R. Hummel, L. J. Gauckler, "Microfabrication of Ceramics by Filling of Photoresist Molds", Advanced Materials, 12, [17] 1261-3 (2000).

[5] E. Kim, Y. N. Xia, G. M. Whitesides, "Micromolding in Capillaries: Applications in Materials Science", Journal of the American Chemical Society, 118, [24] 5722-31 (1996).

[6] E. Kim, Y. N. Xia, G. M. Whitesides, "Polymer Microstructures Formed by Molding in Capillaries", Nature, 376, [6541] 581-4 (1995).

T O P O G R A P H Y M E D I A T E D P A T T E R N I N G 151

[7] X. M. Zhao, Y. N. Xia, G. M. Whitesides, "Soft Lithographic Methods for Nano- Fabrication", Journal of Materials Chemistry, 7, [7] 1069-74 (1997).

[8] M. Heule, L. J. Gauckler, "Gas Sensors Fabricated from Ceramic Suspensions by Micromolding in Capillaries", Advanced Materials, 13, [23] 1790-3 (2001).

[9] H. Yang, P. Deschatelets, S. T. Brittain, G. M. Whitesides, "Microstructures from a Polymeric Precursor Using Soft Lithography", Advanced Materials, 13, [1] 54-8 (2001).

[10] W. S. Beh, Y. Xia, D. Qin, "Formation of Patterned Microstructures of Polycrystalline Ceramics from Precursor Polymers Using Micromolding in Capillaries", Journal of Materials Research, 14, [10] 3995-4003 (1999).

[11] S. Seraji, Y. Wu, N. E. Jewell-Larson, M. J. Forbess, S. J. Limmer, T. P. Chou, G. Z. Cao, "Patterned Microstructure of Sol-Gel Derived Complex Oxides Using Soft Lithography", Advanced Materials, 12, [19] 1421-4 (2000).

[12] M. Heule, L. J. Gauckler, "Miniaturised Arrays of Tin Oxide Gas Sensors on Single Microhotplate Substrates Fabricated by Micromolding in Capillaries", Sensors and Actuators B, 93, [1-3] 100-6 (2003).

[13] M. Heule, "Shaping Ceramics in Small Scale - from Microcomponents to Gas Sensors", PhD Thesis Swiss Federal Institute of Technology, (2003).

[14] M. A. Auger, P. L. Schilardi, I. Caretti, O. Sánchez, G. Benítez, J. M. Albella, R. Gago, M. Fonticelli, L. Vázquez, R. C. Salvarezza, O. Azzaroni, "Molding and Replication of Ceramic Surfaces with Nanoscale Resolution", Small, 1, [3] 300-9 (2005).

[15] O. Azzaroni, P. L. Schilardi, R. C. Salvarezza, J. Manuel-Herrero, C. Zaldo, L. Vázquez, "Surface-Relief Micropatterning of Zinc Oxide Substrates by Micromolding Pulsed-Laser-Deposited Films", Applied Physics A, 81, [6] 1113- 6 (2005).

[16] P. Calvert, "Inkjet Printing for Materials and Devices", Chemistry of Materials, 13, [10] 3299-305 (2001).

[17] K. A. M. Seerden, N. Reis, J. R. G. Evans, P. S. Grant, J. W. Halloran, B. Derby, "Ink-Jet Printing of Wax-based Alumina Suspensions", Journal of the American Ceramic Society, 84, [11] 2514-20 (2001). 152 C H A P T E R 7

[18] R. Teranishi, T. Fujiwara, T. Watanabe, M. Yoshimura, "Direct Fabrication of Patterned PbS and US on Organic Sheets at Ambient Temperature by on-site Reaction Using Inkjet Printer", Solid State Ionics, 151, [1-4] 97-103 (2002).

[19] X. L. Zhao, J. R. G. Evans, M. J. Edirisinghe, J. H. Song, "Direct Ink-Jet Printing of Vertical Walls", Journal of the American Ceramic Society, 85, [8] 2113-5 (2002).

[20] A. Kruusing, S. Leppavuori, A. Uusimaki, B. Petretis, O. Makarova, "Micromachining of Magnetic Materials", Sensors and Actuators A, 74, [1-3] 45-51 (1999).

[21] A. Pique, D. B. Chrisey, J. M. Fitz-Gerald, R. A. McGill, R. C. Y. Auyeung, H. D. Wu, S. Lakeou, V. Nguyen, R. Chung, M. Duignan, "Direct Writing of Electronic and Sensor Materials Uusing a Laser Transfer Technique", Journal of Materials Research, 15, [9] 1872-5 (2000).

[22] M. Geiger, S. Roth, W. Becker, "Influence of Laser-Produced Microstructures on the Tribological Behaviour of Ceramics", Surface & Coatings Technology, 101, [1-3] 17-22 (1998).

[23] C. Buerhop, N. Lutz, R. Weissmann, G. Tomandl, "Surface-Treatment of Glass and Ceramics Using XeCl Excimer-Laser Radiation", Glastechnische Berichte-Glass Science and Technology, 66, [3] 61-7 (1993).

[24] K. Kikuta, K. Takagi, S. Hirano, "Photoreaction of Titanium-based Metal- Organic Compounds for Ceramic Fine Patterning", Journal of the American Ceramic Society, 82, [6] 1569-72 (1999).

[25] V. Z. H. Chan, J. Hoffman, V. Y. Lee, H. Iatrou, A. Avgeropoulos, N. Hadjichristidis, R. D. Miller, E. L. Thomas, "Ordered Bicontinuous Nanoporous and Nanorelief Ceramic Films From Self Assembling Polymer Precursors", Science, 286, [5445] 1716-9 (1999).

[26] M. Heule, U. P. Schonholzer, L. J. Gauckler, "Patterning Colloidal Suspensions by Selective Wetting of Microcontact-Printed Surfaces", Journal of the European Ceramic Society, 24, [9] 2733-9 (2004).

T O P O G R A P H Y M E D I A T E D P A T T E R N I N G 153

[27] M. Heule, J. Schell, L. J. Gauckler, "Powder-based Tin Oxide Microcomponents on Silicon Substrates Fabricated by Micromolding in Capillaries", Journal of the American Ceramic Society, 86, [3] 407-12 (2003).

[28] Y. F. Gao, K. Koumoto, "Bioinspired Ceramic Thin Film Processing: Present Status and Future Perspectives", Crystal Growth & Design, 5, [5] 1983-2017 (2005).

[29] M. J. Madou, "Fundamentals of Microfabrication, The Science of Miniaturization" CRC Press, Boca Raton p. p. 21 (2002).

[30] D. Beckel, A. Dubach, A. R. Studart, L. J. Gauckler, "Spray Pyrolysis of

La0.6Sr0.4Co0.2Fe0.8O3-d Thin Film Cathodes", Journal of Electroceramics, 16, [3] 221-8 (2006).

[31] A. Bieberle-Hütter, D. Beckel, U. P. Muecke, J. L. M. Rupp, A. Infortuna, L. J. Gauckler, "Micro-Solid Oxide Fuel Cells as Battery Replacements", Mstnews, 04/05, 12-5 (2005).

[32] L. J. Gauckler, D. Beckel, B. Buergler, E. Jud, U. P. Muecke, M. Prestat, J. Rupp, J. Richter, "Solid Oxide Fuel Cells: Systems and Materials", Chimia, 58, [12] 837-50 (2004).

[33] Y. Teraoka, Y. Honbe, J. Ishii, H. Furukawa, I. Moriguchi, "Catalytic Effects in Oxygen Permeation through Mixed-Conductive LSCF Perovskite Membranes", Solid State Ionics, 152, 681-7 (2002).

[34] X. Y. Tan, Y. T. Liu, K. Li, "Preparation of LSCF Ceramic Hollow-Fiber Membranes for Oxygen Production by a Phase-Inversion/Sintering Technique", Industrial & Engineering Chemistry Research, 44, [1] 61-6 (2005).

[35] D. Briand, A. Krauss, B. van der Schoot, U. Weimar, N. Barsan, W. Gopel, N. F. de Rooij, "Design and Fabrication of High-Temperature Micro-Hotplates for Drop-Coated Gas Sensors", Sensors and Actuators B, 68, [1-3] 223-33 (2000).

[36] D. Perednis, "Thin Film Deposition by Spray Pyrolysis and the Application in Solid Oxide fuel Cells", PhD Thesis Swiss Federal Institute of Technology, (2003). 154 C H A P T E R 7

[37] O. Wilhelm, S. E. Pratsinis, D. Perednis, L. J. Gauckler, "Electrospray and Pressurized Spray Deposition of Yttria-Stabilized Zirconia Films", Thin Solid Films, 479, [1-2] 121-9 (2005).

8 Micro-Hotplates – a Platform for Micro-Solid Oxide Fuel

Cells

Special high temperature micro-hotplates were investigated for their potential use as a technology platform for miniaturized solid oxide fuel cells (SOFCs). To evaluate the compatibility of these hotplates with typical SOFC materials and processing techniques, spray pyrolysis was used for the deposition of an

SOFC cathode material (La0.6Sr0.4Co0.2Fe0.8O3) onto the micro- hotplate. The resulting microstructures and the electrical conductivity of the thin film (thickness ~500 nm) were characterized. Post-deposition annealing in external furnaces as well as using the integrated heater of the micro-hotplate confirmed that these micro-hotplates are suitable for a maximum operation temperature of 800°C and a long-term operation at 600°C. Very fast heating and cooling rates of several 100°C/min were achieved using the micro-hotplate as a heater, allowing fast processing and aging tests. Bending of the micro-hotplate was found to be critical to the integrity of the thin film, which coats the micro-hotplate. Bending might be reduced using other membrane materials. These membranes are potential components for a micro SOFC in which the integrated heater is used for start-up operation of the fuel cell.

8.1 Introduction

Micro-hotplates generally consist of a thin (~1 µm) dielectric membrane suspended over an opening in a silicon substrate [1-4]. In microsystems technology, such

D. Beckel, D. Briand, A. Bieberle-Hütter, J. Courbat, N. F. de Rooij L. J. Gauckler, Journal of Power Sources, 166, [1] 143-8 (2007). 156 C H A P T E R 8 hotplates are mainly used for sensor applications where the sensing material is deposited onto the membrane and an electrical signal is conducted using contacts integrated on the hotplate [1, 3]. A heater is also integrated in the hotplate, which typically operates at temperatures around 200 to 400°C [5]. In recent years, micro- hotplates operating at temperatures higher than 500°C have been developed for various applications, such as sensors, actuators, infrared sources, micro-reactors or to perform annealing of films or chemical vapor deposition (CVD) on-chip [1, 3, 6, 7].

In the field of solid oxide fuel cells (SOFCs), development currently trends towards lower operating temperatures (from 800 – 1000°C down to 500 – 700°C) [8-10] and to miniaturization of SOFCs [11]. Reduction of the operating temperature is intended to reduce material degradation and facilitate sealing of SOFCs. Miniaturization of SOFCs enables new applications such as battery replacement [11]. With thin film components the ohmic resistances of the cell’s components are reduced, allowing operation at lower temperatures.

These developments led to the situation in which SOFCs and micro-hotplates are now able to operate at similar temperatures, i.e. around 500°C. Thus we investigate a micro-hotplate as platform for a micro SOFC (µSOFCs) with an integrated heater that can be used for start-up operation as well as for processing of SOFC components. Electrical contacts and a thermocouple can also be integrated.

To evaluate the feasibility of this approach, first the compatibility of the thin film components with the micro-hotplates needed to be clarified. Contacting of the µSOFC is simplified if all three components (cathode, electrolyte and anode) are deposited onto one side of the micro-hotplate. Therefore, we deposited in this study the cathode, which is the first layer, directly onto the high temperature micro-hotplate

[1]. We chose La0.6Sr0.4Co0.2Fe0.8O3 (LSCF) as cathode material and spray pyrolysis [12, 13] as deposition technique. The microstructural and mechanical integrity of the LSCF thin film is investigated by high resolution microscopy. We also evaluated the electrical conductivity of the thin film on the micro-hotplate. Realization of an entire µSOFC would then require deposition of the electrolyte and the anode on top of the cathode plus openings in the micro-hotplate membrane to allow gas access to the cathode.

M I C R O – H O T P L A T E S F O R M I C R O – S O F C 157

8.2 Experimental

The LSCF thin films were deposited by spray pyrolysis onto micro-hotplates. Contacting for the electrical measurements was done via patterned electrodes of defined geometry, which were fabricated on the micro-hotplates prior to thin film deposition. A detailed description of each process step is given in the following sections.

8.2.1 Micro-Hotplate Design and Fabrication

The micro-hotplates were fabricated on silicon wafers. They consisted of a dielectric membrane in which a platinum (Pt) heater was embedded and covered on top by electrodes (Fig. 8.1 A)). The heater was made of platinum with a tantalum (Ta) adhesion layer (225 nm thick). The Pt/Ta film was patterned using a lift-off process. In this process a sacrificial lift-off resist layer (LOR MicroChem Corp.) was patterned using a standard S1823 photoresist (Shipley Corp.) on top. After Pt/Ta deposition the sacrificial layer was removed, leaving the Pt/Ta only in the areas where an opening was patterned in the sacrificial layer. More details on the lift-off process are described elsewhere [14]. Two low-stress silicon nitride (SiN) films deposited by low pressure chemical vapor deposition (LPCVD) formed the thermally insulated 1.0 μm-thick membrane of the micro-hotplate. Dense Pt/Ta electrodes 150 nm thick were patterned on top of the membrane using the same lift-off process described above. The dielectric membrane was released using back-side bulk wet etching of the silicon wafer. More details about the fabrication process can be found in a reference [1].

The membrane has an area of 1.0 × 1.0 mm2, and the areas covered by the double meander heater and the electrodes were 530 × 530 μm2 and 400 × 400 μm2, respectively. The electrodes were used to contact the film for the electrical measurements with defined geometries. In a complete fuel cell, current collectors could be fabricated the same way. The design included different geometries for the shape of the electrodes in a four-point configuration. The electrodes were made of four square contact pads with two different gaps and widths. One design has square contacts 180 µm wide and a gap between the square contacts of 40 µm. The other design has square contacts 50 µm wide and a gap between the different square contacts of 300 µm as shown in Fig. 8.1 B). 158 C H A P T E R 8

Fig. 8.1: A): Schematic cross-section view of a micro-hotplate made of a Pt heater in between two SiN thin films, covered on top by electrodes. B): Top view of micro-hotplate with membrane and heater.

The temperature reached at the centre of the micro-hotplates as a function of the input power was calibrated using a micro-thermocouple, type S, made of Pt-PtRh wires with a diameter of 1.3 μm [15]. A temperature of 600°C was reached at a low power of 120 mW.

8.2.2 Thin Film Deposition

LSCF thin films were deposited by air-pressurized spray pyrolysis. For the spray solution, a mixture of metal salts with a molar ratio of LaNO3·6H2O : SrCl2·6H2O :

Co(NO3)2·6H2O : Fe(NO3)3·9H2O = 3:2:1:4 (all from Fluka with purity > 98 %) and a total salt concentration of 0.02 mol/l was dissolved in a solvent composition of 1/3 (volume fraction) ethanol (Scharlau and Merck, purity > 99.5 %) and 2/3 diethylene glycol monobutyl ether (from Fluka and Acros Organic, purity ≥98 %). The resulting solution was pumped (peristaltic pump: Ismatec MS Reglo or syringe pump: Razell Scientific Instruments A 99) with a solution flow rate of 30 ml/h through a nozzle (Badger Air-brush Model 150), which was located 20 cm above the micro-hotplate and atomized by air pressure of 1 bar. These droplets were sprayed through a shadow mask onto the micro-hotplate using an external hotplate at around 270°C in order to evaporate the solvent. The micro-hotplate was fixed in an aluminum holder that also held and aligned the shadow mask. At these conditions a spraying time of 60 min resulted in a layer thickness of ~500 nm to ~1 µm.

Alternatively, the film was deposited onto the micro-hotplate using the integrated heater also at 270°C. No masks were needed, because the film was only deposited M I C R O – H O T P L A T E S F O R M I C R O – S O F C 159 in the hot area. Outside the hot area the liquid spray solution accumulated and was rinsed after deposition. Care has to be taken that the accumulating liquid solution outside the hot area does not short-circuit the wiring, or suddenly floods the chip.

After deposition the film is amorphous and can be crystallized in an additional annealing step using the integrated heater or using an external furnace. Annealing was typically done at 600°C in air.

8.2.3 Thin Film Characterization

The microstructure was characterized by light microscopy (NIKON Optiphot 150) and scanning electron microscopy (SEM Leo 1530 and Philips XL30 ESEM-FEG). The film thickness was determined from cross section micrographs of the film on the micro-hotplate membrane and used to determine the electrical conductivity of the LSCF coating. The membrane deformation of the uncoated and coated micro- hotplates as a function of the input power was measured using an optical profilometer (UBM GmbH) with a resolution of 10 nm.

The micro-hotplate’s maximum power before breakdown was characterized using the Hewlett Packard 4155A semiconductor parameter analyser. The applied voltage of the device increased from 0 to 10 V with steps of 0.05 V and an integration time of 100 ms.

The electrical conductivity of the LSCF films deposited onto the micro-hotplate was measured in air in two-point and four-point probe configurations using the integrated Pt electrodes. The silicon chip was glued to a TO-5 socket and the contact pads of the heater and electrodes connected using wire bonding. The HP 6644A DC power supply was used to warm up the chip using the integrated heater. The control of the power supplied to the heater and the monitoring of the electrical resistance of the LSCF films from the multimeter HP 34401A were performed using Labview.

The electrical conductivity of the LSCF films deposited onto the sapphire bulk substrates, which served as reference samples, was measured in air in a modified tubular furnace. A four-point setup was used and the data was recorded with a multimeter (Keithley 2000) controlled through the Labview software installed on a computer. Flattened Pt wires (Johnson Matthey) were used as electrodes. To ensure 160 C H A P T E R 8 good adhesion of the electrodes, sputtered Pt and Pt paste (Heraeus) connected the electrodes to the film. Additionally, the electrodes were glued to the substrate outside the film using ceramic glue (Firag) with some Pt powder (Johnson Matthey) added for improved adhesion.

8.3 Results and Discussion

8.3.1 Microstructure

A typical SEM micrograph showing the top view and the cross section of an LSCF thin film deposited onto a micro-hotplate is shown in Fig. 8.2. The ridges found in the top view were built during the film deposition following the topology of the micro- hotplate, e.g. the heater and the electrodes. They also form the round features visible in Fig. 8.2 A). The mechanism of formation of these ridges is related to spray pyrolysis and is discussed in more detail elsewhere [12, 16]. The cross section image (Fig. 8.2 B)) shows that the films are about 1 µm thick and that the microstructure is porous.

Fig. 8.2: SEM micrographs of LSCF thin films on micro-hotplates. A) Top view, B) cross section.

Differences in the microstructure are observed with respect to the substrate material, as shown in Fig. 8.3. In the areas of the membrane where the SiN is covered by the Pt electrodes or where the Pt heater is embedded in the SiN membrane, fewer cracks are found than on areas consisting of SiN only. The reasons are the thermal properties of the substrate materials: if the product of thermal conductivity times heat capacity is large, as is the case for Pt, the heat transfer from the substrate to the droplets during film deposition is fast. That means a larger fraction of the non-uniform M I C R O – H O T P L A T E S F O R M I C R O – S O F C 161 sized droplets will show Leidenfrost’s phenomenon, i.e. hovering on their own steam, thus they do not contribute to film formation [17]. As a consequence, less material is deposited leading to thinner films, which reduces crack formation. In the case of a pure SiN substrate, the product of thermal conductivity times heat capacity is small and, hence, cracks easily form due to thicker films. By reducing the solution flow rate and deposition time, crack-free films can also be obtained on SiN substrates. However, if materials with strongly different thermal properties are combined in one substrate, the film thickness and thus the occurrence of cracks varies across the substrate. Thus the design of the metallic lines in the heater is important for the coherence of the film.

50 μm

Pt embedded in SiN below film

Pure SiN below film Pt electrode below film Fig. 8.3: SEM micrograph of LSCF deposited onto a micro-hotplate consisting of different materials.

To observe the microstructure of the films during annealing, the integrated heater of the micro-hotplate was used for different periods and then the microstructure was inspected by SEM. Micrographs at different areas of the sample were taken before annealing and after annealing periods of 5, 20 and 450 min at 600°C. The surface of an as-deposited film shows round grain-type features with a medium diameter of 162 C H A P T E R 8 around 500 nm and rim structures originating from the film formation from droplets, as shown in Fig. 8.4 A). After 5 min of annealing at 600°C, the round grain-type features of the film disappear and small cracks on the surface of the film are formed. After 20 min (Fig. 8.4 B)) or even 450 min annealing, no further changes appear in the microstructure. Generally, fewer cracks are found in the film when the sample is annealed in a furnace and not by using the integrated heater. In this case the whole chip is at the same temperature and so bending of the membrane is expected to be diminished, improving the integrity of the film. Furthermore, no significant microstructural difference between the films deposited onto micro-hotplates or sapphire substrates was observed if both films were annealed in external furnaces.

Fig. 8.4: SEM micrographs of LSCF on micro-hotplates. A) As-deposited. B) Annealed for 20 min.

When the integrated heater is used during film deposition, small LSCF crystals sometimes are found on top of the thin film. These crystals may be formed from liquid spray solution entering the membrane from the surrounding area, as explained in the experimental section. Film quality can be improved by avoiding spill-over of the liquid precursor during deposition.

8.3.2 Mechanical Stability

Bending of the micro-hotplates’ membrane upon heating was determined for samples with and without coating (namely LSCF and w/o coating, respectively Fig. 8.5). Both samples were flat up to 410°C (LSCF) and 340°C (w/o coating), respectively. When exceeding these temperatures, increased bending was observed with increasing temperature. Samples with LSCF coating bent less due to increased thickness of the membrane and thus increased stiffness. Further reduction of bending is expected M I C R O – H O T P L A T E S F O R M I C R O – S O F C 163 when replacing the SiN used for the micro-hotplate membrane by aluminum oxide

(Al2O3), due to the higher Young’s Modulus of Al2O3 of 345 – 409 GPa [18] vs. 230 – 265 GPa of SiN [19].

Fig. 8.5: Bending of micro-hotplates for different temperatures. One sample is coated with LSCF, the other one is without coating (w/o coating).

A maximum power test of an uncoated pure SiN micro-hotplate shows two failure modes: breaking of the membrane and electro-stress migration of the Pt atoms forming the heater, influenced by the elasticity and the strength of the membrane. This observation indicates that the membrane deformation and the migration of the Pt atoms are correlated [20]. A difference is noticed in the maximum power values obtained for the different electrode designs investigated. The maximum powers obtained in these testing conditions are 229 and 245 mW for the smaller and the larger electrodes, respectively. These powers can generate high temperature over 800°C on the micro-hotplate. The fact that the power of 120 mW is enough to heat the device up to 600°C offers a secure margin for the operation of the coated devices to be annealed on-chip.

164 C H A P T E R 8

8.3.3 Conductivity

In Fig. 8.6 A the electrical conductivity of an LSCF thin film deposited onto a micro- hotplate is compared to the conductivity from such a film deposited onto a bulk sapphire substrate as reference. The electrical conductivity was measured during annealing of the samples at 600°C. For the micro-hotplate, the integrated heater was used for annealing, whereas the sample on the bulk substrate was annealed using an external furnace. Both films show increasing conductivity with time with a comparable rate. We attribute this change in conductivity to the increasing amount of crystalline phase fraction in the films with increasing annealing time at 600°C. As was recently shown for spray pyrolysis films, high temperatures or long annealing times are needed in order to transform all amorphous material into a crystalline phase [21]. The crystalline material has a higher electrical conductivity than the amorphous.

After about 30 min of annealing, the film deposited onto the micro-hotplate already reached the conductivity that the film on the bulk substrate reached after 4 hrs of annealing, meaning the film on the flexible micro-hotplate crystallized faster than the film constrained by the rigid sapphire substrate. A similar relation between microstrain and crystallinity has recently been reported [21].

In Fig. 8.6 B), the electrical conductivities of two LSCF films are compared, which were both already annealed using an external furnace, but again one film was deposited onto a micro-hotplate and one onto a bulk sample as a reference. The electrical conductivities for different temperatures during cooling were measured using the integrated heater for the micro-hotplate-based sample and the furnace for the reference sample. With the sample deposited onto a bulk substrate no cooling rate faster than 3°C/min was realized due to the latent heat of the furnace. The micro-hotplate reacted immediately to a change in input power, thus it was cooled stepwise. Both samples show similar electrical conductivity at the target temperature, indicating that the LSCF thin film stays coherent on the micro-hotplate. However, the activation energy differs by a factor of about two. The LSCF thin film on the bulk substrate shows an activation energy of 0.12 eV in the temperature range of 200 – 500°C, which is close to the one reported in literature of 0.10 eV for bulk LSCF in the temperature range of 100 – 500°C [22]. In contrast, the LSCF thin film on the micro- hotplate shows an activation energy of 0.26 eV in the range of 200 – 500°C. The M I C R O – H O T P L A T E S F O R M I C R O – S O F C 165 reason for the difference might be a loss of film integrity upon cooling, which caused higher resistance and thus higher apparent activation energy.

Fig. 8.6: A): Electrical conductivity during annealing of LSCF thin films, deposited onto micro- hotplate and onto bulk substrate (sapphire), respectively. B): Conductivity vs. reciprocal temperature.

For LSCF thin films on bulk substrates, a low degradation of 0.5 to 1 % increase in resistance per 100 hrs was found when the samples were kept at 600°C. However, for LSCF thin films on micro-hotplates the same increase in resistance was found in 166 C H A P T E R 8 about 20 hrs. Based on this pronounced difference between both types of samples, the degradation is most likely not only caused by the thin film material itself. Instead, it may be associated with the entire system. As already pointed out in section 8.3.2, bending of the hotplate might affect the integrity of the thin film. Furthermore, the mismatch in thermal expansion coefficient (TEC) of α = 1.6·10-6/K [23] for the membrane material (SiN) and α = 14·10-6/K [24] for the thin film might also be critical to the film. It is expected that the performance can be significantly improved by using

Al2O3 as the membrane material. Besides a reduction of bending, the TEC of Al2O3 with α = 7.8·10-6/K [25] is also beneficial and should help to maintain the performance for longer time.

Fig. 8.7 Electrical resistance of thin films at 620°C as a function of annealing time and cycling (vertical bars indicate rapid temperature cycles).

Thermal cycling is a critical issue for operation of SOFCs and was addressed with the following experiment. A sample deposited onto a micro-hotplate was exposed to several rapid (10°C/s) temperature cycles from room temperature to 620°C. The resistance was recorded versus time as shown in Fig. 8.7. The vertical lines indicate fast thermal cycling steps to room temperature resulting in high resistance; the other resistance values are taken at 620°C. The resistance increases with every cooling M I C R O – H O T P L A T E S F O R M I C R O – S O F C 167 and heating cycle but stays in the same order of magnitude even after 100 hrs and 23 cycles. The increase in resistance with every cycle is also observed when thin films deposited onto bulk substrates are thermally cycled, although in this case cycling is performed with only 3°C/min. For these samples only a few thermal cycles have been done because each cycle takes 200 times longer compared to the micro- hotplate-based samples. Bending, mismatch in TEC and perhaps aging of the material become more important in thermal cycling than in long-term tests at constant temperature, thus degradation is faster during cycling. The unique ability of micro- hotplates to perform these fast temperature cycles creates the possibility to perform accelerated aging tests with placing heavy demands on the materials. This drastically speeds up degradation tests, which are otherwise very time consuming.

8.4 Summary and Conclusion

The feasibility to combine LSCF thin films, a typical SOFC cathode material, with micro-hotplates, a technology platform initially developed in the field of sensors, has been demonstrated in this work.

The micro-hotplates were able to withstand temperatures above 800°C, which offers a secure margin to the target operating temperature of 600°C.

The functionality of the LSCF thin film under these conditions was proven by obtaining the same electrical conductivity as for reference samples deposited onto bulk substrates. More rapid increase in the resistance of samples deposited onto micro-hotplates as a function of time was attributed to mismatch in TEC and bending of the micro-hotplate. Both might be diminished by using Al2O3 instead of SiN as the membrane material, since Al2O3 offers superior stiffness and a TEC closer to the values of LSCF and other fuel cell materials.

The integrated heater and the low heat capacity of the micro-hotplate allow fast heating rates up to 10°C/s leading to short annealing times. After 5 min of annealing the film on the chip did not further change the microstructure even with longer annealing times. The conductivity data indicated that 30 min annealing on-chip leads to the same result as almost 4 hrs annealing on a bulk substrate, since the flexible membrane allows faster crystallization of the thin film. 168 C H A P T E R 8

When the integrated heater was used during film deposition, selective film growth in the heated areas only was obtained and thus no masking was required. However, flooding of the membrane from the surrounding spray solution outside the heated area has to be avoided.

In conclusion, the micro-hotplate is a suitable platform for µSOFC. Deposition of an electrolyte and anode as well as holes in the membrane for gas access would be the next steps to integrate a whole µSOFC on a micro-hotplate.

8.5 References

[1] D. Briand, A. Krauss, B. van der Schoot, U. Weimar, N. Barsan, W. Gopel, N. F. de Rooij, "Design and Fabrication of High-Temperature Micro-Hotplates for Drop-Coated Gas Sensors", Sensors and Actuators B, 68, [1-3] 223-33 (2000).

[2] D. Barrettino, M. Graf, W. H. Song, K. U. Kirstein, A. Hierlemann, H. Baltes, "Hotplate-Based Monolithic CMOS Microsystems for Gas Detection and Material Characterization for Operating Temperatures up to 500(circle)C", Ieee Journal of Solid-State Circuits, 39, [7] 1202-7 (2004).

[3] M. Heule, L. J. Gauckler, "Miniaturised Arrays of Tin Oxide Gas Sensors on Single Microhotplate Substrates Fabricated by Micromolding in Capillaries", Sensors and Actuators B, 93, [1-3] 100-6 (2003).

[4] I. Simon, N. Barsan, M. Bauer, U. Weimar, "Micromachined Metal Oxide Gas Sensors: Opportunities to Improve Sensor Performance", Sensors and Actuators B, 73, [1] 1-26 (2001).

[5] D. Briand, S. Heimgartner, M. A. Gretillat, B. van der Schoot, N. F. de Rooij, "Thermal Optimization of Micro-Hotplates that have a Silicon Island", Journal of Micromechanics and Microengineering, 12, [6] 971-8 (2002).

[6] R. M. Tiggelaar, J. W. Berenschot, J. H. de Boer, R. G. P. Sanders, J. G. E. Gardeniers, R. E. Oosterbroek, A. van den Berg, M. C. Elwenspoek, "Fabrication and Characterization of High-Temperature Microreactors with Thin Film Heater and Sensor Patterns in Silicon Nitride Tubes", Lab on a Chip, 5, [3] 326-36 (2005). M I C R O – H O T P L A T E S F O R M I C R O – S O F C 169

[7] W. Konz, J. Hildenbrand, M. Bauersfeld, S. Hartwig, A. Lambrecht, V. Lehmann, J. Wöllenstein, "Micromachined IR-source with Excellent Blackbody like Behaviour"; pp. 540-8 in Proceedings of Smart Sensors, Actuators, and MEMS II 5836 Edited by C. Cane, J.-C. Chiao, F.V. Verdu, SPIE, (2005).

[8] C. H. Chen, H. J. M. Bouwmeester, H. Kruidhof, J. E. ten Elshof, A. J.

Burggraaf, "Fabrication of La1-xSrxCoO3-delta Thin Layers on Porous Supports by a Polymeric Sol-Gel Process", Journal of Materials Chemistry, 6, [5] 815-9 (1996).

[9] P. Bohac, L. J. Gauckler, "Chemical Spray Deposition of YSZ and GCO Solid Electrolyte Films", Solid State Ionics, 119, [1-4] 317-21 (1999).

[10] I. Taniguchi, R. C. van Landschoot, J. Schoonman, "Fabrication of

La1-xSrxCo1-yFeyO3 Thin Films by Electrostatic Spray Deposition", Solid State Ionics, 156, [1-2] 1-13 (2003).

[11] A. Bieberle-Hütter, D. Beckel, U. P. Muecke, J. L. M. Rupp, A. Infortuna, L. J. Gauckler, "Micro-Solid Oxide Fuel Cells as Battery Replacements", Mstnews, 04/05, 12-5 (2005).

[12] D. Beckel, A. Dubach, A. R. Studart, L. J. Gauckler, "Spray Pyrolysis of

La0.6Sr0.4Co0.2Fe0.8O3-d Thin Film Cathodes", Journal of Electroceramics, 16, [3] 221-8 (2006).

[13] D. Perednis, L. J. Gauckler, "Thin Film Deposition Using Spray Pyrolysis", Journal of Electroceramics, 14, [2] 103-11 (2005).

[14] MicroChem, LOR Lift-Off resists, www.microchem.com/products/pdf/lor_data_sheet.pdf, (2006).

[15] L. Thiery, D. Briand, A. Odaymat, N. F. de Rooij, "Contribution of Scanning Probe Temperature Measurements to the Thermal Analysis of Micro- Hotplates"; pp. 23-8 in Proceedings of International Workshops on Thermal Investigations of ICs and Systems, Therminic (2004).

[16] D. Beckel, D. Briand, A. R. Studart, N. F. de Rooij, L. J. Gauckler, "Topography Mediated Patterning of Inorganic Materials by Spray Pyrolysis", Advanced Materials, 18, [22] 3015-8 (2006). 170 C H A P T E R 8

[17] U. P. Muecke, G. L. Messing, L. J. Gauckler, "The Leidenfrost Effect During

Spray Pyrolysis of Dense NiO-Ce0.8Gd0.2O1.9-x Thin Films", Submitted to Thin Solid Films, (2006).

[18] J. F. Shackelford, W. Alexander, "CRC Materials Science and Engineering Handbook" CRC Press, Boca Raton p. 790 (2001).

[19] D. Schneider, M. D. Tucker, "Non-Destructive Characterization and Evaluation of Thin Films by Laser-Induced Ultrasonic Surface Waves", Thin Solid Films Papers presented at the 23rd International Conference on Metallurgical Coatings and Thin Films, 290-291, 305-11 (1996).

[20] D. Briand, F. Beaudoin, J. Courbat, N. F. De Rooij, R. Desplats, P. Perdu, "Failure Analysis of Micro-Heating Elements Suspended on Thin Membranes", Microelectronics and Reliability, 45, 1786-9 (2005).

[21] J. L. M. Rupp, A. Infortuna, L. J. Gauckler, "Microstrain and Self-Limited Grain Growth in Nanocrystalline Ceria Ceramics", Acta Materialia, 54, [7] 1721-30 (2006).

[22] L.-W. Tai, M. M. Nasrallah, H. U. Anderson, D. M. Sparlin, S. R. Sehlin,

"Structure and Electrical Properties of La1-xSrxCo1-yFeyO3. Part 2. The System

La1-xSrxCo0.2Fe0.8O3", Solid State Ionics, 76, [3-4] 273-83 (1995).

[23] M. J. Madou, "Fundamentals of Microfabrication, The Science of Miniaturization" CRC Press, Boca Raton p. 300 (2002).

[24] B. C. H. Steele, "Interfacial Reactions Associated with Ceramic Ion Transport Membranes", Solid State Ionics, 75, 157-65 (1995).

[25] J. F. Shackelford, W. Alexander, "CRC Materials Science and Engineering Handbook" CRC Press, Boca Raton p. 643 (2001).

9 Stability of NiO Membranes on Photostructurable Glass

Substrates for Micro Solid Oxide Fuel Cells

NiO thin films were deposited by spray pyrolysis onto photostructurable glass substrates and, ultimately, free- standing NiO membranes with diameters of 100, 200 and 300 μm were fabricated. The membranes are intended to act as simplified anodes or anode current collectors in micro solid oxide fuel cells (μSOFCs) and their differential pressure and thermal stability were characterized. The membranes tolerated a differential pressure between 13700 and 158600 Pa. Smaller membranes showed more pressure tolerance than larger membranes. A membrane diameter of 100 μm and a film thickness of about 500 nm turned out to be a promising geometry for μSOFC membranes. All membranes survived temperatures higher than the intended operating temperature of μSOFCs (350 – 600°C). We attribute the good thermal stability to the match of the thermo-mechanical properties of the substrate and the NiO thin films for the lower temperature regime and the substrate softening at higher temperatures releasing stresses in the thin films. Furthermore, the thermal expansion of the substrate is close to thermal expansion of materials used in SOFCs and circular geometries can be realized using wet etching.

D. Beckel, U. P. Muecke, B. Schoeberle, P. Mueller L. J. Gauckler, submitted to Thin Solid Films, (2007). 172 C H A P T E R 9

9.1 Introduction

Portable electronic devices such as laptops, personal digital assistants (PDAs) or mobile phones have an increasing energy consumption due to the integration of more and more features into one device, thus it becomes difficult to provide sufficient energy for reasonable running times with batteries [1, 2]. This need for small scale energy supply created substantial interest in miniaturized fuel cells, which are an attractive alternative to batteries since they can offer more energy per volume and weight than batteries [3, 4]. Furthermore, they offer the possibility of instant refueling. Compared to polymer-based fuel cells that run on pure hydrogen or diluted methanol, solid oxide fuel cells (SOFCs) offer the advantage to run on a variety of fuels, e.g. hydrogen, CO, or natural gas and do not need precious metal catalysts due to the elevated operating temperatures of 500 to 600°C.

Like large scale SOFCs, micro SOFCs (μSOFCs) consist of two gas compartments, one for fuel and one for air. When the oxygen contained in the air enters the cathode, oxygen is reduced to O2- ions, which are the only charged species that can migrate through the gas-tight electrolyte that separates the air compartment from the fuel 2- compartment. In the anode, the O ions recombine with H2 from the fuel to form water and release their electrons to an external circuit. In a μSOFC all three components (cathode, electrolyte and anode) are fabricated in the form of thin films and combined in a membrane of 150 to 10000 nm thickness that spans over 5 to 500 μm.

Several studies have been conducted to fabricate and characterize single thin film components for μSOFCs regarding their preparation and microstructure [5-16], electrical and electrochemical properties [5, 17-28] and suitability for application in a μSOFC. In some cases a deviation in behavior from bulk materials was found, e.g. self-limiting grain growth [13] and porosity development during annealing [15] was observed. However, the electrochemical performance of thin film electrodes did not differ significantly from thick film electrodes [25, 28]. Few studies have been dedicated to modeling [29] and fabrication of an entire cell [30-32] but further studies are needed to understand the behavior of μSOFCs.

As can be seen from the construction of a μSOFC, the stability of the membrane is a critical issue, since failures in the membrane result in a breakdown of the entire cell: S T A B I L I T Y O F N I O M E M B R A N E S 173 when fuel and air mix through a broken membrane, the driving force for a fuel cell (the difference in oxygen partial pressure between both gas compartments) is diminished. The delicate membrane is exposed to different gas atmospheres on both sides, which creates the risk that the membrane has to withstand a pressure difference. Additionally, the operating temperatures (350 – 600°C) of such devices place heavy demands on the temperature stability of the membrane. This limits the designs of free-standing membranes and their size according to calculations regarding μSOFC membranes [33, 34]. For membranes with 1 μm thickness, fabricated using yttria-stabilized zirconia (YSZ), a typical electrolyte material, a maximum diameter of only 34 μm was calculated for operation at 600°C when Si is used as substrate and intrinsic stress is negligible [34]. However, material behavior sometimes proved to be more complex than expected, e.g. membranes showed buckling although the thin film showed tensile stress before membrane release [35], or initially buckled membranes flattened upon heating, although the thermal expansion coefficient of the membrane material was larger than of the substrate material [36]. These observations show the limits simulating real materials behavior for μSOFCs designs as the local properties of these nanostructured materials behave differently than their micro grain-sized counterparts. Therefore, experimental studies have been conducted to evaluate the thermal stability of such membranes. Free- standing electrolytes served as model for μSOFC membranes and showed limits in operating temperature depending on the material, its preparation and the membrane geometry, in the range of 150 to 500°C [35, 36]. Since power output increases strongly with operating temperature, operating temperatures up to 600°C would be desirable. To our knowledge no experimental studies exist so far on the differential pressure stability of μSOFC membranes.

In this study we investigate the differential pressure and thermal stability of NiO membranes, which can serve as simplified anodes and as current collectors in μSOFCs. We chose the anode layer instead of an electrolyte layer for investigation since we aim to build a μSOFC where all three layers are deposited onto one side of the substrate, which facilitates the μSOFCs fabrication. Instead of Ni or Si, which have been used as substrates in the past [30, 31, 36, 37], we use Foturan®, a photostructurable glass ceramic. Foturan® has so far been used for fabrication of various microreactors [38-41] including reformers for the pretreatment of fuel for 174 C H A P T E R 9 micro fuel cells [42, 43] and for optical applications [44-48]. Bonding of Foturan® and metals, a prerequisite for device packaging, has also been investigated [49] but to our knowledge no publications exist that show Foturan® acting as a supporting structure for thin film membranes up to now.

9.2 Experimental

NiO thin films were deposited by air pressurized spray pyrolysis [11, 12] on photostructurable glass ceramic substrates. The substrate was subsequently selectively etched to release the NiO membranes. The membranes were then characterized with respect to differential pressure and temperature stability. A detailed description of each process step is given in the following sections.

9.2.1 Thin Film Deposition

The salt solution for spray pyrolysis of the NiO thin films was prepared as follows:

0.1 mol of Ni(NO3)2·6H2O (Fluka, purity > 97 %) was dissolved in one liter solvent consisting of 1/10 (vol. fraction) ethanol (Merck, ≥ 99.9 %) and 9/10 tetraethylenglycol (Aldrich, ≥ 99 %). The solution was pumped at a flow rate of 5 ml/h through a nozzle and atomized by air (1 bar) using an air blast nozzle (Binks 460, Glendale Heights, Il, USA). The nozzle was located 39 cm above a heating plate, on which the substrates to be coated were placed. The heating plate was controlled to maintain a constant temperature. The substrate surface temperature was 410°C as measured with a type K surface probe (Omega Model 88108). The spraying time was varied from 30 to 120 min to achieve films of varying thickness. More details on the spray pyrolysis setup can be found elsewhere [11, 12].

9.2.2 Micromachining of the Substrate

Substrates were fabricated from Foturan® (Mikroglas, Mainz, Germany) with a ® thickness of 300 μm and polished surfaces on both sides. Foturan is a SiO2-based glass containing many additives to achieve photosensitivity. More details about the composition can be found elsewhere [50]. The substrate was cut (Disco DAD 321 S T A B I L I T Y O F N I O M E M B R A N E S 175 automatic dicing saw, cutting speed 2 mm/s, spindle rotation 30000 rpm) into pieces of 10 by 10 mm.

A foil mask (3800 dpi, Salinger AG, Zurich, Switzerland) with one round opening of 100, 200 or 300 μm diameter in the middle of the square substrates was used for illumination of the Foturan®. For preliminary tests to fabricate membranes other mask geometries were also used, giving other patterns as shown in Fig. 9.1. A quartz plate was put on top of the foil mask to prevent it from bending. The substrates were illuminated for 180 min using an Hg lamp (LOT Oriel 66905 mercury arc lamp with 69910 power supply) at a set power of 450 W with collimated light. An energy density of 20 J/cm2 at a wavelength of 312 nm is recommended for optimal results by the manufacturer [51] and matches with our experience when no mask is used. However due to the absorption of UV light by the foil mask it was necessary to increase the energy dose (before the mask) to about 1400 J/cm2, which is accounted for with the long exposure time given above. During light exposure, the Ce3+ ions contained in the Foturan oxidize to Ce4+, releasing electrons that reduce the Ag+ ions to Ag atoms [50] used as nucleation agents during the glass to ceramic conversion step.

Fig. 9.1: A) Foturan after selective light exposure and annealing. The brown parts are crystalline, the rest is glassy and transparent (the texture is from the microscope table below). B) NiO membrane on Foturan near the end of etching, the black parts are remaining Foturan crystals. C): Array of NiO membranes.

After light exposure of the substrate, the NiO thin films were deposited and then the Foturan was annealed with heating and cooling rates of 1°C/min and subsequent dwell times of 1 h at 500 and 600°C. During the dwell at 500°C, the Ag atoms agglomerate to form larger nuclei in the radiated areas only. At the 600°C dwell,

Li2SiO3 crystallizes around these nuclei to crystal sizes in the range of 1 to 10 μm 176 C H A P T E R 9

[50]. During the annealing, the chips were pressed between two aluminum oxide plates, giving a pressure of 2.2 g/cm2. The weights were necessary to keep the substrates flat during annealing; however, too much weight bonded the samples to the aluminum oxide plates. After annealing a clear contrast between the brownish crystallized parts and the transparent still glassy part can be seen, as depicted in Fig. 9.1 A).

The chips with the thin films on top were heated for 30 min at 120°C on a heating plate to remove any water from the surface. Then, the thin films were coated with a protective coating (Microposit surface coating FSC-H from Rohm and Haas, Coventry, UK). The protective coating was applied using a brush, so a thick coating with good adherence was achieved. Subsequently, the protective coating was baked for 45 min at 100°C on the hotplate. Etching was carried out in a bath of 10 % HF (diluted from 40 % HF, Merck, suprapur, not buffered), which was continuously stirred at about 200 rpm. The chips were placed on a holder with the protective coating facing upwards. The holder was a Teflon® plate with holes through which the etching bath had access to the areas to be etched. A weight was placed on the top side to prevent the chips from moving in the stirred bath. During etching the HF attacks the crystallized Foturan® area 20 times faster then the surrounding glass. It has been proposed that the crystalline Li2SiO3 dissolves faster than the glass [52], however, we believe the etching process to attack the grain boundary material of the crystallized area rather than the grains, since we frequently observe some crystals remaining in the holes after etching and we found residues in the etching bath when etching larger amounts of Foturan®. We assume that the Li dispersed in the Foturan® stabilizes the ® SiO2 matrix against HF attack. Upon crystallization of the Foturan forming Li2SiO3 crystals, Li diffuses towards these crystals and consequently Li is depleted in the surrounding grain boundaries. The non-stabilized SiO2 matrix is readily attacked by HF, thus the crystals fall out of the substrate but are not necessarily dissolved. This grain boundary etching process led to varying etching times, because the velocity of material removal from the hole depends on the local fluid dynamics of the etching solution, which is not reproducible as long as the fluiddynamics in the etching bath are not properly controlled as it was the case in our experiments. Using ultrasound agitation instead of a stirrer gave more reproducible etching times but could not be used for fabricating membranes, since the ultrasound destroyed the membranes. An S T A B I L I T Y O F N I O M E M B R A N E S 177 average etching time of 22 min for membranes of 100 μm diameter and 17 min for membranes of 200 and 300 μm diameter were necessary. However, variations in the etching time as large as 50 % were observed. The progress of the etching was observed by inspection of the membranes with a light microscope (Nikon Eclipse L200). As shown in Fig. 9.1 B), a clear contrast between the remaining Foturan® crystals and the membrane could be observed. Fig. 9.1 C) shows an array of NiO membranes after etching. The NiO itself is an excellent etch stop for HF: no decrease in NiO film thickness with HF exposure times up to 75 min could be measured and no HF attack on the NiO microstructure was observed as shown in Fig. 9.2.

Fig. 9.2: SEM image of a NiO thin film exposed for 75 min to 10 % HF.

The protective coating was removed by putting the chips in acetone (Merck, purity ≥ 99.8 %) for about 3 to 5 min while stirring slightly by hand. Then, the chips were placed in a second acetone bath followed by a water bath.

9.2.3 Thin Film Characterization

9.2.3.1 Scanning Electron Microscopy and Light Microscopy

Scanning electron microscopy (SEM, Leo 1530 Carl Zeiss SMT) was used to determine the film thickness from cross-sectional images, as shown for example in Fig. 9.3. SEM images were taken with an in-lens detector using 5 kV acceleration 178 C H A P T E R 9 voltage. Light microscopy (Nikon Eclipse L200 and Reichert-Jung Polyvar Met) was used to determine the etching progress and to take images of the entire membrane.

Fig. 9.3: Cross section of NiO thin film as used for determination of the film thickness.

9.2.3.2 Differential Pressure Stability

To evaluate the differential pressure stability of the membrane, a chip was fixed with double-sided adhesive tape on a pressure chamber from a bulge test setup [53] with an opening below the membrane, see Fig. 9.4.

Fig. 9.4: Sketch of the experimental setup for determination of the differential pressure stability of the membranes.

The pressure is applied with dry and filtered air by a digital flow controller (Bronkhorst P-602C) and increased in steps of approximately 1000 Pa. The pressure difference between the chamber and the surrounding is measured independently with a high- precision pressure sensor (Burster 8262). A current meter (Keithley 6485) digitalizes S T A B I L I T Y O F N I O M E M B R A N E S 179 the sensor output. All components are controlled and read-out with LabView. Fracture of the membrane was determined visually.

9.2.3.3 Thermal Stability

Chips with free-standing NiO membranes were placed on a custom-made heating plate and heated at a rate for 20°C/min up to 630°C. The membranes were constantly observed to detect fracture. The temperature was measured with a type K surface probe (Omega Model 88108).

9.3 Results and Discussion

9.3.1 Differential Pressure Stability

Table 9.1 lists the geometry of the membranes and the maximum differential pressure and temperature that the membranes withstood. Fig. 9.5 shows a plot of fracture pressure vs. the ratio of film thickness to membrane diameter. Generally, the thicker the membrane or the smaller its diameter, the more differential pressure it could tolerate. However some scattering of the data is observed. Two reasons related to the thin film preparation may contribute to scattering of the maximum pressure difference that the films can withstand: (i) During spray deposition, droplets of different sizes (5 to 50 μm in diameter [11, 12]) randomly hit the substrate to form the thin film. Thereby local inhomogeneities could be created, which might lead to variations of the intrinsic film stress or even local flaws. Depending on the size of the largest flaw in the stressed film, fracture under external load occurs according to Griffith’s law [54, 55]:

1 E(2γ) σ = · (9.1). F Y c

σF refers to the fracture load of the free-standing membrane; Y is a geometrical factor of the membrane, γ is the effective surface energy for crack initiation, E the elastic modulus and c the maximum crack length in the stressed cross section of the membrane. As the maximum flaw size c differs widely in these films, their maximum load bearing capacity can be expected to differ accordingly. 180 C H A P T E R 9

Fig. 9.5: Maximum differential pressure vs. membrane thickness / diameter.

On the other hand, residual stresses in the films after deposition may contribute to the outside applied stress in an uncontrolled manner. It was not possible to evaluate the amount of residual stress in the film after preparation since the substrate softens at temperatures slightly above the film preparation temperature (glass transition temperature 465°C [51]) and below the film annealing temperature. Therefore substrate curvature measurements would not represent the stress correctly. (ii) Pyrolysis is associated with evaporation and decomposition of precursor material and, therefore, tensile stress can be expected in the films. The thicker the film, the more tensile stresses are created. It was already shown that NiO films fully supported on a substrate and prepared by spray pyrolysis form cracks in the late stage of deposition if they are too thick [11, 12]. This means that there are two counter-balancing phenomena in our loading experiments: on one hand the membrane becomes more stable if it is thicker [33, 34], on the other hand the size of the maximum flaw may become larger and the amount of residual stress in the thin film increases with its thickness. Consequently, there is an optimal thickness for stability. According to Table 9.1, this is about 500 nm. The results presented in Table 9.1 also clearly indicate that smaller membranes are more tolerant to differential pressure than larger S T A B I L I T Y O F N I O M E M B R A N E S 181 ones. The values obtained for membranes with 100 μm diameter (all ≥ 13700 Pa) are sufficient for application in a μSOFC.

Table 9.1: Geometry, maximum differential pressure and temperature of NiO membranes on Foturan® substrates.

Film thickness/ Max temperature Diameter/μm Max pressure/ Pa nm /°C

100 108 - 621

130 209 8000 -

290 106 13700 -

290 108 - > 630

290 104 14300 -

300 122 68600 -

470 212 - > 630

470 111 158600 -

480 130 46000 -

490 100 101900 -

630 226 7200 -

670 302 - > 630

670 302 3000 -

700 98 21000 -

900 122 69300 - 182 C H A P T E R 9

9.3.2 Thermal Stability

Foturan® in the glassy state is recommended by the manufacturer for use up to 450°C. In the fully ceramic state the maximum application temperature is limited to 750°C [51]. In other experiments we clamped larger (24 by 24 mm2) pieces of glassy Foturan® at their edges and annealed them for several hours up to 600°C. The shape of the Foturan® remained stable so for the thermal stability tests of the membrane the Foturan® was maintained in the glassy state. In any case, temperatures exceeding 600°C are not targeted for μSOFCs.

As can be seen from Table 9.1 only one out of four membranes broke in the investigated temperature range up to 630°C. Higher temperatures were not investigated since the substrate material starts to deform substantially at higher temperatures if it is not fully crystalline. Even the one membrane that broke showed enough thermal stability for application in a μSOFC, as it survived up to 621°C. Compared to other investigations, mainly carried out for electrolyte materials on Si substrates, these results are very encouraging.

Nair et al. [35] prepared membranes by sputtering of CeO2, a material that is used as SOFC electrolyte when doped with 10 to 20 at % Gd, Sm, or Y. In this case Si single crystals were used as substrates and membranes of 1.5 μm thickness and 100 by 300 μm2 in size were prepared. Depending on the ratio of oxygen to argon pressure during sputtering, the membranes were either flat or buckled after substrate removal. The flat membranes disintegrated at 150°C, whereas the buckled membranes withstood heating to 220°C multiple times, however, a maximum temperature for these membranes was not given.

Bruschi et al. [37] fabricated membranes by sputtering YSZ and also used Si as a substrate. Depending on their micromachining process they were able to fabricate square membranes with a maximum side dimension of 100 or 170 μm. Specific thermal stability tests were not conducted, but it was reported that cooling from the temperature employed during microfabrication (115°C) was a critical step for the integrity of the membrane.

Baertsch et al. [36] published an extensive study on the thermal stability of YSZ and

Ce0.9Gd0.1O1.95 (CGO) square membranes with 75 to 1000 μm side length prepared by electron beam evaporation and sputtering on Si substrates. Generally they found S T A B I L I T Y O F N I O M E M B R A N E S 183 that YSZ membranes are more robust than CGO membranes and sputtering produces more robust membranes than electron beam evaporation. The maximum temperature their membranes could withstand in case they were prepared by electron beam evaporation was 408°C for a 75 μm membrane. The thermal stability basically decreases with decreasing ratio of film thickness to membrane size. Sputtered membranes survived annealing at 500°C and showed initially compressive stress that turned into tensile stress upon annealing.

Three reasons might be responsible for the good thermal stability of the membranes presented in this work. (i) The Foturan® substrate used in this work softens upon annealing as it has its glass transition temperature at 465°C [51], which offers the unique possibility to reduce stresses caused during thin film deposition and those arising during heating. (ii) The thermal expansion coefficient of Foturan® of α = 8.6·10-6/K in the glassy state and α = 10.5·10-6/K in the ceramic state [51] is only a factor of ~1.6 smaller than that of the NiO membranes of α = 14.1·10-6/K [56]. In contrast, between Si with α = 2.6·10-6/K [57] and YSZ with α = 10.3·10-6/K [56] there is a mismatch of a factor of ~4. (iii) Lastly, we used circular membranes, whereas all other studies employed rectangular membranes with sharp edges in which stress concentrations at the sharp edges in the two dimensional stress state have to be expected susceptible to crack formation. Due to preferential etching along crystal planes in cubic Si single crystals, only rectangular features can be etched with a high aspect ratio when using anisotropic wet etching. Holes in Foturan® on the other hand, can have any shape.

We do not attribute the better thermal stability to the fact that NiO was used in this study, while YSZ was used in most of the other studies [36, 37]. On the contrary, taking into account the material’s properties, the use of YSZ should lead to even more stable membranes than NiO, since the fracture strength of YSZ is 236 MPa [58] compared to ~52 MPa for NiO [59]. The fracture strength of NiO is representative for composite anodes consisting of Ni and YSZ, which have a fracture strength of about 55 MPa [60]. We take the fracture strength for this rough comparison of the material’s properties, since it is the primary criterion when tensile stress is considered for membrane failure [34].

184 C H A P T E R 9

9.4 Summary

Circular NiO membranes were fabricated on Foturan® for application as anode or anode current collector in a μSOFC and their differential pressure and thermal stability were determined. The pressure tolerance for membranes of 100 μm diameter was ≥ 13700 Pa for all membrane thicknesses investigated, which is sufficient to withstand pressure variations occurring during the operation of a μSOFC. Generally, thicker membranes proved to be more stable, however, the membrane thickness should be optimized, since the amount of stress increases with film thickness.

The membranes showed higher thermal stability (≥ 620°C) than membranes fabricated on Si. We attribute this to enhanced stress relaxation during annealing on the soft Foturan® substrate, to better adjusted thermal properties of thin film and substrate and to the circular geometry. Due to their good differential pressure and thermal stability, we expect these membranes to be a reliable basis for entire μSOFCs.

9.5 References

[1] A. Bieberle-Hütter, D. Beckel, U. P. Muecke, J. L. M. Rupp, A. Infortuna, L. J. Gauckler, "Micro-Solid Oxide Fuel Cells as Battery Replacements", Mstnews, 04/05, 12-5 (2005).

[2] D. Nikbin, "Micro SOFCs: Why Small is Beautiful", The Fuel Cell Review, April/May, 21-4 (2006).

[3] H. L. Maynard, J. P. Meyers, "Miniature Fuel Cells for Portable Power: Design Considerations and Challenges", Journal of Vacuum Science and Technology B, 20, [4] 1287-97 (2002).

[4] C. K. Dyer, "Fuel Cells for Portable Applications", Journal of Power Sources, 106, [1-2] 31-4 (2002).

[5] T. Tsai, S. A. Barnett, "Sputter Deposition of Cermet Fuel Electrodes for Solid Oxide Fuel Cells", Journal of Vacuum Science and Technology A, 13, [3] 1073-7 (1995). S T A B I L I T Y O F N I O M E M B R A N E S 185

[6] G. J. la O, J. Hertz, H. Tuller, Y. Shao-Horn, "Microstructural Features of RF- Sputtered SOFC Anode and Electrolyte Materials", Journal of Electroceramics, 13, [1-3] 691-5 (2004).

[7] M. L. Liu, D. S. Wang, "Preparation of La1-zSrzCo1-yFeyO3-x Thin-Films, Membranes, and Coatings on Dense and Porous Substrates", Journal of Materials Research, 10, [12] 3210-21 (1995).

[8] C. Argirusis, T. Damjanovic, G. Borchardt, "Preparation of SOFC Cells by Means of Electrophoretic Deposition", Key Engineering Materials, 314, 101-6 (2006).

[9] C.-Y. Fu, C.-L. Chang, C.-S. Hsu, B.-H. Hwang, "Electrostatic Spray

Deposition of La0.8Sr0.2Co0.2Fe0.8O3 Films", Materials Chemistry and Physics, 91, [1] 28-35 (2005).

[10] I. Taniguchi, R. C. van Landschoot, J. Schoonman, "Fabrication of

La1-xSrxCo1-yFeyO3 Thin Films by Electrostatic Spray Deposition", Solid State Ionics, 156, [1-2] 1-13 (2003).

[11] U. P. Muecke, N. Luechinger, L. J. Gauckler, "Initial Status of Deposition and Film Formation During Spray Pyrolysis of Nickel Oxide, Cerium Gadolinium Oxide and NiO-CGO Thin Films", Submitted to Thin Solid Films, (2006).

[12] U. P. Muecke, G. L. Messing, L. J. Gauckler, "The Leidenfrost Effect During

Spray Pyrolysis of Dense NiO-Ce0.8Gd0.2O1.9-x Thin Films", Submitted to Thin Solid Films, (2006).

[13] J. L. M. Rupp, A. Infortuna, L. J. Gauckler, "Microstrain and Self-Limited Grain Growth in Nanocrystalline Ceria Ceramics", Acta Materialia, 54, [7] 1721-30 (2006).

[14] D. Beckel, D. Briand, A. R. Studart, N. F. de Rooij, L. J. Gauckler, "Topography Mediated Patterning of Inorganic Materials by Spray Pyrolysis", Advanced Materials, 18, [22] 3015-8 (2006).

[15] D. Beckel, A. Dubach, A. N. Grundy, A. Infortuna, L. J. Gauckler, "Solid State

Dewetting of La0.6Sr0.4Co0.2Fe0.8O3±d Thin Films during Annealing", accepted by Journal of the European Ceramic Society, (2007). 186 C H A P T E R 9

[16] D. Beckel, A. Dubach, A. R. Studart, L. J. Gauckler, "Spray Pyrolysis of

La0.6Sr0.4Co0.2Fe0.8O3-d Thin Film Cathodes", Journal of Electroceramics, 16, [3] 221-8 (2006).

[17] J. L. Hertz, H. L. Tuller, "Electrochemical Characterization of Thin Films for MicroSolid Oxie Fuel Cells", Journal of Electroceramics, 13, 663-8 (2004).

[18] L. S. Wang, S. A. Barnett, "Sputter-Deposited Medium-Temperature Solid Oxide Fuel-Cells with Multilayer Electrolytes", Solid State Ionics, 61, [4] 273-6 (1993).

[19] S. Y. Chun, N. Mizutani, "The Transport Mechanism of YSZ Thin Films Prepared by MOCVD", Applied Surface Science, 171, [1-2] 82-8 (2001).

[20] J. L. M. Rupp, L. J. Gauckler, "Microstructures and Electrical Conductivity of Nanocrystalline Ceria based Thin Films", Solid State Ionics, 177, [26-32] 2513- 8 (2006).

[21] T. Suzuki, I. Kosacki, H. U. Anderson, "Microstructure-Electrical Conductivity Relationships in Nanocrystalline Ceria Thin Films", Solid State Ionics, 151, [1- 4] 111-21 (2002).

[22] I. Kosacki, T. Suzuki, V. Petrovsky, H. U. Anderson, "Electrical Conductivity of Nanocrystalline Ceria and Zirconia Thin Films", Solid State Ionics, 136-137, 1225-33 (2000).

[23] G. Chiodelli, L. Malavasi, V. Massarotti, P. Mustarelli, E. Quartarone,

"Synthesis and Characterization of Ce0.8Gd0.2O2-y Polycrystalline and Thin Film Materials", Solid State Ionics, 176, [17-18] 1505-12 (2005).

[24] L. Chen, C. L. Chen, D. X. Huang, Y. Lin, X. Chen, A. J. Jacobson, "High

Temperature Electrical Conductivity of Epitaxial Gd-doped CeO2 Thin Films", Solid State Ionics, 175, [1-4] 103-6 (2004).

[25] C. S. Hsu, B. H. Hwang, "Microstructure and Properties of the

La0.6Sr0.4Co0.2Fe0.8O3 Cathodes Prepared by Electrostatic-Assisted Ultrasonic Spray Pyrolysis Method", Journal of the Electrochemical Society, 153, [8] A1478-A83 (2006). S T A B I L I T Y O F N I O M E M B R A N E S 187

[26] U. P. Muecke, S. Graf, U. Rhyner, L. J. Gauckler, "Microstructure and Electrical Conductivity of Nanocrystalline Ni-and NiO-CGO Thin Films", Submitted to Acta Materialia, (2007).

[27] D. Beckel, D. Briand, A. Bieberle-Hütter, J. Courbat, N. F. De Rooij, L. J. Gauckler, "Micro-Hotplates - a Platform for Micro-Solid Oxide Fuel Cells", Journal of Power Sources, 166, [1] 143-8 (2007).

[28] D. Beckel, U. P. Muecke, T. Gyger, G. Florey, A. Infortuna, L. J. Gauckler, "Electrochemical Performance of LSCF Based Thin Film Cathodes Prepared by Spray Pyrolysis", Solid State Ionics, 178, 407-15 (2007).

[29] J. Fleig, H. L. Tuller, J. Maier, "Electrodes and Electrolytes in Micro-SOFCs: a Discussion of Geometrical Constraints", Solid State Ionics, 174, [1-4] 261-70 (2004).

[30] H. Huang, M. Nakamura, P. Su, R. Fasching, Y. Saito, F. B. Prinz, "High- Performance Ultrathin Solid Oxide Fuel Cells for Low-Temperature Operation", Journal of the Electrochemical Society, 154, [1] B20-B4 (2007).

[31] X. Chen, N. J. Wu, L. Smith, A. Ignatiev, "Thin-Film Heterostructure Solid Oxide Fuel Cells", Applied Physics Letters, 84, [14] 2700-2 (2004).

[32] S. Kang, P. C. Su, Y. I. Park, Y. Salto, F. B. Prinz, "Thin-Film Solid Oxide Fuel Cells on Porous Nickel Substrates with Multistage Nanohole Array", Journal of the Electrochemical Society, 153, [3] A554-A9 (2006).

[33] Y. Tang, K. Stanley, Q. M. J. Wu, D. Ghosh, J. J. Zhang, "Design Consideration of the Micro Thin Film Solid-Oxide Fuel Cells", Journal of Micromechanics and Microengineering, 15, [9] S185-S92 (2005).

[34] V. T. Srikar, K. T. Turner, T. Y. Andrew Ie, S. M. Spearing, "Structural Design Considerations for Micromachined Solid-Oxide Fuel Cells", Journal of Power Sources, 125, [1] 62-9 (2004).

[35] J. P. Nair, E. Wachtel, I. Lubomirsky, J. Fleig, J. Maier, "Anomalous Expansion of CeO2 Nanocrystalline Membranes", Advanced Materials, 15, [24] 2077 - 81 (2003).

[36] C. D. Baertsch, K. F. Jensen, J. L. Hertz, H. L. Tuller, S. T. Vengallatore, S. M. Spearing, M. A. Schmidt, "Fabrication and Structural Characterization of Self- 188 C H A P T E R 9

Supporting Electrolyte Membranes for a Micro Solid-Oxide Fuel Cell", Journal of Materials Research, 19, [9] 2604-15 (2004).

[37] P. Bruschi, A. Diligenti, A. Nannini, M. Piotto, "Technology of Integrable Free- Standing Yttria-Stabilized Zirconia Membranes", Thin Solid Films, 346, [1-2] 251-4 (1999).

[38] K. Yunus, C. B. Marks, A. C. Fisher, D. W. E. Allsopp, T. J. Ryan, R. A. W. Dryfe, S. S. Hill, E. P. L. Roberts, C. M. Brennan, "Hydrodynamic Voltammetry in Microreactors: Multiphase Flow", Electrochemistry Communications, 4, [7] 579-83 (2002).

[39] M. Masuda, K. Sugioka, Y. Cheng, N. Aoki, M. Kawachi, K. Shihoyama, K. Toyoda, H. Helvajian, K. Midorikawa, "3-D Microstructuring Inside Photosensitive Glass by Femtosecond Laser Excitation", Applied Physics A, 76, [5] 857-60 (2003).

[40] Y. Cheng, K. Sugioka, K. Midorikawa, "Microfabrication of 3D Hollow Structures Embedded in Glass by Femtosecond Laser for Lab-on-a-Chip Applications", Applied Surface Science, 248, [1-4] 172-6 (2005).

[41] P. Sichler, S. Buttgenbach, L. Baars-Hibbe, C. Schrader, K. H. Gericke, "A Micro Plasma Reactor for Fluorinated Waste Gas Treatment", Chemical Engineering Journal, 101, [1-3] 465-8 (2004).

[42] T. Kim, S. Kwon, "Catalyst Preparation for Fabrication of a MEMS Fuel Reformer", Chemical Engineering Journal, 123, [3] 93-102 (2006).

[43] T. Kim, S. Kwon, "Design, Fabrication and Testing of a Catalytic Microreactor for Hydrogen Production", Journal of Micromechanics and Microengineering, 16, [9] 1760-8 (2006).

[44] Y. Cheng, H. L. Tsai, K. Sugioka, K. Midorikawa, "Fabrication of 3D Microoptical Lenses in Photosensitive Glass using Femtosecond Laser Micromachining", Applied Physics A, 85, [1] 11-4 (2006).

[45] F. E. Livingston, P. M. Adams, H. Helvajian, "Influence of Cerium on the Pulsed UV Nanosecond Laser Processing of Photostructurable Glass Ceramic Materials", Applied Surface Science, 247, [1-4] 526-36 (2005). S T A B I L I T Y O F N I O M E M B R A N E S 189

[46] S. Juodkazis, K. Yamasaki, V. Mizeikis, S. Matsuo, H. Misawa, "Formation of Embedded Patterns in Glasses using Femtosecond Irradiation", Applied Physics A, 79, [4-6] 1549-53 (2004).

[47] A. A. Bettiol, S. V. Rao, E. J. Teo, J. A. van Kan, F. Watt, "Fabrication of Buried Channel Waveguides in Photosensitive Glass using Proton Beam Writing", Applied Physics Letters, 88, [17] (2006).

[48] M. Kosters, H. T. Hsieh, D. Psaltis, K. Buse, "Holography in Commercially Available Photoetchable Glasses", Applied Optics, 44, [17] 3399-402 (2005).

[49] D. Briand, P. Weber, N. F. de Rooij, "Bonding Properties of Metals Anodically Bonded to Glass", Sensors and Actuators A, 114, [2-3] 543-9 (2004).

[50] T. R. Dietrich, W. Ehrfeld, M. Lacher, M. Kramer, B. Speit, "Fabrication Technologies for Microsystems Utilizing Photoetchable Glass", Microelectronic Engineering, 30, [1-4] 497-504 (1996).

[51] Mikroglas, Material Properties of Foturan, www.mikroglas.de, (2007).

[52] S. Mrotzek, A. Harnisch, G. Hungenbach, H. Strahl, D. Hulsenberg, "Processing Techniques for Photostructurable Glasses", Glass Science and Technology, 76, [1] 22-7 (2003).

[53] B. Schoeberle, D. Moeller, T. Helbling, M. Wendlandt, C. Hierold, "Determination of Mechanical Properties of SU-8 Thin Films with Bulge and Cantilever Test Method"; pp. 195-8 in Proceedings of MicroMechanics Europe Workshop (MME) Edited by (2005).

[54] P. Hing, P. W. McMillan, "The Strength and Fracture Properties of Glass- Ceramics", Journal of Materials Science, 8, 1041-8 (1973).

[55] A. A. Griffith, "The Phenomena of Rupture and Flow in Solids", Philosophical Transactions of the Royal Society of London Series A, 221, 163-98 (1921).

[56] M. Mori, T. Yamamoto, H. Itoh, H. Inaba, H. Tagawa, "Thermal Expansion of Nickel-Zirconia Anodes in Solid Oxide Fuel Cells during Fabrication and Operation", Journal of the Electrochemical Society, 145, [4] 1374-81 (1998).

[57] M. J. Madou, "Fundamentals of Microfabrication, The Science of Miniaturization" CRC Press, Boca Raton p. 205 (2002). 190 C H A P T E R 9

[58] C. S. Monstross, H. Yokokawa, M. Dokiya, "Thermal Stresses in Planar Solid Oxide Fuel Cells due to Thermal Expansion Difference", British Ceramic Transactions, 101, 85-93 (2002).

[59] A. G. Evans, D. Rajdev, D. L. Douglass, "The Mechanical Properties of Nickel Oxide and Their Relationship to the Morphology of Thick Oxide Scales Formed on Nickel", Oxidation of Metals, 4, [3] (1972).

[60] E. Lara-Curzio, M. Radovic, B. Armstrong, C. Walls, M. Lance, P. Tortorelli, S. Waters, L. Walker, A. Murphy, "Reliability and Durability of Materials and Components for Solid Oxide Fuel Cells", Oak Ridge National Laboratory, www.ornl.gov/sci/fossil/Publications/ANNUAL-2003/feaa066.pdf, (2003).

10 Conclusions and Outlook

Thin film cathodes based on La0.6Sr0.4Co0.2Fe0.8O3 (LSCF) and support structures for micro solid oxide fuel cells (µSOFCs) based on glass ceramic Foturan® and Si single crystals have been prepared and characterized. Finally a working µSOFC was achieved together with the work carried out in parallel theses on thin film anodes and electrolytes. The main conclusions and an outlook are given in this chapter.

10.1 Cathodes

Spray pyrolysis was investigated for fabrication of the LSCF based thin film cathodes. The most critical process parameter for successful cathode film fabrication was the ratio of deposition temperature to solvent boiling point, because it determines the drying and decomposition kinetics of the droplets and the growing film. For a spray solution consisting of a total metal salt concentration of 0.01 to 0.04 mol/l of La, Co, Fe nitrates and Sr chloride and a solvent composition of 1/3 vol. fraction ethanol and 2/3 vol. fraction diethylene glycol monobutyl ether and, a ratio of deposition temperature to solvent boiling point in the range of 1.15 to 1.25 (in K) proved to be useful. By keeping this ratio constant, the absolute deposition temperature could be varied by about 100°C, while still keeping a coherent crack-free film. Furthermore, the solvent, the salts, and some of the cations could also be exchanged without loss of film quality. Higher salt concentrations and the use of poor solvents during preparation of the spray solution lead to smoother films. Both parameters lead to stronger precipitation during evaporation of the solvent and the droplets reach the substrate in a wet state. The more precipitates are present in the spreading droplets, the lower is the mobility of the droplets on the substrate because the precipitates obstruct each other and get stuck. Thus the precipitates stay where they randomly hit the substrate and form smooth films. The solution flow rate plays a minor role concerning the film quality as long as it does not exceed a certain limit. The air pressure influences the drying kinetics. Higher pressure leads to faster drying and, 192 C H A P T E R 10 furthermore, to a more uniform droplet distribution and smoother films. In the post deposition annealing step in air, a crystalline film of the desired perovskite phase is already obtained at 650°C having nano-sized grains.

This annealing is also important for the microstructure of the LSCF films. They develop porosity during annealing due to solid-state dewetting. This dewetting was observed for films deposited by spray pyrolysis as well as for those prepared by pulsed laser deposition (PLD). The amount of porosity and pore size can be tailored by the annealing temperature and time. Thereby the defects in the amorphous films are the nuclei of the first pores. The concentration of these defects is determined by the thin film deposition process. Spray pyrolysis leads to more porosity than PLD for the same annealing conditions. The constraint of the thin film by the rigid substrate during annealing supports pore growth and hinders densification. Further tailoring of the porosity can be achieved by choosing precursors that either support densification such as chlorides or those which do not support densification e.g. nitrates. The films show semiconductivity in the temperature range from room temperature to 900°C with a transition to metallic behavior for higher temperatures. The in-plane conductivities are one order of magnitude lower than for micron grain sized samples due to the small grain size. Porosity further reduces the in-plane conductivity.

When determining the performance of these thin films as cathodes by means of the area specific resistance, it was found that the microstructure in combination with the choice of material is the most critical point for performance of the thin film cathode. Smaller grain sizes led to better performance and can be achieved by low annealing temperature. The best performance was achieved with the new material composition

Ba0.25La0.25Sr0.5Co0.8Fe0.2O3±δ which exhibits a very fine microstructure with an average grain size of 35 ± 7 nm. On the other hand, the introduction of a thin but dense cathode layer between the porous cathode and the electrolyte was also very effective. In summary, the microstructures achieved by spray pyrolysis proved to be better for the cathode performance than those achieved by PLD.

Spray pyrolysis can also be used for microstructuring of ceramic thin films on a flat substrate. By choosing the appropriate parameters during spray pyrolysis of metal salt solutions, thin films exhibiting random shaped ridges can be produced. When these solutions are sprayed onto substrates with patterned surfaces, the ridges follow the deliberately formed initial pattern. The lateral dimensions of the resulting ridges C O N C L U S I O N S A N D O U T L O O K 193 were in the range of 1 – 2 µm, which is at least one order of magnitude smaller than that of the initial pattern. The height of the original pattern is amplified by a factor of 30. High feature densities in the plane are achieved, because the number of the initial patterns is doubled. Using this combined approach of standard photolithography for patterning of the substrate with an inexpensive spray technique for the deposition of ceramic thin films on large substrates, microstructured ceramics with feature sizes in the range of the theoretical resolution limit of photolithography, exhibiting an aspect ratio of ~3 can be produced. Other well-established techniques can be used for the initial patterning of the substrate, which imparts great flexibility to this method. Due to the versatility of the spray technique, microstructured films consisting of a wide range of materials can be prepared using the approach outlined here.

In summary thin film cathodes for µSOFCs could successfully be prepared by spray pyrolysis and PLD. Their performance is sufficient to achieve reasonable power output with µSOFCs as temperatures as low as 550°C. Further improvements can be expected using new materials. Therefore new materials such as

Ba0.5Sr0.5Co0.8Fe0.2O3±δ (BSCF) deserve further studies, since basic material properties and behavior are still lacking full understanding. Routes to systematically identify potential interesting cathode materials, such as the structural field map [1, 2] would also be helpful since otherwise the huge amount of possible material combinations that can be achieved by perovskites leaves the search for new materials to a random choice. Another issue that is not fully resolved is the effect of impurities on cathode performance. While generally Pt is a poor cathode compared to mixed ionic electronic conducting perovskites, it proved that very pure, namely Si free, Pt cathodes showed impressive performance [3]. This poses the question if not only the better oxygen permeability is decisive for cathode performance but also a better tolerance to Si contamination. It would also be interesting if very pure, Si free mixed conducting perovskites also show a drastic improve in performance. As, shown above the microstructure, namely a small grain size was found to be very important for good cathode performance. For Pt cathodes this is even more important since a large triple phase area can only be achieved with a fine microstructure for Pt because the bulk remains inactive to the poor ionic conductivity.

194 C H A P T E R 10

10.2 Support Structures

For preparing a µSOFC the compatibility of the materials and the preparation processes are very important. Thin film deposition techniques such as spray pyrolysis and PLD are generally compatible with standard microfabrication techniques and especially the low processing temperatures of 600 to 650°C of the LSCF thin films are well adjusted to microfabrication. The compatibility of LSCF thin film cathodes with special SiN micro-hotplates as a platform for fabricating µSOFCs was also investigated: The SiN micro-hotplates were able to withstand temperature cycles from room temperature up to 800°C, which offers a secure margin to the target operating temperature of 600°C of a µSOFC. The functionality of the LSCF thin film under these conditions was proven by obtaining the same electrical conductivity as for reference samples deposited onto bulk substrates. Aging of cathodes on SiN micro-hotplates was somewhat accelerated and monitored by an increase in the resistance of cathodes compared to those deposited on a solid substrate. This accelerated aging was attributed to a larger mismatch in thermal expansion coefficient and bending of the micro-hotplate. Both might be diminished by using

Al2O3 instead of SiN as the membrane material, since Al2O3 offers superior stiffness and a thermal expansion coefficient closer to the values of LSCF and other fuel cell materials. The integrated heater and the low heat capacity of the micro-hotplate allow fast heating rates up to 10°C/s leading to short annealing times. After 5 min of annealing the film on the chip did not further change the microstructure even with longer annealing times. The conductivity data indicated that 30 min annealing on the micro-hotplate lead to the same result as almost 4 hrs annealing on a bulk substrate, since the flexible membrane allows faster crystallization of the thin film.

A potential substrate for fabrication of µSOFCs is Foturan®, a photostructurable glass-ceramic.

On this substrate it is advantageous to start with the anode as a first layer of the µSOFC, since the anode material can serve as an etching stop. The stability of NiO membranes on Foturan® showed that a membrane with 100 µm in diameter could withstand a gas pressure difference of ≥ 13700 Pa. This should be sufficient in case pressure variations occur during operation of a µSOFC. Generally thicker membranes proved to be more stable; however there is an optimum in membrane thickness, since the amount of residual stress increases with film thickness. C O N C L U S I O N S A N D O U T L O O K 195

The membranes processed on Foturan® showed higher thermal stability (≥ 620°C) than those fabricated on Si single crystal wafers. We attribute this to enhanced stress relaxation during annealing on the soft Foturan® substrate and to a better adjusted thermal expansion coefficient of the functional thin films on their substrate.

The field of µSOFC fabrication just recently opened, so there are many achievements to be expected in the future. After showing the proof of concept, the next steps are further integration of µSOFCs components to a functional system including current collectors, multiple layers stacked in order to increase voltage output and parallel connections of cells for increasing power density. This involves fabrication of many cells per chip, to have a reasonable yield for testing. After optimization of the process yield, the integration of several cells per chip will also be necessary to achieve enough active area. Since the research in this area just begun, it is too early to decide for a specific material or process for either the thin film components or the substrate.

10.3 References

[1] L. J. Gauckler, D. Beckel, B. Buergler, E. Jud, U. P. Muecke, M. Prestat, J. Rupp J. Richter, "Solid Oxide Fuel Cells: Systems and Materials", Chimia, 58, [12] 837-50 (2004).

[2] T. Nakamura, G. Petzow L. J. Gauckler, "Stability of the Perosvkite Phase

LaBO3 (B = V, Cr, Mn, Fe, Co, Ni) in Reducing Atmosphere. 1. Experimental Results", Materials Research Bulletin, 14, [5] 649 (1979).

[3] H. Huang, M. Nakamura, P. Su, R. Fasching, Y. Saito F. B. Prinz, "High- Performance Ultrathin Solid Oxide Fuel Cells for Low-Temperature Operation", Journal of the Electrochemical Society, 154, [1] B20-B4 (2007). 196 C H A P T E R 10

11 Appendix

This chapter gives an overview of the different processes for μSOFCs fabrication that have been tried so far. Also some general observations are given that have been made during processing of Foturan®.

11.1 Process for Fabrication of μSOFCs Based on Foturan®

The miniaturized SOFC was built up on one side of a photostructurable glass ceramic substrate by subsequent thin film deposition of anode contacts, anode, electrolyte and cathode. Double side polished 4” Foturan® wafers (Mikroglas, Mainz, Germany) with 300 μm thickness were used as substrates. A small portion of the substrate was then back-etched to release the fuel cell membrane. Pt contacts were attached to the cathode and the cell was tested. The detailed process steps are given in the following.

11.1.1 Preparation of the Substrate

11.1.1.1 Etching of Alignment Marks

Alignment marks were etched in the backside of the wafer to position all photolithography and shadow masks during cell fabrication. The 1 – 2 μm deep marks were isotropically masked using standard photolithography (1 min spinning of Shipley® Microposit® S1813 photoresist at 3000 rotations per minute (rpm), 1 min prebake at 115°C, 150 mJ/cm2 light exposure through a mask, developing for about

Chapter 11.1 is part of: U. P. Muecke, D. Beckel, A. Bieberle-Hütter, S. Graf, A. Infortuna, J. L. M. Rupp, J. Schneider, P. Mueller, A. Dommann, A. Bernard L. J. Gauckler, submitted to Advanced Functional Materials, (2007). Processing on the wafer level was done at NTB, processing on the chip level was done at ETH. 198 C H A P T E R 1 1

1 min with Microposit® MF 321 developer) and etched in buffered hydrofluoric acid (BHF, for 5 min), see Fig. 11.1 A). For clarity the figures are not to scale but rather show the essential features.

Fig. 11.1: Processing of the substrate on the wafer scale: A) isotropic etching of alignment marks; B) Structuring of aluminum mask, in the areas where membranes are created. C) Deposition of the Cr-Pt contacts; D) lift-off of the Cr-Pt contacts in the membrane area; E) UV exposure of the glass, through a shadow mask, the Cr-Pt contacts define the mask near the membranes. A P P E N D I X 199

11.1.1.2 Fabrication of Cr-Pt Contacts

An Al mask was sputtered (Balzers BAS 450) and patterned (same process as for the alignment marks) to leave circular spots with 100 and 200 μm diameter, see Fig. 11.1 B). A Cr adhesion layer of 30 nm thickness and platinum anode contacts with a thickness of 100 nm were RF-sputtered (custom made sputter machine, 100 W, 80 mm working distance, room temperature) through a 0.3 mm thick Mo mask on the front side of the wafer, as shown in Fig. 11.1 C). Circular holes with 100 and 200 μm diameter were created in the Cr-Pt contacts by lift-off of the Al mask, see Fig. 11.1 D). The Al mask was used instead of a photoresist mask to prevent ridges at the edge of the sputtered Pt after lift-off.

11.1.1.3 Illumination of the Substrate

After deposition of the anode contacts, the wafer was irradiated using UV light (Electronic Visions Group AL 6-2, 312 nm, 60 min, 500 W) through a shadow mask which covered the wafer outside the Cr-Pt anode contacts. The hole in the Cr-Pt contacts served as a precise mask for irradiating the cylindrical parts of the substrate (Fig. 11.1 E) that were later on back-etched to reveal the individual fuel cells. The wafer was then cut (Disco DAD 321) into 24 x 24 mm2 chips with three cells per chip.

11.1.2 Thin Film Deposition

The anode, electrolyte and cathode of the μ-SOFC were deposited by sputtering, pulsed laser deposition or air blast spray pyrolysis. Different electrode/electrolyte combinations as well as different deposition methods were combined to study the influence of single layers on the overall cell performance

11.1.2.1 Anode Deposition

A sputter coater (SCD 050, Balzers, FL) was used to deposit 50 and 200 nm thick platinum films at room temperature, see Fig. 11.2 A). The gas pressure was 0.05 mbar Ar, the sputter current 60 mA and the deposition time between 170 and 680 s. The film thickness was verified by SEM. The Pt films were used as anodes by 200 C H A P T E R 1 1 depositing circular spots of 3 mm diameter through a 0.1 mm Mo mask at the center of each individual cell on a chip. The 200 nm Pt films wrinkled during the crystallization of the Foturan substrate due to internal stresses and 50 nm films were used for most of the cells to reduce the waviness of the Pt surface.

11.1.2.2 Electrolyte Deposition

The electrolyte was deposited through a laser cut stainless steel mask of 200 nm thickness and openings of 6 mm for each cell.

11.1.2.2.1 Pulsed Laser Deposition of Electrolyte

YSZ electrolyte films were deposited by PLD (PLD workstation, Surface) from a 2 (Y2O3)0.08(ZrO2)0.92 target with a 248 nm excimer laser with 4 J/cm fluence, see Fig. 11.2 B. The target-to-substrate distance was 8.5 cm, the chamber pressure 2.66 Pa oxygen and the substrate temperature was 400°C. 54000 pulses at a rate of 10 Hz were used for ~550 nm thick films and 25000 pulses for ~250 nm thick films.

The YSZ target of 30 mm diameter and 5 mm thickness was prepared by uniaxial and isostatic pressing of YSZ powder (TZ-8Y, Tosoh, JP) and sintering at 1550°C for 4 hours with 3°C heating and 5°C cooling rate. The pellet was > 98 % dense after sintering and the average grain size was 1 μm. The target had a tetragonal fluorite phase with lattice constants of a-b = 3.62 Å and c = 5.13 Å [1]

11.1.2.2.2 Electrolyte Fabrication by Spray Pyrolysis

A detailed description of the spray pyrolysis setup used to deposit the

(Y2O3)0.08(ZrO2)0.92 spray pyrolysis electrolyte was given earlier [2]. The working distance was 39 cm, the air pressure 1.0 bar, the precursor flow rate 2.5 ml/h and the substrate surface temperature 410°C in this study. The majority of the sprayed precursor volume was transported by droplets with diameters between 10 and 50 μm. The deposition time was varied between 30 and 90 min. A P P E N D I X 201

Fig. 11.2: Processing on the chip scale, thin film deposition through shadow masks: A) deposition of the anode by sputtering; B) deposition of the electrolyte by pulsed laser deposition; C) deposition of the cathode by spray pyrolysis.

The precursor was prepared by dissolving zirconium-(IV)-acetylacetonate (96 % purity, Fluka, Buchs, CH) and yttrium-III-chloride hexahydrate (99.9 %, Alfa Aesar, Karlsruhe, DE) with the corresponding stoichiometry in a mixture of 10:10:90 vol % ethanol (99.5 %, Scharlau, Barcelona, ES) : polyethylene glycol 600 (PEG 600, purum, Fluka) : tetraethylene glycol (99 %, Aldrich, Steinheim, DE). The PEG 600 was added after complete dissolution of the salts in the ethanol:tetraethlylene glycol mixture at 50-80°C. The crystal water content of the salts was verified by thermogravimetry before weighing. The total salt concentration was 0.1 mol/l. 202 C H A P T E R 1 1

11.1.2.3 Cathode Deposition

The spray pyrolysis setup used to deposit the LSCF (Fig. 11.2 C) cathode is described in detail elsewhere [3]. The working distance was 20 cm, the air pressure 1.0 bar, the precursor flow rate 15 ml/h and the substrate surface temperature 280°C. The deposition time was 30 min, resulting in a film thickness of ~200 nm.

La(NO3)3·6H2O (99 %, ABCR, Karlsruhe, DE), SrCl2·6H2O (99 %, Fluka),

Co(NO3)2·6H2O (98 %, Alfa Aesar and Fluka) and Fe(NO3)3·9H2O (98 %, Fluka) were used to prepare the precursor. The salts were mixed in a ratio of 3.2:1:4 La:Sr:Co:Fe to obtain a material composition of La0.6Sr0.4Co0.2Fe0.8O3. A mixture of 33 vol % ethanol (99.5 %, Scharlau) and 66 vol % diethylene glycol monobutyl ether (99 %, Acros, Scheel, BE) was used as solvent. The total salt concentration was 0.04 mol/l. For some experiments 50 nm Pt was sputtered as cathode instead of LSCF.

11.1.3 Membrane Release

11.1.3.1 Annealing

After the thin film deposition, the chips were heated to 500 and 600°C for 1 hour each with 1°C/min heating and cooling rates in order to crystallize the UV irradiated areas of the substrate. The chips were pressed between two polished alumina plates with a weight of 2.2 g/cm2 during annealing to prevent warping of the substrate. After annealing, a clear contrast between the brownish, crystalline area where the holes will be created and the surrounding transparent glassy state was found.

11.1.3.2 Etching of the Substrate

The substrates were heated for 30 min at 120°C on a hotplate (Faktum) to remove any water. A protective coating (FSC-H, Rohm and Haas, Coventry, UK) was brushed on the thin film side of the substrate to protect the thin films from HF attack and prebaked for 45 min at 100°C, see Fig. 11.3 A). The crystalline areas were removed in a 10 % watery solution of HF (diluted from 40 % HF, suprapur, Merck, Darmstadt, DE, not buffered) under constant stirring. An average etching time of 22 min for 100 μm diameter membranes and 17 min for 200 μm membranes was A P P E N D I X 203 necessary. The photoresist was removed with acetone after etching. A schematic of the cell after processing is shown in Fig. 11.3 B).

A)

B)

C)

Fig. 11.3: Etching of the chips: A) protection of the thin films by a protective coating; B) chips after etching; C) contacted cells for testing.

11.1.4 Contacting

For testing, flat pressed 127 μm Pt wires were attached on both sides of the Cr/Pt anode contacts with Pt paste as current and voltage leads (C3605 P, Heraeus, Hanau, DE). The cathode contact wire was prepared by flat-pressing a 80 μm Pt wire 204 C H A P T E R 1 1 and bending the first ~0.5 mm of the wire by 90°. The tip of the wire was dipped in Pt paste and positioned on the surface of the electrolyte at the position of the fuel cell membrane for cells with a Pt paste cathodes. In case of sputtered Pt and the LSCF cathodes, the contact was placed adjacent to the membrane on top of the cathode material. The diameter of the resulting Pt paste dot was between 0.01 and 0.1 mm2. All wires were attached to the Foturan surface with ceramic glue (Feuerfestkitt, Firag AG, Ebmatingen, CH or Cerama-Bind 830, Aremco, Valley Cottage, NY, USA) to provide mechanical stability (Fig. 11.3 C)).

11.2 Polycrystalline Si Etch Stop

This process was developed as an alternative to the process described in chapter 11.1 with the idea to prevent direct contact between HF and the functional layers.

11.2.1 Process Details:

• Etching of alignment marks on the backside of the Foturan® wafer.

• Photolithography of a positive photoresist such that there are holes only in the areas where the membranes should be.

• Light exposure of the Foturan® through a photoresist mask.

• Stripping of the photoresist with acetone.

• Sputtering of polycrystalline Si (200 nm) on the front side of the Foturan®.

• Fabrication of the Pt electrodes on the front side of the Foturan®.

• Cutting of the wafer.

• Deposition of the active thin films.

• Protection of the thin films with FSC-H.

• Etching holes in the Foturan for the membranes until the polycrystalline Si.

• Etching in agitated 30 % KOH at 50°C for 2 min to remove the polycrystalline Si and open the membranes.

• Stripping of the protective coating. A P P E N D I X 205

11.2.2 Process Evaluation

This process can only be successful when the active layers are less attacked by KOH than by HF. In terms of material removal only the amorphous parts of the functional layers seem to be attacked by HF but HF is suspected to change the electrical properties of the functional layers. A change in electrical properties due to KOH is not known and not expected and KOH does not seem to attack the NiO. 2 min etching time with KOH was estimated as sufficient for removal of the polycrystalline Si layer in preliminary tests. However, during etching it is impossible to see if the polycrystalline Si is really removed since the thin layer is transparent in the light microscope and in contrast to Foturan its grains are too small to be visible in the light microscope. The FSC-H which is used as protective coating for the active layers against HF can also be used as protective coating against KOH at these conditions. It starts dissolving in KOH but due to the huge thickness achieved by brushing the FSC-H on the sample instead of spinning, it lasts for the 2 min. If longer etching times in KOH are required the easiest would be to use ProTEK B2 (from Brewerscience), a KOH resistive coating that can still be removed by acetone. The ProTEK can also be brushed on the sample. When using the same baking program as it was used for the FSC-H, a worse but still sufficient adhesion was achieved and the ProTEK was hard and not flexible. Better results might be achieved by shorter baking or maybe slightly thinner application. If necessary the adhesion might be improved by using the ProTEK primer. The FSC-H can be applied on top of the ProTEK using the same processing as for direct application on the sample. In this configuration the FSC-H first protects against HF, then the ProTEK protects against KOH and there is no need in applying a new coating on the delicate membranes.

This process is more difficult and time consuming than that described in chapter 11.1 but it is still feasible and can be used if needed. One disadvantage is that the Foturan has to be illuminated through the photoresist mask instead of using the Pt as mask. This provides the risk of overillumination which results in loss of the wafer. Overillumination cannot only occur due to human failure but due slight changes in the photoresist. For example, once a new lot of the same photoresist proved to let pass through more of the UV light. Another risk is that of misalignment between holes in the Foturan and holes in the Pt, which is completely excluded when using the Pt as mask. The Foturan cannot be illuminated through the polycrystalline Si, even 206 C H A P T E R 1 1 illuminating 200 times as long as for plain Foturan did not show any sign of illumination on the Foturan. The introduction of a new thin film (polycrystalline Si) is a potential source of new problems. Etching with two etchants lead to longer process time but includes only a small risk of loosing more samples.

11.3 Microfabrication using a Sacrificial Layer

The idea of using a sacrificial layer is very intriguing due to a huge degree of freedom that it would provide for the microstructuring. The idea is using any suitable substrate for fabrication of holes in this substrate by any process. Then the holes should be filled with a sacrificial layer that is able to withstand the thin film deposition conditions (400°C in air for spray pyrolysis, 400°C at 200 mtorr oxygen pressure in PLD). After thin film deposition the sacrificial layer would be removed in order to release the membrane.

11.3.1 Copper as Sacrificial layer

11.3.1.1 Process Details

• Illumination of the Foturan through a quartz plate and foil mask.

• Annealing of the Foturan.

• Etching of the Foturan without any protective coating using 10 % HF and ultrasound for fast and homogeneous etching.

• If needed the Foturan could even be grinded or polished.

• Sticking of the sample with scotch tape on a brass plate.

• Imersing the sample in the cooper plating bath (Cu, H2SO4, H2O, additives).

The help of Dr. Karl Vollmers, Institute of Robotics and Intelligent Systems, with the copper plating is gratefully acknowledged. A P P E N D I X 207

• Plating for 16 hrs at 0.5 A (= 1 A/dm2 of conductive surface) with 0.1 ms forward current pulse and 0.1 ms break. The Cu anode was constantly moved up and down and the bath constantly pumped.

• Plating for another hour with the same settings at 1 A.

• Cold embedding of the sample (Struers Citofix), overnight drying.

• Grinding with 500 until 4000 SiC paper.

• Spraying of NiO at 400°C.

• Annealing for 1 h at 600°C in N2.

• Next intended step would have been etching of the copper with HCl.

11.3.1.2 Process Evaluation

This process failed to achieve free-standing membranes and proved to have several difficulties. There must be a conductive surface at the end of the substrate’s holes, for the copper plating to work. If the area covered by holes is small with respect to the area covered by the substrate, the sample can be glued to a brass plate, providing the conductive surface and it can be removed afterwards without problems. However if a significant portion of the substrate has holes, there is no way of removing the sample afterwards from the plate, since it is strongly bonded by the copper growing through many holes. The plating process itself is somewhat delicate, especially when attempting plating such thick layers since the properties of the plating are adjusted rather qualitatively. Besides Cu, H2SO4 and H2O, there are some additives in the plating bath, such as “brightener” or “carrier”. The amount of additives to be added to the bath is determined in a “Hull cell” experiment. A defined volume of the plating bath is put in a cell with a copper anode and a brass plate to be coated. The resulting plated copper is inspected after some time of plating, depending on the result, i.e. mainly how the surface looks like, a certain amount (depending on the experimenter’s experience) of additives is added. A rough uneven surface of the plated sample caused by dendritic growth can usually be prevented by adding some brightener. Usually such plating baths have a rather big volume (here an 8 l bath was used) so they are not replaced after each experiment and consequently the exact 208 C H A P T E R 1 1 composition (specially regarding the additives) is not known. Therefore, these Hull cell experiments have to be done before each plating experiment. The settings of the plating process (current, pulse time etc.) follow some general trends: low current (and consequently slow deposition) is beneficial for thick layers since a rather even growth is achieved. The difficulty is that the growth rate depends on the local current density, which in turn depends on the hole size and geometry of the whole sample, so usually the current is adjusted to the available area and not all local effects can be taken into account.

Fig. 11.4: Example of fast growth during copper plating through structured Foturan leading to copper “mushrooms”.

In case the growth is too fast (4 A/dm2 of conductive surface), the Cu grows like mushrooms out of the hole as shown in Fig. 11.4. The problem is that these mushrooms are of low density, especially in the center, so upon grinding only a partial filling of the hole is obtained. Pulsing of the current is required to prevent local depletion of the additives in the solution that occurs in fraction of seconds. The cold embedding prior to grinding was done in order to prevent the (soft) copper to be sheared during grinding, which might cause an opening between the copper and the A P P E N D I X 209 substrate and thereby would not allow deposition of a crack-free membranes. The expected maximum hole size between copper and the substrate which can be tolerated during spray pyrolysis is about 2 μm. With the cold embedding it was possible to obtain holes completely filled with Cu leaving no opening as shown in Fig. 11.5.

Fig. 11.5: Holes in Foturan filled with Cu by plating after grinding.

The problems started during thin film deposition (NiO by spray pyrolysis). The copper oxidized, and expanded, so cracks were obtained in the thin film especially at the edges of the membranes. The situation got even worse during annealing, which was done in N2 in order to prevent further oxidation of the copper. A minimum annealing of 1 h at 600°C is required in order to get some crystallinity in the film, because an amorphous film would be dissolved quickly by the HCl which needs to be used to remove the copper. The metallic copper also expanded during annealing, resulting in a complete loss of the film in the areas where the membranes should be and some flakes of the film left over on the substrate as shown in Fig. 11.6. 210 C H A P T E R 1 1

Fig. 11.6: NiO film after annealing for 1 h at 600°C in N2.

Etching of the Cu without removing noticeably material from the NiO thin film should be possible. However, another problem could arise from the use of HCl. Halogenides are known to affect the sintering properties of ceramics, see chapter 5.3.5. The protective coating used for HF (FSC-H) also resists HCl, so protection of the thin films during etching should be feasible. Basically there are no other metals suitable as sacrificial layer: Ni cannot be used since Ni etchants are based on HNO3 which also etches NiO. Sn, Zn and Al have a too low melting point. Precious metals are not feasible; the etchants required to remove the precious metal attack the thin films as well as protective coatings. Besides that, for further development or even production, the use of precious metals as sacrificial layers would mean prohibitive costs.

A P P E N D I X 211

11.3.2 Polymer as Sacrificial Layer

Polymers as sacrificial layers have some advantages: they can be applied in a liquid form, which allows rapid filling of the holes and can be removed without harm to the functional layers.

11.3.2.1 Process Details

• Processing of the substrate as described in chapter 11.3.1.1.

• Placing the sample on a flat Teflon® surface to prevent sticking of the sample to the underground.

• Pouring the liquid polymer (Durimide 115 A, DuPont) over the sample and pressing it firmly in the holes, while keeping it on a hotplate 90 – 100°C.

• Prebake of the sample, ~20 min at 150°C to remove superfluous solvent.

• Curing (30 min at 400°C, 2.5 °C/min heating and1°C/min cooling) of the polymer in nitrogen atmosphere.

• Lapping the samples to obtain an even surface and remove superfluous polymer.

• Thin film deposition.

• Removal of the polymer using oxygen plasma.

11.3.2.2 Process Evaluation

Besides Durimide 115 A several polymer were tested (HD microsystems PI 2545, 2610, 2611 and 2731), with different thermal expansion coefficients (13·10-6/K, 3·10-6/K and 16·10-6/K) close to that of Foturan 8.6·10-6/K. Processing of the polymers was rather straight forward; however complete filling of the holes after polishing was not achieved as. Repeating the polymer deposition did not lead to

The help of Dr. Karim Alchalabi, formerly Laboratory of Solid State Physics now at Bosch, Germany, with polymer selection and processing is gratefully acknowledged. 212 C H A P T E R 1 1 significant improvement. Another try was vacuum exposure of the sample when the polymer was still in the liquid state to prevent shrinking of the polymer by solvent removal. Although more than enough polymer was provided to keep the holes filled, this method did not yield better results. Several other baking and curing procedures were also tried, without substantial improvement. DTA/TG experiments confirmed that most of the tested polymers were able to withstand up to 600°C, so this was not the problem but the insufficient filling of the holes that might be caused by shrinking of the polymer during curing.

11.3.3 Conclusion Sacrificial Layer

The task a material has to fulfill as sacrificial layer is very diverse, thus this approach was not successful here. The sacrificial layer should be easy to apply in small holes (100 to 200 μm diameter), it should fill theses holes with gaps of less than 2 μm, it should be able to withstand the thin film deposition conditions (400°C in air), or even better the annealing conditions (1 h at 600°C in air) for the ceramic thin films without shrinkage, other expansion than the substrate material or oxidation. Furthermore, it should be possible to remove the layer easily without harming the delicate thin film membranes.

11.4 Effect of Film Thickness on Crack Formation and Wrinkling

The likeliest failures of membranes are cracks. Cracks originate from stress in the thin films which is either intrinsic stress from the thin film deposition, or stress caused by a mismatch in thermal expansion coefficient between the thin film materials and the substrate, or both. In thin films the issue of thermal mismatch might be less critical than for bulk materials. Thus, intrinsic stress is probably more severe. It proved that films deposited by spray pyrolysis crack as soon as they reach a certain thickness, which is around 1 μm. With the PLD, also somewhat thicker films could be produced (~1.5 μm), however there was a trade off between density and crack formation: the denser the film, the thinner the maximum thickness without crack formation. From the point of crack formation the films should be as thin as possible; however, thicker films are more robust as membranes. A P P E N D I X 213

During annealing the Foturan® softens (Tg = 465°C), which may lead to stress relaxation in the thin films. That means that the thin films are to some extend flexible on the Foturan® substrate. Especially the Pt films used as contact pads were able wrinkle on the Foturan® after annealing. The thicker the Pt film, the more force it could apply to the Foturan®, the more wrinkling was found. Pt films of 100 nm thickness proved to reduce wrinkling while still having a stable microstructure for the conditions used for cell testing. This significantly improved the cell yield, since the wrinkles are very likely to provoke cracks.

11.5 Unexpected Crystallization of Foturan®

A very interesting observation was made during processing of Foturan®: if the Foturan® was exposed to aluminum during processing, e.g. when aluminum was used as mask for the lift-off process, a crystalline layer of ~2 μm thickness was found after annealing on top of the glassy Foturan®. However, during further annealing crystallization did not proceed further through the glass. The same effect was visible when YSZ, Cr or Ti were in direct contact with Foturan®, but it did not occur for CGO or Pt on the Foturan® surface. Glassy Foturan® that was kept for long time at elevated temperatures (several hours at 500 to 600°C) in a fuel cell environment (H2,

N2 and air atmosphere) showed slow crystallization through the entire wafer. However, when increasing the annealing time in air during selective crystallization to 10 hrs at 600°C, only the radiated areas crystallized and not the entire wafer. Although radiation of the Foturan is done with UV light, the typical deposition conditions for the electrolyte (54000 pulses at 2 J/cm2 target fluence) using a 248 nm UV laser did not bring sufficient energy in the Foturan® to induce crystallization.

11.6 References

[1] A. Infortuna, A. Harvey, L. J. Gauckler, "Microstructures of CGO and YSZ Thin Films by Pulsed Laser Deposition", submitted to Advanced Functional Materials, (2007). 214 C H A P T E R 1 1

[2] U. P. Muecke, N. Luechinger, L. J. Gauckler, "Initial Status of Deposition and Film Formation During Spray Pyrolysis of Nickel Oxide, Cerium Gadolinium Oxide and NiO-CGO Thin Films", Submitted to Thin Solid Films, (2006).

[3] D. Beckel, A. Dubach, A. R. Studart, L. J. Gauckler, "Spray Pyrolysis of

La0.6Sr0.4Co0.2Fe0.8O3-d Thin Film Cathodes", Journal of Electroceramics, 16, [3] 221-8 (2006).

Publications

Peer Reviewed Publications (submitted or already published)

• D. Beckel, U.P. Muecke, B. Schoeberle, P. Müller, L.J. Gauckler “Stability of NiO Membranes on Photostructurable Glass Substrates” submitted to Thin Solid Films (2007).

• A. Bieberle-Hütter, D. Beckel, A. Infortuna, U.P. Muecke, J.L.M. Rupp, L.J. Gauckler, S. Rey-Mermet, P. Muralt, N. Hotz, M.J. Stutz, N.R. Bieri, D. Poulikakos, P. Müller, A. Bernard, R. Gmür, T. Hocker “A Micro-Solid Oxide Fuel Cell System for Battery Replacement”, submitted to Journal of Power Sources (2007).

• U.P. Muecke, D. Beckel, A. Bieberle-Hütter, S. Graf, A. Infortuna, J.L.M. Rupp, J. Schneider, P. Mueller, A. Dommann, A. Bernard, L.J. Gauckler „Micro Solid Oxide Fuel Cells on Glass Ceramic Substrates“, submitted to Advanced Functional Materials (2007).

• D. Beckel, A. Dubach, A.N. Grundy, A. Infortuna, L.J. Gauckler“ Solid State

Dewetting of La0.6Sr0.4Co0.2Fe0.8O3±δ Thin Films during Annealing” accepted for publication in Journal of the European Ceramic Society (2007).

• D. Beckel, A. Bieberle-Hütter, A. Harvey, A. Infortuna, U.P. Muecke, M. Prestat, J.L.M. Rupp, L.J. Gauckler “Thin Films for Micro Solid Oxide Fuel Cells”, Journal of Power Sources doi:10.1016/j.jpowsour.2007.04.070 (2007).

• D. Beckel, U.P. Muecke, T. Gyger, G. Florey, A. Infortuna, L.J. Gauckler “Electrochemical Performance of LSCF Based Thin Film Cathodes Prepared by Spray Pyrolysis”, Solid State Ionics 178 (2007) 407-415.

• D. Beckel, D. Briand, A. Bieberle-Hütter, J. Courbat, N.F. de Rooij, L.J. Gauckler “Micro-Hotplates - a Platform for Micro Solid Oxide Fuel Cells”, Journal of Power Sources, 166 (2007) 143-148 .

• D. Beckel, D. Briand, A.R. Studart, N.F. de Rooij, L.J. Gauckler “Topography Mediated Patterning of Inorganic Materials by Spray Pyrolysis”, Advanced Materials, 18 (2006) 3015-3018. 216 P U B L I C A T I O N S

• D. Beckel, A. Dubach, A.R. Studart, L.J. Gauckler “Spray Pyrolysis of

La0.6Sr0.4Co0.2Fe0.8O3-δ Thin Film Cathodes”, Journal of Electroceramics, 16 (2006) 221-228.

• G. Upper, W. Zhang, D. Beckel, S. Sohn, K. Liu, E. Kiran “Phase Boundaries and Crystallization of Polyethylene in n-Pentane and n-Pentane + Carbon Dioxide Fluid Mixtures” Industrial & Engineering Chemistry Research 45 (2006) 1478-1492.

• L.J. Gauckler, D. Beckel, B.E. Buergler, E. Jud, U.P. Muecke, M. Prestat, J.L.M. Rupp, J. Richter “Solid Oxide Fuel Cells: Systems and Materials”, Chimia, 58 (2004) 837-850.

Patent Applications

• D. Beckel, L.J. Gauckler; „Poröser keramischer Dünnfilm“; WO 2007/056876. Publication date: 24.05.2007 (priority 21.11.2005).

• L.J. Gauckler, D. Beckel, U.P. Muecke, P. Müller, J.L.M. Rupp; WO 2007/045113 „Verbund eines Dünnfilms und eines Glaskeramischen Substrats als Miniaturisiertes Elektrochemisches Gerät“. Publication date: 26.04.2007 (priority 19.10.2005).

• L.J. Gauckler, D. Beckel, U.P. Muecke, P. Müller, J.L.M. Rupp; WO 2007/045111 „Dünnfilm und damit hergestelltes Verbundelement“. Publication date: 26.04.2007 (priority 19.10.2005).

Selected other Publications and Presentations

• D. Beckel, U.P. Muecke, J.L.M. Rupp, A. Infortuna, A. Bieberle-Hütter, P. Mueller, A. Bernard, L.J. Gauckler “Thin Film Cathodes for Micro Solid Oxide Fuel Cells”, Euromat07, Nürnberg, Germany, September 2007 (Talk).

• A. Bieberle-Hütter, D. Beckel, A. Infortuna, U.P. Muecke, J.L.M. Rupp, L.J. Gauckler, S. Rey-Mermet, P. Muralt, N. Hotz, M.J. Stutz, N.R. Bieri, D. Poulikakos, P. Müller, P. Heeb, A. Bernard, R. Gmür, T. Hocker “A Micro- P U B L I C A T I O N S 217

Solid Oxide Fuel Cell System for Battery Replacement”, Euromat07, Nürnberg, Germany, September 2007 (Talk).

• U.P. Muecke, D. Beckel, S. Graf, A. Bieberle-Hütter, A. Infortuna, L. Schlagenhauf, J. Schneider, J.L.M. Rupp, P. Mueller, A. Domann, A. Bernard, L.J. Gauckler “Electrical Characterization of Micro-Solid Oxide Fuel Cells”, Solid State Ionics 16 (SSI-16), Shanghai, China, July 2007 (Talk).

• A. Bieberle-Hütter, D. Beckel, A. Infortuna, U.P. Muecke, J.L.M. Rupp, L.J. Gauckler, S. Rey-Mermet, P. Muralt, N. Hotz, M. J. Stutz, N. R. Bieri, D. Poulikakos, P. Müller, P. Heeb, A. Bernard, R. Gmür, T. Hocker “A Micro- Solid Oxide Fuel Cell System for Battery Replacement”, Solid State Ionics 16 (SSI-16), Shanghai, China, July 2007 (Talk).

• D. Beckel, D. Briand, J. Courbat, A. Bieberle-Hütter, N.F. de Rooij, L.J. Gauckler “Micro-Hotplate Devices for Micro-SOFC”, in: Proceedings of 10th International Symposium on Solid Oxide Fuel Cells (SOFC-X), Nara, Japan, 2007 ECS Transaction 7 [1] pp.412-417, (Talk and proceedings).

• D. Beckel, U.P. Muecke, G. Florey, T. Gyger, A. Dubach, A. Infortuna, L.J. Gauckler “LSCF Thin Film Cathodes Deposited by Spray Pyrolysis for Micro-SOFC”, in: Proceedings of 10th International Symposium on Solid Oxide Fuel Cells (SOFC-X), Nara, Japan, 2007, ECS Transaction 7 [1] pp.1139-1145, (Poster and proceedings).

• J.L.M. Rupp, U.P. Muecke, D. Beckel, A. Bieberle-Hütter, A. Infortuna, S. Rey- Mermet, P. Muralt, L.J. Gauckler “ONEBAT: Micro-Solid Oxide Fuel Cells for Battery Replacement in Portables” in: Proceedings of the 10th International Symposium on Solid Oxide Fuel Cells (SOFC-X), Nara, Japan, 2007, ECS Transaction 7 [1] pp. 887-890, (Talk and proceedings).

• J.L.M. Rupp, D. Beckel, A. Bieberle-Hütter, A. Harvey, A. Infortuna, U.P. Muecke, T. Ryll, B. Scherrer, R. Tölke, L.J. Gauckler “Micro-Solid Oxide Fuel Cells for Battery Replacement”, 8th International Symposium on Systems with Fast Ionic Transport (ISSFIT), Vilnius, Lithuania, May 2007 (Talk).

• D. Beckel, D. Briand, A. Bieberle-Hütter, J. Courbat, N.F. de Rooij, L.J. Gauckler “Electrical Conductivity of LSCF Thin Films for SOFC on Micro 218 P U B L I C A T I O N S

Hotplates”, in: Proceedings of 7th European Solid Oxide Fuel Cell Forum, Lucerne, Switzerland, July 2006, P0603. (Poster and Proceedings).

• D. Beckel, U.P. Muecke, G. Florey, A. Dubach, A. Infortuna and L.J. Gauckler “Thin Film Cathodes for Use in Micro Solid Oxide fuel Cells“, 3rd SOFC Meeting at 30th International Conference on Advanced Ceramics and Composites, Cocoa Beach, Florida, USA, January 2006 (Talk).

• A. Infortuna, D. Beckel, A. Bieberle-Hütter, U.P. Muecke, J.L.M. Rupp, L.J. Gauckler “Microstructural Study of Metal-Oxide Thin Films for SOFC Applications, Grown by Pulsed Laser Deposition and Spray Pyrolysis“, 9th International Ceramic Processing Science Symposium, Coral Springs, Florida, USA, January, 2006 (Talk).

• A. Bieberle-Hütter, D. Beckel, U.P. Muecke, J.L.M. Rupp, A. Infortuna, L.J. Gauckler “Micro-Solid Oxide Fuel Cells as Battery Replacements”, Mstnews, 04/05 (2005) 12-15.

• D. Beckel, A. Mariaux, A. Dubach, L.J. Gauckler “La1-xSrxCoyFe1-yO3-δ Thin Films for Micro Solid Oxide Fuel Cell Cathodes“, in: Proceedings of 206th Meeting of The Electrochemical Society 2004 - Ionic and Mixed Conducting Ceramics V, Honolulu, HI, USA, 2004, (Talk and proceedings, accepted for publication).

• U.P. Muecke, D. Beckel, A.O.J. Ganz, P. Mueller, A. Dommann, L.J. Gauckler “Nickel - Gadolinia Doped Ceria Cermet Thin Films for Micro Solid Oxide Fuel Cells“, in: Proceedings of 206th Meeting of The Electrochemical Society 2004 - Ionic and Mixed Conducting Ceramics V, Honolulu, HI, USA, 2004, (Talk and proceedings, accepted for publication).

• D. Beckel, A. Dubach, A. Mariaux, L.J. Gauckler ”LSCF Cathodes for Micro- Solid Oxide Fuel Cells” 55th Annual Meeting of the International Society of Electrochemistry, Thessaloniki, Griechenland,. September, 2004 (Poster).

• G. Upper, D. Beckel, W. Zhang, E. Kiran “High Pressure Crystallization in Supercritical or Dense Fluids” in: Proceedings of the 6th International Symposium on Supercritical Fluids, Versailles, France, 2003, Vol. 3 Materials Processing, pp. 1509-1514. (Talk and proceedings). P U B L I C A T I O N S 219

• T. Rager, D. Beckel, C. Noirtin, P. Muff, S. Schweri, O. Haas, J. Huslage, G.G. Scherer “Radiation-Grafted Polymer Films with Improved Mechanical Properties“ Paul Scherrer Institute Scientific Report Vol.5 General Energy 2001, 5, 97-98. 220 P U B L I C A T I O N S

Curriculum Vitae

Daniel Beckel, born 05.05.1977, citizen of Germany

PhD thesis

Jan. 2003 – present „Thin Film Cathodes for Micro Solid Oxide Fuel Cells”, ETH Zurich, Supervison: Prof. Dr. L.J. Gauckler.

Studies

Oct. 1997 – Sept. 2002 Process Engineering, University of Karlsruhe (TH), Germany.

Diploma thesis

Mar. 2002 – Sept. 2002 “Experimental Investigations of the Polymer Phase Behavior and Crzystallization of Polyethylene in Dense Fluids” Virginia Polytechnic Institute & State University, Blacksburg, USA, Prof. Dr. E. Kiran.

Internships

Oct. 2001 Workshop training, Elastomer Technik Gedern, Gedern, Germany.

Aug. 2001 – Sept. 2001 Radiation Grafting of Polymer Membranes for Fuel Cells, Paul Scherrer Institute, Villigen, Switzerland, Dr. T. Rager.

Mar. 2001 – Apr. 2001 Workshop training, Fraunhofer Institute for Solar Energy Systems (ISE), Freiburg, Germany.

Project work

Oct. 2000 – Mar. 2001 “Auslegung eines Prototyps für ein Brennstoffzellen Block Heizkraftwerk“ Fraunhofer ISE, Freiburg, Germany, Dr. P. Hübner.

School

Sept. 1984 – Jun. 1997 Freie Waldorfsschule Würzburg, Germany.