Solid Oxide Fuel Cell Lifetime and Reliability. Critical Challenges in Fuel Cells [1st Edition] 9780128097243, 9780081011027

Solid Oxide Fuel Cell Lifetime and Reliability: Critical Challenges in Fuel Cells presents in one volume the most recent

853 130 20MB

English Pages 234 [222] Year 2017

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Solid Oxide Fuel Cell Lifetime and Reliability. Critical Challenges in Fuel Cells [1st Edition]
 9780128097243, 9780081011027

Table of contents :
Content:
Front-matter,Copyright,List of ContributorsEntitled to full textChapter 1 - An Introduction to Solid Oxide Fuel Cell Materials, Technology and Applications, Pages 1-18, Samuel J. Cooper, Nigel P. Brandon
Chapter 2 - Solid Oxide Fuel Cell Electrolytes—Factors Influencing Lifetime, Pages 19-35, Alan Atkinson
Chapter 3 - The Impact of Fuels on Solid Oxide Fuel Cell Anode Lifetime: The Relationship Between Fuel Composition, Fuel Impurities, and Anode Lifetime and Reliability, Pages 37-50, Kazunari Sasaki
Chapter 4 - The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime, Pages 51-77, Tony Wood, Douglas G. Ivey
Chapter 5 - Microstructural Degradation: Mechanisms, Quantification, Modeling and Design Strategies to Enhance the Durability of Solid Oxide Fuel Cell Electrodes, Pages 79-99, Farid Tariq, Enrique Ruiz-Trejo, Antonio Bertei, Paul Boldrin, Nigel P. Brandon
Chapter 6 - Cathode Degradation From Airborne Contaminants in Solid Oxide Fuel Cells: A Review, Pages 101-119, Ashish Aphale, Chiying Liang, Boxun Hu, Prabhakar Singh
Chapter 7 - Lifetime Issues for Solid Oxide Fuel Cell Interconnects, Pages 121-144, Manuel Bianco, Markus Linder, Yngve Larring, Fabio Greco, Jan Van herle
Chapter 8 - Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells, Pages 145-171, Joongmyeon Bae
Chapter 9 - Life and Reliability of Solid Oxide Fuel Cell-Based Products: A Review, Pages 173-191, Subhasish Mukerjee, Rob Leah, Mark Selby, Graham Stevenson, Nigel P. Brandon
Chapter 10 - New Materials for Improved Durability and Robustness in Solid Oxide Fuel Cell, Pages 193-216, Mark Cassidy, Dragos Neagu, Cristian Savaniu, Paul Boldrin
Index, Pages 217-223

Citation preview

Solid Oxide Fuel Cell Lifetime and Reliability

Solid Oxide Fuel Cell Lifetime and Reliability Critical Challenges in Fuel Cells

Edited by

Nigel P. Brandon, Enrique Ruiz-Trejo and Paul Boldrin Imperial College London, London, United Kingdom

Academic Press is an imprint of Elsevier 125 London Wall, London EC2Y 5AS, United Kingdom 525 B Street, Suite 1800, San Diego, CA 92101-4495, United States 50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom Copyright r 2017 Elsevier Ltd. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions. This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein). Notices Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary. Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility. To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress ISBN: 978-0-08-101102-7 For Information on all Academic Press publications visit our website at https://www.elsevier.com/books-and-journals

Publisher: Joe Hayton Acquisition Editor: Raquel Zanol Editorial Project Manager: Mariana Kuhl Production Project Manager: Kiruthika Govindaraju Cover Designer: Greg Harris Typeset by MPS Limited, Chennai, India

List of Contributors Ashish Aphale University of Connecticut, Storrs, CT, United States Alan Atkinson Imperial College London, London, United Kingdom Joongmyeon Bae Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea Antonio Bertei Imperial College London, London, United Kingdom Manuel Bianco Swiss Federal Institute of Technology in Lausanne Valais, Sion, Switzerland Paul Boldrin Imperial College London, London, United Kingdom Nigel P. Brandon Imperial College London, London, United Kingdom Mark Cassidy University of St Andrews, St Andrews, United Kingdom Samuel J. Cooper Imperial College London, London, United Kingdom Fabio Greco Swiss Federal Institute of Technology in Lausanne Valais, Sion, Switzerland Boxun Hu University of Connecticut, Storrs, CT, United States Douglas G. Ivey University of Alberta, Edmonton, AB, Canada Yngve Larring SINTEF Materials and Chemistry, Oslo, Norway Rob Leah Ceres Power Ltd, Horsham, United Kingdom Chiying Liang University of Connecticut, Storrs, CT, United States Markus Linder Zurich University of Applied Sciences, Winterthur, Switzerland Subhasish Mukerjee Ceres Power Ltd, Horsham, United Kingdom Dragos Neagu University of St Andrews, St Andrews, United Kingdom Enrique Ruiz-Trejo Imperial College London, London, United Kingdom Kazunari Sasaki Kyushu University, Fukuoka, Japan Cristian Savaniu University of St Andrews, St Andrews, United Kingdom Mark Selby Ceres Power Ltd, Horsham, United Kingdom Prabhakar Singh University of Connecticut, Storrs, CT, United States Graham Stevenson Imperial College London, London, United Kingdom

ix

x

List of Contributors

Farid Tariq Imperial College London, London, United Kingdom Jan Van herle Swiss Federal Institute of Technology in Lausanne Valais, Sion, Switzerland Tony Wood Fuel Cell Energy, Calgary, AB, Canada

Chapter 1

An Introduction to Solid Oxide Fuel Cell Materials, Technology and Applications Samuel J. Cooper and Nigel P. Brandon Imperial College London, London, United Kingdom

Chapter Outline A Brief History of Solid Oxide Fuel Cells Solid Oxide Fuel Cell Fundamentals Activation Losses Ohmic Losses Concentration Losses Crossover Losses Solid Oxide Fuel Cell Design Solid Oxide Fuel Cell Operating Temperature and Materials

1 3 4 4 4 5 5

Materials Selection Microstructural Design Commercially Available Solid Oxide Fuel Cells Current Technology Status Introduction to Degradation Physical Chemical Conclusions References

8 9 11 12 13 14 15 15

7

This chapter aims to give the reader an overview of solid oxide fuel cell (SOFC) technology in terms of both the fundamental theory and real world applications. It concludes with an introduction to the various degradation mechanisms common to many fuel cell systems today, which are discussed in detail in the following chapters of this book.

A BRIEF HISTORY OF SOLID OXIDE FUEL CELLS Fuel cells are a family of electrochemical devices, which generate electricity by promoting a redox reaction across an ionically conductive membrane. Although fuel cells were first reported in 1839 by Sir William Grove, it was not until 1961, when NASA began Project Gemini, that they found their first practical application [1]. Fuel cells are typically named in terms of two key Solid Oxide Fuel Cell Lifetime and Reliability. DOI: http://dx.doi.org/10.1016/B978-0-08-101102-7.00001-5 Copyright © 2017 Elsevier Ltd. All rights reserved.

1

2

Solid Oxide Fuel Cell Lifetime and Reliability

characteristics: the mobile ion and the electrolyte material, with the operating temperature also being used to subclassify in some cases. SOFCs are named after their ion conducting, ceramic oxide electrolyte and their history is tied to some of the great names in science and engineering. Faraday’s early investigations of conduction in ceramics in the 1830s [2], led him to classify conductors into two categories, although the exact mechanism for these two modes of conduction was unknown. It was not until much later, in the 1890s, when Walther Nernst observed the significantly increased conductivity of mixed oxides over their pure constituents that the first technological implication of ion conduction in solids was conceived. Although ultimately not a commercial success, due in part to its high cost, the “Nernst Glower” was nearly twice as efficient as the carbon filament lamps of the day [3]. The device consisted of a ceramic oxide rod made of yttria-doped zirconia (often referred to as the “Nernst Mass”) which, after preheating to around 1000 C, would begin to conduct under load; this in turn led to the temperature increasing further, causing the rod to glow. The 1930s saw the conceptual development of ion conduction through lattice defects by Schottky [4] and Frenkel [5], which led to the submission of the first SOFC patent through Siemens and Halske [6]. The first cell beginning to resemble a modern configuration was proposed by Baur and Preis [7], who used the “Nernst Mass” for the electrolyte in combination with metal oxide electrodes. Although the system was a failure due to high Ohmic losses, it spurred a new wave of investigation into conducting mixed oxides. Over the following 30 years, Kiukkola and Wagner [8] and many others [9,10] undertook a systematic investigation into ionconducting electrode materials in order to find structures that had both the mechanical and electrochemical properties required for a durable fuel cell. By 1970 the adoption of electroceramics for a broad range of other industrially relevant applications, such as sensors (e.g., lambda sensors that are widely used today to measure the air/fuel ratio in engine exhaust gases) and oxygen separation membranes, led to key advances in materials processing and the materials supply chain. Other related advances, for example in the semiconductor industry, resulted in processes emerging such as electrochemical vapor deposition [11]. This allowed for much thinner layers of highpurity material to be deposited, which not only had the potential to reduce Ohmic losses, but also opened the possibility of using materials previously deemed too costly. Following the first and second oil crises of the 1970s, which cumulatively led to a 10-fold increase in the price of oil [12], governments from fuel importing nations began to invest more heavily in the research and development of alternative energy technologies [13]. Since the early 1990s, a sequence of SOFC companies predominantly from the United States, Western Europe, and Japan have emerged aiming at bringing a range of SOFC configurations to market.

An Introduction to Solid Oxide Fuel Cell Materials Chapter | 1

3

These companies are developing technologies largely focussed on the distributed generation market. G

G

G

Residential combined heat and power (c. 1 kWe) - e.g., Solid Power, Ceres Power Commercial grid-independent generators (c. 100 kWe) - e.g., Bloom Energy Industrial SOFC gas turbine hybrids (c. 1 MWe) - e.g., LG Fuel Cell Systems

Common to all of these applications is the necessity for the devices to operate for extended periods (510 years) without requiring significant maintenance or replacement. It is also critical for the cells, stacks, and systems to be able to withstand the inevitable shut down events, which poses a particular problem for SOFCs due to their high operating temperature and brittle ceramic components. State of the art SOFC devices can already achieve electrical efficiencies of above 50% and combined heat and power systems exist with total efficiencies in excess of 90%. These two metrics are very impressive on their own, but in combination with the lack of NOx/SOx or particulates in the exhaust stream and the low noise/vibration of these systems, the appeal of SOFC devices is clear. However, SOFCs will not be able to fully deliver on their potential until the degradation issues key to lifetime are resolved, which is the subject of this book.

SOLID OXIDE FUEL CELL FUNDAMENTALS The Nernst potential, ENernst, of an SOFC is a function only of the physical properties and chemical composition of its two incoming gas streams (fuel and oxidant). It can be determined using the Nernst equation, which is the sum of the standard cell potential E0 and a term that describes the activity at the specific conditions in question, ! 1=2 P P RT H 2 O 2 ln ENernst 5 E0 1 ð1:1Þ 2F PH 2 O where R is the universal gas constant, T is the temperature, F is the Faraday constant, and Px is the normalized partial pressure of species x. The standard cell potential term, E0, is calculated as the difference between the equilibrium potentials of the two reduction/oxidation (redox) reactions under standard conditions: 2H1 1 2e2 "H2

ðE0 5 0 VÞ

1 O2 1 2H1 1 2e2 "H2 O 2

ðE0 5 1:23 VÞ

ð1:2Þ ð1:3Þ

4

Solid Oxide Fuel Cell Lifetime and Reliability

For the hydrogenoxygen redox couple under standard conditions, the cell potential is 1.23 V. As a current is drawn, the system moves away from equilibrium and the potential between the two electrodes decreases. The Nernst potential describes an idealized reaction, which is a useful reference when quantifying the four main categories of losses (overpotentials) in SOFCs: activation losses, Ohmic losses, concentration losses, and crossover losses.

Activation Losses Activation losses can be considered as the potential required to drive the reaction forward at the required rate, noting that the high operating temperature of SOFCs significantly improves the reaction kinetics. The ButlerVolmer equation quantifies the effect of the charge transfer processes at each electrode on the total current density, j,      αa nFη αc nFη j 5 j0 exp 2 exp ð1:4Þ RT RT where n is the number of electrons involved in each electrode reaction, η is the activation overpotential, and αa and αc are the anodic and cathodic charge transfer coefficients, respectively. The activation overpotential is described by the relation, η 5 Eelectrode 2 ENernst

ð1:5Þ

which is the difference between real and equilibrium potentials, specified at each electrode. The magnitude of this overpotential increases at each electrode with the current, thus reducing the overall potential of the cell.

Ohmic Losses These are caused by the resistance to flow of electrical current through the cell. Typically the ionic transport, as opposed to electronic, is the most significant contribution to this overpotential. The intrinsic conductivities of the various materials, the cell and stack geometry, and the convolution of the conduction paths in the porous electrodes, all need to be considered.

Concentration Losses The electrochemical reactions typically only occur in a region close to the electrodeelectrolyte interface, which means the gases must first travel through much of the porous electrodes. At high current densities, this can become the rate limiting step for the system.

An Introduction to Solid Oxide Fuel Cell Materials Chapter | 1

EOCV

Cell voltage, E / V

Ohmic

5

Conce ntratio losses n loss

es

Activation losses

Current density, j / A cm–2 FIGURE 1.1 Plot of current density against cell voltage, illustrating the breakdown of cell performance by loss type.

Crossover Losses This category covers two fairly distinct sources of loss. First, electrolytes, either through porosity or cracking, may be gas permeable, which means that some of the fuel is either exhausted or locally combusted. Second, internal electrical currents may occur in the electrolyte if it is not a perfect electronic insulator. These two losses are responsible for the difference between the theoretical Nernst potential and measured open circuit voltage (OCV). The relative significance of each of these types of loss is dependent on the load applied. The graph in Fig. 1.1 plots the cell voltage as a function of current density and is labeled with the contributions of the first three sources of loss described above. The redox reaction in an SOFC is split into two half-cell reactions (see Eqs. (1.2) and (1.3)), with one occurring at each electrode and completed by the transport of mobile ions and electrons around separate paths.

SOLID OXIDE FUEL CELL DESIGN The electrochemically active components of conventional SOFCs comprise two porous electrodes, an anode and a cathode, separated by a dense electrolyte. Each of these components must exhibit certain characteristics for the system to function effectively; for example, in a typical SOFC the electrolyte must be gas tight and conductive to ions, but not to electrons. The performance of the anode and cathode is strongly influenced by both their material composition and their porous microstructure [14]. Both electrodes support electrochemical reactions and must also allow for conduction through their bulk and diffusion through their pores. A schematic of a planar cell assembly

6

Solid Oxide Fuel Cell Lifetime and Reliability

with interconnects can be seen in Fig. 1.2. Interconnects are used to collect the current and guide the gas flows, but are also required for stacking multiple cells in series. Early fuel cells were predominantly tubular in design, in part because these systems were much easier to seal, and some developers continue to pursue this design; however, most commercially available systems today are in the planar configuration due to manufacturing considerations, optimal volumetric power density, and the ease of cell stacking. Stacking allows the system to have a higher voltage (series stack) or current (parallel stack) than a single cell. The schematic in Fig. 1.2 shows a series configuration of planar cells. Several other cell geometries have also been developed, a selection of which are shown in Fig. 1.3, including the no longer pursued bell-and-spigot type, which is a means of serial stacking a tubular design with repeat frustum units [15]; the tubular design, which is a tubular cell with diameters in the range of 130 mm [16]; and the flat-tubular geometry, which combines the sealing and mechanical advantages of a tube, whilst maintaining the “stackability” of planar cells [17]. In addition to the geometry of the whole assembly, the relative thickness of each layer of the cell must be optimized. Typically, one of the four layers in the systems will be used as a support onto which the remaining layers can be deposited, as shown in Fig. 1.4. The supporting layer will inevitably be thicker than the others and so must be carefully optimized to minimize the Cathode Electrolyte Anode Interconnect

One cell Fuel Current Air FIGURE 1.2 SOFC schematic of a single cell between two interconnects, showing the passage of fuel and air streams relative to the electrodes.

FIGURE 1.3 Schematic representation of SOFC cell geometries, including (lr) planar, flat tubular, tubular, bell-and-spigot.

An Introduction to Solid Oxide Fuel Cell Materials Chapter | 1

7

Anode Electrolyte Cathode Porous current collector Metal interconnect

FIGURE 1.4 Schematic representation of common cell support configurations in SOFCs.

resulting transport overpotential. Over the years, systems have been developed and operated with each of the four layers chosen as the support; however, the cathode is the least common choice (often as it would result in a significant increase in material costs). A key consideration in the design of SOFC systems is the operating temperature. Operation at increasing temperature not only offers several desirable properties such as higher efficiency and the potential for integration with gas turbines, but also many disadvantages, including higher system costs. Therefore the selection of optimum operating temperature is often related to the end use application.

SOLID OXIDE FUEL CELL OPERATING TEMPERATURE AND MATERIALS Early SOFC devices tended to operate at high temperatures (HTs; 7501000 C), as this was required to stimulate a practical level of ionic conductivity in the electrolytes available at the time (such as yttria-stabilized zirconia (YSZ)) [18]. In ceramic materials, the ionic conductivity and surface exchange kinetics typically increase with temperature, which means that, in theory, HT-SOFCs will operate more efficiently than lower temperature systems [19]. Also, HT operation more readily allows for internal reforming to take place, increasing system efficiency on a wide range of fuels. However, HT devices have several practical disadvantages: first, an HT stack requires a balance of plant (BoP) able to maintain and withstand the temperatures it contains, which often increases the cost of the materials used [20]. Second, to minimize thermo-mechanical damage, SOFCs must be heated gradually before operation, which means that HT-SOFCs take longer to start-up [20]. Third, SOFC interconnects must be resistant to the highly oxidizing and reducing atmospheres; as well as compatible, both chemically and in terms of thermal expansion, with the other components of the cell [21]. This requires either ceramics or nonstandard steels with surface coatings, both of which add cost. Finally, SOFCs need to be gas tight and sealing becomes increasingly difficult at higher temperatures [22], especially during thermal transients. Steele defined a category of SOFC, referred to as intermediate temperature SOFCs (IT-SOFCs), which operate in the 500750 C range [23]. In order to function effectively at these lower temperatures, which Steele believed to be important for commercially viable systems, components

8

Solid Oxide Fuel Cell Lifetime and Reliability

needed to be developed to compensate for the inherently lower activities and conductivities. The two main avenues of investigation, often in combination, were chemical and microstructural augmentation.

Materials Selection The Arrhenius plot shown in Fig. 1.5 (data from [24]) illustrates the strong temperature dependence of conductivity in many common SOFC electrolyte materials. The area-specific resistance (ASR), which is a key metric for SOFC comparison, has an inverse relationship with the conductivity. Wachsman and Lee [25] suggest that, assuming the electrolyte to contribute c. 60% of the total resistance, a viable SOFC would require an ASRelectrolyte , 0.15 Ω cm2. This is reflected in the secondary vertical axis of the graph, which shows the maximum electrolyte thickness allowed by this target. Novel deposition techniques such as suspension spraying [26], colloidal deposition [27], spray pyrolysis [28], chemical vapor deposition [29], and many others have been employed in recent years to try and produce thin electrolytes and interlayers that densify upon sintering, in order to remain gas tight. However, each of these techniques is only suitable for a subset of the materials in Fig. 1.5. YSZ is the most widely used electrolyte because it matches sufficient oxygen ion conductivity with excellent chemical stability, low electronic Temperature, T / °C 1000 900 0

800

700

600

500

400 1500

IT-SOFC –0.5

150

log (σ / S cm–1)

–1.5

LSG

MC

YDC

–2 YS Z

–2.5

CG O

CD

15 SSZ

C

Ca

SZ

–3

Maximum Electrolyte Thickness / µm

YSB

–1

1.5

–3.5 –4

0.15 0.8

0.9

1.0

1.1

1.2

1.3

1.4

1000/T / K–1 FIGURE 1.5 Arrhenius plot of specific ionic conductivity against reciprocal temperature for selected SOFC electrolytes, overlaid with the operational range of different interconnect materials, recreated from a similar figure in [20]. The dashed section of the CGO line illustrates its transition to a mixed conduction regime. The right axis indicates the maximum electrolyte thickness to achieve suggested 0.15 Ohm cm2.

An Introduction to Solid Oxide Fuel Cell Materials Chapter | 1

9

conductivity, reasonable mechanical properties, acceptable cost and relative ease of processing, in particular in thin layers [20]. However, other electrolyte materials are also used by some developers, for example, ceria-based systems, typically when operating temperatures are lower. SOFC anodes are often cermets, which are a composite of an ionically conductive ceramic phase and an electronically conductive metallic phase. Many SOFC cathodes are also composites; however, both phases are usually ceramic. Mixed ionic electronic conductors (MIECs) are conductive to both ions and electrons, and can be used instead of (or as part of) a composite electrode. As the electrochemically active region of the electrodes is typically close to the interface with the electrolyte, they need to be sufficiently porous to allow adequate gas transport. However, this can lead to issues with mechanical stability. The interconnects between each cell must be made from electronically conductive materials that can withstand the harsh chemical environment. For conventional HT cells, this often means using machined ceramics; however, IT-SOFCs are able to utilize steel instead, which can significantly reduce the stack costs and increase mechanical strength. Another important criteria in the selection of materials for SOFCs is chemical compatibility, during both fabrication and operation. Considering each material separately, a YSZlanthanum strontium cobalt ferrite (LSCF) cell appears well optimized for price and performance; however, these two materials can chemically react to form an interfacial layer, which is highly resistive to ionic transport. This means that an interlayer (typically gadolinium-doped ceria (CGO)) is required to separate these materials [30]. CGO itself is not suitable as a single layer electrolyte for applications above 600 C, as it becomes a mixed conductor, which causes short circuit currents and overpotentials [23], lowering the OCV and reducing efficiency.

Microstructural Design In addition to adjusting the materials chemistry to increase electrochemical kinetics and/or ionic/electronic conductivity, increased performance can also be achieved through optimization of the electrode microstructure. The transport and electrochemical processes occurring in SOFC electrodes are closely coupled, which means that the system requires a balance of potentially conflicting geometrical properties, such as high surface area and low tortuosity factor. The tortuosity factor is a measure of the degree of obstruction to transport through a system, caused by the convolution of its geometry. For composite materials, the redox reactions at each electrode are thought to occur close to the triple phase boundaries (TPBs) [31]. TPBs are lines in the geometry at which all three phases are in contact. For a TPB to be electrochemically active, each of the three phases needs to be connected to a path linking it to the circuit, as illustrated in Fig. 1.6. Alternatively, porous MIEC electrodes have primarily dual-phase boundaries, which are simply their surface.

10

Solid Oxide Fuel Cell Lifetime and Reliability Not to scale

Current collector Electronic phase Composite electrode

Active TPB Ionic phase Inactive TPB

Electrolyte FIGURE 1.6 Model of a composite porous SOFC electrode, highlighting phase composition and TPB activity.

IT-SOFC cathodes can suffer from low surface activity, which can be addressed in part by the production of high surface area nano-powders. Solgel thermolysis [32], aqueous gel-casting [33], and thermal spray drying [34] can all produce extremely fine powders, which in turn can be sintered into high surface area electrodes. However, even with these fine powders, conventional electrode fabrication techniques, such as screen-printing, will still result in a broadly random structure, which may not be optimal. Furthermore, initially nanostructured electrodes can be more susceptible to microstructural evolution and coarsening, meaning that their improved performance may be only short lived. Fig. 1.7A shows a scanning electron microscope (SEM) image of a section through a typical cermet anode produced by screen-printing [35] and Fig. 1.7B a 3D rendering of the same anode, reconstructed from focused ion beam (FIB) serial sectioning data. These standard microstructures are commonly modeled as randomly packed spheres which, as discussed by Chueh et al., can lead to high transport resistances [36]. Due to their low aspect ratio, spheres are also predisposed to low intrinsic surface area and poor phase connectivity. In recent years the concept of microstructural design and optimization has gained increasing interest [3739]. This was prompted in part by the emergence of a range of imaging techniques, such as X-ray nano-tomography and FIBSEM, capable of geometric characterization at the relevant length-scales. Perhaps the simplest approach to geometrical optimization is adjusting the volume fraction of each phase and, although this has been investigated for many decades, the emergence of new characterization approaches means that the work continues today [40]. Varying the relative size of component powders is also still investigated frequently [41,42]. The particle size distributions (PSDs) of each phase (including pore phases, where pore formers can be used to control this parameter) have been correlated with expected

An Introduction to Solid Oxide Fuel Cell Materials Chapter | 1

Z

11

2 µm

X Y

Nickel YSZ

2 µm (A)

(B)

FIGURE 1.7 (A) SEM image and (B) FIBSEM reconstruction of a NiYSZ anode produced by screen-printing. Reproduced from Kishimoto M, Iwai H, Saito M, Yoshida H. Quantitative evaluation of solid oxide fuel cell porous anode microstructure based on focused ion beam and scanning electron microscope technique and prediction of anode overpotentials. J Power Sources 2011;196(10):455563 with permission from Elsevier.

cell performance [43]. The effect of the PSD, which is a commonly quoted metric when quantifying powders, may be further improved by using bi-modal distributions. The next most commonly implemented geometrical augmentation is probably functional grading. The coupled electrochemistry and transport phenomena in SOFC electrodes cause different processes to be rate limiting at different points through the electrode thickness. Near the electrolyte interface, high surface area and TPB density are required to speed up the electrochemical reactions, with direct percolating transport paths in the ceramic phase to minimize Ohmic losses. This is often referred to as the electrode “active layer.” As we move further back toward the interconnect, the effective gas transport and percolation of the electronic phase become more important. By fabricating electrodes using multiple thin layers, which is again possible using screen-printing, different inks can be used for each layer to suit the requirements of the system at that depth. Beyond these relatively conventional methods of microstructural augmentation, a variety of more elaborate techniques have been developed, including freeze-casting [44], electrostatic spray deposition [45], physical phaseinversion [46], infiltration [47], and the “breath figures method” [48], amongst many others.

COMMERCIALLY AVAILABLE SOLID OXIDE FUEL CELLS CURRENT TECHNOLOGY STATUS The following is a list of some of the key SOFC developers active at the time of writing: G G

Protonex (United States) Ceres Power (United Kingdom)

12 G G G G G G G

Solid Oxide Fuel Cell Lifetime and Reliability

Bloom Energy (United States) FuelCell Energy (United States) LG Fuel Cell Systems (United States) Hexis (Switzerland) Kyocera (Japan) SolidPower (Italy) FCO Power (Japan)

Between 2008 and 2016, Bloom Energy installed over 140 MWe of SOFC units in the 100 kWe range, including a single 1 MWe unit in 2014 [49]. Mitsubishi Heavy Industries are currently trialing a 200 kWe SOFC, hybridized with a micro gas turbine, which reportedly showed limited degradation and a system efficiency of over 54% [50]. Several large scale SOFC demonstration plants are also receiving government support, including a 400 kWe system produced by FuelCell Energy [51]. Since the widespread adoption of renewable energy sources, interest has been growing in reversible SOFCs as a mechanism for grid balancing [52,53], through the production and storage of hydrogen. The rapid response of these reversible SOFCs gives them a significant advantage over the conventional hydroelectric and gas turbine solutions [54]. Perhaps the most successful fuel cell initiative is the government supported ENE-FARM project in Japan, which is responsible for the installation of more than 160,000 residential systems to date. Although currently only around 20% of these are based on SOFC technology, this ratio is expected to rise as the system price continues to decrease and more data on lifetime becomes available.

INTRODUCTION TO DEGRADATION The focus of this book is on the lifetime and degradation of SOFC systems. This final section of the introduction will give a broad overview of the various challenges still facing SOFCs today. Japan’s New Energy and Industrial Technology Development Organization (NEDO) is currently running a longterm durability study of SOFCs from six manufacturers and has a target of 10% performance loss over 10 years (i.e., 0.11%/1000 h) [55]. Similarly, the US Department of Energy’s (DOE) Solid State Energy Conversion Alliance (SECA) has a target of 0.2%/1000 h over a four-and-a-half-year lifetime [56]. Both of these targets are accompanied by initial efficiency and total stack cost goals to ensure the resulting research is relevant to market needs. From a commercial perspective, durability can be reframed as stack sizing issue (i.e., drawing the same total power from a stack with more cells can help reduce degradation), which allows a costbenefit analysis of cell development to be performed. For example, a study by the US DOE’s

An Introduction to Solid Oxide Fuel Cell Materials Chapter | 1

13

National Energy Technology Laboratory (NETL) suggested that a degradation rate of 0.25%/1000 h (with 5-year stack life) would necessitate a 25% increase in the initial required capacity of an SOFC system in order to meet a specified power demand [57]. SOFCs are highly coupled systems and so it is not always possible to consider any one feature in isolation. Furthermore, the solution to a problem ultimately may be found anywhere from the design stage, through to the manufacture or even the operation of the system. The major issues that will be discussed in this book are introduced below and have been broadly (and somewhat arbitrarily) separated into physical and chemical categories.

Physical Recent advances in imaging technology have allowed direct observation of the changes that occur in the microstructure of SOFC electrodes under various operating conditions. One of the most dramatic morphological modifications occurs when the nickel in common cermet anodes is oxidized (or conversely when the oxide is reduced), as this results in a c. 70% increase in volume of this phase. This growth must be accommodated by the surrounding system and, if induced rapidly, may cause the ceramic network to fail. This makes many SOFC systems sensitive to the “emergency stop” or E-stop; however, some resilience to this redox cycling is a necessity for a commercially viable system (see Chapter 4: The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime). Even without the E-stop, coarsening (or agglomeration) of nickel in cermet anodes in a commonly discussed mechanism for performance loss, as it in turn precipitates a reduction is surface area and TPB length [58]. This effect is accelerated in higher temperature devices, but can potentially be influenced by the presence (intentional or otherwise) of impurities in the metal, which is particularly critical to infiltrated electrodes. Microstructure evolution is also expected to occur in the ceramic phases, especially those that have been nanostructured, again resulting in reduced surface area and performance loss, although typically to a lesser extent than with nickel (see Chapter 5: Microstructural Degradation: Mechanisms, Quantification, Modeling and Design Strategies to Enhance the Durability of Solid Oxide Fuel Cell Electrodes). Gas tightness is required for efficient SOFC operation, but using thick electrolytes to ensure this will result in high Ohmic losses. However, thinner electrolytes are inevitably less mechanically robust and can be susceptible to cracking, especially if exposed to rapid thermal cycling. These thin electrolytes will also be more susceptible to pinhole defects, which can permit gas crossover losses in an otherwise robust cell (as discussed in Chapter 2: Solid Oxide Fuel Cell Electrolytes—Factors Influencing Lifetime).

14

Solid Oxide Fuel Cell Lifetime and Reliability

Delamination, similar to cracking, is aggravated by rapid thermal cycling, as some mismatch between the thermal expansion coefficients of the various layers is inevitable [59]. However, it has also been suggested that this effect can result from a gradual degradation of the interfacial layer due to chemical processes. Particularly in the case of reversible SOFCs (i.e., cells that can act as electrolyzers under a reversed polarity), the oxygen electrode can delaminate at high current densities due to the high pressure gradient of the oxygen produced [60]. Beyond the cell itself, sealing is required to contain the flow of the two gas streams. The design of robust and long-lived seals for HT applications has been investigated for many applications besides fuel cells; however, as is typical of SOFC, this problem cannot be treated in isolation as the chemical composition of the seal must be chosen so as not to poison the cell (see Chapter 7: Lifetime Issues for Solid Oxide Fuel Cell Interconnects).

Chemical One of the most attractive features of SOFCs is their fuel flexibility (especially in comparison to low-temperature polymer electrolyte fuel cells); however, it is common for commercial fuels to contain impurities. Depending on the cell chemistry deployed and the operating temperature, some of these impurities can react with the electrode materials and bind to their surface or integrate into the lattice. When this reaction results in a decrease in cell performance, it is referred to as poisoning. The two electrodes, as well as necessarily being exposed to different gas environments are also sensitive to different poisons. Common poisons for cermet anodes include sulfur-containing compounds and chlorine, both of which are likely to be found in commercial hydrocarbon fuels (discussed in Chapter 3: The Impact of Fuels on Solid Oxide Fuel Cell Anode Lifetime: The Relationship Between Fuel Composition, Fuel Impurities, and Anode Lifetime and Reliability). For the cathode, perhaps the most frequently discussed poison is chromium, which can originate from metallic components in the system (especially interconnects). Even without the effect of poisons, chemical decomposition in SOFC materials can occur during operation. For example, the stoichiometry of strontium containing perovskite cathodes (such as the commonly used LSCF) has been found to change during operation [61], which is likely to affect its surface exchange and bulk diffusion properties (see Chapter 6: Cathode Degradation from Airborne Contaminants in Solid Oxide Fuel Cells: A Review). The deposition of carbon in the pores or “coking” can be a cause of performance degradation in many SOFCs powered by carbon-containing fuels [62]. If formed, the carbon both inhibits gas flow and obstructs active surface area (see Chapter 3: The Impact of Fuels on Solid Oxide Fuel Cell Anode Lifetime: The Relationship Between Fuel Composition, Fuel Impurities, and Anode Lifetime and Reliability).

An Introduction to Solid Oxide Fuel Cell Materials Chapter | 1

15

As discussed previously, although some materials have attractive properties in isolation, they may not be ideal when used together if they react to form low performance layers at their interface. This chemical incompatibility has been addressed in many systems by the application of an interlayer, deposited between the two desired materials, that is chemically compatible with both. Protective coatings may also be used, in particular on metallic interconnects, to manage surface reactions and mitigate accompanying degradation (see Chapter 7: Lifetime Issues for Solid Oxide Fuel Cell Interconnects). In addition to mitigating degradation effects for conventional materials, a significant body of research is focused around the development on novel materials, where the structure and chemistry are carefully tuned at the nanoscale [63] (see Chapter 10: New Materials for Improved Durability and Robustness in Solid Oxide Fuel Cell). Beyond the stack itself, the BoP required to control the temperature and gas stream compositions is composed of a variety of more conventional components (see Chapter 9: Life and Reliability of Solid Oxide Fuel Cell-Based Products: A Review). The heat-exchangers, blowers, control circuitry, and wiring are all susceptible to degradation as in any conventional system, but have benefitted from significant research over the past century. Perhaps the most sensitive and least well understood component of the BoP is the reformer used in many hydrocarbon-fueled systems (discussed in Chapter 8: Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells). As it is a catalytically active component, it is also susceptible to various poisons and must have a carefully controlled operating temperature to minimize degradation.

CONCLUSIONS SOFCs have a range of highly desirable properties when compared to either conventional internal combustion engines or even to low-temperature fuel cells. However, the main obstacle to wider commercial adoption are the degradation challenges presented by lifetime and cycling. The previous section introduces some of the key factors influencing degradation, which are discussed in more detail in the following chapters.

REFERENCES [1] Warshay M, Prokopius PR. The fuel cell in space: yesterday, today and tomorrow. Grove Anniv Fuel Cell Symp 1989;29(1):193200. [2] Faraday M. Siebente reihe von experimental-untersuchungen u¨ber elektricita¨t. Ann Phys 1834;109(17-22):30131. [3] Nernst W. Elektrische lampe mit nernstschem gluhkorper. Berlin AE-G, editor. Germany; 1899. ¨ ber stromliefernde prozesse im konzentrationsgefa¨lle fester elektrolyte. [4] Schottky W. U Wissenschaftliche Vero¨ffentlichungen der Siemens 1935;14:119.

16

Solid Oxide Fuel Cell Lifetime and Reliability

[5] Frenkel J. A general theory of heterophase fluctuations and pretransition phenomena. J Chem Phys 1939;7(7):53847. [6] Schottky W. Zur Stromlieferung geeignetes galvanisches Element bzw. Elementkette. Google Patents; 1937. ¨ ber brennstoff-ketten mit festleitern [Fuel cells with solid conductors]. [7] Baur E, Preis H. U Z Elektrochem. Wiley Online Library 1937;43(9):72732. [8] Kiukkola K, Wagner C. Measurements on galvanic cells involving solid electrolytes. J Electrochem Soc 1957;104(6):379. [9] Palguyev SF, Volchenkova ZS. Das problem der festen elektrolyte fu¨r brennstoffzellen. Tr Inst Khim, Akad Nauk SSSR, Ural Fil 1958;2:183200. [10] Weissbart J, Ruka R. A solid electrolyte fuel cell. J Electrochem Soc 1962;109(8):723. [11] Isenberg AO, Pabst WA, Archer DH. Preparation of thin-film electrolytes for hightemperature fuel cells by chemical vapor deposition. J Electrochem Soc 1970;C197. [12] Hamilton JD. Historical oil shocks. Cambridge, MA: National Bureau of Economic Research; 2011. [13] Lund H. Renewable energy strategies for sustainable development. Energy 2007;32 (6):91219. [14] Bansal NP. Modeling solid oxide fuel cells. Techniques. Spain: Springer; 2008. [15] Minh N. Solid oxide fuel cell technology—features and applications. Solid State Ionics 2004;174(14):2717. [16] Lawlor V, Griesser S, Buchinger G, Olabi AG, Cordiner S, Meissner D. Review of the micro-tubular solid oxide fuel cell. Part I. Stack design issues and research activities. J Power Sources 2009;193(2):38799. [17] Lim TH, Park JL, Lee SB, Park SJ, Song RH, Shin DR. Fabrication and operation of a 1 kW class anode-supported flat tubular SOFC stack. Int J Hydrogen Energy 2010;35 (18):968792. [18] Wang Z, Sun K, Shen S, Zhang N, Qiao J, Xu P. Preparation of YSZ thin films for intermediate temperature solid oxide fuel cells by dip-coating method. J Membr Sci 2008;320 (12):5004. [19] Ivers-Tiffe´e E, Weber A, Herbstritt D. Materials and technologies for SOFC-components. J Eur Ceram Soc 2001;21(1011):180511. [20] Brett DJL, Atkinson A, Brandon P, Skinner SJ, Brandon NP, Skinner SJ. Intermediate temperature solid oxide fuel cells. Encycl Inorg Bioinorg Chem 2008;37(8):1568. [21] Stambouli AB, Traversa E. Solid oxide fuel cells (SOFCs): a review of an environmentally clean and efficient source of energy. Renew Sustain Energy Rev 2002;6(5):43355. [22] Singh RN. Sealing technology for solid oxide fuel cells. Int J Appl Ceram Technol 2007;4(2):13444. [23] Steele BCH. Material science and engineering: the enabling technology for the commercialisation of fuel cell systems. J Mater Sci 2001;36(5):105368. [24] Wincewicz KC, Cooper JS. Taxonomies of SOFC material and manufacturing alternatives. J Power Sources 2005;140(2):28096. [25] Wachsman ED, Lee KT. Lowering the temperature of solid oxide fuel cells. Science 2011;334(6058):9359. [26] Lin B, Sun W, Xie K, Dong Y, Dong D, Liu X, et al. A cathode-supported SOFC with thin Ce0.8Sm0.2O1.9 electrolyte prepared by a suspension spray. J Alloys Compd 2008;465 (12):28590. [27] de Souza S, Visco SJ, De Jonghe LC. Thin-film solid oxide fuel cell with high performance at low-temperature. Solid State Ionics 1997;98(12):5761.

An Introduction to Solid Oxide Fuel Cell Materials Chapter | 1

17

[28] Perednis D. Solid oxide fuel cells with electrolytes prepared via spray pyrolysis. Solid State Ionics 2004;166(34):22939. [29] Will J, Mitterdorfer A, Kleinlogel C, Perednis D, Gauckler LJ. Fabrication of thin electrolytes for second-generation solid oxide fuel cells. Solid State Ionics 2000;131(1):7996. [30] Park S-Y, Ahn JH, Jeong C-W, Na CW, Song R-H, Lee J-H. NiYSZ-supported tubular solid oxide fuel cells with GDC interlayer between YSZ electrolyte and LSCF cathode. Int J Hydrogen Energy 2014;39(24):12894903. [31] Adler SB. Factors governing oxygen reduction in solid oxide fuel cell cathodes. Chem Rev 2004;104(10):4791843. [32] Subramania A, Saradha T, Muzhumathi S. Synthesis of nano-crystalline (Ba0.5Sr0.5) Co0.8Fe0.2O3-δ cathode material by a novel sol-gel thermolysis process for IT-SOFCs. J Power Sources 2007;165(2):72832. [33] Cheng CS, Zhang L, Zhang YJ, Jiang SP. Synthesis of LaCoO3 nano-powders by aqueous gel-casting for intermediate temperature solid oxide fuel cells. Solid State Ionics 2008;179 (7):2829. [34] Serra JM, Uhlenbruck S, Meulenberg WA, Buchkremer HP, Sto¨ver D. Nano-structuring of solid oxide fuel cells cathodes. Top Catal 2006;40(14):12331. [35] Kishimoto M, Iwai H, Saito M, Yoshida H. Quantitative evaluation of solid oxide fuel cell porous anode microstructure based on focused ion beam and scanning electron microscope technique and prediction of anode overpotentials. J Power Sources 2011;196 (10):455563. [36] Chueh CC, Bertei A, Pharoah JG, Nicolella C. Effective conductivity in random porous media with convex and non-convex porosity. Int J Heat Mass Transf 2014;71:1838. [37] Shearing P, Brett D, Brandon N. Towards intelligent engineering of SOFC electrodes: a review of advanced microstructural characterisation techniques. Int Mater Rev 2010;55 (6):34763. [38] Kenney B, Karan K. Engineering of microstructure and design of a planar porous composite SOFC cathode: a numerical analysis. Solid State Ionics 2007;178(34):297306. [39] Kishimoto M, Lomberg M, Ruiz-Trejo E, Brandon NP. Numerical modeling of nickelinfiltrated gadolinium-doped ceria electrodes reconstructed with focused ion beam tomography. Electrochim Acta 2016;190:17885. [40] Riazat M, Baniassadi M, Mazrouie M, Tafazoli M, Moghimi Zand M. The effect of cathode porosity on solid oxide fuel cell performance. Energy Equip Syst 2015;3(1):2532. [41] Moller P, Kanarbik R, Kivi I, Nurk G, Lust E. Influence of microstructure on the electrochemical behavior of LSC cathodes for intermediate temperature SOFC. J Electrochem Soc 2013;160(11):F124553. [42] Harris WM, Nelson GJ, Lombardo JJ, Cocco AP, Izzo JR, Chiu WKS, et al. Analysis of solid oxide fuel cell LSM-YSZ composite cathodes with varying starting powder sizes. ASME 2011 international mechanical engineering congress and exposition. Denver, Colorado, USA: American Society of Mechanical Engineers; 2011. p. 14. Available from: http://proceedings.asmedigitalcollection.asme.org/proceeding.aspx? articleid=1643183. [43] Fukui T, Murata K, Ohara S, Abe H, Naito M, Nogi K. Morphology control of NiYSZ cermet anode for lower temperature operation of SOFCs. J Power Sources 2004;125 (1):1721. [44] Lichtner AZ, Jauffre`s D, Roussel D, Charlot F, Martin CL, Bordia RK. Dispersion, connectivity and tortuosity of hierarchical porosity composite SOFC cathodes prepared by freeze-casting. J Eur Ceram Soc 2015;35(2):58595.

18

Solid Oxide Fuel Cell Lifetime and Reliability

[45] Sar J, Charlot F, Almeida A, Dessemond L, Djurado E. Coral microstructure of graded CGO/LSCF oxygen electrode by electrostatic spray deposition for energy (IT-SOFC, SOEC). Fuel Cells 2014;14(3):35763. [46] Li T, Wu Z, Li K. A dual-structured anode/Ni-mesh current collector hollow fibre for micro-tubular solid oxide fuel cells (SOFCs). J Power Sources 2014;251:14551. [47] Kishimoto M, Lomberg M, Ruiz-Trejo E, Brandon NP. Enhanced triple-phase boundary density in infiltrated electrodes for solid oxide fuel cells demonstrated by high-resolution tomography. J Power Sources 2014;266:2915. [48] Li J, Zhang N, Ni D, Sun K. Preparation of honeycomb porous solid oxide fuel cell cathodes by breath figures method. Int J Hydrogen Energy 2011;36(13):76418. [49] Bloom Energy. 1 MW Bloom Energy project at Maxim San Jose Headquarters [Internet]. 2016 [cited 2016 Dec 14]. p. 1. Available from: http://www.bloomenergy.com/customerfuel-cell/maxim-integrated-installs-fuel-cells/. [50] Ando Y, Oozawa H, Mihara M, Irie H, Urashita Y, Ikegami T. Demonstration of SOFCmicro gas turbine (MGT) hybrid systems for commercialization. Mitsubishi Heavy Ind Tech Rev 2015;52(4). [51] DOE selects research projects to advance SOFC technology. Fuel Cells Bull 2015;123. [52] Rajashekara K, Rathore AK. Power conversion and control for fuel cell systems in transportation and stationary power generation. Electr Power Components Syst 2015;43 (12):137687. [53] Graves C, Ebbesen SD, Mogensen M. Co-electrolysis of CO2 and H2O in solid oxide cells: performance and durability. Solid State Ionics 2011;192(1):398403. [54] Kusumi N, Hino N, Thet AK. A new concept for power grid stabilization using a motor-assisted variable speed gas turbine system. ASME 2014 international mechanical engineering congress and exposition. Quebec, Canada: American Society of Mechanical Engineers; 2014. Available from: http://proceedings.asmedigitalcollection.asme.org/ proceeding.aspx?articleid=1643183. [55] Yokokawa H. Current status of rapid evaluation of durability of six SOFC stacks within NEDO project. ECS Trans 2015;68(1):182736. [56] Gerdes K, Richards G. Advanced fuel cell research at NETL; 2012. [57] Thijssen J. Natural gas-fueled distributed generation solid oxide fuel cell systems: projection of performance and cost of electricity. 2009. [58] Holzer L, Iwanschitz B, Hocker T, Keller L, Pecho O, Sartoris G, et al. Redox cycling of Ni-YSZ anodes for solid oxide fuel cells: influence of tortuosity, constriction and percolation factors on the effective transport properties. J Power Sources 2013;242:17994. [59] Keane M, Mahapatra MK, Verma A, Singh P. LSM-YSZ interactions and anode delamination in solid oxide electrolysis cells. Int J Hydrogen Energy 2012;37(22):1677685. [60] Virkar AV. Mechanism of oxygen electrode delamination in solid oxide electrolyzer cells. Int J Hydrog Energ 2010;35(18):952743. [61] Oh D, Gostovic D, Wachsman ED, Simner SP, Anderson MD, Engelhard MH, et al. Mechanism of La0.6Sr0.4Co0.2Fe0.8O3 cathode degradation. J Mater Res 2012;27 (15):19929. [62] Yang L, Wang S, Blinn K, Liu M, Liu Z, Cheng Z, et al. Enhanced sulfur and coking tolerance of a mixed ion conductor for SOFCs: BaZr0.1Ce0.7Y0.2xYbxO3δ. Science 2009;326(5949):1269. [63] Irvine JTS, Neagu D, Verbraeken MC, Chatzichristodoulou C, Graves C, Mogensen MB, et al. Evolution of the electrochemical interface in high-temperature fuel cells and electrolysers. Nat Energy 2016;1(1):15014.

Chapter 2

Solid Oxide Fuel Cell Electrolytes—Factors Influencing Lifetime Alan Atkinson Imperial College London, London, United Kingdom

Chapter Outline Introduction Structural Stability of Electrolytes Chemical Interactions La1xSrxMnO3/Yttria-Stabilized Zirconia Interactions Ce1xGdxO2x/2Yttria-Stabilized Zirconia Interdiffusion in Bilayer Electrolytes

19 20 21 21

22

Mechanical Degradation Sources of Stress Mechanical Failure Slow Crack Growth Creep Thermal and Redox Cycling Closing Remarks References

24 24 27 29 31 33 33 34

INTRODUCTION The processes influencing long-time degradation of the electrolyte have varied origins. The most fundamental is the inherent chemical stability of the electrolyte itself, which can lead to a gradual decrease in its conductivity and hence increase in its contribution to the cell area-specific resistance (ASR). The second comes from chemical interactions between the electrolyte and the materials it is in contact with, particularly the cathode. This can lead to reaction zones at interfaces that have low ionic conductivity and again increase the cell ASR. Finally, there is possible mechanical instability of the electrolyte arising from thermal gradients, thermal cycles, or redox cycles. This can cause fracture of the electrolyte leading to direct combustion of fuel or even catastrophic failure. In this chapter, all three degradation routes are surveyed, with emphasis on mechanical degradation.

Solid Oxide Fuel Cell Lifetime and Reliability. DOI: http://dx.doi.org/10.1016/B978-0-08-101102-7.00002-7 Copyright © 2017 Elsevier Ltd. All rights reserved.

19

20

Solid Oxide Fuel Cell Lifetime and Reliability

STRUCTURAL STABILITY OF ELECTROLYTES In principle electrolyte compositions are selected to provide a high and stable oxygen ion conductivity. However, this is not easily achieved in practice because atomistic structural changes can take place slowly over long periods of time at operating temperatures that are lower than the temperatures at which the electrolytes are processed by sintering. Examples of these changes are illustrated in Fig. 2.1 [1] for zirconia electrolytes and typically involve the slow rearrangement of point defects. In the case of zirconia trivalent cation dopants are added to stabilize the cubic structure and introduce mobile oxygen vacancies. The most widely used stabilized zirconia electrolyte contains 8 mol% Y2O3 (8YSZ). It is well known for stabilized zirconia, that the oxygen vacancies are attracted to the trivalent substitutional cations (which have opposite effective electrical charge) to form defect clusters in which the oxygen vacancies are strongly bound and do not contribute to high oxygen ion conduction. These clusters capture a larger fraction of the total oxygen vacancies as the concentration of trivalent ions increases and as the temperature decreases. However, the rate at which larger stable clusters form is controlled by the migration of the slowly diffusing trivalent cations and therefore they form only slowly at solid oxide fuel cell (SOFC) operating temperatures leading to a gradual decrease in ionic conductivity. This is exacerbated in 8YSZ because the composition is on the edge of the cubic phase field at the sintering temperature and falls

FIGURE 2.1 Evolution of ionic conductivity of zirconia electrolytes with time at temperature. After Mu¨ller AC, Weber A, Ivers-Tiffe´e E. Degradation of zirconia electrolytes. In: Proceedings of the 6th European solid oxide fuel cell forum, 2004, p. 12318.

Solid Oxide Fuel Cell Electrolytes—Factors Influencing Lifetime Chapter | 2

21

into the tetragonal phase field at typical operating temperatures. Therefore this composition in principle is inherently unstable in operation and microdomains of clusters form with similar characteristics to the local tetragonal arrangement [2]. As shown in Fig. 2.1, this results in a substantial drop in conductivity of 8YSZ over 3000 h at 950 C. At the lower temperature of 861 C the reduction is much slower reflecting the high activation energy for cation diffusion in zirconia. Also shown in Fig. 2.1 are data for 10YSZ and 10Sc1CeSZ (10 mol% Sc2O3, 1 mol% CeO2, 89 mol% ZrO2). Both are fully stabilized in the cubic structure and as a result show much less degradation in conductivity than does 8YSZ. For electrolyte-supported cells, 3YSZ is often preferred because it has much better mechanical properties, even though it has much lower conductivity. Again 3YSZ also shows a slow degradation in conductivity at 950 C. In all cases the degradation was fitted to an exponential decay toward a long-term stable value [1]. In the case of 8YSZ, this was B50% of the initial conductivity, but was smaller for the other materials. Degradation at lower temperatures over much longer times is uncertain given that the process is so slow because of the high activation energy. Nevertheless, the thermodynamic driving force for degradation is still present, and probably even increased, so that degradation cannot be discounted, particularly for 8YSZ.

CHEMICAL INTERACTIONS These potentially deleterious effects are encountered mainly during cell fabrication at temperatures (11001500 C) significantly higher than operating temperatures (6001000 C). Strategies to avoid the unwanted interactions are therefore aimed at fabrication processes and success can be evaluated immediately after fabrication. However, although operating temperatures are much lower than fabrication temperatures, the exposure times are much longer and there are steady DC electrical fields that can accelerate cation drift. Therefore slow degradation caused by chemical interactions during operation cannot be ruled out. Some important examples are discussed below.

La1xSrxMnO3/Yttria-Stabilized Zirconia Interactions It is well known that adverse reactions can occur between stabilized zirconia electrolytes and many common perovskite-structure cathodes. For example, the perovskite La1xSrxMnO3 (LSM) can react with zirconia electrolytes to give as products La2ZrO7 and SrZrO3 that have low ionic conductivity and increase the ohmic losses. This interaction also has implications for the longterm stability of LSM/YSZ composite cathodes. This LSM/YSZ interaction can be inhibited by using A-site deficient LSM that decreases the activities of La and Sr oxides [3]. Nevertheless, thermodynamic analysis [4] suggests that the reaction will still occur and is promoted by cathodic bias in

22

Solid Oxide Fuel Cell Lifetime and Reliability

operation which, as it corresponds to more reducing conditions, reduces the stability of the LSM. The reaction between A-site deficient LSM and YSZ is reported to be controlled by the diffusion of Mn into the YSZ which increases the A-site occupancy at the interface between the LSM and YSZ where reaction occurs to give the aforementioned low-conductivity products [5]. Chen et al. [4] showed that when La0.65Sr0.30MnO3 electrodes were fired onto YSZ at 1400 C for 1 h an amorphous interdiffusion zone B70 nm thick was formed. Subsequent annealing at 1000 C for 1000 h crystallized this zone with reaction products and it thickened to 100 nm. The parabolic rate constant for the reaction zone thickness is given approximately by   2147 kJmol21 kðm2 s21 Þ 5 1:2 3 10213 exp ð2:1Þ RT Brant et al. [6] studied the effect of the LSM/YSZ interaction on the electrode ASR. They report that annealing at 1400 C for 4 h produced a contribution to the ASR attributable to the reaction products of B14,000 Ω cm2 at 900 C. From Eq. (2.1) the reaction zone is estimated to have a thickness of 210 nm showing the highly resistive nature of the reaction product layer. Although this analysis is approximate, nevertheless it is sufficient to indicate that chemical interaction between LSM cathodes and YSZ electrolytes cannot be discounted over long times even at moderate operating temperatures. Malzbender et al. [7] carried out a post-test analysis of a stack with 8YSZ electrolyte and La0.65Sr0.30MnO3 cathode that had been operated for 19,000 h at 800 C under a steady current load of 0.5 A cm22. Mn was found to have penetrated the YSZ grain boundaries and generated cracks within the YSZ. There was no evidence of reaction products at the LSM/YSZ interface contrary to what might have been expected from the higher temperature experiments described earlier. In that particular stack, it was not possible to separate different contributions to the overall degradation rate (B0.01 V in 1000 h) but the electrolyte cracks induced by Mn diffusion were held responsible for the eventual mechanical failure of the stack.

Ce1xGdxO2x/2Yttria-Stabilized Zirconia Interdiffusion in Bilayer Electrolytes For operating temperatures below B800 C, more active cathodes than LSM have entered widespread use. These often contain cobalt (e.g., La1xSrxCoO3 (LSC) and La12xSrxCo12yFeyO3 (LSCF)), but their superior cathode performance compared with LSM is correlated with lower chemical stability. If they are used in direct contact with zirconia electrolytes, they react more rapidly than does LSM to produce the zirconate low-conductivity phases at the interface. In order to prevent, or slow down, this interaction, it has become common to use a bilayer electrolyte. This typically comprises a so-called

Solid Oxide Fuel Cell Electrolytes—Factors Influencing Lifetime Chapter | 2

23

barrier layer of doped ceria (e.g., Ce1xGdxO2x/2 (CGO)), with which the cobalt-containing cathodes do not react, between the cathode and zirconia to block cation diffusion from the electrode into the zirconia. Another use of a bilayer electrolyte is when a doped ceria main electrolyte is used because of its higher ionic conductivity at lower operating temperatures. However, doped ceria has appreciable electronic conductivity under fuel side conditions, which results in an electronic leakage current that reduces cell efficiency. A thin layer of doped zirconia serves to block the electronic current without increasing too much the ASR of the combination. Tsoga et al. [8] studied interdiffusion between CGO (x 5 0.2) and 8YSZ. These materials have the same fluorite structure and therefore form a full range of solid solution. They found that the interdiffusion zone had a width of several microns after fabrication and rather surprisingly the width did not seem to vary with sintering conditions ranging from 3 h at 1200 C to 10 h at 1500 C. (More recently Gao et al. [9] report that firing temperature does have a significant effect on the extent of the interdiffusion zone as would be expected.) Homogeneous bulk solid solution samples of equimolar CGO and YSZ were also prepared and found to have ionic conductivity 1 to 2 orders of magnitude lower than YSZ. The self-diffusion coefficient of Zr in the 8YSZ lattice is reported by Kilo et al. [10] to be   2430 kJmol21 DZr=8YSZ ðm2 s21 Þ 5 3:3 3 1026 exp ð2:2Þ RT in the temperature range 11251460 C. The activation energy in Eq. (2.2) is much greater than that for the LSM/YSZ reaction parabolic rate constant in Eq. (2.1) and therefore interdiffusion between CGO and YSZ is unlikely to cause degradation at typical operating temperatures even over long times. As an example of a tetravalent cation, grain boundary diffusion of Ti41 in YSZ has been reported to be many orders of magnitude faster than lattice diffusion and has a lower activation energy of 340 kJ mol21 [11]. Even this is very high and it is unlikely that significant CGO/YSZ solid solutions would form by grain boundary diffusion at cell operating temperatures and lifetimes. As mentioned earlier, one of the reasons for using bilayer electrolytes is to prevent reaction between cobalt-containing cathodes and zirconia electrolytes. Therefore it is necessary to consider how well a doped ceria barrier layer performs in this regard over the long term. Khan et al. [12] showed that the density of a CGO barrier layer was a crucial parameter for slowing down degradation from a cathode containing LSCF. Their CGO barrier layers were formed by slurry coating on ScSZ and sintering at 1400 C for 5 h. The degradation of cathode ASR was B0.1 Ω cm2 after 1000 h at 900 C and was attributed to Co and Sr diffusion through the CGO barrier. Matsui et al. [13] studied the microstructural changes in a similar bilayer electrolyte formed by sintering an Sm-doped

24

Solid Oxide Fuel Cell Lifetime and Reliability

ceria (SDC) layer onto a thicker 8YSZ electrolyte for 5 h at 1200 C which gave a porous SDC barrier layer. The LSCF cathode was exposed in this arrangement at 1000 C and 0.3 A cm22 and degradation in ohmic resistance was B0.15 Ω cm2 after 400 h. High test temperatures were used in these studies to accelerate the degradation, but there is insufficient information at present to estimate their likely influence at lower temperatures and much longer times.

MECHANICAL DEGRADATION Sources of Stress Thermal Stresses Thermal expansion is a common cause of stress in SOFC cells and stacks. The thermal expansion stress-free strain, εt, is given by εt 5 αðT 2 T0 Þ

ð2:3Þ

where α is the coefficient of thermal expansion (CTE) and T0 is a reference temperature. This thermal strain can give rise to a stress if the material is constrained in some way, such as the temperature change not being uniform or being bonded to a material with a different CTE. Residual stresses are due to differences in thermal expansion coefficients of the different component materials as they cool from high manufacturing temperatures where they are almost stress-free. The residual stresses are therefore maximum at room temperature. It is often the case that one component in the structure is far stiffer mechanically than the rest. In such cases the other components will be forced (as long as they are adherent) to be compatible with the stress-free strains and displacements of the stiffest component, which remains almost stress-free because of its greater stiffness. In planar structures (that are also constrained to remain planar) the stress state remote from edges and discontinuities is equi-biaxial in the plane. These stresses give an elastic strain contribution εe 5

σð1 2 νÞ E

ð2:4Þ

where E is Young’s modulus and ν is Poisson’s ratio of the appropriate layer in the structure. Although the different geometry of tubular cells changes the details (the strains are axial and tangential) the same principles and general conclusions apply. The total strain is εtot 5 εt 1 εe

ð2:5Þ

and must be equal for all the components in the planar structure. When these simplifying assumptions are not met (for example, if a planar cell is allowed to bend) then the stresses can still be calculated from the strains by using the equilibrium conditions that total force and total moment must be zero.

Solid Oxide Fuel Cell Electrolytes—Factors Influencing Lifetime Chapter | 2

25

TABLE 2.1 Typical Thermo-Mechanical Properties of Some SOFC Materials (at 25 C Except for CTE) Material

CTE (ppm K21) (251000 C)

Young’s Modulus, E (GPa)

Fracture Toughness, KIc (MPa m1/2)

Critical Stress, σc (MPa)

8YSZ

10.5

215

1.7

416

3YSZ

10.8

211

5.7

770

CGO

12.5

200

0.6

120

Porous LSM

12.4

35

0.7

46

Porous LSCF

15.3

45

0.6

NiO/YSZ

12.8

70

1.7

90

Ni/YSZ

12.5

30

2.5

75

Crofer 22APU

12.7

220

270a

a

Yield stress.

Residual stresses in electrolytes can vary greatly depending on the cell materials and geometry. Some typical values of thermal expansion coefficients for common SOFC materials are presented in Table 2.1. It is clearly seen that the CTE for doped zirconia is significantly lower than for other common SOFC materials. As a result, thin zirconia electrolytes supported on any of the other materials in Table 2.1 will have compressive residual stresses. These are beneficial for the integrity of the electrolyte since they help to suppress fracture. Values of residual stresses in anode-supported 8YSZ electrolytes measured at room temperature by X-ray diffraction are typically 600 MPa (negative sign indicates compressive stress) before reduction and 450 MPa after reduction [14]. The stress before reduction is consistent with cooling from a stress-free temperature of B1200 C and the lower magnitude of stress after reduction is the result of some plastic deformation partially relieving the stress during the reduction process. This has also been reported by Frandsen et al. [15] who suggest that this is due to relaxation of internal microstresses as the stiff NiO is transformed to softer Ni and the porosity increases. Malzbender et al. [16] showed that the residual electrolyte stress in similar cells before reduction gradually reduced in magnitude with increasing temperature and became almost zero at 900 C, again indicating some stress relief at the higher temperatures.

26

Solid Oxide Fuel Cell Lifetime and Reliability

X2

X3

X1

X

Stress (MPa) +3 +2 +2 +2 +1 +1 +1 +1 +1 +8 +6 +4 +2 +0 –3

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

981e+01 500e+01 292e+01 083e+01 875e+01 667e+01 458e+01 250e+01 042e+01 333e+00 250e+00 167e+00 083e+00 000e+00 721e+00

FIGURE 2.2 Distribution of maximum principal stress in the anode support of an anodesupported cell, at 800 C and 300 mA cm22, calculated by FEM [17]. The upper contour plot shows the plan view and the lower contour plot shows the cross section at the mid point. The dashed line is a symmetry plane.

Stresses can also arise from temperature gradients in a plate of a single material because the thermal strain is different at different locations. These stresses have been studied extensively in the fuel cell context by finite element modeling (FEM) approaches. These are usually combined with fluid dynamics, electrochemical, and heat transfer models to obtain temperature distributions from which the stress distributions are calculated and have been applied to both steady-state and transient conditions (thermal cycles and power cycles). The results are very specific to particular cell and stack geometries and operating conditions. Fig. 2.2 shows an example of a simulated stress distribution in the anode support of a cell under operating conditions. The cell is operated in co-flow geometry with the fuel (hydrogen) and air entering from the bottom of the plan view. The maximum tensile stresses are of the order of 40 MPa, which can be compared with a typical strength of 75 MPa for a Ni/YSZ support in Table 2.1.

Solid Oxide Fuel Cell Electrolytes—Factors Influencing Lifetime Chapter | 2

27

Chemically Induced Stresses Some SOFC materials can develop strains (change dimensions) as a result of changes to their chemical environment. Such chemical strains must be added to the thermal and elastic strains in Eq. (2.1). Probably the most important of these is the irreversible expansion of Ni-based anodes if the Ni is oxidized to NiO, either as a result of excessive fuel utilization or a fuel supply failure. The general features of this so-called “redox” can be found in review articles such as [18,19] and in Chapter 4, The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime of the current volume. The expansion strain can be very large (of the order of 1%) and depends mainly of the degree of oxidation, the Ni content, and the porosity of the anode. The consequences for the integrity of the electrolyte depend on the configuration of the cell and particularly on the thickness of the anode. The threat to the electrolyte is therefore highest for anode-supported thin electrolytes which can fracture once a critical strain is exceeded. Another example of a chemically induced stress occurs with doped ceria electrolytes. On the anodic side of a doped ceria electrolyte, partial reduction of Ce41 to Ce31 occurs which increases as the temperature increases. This not only results in the electronic conductivity discussed earlier, but also causes an expansion of the fluorite lattice that is not uniform across the electrolyte thickness. The resulting stresses can be severe and effectively restrict the maximum operating temperature of doped ceria electrolytes to B650 C [20]. In addition to all of the above, are stresses imposed as a result of external forces such as stack compression. These must be added to the residual stresses to assess whether damage is likely, bearing in mind that residual stresses are usually maximum at room temperature and tend to be relaxed at operating temperature.

Mechanical Failure After estimating the likely stresses in the cell components, it is necessary to evaluate how the structure will respond to the stresses. For the brittle materials, such as the electrolyte, there is the possibility of failure by cracking. In addition, cracks in different locations will have different consequences depending on their size and state of stress. For the electrolyte the most severe will be a crack across the thickness which would allow direct combustion of fuel creating a “hot spot.” Even if this is local at first, the heat generated by the combustion will probably be sufficient to extend the local crack into a large scale failure. Brittle failure of ceramics involves first the initiation of a crack (e.g., at a defect or sharp geometrical feature) and then the propagation of the crack driven by the stress field, which changes as the crack propagates. Although the underlying physical processes are the same in both stages, the initiation is usually described in terms of a critical value

28

Solid Oxide Fuel Cell Lifetime and Reliability

of stress (strength, σc) and the propagation by a critical energy release per unit area of crack extension, Gc, equal to the energy required to create the new surfaces of the crack and any energy dissipation processes involved in fracture (such as some plasticity near the crack tip). The critical energy release depends on the direction of stress compared with the direction of the crack propagation and is referred to as the mode of loading. The simplest mode of loading is Mode I, or opening mode, in which the stress is perpendicular to the crack. The local stresses in the vicinity of the crack tip are characterized by the stress intensity K, and for opening mode loading this is denoted KI. When this reaches a critical value at which the crack extends this is KIc, also known as the fracture toughness, and is a property of the material. Typical values of fracture toughness for some SOFC materials are listed in Table 2.1. G and K are related by ^ K 2 5 EG

ð2:6Þ

where Eˆ 5 E for plane stress and E/(1  ν ) for plane strain. The strength is not a true property of the material since it depends on the defects that initiate the fracture. For a crack, of length 2c, under constant farfield tensile stress, the critical stress (strength) to extend the crack is given by 2

σc 5

KIc Yc1=2

ð2:7Þ

where Y is a dimensionless parameter of order unity that depends on the crack shape. For this reason the strength is a statistical parameter with a distribution of values reflecting the distribution of defect sizes and leads to a probability of fracture initiation. Although ideally one should consider both initiation and propagation together, it is common to simplify the analysis by assuming that one is dominant. Thus in a strength-based analysis, it is assumed that once a crack is initiated it can propagate easily. Conversely an energy-based analysis assumes that there will always be a defect large enough to initiate a crack and that the controlling criterion is the energy available to propagate it. In either case the assumption that one of the steps is easy can lead to a pessimistic appraisal of failure. The strength criterion is easier to apply in complicated situations because it does not require knowledge of the stress distribution as the crack propagates. However, it will be pessimistic in cases where a large tensile stress region is very localized or is surrounded by a compressive region that is not relaxed as the crack propagates toward it. The statistical nature of a strength-based assessment is usually done using a Weibull distribution for the strength, which in its simplest form gives a probability of failure initiated at defects in the volume of the material  ð  m  1 σ Pf 5 1 2 exp 2 dV ð2:8Þ V0 σ0

Solid Oxide Fuel Cell Electrolytes—Factors Influencing Lifetime Chapter | 2

29

where σ0 is the Weibull characteristic stress, V0 is a calibration volume, and m is the Weibull modulus, which defines the spread of the strength distribution. These parameters are obtained in well-defined tests and depend not only on the material, but also on its microstructure and method of fabrication. A consequence of Eq. (2.8) is that the failure probability increases with the size of the stressed volume, which reflects the probability of finding ever larger defects as the volume increases. This is often unrealistic, especially in materials with small dimensions. In such cases Eq. (2.8) may be modified by introducing a threshold stress (corresponding to a maximum defect size) below which no failure occurs. For an example, see [21]. Further complications arise when the state of stress is multiaxial. One common way of dealing with this is to use the maximum principal stress as the variable in Eq. (2.8). This is why the stress that is plotted in Fig. 2.2 is the maximum principal stress. The rationale for this is that the crack will propagate normal to the stress which is as close as possible to Mode I loading. The critical component for the strength-based analysis depends on the cell configuration. For an electrolyte-supported cell, it is the electrolyte itself and this is why such cells are often made with 3YSZ rather than 8YSZ. Even though 3YSZ has the lower conductivity, it has much greater toughness and strength than 8YSZ. Conversely for an anode-supported electrolyte the critical component in most scenarios is the support itself because if the support breaks the electrolyte will not survive. However, the situation is different when considering electrolyte fracture as a result of redox expansion of the support. In this case the stress in the support is small, but the chemical strain is large. This situation is more easily analyzed using the energy approach and results in a so-called channel crack with an energy of propagation condition [22] Gch 5 ω

σ2 h . Gc E^

ð2:9Þ

In this expression, σ is the stress in the electrolyte, h its thickness, and ω is a numerical factor that depends on the relative elastic properties of the electrolyte and support.

Slow Crack Growth The problem with an approach based on elastic properties and fast fracture is that there is no time-dependent phenomenon involved and so it is incapable of accounting for a slow degradation or a lifetime of the device. Timedependence can enter in a variety of ways and increases the complexity considerably. These include: changes in properties of materials with exposure, slow (subcritical) crack growth, and creep. Changes in material properties can be caused by coarsening, reactions between phases or with the environment, or as a result of cycling. For

30

Solid Oxide Fuel Cell Lifetime and Reliability

example, if one of the electrodes loses activity at different rates at different locations as a result of coarsening or contamination, this will gradually change the temperature distribution in the stack and hence the stress distribution. Oxidation of steel interconnects produces a reaction product scale that thickens gradually which can lead to weakening of seals. Redox and thermal cycling lead to redistribution of Ni in Ni-based anodes resulting in changes in their thermo-mechanical properties [14]. Slow sub-critical crack growth is a common feature in ceramics and glass in which a crack that is not in a critical state for fast fracture will slowly extend under the combined action of stress and the environment weakening the atomic bonding at the crack tip. The crack growth velocity is given by v 5 AKIn (for Mode I loading). As the crack grows the stress intensity increases, the velocity increases, and the crack elongates until KI reaches KIc and then rapid fracture occurs. When combined with the Weibull statistical strength distribution, it leads to a strengthprobabilitytime (SPT) relationship. Slow crack growth data for 8YSZ electrolyte in laboratory air at room temperature are shown in Fig. 2.3 [23]. In this case the exponent n B 20 and there appears to be a threshold stress intensity below which there is no detectable crack growth. Choi and Bansal [24] also report slow crack growth in 10YSZ at 1000 C, but with a lower exponent of 8. The time to failure at a stress, σ, is given by [25] "  n22 # σn22 σ c ð2:10Þ tf 5 B n 1 2 σc σ where σc is the strength at zero time and

FIGURE 2.3 Crack growth velocity as a function of stress intensity for 8YSZ in air at 25 C [23].

Solid Oxide Fuel Cell Electrolytes—Factors Influencing Lifetime Chapter | 2

31

FIGURE 2.4 The calculated effect of subcritical crack growth on the failure probability of a typical 8YSZ tape-cast electrolyte plate after 10,000 h at room temperature or 1000 C.

B5

2 AY 2 ðn 2 2Þ

KIc22n

ð2:11Þ

In this equation, Y is the geometric parameter for the strength-controlling defect in Eq. (2.7). Combining Eq. (2.10) with a typical Weibull distribution for a 5 3 5 cm2 tape-cast 8YSZ electrolyte plate gives the SPT diagram of Fig. 2.4. This predicts that over long timescales of several thousand hours, slow crack growth reduces the effective strength of the YSZ by B30% at room temperature. At 1000 C for 10YSZ the mean initial strength is lower and the slow crack growth exponent n is also lower [24]. Assuming the same Weibull modulus the calculated SPT relationship is also shown in Fig. 2.4. It illustrates that at high temperatures the effective strength for a given failure probability is reduced by a combination of a lower short-term characteristic strength and a greater susceptibility to subcritical crack growth. The combined effect is equivalent to strength being reduced to approximately only one-third of its short-term value.

Creep At high-temperature, materials subjected to stresses for extended time periods undergo creep processes in which extra strains are produced. The creep strain, εc, must be added to the thermal and elastic strains in Eq. (2.5) to give the total strain. Creep generally progresses through an initial transient stage (primary creep) in which the creep rate decreases with time at constant stress, to a steady state of constant creep rate (secondary creep) and finally

32

Solid Oxide Fuel Cell Lifetime and Reliability

at an accelerating rate leading to rupture (tertiary creep). In the steady-state creep region the rate of change in strain with time for a material may be described by,   Qc ε_ c 5 Aσnc L2p PO2 m exp 2 ð2:12Þ RT where σ is the applied stress, nc is the stress exponent, L is the grain size, p is the grain size exponent, PO2 is the oxygen partial pressure, m is the oxygen partial pressure exponent, Qc is the creep activation energy, R is the universal gas constant, and T is the absolute temperature. The rate is controlled by diffusion of the slowest element in a compound which, for SOFC electrolytes, is usually the most highly charged cation (i.e., Zr in YSZ). From the point of view of long-term stability of the electrolyte, creep can be beneficial or detrimental. The beneficial effect comes from the relaxation of some stresses in the electrolyte, whereas detrimental effects come from extra strains in other components such as supports or bipolar plates that can increase stress in the electrolyte. Greco et al. [26] provide an example of a thermo-mechanical analysis of an operating stack incorporating creep (but not slow crack growth) leading to lifetime estimation using the strength approach for failure outlined above. A typical result is shown in Fig. 2.5. Fig. 2.5 shows that in this example creep has a large effect on reducing the failure probability of the anode support, but gradually increases the failure probability of the electrolyte. The different power levels have most effect in the transient conditions and particularly at the lower power levels when the cell temperature is lower and creep is not so effective. 20

1E-00 18

1E-02 1E-04 Anode Electrolyte Comp. layer

1E-06 1E-08 1E-10

16 14

Current intensity (A)

Probability of failure (%)

1E-02

12

1E-12 1E-14 0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10 10000

Time (h)

FIGURE 2.5 Failure probability of cell constituents from a simulation of an anode-supported cell in a repeat unit in which creep processes were considered at a nominal temperature of 850 C. The cell was assumed to deliver a sequence of output current levels as shown by the solid curve [26]. The “compatibility layer” in the diagram is a barrier layer of doped ceria to prevent chemical interaction between the LSCF cathode and the 8YSZ electrolyte.

Solid Oxide Fuel Cell Electrolytes—Factors Influencing Lifetime Chapter | 2

33

Thermal and Redox Cycling In thermal cycling the change in temperature causes thermal expansion mismatch strains between different materials and the temperature gradients cause additional stresses. The severity of the transient stresses depends on the detailed geometry and the rates of temperature change. In principle the approaches outlined above can be applied to thermal cycling, but to predict a thermal cycling lifetime the analysis must include time-dependent mechanical processes. This is because it is generally observed that cells will fail after some number of nominally identical thermal cycles. However, an analysis only including thermo-elastic effects will predict either failure on the first cycle or survival for an infinite number of cycles. The reason for survival for a finite number of cycles and then failure is not understood, but might be indicative of subcritical crack growth in the structure. In addition the cycling can result in changes to the properties of the materials as a consequence of nonreversible plastic deformation that increases incrementally with each cycle. This is common in metals and is known as ratcheting. For example, it has been found that the CTE of Ni/YSZ anode supports increases with thermal cycling, probably due to plastic redistribution of nickel in the porous microstructure [14]. This gradually increases the compressive residual stress in the electrolyte, which should increase its resistance to failure, unless it becomes so great that the electrolyte delaminates from the support. Another cycling threat comes from redox cycling. This is an example of a chemical strain that is not fully reversible because the distribution of nickel in the anode or, anode support, is slightly different with each cycle and the redox strain increases with each cycle [27] (see Chapter 4: The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime for further details).

CLOSING REMARKS The lifetime of an SOFC electrolyte can be compromized by either chemical or mechanical processes. In both cases, it is very difficult to predict the lifetime. The chemical threats are controlled by slow diffusion of cations and, although these effects are reasonably well understood for the high temperatures and short times experienced in cell fabrication, extrapolation to very long times at the much lower operating temperatures and operational electric fields is uncertain. The methodologies for assessing mechanical threats, based on detailed FEM of specific designs, are well understood. They result in either a statistical probability of failure, based on the variability of strength in a family of nominally identical components, or on a deterministic analysis of failure based on fracture mechanics applied to a known defect. (A mixture of the two approaches is also possible.) However, these approaches do not allow a lifetime to be predicted unless time-dependent phenomena are included, such as subcritical crack growth, creep, or changes in material properties with time.

34

Solid Oxide Fuel Cell Lifetime and Reliability

REFERENCES [1] A.C. Mu¨ller, A. Weber and E. Ivers-Tiffe´e, Degradation of zirconia electrolytes. In: Proceedings of the 6th European solid oxide fuel cell forum, 2004, p. 12311238. [2] Gibson IR, Dransfield GP, Irvine JTS. Influence of yttria concentration upon electrical properties and susceptibility to ageing of yttria-stabilised zirconias. J Euro Ceram Soc 1998;18:6617. [3] Kawada T, Sakai N, Yokokawa H, Dokiya M, Anzai I. Reaction between solid oxide fuel cell materials. Solid State Ionics 1992;50:18996. [4] Chen M, Nicholas Grundy A, Hallstedt B, Gauckler LJ. Thermodynamic modeling of the La-Mn-Y-Zr-O system. Calphad 2006;30:489500. [5] Yang C-CT, Wei W-CJ, Roosen A. Reaction kinetics and mechanisms between La0.65Sr0.3MnO3 and 8 mol% yttria-stabilized zirconia. J Am Ceram Soc 2004;87:111016. [6] Brant MC, Matencio T, Dessemond L, Domingues RZ. Electrical degradation of porous and dense LSM/YSZ interface. Solid State Ionics 2006;177:91521. [7] Malzbender J, Batfalsky P, Vaßen R, Shemet V, Tietz F. Component interactions after long-term operation of an SOFC stack with LSM cathode. J Power Sources 2012;201:196203. [8] Tsoga A, Gupta A, Naoumidis A, Skarmoutsos D, Nikolopoulos P. Performance of a double-layer CGO/YSZ electrolyte for solid oxide fuel cells. Ionics 1998;4:23440. [9] Gao Z, Kennouche D, Barnett SA. Reduced-temperature firing of solid oxide fuel cells with zirconia/ceria bi-layer electrolytes. J Power Sources 2014;260:25963. [10] Kilo M, Borchardt G, Lesage B, Kaitasov O, Weber S, Scherrer S. Cation transport in yttria stabilized cubic zirconia: Zr-96 tracer diffusion in (ZrxY1x)O2x/2 single crystals with 0.15 , 5 x , 5 0.48. J Euro Ceram Soc 2000;20:206977. [11] Kowalski K, Bernasik A, Sadowski A. Bulk and grain boundary diffusion of titanium in yttria-stabilized zirconia. J Euro Ceram Soc 2000;20:9518. [12] Khan MZ, Mehran MT, Song R-H, Lee J-W, Lee S-B, Lim T-H, et al. Effect of GDC interlayer thickness on durability of solid oxide fuel cell cathode. Ceram Int 2016;42:697884. [13] Matsui T, Komoto M, Muroyama H, Eguchi K. Interfacial stability between air electrode and ceria-based electrolyte under cathodic polarization in solid oxide fuel cells. Fuel Cells 2014;14:10227. [14] Sun B, Rudkin RA, Atkinson A. Effect of thermal cycling on residual stress and curvature of anode-supported SOFCs. Fuel Cells 2009;9:80513. [15] Frandsen HL, Makowska M, Greco F, Chatzichristodoulou C, Ni DW, Curran DJ, et al. Accelerated creep in solid oxide fuel cell anode supports during reduction. J Power Sources 2016;323:7889. [16] Malzbender J, Fischer W, Steinbrech RW. Studies of residual stresses in planar solid oxide fuel cells. J Power Sources 2008;182:5948. [17] Clague R, Marquis AJ, Brandon NP. Finite element and analytical stress analysis of a solid oxide fuel cell. J Power Sources 2012;210:22432. [18] Sarantaridis D, Atkinson A. Redox cycling of Ni-based solid oxide fuel cell anodes: a review. Fuel Cells 2007;7:24658. [19] Faes A, Hessler-Wyser A, Zryd A, Van herle J. A review of redox cycling of solid oxide fuel cells anode. Membranes 2012;2:585.

Solid Oxide Fuel Cell Electrolytes—Factors Influencing Lifetime Chapter | 2

35

[20] Atkinson A. Chemically-induced stresses in gadolinium-doped ceria solid oxide fuel cell electrolytes. Solid State Ionics 1997;95:24958. [21] Malzbender J, Steinbrech RW. Threshold fracture stress of thin ceramic components. J Euro Ceram Soc 2008;28:24752. [22] Beuth JL. Cracking of thin bonded films in residual tension. Int J Solids Struct 1992;29:165775. [23] Atkinson A, Selc¸uk A. Mechanical behaviour of ceramic oxygen ion-conducting membranes. Solid State Ionics 2000;134:5966. [24] Choi SR, Bansal NP. Flexure strength, fracture toughness, and slow crack growth of YSZ/ alumina composites at high temperatures. J Am Ceram Soc 2005;88:147480. [25] Munz D, Fett T. Ceramics—mechanical properties, failure behaviour, materials selection. In: Zunger A, et al., editors. Materials Science. Berlin Heidelberg: Springer Verlag; 1999. [26] Greco F, Frandsen HL, Nakajo A, Madsen MF, Van herle J. Modelling the impact of creep on the probability of failure of a solid oxide fuel cell stack. J Euro Ceram Soc 2014;34:2695704. [27] Klemenso T, Chung C, Larsen PH, Mogensen M. The mechanism behind redox instability of anodes in high-temperature SOFCs. J Electrochem Soc 2005;152:A218692.

Chapter 3

The Impact of Fuels on Solid Oxide Fuel Cell Anode Lifetime: The Relationship Between Fuel Composition, Fuel Impurities, and Anode Lifetime and Reliability Kazunari Sasaki Kyushu University, Fukuoka, Japan

Chapter Outline Introduction Fuel Compositions Power Generation Characteristics for Various Fuels Fuel Impurities

37 38 39 41

Anode Lifetime Reliability Outlook References

43 46 49 49

INTRODUCTION Fuel cells have been increasingly accepted as environmentally compatible, efficient energy conversion systems. In particular, solid oxide fuel cells (SOFCs) may be regarded as the most flexible fuel cells with respect to their flexibility in selecting the types of fuels able to be supplied directly to the fuel electrodes [1 6]. Because of their multifuel capability, not only hydrogen and carbon monoxide but also various kinds of fuels may be used via internal reforming and/or via simple external reforming, including natural gas (consisting mainly of CH4 with a small amount of other hydrocarbons such as C2H6), liquefied petroleum gas (LPG, consisting mainly of C3H8 with C4H10), naphtha (consisting mainly of C5 and C6 hydrocarbons), gasoline (consisting mainly of hydrocarbons with carbon numbers around 8), kerosene (consisting mainly Solid Oxide Fuel Cell Lifetime and Reliability. DOI: http://dx.doi.org/10.1016/B978-0-08-101102-7.00003-9 Copyright © 2017 Elsevier Ltd. All rights reserved.

37

38

Solid Oxide Fuel Cell Lifetime and Reliability

of hydrocarbons with carbon numbers around 10 12), alcohols, biogas (consisting mainly of CH4, CO2, and H2O), coal gas (consisting mainly of CO and H2), and coke oven gas. The aim of this chapter is therefore to examine and compare power generation characteristics of SOFCs for various types of fuels and for various operational conditions. First, we consider the thermochemical stability of fuels to understand chemical equilibria, which is helpful to adjust initial fuel compositions. In particular, coke formation must be prevented so that a sufficient amount of water vapor has to be added for SOFCs to operate outside the carbon deposition region. The power generation characteristics of SOFCs operated with various fuels via internal reforming or via external reforming are then described. Understanding on SOFC durability issues is summarized, associated with the high-temperature operational nature of SOFCs: (1) extrinsic degradation caused by fuel impurities partly due to the fuel flexibility of SOFCs, (2) intrinsic degradation caused by high-temperature fabrication and operation up to a decade, and (3) cycle durability in practical operations associated with temperature and fuel changes.

FUEL COMPOSITIONS Thermochemical calculations can be made to derive the thermochemically expected fuel compositions. These calculations were performed by assuming a reactor to which a (mixed) fuel gas was supplied, and the amounts of gas, liquid, or solid products in thermodynamic equilibrium were numerically derived, in the temperature range between 100 C and 1000 C [4,5]. The minimum amounts of H2O essential to prevent carbon deposition are shown in Fig. 3.1A for hydrocarbons and Fig. 3.1B for alcohols. While S/C (steam:carbon ratio) of 1.5 is enough for CH4, higher S/C is needed with increasing carbon number of hydrocarbons. For alcohols, less H2O is needed at elevated temperatures. It has been found that the major

FIGURE 3.1 Minimum S/C ratio needed to prevent carbon deposition in thermochemical equilibrium (A) for hydrocarbons and (B) for alcohols [4].

The Impact of Fuels on Solid Oxide Fuel Cell Anode Lifetime Chapter | 3

39

FIGURE 3.2 The C H O ternary diagrams: (A) positions of SOFC-related fuels and (B) carbon deposition limit lines at various temperatures [5].

constituents in fuel gases are H2(g), H2O(g), CO(g), CO2(g), CH4(g), and C(s). Since their compositions depend solely on the C H O ratio, we can plot, on such C H O diagrams, parameters relevant to operational conditions, including carbon deposition region, gas partial pressures, and electromotive force. Positions of various SOFC-related species are plotted in Fig. 3.2A. As examples, carbon deposition region at temperatures between 100 and 1000 C is shown in Fig. 3.2B. Various other C H O diagrams are reported elsewhere [5].

POWER GENERATION CHARACTERISTICS FOR VARIOUS FUELS Model SOFC cells may be used to compare power generation characteristics for various SOFC fuels, with sintered zirconia-based electrolyte plates and Ni-based cermet anodes. Various kinds of fuels may be supplied to SOFCs. Analysis of power generation characteristics by varying operational parameters, including H2-to-CO ratio, type of the carrier gas such as He, N2, and Ar, temperature, fuel-to-carrier gas ratio, and water vapor concentration, is of technological interest. As selected typical examples, Fig. 3.3A shows the I V characteristics of a cell at 1000 C for different H2-to-CO ratios [7]. It has been experimentally confirmed that the use of CO-rich gases results in comparable performance to that of H2-rich gases and thus mixed gas such as coal gas may be useful as a SOFC fuel. I V characteristics depend on the carrier gas, indicating the importance of gas transport in porous anodes for anodic polarization [7]. I V characteristics for alcohols (including methanol, ethanol, and propanol) are shown in Fig. 3.3B. For all these alcohol-based fuels the C H O ratio was fixed to the same value for comparison. It has been demonstrated that direct-alcohol SOFCs can be realized, at least, for alcohols with carbon number up to 4. In the case that methanol was directly supplied, the I V characteristics

40

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 3.3 The I V characteristics of cells for (A) different H2-to-CO ratios [7], (B) with alcohol-based fuels [8,9], (C) CH4 or C2H6 as a fuel at the S/C ratio of 3.5 [6], and (D) simulated biogas at 1000 C [10].

were similar to those with the simulated reformed gas. However, with increasing carbon number of alcohols, a decrease in cell voltage was observed. From gas chromatography, it has been revealed that the compositions of simulated reformed gas and methanol-derived (internal reforming) fuel gas were almost the same, explaining the similar I V characteristics. However, with increasing carbon number of alcohols and/or with decreasing operational temperature, a decrease in H2 and CO gas concentrations and an increase in concentrations of by-products (partial oxidation products) in the exhaust gas were observed, associated with a decrease in I V characteristics [9]. Fig. 3.3C shows the I V characteristics of an SOFC operated with methane- and ethane-based fuel gases [6]. It has been found that carbon deposition was observed when ethane and ethylene were directly supplied as fuels, associated with a decrease in cell voltage with time. Therefore direct hydrocarbon fuel cells may suffer from carbon deposition with increasing carbon number of hydrocarbon-based fuel gases. Biogas can be produced from biomass via various procedures such as fermentation and thermal decomposition, the use of which may enable the realization of a carbon-neutral zero-emission energy system with SOFC

The Impact of Fuels on Solid Oxide Fuel Cell Anode Lifetime Chapter | 3

41

technology. Fig. 3.3D shows I V characteristics of an SOFC operated with simulated biogas to which H2O was added, exhibiting satisfactory electrochemical performance. It should be, however, noted that real biogas usually contains a nonnegligible amount of impurities such as H2S, leading to a decrease in cell voltage at elevated temperatures and irreversible degradation at lower operational temperatures. The removal of such impurities from biogas may therefore be needed.

FUEL IMPURITIES SOFCs can be regarded as the most flexible fuel cells, so that the use of various types of practical fuels has been considered [1,10]. As described in Fig. 3.4, practical fuels, however, contain minor constituents as impurities. Sulfur compounds are common fuel impurities in various fossil fuels and in biogas. Phosphorus and halogen compounds are contained in coal gas.

FIGURE 3.4 Possible impurity contaminations of SOFCs.

42

Solid Oxide Fuel Cell Lifetime and Reliability

Siloxane is a typical impurity in digester gas. Aromatic compounds, easily forming graphene and thus coke (deposited carbon) on the anodes, may be contained in various petroleum-related fuels. In addition, due to a high operational temperature of SOFCs, impurities with a high vapor pressure can also be evaporated from the system components, causing contamination in the SOFC systems. Inexpensive lower purity raw materials for cost reduction and conventional materials such as steels containing sulfur and phosphorus could even be an origin of fuel impurities. Such extrinsic impurities can cause chemical degradation of SOFCs, as reported, e.g., for sulfur [11 14] and other compounds [15]. In fact, as reported by Yokokawa et al. [16], it has been confirmed that various impurities do exist in SOFC systems. In this section, chemical degradation of SOFCs is considered based on thermochemical calculation, long-term poisoning tests, and microstructural observations. Durability of SOFCs upon poisoning by various impurity species has been characterized in many studies. H2S poisoning has been analyzed with respect to various operational conditions [13]. As an example, Fig. 3.5 shows the cell voltage, anodic polarization, and the ohmic loss on the anode side, measured for the 50%-prereformed methane-based fuel with S/C 5 2.5 containing 5 ppm H2S. An initial cell voltage drop followed by a quasisteady-state cell voltage can be observed associated mainly with an increase in anodic polarization and almost constant IR loss, up to 1000 h. The cell voltage drop is almost reversible for such a low H2S impurity concentration, even after 1000-h poisoning. Poisoning mechanisms of sulfur, a common fuel impurity, have been extensively analyzed for various fuels, including H2, H2 CO, CH4, partially reformed CH4, and simulated biogas. For relatively low concentrations (ppm level) of sulfur, reversible processes associated with adsorption/desorption of sulfur have been considered as the predominant mechanism. At a higher sulfur concentration and/or a lower operational temperature, an irreversible

FIGURE 3.5 Cell voltage, anodic polarization, and anode-side IR loss, measured at 200 mA cm during H2S (5 ppm) poisoning at 800 C for the 50%-prereformed CH4 fuel with S/C 5 2.5 [13].

2

The Impact of Fuels on Solid Oxide Fuel Cell Anode Lifetime Chapter | 3

43

FIGURE 3.6 Microstructural change of anodes by various impurities: (A) Cl2, (B) siloxane, (C) phosphorus, and (D) boron [15].

degradation was observed associated with the oxidation of Ni due to a large anodic polarization. Sulfur poisoning to internal reforming reactions is also serious, so that much larger cell voltage drop has been observed in case CH4-rich fuels were supplied. In addition, it has been suggested that the formation of Ni3S2 (melting point: 787 C) will be possible for hydrogen-poor fuels. Poisoning phenomena have been analyzed for other impurities, including chlorine, siloxane, phosphorus, and boron on the anode side. Fig. 3.6 shows field emission scanning electron microscope micrographs of anodes poisoned by these impurities of ppm levels [15]. Chlorine poisoning was associated with the sublimation of NiCl2 that could be precipitated on zirconia surfaces, while siloxane poisoning led to silica deposition. Phosphorus is reactive with Ni to form a eutectic compound. The presence of boron accelerated the grain growth of Ni in the cermet anodes.

ANODE LIFETIME Durability of SOFCs is one of the most important requirements for commercialization. SOFC anode degradation phenomena may be caused by both extrinsic and intrinsic origins besides by thermal/redox cycling.

44

Solid Oxide Fuel Cell Lifetime and Reliability

Even through fuel flexibility is one of the major advantages of SOFCs, it means that various kinds of practical fuels may be applied and various impurities can often flow into the cell/stacks. Based on various experimental studies as well as thermochemical studies, typical degradation mechanisms caused by external species may be compiled as shown in Fig. 3.7 [17]. Adsorption-related phenomena are reversible and thus recoverable, but the mechanism associated with sublimation leads to a loss of catalytic electrode materials and the mechanism coupled with deposition of additional phases can affect electrode properties. For example, serious carbon formation can lead to a loss of Ni acting as the catalysts to form carbon nanotubes and can, e.g., cause delamination of current collectors. Reactive species with Ni can cause serious change in microstructure due to catalytically inactive secondary phase formation.

FIGURE 3.7 Typical extrinsic chemical degradation mechanisms of SOFC anodes [17].

The Impact of Fuels on Solid Oxide Fuel Cell Anode Lifetime Chapter | 3

45

External impurities can be removed by using pretreatment systems (e.g., desulfurizer to remove sulfur-based odorant) [See Chapter 8 in this book] and/or using clean fuels and high-purity raw materials. However, intrinsic chemical degradation phenomena become obvious for longer operation, where interdiffusion from neighboring components and associated interfacial chemical phenomena such as the formation of insulating phases may dominate the long-term durability. Based on degradation rate data, we have estimated concentration thresholds of various impurities under a simple assumption that degradation rate may, in a first approximation, depend linearly on the impurity concentration. We have also set a target of 10% voltage degradation acceptable during 40,000 h (a New Energy and Industrial Technology Development Organization target, 0.25%/1000 h) to estimate such tolerant impurity concentrations. The estimated values are listed in Table 3.1, suggesting that major common impurities such as sulfur and chlorine may have a concentration threshold of ca. 1 ppm but reactive impurities such as phosphorus should be eliminated down to ppb levels. Additional studies have been made using larger cells under development by companies to check durability against typical impurities such as sulfur, indicating a similar tendency as we found using single button cells [18]. For microscopic observations, microsampling using the focused ion beam (FIB) technique is useful. [Discussed in more detail in Chapter 5 of this book.] This enables us to pick up samples of interest for high-resolution

TABLE 3.1 Impurity Concentration Threshold (Tolerant Concentration), Experimentally Estimated (Tentative, Assuming Linear Dependence) Impurities

Impurity Concentration Threshold and its Precondition (0.2 A cm22)

Sulfur (S) (H2S, COS, CH3SH)

700 ppb

50%-Prereformed CH4 fuel at 800 C, provided the reversible voltage loss is excluded

Chlorine (Cl) (Cl2, HCl)

400 ppb

3%-Humidified H2 fuel at 800 C

Phosphorus (P)

3 ppb

3%-Humidified H2 fuel at 800 C

2 ppb

50%-Prereformed CH4 fuel (S/C 5 2.5) at 800 C

Boron (B)

5 ppb

3%-Humidified H2 fuel at 800 C

Siloxane (Si) (D5)

2 ppm

3%-Humidified H2 fuel at 800 C for a 1 kW SOFC stack having an anode area of 5000 cm2

Sulfur (SO2)

4 ppm

Air (dry) for La1 xSrxMnO3

0.5 ppm

Air (dry) for La12xSrxCo12yFeyO3

46

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 3.8 Possible advanced visualization of SOFCs. Courtesy of NEXT-FC, Kyushu University.

microstructural and elemental characterizations by eliminating signals and noise from the background phases. Visualization of such changes from atomic and nano-meter levels to stack and system levels becomes increasingly important, as illustrated in Fig. 3.8, accelerating materials and system development for wider commercialization of SOFCs. Atomic-resolution scanning transmission electron microscopes enable detailed studies on microstructural and chemical changes in the cells, while environmental transmission electron microscope enables in situ or even in operando observation of various processes in SOFC operation. FIB-SEMs make three-dimensional analysis of complicated porous electrode microstructure. Low-energy ion scattering enables the chemical analysis of the outermost surfaces determining catalytic and surface exchange activities, while secondary ion mass spectrometry can obtain surface depth profile of specific elements. X-ray computed tomography enables nondestructive analysis of cells and stacks.

RELIABILITY Continuous SOFC stack operation with constant power output generally results in a gradual degradation in performance during long-term operation, while SOFCs may have to run for up to a decade. However, SOFCs can suffer to a greater degree from changes in operation conditions. Due to thermal expansion mismatch between the different components, the cells can suffer from mechanical degradation processes, such as delamination and crack formation with simple thermal cycling. Changes in atmosphere can result in more serious degradation. The influences of thermal cycling and

The Impact of Fuels on Solid Oxide Fuel Cell Anode Lifetime Chapter | 3

47

current density cycling on cell degradation have been investigated [19 21]. Redox cycles are typically associated with oxidation of Ni particles at the anode and the consequent change in volume [22 29]. Furthermore, redox cycling results in the formation of Ni hydroxides at a certain vapor pressure in an oxidizing atmosphere, with a high water vapor concentration [30,31]. In real residential SOFC power units, cells and stacks are not subjected to these different conditions independently; the changes occur much more dynamically. Therefore cycle durability studies should be performed using realistic operation protocols that simulate typical operational conditions. In order to improve reliability of SOFCs, the influence of thermal cycling on SOFC cell performance may be characterized using three different protocols; hot standby, cold standby, and shutdown. Practical fuels such as partially reformed methane gas can be used, while humidified H2 fuel can also be used for such cycle durability tests. During short time periods where power is not required (e.g., at night), residential SOFC systems are cooled down to a lower temperature, and a small amount of electricity is continuously generated to maintain minimum required system operation. The hot standby protocol simulates such conditions. No interruption in fuel supply occurs during the hot standby protocol. The cold-standby protocol simulates longer term shutdown conditions, e.g., when the user is absent for extended periods of time. In this situation the system is stopped in a controlled manner, and cooled down naturally. The shutdown protocol simulates emergency shutdown, e.g., during a blackout, or in an earthquake. In this situation the fuel supply shuts off, and the temperature changes very quickly. These can be investigated by varying the number of thermal cycles, cooling rate, and fuel composition to systematically clarify the SOFC degradation mechanisms caused by realistic start-stop cycle operations. Details of each cycle protocol are shown in Fig. 3.9 [32]. The cell performance degraded during repeated cold standby and shutdown modes, whilst negligible degradation was observed in hot standby. Cycle-induced degradation was significantly influenced by both the cool down rate (i.e., the period of time during cooling), and the presence of H2S in the fuel feedstock. As summarized in Fig. 3.10, possible degradation mechanisms include: (1) Ni agglomeration, (2) Ni oxidation (accelerated by the presence of H2S), (3) Ostwald ripening, (4) Ni precipitation on anode zirconia grains, and (5) Ni precipitation at electrolyte grain boundaries. Other mechanisms such as mechanical degradation induced by thermal expansion mismatch may also occur. The main origin of cell performance degradation is the interruption of fuel supply, which promotes Ni oxidation at the anode. In the case of sulfur poisoning, the water vapor concentration increased on the anode side, and Ni grains were thus easily oxidized (H2S deactivates both electrode and internal reforming reactions). Ni oxidation may also be accelerated by insufficient sealing after thermal cycling. In addition, a decrease in the amount of Ni

48

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 3.9 (A) Hot standby, (B) cold standby, and (C) shutdown cycle protocols [32].

FIGURE 3.10 Schematic drawing of the degradation mechanisms occurring during cycle durability tests, including (i) Ni agglomeration, (ii) Ni oxidation accelerated by H2S, (iii) Ostwald ripening, (iv) Ni precipitation on anode zirconia grains, and (v) Ni precipitation at electrolyte grain boundaries [32].

around the triple-phase boundary, and consequent Ni grain coarsening around the anode surface were identified, explained by the Ostwald ripening mechanism (via Ni(OH)2 in the gas phase). Precipitation of Ni on the surface of zirconia particles, and at grain boundaries in the electrolyte was also identified. All these phenomena lead to a decrease in electrode activity and consequent cell performance degradation during cycling.

The Impact of Fuels on Solid Oxide Fuel Cell Anode Lifetime Chapter | 3

49

OUTLOOK While commercialization of SOFC systems has been started for industrial and residential applications, there still exist technological issues related to durability and reliability of SOFCs [33]. Durability issues are decisive in commercialization of fuel cells systems, and degradation processes depend strongly on the location and operational conditions of the systems. Fundamental understanding on degradation mechanisms and the advances in analytical techniques may be helpful to have a solid basis for this promising technology.

REFERENCES [1] Steele BCH, Heinzel A. Materials for fuel-cell technologies. Nature 2001;414:345 52. [2] Craciun R, Park S, Gorte RJ, Vohs JM, Wang C, Worrell WL. A novel method for preparing anode cermets for solid oxide fuel cells. J Electrochem Soc 1999;144:4019 22. [3] Larminie J, Dicks A. Fuel cell systems explained. Chichester: John Wiley & Sons; 2000. [4] Sasaki K, Teraoka Y. Equilibria in fuel cell gases: I. Equilibrium compositions and reforming conditions. J Electrochem Soc 2003;150(7):A878 84. [5] Sasaki K, Teraoka Y. Equilibria in fuel cell gases: II. The C H O ternary diagrams. J Electrochem Soc 2003;150(7):A885 8. [6] Sasaki K, Kojo H, Hori Y, Kikuchi R, Eguchi K. Direct-alcohol/hydrocarbon SOFCs: comparison of power generation characteristics for various fuels. Electrochemistry 2002;70(1):18 22. [7] Sasaki K, Hori Y, Kikuchi R, Eguchi K, Ueno A, Takeuchi H, et al. Current voltage characteristics and impedance analysis of solid oxide fuel cells for mixed H2 and CO gases. J Electrochem Soc 2002;149(3):A227 33. [8] Sasaki, K., Watanabe, K., Shiosaki, K., Susuki, K., and Teraoka, Y. (2003c) Power generation characteristics of SOFCs for alcohols and hydrocarbon-based fuels. Proceedings Of the 8th international symposium solid oxide fuel cells. Electrochem. Soc. Proc. Vol. 2003-07. Pennington, NJ: Electrochem. Soc., p. 1295. [9] Sasaki K, Watanabe K, Teraoka Y. Direct-alcohol SOFCs: current voltage characteristics and fuel gas compositions. J Electrochem Soc 2004;151(7):A965 70. [10] Sasaki K, Watanabe K, Shiosaki K, Susuki K, Teraoka Y. Multi-fuel capability of solid oxide fuel cells. J Electroceram 2004;13:669 75. [11] Matsuzaki Y, Yasuda I. The poisoning effect of sulfur-containing impurity gas on a SOFC anode: Part I. Dependence on temperature, time, and impurity concentration. Solid State Ionics 2000;132:261 9. [12] Sasaki K, Susuki K, Iyoshi A, Uchimura M, Imamura N, Kusaba H, et al. H2S poisoning of solid oxide fuel cells. J Electrochem Soc 2006;153:A2023 9. [13] Sasaki K, Adachi S, Haga K, Uchikawa M, Yamamoto J, Iyoshi A, et al. Fuel impurity tolerance of solid oxide fuel cells. ECS Trans 2007;7(1):1675 83. [14] Mukerjee S, Haltiner K, Kerr R, Chick L, Sprenkle V, Meinhardt K, et al. Solid oxide fuel cell development: latest results. ECS Trans 2007;7(1):59 65. [15] Haga K, Adachi S, Shiratori Y, Itoh K, Sasaki K. Poisoning of SOFC anodes by various fuel impurities. Solid State Ionics 2008;179(27-32):1427 31. [16] Yokokawa H, Watanabe T, Ueno A, Hoshino K. Investigation on degradation in long-term operations of four different stack/modules. ECS Trans 2007;7(1):133 40.

50

Solid Oxide Fuel Cell Lifetime and Reliability

[17] Sasaki K, Haga K, Yoshizumi T, Minematsu D, Yuki E, Liu RR, et al. Chemical durability of solid oxide fuel cells: influence of impurities on long-term performance. J Power Sources 2011;196(22):9130 40. [18] Sasaki K, Yoshizumi T, Haga K, Yoshitomi H, Hosoi T, Shiratori Y, et al. Chemical degradation of SOFCs: external impurity poisoning and internal diffusion-related phenomena. ECS Trans 2013;57(1):315 23. [19] Bujalski W, Paragreen J, Reade G, Pyke S, Kendall K. Cycling studies of SOFCs. J Power Sources 2006;157:745 9. [20] Bujalski W, Dikwal CM, Kendall K. Cycling of three solid oxide fuel cell types. J Power Sources 2007;171:96 100. [21] Guan Y, Gong Y, Li W, Gelb J, Zhang L, Liu G, et al. Quantitative analysis of micro structural and conductivity evolution of Ni-YSZ anodes during thermal cycling based on nano-computed tomography. J Power Sources 2011;196:10601 5. [22] Werber T. Joining of nickel powder grains by thermal oxidation. Solid State Ionics 1990;42:205 11. [23] Waldbilling D, Wood A, Ivey DG. Electrochemical and microstructural characterization of the redox tolerance of solid oxide fuel cell anodes. J Power Sources 2005;145:206 15. [24] Liu B, Zhang Y, Tu B, Dong Y, Cheng M. Electrochemical impedance investigation of the redox behaviour of a Ni YSZ anode. J Power Sources 2007;165:114 19. [25] Fujita K, Somekawa T, Horiuchi K, Matsuzaki Y. Evaluation of the redox stability of segmented-in-series solid oxide fuel cell stacks. J Power Sources 2009;193:130 5. [26] Laurencin J, Delette G, Sicardy O, Rosini S, Lefebvre-Joud F. Impact of ‘redox’ cycles on performances of solid oxide fuel cells: case of the electrolyte supported cells. J Power Sources 2010;195:2747 53. [27] Jeangros Q, Faes A, Wagner JB, Hansen TW, Aschauer U, van Herle J, et al. In situ redox cycle of a nickel YSZ fuel cell anode in an environmental transmission electron microscope. Acta Mater 2010;58:4578 89. [28] Sumi H, Kishida R, Kim J, Muroyama H, Matsui T, Eguchi K. Correlation between microstructural and electrochemical characteristics during redox cycles for Ni YSZ anode of SOFCs. J Electrochem Soc 2010;157(12):B1747 52. [29] Pihlatie MH, Kaiser A, Mogensen M, Chen M. Electrical conductivity of Ni YSZ composites: degradation due to Ni particle growth. Solid State Ionics 2011;189:82 90. [30] Holzer L, Iwanschitz B, Hocker T, Munch B, Preatat M, Wiedenmann D, et al. Microstructure degradation of cermet anodes for solid oxide fuel cells: quantification of nickel grain growth in dry and in humid atmospheres. J Power Sources 2011;196:1279 94. [31] Holzer L, Iwanschitz B, Hocker T, Keller L, Pecgo O, Sartoris G, et al. Redox cycling of Ni YSZ anodes for solid oxide fuel cells: influence of tortuosity, constriction and percolation factors on the effective transport properties. J Power Sources 2013;242:179 94. [32] Hanasaki M, Uryu C, Daio T, Kawabata T, Tachikawa Y, Lyth SM, et al. SOFC durability against standby and shutdown cycling. J Electrochem Soc 2014;161(9):F850 60. [33] Sasaki K, Li H-W, Hayashi A, Yamabe J, Ogura T, Lyth SM, editors. Hydrogen energy engineering: a Japanese perspective. Japan: Springer; 2016.

Chapter 4

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Tony Wood1 and Douglas G. Ivey2 1

Fuel Cell Energy, Calgary, AB, Canada, 2University of Alberta, Edmonton, AB, Canada

Chapter Outline Introduction Anode-Supported Solid Oxide Fuel Cell Manufacture and Microstructure Kinetics of Redox Cycling Mechanical Considerations Impact on Electrochemical Performance

51 53 54 56

Microstructural Changes Solutions to Redox Cycle Degradation Summary Acknowledgements References

60 70 74 74 74

57

INTRODUCTION Solid oxide fuel cell (SOFC) anodes are often fabricated as mixed ceramic oxides with a reduction step prior to operation to form a cermet material. Most commonly a good oxygen ion conducting ceramic (such as yttria-stabilized zirconia (YSZ) or samaria-doped ceria) is blended with a metal oxide chosen to give a desired metallic component after the reduction process [1]. The most widely used metallic component is nickel for three main reasons: 1. Nickel is a good electronic conductor, imparting electrical conductivity to the electrode. 2. Nickel has high catalytic activity for internal reforming of methane and for the water gas shift reaction to convert carbon monoxide in the fuel to hydrogen. 3. At triple-phase boundary (TPB) sites where the metal, oxygen ion conducting ceramic and gas phases meet, nickel has high electrochemical activity for the reaction of oxide ions with the fuel gas. However, nickel has also some disadvantages for SOFC use. It is prone to carbon deposition at high carbon activities, so the fuel composition must Solid Oxide Fuel Cell Lifetime and Reliability. DOI: http://dx.doi.org/10.1016/B978-0-08-101102-7.00004-0 Copyright © 2017 Elsevier Ltd. All rights reserved.

51

52

Solid Oxide Fuel Cell Lifetime and Reliability

be managed at all times to prevent this. In the event of fuel loss, nickel can be easily reoxidized with an associated dimensional increase of 69% by volume from bulk nickel back to the as-fabricated nickel oxide [2]. Since the anode structures also contain ceramic material and porosity that are not affected by loss of fuel, the actual volumetric change is far smaller than this value but still significant. This reduction and oxidation process is known as a redox cycle. Since nickel oxide must be reduced back to nickel metal again prior to operation and every time there is a loss of fuel the redox process will be repeated, there is the potential for several redox cycles to occur during an SOFC’s operational lifetime. These cycles have been found to severely degrade SOFC lifetime, such that redox cycle degradation has become a major research topic. SOFCs can be categorized based on many different criteria, such as the geometry (planar or tubular), the system size (,1 kWe, 110 kWe, and 100 kWe to 1 MWe), or application (automotive, stationary base load, and stationary distributed power). One of the favored ways that scientists and engineers classify SOFCs is by the component of the cell that provides the most mechanical support. This leads to the four most widely reported categories: 1. Electrolyte-supported cells with a relatively thick electrolyte (50200 μm) and a thinner anode (550 μm). Typically YSZ is used as the electrolyte and operates at high temperatures, from 850 to 1000 C. 2. Cathode-supported cells with a relatively thin electrolyte (120 μm) and a thinner anode (550 μm). Again, YSZ is used as the electrolyte and operates at 8501000 C. 3. Anode-supported cells (ASCs) have a relatively thin electrolyte at (120 μm) and a thinner anode functional layer (AFL) (550 μm), with a thicker anode substrate (2001500 μm). The thin electrolyte allows for intermediate operating temperatures, in the 600850 C range. 4. Substrate-supported cells with a 120 μm thick electrolyte and a 550 μm thick anode. The operating temperature range depends on the electrolyte material, but falls within 5001000 C. Cathode-supported cells have been extensively developed by SiemensWestinghouse for large scale power generation and integration with gas turbines. At this scale, system solutions exist to prevent reoxidation of nickel during startup and shutdown that are not prohibitively expensive. Hence, there is very little literature on redox degradation for this cell architecture. The other three designs have all been considered for small-scale systems, where larger scale system level solutions for startup and shutdown would be prohibitively expensive or would not fit within volume constraints. As such, redox degradation for these architectures has been more widely studied. Sarantaridis and Atkinson [3] showed numerically (with certain assumptions) that mechanical failures occur by different mechanisms and result in different allowable anode expansions (due to oxidation of nickel) before failure

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Chapter | 4

53

for these three types of cells. Electrolyte-supported cells with 10 μm thick anodes can likely tolerate 0.5% anode expansion before failing. Substratesupported cells can probably tolerate 1.0% anode expansion for the same anode thickness before failing by the same mechanism. However, ASCs can likely only tolerate ,0.1% anode expansion before tensile cracking of the electrolyte occurs. It is for this reason that ASCs have been the most studied for redox degradation and this chapter will, therefore, focus on redox degradation of anode-supported SOFCs. The literature for redox cycling of SOFCs was scarce until around the turn of the millennium. Since then, hundreds of research papers, including five review papers, tens of PhD theses, and dozens of patents have been published on the subject. This does not include publications dealing with alternate ceramic anodes, which are discussed in Chapter 10, New Materials for Improved Durability and Robustness in Solid Oxide Fuel Cell. The sheer volume of literature precludes full coverage of the research and the reader is encouraged to seek out the references cited for a more detailed review of the concepts presented. In this chapter, microstructural effects of redox cycling will be the primary topic of interest, with cell manufacturing, kinetics of reduction and oxidation of nickel, mechanical effects and electrochemical effects summarized to provide background information and context. Finally, potential solutions to redox cycle degradation are presented along with future research directions.

ANODE-SUPPORTED SOLID OXIDE FUEL CELL MANUFACTURE AND MICROSTRUCTURE ASCs are typically fabricated using anode substrates made from nickel oxide and YSZ in the range of 4570% (w/w) nickel oxide by methods such as Tape-casting, Screen-printing, and Co-firing (e.g., the TSC process at Versa Power Systems) [4]. This corresponds to a range of 3766% (w/w) or 2957% (v/v) nickel metal after complete reduction to a functional nickelYSZ cermet structure, ignoring porosity. In practice, porosity has a significant influence on many critical properties of the anode substrate such as electrical conductivity, gas diffusion parameters and, most relevant to this chapter, mechanical properties. Theoretically a fully dense anode made from known volume fractions of nickel oxide and YSZ will have a density, predicted by the rule of mixtures, equal to the sum of the products of density and volume fraction for each of the two phases. By measuring the actual density of anodes, the porosity can be calculated since it is responsible for the difference between the theoretical and measured density. Fig. 4.1 shows the porosity of an as prepared ASC as a function of nickel oxide content of the anode substrate at a measured density of 5.4 g cm23 after sintering, as well as the calculated nickel, YSZ, and porosity volume fractions after reduction of the nickel oxide to nickel metal assuming no bulk volumetric

54

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 4.1 Chart showing porosity, nickel and YSZ volume % as a function of initial nickel oxide content for an ASC at a fixed measured density of 5.4 g cm23.

change of the cell (i.e., the practical observation). This chart can be reproduced for any measured cell density or rearranged to show these calculations as a function of cell density for a given cell nickel oxide content. As the main structural support of ASCs, the ratios of these values are a key part of understanding mechanical failures and stresses on the cell during redox cycling and likely account for many of the discrepancies reported in the literature regarding the effects of redox cycling on ASCs, as noted in a review article by Faes et al. [5].

KINETICS OF REDOX CYCLING In order to understand the impact of redox cycling, it is first necessary to review the chemistry of reduction and oxidation of nickel, including reaction rates and the impact of microstructure and substrate on reaction rates. One of the first things to note is the energy change. Reduction of nickel oxide is shown in Eq. (4.1). While hydrogen oxidation is exothermic, NiO reduction is endothermic and the net result is a very slightly exothermic reduction reaction (ΔH 5 13 kJ mol1 at 800 C) [6]. Reducing a cell is unlikely to cause damage due to this relatively small exotherm. NiO 1 H2 -Ni 1 H2 O

ð4:1Þ

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Chapter | 4

55

On the other hand, oxidation of nickel (Eq. (4.2)) is highly exothermic (ΔH 5 239.8 kJ mol1) [6]. Ni 1 12 O2 -NiO

ð4:2Þ

For large cells oxidized rapidly (e.g., at high temperature), this could be a mechanism of degradation. It is clear that the kinetics of oxidation cannot only provide insight into how to modify reoxidation to make it less damaging by understanding the mechanism and limiting factors, but also provide guidance on conditions to avoid that would lead to rapid and catastrophic oxidation. With this in mind, Waldbillig et al. [7] performed an in-depth study of oxidation and reduction rates of Versa Power Systems anode-supported TSC2 cells using thermogravimetric analysis between 400 C and 850 C. The study showed that reduction of the NiO/YSZ anode samples followed linear kinetics with an activation energy of 78 kJ mol21. Several other studies have concluded that NiO reduction is first order and the rate has a linear dependence with the partial pressure of hydrogen [810]. There is little effect of the YSZ substrate observed for reduction kinetics for the most part (at least of relevance to redox cycle degradation). Oxidation of the reduced Ni/YSZ anode followed parabolic kinetics at temperatures lower than 700 C. A divergence from parabolic kinetics was seen at higher temperatures (700850 C). This divergence has been seen in many other kinetic studies and is usually attributed to short-circuit diffusion mechanisms. An activation energy of 87 kJ mol21 was calculated for oxidation. Oxidation rates are clearly more complicated. Taking a simplified look at metallic nickel oxidation (to eliminate substrate and Ni/YSZ microstructural effects) a parabolic trend is seen at temperatures greater than 1000 C with activation energies typical of the outward bulk diffusion of nickel in NiO. However, at temperatures less than 1000 C, many studies have observed that the kinetics tend to diverge from parabolic behavior and the activation energy for reoxidation decreases [1118]. This trend, like the Ni/YSZ study at 700850 C, indicates that a short circuiting mechanism, such as transport along grain boundaries or dislocations, begins to dominate the kinetics at temperatures less than 1000 C. This observation is supported by the fact that nickel oxidation kinetics seem to be very sensitive to the preparation procedures. Different impurity levels and preparation procedures will lead to a different concentration of short-circuit paths and may explain the large spread in literature values of oxidation activation energy. Divergence from parabolic rates may be caused by inward diffusion of oxygen along grain boundaries, the gaseous transport of oxygen through transient microfissures or grain growth within the film during oxidation, which reduces the number of grain boundaries and thus the amount of grain boundary diffusion. Studying these effects in more detail could lead to processing methods and additives that slow the fundamental oxidation kinetics. While this would not allow any amount of air to flow into the fuel side of a system for any amount of time, it could open

56

Solid Oxide Fuel Cell Lifetime and Reliability

up low-cost system redox solutions. For example, a slower oxidation rate could be the difference between allowing a safe, controlled shutdown in a reasonable time frame when fuel is lost, simply by cooling with an air purge, and requiring expensive, complicated, bulky purge gas solutions [19].

MECHANICAL CONSIDERATIONS The key SOFC failure mechanism attributed to redox events, first highlighted by Cassidy et al. [20] and subsequently by many researchers [2126], is mechanical failure of the electrolyte due to stresses in the anode that form upon reoxidation. This is shown clearly in Fig. 4.2 from Waldbillig et al. [2] with a characteristic transverse electrolyte crack after redox cycling an ASC. The maximal strain of a planar ASC before cracking the thin electrolyte in this manner can be calculated using Eq. (4.3) [3,5]. GASC 5 πσ2 h=½E=ð1 2 v2 Þ 5 πh½εox E=ð12vÞ2 =½E=ð1 2 v2 Þ 5 πhεox E=ð1 1 vÞ=ð1 2 vÞ . GC

ð4:3Þ

where E is the Young’s modulus of the electrolyte, v is Poisson’s ratio, h is the electrolyte thickness, εox is the oxidation strain, and GASC and Gc are the stored and critical energy release rate, respectively. It is important to note that this relationship predicts that decreasing electrolyte thickness increases redox stability (higher redox strain limit). Using typical values, the authors [5] estimated that an electrolyte thickness of about 2 μm gives a redox strain limit of 0.2% before electrolyte fracture. It should be noted, however, that microstructure plays a critical role, as demonstrated by Waldbillig et al. [7]. The anode of most ASCs consists of two nickel-cermet layers that have differing functions and, hence, different microstructures (Fig. 4.3). The different microstructures are typically a fine particle size nickel and ceramic phase for the AFL (typically a thin

FIGURE 4.2 SEM BSE image of a fresh fractured ASC after redox cycling. Note the large vertical crack in the electrolyte [2]. Reproduced by permission of Elsevier.

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Chapter | 4

57

screen-printed layer, 550 μm thick, adjacent to the electrolyte) and a coarse particle size nickel with finer particle size ceramic phase for the anode substrate. These microstructures have been tested for dimensional change upon reduction and oxidation using a thermomechanical analyser. Fig. 4.4 shows a comparison of the dimensional change for the two microstructures during heating to 750 C and reoxidation in air. The coarse, porous microstructure representative of the anode substrate shows an expansion of about 0.2% on heating, but no measurable expansion due to reoxidation. In contrast the finer, denser AFL microstructure shows an expansion about 10 3 higher (2.3%) and the samples after oxidation show the effect clearly, with the anode substrate sample intact and the AFL sample fractured (inset Fig. 4.4).

IMPACT ON ELECTROCHEMICAL PERFORMANCE When kinetic studies of nickel oxidation are considered, as well as the large amount of nickel present in a large ASC, it becomes clear that it could take a reasonable amount of time and air flow to fully reoxidize an ASC in an actual SOFC device. This time frame and amount of air flow may be sufficient that, in fact, cells do not get fully reoxidized during use, but only partially reoxidized (a partial redox cycle). With this assumption, Waldbillig et al. [2] proposed a reoxidation or redox depth approach in order to determine the length of partial redox cycle times for in situ testing of the effects of redox cycling on electrochemical performance of planar ASCs. This analysis uses the amount of nickel within the cell and the flow rate of air to predict the amount of time, it would take to oxidize the cell to a specific redox depth (50% redox depth means that 50% of the nickel in the cell has been

FIGURE 4.3 SEM BSE image of anode-supported SOFC half-cell showing the anode substrate layer (coarse microstructure) and AFL (fine microstructure).

58

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 4.4 Thermomechanical analysis comparison of anode substrate and AFL microstructures during heating to 750 C and reoxidation in air. Inset shows samples after test (graduations in mm) [7]. Reproduced by permission of Elsevier.

oxidized and 50% remains as metallic nickel; 100% redox depth represents a complete redox cycle). The approach assumes that all the oxygen in the air feed to the test is used up to oxidize nickel (i.e., fast kinetics). This was confirmed to 60% oxidation depth by measuring the length of time taken before oxygen was detected in the outlet gas using gas chromatography. Redox cycles to an oxidation depth of 10%, 20%, 30%, 60%, 120% (excess air flow) and 180% (excess air flow) were performed corresponding to 20, 40, 60, 120, 240, and 360 min at 120 mL min21 air flow into the test jig. The reason for the excess air is that as a full redox cycle (100% redox depth) is approached, not all the oxygen is used to oxidize nickel with the test arrangement utilized, so 120% is close to or represents a full redox cycle and 180% is a full redox cycle with almost double the oxygen required to oxidize all nickel in the cell supplied during the test. Baseline electrochemical testing was performed comparing initial currentvoltage curves and steady-state degradation testing with tests after redox cycles to varying degrees of oxidation or redox depth. Fig. 4.5 shows the effect of redox cycles to different oxidation depths on the VJ curves for an ASC with the percentage degradation tabulated (inset). Little degradation occurs for redox cycles with less than 30% oxidation depth. The largest amount of degradation occurs after the first complete (100% oxidation depth) cycle. A slight drop in open circuit voltage also begins to occur after the first complete redox cycle, most likely indicating electrolyte cracking occurs somewhere above 60% oxidation

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Chapter | 4

59

FIGURE 4.5 Currentvoltage curves to 0.74 A cm22 for tests with various redox times at 750 C; inset is data summary table [2]. Reproduced by permission of Elsevier.

TABLE 4.1 Influence of Temperature and Time of Reoxidation on the Electrochemical Performance and the Linear Expansion of Tubular ASCs [27] T ( C)

52 Redox Cycles of 5 min, Δi/i at 0.5 V

After First Full Oxidation Cycle, Δi/i at 0.5 V

Expansion During First Full Oxidation

Time to Full Oxidation (h)

600

20.38%/cycle

235%

0.20%

4.5

700

20.42%/cycle

261%

0.33%

3.0

800

20.44%/cycle

272%

0.46%

0.5

of the anode. Electron microscopy analysis showed that the microstructure is irreversibly changed by the first redox cycle (see the Microstructural Changes section). Researchers at the University of Birmingham studied the behavior of tubular ASCs with 200 μm anode thickness, 15 μm electrolyte thickness, and an external diameter of 2 mm [27]. Electrochemical performance degradation and linear expansion were studied as a function of temperature (at 600, 700, and 800 C) and oxidation time of redox cycles (5 min and full redox cycle). The degradation increases with increasing temperature after a full redox cycle. The cell no longer worked despite the relatively small expansion (see Table 4.1). Other studies also showed high degradation of cell performance

60

Solid Oxide Fuel Cell Lifetime and Reliability

under redox cycles [28]. In these studies the electrolyte is deposited exterior to the anode support. A small expansion of the support upon reoxidation will then create large tensile stresses in this layer. If the electrolyte was deposited inside the support, as for the Siemens-Westinghouse tubular design, then it could be under compressive stress when the anode expands. This demonstrates the effect of cell design on redox tolerance. Clearly many factors can affect electrochemical performance degradation and a full description is beyond the scope of this volume. For a more comprehensive review the reader is referred to an excellent summary by Faes et al. [5].

MICROSTRUCTURAL CHANGES It is important to understand the microstructural changes that occur to the anode and the anode/electrolyte during redox cycling and how these affect cell integrity and performance. This section will provide an overview of the work done in the last 10 1 years, including more recent methods involving 3D imaging and in situ studies. On a related note, Chapter 5, Microstructural Degradation: Mechanisms, Quantification, Modeling and Design Strategies to Enhance the Durability of Solid Oxide Fuel Cell Electrodes examines the role of microstructure on SOFC lifetime. Much of the initial work on microstructural analysis was done by Waldbillig et al. [2,7,29,30] just over 10 years ago. They utilized electron microscopy techniques, both scanning and transmission electron microscopy (SEM and TEM), to study ASCs with the anode consisting of a Ni/YSZ cermet. Experiments were done ex situ; some of the more recent work by other authors has been done in situ, as discussed later in this section. Two types of samples were examined by Waldbillig et al. [2], i.e., actual cells that were fractured after various redox treatments and imaged as cross sections in the SEM and anode substrates from which TEM samples were prepared and subjected to redox cycles. For the latter a combination of mechanical polishing and ion milling methods were used to generate electron transparent samples. These samples were reduced (5% H2/95% N2) and oxidized (air) at 700 C for varying times. Virtually all microstructural changes involved nickel, with little apparent effect on YSZ. During the first reduction cycle the nickel that formed grew epitaxially on NiO and the volume change (decrease) was accommodated by intragranular nanopores (B50 nm in diameter) (Fig. 4.6A). The satellite spots in the electron diffraction pattern are due to double diffraction effects associated with the presence of both nickel and NiO and the orientation relationship between the two. Epitaxial growth occurred despite the large mismatch (15.6%) in lattice parameters between nickel and NiO. Reoxidation of the reduced structure resulted in the formation of polycrystalline NiO, with grains (B50 nm in size) replacing larger ( . 1 μm) nickel single crystals (Fig. 4.6B). The grains were randomly oriented and much of the nanoporosity disappeared. The grain refinement

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Chapter | 4

61

FIGURE 4.6 TEM images and diffraction patterns (insets) of anode substrates: (A) partially reduced NiO/YSZ; (B) reoxidation of the sample in (A); and (C) after the second reduction [2]. Reproduced by permission of Elsevier.

during reoxidation was likely due to the nanopores in the individual nickel grains, which acted as nucleation sites for NiO. After the second reduction process, the resultant nickel was polycrystalline with no preferred orientation effects (Fig. 4.6C).

62

Solid Oxide Fuel Cell Lifetime and Reliability

The TEM samples were compared with bulk anode samples that underwent similar redox cycles. The reduced bulk samples had increased porosity, as expected, but the porosity was not intragranular. Reoxidized samples had a spongy appearance with smaller porosity than the as-fabricated anode structures. The difference in morphology between the TEM samples and the bulk anode samples, with cycling, was attributed to the thin nature (,200 nm thick) of the TEM samples. Actual fuel cells were also tested for comparison; similar microstructural changes relative to the bulk anode substrate samples were observed, i.e., reoxidized samples had the same spongy NiO morphology. Cracking of the thin electrolyte was detected in some instances as well as localized damaged regions in the AFL layer. These regions remained oxidized even after reduction. A few years after Waldbillig’s work, Faes et al. [31] used an environmental TEM (ETEM) to study redox of nickel in a Ni/YSZ anode. TEM samples were prepared from anode/electrolyte half cells using a dual-beam focused ion beam (FIB)SEM instrument. The anode was initially NiO/YSZ. Reduction was done in 1.4 mbar H2 in the ETEM by heating between 300 C and 500 C, while subsequent oxidation was done in 3.2 mbar O2 by heating between 250 C and 500 C. Reduction began at B320 C at the NiO/YSZ interface and proceeded toward the center of NiO grains, through an autocatalytic process, generating nanoporosity and epitaxial growth of Ni similar to that reported earlier by Waldbillig et al. [2] in their ex situ experiments. At higher temperatures (.410 C), reduction also took place at NiO free surfaces. Coalescence of nanopores occurred at temperatures approaching 500 C. Some isolated NiO regions were retained at temperatures as high as 480 C. Subsequent oxidation started at B350 C at the Ni surface, closing off the porosity as more nickel transformed to NiO. The NiO was polycrystalline (nanocrystals) and much of the porosity from reduction was trapped in the reoxidized material. This led to NiO extending out of the plane of the TEM foil creating a rough surface. The same group studied the reduction of NiO particles (no YSZ) by heating in an ETEM [32]. Samples were prepared by placing a few drops of a NiO/isopropanol suspension onto a Cu grid covered with a thin, electron transparent SiO2 coating and evaporating the isopropanol. Particles used for imaging were less than a few hundred nanometers in size. The reducing environment was 1.3 mbar of H2 at a flow rate of 2 mL min1. Electron diffraction and electron energy loss spectroscopy (EELS) were utilized to characterize the structural and chemical changes in NiO during reduction and to explore the reaction kinetics. The authors were able to determine an activation energy for NiO reduction of B70 kJ mol21, which is within the range (10150 kJ mol21) reported in the literature [8] and close to the value (78 kJ mol21) reported earlier by Waldbillig et al. [7]. They proposed a model for NiO reduction based on their observations and previous studies [33]. Hydrogen (H2) is adsorbed on the surface of NiO and dissociates on

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Chapter | 4

63

two Ni atoms, which neighbor an oxygen vacancy. Ni clusters form into larger domains at 350400 C; H2 adsorbs directly on these clusters and H2O desorbs. The formation of nickel domains is dependent on the number of oxygen vacancies at the surface, which depends on both temperature and time. Both epitaxial and nonepitaxial growth of nickel on NiO were observed, with epitaxial behavior more prevalent at slower heating rates. The nickel domains grow through the displacement of interfaces and pores form within larger particles. To accommodate the decrease in volume during reduction, smaller particles shrink leading to an irregular structure. Water molecules can accumulate around the nickel particles blocking H2 access and preventing further reaction. At higher temperatures ( . 600 C), larger nickel grains grow at the expense of smaller ones to minimize the surface energy. Note that there are differences in the reduction process for individual nickel particles and nickel in a fuel cell anode. For the anode, the process begins at the NiO/YSZ interface and involves the movement of oxygen from NiO to YSZ, which produces oxygen vacancies in NiO. Jeangros et al. [32] also discussed some of the artifacts that can arise in the ETEM. Electron beam damage can be induced by the high energy beam, generating Ni3O4 (spinel) at room temperature in low-vacuum conditions and causing faceting, erosion and knock-on damage. Any hydrocarbons in the microscope can be cracked and deposited on the specimen as a carbon layer (Fig. 4.7); this carbon layer can influence the reduction process. Localized heating by the electron beam can also accelerate the reactions in the ETEM. In other more recent work by Jeangros et al. [34], the authors did not look at redox cycling but only at initial activation of the anode through reduction of NiO to nickel in a 55 wt% NiO45 wt% YSZ anode. In particular, they examined the effect of initial NiO/Ni phase boundary symmetry on the resultant FIGURE 4.7 HRTEM image (in 1.3 mbar of H2) from a Ni nanoparticle oriented along [110] on the surface of a NiO grain at 500 C. Some blurring is present due to thermal drift. An amorphous layer (likely carbon) is present on the surface of the larger NiO grain [32]. Reproduced by permission of Springer.

64

Solid Oxide Fuel Cell Lifetime and Reliability

connectivity of Ni grains using conventional imaging and 3D tomography in the TEM. TEM samples, prepared using FIB methods, were activated in an ETEM operated at 300 kV by heating from 300 to 604 C in 1.3 mbar H2. 3D images were obtained for samples after reaction in the ETEM, by tilting from 65 to 165 in increments of 1 and collecting 2D scanning transmission electron microscopy (STEM) images at each step. Suitable software was utilized to combine the images. Other techniques, such as electron diffraction, energy dispersive X-ray analysis, and EELS, were used to characterize the samples. Automated crystal orientation maps were obtained to assess the effects of grain orientation and grain boundary symmetry. Two types of Ni/NiO interfaces were observed during reduction in the ETEM, coherent and incoherent. Nickel and NiO tended to remain in contact for coherent interfaces, but detached for incoherent interfaces. The type of reduction influenced the connectivity and, as a consequence, the electronic conductivity of the anode. Twinning of nickel grains during reduction was also observed. Impurities, such as aluminum and silicon present in the raw materials, tended to segregate to incoherent Ni/NiO interfaces. There they formed a thin glassy film within the resultant voids (Fig. 4.8). Nickel nanoparticles formed in these regions, which the authors attributed to nickel hydroxide evaporation and subsequent condensation. The segregated impurities may have blocked active sites in the anode, which is an indication of the importance of raw material purity for fabrication of anodes. Several studies have utilized the SEM to study redox behavior. One of the earlier and more detailed studies was done by Sarantaridis et al. [35]. They used a dual-beam FIB/SEM to examine microstructural changes in Ni/YSZ anodes during redox cycling and to relate them to mechanical property changes. FIB methods were used to section and polish samples for imaging; only 2D images were obtained, i.e., no 3D reconstruction was done. Anodes were reduced at 900 C (5% H2) and underwent interrupted oxidation at 800 C in air. They confirmed the irreversible expansion on oxidation (B0.1%) reported by many others, as well as a reversible change in the

FIGURE 4.8 (A and B) STEM dark field images of the Ni/YSZ microstructure after reduction up to 604 C. X-ray maps are shown in (B) using the Ni Kα, O Kα, Si Kα, and Al Kα peaks [34]. Reproduced by permission of Elsevier.

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Chapter | 4

65

elastic modulus of the composite during redox cycling. The amount of expansion depended on the oxidation process. No macroscopic cracking of the YSZ skeleton was observed, but there was evidence of some warping for interrupted oxidation experiments. Closed porosity was found after oxidation and was estimated to be about 5% of the total volume of the cermet. The mechanism for porosity formation was attributed to a combination of nickel sintering and intrinsic oxidation properties. They attributed the expansion during oxidation to the closed porosity. Sarantaridis et al. [35] also studied the oxidation of individual nickel particles (515 μm in size), which was done at times ranging from 30 to 300 s at 800 C. Sectioning of the nickel particles, using FIB polishing, showed that partially oxidized nickel had more porosity relative to fully oxidized nickel (Fig. 4.9). They explained this behavior as being due to nickel diffusion dominating in the early stages and oxygen diffusion becoming important in latter stages. Toros [36] has used finite element, micro-level modeling to examine reoxidation behavior in the Ni/YSZ AFL at temperatures from 800 to 950 C. He first verified his model through comparison with work done previously by Sarantaridis et al. [35] on oxidation of nickel particles. He modeled both the microstructure and the stress distribution in the anode. Reoxidation was not complete at 800 C after 2 h; 95.6% of Ni was transformed. During reoxidation the stresses generated were large. The model showed good agreement with experimental results; the rate of reoxidation decreased with time due to NiO coverage on nickel. Stress increased with exposure time, due to the volume expansion accompanying oxidation. Also, there were higher stresses at higher reoxidation temperatures. The estimated stresses were higher than the fracture strength of the anode. Klemensø et al. [37] studied redox cycling of a Ni/YSZ cermet in situ using an environmental SEM (ESEM). Samples were reduced in H2 (9 vol%)/N2 at 1000 C for 1 h or 96 h and then reoxidized in air for 1 h. Reduction of NiO led to increased porosity, shrinking, and rounding of the nickel particles. The rounding process was faster for smaller particles. Reoxidation led to redistribution of NiO and its morphology depended on the oxidation temperature. For rapid oxidation, NiO particles split into two to four particles; these grew into adjacent voids. For slower oxidation, NiO grew as a surface layer around the Ni. Another method for examining microstructural changes to the anode that has gained traction in recent years is 3D tomography through serial sectioning by combined FIB milling and SEM. This is most commonly done in a dualbeam FIB/SEM instrument. Thin layers of material (of the order of 10 nm) are ion milled from the surface with a Ga ion beam. Typically the anode porosity is filled with either epoxy (vacuumed impregnated) or photoresist prior to FIB milling. SEM images are taken after each milling step and these are combined through suitable software to generate a 3D image of the sample.

66

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 4.9 FIB cross-section images (SE) of Ni particles oxidized at 800 C for (A) 30 s, (B) 60 s, (C) 90 s, (D) 180 s, and (E) 300 s. (F) The same particle shown in (E) imaged using secondary ions [35]. Reproduced by permission of The Electrochemical Society.

Holzer at al. [38,39] have conducted detailed studies of redox cycling for Ni/YSZ anodes using the serial sectioning method; their results were coupled with simulations. Most of the previous 3D imaging of SOFC anodes has been used to quantify electrochemical reaction sites and to quantify active and inactive TPBs. Holzer et al. studied the effects of redox cycling on transport properties, which are related to the M factor,

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Chapter | 4

67

where M essentially means microstructural effects, through the following equation: σeff 5 σ0 M

ð4:4Þ

σeff and σ0 are the effective and intrinsic transport properties. The M factor depends on four separate microstructure parameters, i.e., phase volume fraction (φ), percolation factor (P), constriction factor (β), which is the ratio of cross sectional areas at bottlenecks and bulges for a cylindrical system, and inverse tortuosity (1/τ) that describes variations in lengths of transport pathways. The M factor can be expressed as a product of these four parameters with appropriate weighting factors. In the more recent paper [39] the following equation is given for M: M 5 ðφPÞa β b =τ c

ð4:5Þ

The exponents a, b, and c are empirically determined constants. Holzer et al. considered anodes with both fine and coarse microstructures. Finer anodes had higher initial electronic and ionic conductivities relative to coarse anodes. This correlated with a higher M factor, for both nickel and YSZ, for the fine anodes compared with the coarse ones. The lower nickel M factor for the coarse anodes was a consequence of its higher tortuosity and lower effective volume fraction. For YSZ the lower M factor for the coarse anodes was attributed to their higher tortuosity and lower constrictivity (smaller bottlenecks). Redox cycling was done at 950 C over a total of eight complete cycles. Anode degradation was dependent on the initial anode microstructure, but was mainly due to nickel coarsening and loss of connectivity in the YSZ (Fig. 4.10). Nickel coarsening led to loss of percolation and increased tortuosity and up to a 50% drop in electronic conductivity. The effect was not as pronounced in the coarse anode because of its larger pores. For YSZ the loss in connectivity was more pronounced for the coarse anodes relative to the fine anodes. The constrictivity (β) decreased and the tortuosity increased leading to reduced percolation. Holzer et al. suggested that a combined fine/coarse microstructure, as well as the addition of relatively coarse pore formers, could improve redox tolerance. The larger pores could better accommodate the volume changes during redox. Jiao and Shikazono [4042] used phase field modeling, coupled with 3D reconstructions with a dual-beam FIB/SEM, to study NiO reduction in the anode. Reduction was done at 500, 800, and 1000 C. There were a large number of uniformly distributed inner pores when NiO was reduced at 500 C, due to volume contraction. For anodes reduced at 800 and 1000 C the nickel particles were fairly dense with much less porosity. Volume reduction at higher temperatures occurred by sintering, which led to densification. Anodes reduced at 800 C had the best initial performance, while anodes reduced at 1000 C had the most stable performance (over a period of 100 h). Samples reduced at 500 C had the highest degradation rate for both Ohmic

68

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 4.10 3D view of (a) YSZ, (b) Ni, and (c) Ni 1 YSZ in the coarse sample before (A) and after (B) redox cycling. Degradation during redox involves disintegration of YSZ particles into smaller particles (a). Ni agglomeration results in a less granular structure and Ni partially overgrows the small, disconnected YSZ islands (c) [39].

and polarization resistances. This is due to the poor adhesion at the Ni/YSZ interface and enhanced sintering of nickel as a result of the large number of submicron pores that formed during reduction at 500 C. Jiao and Shikazono also noted that the size of the reconstruction is important; regions ,30 μm in size may not be large enough to be representative. Brus et al. [43] took a different approach and studied the change in the anode morphology during fuel starvation for a Ni/YSZ ASC using 3D reconstruction via a dual-beam SEM/FIB after a redox cycle. The oxidation process can be classified into two types: one is through chemical reaction (2Ni 1 O2 5 2NiO), which is the one most commonly studied, and the other is through electrochemical reaction (Ni 1 O22 5 NiO 1 2e2).

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Chapter | 4

69

Chemical oxidation is easier to study, but is not completely representative of what occurs in the anode during operation. Electrochemical oxidation tends to occur in the vicinity of the electrolyte/anode interface. Fuel starvation can occur when the fuel is consumed faster than it can be supplied to the anode. When this occurs, the cell may be strongly polarized and oxygen ions can continue to arrive from the cathode side leading to electrochemical oxidation. The effect of this phenomenon on anode microstructure is not well understood. Brus et al. [43] used button cells consisting of Ni/YSZ (anode), YSZ (electrolyte), and GDC/LSCF (cathode). Fuel starvation was simulated by stopping the fuel supply under the polarized condition. Nickel near the electrolyte was oxidized even though O2 was not supplied to the anode side, which indicates that oxygen must have come from the cathode side. The result was electrochemical oxidation of nickel. Oxidation led to an increase in the tortuosity factor for nickel (approximately double) and a decrease in the tortuosity factor for the pores, as well as a reduction in nickel particle and pore size and connectivity from 97 to 87%. Other techniques, such as X-ray diffraction (XRD), have been used for in situ study of the redox process. Hagen et al. [44] used high energy synchrotron radiation (71 keV X-rays) to examine both full cells and half cells (NiO/YSZ). The reduction and oxidation processes were done in the 836880 C temperature range and changes in NiO (200) and nickel (111) reflection intensities were monitored. They found that nickel particles further away from the surface were reduced or reoxidized more slowly than those close to the surface. Richardson et al. [8] studied the reduction of NiO particles in H2, with and without steam, at 175300 C by in situ hot stage XRD. Like Hagen et al., changes in the NiO (200) and nickel (111) peaks were monitored. Their results showed an induction period, followed by pseudo-first-order kinetics up to a fraction transformed of B0.8; after this point inaccessibility of NiO slowed down the rate. Water vapor adsorbed on the surface was believed to be responsible, since the addition of H2O to the reducing gas increased the induction period and slowed the reduction rate. There was an increase in crystallite size, from B3 nm for NiO to B20 nm for reduced nickel. Apparent activation energies were determined as 126 6 27 and 85 6 6 kJ mol21 for reductions without and with added water, respectively. This is within the wide range (10150 kJ mol21) reported in the literature [8]. For comparison, Jeangros et al. [32] determined an activation energy of B70 kJ mol21 for NiO particle reduction in the TEM. This is close to the latter value for Richardson et al. [8], but reduction was done in low-pressure H2 only. Other work has been done using X-rays, but these studies have focused on long-term effects on microstructure, not redox cycling. Chen-Wiegart et al. [45] used synchrotron X-ray nano-tomography, which has sub 100 nm

70

Solid Oxide Fuel Cell Lifetime and Reliability

resolution, to study coarsening of nickel in NiYSZ anodes. They alternatively heated and imaged their samples in humidified H2 fuel. They looked at two different Ni/YSZ ratios (65:35 wt% and 50:50 wt%) and found that the nickel coarsening rate increased with increasing Ni/YSZ ratio. They attributed the coarsening effect to an increase in pore volume and a less “obtrusive YSZ network,” which facilitate coarsening. The data were fit to a power-law coarsening model, from which a mass transport coefficient was determined (KD); KD was 27 times higher for the 65:35 sample relative to the 50:50 sample. Kennouche et al. [46] used transmission X-ray microscopy, with a synchrotron light source, to study high temperature microstructural changes in NiYSZ anodes (50:50 NiO:YSZ). The samples were aged for 48 h at 1050 C ex situ; the high temperature was used to accelerate the process. One advantage of this method over the 3D sectioning methods already discussed is that the same region can be imaged before and after aging and imaging can be done in real time. Anode-supported half cells were first reduced in 5% H2/3% H2O/Ar at 800 C. Samples were prepared by fracturing, mechanical polishing, and FIB polishing to produce a cylinder with B40 μm diameter and B80 μm height. The samples were then aged in 5% H2/3% H2O/Ar to simulate anode fuel. The authors observed nickel grain coarsening, but Ni loss by volatization was not significant. The tortuosity for nickel and YSZ did not change; the values were B1.6 and B1.1, respectively. The pore tortuosity increased from 1.45 to 1.88. There was an increase in average nickel and pore sizes. Active TPBs decreased during the first 24 h from 5.83 to 2.29 μm22—to 2.07 μm22 after 49 h. Nickel coalescence involved elimination of a YSZ region. YSZ acted, in general, to limit coarsening.

SOLUTIONS TO REDOX CYCLE DEGRADATION A review of redox solutions has been made by Wood et al. for small-scale residential and industrial power generation (310 kWe) based on ASCs [19,29]. The solutions are divided into two main categories, system solutions and materials (including cell and stack design) solutions. In this volume, we are mainly concerned with the latter and, in particular, passive solutions based on cell material and microstructure modifications that could provide a low-cost redox degradation solution. The amount of redox-induced degradation is extremely dependent on temperature as demonstrated by Wood et al. [19,29] and shown in Fig. 4.11. Degradation is 2 3 lower at 700 C and 5 3 lower at 600 C compared with baseline results at 750 C. A series of microstructural approaches was also proposed by the same authors and by Waldbillig et al. [30] and is summarized in Table 4.2 along with the temperature effects in order of increasing effectiveness of the redox solution (highest redox degradation is found at the top of the table and lowest rate at the bottom). One solution

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Chapter | 4

71

FIGURE 4.11 Temperature effects on redox cycle degradation [19].

involves an oxidation barrier such as a fine microstructured nickel layer printed onto the bottom of the anode substrate that would oxidize and densify somewhat, limiting air ingress to more important cell components. The other solutions are all variations of modifying the anode microstructure, including a multi-layered AFL with decreasing nickel content in layers toward the electrolyte, increasing AFL porosity and modifying the microstructure through a controlled, low temperature conditioning redox cycle (at 550 C). Klemensø [47] and Klemensø and Sørensen [48] confirmed that lowering the temperature is effective and added that decreasing the anode support thickness will increase the redox stability. For an anode support thickness of 300 μm, AFL and electrolyte thicknesses of 10 μm, the maximal strains before electrolyte cracking using Eq. (4.3) are 0.2% for the anode support and 0.7% for the AFL at 650 C, and 0.2% for the anode support and 0.25% for the AFL at 800 C. More recently, Wang et al. [49] showed that nickeliron alloys display increasing redox tolerance as iron content is increased and temperature is decreased for metal supported SOFCs. It is possible that this approach could be applied to ASC technology as well. A popular approach for planar [5052] and tubular ASC [53] architectures has been to use a ceramic or inert support and infiltrate nickel into the microstructure such that redox expansion is more easily tolerated without damaging the ceramic structural support. Unfortunately, this approach requires many steps of infiltration and heat treatment (often more than 10) in order to fabricate a cell with reasonable performance. Small amounts of chemical additives to limit nickel coarsening [54,55], alloys of nickel with higher oxidation resistance to slow

TABLE 4.2 Effect of Oxidation Temperature and Anode Microstructural Modifications on Redox Cycle Degradation—Ranked in Order of Increasing Effectiveness (Baseline 750 C is Least Effective and 600 C Redox Temperature is Most Effective) [30] Test Type

Degradation (%) VJ Curve Tests

Degradation (%) Steady-State Hold Tests

Baseline (750 C)

10.8

8.5





Oxidation barrier

8.8

7.9

19

7

45% graded AFL

7.8

7.6

28

11



% Enhancement VJ Curve tests

% Enhancement Steady-State Hold Tests

700 C redox

5.4

3.2

50

62

AFL pore formers

4.2

3.5

61

59

550 C preoxidation

3.3

2.2

69

74



3.2

1.9

70

78



2.2

0.9

80

89

650 C redox 600 C redox

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Chapter | 4

73

down the oxidation kinetics [56], and additives to make a protective layer on the nickel [56] have all been proposed in the patent literature as potential solutions. Although the focus has been on materials solutions, an interesting solution proposed by Wood et al. [19,29] is the use of a steam purge in the event of system shutdown, whereby the latent heat in the system would be used to evaporate water from a storage reservoir in the system. By storing liquid water (or using a local supply) the storage volume and cost compared with compressed gas storage is dramatically reduced and, importantly, a significant effect on redox cycle degradation has been demonstrated using steam purge instead of air purge (Fig. 4.12). For a 6 h air purge, degradation is around 11% compared with no measured degradation over the same time frame for two tests using a steam purge. This is a relatively simple approach that does not add significant cost or complexity and allows the cell to be optimized for electrochemical performance and long lifetimes. However, long-term exposure of ASCs to high steam concentrations in solid oxide electrolysis cell (SOEC) testing has shown degradation due to the nickelcermet electrode coarsening and even loss of nickel as volatile nickel hydroxide at higher temperatures (950 C and above) [5759]. These phenomena can be studied using the same approaches applied to redox cycling in air and are, therefore, highly relevant to the work described in this chapter. More study of the kinetic and microstructural effects of exposure to high steam content is required for implementation of this approach (and for development of SOEC technology).

FIGURE 4.12 Comparison of the effects of steam and air purges on redox cycle degradation [19].

74

Solid Oxide Fuel Cell Lifetime and Reliability

SUMMARY Excellent progress has been made in the understanding of redox cycle degradation phenomena over the past decade or so, especially regarding microstructural evolution. Electron microscopy techniques have been applied for both ex situ and in situ studies of redox behavior. Results vary with experimental conditions and the types of samples examined; however, there are some common features. During the first reduction cycle both epitaxial and nonepitaxial growth of nickel on nickel oxide can occur, along with the formation of nanopores to accommodate the large decrease in volume. Epitaxial growth is more prevalent at slower heating rates and reduction initiates at NiO/YSZ interfaces. Nickel coarsening occurs during fuel cell operation. Reoxidized samples exhibit irreversible expansion (B0.1%), as well as closed porosity. Nickel diffusion tends to dominate in the early stages of oxidation, while oxygen diffusion becomes important later. The reoxidation rate decreases with time due to NiO coverage on nickel and reoxidation may not go to completion. The research in this field has led to dozens of patents [e.g., 4751] and proposed solutions that can be found in the literature [e.g., 19,29,30,5559]. Hundreds of research papers, five review articles, and tens of PhD theses have been devoted to the subject and, while not exhaustive, this chapter has attempted to summarize a significant portion of these. One key related area of focus for future research has been identified, i.e., the effects of steam redox cycling (and prolonged exposure of the anode to steam). Many of the same techniques applied to study redox cycling in air, as described in this chapter, are directly applicable to studying these degradation mechanisms. As interest in solid oxide electrolysis grows, this research will become increasingly important.

ACKNOWLEDGEMENTS The authors would like to thank Ms. Aliesha Johnson (University of Alberta) for her assistance in the literature search.

REFERENCES [1] Minh NQ, Takahashi T. Science and technology of ceramic fuel cells. Amsterdam. Elsevier Science; 1995. [2] Waldbillig D, Wood A, Ivey DG. Electrochemical and microstructural characterization of the redox tolerance of solid oxide fuel cell anodes. J Power Sources 2005;145:20615. [3] Sarantaridis D, Atkinson A. Redox cycling of Ni-based solid oxide fuel cell anodes: a review. Fuel Cells 2007;3:324658. [4] Tang E, Martell F, Brule R, Borglum B, Ghosh D. Global thermoelectric’s integrated cell manufacturing of planar SOFCs. In: Huijsmans J, editor. Fifth European solid oxide fuel cell forum. Lucerne, Switzerland; 2002. p. 26.

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Chapter | 4

75

[5] Faes A, Hessler-Wyser A, Zryd A, Van Herle J. A review of redox cycling of solid oxide fuel cell anode. Membranes 2012;2:585664. [6] Weast RC. CRC handbook of chemistry and physics. 53rd ed Cleveland: CRC; 1973. [7] Waldbillig D, Wood A, Ivey DG. Thermal analysis of the cyclic reduction and oxidation behaviour of SOFC anodes. Solid State Ionics 2005;176(910):84759. [8] Richardson JT, Scates R, Twigg M. X-ray diffraction study of nickel oxide reduction by hydrogen. Appl Catal A, Gen 2003;246:137. [9] Rao YK, Rashed AH. Kinetics of reduction of nickel oxide with heliumhydrogen gas mixtures in the range 300400 C. Trans Inst Mining Metall: Sect C 2001;110:16. [10] Tikekar N, Armstrong T, Virkar A. Reduction and re-oxidation kinetics of nickel-based solid oxide fuel cell anodes. In: Singhal SC, Dokiya M, editors. The Electrochemical Society Proceedings Series, Proc. solid oxide fuel cells; 2003. Vol. VIII. p. 670. [11] Stathis G, Simwonis D, Tietz F, Moropoulou A, Naoumides A. Oxidation and resulting mechanical properties of Ni/8Y2O3-stabilized zirconia anode substrate for solid-oxide fuel cells. J Mater Res 2002;17:9518. [12] Peraldi R, Monceau D, Pieraggi B. Correlations between growth kinetics and microstructure for scales formed by high-temperature oxidation of pure nickel. II. Growth kinetics. Oxid Met 2002;58(3-4):27595. [13] Suwanwatana W, Yarlagadda S, Gillespie JW. An investigation of oxidation effects on hysteresis heating of nickel particles. J Mater Sci 2003;38:56573. [14] Karmhag R, Niklasson G, Nygren M. Oxidation kinetics of nickel nanoparticles. J Appl Phys 2001;89:301217. [15] Karmhag R, Niklasson G, Nygren M. Oxidation kinetics of large nickel particles. J Mater Res 1999;14:30518. [16] Karmhag R, Niklasson G, Nygren M. Oxidation kinetics of small nickel particles. J Appl Phys 1999;85:118691. [17] Haugsrud R. On the high-temperature oxidation of nickel. Corros Sci 2003;45:21135. [18] Czerwinski F, Smeltzer WW. The early-stage oxidation kinetics of CeO2 sol-coated nickel. J Electrochem Soc 1993;140:260615. [19] Wood A, Pastula M, Waldbillig D, Ivey DG. Initial testing of solutions to redox problems with anode-supported SOFC. J Electrochem Soc 2006;153(10):A192934. [20] Cassidy M, Lindsay G, Kendall K. The reduction of nickel-zirconia cermet anodes and the effects on supported thin electrolytes. J Power Sources 1996;61:18992. [21] Ettler M, Timmermann H, Malzbender J, Weber A, Menzler NH. Durability of Ni anodes during re-oxidation cycles. J Power Sources 2010;195:545267. [22] Malzbender J, Steinbrech RW. Advanced measurement techniques to characterize thermomechanical aspects of solid oxide fuel cells. J Power Sources 2007;173:607. 2007. [23] Sarantaridis D, Rudkin RA, Atkinson A. Oxidation failure modes of anode-supported solid oxide fuel cells. J Power Sources 2008;180:70410. [24] Laurencin J, Delette G, Morel B, Lefebvre-Joud F, Dupeux M. Solid oxide fuel cells damage mechanisms due to Ni-YSZ re-oxidation: case of the anode-supported cell. J. Power Sources 2009;192:34452. [25] Klemensoe T, Mogensen M. Ni-YSZ solid oxide fuel cell anode behavior upon redox cycling based on electrical characterization. J Am Ceram Soc 2007;90:35828. [26] Liu B, Zhang Y, Tu B, Dong Y, Cheng M. Electrochemical impedance investigation of the redox behaviour of a Ni-YSZ anode. J Power Sources 2007;165:11419. [27] Dikwal CM. Cycling studies of micro-tubular solid oxide fuel cells. PhD Thesis. Birmingham, UK: University of Birmingham; 2009.

76

Solid Oxide Fuel Cell Lifetime and Reliability

[28] Pusz J, Smirnova A, Mohammadi A, Sammes NM. Fracture strength of micro-tubular solid oxide fuel cell anode in redox cycling experiments. J Power Sources 2007;163:9006. [29] Wood A, Pastula M, Waldbillig D, Ivey D. Initial testing of solutions to redox problems with anode-supported solid oxide fuel cells (SOFC). In: Singhal SC, Mizusaki J, editors. Proceedings of SOFC-IX, Quebec, Canada. Electrochemical Society Proceedings; 2005, p. 571583. [30] Waldbillig D, Wood A, Ivey D. Enhancing the redox tolerance of anode-supported SOFC by microstructural modification. J. Electrochem Soc 2007;154(2):B1338. [31] Faes A, Jeangros Q, Wagner JB, Hansen TW, Van Herle J, Brisse A, et al. In situ reduction and oxidation of nickel from solid oxide fuel cells in a transmission electron microscope. ECS Trans 2009;25(2):198592. [32] Jeangros Q, Hansen TW, Wagner JB, Damsgaard CD, Dunin-Borkowski RE, He´bert C, et al. Reduction of nickel oxide particles by hydrogen studied in an environmental TEM. J Mater Sci 2013;48:2893907. [33] Rodriguez JA, Hanson JC, Frenkel AI, Kim JY, Perez M. Experimental and theoretical studies on the reaction of H2 with NiO: role of O vacancies and mechanism for oxide reduction. J Am Chem Soc 2002;124(2):34654. [34] Jeangros Q, Aebersold AB, He´bert C, Van Herle J, Hessler-Wyser AA. TEM study of Ni interfaces formed during activation of SOFC anodes in H2: Influence of grain boundary symmetry and segregation of impurities. Acta Mater 2016;103:4427. [35] Sarantaridis D, Chater RJ, Atkinson A. Changes in physical and mechanical properties of SOFC NiYSZ composites caused by redox cycling. J Electrochem Soc 2008;155(5): B46772. [36] Toros S. Microstructural finite element modeling of redox behavior of NiYSZ based ceramic SOFC anodes. Ceram Int 2016;42:891524. [37] Klemensø T, Appel CC, Mogensen M. In situ observations of microstructural changes in SOFC anodes during redox cycling. Electrochem Solid-State Lett 2006;9(9):A4037. [38] Holzer L, Iwanschitz B, Hocker Th, Keller L, Pecho O, Sartoris G, et al. Redox cycling of Ni/YSZ anodes for solid oxide fuel cells: influence of tortuosity, constriction and percolation factors on the effective transport properties. J Power Sources 2013;242:17994. [39] Pecho O, Stenzel O, Iwanschitz B, Gasser P, Neumann M, Schmidt S, et al. 3D Microstructure effects in NiYSZ anodes: prediction of effective transport properties and optimization of redox stability. Materials 2015;8:555485. [40] Jiao Z, Shikazono N. 3D reconstruction size effect on the quantification of solid oxide fuel cell nickelyttria-stabilized-zirconia anode microstructural information using scanning electron microscopy-focused ion beam technique. Sci Bull 2016;17. [41] Jiao Z, Shikazono N. Quantitative study on the correlation between solid oxide fuel cell Ni-YSZ composite anode performance and reduction temperature based on threedimensional reconstruction. J Electrochem Soc 2015;162(6):F5718. [42] Jiao Z, Shikazono N. Simulation of the reduction process of solid oxide fuel cell composite anode based on phase field method. J Power Sources 2016;305:1016. [43] Brus G, Miyoshi K, Iwai H, Saito M, Yoshida H. Change of an anode’s microstructure morphology during the fuel starvation of an anode-supported solid oxide fuel cell. Int J Hydrogen Energy 2015;40:692734. [44] Hagen A, Poulsen HF, Klemensø T, Martins RV, Honkima¨ki V, Buslaps T, et al. A depth-resolved in situ study of the reduction and oxidation of Ni-based anodes in solid oxide fuel cells. Fuel Cells 2006;6(5):3616.

The Impact of Redox Cycling on Solid Oxide Fuel Cell Lifetime Chapter | 4

77

[45] Chen-Wiegart YK, Kennouche D, Cronin JS, Barnett SA, Wang J. Effect of Ni content on the morphological evolution of NiYSZ solid oxide fuel cell electrodes. Appl Phys Lett 2016;108:083903. [46] Kennouche D, Chen-Wiegart YK, Yakal-Kremski KJ, Wang J, Gibbs JW, Voorhees PW, et al. Observing the microstructural evolution of Ni-Yttria-stabilized zirconia solid oxide fuel cell anodes. Acta Mater 2016;103:20410. [47] Klemensø T. Relationships between structure and performance of SOFC anodes. PhD Thesis. Technical University of Danemark, Risoe National Laboratory, Topsoe Fuel Cell, Risø, Denmark 2005. [48] Klemensø T, Sørensen BF. Evaluating redox stability of Ni-YSZ supported SOFCs based on simple layer models. In: Singh P, Bansal NP, editors. Advances in solid oxide fuel cells IV. Westerville, OH: The American Ceramic Society; 2009. p. 8192. [49] Wang X, Li K, Jia L, Zhang Q, Jiang SP, Chi B, et al. Porous NiFe alloys as anode support for intermediate temperature solid oxide fuel cells: I. Fabrication, redox and thermal behaviors. J Power Sources 2015;277:4749. [50] Jasinski P, Suzuki T, Petrovsky V, Anderson HU. Nanocomposite nickel ceria cermet with low nickel content for anode-supported SOFCs. Electrochem Sol State Lett 2005;8: A21921. [51] Zhu X, Lue Z, Wei B, Zhang Y, Huang X, Su W. Fabrication and evaluation of a Ni/ La0.75Sr0.25Cr0.5Fe0.5O3 2 delta co-impregnated yttria-stabilized zirconia anode for singlechamber solid oxide fuel cells. Int J Hydrogen Energy 2010;35:6897904. [52] Buyukaksoy A, Petrovsky V, Dogan F. Redox stable solid oxide fuel cells with NiYSZ cermet anodes prepared by polymeric precursor infiltration. J Electrochem Soc 2012;159: B2324. [53] Panthi D, Choi B, Tsutsumi A. Performance improvement and redox cycling of a microtubular solid oxide fuel cell with a porous zirconia support. Int J Hydrogen Energy 2015;40:1058895. [54] Robert G, Kaiser AF-J, Batawi E. Structured body for an anode used in fuel cells. U.S. Patent 20030165726 A1. 4 September 2003. [55] Larsen P.H, Chung C, Mogensen M. Redox-stable anode. WO Patent 2006079558A1. 3 August 2006. [56] Mukerjee S, Grieve M.J, Keegan K.R. Methods for preventing anode oxidation in a fuel cell. Eur. Patent 1263071. 15 April 2002. [57] Khan MS, Lee S-B, Song R-H, Lee J-W, Lima T-H, Park S-J. Fundamental mechanisms involved in the degradation of nickelyttria stabilized zirconia (NiYSZ) anode during solid oxide fuel cells operation: a review. Ceram Int 2016;42:3548. [58] Jiao Z, Takagi N, Shikazono N, Kasagi N. Study on local morphological changes of nickel in solid oxide fuel cell anode using porous Ni pellet electrode. J Power Sources 2011;196:101929. [59] Matsui T, Kishida R, Muroyama H, Eguchi K. Comparative study on performance stability of Ni-oxide cermet anodes under humidified atmospheres in solid oxide fuel cells. J Electrochem Soc 2012;159:F45660.

Chapter 5

Microstructural Degradation: Mechanisms, Quantification, Modeling and Design Strategies to Enhance the Durability of Solid Oxide Fuel Cell Electrodes Farid Tariq, Enrique Ruiz-Trejo, Antonio Bertei, Paul Boldrin and Nigel P. Brandon Imperial College London, London, United Kingdom

Chapter Outline Introduction Microstructural Degradation Mechanisms Impedance for Identifying Changes in Microstructure Using Electrode Imaging and Quantification to Measure Degradation Introduction to Approaches

79 80 81 84 84

3D Imaging Applied to Measuring Solid Oxide Fuel Cell Electrode Degradation Modeling of Microstructural Degradation Microstructural Design Strategies Conclusions References

86 91 94 95 95

INTRODUCTION Electrode microstructure is one of the main factors determining the performance and durability of solid oxide fuel cells (SOFCs). In the electrodes, hydrogen or oxygen is electrochemically converted and therefore these porous structures need to contain as many reaction sites as possible to promote the electrochemical reactions. However, the electrode efficacy depends not only on the catalytic activity of the materials, but also on the rate at which charged species (electrons and ions) and chemical species (e.g., H2, O2, and H2O) reach or leave these reaction sites, called three-phase or triple-phase boundaries (TPBs). Solid Oxide Fuel Cell Lifetime and Reliability. DOI: http://dx.doi.org/10.1016/B978-0-08-101102-7.00005-2 Copyright © 2017 Elsevier Ltd. All rights reserved.

79

80

Solid Oxide Fuel Cell Lifetime and Reliability

It is widely recognized that transport and reaction phenomena are significantly influenced by the electrode microstructure, which determines the effective transport properties of each of the phases, as well as the TPB density and distribution. Diffusion of gas molecules to and from each TPB requires sufficient interconnected porosity of appropriate tortuosity, while efficient charge transport requires a percolating network of ionic and electronic conductors. The initial microstructure of an electrode depends upon the manufacturing process—deposition method, powder morphology, initial composition, sintering temperature, reduction temperature, atmosphere, etc. All these variables affect the microstructure of the electrode and must therefore be carefully controlled to avoid any operation that leads to insufficient TPB density or poor charge transport (e.g., loss of percolation in the electronic or ionic phase) or mass transport (e.g., loss of porosity). In principle the electrochemical performance can be enhanced by optimizing the electrode microstructure during fabrication. However, during operation of the SOFC, the microstructure degrades leading to losses in power density and, in some cases, the complete failure of the electrode. The microstructural degradation therefore depends not only upon how the electrode was fabricated but also critically on how the cell is operated. In anodes the metallic phase, usually nickel, is the most vulnerable component during operation at elevated temperature. Metallic particles are prone to coarsening, oxidation, and poisoning. These processes alter the electrode microstructure by reducing the number of TPB sites and the connectivity of the phases, causing a reduction in electrochemical performance. Moreover the evolution of the microstructure causes internal mechanical stresses, which can lead to the failure of the electrode. SOFC anodes are typically prepared by mixing thoroughly NiO and yttria-stabilized zirconia (YSZ) and applying them to a substrate using ceramic forming techniques (e.g., tape casting and screen printing) followed by cosintering. Large volumes of nickel (30 50 vol%) are needed to achieve adequate electronic conductivity. This makes Ni-based anodes sensitive to microstructural change. In this chapter the degradation mechanisms affecting the microstructure of SOFC electrodes are reviewed and discussed, focussing mainly on cermet anodes. The experimental and modeling techniques used to quantify the contribution of microstructural degradation are described, and strategies to enhance the robustness of the microstructure are proposed.

MICROSTRUCTURAL DEGRADATION MECHANISMS Anodes can degrade for a variety of reasons. 1. Nickel microstructural evolution. For Ni/YSZ cermet anodes, Ni coarsens and sinters at high temperature with time [1 3] leading to loss of

Microstructural Degradation Chapter | 5

81

percolation, hindering transfer of electrons to and from TPB and reducing active TPB [2] decreasing the electrochemical activity. Nickel can also be lost by volatilization under high steam concentration [4]. The degradation depends upon the nature of the interaction of Ni with YSZ at the specific conditions of temperature, pressure, atmosphere, etc. 2. Changes in the volume of nickel phases during redox cycling. As prepared initially the anodes contain NiO and are reduced in situ to nickel metal before operation. Although there is no macroscopic shrinkage of the NiO/YSZ anode at this step, the conditions of reduction are critical for the final performance of the anode. This indicates the importance of the microstructural changes during reduction [5 8]. It is however during oxidation that most undesirable changes happen as oxidation always leads to an overall expansion: with a molar volume ratio Ni/NiO of 0.6 and an increase of 69.2 vol% during oxidation (Ni - NiO), the cermet can undergo delamination, or induce cracking in other components: in particular, thin electrolytes [9] in anode-supported cells are more vulnerable than metal or electrolyte supported cells. After several redox cycles, micro-fractures on ceramics are visible [10]. It is known that the YSZ network is fundamental for the stability of the anode [10,11]. Fig. 5.1 shows a graphic sketch of the degradation associated with redox cycling where the different microstructural changes can be visualized [11]. In this simple sketch the influence of the stability of the ceramic network and the size of large NiO can be inferred. 3. Carbon and sulfur poisoning. Nickel loses catalytic activity as a consequence of poisoning through carbon or sulfur contained in the fuel and although the activity can be regenerated in some cases, this is not possible when the microstructure is irreversibly damaged [12]. One wellknown case of heavy poisoning in catalysis is the encapsulation of nickel with carbon [13,14] that in the extreme case engulfs the nickel and separates it from YSZ. The volume change in nickel when forming different phases (for example carbides or sulfides) can also produce mechanical failure of the nickel agglomerates [13] leading to mechanical failure of the anode, especially if the nickel is a fundamental part of the mechanical structure of the anode.

IMPEDANCE FOR IDENTIFYING CHANGES IN MICROSTRUCTURE Electrochemical impedance spectroscopy (EIS) is the most common tool for in situ monitoring in real time of SOFCs, and thus is vital for any study of microstructural change and consequent impact on fuel cell performance. Each electrochemical process occurring in a cell produces a feature in a certain range of frequencies in the impedance spectrum that can be deconvoluted. Typically there are several components that are present in every cell.

82

Solid Oxide Fuel Cell Lifetime and Reliability

(A) Sintered state

Electrolyte

Current collector (C) Reduced state Electrolyte

Current collector

YSZ NO N

(B) Immediate reduced state

Electrolyte

Current collector (D) After 1st reoxidation Electrolyte

Current collector

FIGURE 5.1 Graphic sketch illustrating the redox mechanism in Ni/YSZ cermets. The sketch illustrates (A) the sintered state, (B) the short-term reduced state, (C) the long-term reduced state, and (D) the reoxidized state. The arrows in (D) point to a crack in the thin electrolyte and a failure in the ceramic network of the cermet, respectively [11]. Reproduced with permission from the Electrochemical Society.

The series or Ohmic resistance (Rs) contains the resistances of all the components of the cell, especially the electrolyte, interconnects, and current collectors. Each electrode has a polarization resistance (Rp), which is related to the rate of the gas solid reactions and the TPB length. In addition there is a process related to gas diffusion, and in cells running on carbon fuel there may be processes relating to reforming reactions. Detailed analysis of the impedance spectrum may be able to separate these processes further.

Microstructural Degradation Chapter | 5

83

There are two main methods for analyzing an impedance spectrum. Equivalent circuit analysis makes assumptions about the number of processes, and approximates each one as a circuit element, and the resistance, capacitance, etc. of that element can be calculated [15]. Alternatively the impedance spectrum can be analyzed using the distribution of relaxation times (DRTs), which unlike the equivalent circuit analysis makes no assumptions about the number of processes [16]. Equivalent circuit modeling is a more established technique, easier to perform and more tolerant to lowquality data, while DRT is better able to isolate and quantify processes which overlap in the impedance spectrum and demands higher quality data. Electrodes are subject to a number of microstructural changes that can be detected by EIS. Typical changes that affect electrodes are delamination and sintering. Delamination causes the series resistance and the polarization resistances of the anode and cathode to increase equally, as the delamination effectively deactivates part of the electrolyte and the electrocatalytically active areas on the electrodes on both sides [17]. The polarization resistance of the delaminated electrode will increase slightly more than the resistance of the electrode on the other side, due to the finite in-plane resistance of SOFC electrodes. During sintering, there is an overall electrode surface area loss causing an increase in polarization resistance on one electrode only [18 20], and can also cause a corresponding shift in peak frequency in that electrode [17]. Plotting the ratios R0p =Rp and R0s =Rs against each other can help visualize these changes [21], with the slope of the graph identifying the process (Fig. 5.2) responsible for degradation. In some cases, cell

FIGURE 5.2 Graph of R0p =Rp against R0s =Rs showing the changes in a selection of four cells in a 13-cell stack during a test. The samples were planar anode-supported SOFCs (Ni YSZ/YSZ/ LSM YSZ). Cells were designated RU1 RU#13. RU#6 and RU#11 show electrode only degradation, with improvements in Rs, possibly caused by improved contacts between different components. RU#2 shows equal degradation of Rs and Rp, which can be indicative of delamination, while RU#4 shows large degradation of Rs, indicative of changes in the interconnects or electrolyte. Data from R. R. Mosbæk, J. Hjelm, R. Barfod, J. Høgh and P. V. Hendriksen, Electrochemical characterization and degradation analysis of large SOFC stacks by impedance spectroscopy. Fuel Cells, 2013, 13, 605 611.

84

Solid Oxide Fuel Cell Lifetime and Reliability

performance may increase under initial operation, due to improvement of conducting pathways through the anode, or by improvements in the contacts between different elements of the cell, and these changes can be followed through EIS [22].

USING ELECTRODE IMAGING AND QUANTIFICATION TO MEASURE DEGRADATION Introduction to Approaches Indirect measures of microstructural degradation occurring within fuel cell electrodes (e.g., EIS) provide significant in situ real-time insights but are often limited by the difficulty of relating measured observations with structurally influenced effects or their changes [15]. Therefore quantifying the microstructural degradation with direct imaging techniques and relating this to the decrease in electrochemical performance is important. Direct imaging (e.g., microscopy) helps to provide such insight and to track changes in the electrode micro/nanostructure and composition. Fundamental studies utilize direct ex situ two-dimensional (2D) imaging (e.g., scanning electron microscope (SEM)) to provide this insight (Fig. 5.3). However, 2D images inadequately describe the complex microstructure of fuel cell electrodes. This issue has been addressed through the increasing development and use of three-dimensional (3D) tomographic imaging techniques to probe the 3D nano/microstructure and reconstruct it, though at an increasing commitment of acquisition time (Fig. 5.3). Furthermore, it is possible to track nano/structural changes through time-resolved 3D-imaging (so-called “4D” tomography). The specific details in all cases often depend on the nature of the phenomena to be tracked, resolution required, and whether the imaging is itself destructive (e.g., focused ion beam (FIB) SEM tomography) or not (e.g., X-ray or transmission electron). While this information is a significant step forward in understanding electrode structure and causal relationships affecting degradation, the information

FIGURE 5.3 Schematic of different steps in understanding the fundamental structure of an electrode; each with increasing accuracy of its physical morphology.

Microstructural Degradation Chapter | 5

85

remains primarily visual and qualitative. An approach to overcome this in 4D datasets is to utilize digital volume correlation, which correlates the movement of features to provide a vector measure of change between the features across different 3D datasets, provided acquisition is nondestructive. However, this is only effective for large volumes, provided little movement occurs, and features remain identifiable. This technique has been applied more frequently on battery electrodes [23]. Further details on the microscopic instruments themselves and their capabilities applied to solid oxide fuel cell electrodes are present in the literature [24,25]. Combination of the different 3D tomographic techniques enables scales of large orders of magnitude (104 105) to be combined altogether for a single given sample improving the understanding of the microstructural changes [26]. Improved accuracy of measurement of the electrode nano/microstructures is possible through image segmentation of different phases, and basic analysis (Fig. 5.3, Stage III). Successful segmentation of a given phase enables the subsequent calculation of its contribution to electrode volume fractions, surface areas, and TPBs [27 29], providing the first quantitative step in relating morphology and electrochemical performance [30,31]. This comes at a further commitment of time involved in the segmentation of the images [32,33], especially if several samples or 3D scans are compared. Image segmentation is often critical and nontrivial and currently computer routines lack sufficient artificial intelligence to perform complex segmentation completely independently of human input. While information such as volume fractions, phase surface areas, TPBs, and some feature sizes are relatable to diffusion, electrode kinetics, poisoning, thermal, mechanical, and sintering/ aging effects, it is difficult to deconvolute the different contributions of each form of degradation to such simple metrics alone. The ability to track these metrics between different samples or with time only provides additional insights, provided that the contributions from other degradation mechanisms be minimized, which requires specific experimental approaches or the use of sophisticated rigs [34]. Nevertheless, the value of being able to relate electrode structural metrics directly with electrode degradation experimental measurements has spawned significant interest [23,24,26,27,30,35 40]. Additional metrics such as connectivity or tortuosity are often highly valuable inputs for modeling (Fig. 5.3, Stage IV) [41,42]. These are typically used in mesoscale models to simulate electrochemical performance and relate microstructural changes to electrode degradation, as discussed in the Modeling of Microstructural Degradation section. It is also possible to use the 3D electrode structure as a direct geometric input into 3D microscale models (see Fig. 5.4 and Modeling of Microstructural Degradation section), provided the volume size is representative for the electrode degradation phenomena being studied [34,38,43,44]. Recent developments have demonstrated the use of advanced quantitative imaging to directly relate metrics with specifically identifiable microstructural

86

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 5.4 Using 3D imaged data of a porous channel within a ceramic flow plate as real geometric input for models.

features. This makes it possible to deconvolute different microstructural contributions and relate them quantifiably to particular forms of degradation [33] (Fig. 5.3, Stage V). In the approach the electrode is no longer regarded as a monolithic block of a few phases. Algorithms are used to identify particles of different shapes and sizes, neighbors, necks, and interfaces that relate between all different phases present; their convolution giving rise to observable performance properties as shown in Fig. 5.5. The microstructure is taken from 3D images, segmented into phases, and then particle shape, sizes, neighbors, necks, and interfaces are calculated. The specific microstructural attributes correspond to specific properties that were both measured and modeled (Fig. 5.5). In turn this approach makes degradation mechanisms relatable to nano- and microstructural changes, offers statistical tracking of changes, and an ability to optimize electrode designs [33]. There have been academic papers in this area [4,33,45], but the full scope remains unexplored due to the expertise needed and time required to extract meaningful values. Commercial companies have likewise been primarily focussed on Stage III and visualizing data [46], although new functions are now being introduced and new software being developed that is focussed on quantifiably measuring SOFC electrode degradation using 3D imaged data [47].

3D Imaging Applied to Measuring Solid Oxide Fuel Cell Electrode Degradation Coarsening, Sintering, and Dewetting Some of the primary microstructural changes that occur are the processes of coarsening, sintering, and dewetting of the different phases, most critically in nickel. This can lead to loss in mechanical robustness, delamination, change in TPB densities and exposure or loss of active surfaces. It can also affect reactive species transport within the microstructure. The contact of different phases (e.g., Ni, YSZ, and pores) between each other affect the TPBs formed, which is why interface measurements are important. Typically elevated sintering or operating temperatures lead to greater agglomeration of nickel in anodes. Studies of microstructural degradation after 100 h operation (following sintering at 1400, 1450, and 1500 C) in humidified (3% steam) hydrogen found the lowest polarization resistance in

FIGURE 5.5 Relating microstructural attributes (A) cross-sectional image of typical Ni ScSZ in original format that is (B) segmented into different phases of Ni, ScSZ, and pore (white, gray, and black, respectively) and (C) identification of necks and interfaces. One example then reveals the nature of (D) Ni particles, (E) Ni Ni necks, and (F) interface of Ni particles with ScSZ that may be calculated within the acquired spatial domain. The changes can then be followed with examples of how these change shown for (G) Ni30 and (H) Ni50 particles. This wealth of information can then be correlated against real data from experiments, examples shown for conductivity and nano-indentation (I and J) or those from models. The relationship of specific properties of the microstructure revealed relationships between Ni Ni necks and conductivity and ScSZ ScSZ necks and Young’s modulus. This provides an example of relating microstructure to properties. More details can be found in [33]. Reproduced with permission from Elsevier.

88

Solid Oxide Fuel Cell Lifetime and Reliability

the anode with the highest active TPB density (sintered at 1450 C). The same electrode structure also had the lowest performance degradation rate. Studies investigating wetting behavior at the same temperatures between different phases within the anode (Ni YSZ pore, see Fig. 5.6) found little change in contact angles following 100, 250, and 650 h operation, following sintering at the same temperatures. However, the study reported significant differences in contact angles following reduction with electrodes sintered at different temperatures, though the study did not elucidate reasons for this. This was particularly prominent for the YSZ and pore phases [48 50]. Likewise for cathodes, studies have focussed upon the densification of the electrode microstructure and its relationship to degradation. These either involve sintering studies following fabrication or involved following measures of volume fractions, surface areas and TPB densities after having run samples under given operating conditions to obtain EIS measurements. La0.6Sr0.4Co0.2Fe0.8O3 cathodes often show measurable variation in the electrode structure within 1 2 h of being held at sintering temperatures of 850 1100 C. As before the progression of coarsening results in decreasing surface area and TPB densities. At temperatures over 1200 C some studies for La0.85Sr0.15MnO3 δ (LSM) reveal a significant increase in densification and loss of porosity, becoming most pronounced at 1450 C, the melting temperature of LSM. Composite LSM YSZ electrodes were found to behave in a more complex manner with a peak performance following sintering at 1175 C. This was attributed to a trade-off between lower LSM phase percolation, high YSZ tortuosity below this temperature and improved YSZ tortuosity but loss of TPB density above this temperature. Studies with comparison to EIS report polarization resistance changes with respect to surface areas, TPB densities and tortuosity. As before, the metrics are an early insight into potential mechanisms behind electrode degradation [36,51,52].

Changes in Volume and Redox Cycling Studies of redox cycling have rapidly utilized in situ or time-resolved imaging to follow microstructural evolution of dynamic processes. This has been carried out using transmission electron microscopes, FIB SEMs, and X-rays [25,26,49]. Oxidation of Ni YSZ anodes at temperatures below 500 C held for 10 min revealed little change in electrode structure under normal atmospheric conditions. However, when held for 10 min at temperatures of 500 C and particularly 700 C significant microstructural changes were found to occur. The Ni phase was found to coarsen; however, there was observable NiO film growth around the Ni particles as well as increased delamination and porosity around the Ni particles themselves. Overall porosity decreases from ca. 23 to 15 vol.% indicative of electrode densification [26]. Detailed 3D reconstruction (Fig. 5.7) of a Ni particle demonstrates the growth of distinguishable and porous NiO film (ca. 500 nm thickness). This

Nr’s contact angle

θNi

γNi/Pore θPore

θYSZ

γYSZ/Pore

25 Reduction Probability density (%)

γNi/YSZ

Pore

Reduction Probability density (%)

Ni

25 100h discharge

20

650h discharge 15 10 5 0

YSZ

0

50

100 150 YSZ’s contact angle

200

250

100h discharge

20

650h discharge 15 10 5 0

0

50

100 150 YSZ’s contact angle

200

250

FIGURE 5.6 3D reconstructed microstructure following sintering at 1450 C, after reduction (top-left), and running for 100 h (top-middle) and 250 h (top-right). The bottom row illustrates the approach in calculating wetting angles for each phase and wetting angle changes observed in the electrodes for YSZ and pore phase [50]. Reproduced with permission from Elsevier.

90

Solid Oxide Fuel Cell Lifetime and Reliability

is reported to block catalytically active surface sites, facilitate the measured reduction in bulk electrode porosity as well as increase localized mechanical stresses in the electrode beyond those due to thermal mismatch (Fig. 5.8). The calculated reduction in Ni/pore interfacial area was 11.5 for this particular particle. The NiO layer is not homogenous and demonstrates porosity between itself and the Ni interface [26]. Extended investigations over 100 h operation and 1 10 redox cycles revealed increased TPB densities after the redox cycle and a loss of TPB density after running the cell under constant current of 0.2 A cm22 for 20 h

FIGURE 5.7 Volume rendering from detail of a single Ni particle shown in cross-sectional view. (A) 3D rendering showing the growth of a NiO film (red) on a Ni particle (blue). (B) In the same orientation the isolated NiO film shows internal porosity. (C) The same particle rotated shows blocking of the Ni surface by NiO and a decrease in the Ni/pore interfacial area [26].

(A)

(B)

A σP1

A

(MPa) 200

0 YSZ phase Ni phase Distribution of phase in sample Section A-A Maximum principal stress FIGURE 5.8 Distribution of phases and simulation of stress build up caused by thermal mismatch between Ni and YSZ within a 3D electrode. Redox cycling would serve to increase microstructural stresses and degradation [34]. Reproduced with permission from Elsevier.

Microstructural Degradation Chapter | 5

91

FIGURE 5.9 The 3D structure of a cylindrical sample following sulfur poisoning, the diameter of the cylinder is about 12 µm (A). A second view is presented in (B), and a magnified picture of a particular region is shown in (C). The region shown in (C) is also used to show the distributions of the Ni phase in green (D) and the Ni S phase that formed in red (E) [54].

[53]. Other studies found stronger correlation between the number of cycles and loss of TPB densities as a degradation mechanism. In all cases the volumetric change observed is irreversible, measurable, and detrimental.

Changes Caused by Poisoning Impurities such as H2S cause sulfur poisoning through adsorption onto Ni electrode surfaces. Sulfidation of Ni can produce bulk Ni S-based compounds causing irreversible poisoning. The domination of either mechanism will likely depend on time, temperature, and concentration of exposure of the Ni surface to S within the electrode. However, due to the complexity of both the electrochemistry and imaging needed, relatively few studies have been conducted to follow the 3D electrode degradation. Nevertheless, through a combination of both X-ray tomography and fluorescence, recent studies have revealed the deposition of sulfur within a Ni YSZ electrode exposed to 100 ppm H2S at 800 C for 1 h (Fig. 5.9) [54].

MODELING OF MICROSTRUCTURAL DEGRADATION Degradation models can be classified into two categories: microscale models and mesoscale models, as illustrated in Fig. 5.10. Microscale models simulate the evolution of the electrode 3D microstructure (e.g., reproducing the nickel coarsening in cermet anodes). Two approaches have been proposed: phase field method and cellular automata/ kinetic Monte Carlo methods. In the phase field method [30,48,55] the

92

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 5.10 Schematic illustration of microstructural degradation models: (A) microscale models simulate the evolution of the microstructure; (B) mesoscale models solve continuum transport and reaction equations within the electrode domain and simulate the decrease in electrochemical performance upon variation of microstructural parameters.

Microstructural Degradation Chapter | 5

93

Cahn Hilliard evolution equation is solved to minimize the free energy of the interfaces, taking into account the thermodynamics and kinetics of the coarsening without any empirical parameters. Phase field simulations have confirmed the reduction in TPB density and Ni surface area observed experimentally. Recently the phase field method has been extended to take into account the reduction of nickel oxide [56], making it suitable to simulate redox cycling. The nickel particle evolution can alternatively be simulated by using cellular automata [57,58] and kinetic Monte Carlo simulations [59 61]. Despite the physics of the coarsening not being described from first principles, these models allow fast computation in large domains and include additional mechanisms such as nickel evaporation/deposition. In general the predictions of microscale models are sensitive to specific parameters related to the wetting properties of the phases, thus more efforts must be dedicated to validate these models by observing the changes occurring in the same region of the electrode [62]. Mesoscale models simulate the effect of a microstructural degradation on electrode performance. Transport and reaction phenomena are solved with a set of macrohomogeneous conservation equations, with particle-level details represented through effective properties. In some studies the degradation has phenomenologically been simulated by reducing the catalytic surface area [17] or the TPB density [63], for example, as a consequence of nickel dusting from carbon deposition [64]. Nevertheless, microstructural properties are strongly correlated, thus percolation models have been adopted to incorporate the effect of nickel coarsening on the different effective properties [1,65]. The coupling between different effective properties can be described more accurately by using data obtained from tomographic reconstruction, allowing also for taking into account the distribution of microstructural properties within the electrode volume [66]. Although a proper validation at different length scales is still lacking, mesoscale models are expected to assist the interpretation of experimental data. Physically based impedance simulations can be used as a diagnostic tool to deconvolute the different contributions present in EIS data. Any change in impedance upon degradation can be quantitatively linked to the variation in a specific microstructural parameter (e.g., TPB density, tortuosity factor, etc.). As an example, mesoscale simulations have been used to distinguish the impedance contribution due to a decrease in catalytic surface area from a delamination [17]. However, more efforts are necessary to couple mesoscale models with microscale ones, for example, by integrating the results of phase field simulations into electrochemical models [67,68]. Finally, mechanical models have been proposed to simulate the mechanical stress arising within the electrode microstructure [34,37,69]. These models are expected to be coupled with microscale evolution models [32,70] and

94

Solid Oxide Fuel Cell Lifetime and Reliability

eventually with macroscale thermomechanical models [71,72], allowing for the prediction of the mechanical failure of the system. There remains great potential in modeling microstructural evolution and its link to performance degradation.

MICROSTRUCTURAL DESIGN STRATEGIES The importance of the microstructural features (percolation, porosity, TPB, etc.) in the performance of the electrodes has been outlined in the previous sections. Electrochemical characterization and tomographic studies in combination with modeling provide a powerful set of tools to understand degradation processes. Finally this acquired knowledge can be used for the design of a new generation of SOFC electrodes. A new design approach is to manufacture a scaffold of the electrolyte followed by incorporation with an active catalytic phase by infiltration or any other method [73 75]. One of the key points of this strategy is that the microstructure of both phases can be independently designed. For example, porosity can be controlled during scaffold fabrication and the amount of infiltrated nickel necessary for percolation can be kept to a minimum: therefore the stresses that lead to fractures and cracking in anodes during redox cycling can be minimized as there is enough porosity to allow the nickel phase to expand [76]. An improved electrochemical activity has been clearly linked to the larger TPB density obtained by impregnation [77]. The problem of degradation by nickel coarsening and sintering depends upon the nature of the interaction between Ni and YSZ. Since the nonwetting nature of nickel on YSZ leads to an agglomeration of the metal, the wetting can be improved by alloying of nickel with another metal or modification of the YSZ substrate [78] both of which can be carried out at different fabrication steps. A radical solution to avoid the problem of degradation by nickel coarsening while keeping the catalytic activity is to use a ceramic scaffold that exhibits oxygen ion and electron conductivity (e.g., Ni-infiltrated Gddoped ceria) or an all-ceramic anode with either infiltrated or ex-solved nickel (e.g., La0.4Sr0.4TiO3-based materials) [79]. Other degradation problems such as damage by redox cycling can be addressed by the incorporation of sacrificial layers. During operation, in the case of a sudden expected air leak, an extra nickel-rich layer can consume the leaked oxygen while giving enough time for the whole cell to cool down and avoid damage to the most electrochemically active layer [80]. A similar idea can be used to minimize the microstructural change associated with severe carbon or sulfur poisoning: an external sacrificial layer with the function of oxidizing carbon to CO2 or to oxidize H2S while protecting the more active area at the interface with the electrolyte.

Microstructural Degradation Chapter | 5

95

CONCLUSIONS Changes in the microstructure of electrodes are important in defining SOFC lifetime. This degradation depends not only on its initial microstructure, which in turn depends on the fabrication process, but also on the conditions of operation. To gain a clear understanding of the mechanism of degradation, the microstructural properties (e.g., TPBs, porosity, and tortuosity) along with the electrochemical responses need to be quantified. Tomographic techniques are being increasingly used to provide this information and at the same time models use these datasets to describe the electrochemical responses measured and, most importantly, to link the change in microstructure to the performance degradation. Although considerable progress has been made in fabrication, imaging, and modeling, there is still a need for high-quality imaging and microstructural quantification, which can be coupled with impedance spectra studies through physically based models in order to decouple and quantify different degradation mechanisms and assess the sensitivity to microstructural parameters. The knowledge gained must then be used to design new electrodes that can extend the lifetime of SOFCs.

REFERENCES [1] Tanasini P, Cannarozzo M, Costamagna P, Faes A, Van Herle J, Hessler-Wyser A, et al. Experimental and theoretical investigation of degradation mechanisms by particle coarsening in SOFC electrodes. Fuel Cells 2009;9:740 52. [2] Kennouche D, Chen-Wiegart Y-c K, Cronin JS, Wang J, Barnett SA. F1293-F1304. Threedimensional microstructural evolution of Ni-Yttria-stabilized zirconia solid oxide fuel cell anodes at elevated temperatures. J Electrochem Soc 2013;160. [3] Lee J-H, Moon H, Lee H-W, Kim J, Kim J-D, Yoon K-H. Quantitative analysis of microstructure and its related electrical property of SOFC anode, Ni YSZ cermet. Solid State Ionics 2002;148:15 26. [4] Holzer L, Iwanschitz B, Hocker T, Mu¨nch B, Prestat M, Wiedenmann D, et al. Microstructure degradation of cermet anodes for solid oxide fuel cells: quantification of nickel grain growth in dry and in humid atmospheres. J Power Sources 2011;196:1279 94. [5] Faes A, Jeangros Q, Wagner JB, Hansen TW, Van Herle J, Brisse A, et al. In situ reduction and oxidation of nickel from solid oxide fuel cells in a transmission electron microscope. ECS Trans 2009;25:1985 92. [6] Pihlatie M, Kaiser A, Mogensen M. Mechanical properties of NiO/Ni YSZ composites depending on temperature, porosity and redox cycling. J Eur Ceram Soc 2009;29:1657 64. [7] Radovic M, Lara-Curzio E. Mechanical properties of tape cast nickel-based anode materials for solid oxide fuel cells before and after reduction in hydrogen. Acta Mater 2004;52:5747 56. [8] Sarantaridis D, Atkinson A. Redox Cycling of Ni-Based Solid Oxide Fuel Cell Anodes: A Review. Fuel Cells 2007;7:246 58. [9] Cassidy M, Lindsay G, Kendall K. The reduction of nickel- zirconia cermet anodes and the effects on supported thin electrolytes. J Power Sources 1996;61:189 92.

96

Solid Oxide Fuel Cell Lifetime and Reliability

[10] Fouquet D, Mu¨ller A, Weber A, Ivers-Tiffe´e E. Kinetics of oxidation and reduction of Ni/ YSZ cermets. Ionics 2003;9:103 8. [11] Klemensø T, Chung C, Larsen PH, Mogensen M. A2186-A2192. The mechanism behind redox instability of anodes in high-temperature SOFCs. J Electrochem Soc 2005;152. [12] Boldrin P, Ruiz-Trejo E, Mermelstein J, Bermúdez Menéndez JM, Ramı ́rez Reina TS, Brandon NP. Strategies for carbon and sulfur tolerant solid oxide fuel cell materials, incorporating lessons from heterogeneous catalysis. Chem Rev 2016;116:13633 84. [13] Bartholomew CH. Mechanisms of catalyst deactivation. Appl Catal A Gen 2001;212:17 60. [14] Takeguchi T, Kani Y, Yano T, Kikuchi R, Eguchi K, Tsujimoto K, et al. Study on steam reforming of CH4 and C 2 hydrocarbons and carbon deposition on Ni-YSZ. J Power Sources 2002;112:588 95. [15] Lvovich VF. Impedance spectroscopy: applications to electrochemical and dielectric phenomena. Hoboken New Jersey. 1 edn Wiley; 2012. [16] Schichlein H, Feuerstein M, Mu¨ller A, Weber A, Kru¨gel A, Ivers-Tiffe´e E. System identification: a new modelling approach for SOFC single cells. SOFC VI Electrochem Soc Proc Ser 1999;1069 77. [17] Gazzarri JI, Kesler O. Electrochemical AC impedance model of a solid oxide fuel cell and its application to diagnosis of multiple degradation modes. J Power Sources 2007;167:100 10. [18] Laurencin J, Delette G, Sicardy O, Rosini S, Lefebvre-Joud F. Impact of ‘redox’cycles on performances of solid oxide fuel cells: case of the electrolyte supported cells. J Power Sources 2010;195:2747 53. [19] Komatsu T, Yoshida Y, Watanabe K, Chiba R, Taguchi H, Orui H, et al. Degradation behavior of anode-supported solid oxide fuel cell using LNF cathode as function of current load. J Power Sources 2010;195:5601 5. [20] Hauch A, Mogensen M, Hagen A. Ni/YSZ electrode degradation studied by impedance spectroscopy—Effect of p (H2O). Solid State Ionics 2011;192:547 51. [21] Mosbæk RR, Hjelm J, Barfod R, Høgh J, Hendriksen PV. Electrochemical characterization and degradation analysis of large SOFC stacks by impedance spectroscopy. Fuel Cells 2013;13:605 11. [22] Choi JH, Lee T, Choi M, Yoo Y-S, Baek S-W, Bae J. Long-term performance of anodesupported SOFC integrated with metal interconnect by joining process. Int J Hydrog Energy 2010;35:4285 91. [23] Eastwood DS, Yufit V, Gelb J, Gu A, Bradley RS, Harris SJ, et al. Lithiation-induced dilation mapping in a lithium-ion battery electrode by 3D X-ray microscopy and digital volume correlation. Adv Energy Mater 2014;4. [24] Uchic M, Holzer L, Inkson B, Principe E, Munroe P. Three-dimensional microstructural characterization using focused ion beam tomography. MRS Bull 2007;32:408 16. [25] Jeangros Q, Faes A, Wagner JB, Hansen TW, Aschauer U, Van herle J, et al. In situ redox cycle of a nickel YSZ fuel cell anode in an environmental transmission electron microscope. Acta Mater 2010;58:4578 89. [26] Shearing PR, Bradley RS, Gelb J, Tariq F, Withers PJ, Brandon NP. Exploring microstructural changes associated with oxidation in Ni YSZ SOFC electrodes using high resolution X-ray computed tomography. Solid State Ionics 2012;216:69 72. [27] Wilson JR, Kobsiriphat W, Mendoza R, Chen H-Y, Hiller JM, Miller DJ, et al. Threedimensional reconstruction of a solid-oxide fuel-cell anode. Nat Mater 2006;5:541 4.

Microstructural Degradation Chapter | 5

97

[28] Iwai H, Shikazono N, Matsui T, Teshima H, Kishimoto M, Kishida R, et al. Quantification of SOFC anode microstructure based on dual beam FIB-SEM technique. J Power Sources 2010;195:955 61. [29] Shearing PR, Golbert J, Chater RJ, Brandon NP. 3D reconstruction of SOFC anodes using a focused ion beam lift-out technique. Chem Eng Sci 2009;64:3928 33. [30] Chen H-Y, Yu H-C, Cronin JS, Wilson JR, Barnett SA, Thornton K. Simulation of coarsening in three-phase solid oxide fuel cell anodes. J Power Sources 2011;196:1333 7. [31] Kishimoto M, Iwai H, Saito M, Yoshida H. Quantitative evaluation of solid oxide fuel cell porous anode microstructure based on focused ion beam and scanning electron microscope technique and prediction of anode overpotentials. J Power Sources 2011;196 4555 63. [32] Abdullah T, Liu L. Phase field simulation coupling microstructural evolution and crack propagation during performance degradation of solid oxide fuel cells. ECS Trans 2015;66:207 17. [33] Tariq F, Kishimoto M, Yufit V, Cui G, Somalu M, Brandon NP. 3D imaging and quantification of interfaces in SOFC anodes. J Eur Ceram Soc 2014;34:3755 61. [34] Clague R, Shearing PR, Lee PD, Zhang Z, Brett DJL, Marquis AJ, et al. Stress analysis of solid oxide fuel cell anode microstructure reconstructed from focused ion beam tomography. J Power Sources 2011;196:9018 21. [35] Kishimoto M, Iwai H, Saito M, Yoshida H. Three-dimensional simulation of SOFC anode polarization characteristics based on sub-grid scale modeling of microstructure. J Electrochem Soc 2012;159:B315 23. [36] Gostovic D, Smith JR, Kundinger DP, Jones KS, Wachsman ED. Three-dimensional reconstruction of porous LSCF cathodes. Electrochem Solid-State Lett 2007;10:B214 17. [37] Chen Z, Wang X, Giuliani F, Atkinson A. Analyses of microstructural and elastic properties of porous SOFC cathodes based on focused ion beam tomography. J Power Sources 2015;273:486 94. [38] Carraro T, Joos J, Ru¨ger B, Weber A, Ivers-Tiffe´e E. 3D finite element model for reconstructed mixed-conducting cathodes: I. Performance quantification. Electrochim Acta 2012;77:315 23. [39] Joos J, Ender M, Rotscholl I, Menzler NH, Ivers-Tiffe´e E. Quantification of double-layer Ni/YSZ fuel cell anodes from focused ion beam tomography data. J Power Sources 2014;246:819 30. [40] Wilson JR, Gameiro M, Mischaikow K, Kalies W, Voorhees PW, Barnett SA. Threedimensional analysis of solid oxide fuel cell Ni-YSZ anode interconnectivity. Microscopy Microanal 2009;15:71 7. [41] Jin X, Xue X. Micro modeling study of cathode/electrolyte interfacial stresses for solid oxide fuel cells. J Electrochem Soc 2013;160:F815 23. [42] Ferrari T, Bertei A, Nicolella C. Estimation of 3D effective properties from 2D cross sections in porous electrodes. ECS Trans 2015;68:2991 3001. [43] Roussel D, Lichtner A, Jauffre`s D, Villanova J, Bordia RK, Martin CL. Strength of hierarchically porous ceramics: discrete simulations on X-ray nanotomography images. Scripta Mater 2016;113:250 3. [44] Joos J, Carraro T, Weber A, Ivers-Tiffe´e E. Reconstruction of porous electrodes by FIB/ SEM for detailed microstructure modeling. J Power Sources 2011;196:7302 7. [45] Vivet N, Chupin S, Estrade E, Piquero T, Pommier PL, Rochais D, et al. 3D Microstructural characterization of a solid oxide fuel cell anode reconstructed by focused ion beam tomography. J Power Sources 2011;196:7541 9.

98

Solid Oxide Fuel Cell Lifetime and Reliability

[46] VSG3D F, Avizo, , http://www.fei.com/software/avizo3d/ . [accessed January 2016). [47] I. E. Q. I. Division, IQM Elements QuiQ 3D, , http://www.iqm-elements.com/software. html . [January 2016]. [48] Jiao Z, Shikazono N. Simulation of solid oxide fuel cell anode microstructure evolution using phase field method. J Electrochem Soc 2013;160:F709 15. [49] Jiao Z, Shikazono N. Simulation of nickel morphological and crystal structures evolution in solid oxide fuel cell anode using phase field method. J Electrochem Soc 2014;161: F577 82. [50] Sun X, Hasegawa Y, Kohno H, Jiao Z, Hayakawa K, Okita K, et al. Calculation of contact angles at triple phase boundary in solid oxide fuel cell anode using the level set method. Mater Character 2014;96:100 7. [51] Smith JR, Chen A, Gostovic D, Hickey D, Kundinger D, Duncan KL, et al. Evaluation of the relationship between cathode microstructure and electrochemical behavior for SOFCs. Solid State Ionics 2009;180:90 8. [52] Scott Cronin J, Muangnapoh K, Patterson Z, Yakal-Kremski KJ, Dravid VP, Barnett SA. Effect of firing temperature on LSM-YSZ composite cathodes: a combined threedimensional microstructure and impedance spectroscopy study. J Electrochem Soc 2012;159:B385 93. [53] Shimura T, Jiao Z, Hara S, Shikazono N. Quantitative analysis of solid oxide fuel cell anode microstructure change during redox cycles. J Power Sources 2014;267:58 68. [54] Harris WM, Lombardo JJ, Nelson GJ, Lai B, Wang S, Vila-Comamala J, et al. Threedimensional microstructural imaging of sulfur poisoning-induced degradation in a Ni-YSZ anode of solid oxide fuel cells. Sci Rep 2014;4:5246. [55] Li Q, Liang L, Gerdes K, Chen L-Q. Three-dimensional microstructural imaging of sulfur poisoning-induced degradation in a Ni-YSZ anode of solid oxide fuel cells. Appl Phys Lett 2012;101:033909 033909 [56] Jiao Z, Shikazono N. Simulation of the reduction process of solid oxide fuel cell composite anode based on phase field method. J Power Sources 2016;305:10 16. [57] Wang X, Atkinson A. Modeling microstructure evolution of Ni cermet using a cellular automaton approach. J Electrochem Soc 2014;161:F605 14. [58] Wang X, Atkinson A. Simulation and prediction of 3-D microstructural evolution and long term performance of Ni-YSZ anode. ECS Trans 2015;68:2867 73. [59] Hara S, Ohi A, Shikazono N. Sintering analysis of sub-micron-sized nickel powders: kinetic Monte Carlo simulation verified by FIB SEM reconstruction. J Power Sources 2015;276:105 12. [60] Hara S, Shikata K, Shikazono N, Izumi S, Sakai S. Monte Carlo study on the constraint effect of YSZ phase on Ni sintering in Ni-YSZ composite system. ECS Trans 2013;57:2857 63. [61] Zhang Y, Ni M, Yan M, Chen F. Thermal aging stability of infiltrated solid oxide fuel cell electrode microstructures: a three-dimensional kinetic monte carlo simulation. J Power Sources 2015;299:578 86. [62] Kennouche D, Chen-Wiegart Y-c K, Yakal-Kremski KJ, Wang J, Gibbs JW, Voorhees PW, et al. Observing the microstructural evolution of Ni-Yttria-stabilized zirconia solid oxide fuel cell anodes. Acta Mater 2016;103:204 10. [63] Lay-Grindler E, Laurencin J, Delette G, Aicart J, Petitjean M, Dessemond L. Micro modelling of solid oxide electrolysis cell: From performance to durability. Int J Hydrog Energy 2013;38:6917 29.

Microstructural Degradation Chapter | 5

99

[64] Yurkiv V. Reformate-operated SOFC anode performance and degradation considering solid carbon formation: a modeling and simulation study. Electrochim Acta 2014;143:114 28. [65] Nakajo A, Tanasini P, Diethelm S, Van herle J, Favrat D. Electrochemical model of solid oxide fuel cell for simulation at the stack scale II: implementation of degradation processes. J Electrochem Soc 2011;158:B1102 18. [66] Pecho O, Mai A, Mu¨nch B, Hocker T, Flatt R, Holzer L. 3D microstructure effects in Ni-YSZ anodes: influence of TPB lengths on the electrochemical performance. Materials 2015;8:7129 44. [67] Abdeljawad F, Vo¨lker B, Davis R, McMeeking RM, Haataja M. Connecting microstructural coarsening processes to electrochemical performance in solid oxide fuel cells: An integrated modeling approach. J Power Sources 2014;250:319 31. [68] Jiao Z, Shimura T, Shikazono N. Numerical assessment of SOFC anode polarization with microstructure evolution. ECS Trans 2015;68:1281 9. [69] Celik S, Ibrahimoglu B, Toros S, Mat MD. Three dimensional stress analysis of solid oxide fuel cell anode micro structure. Int J Hydrog Energy 2014;39:19119 31. [70] Abdeljawad F, Haataja M. Microstructural coarsening effects on redox instability and mechanical damage in solid oxide fuel cell anodes. J Appl Phys 2013;114:183519. [71] Greco F, Frandsen HL, Nakajo A, Madsen MF, Van herle J. Modelling the impact of creep on the probability of failure of a solid oxide fuel cell stack. J Eur Ceram Soc 2014;34:2695 704. [72] Pihlatie MH, Frandsen HL, Kaiser A, Mogensen M. Continuum mechanics simulations of NiO/Ni YSZ composites during reduction and re-oxidation. J Power Sources 2010;195:2677 90. [73] Ruiz-Trejo E, Irvine JTS. Electrolysis of CO2 in a proton conducting membrane. Solid State Ionics 2013;252:157 64. [74] Ruiz-Trejo E, Atkinson A, Brandon NP. Metallizing porous scaffolds as an alternative fabrication method for solid oxide fuel cell anodes. J Power Sources 2015;280:81 9. [75] Ruiz-Trejo E, Azad AK, Irvine JTS. A 60-second microwave-assisted synthesis of nickel foam and its application to the impregnation of porous scaffolds. J Electrochem Soc 2015;162:F273 9. [76] Busawon A, Sarantaridis D, Atkinson A. Ni infiltration as a possible solution to the redox problem of SOFC anodes. Electrochem Solid-State Lett 2008;11:B186 9. [77] Kishimoto M, Lomberg M, Ruiz-Trejo E, Brandon NP. Enhanced triple-phase boundary density in infiltrated electrodes for solid oxide fuel cells demonstrated by high-resolution tomography. J Power Sources 2014;266:291 5. [78] Tsoga A, Nikolopoulos P, Naoumidis A. Modified YSZ/Ni structures with improved stability. Ionics 2, 427 434. [79] Neagu D, Tsekouras G, Miller DN, Me´nard H, Irvine JTS. In situ growth of nanoparticles through control of non-stoichiometry. Nat Chem 2013;5:916 23. [80] Wood A, Pastula M, Waldbillig D, Ivey D. Initial testing of solutions to redox problems with anode-supported SOFC. J Electrochem Soc 2006;153:A1929 34.

Chapter 6

Cathode Degradation From Airborne Contaminants in Solid Oxide Fuel Cells: A Review Ashish Aphale, Chiying Liang, Boxun Hu and Prabhakar Singh University of Connecticut, Storrs, CT, United States

Chapter Outline Introduction Degradation in Solid Oxide Fuel Cell Systems Cathode Materials Long-Term Degradation in the Cathode

102 103 104 106

Approaches for the Mitigation of Chromium-Assisted Cathode Degradation Summary and Outlook Acknowledgement References

Solid Oxide Fuel Cell Lifetime and Reliability. DOI: http://dx.doi.org/10.1016/B978-0-08-101102-7.00006-4 Copyright © 2017 Elsevier Ltd. All rights reserved.

112 113 114 114

101

102

Solid Oxide Fuel Cell Lifetime and Reliability

INTRODUCTION The need for clean and efficient energy technologies for global deployment under carbon constrained conditions has become a topic of debate among researchers and policy-makers in both developed and developing nations [1]. The drive for the development of technologies stems from achieving higher electrical efficiency with low emissions, the ability to reduce and capture carbon, as well as utilize and prolong the life of existing local resources. One such technology is solid oxide fuel cells (SOFCs), which directly converts the chemical energy of the fuels (coal derived, biomass, gaseous hydrocarbons, and hydrogen) into electricity and high-quality heat in the temperature range of 6001000 C [2,3]. SOFC power systems have found applications in centralized (MWe class) and distributed (large kWe class) power generation [46]. The key advantages of SOFCs are high efficiency, hybridization, modular design, small CO2 foot print per kWh of generated electricity and fuel flexibility [2,7]. A typical SOFC stack consists of several single cells as shown in Fig. 6.1. The region connecting electrolyte, gaseous reactants as well as the electrocatalyst material is called the triple-phase boundary (TPB)—and its presence is important for any electrochemical reactions to occur [8]. The cathode electrode facilitates the electrochemical reduction of oxygen: 1 2 O2

1 2e 5 O5

ðoxygen reduction; cathodic reactionÞ

ð6:1Þ

Extensive research has been conducted and reported widely in the literature as it pertains to the development of cathode materials that must meet key system requirements such as high electrocatalytic activity for oxygen reduction reaction (ORR), chemical stability and a matching coefficients of thermal expansions (CTE) with the adjoining cell component materials. Cathodes such as La1xSrxMnO3δ (LSM), La1xSrxFeO3δ (LSF),

FIGURE 6.1 Typical SOFC single cell consisting of porous electrodes and dense electrolyte, along with a simplified illustration of the TPB reaction sites and transport of oxygen ion across the state-of-the-art LSM cathode and YSZ electrolyte.

Cathode Degradation From Airborne Contaminants Chapter | 6

103

La1xSrxCoO3δ (LSC), and La1xSrxCo1yFeyO3δ (LSCF) have been widely studied, electrochemically tested and utilized in power systems [9]. Varying the compositions as well as oxygen stoichiometry enables the optimization of the electronic and catalytic properties of the cathode [10] while changes in the oxygen stoichiometry in the lattice leads to anomalous shrinkage and influences the dimensional stability [11,12]. LSM has been the stateof-the-art cathode material as it satisfied the above requirements suitable for high-temperature SOFCs operating between 800 C and 1000 C [10]. On the other hand, for intermediate-temperature SOFCs (IT-SOFCs) operating in the range of 600800 C, materials optimization becomes necessary due to limited performance in ORR at the cathode. The introduction of composite electrodes offering mixed ionic and electronic conductivities (MIECs) such as LSCF allows an increase in the active area for oxygen reduction, which in a pure electronic conductor is limited to the TPB. The substantial ionic conductivity of MIECs creates pathways for oxygen ion migration through the MIEC itself and increases the reaction area [13]. The ability of SOFCs to maintain long-term performance remains a major bottleneck toward large scale commercialization [14,15]. In order to compete with the existing technologies the current research priority is to develop materials and processes to reduce the cost and improve long-term performance [1618].

Degradation in Solid Oxide Fuel Cell Systems For wide scale implementation of fuel cell technology in centralized or distributed power generation applications, a long-term operation of at least 40,000 h [19], with a maximum degradation rate of 0.2% per 1000 h is required as indicated by the US Department of Energy (USDOE) [20]. Degradation is defined as a performance drop in physical, chemical, and/or mechanical activity of the materials sometimes leading to permanent failure. Often, one degradation event triggers a chain of several degradations, thereby reducing the SOFC reliability [21]. In the SOFC stack, both electrochemically active and inactive components such as anode, cathode, electrolyte membranes, interconnect (IC), and seals are assembled together in electrical series connection and requires matching CTE, chemical and mechanical stability as well as phase stability [22]. Exposure of commonly used perovskite cathode materials to “real world” air containing minor constituents such as H2O, CO2, CrOx, and SOx has shown tendency to the exsolution of A-site alkaline earth dopant from the host matrix, coverage of surface (solidgas) and interface (solidsolid) with undesirable reaction products. The perovskite cathode also shows a tendency to react with both the electrolyte and gas phase impurities resulting in the formation of a resistive pyrochlore (La2Zr2O7) at the electrolyte interface and dopant oxides and hydroxides at the exposed surface [23,24]. Metallic alloys used in the cell to cell IC and the balance of plant (BoP) subsystems

104

Solid Oxide Fuel Cell Lifetime and Reliability

(air and fuel handling systems, pipes, and ducts manifolds) show metal corrosion as well as evaporation of Cr vapor that triggers cathode degradation [2528]. The stable nature of the electrical performance has been evaluated by analyzing the changes in the morphology of the component material and their chemical interactions with each other and with the existing environment [29]. Long-term stack and system tests indicate that the cathode degradation remains a major contributor to the performance deterioration in SOFC power systems [30]. Fig. 6.2A shows contributions from various losses namely anodic and cathodic polarization, Nernst loss [31], and losses due to IR drop [32]. It is evident from Fig. 6.2B that the losses from the cathode polarization has the highest contribution amongst all the other losses during 10,000 h operation of the stack. In general the performance degradation arises from an increase in the contact resistance between electrodes and current collectors, the instability of the contact between electrode, electrolyte, and IC materials [33], delamination of the bonded interfaces due to mismatches in thermal expansion coefficients [34], and the presence of gas phase contaminants leading to unwanted reaction products formation and surface coverage.

Cathode Materials Doped perovskites (ABO3, where A-site is occupied by rare and alkaline earth elements and B-site with transition metal) have been extensively used as the cathode in SOFCs (Fig. 6.3). Full or partial substitution of A or B cations and the formation of oxygen vacancies is possible [35,36]. The bulk ionic conductivity increases by optimizing the selection of A and B cations and introduction of oxygen ion vacancies [37]. Oxygen excess chemistry can also readily be achieved in the structure because of cation nonstoichiometry [38].

FIGURE 6.2 Long-term performance degradation (A) contribution of electrochemical losses in SOFC; adopted from Yokokawa H. 11th international conference on ceramic materials components for energy and environmental applications, Japan, 2014, (B) comparative increase in the losses during 10,000 h operation.

Cathode Degradation From Airborne Contaminants Chapter | 6

105

FIGURE 6.3 Unit cell of perovskites with ABO3 structure. The larger spaces are occupied by A-site cations and B-site cations in the center occupy smaller octahedral holes.

Any distortion in the structure of perovskites can be represented by the Goldschmidt tolerance factor (t): pffiffiffi ð6:2Þ t 5 ðrA 1 rO Þ= 2ðrB 1 rO Þ where rA, rB, and rO are the ionic radii of A, B, and O ions, respectively. An ideal cubic structure has a tolerance factor of unity; however, most of the perovskites are distorted and do not have cubic symmetry. Perovskites with ABO3 type structure display several interesting physical characteristics such as higher thermal conductivity [38], wide variations in electrical properties, for example, dielectric constant, metallic conductivity, or semiconduction. The conductivity can be further enhanced by partial substitution of the lanthanide (A-site ion) by a divalent ion, such as in the case of La1xSrxMnO3, where at x 5 0 Mn ions are Mn31 and increasing x results in the creation of Mn41 holes, thus increasing the conductivity [38]. The electrical performance of the cathode can be further improved by optimizing the microstructure such as utilizing nanostructured high surface area (HSA) electrode materials [39]. These HSA electrodes can be fabricated using processes such as chemical vapor deposition and wet impregnation to form a connected network of solid particles. However, the nano-sized particles tend to agglomerate at the high temperatures thereby decreasing the performance with time [36]. Sr-doped lanthanum manganite (LaMnO3) cathodes are the state-of-theart material for SOFCs operating in the 8001000 C temperature range. The ionic and electronic properties of the cathode can be tailored by making rational changes to either oxygen excess or deficient nonstoichiometries. Sr, because of its matching size with La, is most commonly used as an A-site dopant for the oxidation of manganese ions and increase in the electron-hole concentration for improvement in the electrical conductivity. The optimized Sr concentration level is commonly limited to 30 mol% due to the formation of unwanted insulating phases such as SrZrO3 [40] at the electrolyte interface

106

Solid Oxide Fuel Cell Lifetime and Reliability

during cell operation. At high temperatures, LaMnO3 reacts with YSZ to form La2Zr2O7, and substitution of Sr has shown to decrease the reactivity of LSM and YSZ [41]. Lanthanum ferrite (LaFeO3)-based cathodes such as Sr-doped LaFeO3 (LSF) are used at a relatively lower operating temperature of 600800 C. Due to the stability of Fe31 ions, the reactivity with YSZ electrolyte is significantly reduced [42]. Although pure LSF is thermodynamically more stable, its electrochemical performance is inferior to ferrocobaltite-based cathodes [43] due to interdiffusion of Zr and La [44]. LSCF has superior oxygen reduction kinetics as they have rapid oxygen incorporation in the lattice and is only limited by oxygen surface coverage [45,46] and higher electrical conductivity between 600 C and 800 C [47,48]. The increase in the electronic conductivity is influenced by the concentration of Fe and Co at the B-site. Cobaltite-based cathodes (LSC) exhibit higher electronic density of states so as to enhance the electron transfer between surface cation and potentially catalyzed species. Sr ions in La1xSrCoO3δ are occupied at the La lattice sites, creating electron holes. The incorporation of Sr ions is compensated by the formation of positive charges, comprising Co:Co and oxygen vacancies [VO:: , which maintains the overall electro-neutrality. Cobalt-based materials display a decrease in the cathode polarization resistance because of higher ionic and electronic conductivities [36].

Long-Term Degradation in the Cathode Overall cathode degradation (structural and electrical) is attributed to cationic interdiffusion, localized densification and compound formation mainly due to solidsolid and solidgas interactions at the cell and stack [15] level. During the solidsolid interactions, the interfacial reactions often lead to the formation of thermochemically stable reaction products. The interfacial compounds usually demonstrate poor electrical and electrocatalytic properties as well as different CTE that adversely impacts the electrical and mechanical stability [43]. For example, the formation of La2Zr2O7 at the cathode/electrolyte interface increases the cell resistance [49,50] and delamination at the LSM/YSZ interface [24,51]. One approach to overcome the issue of zirconate formation between cathode and electrolyte is the use of thin reaction barrier layer that prevents chemical interaction between two adjacent phases. The use of a ceria reaction barrier layer was first demonstrated at Westinghouse [52] in tubular air electrode cells fabricated from doped LSM. A gadolinium (Gd)-doped ceriabased interlayer has also been used between LSCF electrode and 8YSZ electrolyte to block the direct contact between LSCF and YSZ [53,54]. Mn-doped YSZ has also been used to mitigate and/or slow down the manganese diffusion and zirconate formation. In the long run, manganese eventually diffuses from LSM to the Mn-doped YSZ interlayer to bulk YSZ [55]. Electrical performance degradation also arises from the formation of porosity

Cathode Degradation From Airborne Contaminants Chapter | 6

107

at the YSZ bulk and grain boundaries at the interface, which arises due to high oxygen partial pressure in YSZ near the interface [56]. Degradation induced by the solidgas interactions results from undesirable reaction products formation due to interactions between gaseous constituents and the cathode. For “real world” air atmosphere the reactions usually involve interactions of minor intrinsic (H2O, CO2, and SOx) and extrinsic (CrOx and CrOx(OH)x) gas phase impurities with the cathode and subsequent formation of reaction products, and surface and TPB coverage. For example, a slight degradation in the performance of LSM cathode under long-term exposure of SO2 containing air has been reported due to the formation of SrSO4 at the TPBs and the LSM cathode surface [57]. Cathode degradation also arises due to interactions with gas phase chromium vapors generated from the metallic ICs, and BoP components—the process commonly termed as “chromium poisoning” [58]. It is noted that the simultaneous presence of impurities in air has a tendency to exaggerate the effect of chromium and moisture on cathode degradation [57,59,60].

Impurities in the Air Stream (Intrinsic Impurities) “Real world” air atmosphere composition varies widely (rural vs. urban, inland vs. seacoast, and summer vs. winter) and may contain differing levels of H2O, CO2, SOx, and particulate matter. USEPA, under the Clean Air Act [61] has set a National Ambient Air Quality Standards for pollutants considered harmful for public health and the environment and names six common pollutants. The current air quality standards for SO2 is 75 ppb [62]. The role of contaminants present in the air atmosphere, contacting the electrode surface, has been examined in detail during the evaluation of a number of cathode materials under simulated SOFC operating conditions and the findings indicate that the presence of intrinsic impurities accelerates performance degradation and remains the major common contributor of SOFC performance loss [25,63,64]. Degradation in the Presence of Water Vapor The presence of water vapor in air shows cathode degradation to be strongly dependent on the humidification level as well as temperature [65]. Higher concentrations of water vapor at 800 C have been shown to cause a larger cell voltage drop because of an increase in the cathodic IR losses. Experiments show an increase in polarization resistance at higher frequencies which signify the hindrance in the transfer of oxygen ion across the LSM/ YSZ interface. Remarkably, upon removal of the moisture, a partial gain in the performance was observed during the experiment [66]. Electrochemical impedance spectroscopy analyses, conducted under a range of humidification levels (050%), revealed an increase in nonohmic resistance with increases in humidity levels, temperature, or cathodic bias [64]. Cathode delamination led to a decrease in electrochemical performance

108

Solid Oxide Fuel Cell Lifetime and Reliability

mainly due to the strontium oxide/hydroxides segregation at the electrode surface and cathodeelectrolyte interface and loss of effective interfacial contact area [59,67,68]. The competitive adsorption of H2O over O2 molecule on the LSM surface leads to the formation of lanthanum/manganese cation vacancies in the lattice. The metal ion segregates due to the increased cation vacancies and thus SrO formation is accelerated at higher temperatures and water contents [64]. A combined approach using DFT calculations and experiments provides in-depth understanding of the interaction between moisture and lanthanum manganite cathodes. The first-principles thermodynamic approach demonstrated the tendency of the cation segregation in the presence of moisture, where the (La, A)O-terminated (001) surfaces energetically favored the dissociative adsorption of water molecules. This interplay of moisture and surface oxygen vacancies modulates the cationic segregation, which causes degradation in catalytic activity at the surface [69]. Degradation in the Presence of Carbon Dioxide The presence of CO2 in ambient air also adversely affects the cathode performance as CO2 competes with the O2 for ORR [70], thus reacting with the surface SrO and forming SrCO3 [71,72]. CO2-related microstructural as well as electrochemical degradation is observed at B400 ppm CO2 concentrations during long-term operation. Electrochemical performance degradation in the LSM cathode, however, was negligible after initial operational period of 20 h [63]. It was also observed that CO2 does not affect the preactivated LSM cathode performance mainly due to the absence of SrO at the free LSM surface. The reversibility of the LSM performance has been studied in the presence of a combined CO2H2O atmosphere [73]. Performance reversibility in the LSM cathode was observed when exposed to dry air, after initial exposure to H2OCO2 air atmosphere. A decrease in SrO segregation was observed along with an increase in the electrochemical performance. It is postulated that in the presence of dry air atmosphere, SrO nanoparticles can be incorporated back into the LSM lattice thereby mitigating LSM degradation. The interactions of CO2 with the cathode materials were predicted using the CALculation of Phase Diagram (CALPHAD) approach over a wide range of temperatures, and partial pressures of CO2 and O2 [72]. The formation of SrCO3 on the surface and TPBs was predicted under 10 vol.% CO2air atmosphere. Analysis also indicated that the formation of SrCO3 remained more favorable at lower temperatures, high Sr concentration and low PO2 conditions. Interestingly, no secondary phase formation was predicted in the air containing 21 vol.% oxygen with ,400 ppm CO2. Although exposure to CO2 does not change electronic/ionic carriers concentration, the electrical degradation occurs due to the formation of secondary phases such LaZr2O7, SrZrO3, and/or SrCO3 at the LSM/YSZ interface, blocking the ion transfer paths [68,73].

Cathode Degradation From Airborne Contaminants Chapter | 6

109

Degradation in the Presence of Sulfur Dioxide Sulfur dioxide (SO2) is a USEPA designated criteria pollutant and remains in trace amounts in the air. Its presence has not received large attention until recently as a potential contaminant for SOFC electrodes [57]. Experiments conducted at 0.1 ppm SO2 led to negligible voltage drop at 800 C over 1000 h for LSM cathodes [74]. However, upon examining sulfur poisoning in an accelerated mode, experiments have shown that significant sulfur deposits on free electrode surfaces in the form of SrSO4 at 800 C [57]. A detailed study on the sulfur poisoning mechanisms for the LSCF cathode as a function of SO2 concentrations has been reported [75]. The presence of 0.1 ppm SO2 in air showed little or no change in the ohmic and nonohmic polarization resistances. With increase in the SO2 concentration to 1 ppm and above, extensive SrSO4 formation and coverage of the free surface of the cathode was readily observed along with significant increase in the ohmic resistance. Fig. 6.4 schematically shows the surface coverage and formation of SrSO4 along with blocking of the access of oxygen thereby increasing the polarization resistance. Increase in the polarization resistance is also postulated to be due to decrease in the oxygen vacancies [75] concentration near the cathode surface. A more recent study on LSM using atomic resolution scanning transmission electron microscopy describes an early stage of sulfur poisoning showing atomic edges and grain boundaries with strong sulfur signal and the growth of SrSO4 nanoparticles at the grain boundaries due to faster Sr ion transport compared with the bulk material [76].

FIGURE 6.4 Schematic illustration of sulfur poisoning due to surface coverage of SrSO4 on LSCF cathode.

110

Solid Oxide Fuel Cell Lifetime and Reliability

Evaporation of Chromium Impurities under Solid Oxide Fuel Cell System Conditions A number of commercially available nickel- and iron-based austenitic and ferritic alloys have been conventionally used for the fabrication of cell ICs as well as a number of BoP components and subsystems for air handling and thermal management. The above alloys also offer cost effectiveness, high-temperature stability, and ease of fabrication. Long-term oxidation and corrosion resistance is commonly achieved by the formation of protective chromia (Cr2O3) surface scales at high temperatures. When exposed to humidified air atmosphere, these alloys, however, show significant chromium evaporation at SOFC operating temperatures [7779]. The oxidation rates from chromia containing metals and alloys at 800 C and 1000 C are B15% higher when 1% (1000 Pa) water vapor is added in argon/oxygen mixtures when compared with dry air [78,8082]. Formation of gaseous chromium species (in equilibrium with Cr2O3) is presented as follows: 1:5O2 ðgÞ 1 Cr2 O3 ðsÞ 5 2CrO3 ðgÞ

ð6:3Þ

O2 ðgÞ 1 4H2 OðgÞ 1 2Cr2 O3 ðsÞ 5 4CrOðOHÞ2 ðgÞ

ð6:4Þ

1:5O2 ðgÞ 1 2H2 OðgÞ 1 Cr2 O3 ðsÞ 5 2CrO2 ðOHÞ2 ðgÞ

ð6:5Þ

It is noted that all gaseous chromium species are present in Cr61 valence state and remain prone to reduction to lower valence state due to electrochemical reduction as well as thermal reduction during cool down or thermal cycling. CrO2(OH)2 is the most predominant gaseous constituent in the humidified air in 6001000 C temperature range. CrO3 becomes predominant in dry air and with increasing temperature. Chromium evaporation rates from various alumina and chromia forming alloys such Aluchrom YHf, Nicrofer 6025 HT, and AISI 310S were examined under different temperature and moisture levels. Results revealed that Aluchrom alloys showed the lowest chromium evaporation rate followed by Nicrofer 6025. AISI 310S showed the highest chromium evaporation rate with the cracking and spallation of scales [83]. Degradation in the Presence of Chromium Vapor: Chromium Poisoning The reaction process involved in the gaseous chromium-assisted degradation of the cathode is schematically shown in Fig. 6.5. The incoming chromium vapor species have a tendency to reduce at the electrochemically active TPB to form Cr2O3(s). This process reduces the electrode activity and ORR reaction, thereby decreasing the electrochemical activity. Electrochemical studies on LSM/YSZ demonstrate an increase in the cell over-potential attributed to

Cathode Degradation From Airborne Contaminants Chapter | 6

111

FIGURE 6.5 Chromium poisoning mechanism. Chromium vapor at the TPB site reduces to form a solid compound such as Cr2O3 blocking active site for oxygen reduction.

increased charge-transfer resistance and diffusion resistance during chromium poisoning [84]. Sr-enriched surfaces and interfaces, as reported above, serve as nuclei for SrCrO4 formation. The presence of Mn21 in the LSM cathode due to the PO2 gradient under polarization serves as nuclei for MnCr spinel formation. Studies on the LSM cathode at 750 C reveal an overall decrease in the TPB length at the interface. Such changes originate from variations in the interfacial structure and chemistry such as flattening and size reduction of the nano-sized LSM craters and the presence of insulating phases [68]. In an effort to explain the degradation process, experiments at 1000 C and 3% H2O shows the deposition of chromium in the pores across the cathode/electrolyte interface and the diffusion of chromium to the electrode side when the discharge current is interrupted. An increase in polarization was observed due to chromium filling the pores, increased resistance to the diffusion of oxygen gas and a decrease in the electrode reaction sites, which is related to the chromium intensity at the interface [85].

Degradation in “Real world” Air Atmosphere Operation of the SOFC systems under the “real world” air atmosphere accelerates the chromium evaporation from the oxide scales that also lead to rapid electrical performance degradation of the cell and stack due to chromium and sulfur poisoning of the cathode (Fig. 6.6). Combined effects of the gaseous sulfur and chromium contaminants on the SOFC cathode have been studied [86]. Thermodynamic analysis predicts formation of SrSO4 and SrCrO4 phases that can also coexist under cell and stack operating conditions. Conjoint formation of S and Cr containing compounds have been reported to significantly increase the electrode resistance as well as assist in the densification of the porous electrode structure [87] and alteration of the cathode morphology due to Sr depletion [88].

112

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 6.6 Cathode degradation due to the “real world” air atmosphere containing intrinsic and extrinsic contaminants.

Approaches for the Mitigation of Chromium-Assisted Cathode Degradation Chromium poisoning remains a major cause of long-term degradation [87,89]. Several approaches including (1) minimization of chromium evaporation through the use of modified alloy chemistry, (2) surface coatings, (3) chromium getters, and (4) chromium tolerant cathodes have been proposed to minimize and mitigate the long-term performance degradation and poisoning of the cathode. Fig. 6.7 shows the role of alloy chemistry (A) on chromium evaporation, the use of electrically conducting coating (B) on interconnection and the use of chromium getter (C) for chromium capture. Various approaches currently utilized and under investigation are briefly described below. a. Use of alloys and coatings: Commonly practiced materials selection approach includes use of alumina scale forming ferritic and austenitic (iron and nickel based) alloys for the fabrication and assembly of the BoP components. In addition, use of aluminization surface coatings (leading to the formation of alumina scale) has also been considered to reduce Cr evaporation. b. Electronically conducting coating for IC: Electrically conducting CoMn spinel coatings (Fig. 6.7A) have been commonly used to reduce the evaporation of chromium from the alloy surface [26,77,9295]. Long-term stability of coatings and alloys, however, remains a concern due to interdiffusion of alloying constituents, porosity formation at the metal-coating interface and microcracking and fissuring. c. Use of chromium getter: Chromium getters capture gaseous chromium species present in the air stream and significantly lower the equilibrium partial pressure due to the formation of thermochemically stable compounds. Porous ceramic supported getters have shown excellent chromium capture tendency and are currently subject of long-term validation studies [83,96]. Fig. 6.7C shows a fabricated chromium getter using complex oxide compounds, and profile of the captured chromium along

Cathode Degradation From Airborne Contaminants Chapter | 6

113

FIGURE 6.7 Approaches for mitigation of chromium poisoning. (A) Chromium evaporation rates as a function of temperature and water vapor on different alloy materials [26], (B) crosssection SEM image and elemental mapping profile on manganese cobalt spinel coating [90], and (C) digital micrograph of Cr getter and Cr distribution profile after 500 h Cr capture test [91].

the getter length after 500 h Cr capture test. The viability of Cr getters has been examined on the LSM/YSZ/Pt half-cell for 100 h in the presence of chromium vapor in 3% H2O air atmosphere. d. Use of chromium tolerant cathode: This approach relies on the modification of cathode chemistry or the use of alternate cathode formulations that offer resistance to TPB poisoning and undesirable surface and interface compound formation [28,97,98]. In this context, cathodes such as LaNi0.6Fe0.4O3 and La0.6Ba0.4Co0.2Fe0.8O3 have been identified as resistant to chromium poisoning. Infiltrating BaO nanoparticles into LSCF electrode inhibits the formation of SrCrO4 and forms the thermodynamically stable and conducting BaCrO4 compound.

SUMMARY AND OUTLOOK The ability of SOFC components to remain chemically and structurally stable under the complex SOFC exposure environment remains a challenge

114

Solid Oxide Fuel Cell Lifetime and Reliability

during long-term operation under nominal and transient systems operating conditions. Degradation in the cathode performance contributes the most due to increase in the contact resistance between electrode and current collector, mismatches in the thermal expansion between bonded interfaces as well as the formation of undesirable reaction products at the surface (solidgas) and interface (solidsolid). The exposure of perovskite cathodes to “real world” air containing impurities such as H2O, CO2, Cr, and SOx has shown the tendency toward A-site alkaline earth dopant exsolution from the host matrix, morphological changes, and electrocatalytically inactive and insulating compound formation. Chromium poisoning of the cathode is currently considered to be one of the major challenges affecting the performance. Increase in the polarization resistance, decrease in the diffusion of oxygen gas and electrode reaction sites have been reported for cathode under the influence of chromium vapor. Modifications of alloy chemistry and surface coatings such as the aluminizing of BoP components and CoMn spinel coatings for IC alloys have been widely studied and used in the state of the art SOFC systems. Although these approaches have been able to minimize chromium evaporation, alloy modifications and long-term reliability of coatings remain a concern. Cost effective methods to mitigate chromium poisoning are topics of interest. Nonnoble and nonstrategic materials containing chromium getters have been found effective for the capture of chromium from the cell IC and BoP components. Further development of getter materials and design remains a topic of interest for combined capture of Cr and S containing gaseous species present in air.

ACKNOWLEDGEMENT Authors sincerely acknowledge financial support from USDOE (Office of Fossil Energy Grant # DE-FE-0023385). The authors also thank graduate students and technical staff at the Center for Clean Energy Engineering for their help with experiments and characterization.

REFERENCES [1] Stambouli AB, Traversa E. Solid oxide fuel cells (SOFCs): a review of an environmentally clean and efficient source of energy. Renew Sustain Energy Rev 2002;6:43355. [2] Taroco HA, Santos J, Domingues R, Matencio T. Advances in ceramics—synthesis and characterization, processing and specific applications. InTech; 2011. [3] Shaikh SPS, Muchtar A, Somalu MR. A review on the selection of anode materials for solid-oxide fuel cells. Renew Sustain Energy Rev 2015;51:18. [4] Choudhury A, Chandra H, Arora A. Application of solid oxide fuel cell technology for power generation—a review. Renew Sustain Energy Rev 2013;20:43042. [5] Hawkes A, Staffell I, Brett D, Brandon N. Fuel cells for micro-combined heat and power generation. Energy Environ Sci 2009;2:72944.

Cathode Degradation From Airborne Contaminants Chapter | 6

115

[6] Wachsman ED, Marlowe CA, Lee KT. Role of solid oxide fuel cells in a balanced energy strategy. Energy Environ Sci 2012;5:5498509. [7] Sanchez R. Technology readiness assessment guide—DOE directives, delegations, and requirements. Washington, DC: 2011. [8] O’Hayre R, Barnett DM, Prinz FB. The triple phase boundary. J Electrochem Soc 2005;152:A439. [9] Richter J, Holtappels P, Graule T, Nakamura T, Gauckler LJ. Materials design for perovskite SOFC cathodes. Monatsh Chem 2009;140:985. [10] Irshad M, Siraj K, Raza R, Ali A, Tiwari P, Zhu B, et al. A brief description of high temperature solid oxide fuel cell’s operation, materials, design, fabrication technologies and performance. Appl Sci 2016;6:75. [11] Adler SB. Chemical expansivity of electrochemical ceramics. J Am Ceram Soc 2004;84:211719. [12] Kharton VV, Kovalevsky AV, Avdeev M, Tsipis EV, Patrakeev MV, Yaremchenko AA, et al. Chemically induced expansion of La2NiO4 1 δ-based materials. Chem Mater 2007;19:202733. [13] Fleig J. On the width of the electrochemically active region in mixed conducting solid oxide fuel cell cathodes. J Power Sources 2002;105:22838. [14] Brandon NP, Skinner S, Steele BCH. Recent advances in materials for fuel cells. Annu Rev Mater Res 2003;33:183213. [15] Singh P, Minh NQ. Solid oxide fuel cells: technology status. Int J Appl Ceram Technol 2005;1:515. [16] Steele BC, Heinzel A. Materials for fuel-cell technologies. Nature 2001;414:34552. [17] Minh N. Solid oxide fuel cell technology-features and applications. Solid State Ionics 2004;174:2717. [18] Dicks A. Advances in catalysts for internal reforming in high temperature fuel cells. J Power Sources 1998;71:11122. [19] Blum L, Batfalsky P, Fang Q, de Haart LGJ, Malzbender J, Margaritis N, et al. SOFC stack and system development at Forschungszentrum Ju¨lich. J Electrochem Soc 2015;162: F1199205. [20] Gerdes K., Richards G. Fuel cells. www.netl.doe.gov (2012). https://www.netl.doe.gov/ File.Library/Research/onsite research/research capabilities/R-D182.pdf. [21] Yokokawa H, Tu H, Iwanschitz B, Mai A. Fundamental mechanisms limiting solid oxide fuel cell durability. J Power Sources 2008;182:40012. [22] Tietz F. Thermal expansion of SOFC materials. Ionics (Kiel) 1999;5:12939. [23] Keqin H, Robin T, Goodenough JB, Milliken C. Superior perovskite oxide-ion conductor; strontium- and magnesium-doped LaGaO3. III. Performance tests of single ceramic fuel cells. J Am Ceram Soc 1998;81:25815. [24] Keane M, Mahapatra MK, Verma A, Singh P. LSMYSZ interactions and anode delamination in solid oxide electrolysis cells. Int J Hydrogen Energy 2012;37:1677685. [25] Schuler JA, Gehrig C, Wuillemin Z, Schuler AJ, Wochele J, Ludwig C, et al. Air side contamination in solid oxide fuel cell stack testing. J Power Sources 2011;196:722531. [26] Ge L, Verma A, Goettler R, Lovett D, Raman RKS, Singh P. Oxide scale morphology and chromium evaporation characteristics of alloys for balance of plant applications in solid oxide fuel cells. Metall Mater Trans A 2013;44:193206. [27] Zeng Z. Corrosion of metallic interconnects for SOFC in fuel gases. Solid State Ionics 2004;167:916.

116

Solid Oxide Fuel Cell Lifetime and Reliability

[28] Jiang SP, Chen X. Chromium deposition and poisoning of cathodes of solid oxide fuel cells  a review. Int J Hydrogen Energy 2014;39:50531. [29] Tietz F, Mai A, Sto¨ver D. From powder properties to fuel cell performance  a holistic approach for SOFC cathode development. Solid State Ionics 2008;179:150915. [30] Yokokawa H. Towards comprehensive description of stack durability/reliability behavior. Fuel Cells 2015;15:65268. [31] Standaert F, Hemmes K, Woudstra N. Analytical fuel cell modeling; non-isothermal fuel cells. J Power Sources 1998;70:18199. [32] Yokokawa H. 10-Year cooperative investigations on durability/reliability of stationary SOFCs among industries, research institutes and universities within NEDO SOFC projects. Proceedings of the 11th international conference ceramic materials components energy environmental application, Japan; 2014. [33] Hsiao Y, Selman J. The degradation of SOFC electrodes. Solid State Ionics 1997;98:338. [34] Nguyen BN, Koeppel BJ, Ahzi S, Khaleel MA, Singh P. Crack growth in solid oxide fuel cell materials: from discrete to continuum damage modeling. J Am Ceram Soc 2006;89:135868. [35] Boukamp BA. Fuel cells: the amazing perovskite anode. Nat Mater 2003;2:2946. [36] Sun C, Hui R, Roller J. Cathode materials for solid oxide fuel cells: a review. J Solid State Electrochem 2009;14:112544. [37] Adler SB. Factors governing oxygen reduction in solid oxide fuel cell cathodes. Chem Rev 2004;104:4791843. [38] Pen˜a MA, Fierro JLG. Chemical structures and performance of perovskite oxides. Chem Rev 2001;101:19812018. [39] Antonietti M, Ozin GA. Promises and problems of mesoscale materials chemistry or why meso? Chem-A Eur J 2004;10:2841. [40] Clausen C, Bagger C, Bilde-Sørensen JB, Horsewell A. Microstructural and microchemical characterization of the interface between La0.85Sr0.15MnO3 and Y2O3-stabilized ZrO2. Solid State Ionics 1994;70:5964. [41] Ralph JM, Schoeler AC, Krumpelt M. Materials for lower temperature solid oxide fuel cells. J Mater Sci 2001;36:116172. [42] Ralph JM, Rossignol C, Kumar R. Cathode materials for reduced-temperature SOFCs. J Electrochem Soc 2003;150:A1518. [43] Yokokawa H, Sakai N, Horita T, Yamaji K, Brito ME, Kishimoto H. Thermodynamic and kinetic considerations on degradations in solid oxide fuel cell cathodes. J Alloys Compd 2008;452:417. [44] Anderson MD, Stevenson JW, Simner SP. Reactivity of lanthanide ferrite SOFC cathodes with YSZ electrolyte. J Power Sources 2004;129:18892. [45] Armstrong EN, Duncan KL, Oh DJ, Weaver JF, Wachsman ED. Determination of surface exchange coefficients of LSM, LSCF, YSZ, GDC constituent materials in composite SOFC Cathodes. J Electrochem Soc 2011;158:B492. [46] Wachsman ED, Lee KT. Lowering the temperature of solid oxide fuel cells. Science 2011;334:9359. [47] Kojima I, Adachi H, Yasumori I. Electronic structures of the LaBO3 (B 5 Co, Fe, Al) perovskite oxides related to their catalysis. Surf Sci 1983;130:5062. [48] Petrov AN, Kononchuk OF, Andreev AV, Cherepanov VA, Kofstad P. Crystal structure, electrical and magnetic properties of La1 2 xSrxCoO3 2 y. Solid State Ionics 1995;80:18999.

Cathode Degradation From Airborne Contaminants Chapter | 6

117

[49] Lau S, Singhal SC. Potential electrode/electrolyte interactions in solid oxide fuel cells. United States Natl Assoc Corros Eng 1985;20:70. [50] Yokokawa H. Understanding materials compatibility. Annu Rev Mater Res 2003;33:581610. [51] Virkar AV, Nachlas J, Joshi AV, Diamond J. Internal precipitation of molecular oxygen and electromechanical failure of zirconia solid electrolytes. J Am Ceram Soc 1990;73:338290. [52] Singh P., Ruka R.J. Oxide modified air electrode surface for high temperature electrochemical cells. US5106706 A, 1992. [53] Mai A, Haanappel VAC, Uhlenbruck S, Tietz F, Sto¨ver D. Ferrite-based perovskites as cathode materials for anode-supported solid oxide fuel cells: Part I. Variation of composition. Solid State Ionics 2005;176:134150. [54] Yang M, Yan A, Zhang M, Hou Z, Dong Y, Cheng M. Effects of GDC interlayer on performance of low-temperature SOFCs. J Power Sources 2008;175:34552. [55] Li N, Keane M, Mahapatra MK, Singh P. Mitigation of the delamination of LSM anode in solid oxide electrolysis cells using manganese-modified YSZ. Int J Hydrogen Energy 2013;38:6298303. [56] Knibbe R, Traulsen ML, Hauch A, Ebbesen SD, Mogensen M. Solid oxide electrolysis cells: degradation at high current densities. J Electrochem Soc 2010;157:B1209. [57] Xiong Y, Yamaji K, Horita T, Yokokawa H, Akikusa J, Eto H, et al. Sulfur poisoning of SOFC cathodes. J Electrochem Soc 2009;156:B588. [58] Hu B, Krishnan S, Liang C, Heo JS, Aphale AN, Ramprasad R, et al. Experimental and thermodynamic evaluation of La1  xSrxMnO3 6 δ and La1  xSrxCo1  yFeyO3  δ cathodes in Cr-containing humidified air. Int J Hydrogen Energy 2017; (In Press). http://dx. doi.org/10.1016/j.ijhydene.2017.01.040 [59] Fergus JW. Effect of cathode and electrolyte transport properties on chromium poisoning in solid oxide fuel cells. Int J Hydrogen Energy 2007;32:366471. [60] Arregui A, Rodriguez-Martinez LM, Modena S, Bertoldi M, van Herle J, Sglavo VM. Stability of ferritic perovskite cathodes in anode-supported solid oxide fuel cells under different processing and operation parameters. Electrochim Acta 2011;58:31221. [61] U.S. Environmental Protection Agency, Summary of the Clean Air Act (2016), (available at https://www.epa.gov/laws-regulations/summary-clean-air-act). US EPA O. Summary of the Clean Air Act. [62] US EPA, OAR, O. of A. Q. P. and S. (2015). National Ambient Air Quality Standards (NAAQS) | Technology Transfer Network | US EPA. Retrieved from http://www3.epa. gov/ttn/naaqs/criteria.html. [63] Hu B, Mahapatra MK, Keane M, Zhang H, Singh P. Effect of CO2 on the stability of strontium doped lanthanum manganite cathode. J Power Sources 2014;268:40413. [64] Hu B, Keane M, Mahapatra MK, Singh P. Stability of strontium-doped lanthanum manganite cathode in humidified air. J Power Sources 2014;248:196204. [65] Liu RR, Kim SH, Taniguchi S, Oshima T, Shiratori Y, Ito K, et al. Influence of water vapor on long-term performance and accelerated degradation of solid oxide fuel cell cathodes. J Power Sources 2011;196:70906. [66] Nielsen J, Mogensen M, Sofc LSM. YSZ cathode degradation induced by moisture: an impedance spectroscopy study. Solid State Ionics 2011;189:7481. [67] Nielsen J, Hagen A, Liu YL. Effect of cathode gas humidification on performance and durability of solid oxide fuel cells. Solid State Ionics 2010;181:51724.

118

Solid Oxide Fuel Cell Lifetime and Reliability

[68] Liu YL, Hagen A, Barfod R, Chen M, Wang HJ, Poulsen FW, et al. Microstructural studies on degradation of interface between LSMYSZ cathode and YSZ electrolyte in SOFCs. Solid State Ionics 2009;180:1298304. [69] Sharma V, Mahapatra MK, Krishnan S, Thatcher Z, Huey BD, Singh P, et al. Effects of moisture on (La, A)MnO3 (A 5 Ca, Sr, and Ba) solid oxide fuel cell cathodes: a first-principles and experimental study. J Mater Chem A 2016;4:560515. [70] Zhao Z, Liu L, Zhang X, Wu W, Tu B, Ou D, et al. A comparison on effects of CO2 on La0.8Sr0.2MnO31δ and La0.6Sr0.4CoO32δ cathodes. J Power Sources 2013;222:54253. [71] Ponce S, Pen˜a MA, Fierro JLG. Surface properties and catalytic performance in methane combustion of Sr-substituted lanthanum manganites. Appl Catal B Environ 2000;24:193205. [72] Darvish S, Asadikiya M, Hu B, Singh P, Zhong Y. Thermodynamic prediction of the effect of CO2 to the stability of (La0.8Sr0.2)0.98MnO3 6 δ system. Int J Hydrogen Energy 2016;41:1023948. [73] Hu B, Mahapatra MK, Singh P. Performance regeneration in lanthanum strontium manganite cathode during exposure to H2O and CO2 containing ambient air atmospheres. J Ceram Soc Japan 2015;123:199204. [74] Liu RR, Kim SH, Shiratori Y, Oshima T, Ito K, Sasaki K. The influence of water vapor and SO2 on the durability of solid oxide fuel cell. ECS Trans. 2009;vol. 25:285966 ECS [75] Wang F, Yamaji K, Cho D-H, Shimonosono T, Kishimoto H, Brito ME, et al. Sulfur poisoning on La0.6Sr0.4Co0.2Fe0.8O3 cathode for SOFCs. J Electrochem Soc 2011;158:B1391. [76] Daio T, Mitra P, Lyth SM, Sasaki K. Atomic-resolution analysis of degradation phenomena in SOFCS: a case study of SO2 poisoning in LSM cathodes. Int J Hydrogen Energy 2016;41:1221421. [77] Collins C, Lucas J, Buchanan TL, Kopczyk M, Kayani A, Gannon PE, et al. Chromium volatility of coated and uncoated steel interconnects for SOFCs. Surf Coat Technol 2006;201:446770. [78] Saunders SRJ, Monteiro M, Rizzo F. The oxidation behaviour of metals and alloys at high temperatures in atmospheres containing water vapour: a review. Prog Mater Sci 2008;53:775837. [79] Gindorf C, Singheiser L, Hilpert K. Chromium vaporisation from Fe, Cr base alloys used as interconnect in fuel cells. Steel Res 2001;72:52833. [80] Jacob YP, Haanappel VAC, Stroosnijder MF, Buscail H, Fielitz P, Borchardt G. The effect of gas composition on the isothermal oxidation behaviour of PM chromium. Corros Sci 2002;44:202739. ˙ J, Michalik M, Schmitz F, Kern T-U, Singheiser L, Quadakkers WJ. The effect of [81] Zurek water-vapor content and gas flow rate on the oxidation mechanism of a 10%Cr-ferritic steel in ArH2O mixtures. Oxid Met 2005;63:40122. [82] Fujii CT, Meussner RA. The mechanism of the high-temperature oxidation of iron-chromium alloys in water vapor. J Electrochem Soc 1964;111:1215. [83] Singh P., Liang C., Hu B., Mahapatra M.K., Jun B. Chromium poisoning in high temperature (6001000 C) electrochemical systems. TMS 2016 145th annual meeting & exhibition, Nashville, Tennessee; 2016. [84] Matsuzaki Y, Yasuda I. Electrochemical properties of a SOFC cathode in contact with a chromium-containing alloy separator. Solid State Ionics 2000;132:2718. [85] Taniguchi S, Kadowaki M, Kawamura H, Yasuo T, Akiyama Y, Miyake Y, et al. Degradation phenomena in the cathode of a solid oxide fuel cell with an alloy separator. J Power Sources 1995;55:739.

Cathode Degradation From Airborne Contaminants Chapter | 6

119

[86] Schuler JA, Yokokawa H, Calderone CF, Jeangros Q, Wuillemin Z, Hessler-Wyser A, et al. Combined Cr and S poisoning in solid oxide fuel cell cathodes. J Power Sources 2012;201:11220. [87] Yokokawa H, Horita T, Sakai N, Yamaji K, Brito M, Xiong Y, et al. Thermodynamic considerations on Cr poisoning in SOFC cathodes. Solid State Ionics 2006;177:31938. [88] Mori M, Hiei Y, Sammes NM. Sintering behavior of Ca- or Sr-doped LaCrO3 perovskites including second phase of AECrO4 (AE 5 Sr, Ca) in air. Solid State Ionics 2000;135:7438. [89] Yang M, Bucher E, Sitte W. Effects of chromium poisoning on the long-term oxygen exchange kinetics of the solid oxide fuel cell cathode materials La0.6Sr0.4CoO3 and Nd2NiO4. J Power Sources 2011;196:731317. [90] Ge L. Chromium evaporation of metallic component materials in solid oxide fuel cell (SOFC). Doctoral Dissertations 2014;329 http://digitalcommons.uconn.edu/dissertations/ 329 [91] Liang C., Hu B., Venkataraman M.B., Aphale A.N., Manthina V., Mahapatra M.K., et al. Design and fabrication of getter for mitigation of chromium poisoning in SOFC power systems. Unpublished work. [92] Lee S-I, Hong J, Kim H, Son J-W, Lee J-H, Kim B-K, et al. Highly dense MnCo spinel coating for protection of metallic interconnect of solid oxide fuel cells. J Electrochem Soc 2014;161:F138994. [93] Choi JP, Scott Weil K, Matt Chou Y, Stevenson JW, Gary Yang Z. Development of MnCoO coating with new aluminizing process for planar SOFC stacks. Int J Hydrogen Energy 2011;36:454956. [94] Yang Z., Xia G., Wang C., Nie Z., Templeton J., Stevenson J., et al. Investigation of AISI 441 ferritic stainless steel and development of spinel coatings for SOFC interconnect applications. Richland, WA; 2008. [95] Yang Z, Xia G, Stevenson JW. Mn1.5Co1.5O4 spinel protection layers on ferritic stainless steels for SOFC interconnect applications. Electrochem Solid-State Lett 2005;8:A16870. [96] Liang C, Hu B, Aphale A, Rodriguez W, Uddin MA, Singh P. Mitigation of chromium assisted degradation of LSM cathode in SOFC. Meet Abstr 2016; MA201602:12321232. [97] Zhen YD, Tok AIY, Jiang SP, Boey FYC. La(Ni, Fe)O3 as a cathode material with high tolerance to chromium poisoning for solid oxide fuel cells. J Power Sources 2007;170:616. [98] Chen K, Ai N, O’Donnell KM, Jiang SP. Highly chromium contaminant tolerant BaO infiltrated La0.6Sr0.4Co0.2Fe0.8O3 2 δ cathodes for solid oxide fuel cells. Phys Chem Chem Phys 2015;17:48704.

Chapter 7

Lifetime Issues for Solid Oxide Fuel Cell Interconnects Manuel Bianco1, Markus Linder2, Yngve Larring3, Fabio Greco1 and Jan Van herle1 1

Swiss Federal Institute of Technology in Lausanne Valais, Sion, Switzerland, 2Zurich University of Applied Sciences, Winterthur, Switzerland, 3SINTEF Materials and Chemistry, Oslo, Norway

Chapter Outline Introduction Metal Interconnects Degradation Solutions to Decrease IC Degradation

121 122 124 132

Lifetime Behavior of Stacks and Cells Tested in Operating Conditions Conclusion Acknowledgements References

136 141 141 141

INTRODUCTION A single solid oxide fuel cell (SOFC) generates 1 V at open circuit voltage conditions; to generate useful power it is necessary to stack cells through interconnects (ICs), which must collect current, keep gas flows separated and provide additional mechanically stability to the stack. According to the stack geometry, tubular, or planar, the IC can assume different shapes: long and narrow in the first case while wide and thin for the latter. The material class varies: metal is preferred for intermediate temperature SOFC (600800 C), while a ceramic is employed for the highest continuous operating temperatures .900 C. However, due to higher manufacturing complexity and involved cost, R&D activities in fact focus on metallic ICs even for the higher operating temperatures. Regardless of the geometry or the material, an SOFC IC must accomplish different tasks: divide cathode (air) and anode (fuel) streams (gas impermeability), distribute gas flows to optimize fuel utilization (easy machining, stamping, etc.), collect electrical current generated by the cell (low electrical contact resistivity ,10 mΩ  cm2), guarantee mechanical stability (creep resistance),

Solid Oxide Fuel Cell Lifetime and Reliability. DOI: http://dx.doi.org/10.1016/B978-0-08-101102-7.00007-6 Copyright © 2017 Elsevier Ltd. All rights reserved.

121

122

Solid Oxide Fuel Cell Lifetime and Reliability

show sufficient chemical stability and inertia with respect to the other components (low corrosion), avoid residual stress potentially damaging to the ceramic cell (coefficient of thermal expansion (CTE) compatible with the cell 1213 3 1026 K21 [1]), and transfer heat from the cathode to the anode in case of fuel reforming (good thermal conductivity .5 W m21 K21 [2]). Balancing all criteria ultimately leaves a limited choice of potential materials: LaCrO3-based perovskites [3], Cr-based alloy, and ferritic stainless steels (FSSs) [4], from the higher to the lower operating temperatures, respectively. Yet, no material completely fulfills all the features requested. In addition, the SOFC requirements for stationary applications—at least 40,000 h at operating temperature—lead to IC lifetime issues, such as mechanical deformation, corrosion, surface spallation, which impact the stack lifetime. To mitigate such detrimental processes, metal ICs (MICs) have very specific chemical compositions and are combined with protective oxide coatings like a spinel or a perovskite layer deposited by different methods. The IC degradation mechanisms provide an important contribution to the overall stack degradation, especially after prolonged stack operation times (.10,000 h). This is in turn due to improvements in the stability of the electrodes, and the fact that the major part of electrolyte degradation (a reduction in ionic conductivity) occurs within the first 4000 h [5]. This chapter will first introduce theoretical aspects on the physical phenomena causing lifetime issues in SOFC ICs. The second part presents, through a compilation table, the main results on the behavior of MICs tested for extended periods in real stacks, in both industrial and academic environments.

METAL INTERCONNECTS The two main types of alloy used today in SOFC technology are FSS and Cr-alloy e.g. Chromium-Iron-Yttrium (CFY). Stainless steels are chosen because of their ability to grow a passivating oxide layer on the surface, preventing the formation of brittle and low electrical conductivity hydrated iron oxide phases. The passivating film could be made of chromium and/or aluminum oxides. Both oxides are effective in slowing the corrosion kinetics down, but the electrical insulating properties of aluminum oxide (σAl2 O3 B 10210 S cm21 at 700 C [6]) leaves chromium oxide forming steel (σCr2 O3 B 1021 S cm21 at 750800 C [7,2]) as the best choice for an SOFC IC [8] (the conductivity values are not absolute because at the IC operating temperature the oxides are extrinsic semiconductors). The chromium content needed to ensure a continuous and homogeneous passivation layer is 1819 wt.%, but, considering the risk of Cr breakaway oxidation (depletion of the protecting element), 2225 wt.% is considered a safer Cr content for the IC [4,8]. Following this, only austenitic stainless steel or

Lifetime Issues for Solid Oxide Fuel Cell Interconnects Chapter | 7

123

FSS can be used in an SOFC stack, as martensitic stainless steel contains an insufficient amount of chromium. The γ-austenitic phase is face-centered cubic while the α-ferritic phase is body-centered cubic. This difference causes different CTEs: 1820 3 1026 K21 for the γ-phase vs. 1114 3 1026 K21 for the α-phase, in the SOFC temperature range of 25800 C [8], the FSS therefore being more compatible with the ceramic cell CTE (10.512.5 3 1026 K21 at 30800 C [1]). On the other hand, FSS has worse creep resistance and high-temperature mechanical strength than austenitic steel. High Cr alloy also creates a protective passivating surface layer, and its CTE value is close to that of the cathode (for CFY produced by Plansee: 8.910.5 3 1026 K21 at 300800 C [9]). Different alloy compositions have been tested and improved in order to optimize for the property criteria mentioned in the introduction. The current state-of-the-art is specialized SOFC alloys: Crofer 22 APU or H, Sandvik Sanergy HT, Plansee ITM, and CFY. In addition, commercial FSS grade AISI 441/K41 is also a widespread, SOFC nonspecific alternative due to its lower cost, while it has lower corrosion resistance and worse mechanical properties. The alloy compositions are given in Table 7.1.

TABLE 7.1 Elemental Compositions of Metal Interconnect Substrates (Weight Percentage) K41/AISI 441 [10]

Sanergy HT [10]

Crofer 22 APU [11]

Crofer 22H [10]

CFY [9]

IT-11 or ITM [12]

Fe

Bal.

Bal.

Bal.

Bal.

5.0

71.80

Cr

18.0

21.2

20.024.0

20.024.0

, 95.0

26.40

Y









, 1.0

0.08

C

0.012

0.04

0.00.03

0.00.03



0.009

Mn

0.30

0.30

0.300.80

0.00.80





Si

0.35

0.12

0.5

0.10.60



0.01

Al



0.017

0.5

0.00.10



0.02

Mo



0.96









Nb

0.45

0.71



0.21.0





Ti

0.17

0.09

0.030.20

0.020.20





W







1.03.0





124

Solid Oxide Fuel Cell Lifetime and Reliability

Degradation For the SOFC IC, corrosion and mechanical deformation are the most pronounced degradation phenomena. The former increases the electrical resistivity of the IC because of the poorer electrical conductivity of the thermally grown oxide layer, while the latter deforms the IC shape, potentially reducing the contact area with the cell.

Atmospheric Corrosion Among the different definitions of corrosion, the one by L.L. Shreir is taken into consideration [13]: “the reaction of an engineering constructional metal (material) with its environment with a consequent deterioration in properties of the metal (material).” At microscopic level, corrosion is a two-step nucleation and growth process: oxidative species are adsorbed and react with the metal surface, starting oxide nucleation followed by two-dimensional lateral growth and finally a three-dimensional one. When the oxide layer is compact and continuous, growth becomes a diffusion-controlled phenomenon. Based on the hypothesis of a diffusion-controlled process, Wagner developed a model derived from Fick’s law [14]. The scale thickness growth (Δxox Þ with time is proportional to a kinetic constant and follows a parabolic law: pffiffiffiffiffiffiffi Δxox 5 2 Kp t ð7:1Þ with Kp the corrosion or parabolic rate constant, explicitly defined as: ð pvO 2 Kp 5 k DM d ln pO2 p0O

ð7:2Þ

2

where k is a constant, DM is the diffusion coefficient for the dominant diffusive species, p0O2 is the oxygen partial pressure at the metal/oxide interface, and pvO2 is the oxygen partial pressure on top of the scale (for a detailed approach, cf. [1416]). Fig. 7.1 gives an example for Cr2O3 scale growth on CFY IC cross sections at different time steps operated in SOFC stacks

FIGURE 7.1 Oxide scale growth on CFY ICs at the cathode side operated with air in Hexis SOFC stacks at 900 C with catalytically partially oxidized (CPOx) reformed natural gas at the anode.

Lifetime Issues for Solid Oxide Fuel Cell Interconnects Chapter | 7

125

up to the first target lifetime of 40,000 h. The obtained scale growth for the investigated samples exhibits, more or less coincidentally, a parabolic behavior [17]. Correlation in Eq. (7.1) is a priori interesting because with a known Kp value obtained after a short experimental period of 1000 h, it is possible to predict the oxide layer thickness for longer periods, e.g., years (.10,000 h). Such a minimal observation time is necessary because within the first few hundred hours of testing various interfering oxide formation mechanisms are involved in scale formation, which complicate a reliable extrapolation [18]. In addition, considering a simple model where the resistivity of the MIC is directly proportional to the scale thickness, resistivity forecasting is possible too. In reality, this basic approach is oversimplifying, as shown in Fig. 7.2, where the fitted exponent n significantly deviates from purely parabolic behavior (n 5 0.5). Therefore other parameters must be taken into account, such as scale morphology, oxide scale composition (always containing various impurities), and transient operating conditions (e.g., redox and thermal cycles). Comparing the evolution of the area-specific resistance (ASR), from conductivity measurements, with the evolution of the Cr2O3 scale thickness, the trends are different and nonlinear, with a much more pronounced variation at the beginning of exposure, i.e., for observation times ,5000 h. This behavior is related to the morphology of thermally grown oxide scales between the coating and the metallic substrate (on the cathode side). According to Ohm’s law the electrical current follows the paths of least resistance, which is necessarily through the bridges where the oxide scale is thinnest. This kind of current bridges can lead to a significant reduction of ASR, considering that the electrical conductivity of the coatings and the IC is orders of magnitude

FIGURE 7.2 ASR trend obtained from a La1-xSrxMnO3 (LSM)-coated CFY sample in air at 850 C. Reprinted with permission from 11th European SOFC and SOE Forum Proceedings, 2014.

126

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 7.3 Boxplot of the morphology effect obtained from a large amount of SEM image cross sections from heat exposed CFY samples in air at 850 C. The inset shows the timedependent behavior of the M factor. Reprinted with permission from 11th European SOFC and SOE forum proceedings 2014.

higher than the formed Cr2O3. This provides local transversal electrical current pathways that amplify the bridging effect. Nevertheless, the impact of this bridging effect is time-dependent, in other words depending on the overall scale thickness. With increasing scale thickness the effect becomes less relevant, since the influence of the morphology effect, related to the measured ASR compared with the ASR predicted based on a uniform featureless oxide layer, decreases. Fig. 7.3 shows the decreasing trend from the morphology effect derived from scanning electron microscope (SEM) images of CFY samples, where the morphology effect tends asymptotically to 1 (at this point morphological effects are nonexistent [19]). The morphology factor MðtÞ is quantified based on a comprehensive amount of SEM images. For that purpose the resulting ASRx ðtÞ, derived from the mean oxide scale thickness, is set in relation to the calculated ASRsim ðtÞ, determined from finite element simulation: MðtÞ 5

ASRx ðtÞ ASRsim ðtÞ

ð7:3Þ

The calculation mesh used for the finite element computation is automatically generated on the basis of SEM image contrast. Fig. 7.4 compares the ASR prediction based on oxide scale formation xðtÞ with and without considering the morphology factor. Taking the morphology factor into account gives a significantly better fit with the experimentally determined ASR data, thus a more reliable degradation prediction based on oxide scale formation. Yet, even for M 5 1 the relation between scale thickness and resistivity is not straightforward: impurities and interaction with adjacent components, such as current pick-up mesh and coatings, influence the electrical

Lifetime Issues for Solid Oxide Fuel Cell Interconnects Chapter | 7

127

FIGURE 7.4 ASR prediction with and without morphology factor. Diamonds represent experimental ASR data from LSM-coated CFY sample heat exposed in air at 850 C. Reprinted with permission from M. Linder, T. Hocker, L. Holzer, K.A. Friedrich, B. Iwanschitz, A. Mai, et al., Model-based prediction of the ohmic resistance of metallic interconnects from oxide scale growth based on scanning electron microscopy, J Power Sources 2014;272:595605.

conductivity of the Cr2O3 scale. Impurities dissolved in thermally grown Cr2O3 scales originate from alloying and/or the stack manufacturing process. This may improve the electrical conductivity of the formed oxide scale. Interaction with adjacent components can, e.g., lead to the desirable formation of spinel phases like (Cr,Mn)3O4 or NiCr2O4 [20,21]. Given this complex interplay it is obvious why sufficient testing time is essential for the reliable prediction of ASR trends and degradation, respectively [19]. Considering the alloys described in Table 7.1, for Eq. (7.2), the majority diffusive species might be Cr31 or O22 ions, or their vacancies, with interstitial Cr-cations commonly accepted as the main diffusive species and moving preferably along grain boundaries in the temperature range in which SOFC ICs operate [15,16]. On the effect of oxygen partial pressure on oxidation rate instead, there is no overall consensus. Various experiments demonstrated the oxidation rate to be independent of pO2 [15,22,23], while others report an interdependence [16]. Referring to Eq. 7.2, the most agreed idea is that below 1000 C the major diffusing species is interstitial Cr which is related to the oxygen partial pressure at the metal/scale interface, that in turn can be taken as constant for a given temperature. Therefore Kp should be independent of pO2 [23]. However, the explanation for these different behaviors is not final. Water partial pressure influences the growth kinetics of the corrosion process. In general, Kp is found to increase in the presence of humidity both in anodic and cathodic conditions. Faster diffusion of ion hydroxide OH2 com˚ vs. 1.4 A ˚ ionic radius) through the scale is the prevailing pared to O22 (0.95 A

128

Solid Oxide Fuel Cell Lifetime and Reliability

accepted explanation for this [24]. At the same time, water in the gas flows reacts with Cr2O3 producing at SOFC operating conditions the gaseous species CrO2(OH)2. This process in turn depletes Cr from the substrate and thins the scale. The predominance of one of the two concurrent processes is influenced by the deposition of a coating or by the alloy composition: for example, Mn migrates toward the alloy surface in the first stages of scale formation, forming a superficial (Mn,Cr)3O4 spinel layer that decreases Cr evaporation. Water vapor influences the structural scale properties too: porosity in the thermally grown oxide layer is found to be more homogenously distributed through the thickness, compared to a scale grown in a dry environment, where the pores are instead concentrated at the alloy/scale interface [24]. The aforementioned studies refer to samples studied in a single atmosphere. In a real stack the ICs work in a dual atmosphere. Experiments conducted with FSS alloys placed between reducing and oxidizing environments found an increase in Fe concentration in the thermally grown oxide layer at the air side, attributed to the proton migration from fuel to air side [25]. This behavior could be an important drawback for SOFCs working at temperatures around 550600 C, as it has been demonstrated that AISI 441 exposed to dual atmosphere at 600 C suffered severe breakaway corrosion, where the rate of corrosion rapidly increases due to break up of the protective Cr2O3 coating [26]. Different explanations have been proposed, e.g., that hydrogen would increase internal oxidation reducing chromium supply and in turn causing breakaway corrosion, but no theory is universally accepted for the moment. Note that most of the studies cited above used uncoated substrates, in order to accelerate the degradation kinetics and decouple the influence of a coating.

Sealing Corrosion In anode-supported and electrolyte-supported SOFC, glass sealants are used to ensure the gas tightness between the cell and the ICs in the stacks. Bariumcalciumaluminosilicate (BCAS)-based glasses are commonly used [27] for this purpose. To tune the glass properties, minor quantities of elements like Pb, V, B, Mg are added. Their presence might cause the acceleration of corrosion kinetics at the glass/steel interface. In the stack failure described in [28], for example, a stack experienced electrical short circuiting between metal plates after only a few hundred hours of operation at 800 C. The post-test characterization correlated the failure to the presence of iron oxide nodules at the cathode side. These bumps grew rapidly and electrically connected the ICs; the electrical conductivity was further enhanced by the presence of metallic iron inside the nodules. At the anode side instead, internal oxidation along grain boundaries in the steel microstructure lead to both material cracking and bulging of the steel surface of the order of tens of μm.

Lifetime Issues for Solid Oxide Fuel Cell Interconnects Chapter | 7

129

In both cases the phenomena appeared only at the triple-phase boundary metal/glass-sealing/gas (air or fuel). Another study helped to understand the corrosion process [27]: reduction of PbO (contained in the glass) frees Pb that reacts with the steel surface and scavenges chromium, increasing the oxidation kinetics. Fast Cr2O3 growth continues until there is not enough Cr left and breakaway corrosion occurs. Eventually, iron oxides grow and form an electrical conductive pathway leading to a short circuit. Besides Pb, Ba is an active element toward the Cr contained in the steel, leading to the formation of BaCrO4; however, posttest analysis did not show a role for this compound in the stacks failure (cf., Table 7.2). A picture of the yellowish barium chromate at the sealing/air interface is given in Fig. 7.5A and B. It is taken from a recent study where the influence of polarization on the MIC/sealant interface degradation is analyzed [29]: three different stacks, made of five FSS plates joined together by BaSi-based glass were polarized at three different voltage potentials of 0, 1, and 6 V. The steel/sealant interface simulates the interaction of a real IC with the glass during stack operation. Fig. 7.5DF demonstrate clearly a correlation between porosity and applied voltage potential; in particular, porosity accumulates at the interface. The failure of gas tightness is highlighted by the appearance of barium chromate (yellow) all along the sealant surface (Fig. 7.5C). An alternative to glass-sealing are phyllosilicates-based gaskets, containing Mg, Fe, Al, Si, and K, which provide a gas barrier if mechanically

FIGURE 7.5 BaSi base sealant in between FSS plates. (D, E, F) Effect of the increased polarization intensity between the MIC pairs on the sealing porosity at the sealing/MIC interface. (A, B, C) The consequence of this increased porosity: yellow BaCrO4 diffuses also at the glass/ steel interface. Reprinted with permission from 12th European SOFC and SOE forum proceedings 2016 (Poitel S, Antonnetti Y, Wuillemin Z, Van Herle J, He´bert C. XII EFCF proceedings, 2016, B05, pp. 8897).

130

Solid Oxide Fuel Cell Lifetime and Reliability

compressed. The interaction between a phyllosilicate material (Thermiculite XJ766) and Crofer 22 APU at 800 C for hundreds of hours reveals a corrosion behavior similar to that of glass/steel [30]: Cr corrosion kinetics at the steel/Thermiculite/air interface is faster than expected for Crofer 22 APU, with a scale thickness attaining 2025 μm vs. a usual value of 23 μm. In addition, Cr depletion from the steel substrate caused Mg/Mn/Cr-enriched Fe oxide to grow. At the steel/Thermiculite/fuel side there was an increase in corrosion too, in the form of 810 μm thick Fe/Cr/Mg/Mn-oxide, but no crack formation typical of the interaction with a BCAS glass seal was apparent at the anode side. An important result emerging from various post-test analysis is that the most sensible location for corrosion degradation is always the triple-phase boundary alloy/sealant/gas (air or fuel) [27,28,31]. The use of a hybrid seal such as vermiculite-based materials, coupled with a BCAS glass, gives similar results in terms of interaction at the three-phase boundary [32].

Mechanical Stress As mentioned in the introduction, FSS has been chosen as IC material among others because of their good CTE compatibility with the ceramic components in an SOFC stack. Yet the difference in lattice parameters between steel substrate, thermally grown oxide and protective coating can cause residual mechanical stresses during heating or cooling of the stack and lead to spallation of the scale or the protective layer. During heating and scale growth in fact the scale is subjected to tensile stresses because of the higher thermal expansion of the steel substrate compared to that of the scale. Conversely, compression stresses in the scale arise during cooling. Understanding of the thermal stresses in the stack components is of high relevance to meet the reliability requirements of SOFC systems. In addition, due to the increasing interest to fit the SOFC systems operation to the user power demand, SOFC stacks have to withstand thermal cycling during their lifetime. Thermal cycling exposes the stack components to faster mechanical degradation, caused not only by the mismatch of thermal expansion between the materials but also by thermal gradients during heat up/cool down. In the literature the approach to the problem is heterogeneous, with different testing and simulation conditions, but there is a convergence to use the Griffith fracture approach: a certain energy (given by shear stress in this case) threshold must be overcome in order to cause failure of the ceramic materials. The interfacial strength is measured experimentally using either an indentation test (e.g. Rockwell) or a four-point bending test. Knowing the other mechanical properties of the materials studied, it is possible to correlate this value to a certain oxide layer thickness, which in turn, coupled with a kinetic

Lifetime Issues for Solid Oxide Fuel Cell Interconnects Chapter | 7

131

FIGURE 7.6 Plot of the Von Mises stress field at high temperature in a metallic IC manufactured by SOLIDpower S.p.A. The stress field is calculated by FEA simulating the stack production process steps, followed by coflow operation. Operating conditions used to generate the simulated temperature profile are Tfuel,in 5 700 C, Tair,in 5 700 C, partial prereforming of 0.5, Fuel Utilization 5 0.85, and Tmax,cell 5 827 C. Regions with high stresses indicate possible plastic deformation as well as risk of buckling.

curve for oxide layer growth, could forecast the adhesion lifetime of the ceramic layers (oxide layer or protective coating). As an example, Liu et al. forecasted the lifetime of coated (with MnCo2O4) and uncoated Crofer 22 APU substrates [33]. The samples are first oxidized at 800 C and then cooled to room temperature to simulate the stress due to a turning off of the system. Then a Rockwell indentation test is performed. From these test results, they found the strength at the interface (both at metal/scale and at scale/coating) which is the maximum shear stress tolerated by the samples. The experimental data were then used in a finite element modeling to obtain the interfacial strength as a function of the scale thickness. The model predicts a lifetime of about 4800 h for uncoated Crofer 22 APU and ca. 15,000 h when coated with MnCo2O4. These results reflect a critical scale thickness of 11.4 and 4.2 μm, respectively. Models like the aforementioned one estimate stress inside the MIC with finite element simulations. Fig. 7.6 shows an example. A temperature profile was imported into a finite elements analysis (FEA) model obtained from combined thermo-electrochemical and fluid dynamics models. Fig. 7.6 depicts the Von Mises stresses in the metallic IC of an SOFC stack model simulated in coflow configuration. The center of the IC is exposed to the highest temperatures, because of the electrochemical reactions occurring at the cell. Considering that stresses in SOFC stacks are generated, among others, by thermal gradients and thermal expansion mismatch between the

132

Solid Oxide Fuel Cell Lifetime and Reliability

components, the highest Von Mises stresses are located at the central area of the IC, as shown in Fig. 7.6. The IC is likely to buckle in this area, especially if it is relatively thin. In turn, buckling may provoke (1) on one face, damage of the adjacent elements, i.e., inelastic deformation of the gas distribution layers and (2) detachment of the interface, on the opposite face. As a result, the stack reliability is affected also by the stresses in the IC. To investigate thermo-mechanical reliability of SOFC stacks upon operation, scenarios of long-term operation, thermal cycling, and a combination of them have been simulated [34]. Using the stress states simulated by the FEA model, mechanical reliability analysis of SOFC stacks was then carried out either by applying fracture mechanics or by investigation of the contact pressure between the stack components. Contact pressure influences the electrical current pathways between IC and cell and thus the stack performance and durability [34,35]. Nonetheless, thermo-mechanical degradation of ICs in real stacks appears to be less dramatic than that predicted by modeling: in a stack run for 35,000 h and containing Fe-doped MnCo2O4 (MCF) coated Crofer 22 APU ICs (same steel substrate as in [33]), no evidence of delamination at the steel/scale or scale/MCF interface occurred, even though the average scale thickness was between 3 μm and 5 μm [36] and the model of Liu et al. predicted delamination at 4.2 μm scale thickness for coated surfaces.

Solutions to Decrease IC Degradation Tuning the alloy composition is one strategy adopted to mitigate MIC degradation. Reactive elements addition in the alloy affects the oxygen and chromium diffusion coefficients in thermally grown Cr2O3, decreasing its growth rate [37,38]. The addition of reactive elements (e.g., small amounts of Ce, La, or Y) improves as well the scale adhesion to the alloy substrate and therefore the resistance to thermal cycling. Another important alloying element is Mn. Its fast diffusion across the chromia scale [16] and interaction with chromium leads to the creation of a preferred (Cr,Mn)3O4 spinel phase at the gas interface of the thermally grown oxide scales. This spinel phase positively influences the electrical conductivity [2] and in addition reduces the undesirable Cr evaporation. Alloy composition alone is not sufficient to achieve the desired IC lifetime. The deposition of protective coatings on the alloy surface has been established as a sine qua non for long-term SOFC application. Spinel and perovskite coatings are commonly used. Their deposition techniques play an important role, as they lead to different microstructures, in particular the coating density. The effectiveness of a protective coating in improving the electrical contact is shown in Fig. 7.7, where ASR results for coated (solution based) and uncoated alloys substrates are compared. The difference is

Lifetime Issues for Solid Oxide Fuel Cell Interconnects Chapter | 7

133

FIGURE 7.7 ASR graph of alloys ITM (26 at.% Cr) and Crofer 22 APU (22 at.% Cr) with and without protective coating (MnCo2O4). In air at 800 C using 250 mA cm2, in contact with LSCM5555 powder LSM tablet with Pt electrode (tablet and electrode resistance subtracted).

FIGURE 7.8 Electrical resistivity performances of FSS substrates coated with Fe-doped MCO deposited by different methods. High coating density techniques (APS and PVD) produce the best results. Testing conditions: 700 C, 3 vol.% H2O(g), 0.4 A cm22. ALD: atomic layer deposition.

bigger than one order of magnitude, and a similar result is found for the same coating on top of different steels. With the coating material being equal, high coating density deposition methods like physical vapor deposition (PVD) or atmospheric plasma spray (APS) give better results, as demonstrated in Fig. 7.8 on Sandvik Sanergy HT steel substrates. Fig. 7.9 illustrates the microstructures that help explain the results seen in Fig. 7.8. Fig. 7.9A shows the MnCo2O4 protective coating deposited by wet

134

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 7.9 Thermal oxide grown on FSS substrates coated with different coating techniques but same protective layer (MnCo2O4): (A) WPS, (B) APS, and (C) PVD. High-density coatings lead to the best results in terms of minimum scale thickness, which in turn gives lower contact resistive loss for the samples (B) and (C).

powder spray (WPS) to remain porous, leading to a thicker Cr2O3 scale grown on top of the FSS substrate, with respect to the APS and PVD deposited dense coatings, in Fig. 7.9B and Fig. 7.9C, respectively. PVD coatings on FSS lead to the lowest contact resistance, Fig. 7.8, thus represent the state-of-the-art [11,39]. The ΔpO2 through a Cr2O3 layer is very large going from metallic Cr at the metal/scale interface to air. Application of a dense coating (e.g., spinel layer) therefore reduces the oxygen gradient over the Cr2O3 layer significantly (going from metallic Cr to the decomposition pressure of the spinel), lowering the gradient and oxygen transport in this layer. Adding a second protective layer (e.g. as a perovskite) might then reduce the oxygen gradient over the spinel layer in the same way. These coatings have also different transport properties. In spinels often cation diffusion is predominant, while in perovskites often oxygen ions are the predominant diffusing species. The spinel thus limits the oxygen transport while a perovskite can limit cation diffusion. The total gradient is the same as for uncoated steel, but since the coating layers restrict diffusion efficiently, a protection is obtained which can withstand several thousands of hours before Cr reaches the surface; since the Cr concentration in the surface is then very low and stabilized in a structure, only restricted Cr evaporation is expected. Cr retention properties of spinels are depicted in Fig. 7.10. In this case too, the denser coatings such as PVD and APS show the best performances. The dense spinels protected a 200 μm La1xSrxCoO3 (LSC) cathode from Cr poisoning for 1000 h at 700 C (Fig. 7.10B) while the one deposited with WPS let chromium to poison the LSC, producing the low electrically conductive phase SrCrO4 (light grey layer at coatingcathode interface in Fig. 7.10A). The effectiveness of coatings has been demonstrated in the long term on stacks (Table 7.2). Currently the main drawback of these coatings and their deposition techniques are the associated production cost and the environmental impact, Co being a hazardous and critical element [40].

Lifetime Issues for Solid Oxide Fuel Cell Interconnects Chapter | 7

135

FIGURE 7.10 LSC cathode material in contact with FSS substrate coated by MCO spinel: Cr diffusion profile expressed in atomic percentage. (A) MCO deposited via WPS: Cr poisoning of the LSC, (B) MCO deposited via PVD: excellent Cr retention. Pd is a counter contact plate in the ASR test set-up. Dark red line represents 1% atomic threshold.

From the mechanical point of view, the poor creep resistance of FSSs is improved with the addition of an amount of Nb and/or W. These elements lead to Laves phases, intermetallic compounds with the composition AB2 (e.g. Fe2Nb), which segregate at the grain boundaries and induce precipitation hardening [41]. An example of Laves phases in FFS is given in Fig. 7.11: Crofer 22H (cf., Table 7.1) segregates NbW containing precipitates at the grain boundary, Crofer 22 APU having a similar

136

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 7.11 Comparison of Crofer 22 H (A) vs. Crofer 22 APU (B) microstructure after thermal ageing at 800 C. The white Laves phases in Crofer 22H due to Nb/W segregate mostly at the grain boundaries.

FIGURE 7.12 Comparison of steel/scale interface for Crofer 22H (left) and APU (right). Steels aged for 3000 h at 800 C in air and current density of 0.35 A cm22. Laves phases accumulation below the scale in Crofer 22H is clearly visible.

composition to Crofer H but without Nb and W, shows no sign of these phases. The precise correlation between Laves phase precipitation and creep strengthening has not been found yet, as other precipitates strengthen the FFS microstructure too. In addition, Ostwald ripening of Laves phases can decrease the mechanical properties of the material leading to a ductile-tobrittle transition. Coarsening of the Laves phases must be delayed to ensure long life of the IC [42]. Laves phases segregate below the scale too (Fig. 7.12); this, together with their affinity for Si, may be another cause of IC electrical behavior degradation [39].

LIFETIME BEHAVIOR OF STACKS AND CELLS TESTED IN OPERATING CONDITIONS Table 7.2 summarizes long-term results on present state-of-the-art stacks, with a focus on the observed MIC degradation phenomena.

TABLE 7.2 Long-Term Tests of Stacks or Cells Based on Different Structural Technology IC Alloy and Coating

Testing Conditions

Cell

Stack Degradation

B17,000 h 700 C 0.5 A cm2

Anode: NiYSZ (H2/3%H2O) Cathode: LSCF (air)

8-12 mV kh1 , 13 kh 15-20 mV kh1 . 13 kh

Interconnect Behavior

Anode Supported 1 [43]

Crofer 22 APU Mn2O3 (WPS)

G

G

G

2 [44]

Crofer 22 APU MCF (APS)

19,000 h 800 C 0.5 A cm2

Anode: NiYSZ (H2/3%H2O) Cathode: LSM (air)

4 mV kh1

G G

G

G

G

3 [36]

Crofer 22 APU MCF (APS)

B35,000 h 700 C 0.5 A cm2

Anode: NiYSZ (H2/25% H2O) Cathode: LSCF(air)

0.3%/kh

G

G

G

Mn2O3 let some Cr migrate - SrCrO4 grown on LSCF in contact with IC - progressive stack degradation At cathode side: 4 μm scale thickness; local Cr2O3 breakaway with Fe-oxide spots (but no short-circuits between ICs). At anode side: 2 μm scale thickness; Ni diffusion into Crofer 22 APU - austenite phase creation Small cracks in MCF but no delamination Thicker scale at the uncoated anode side (12 μm) than at the coated cathode side (3 μm) Micropores accumulation in the MCF at the IC/MCF interface Humidity present at anode side leads to porosity in the scale MCF APS deposited stop Cr migration towards cathode At cathode side: 35 μm thick CrMn scale. No spallation at the steel/scale/MCF interface. At anode side: Ni diffusion (from contacting wires) lead to a 50100 μm wide austenized zone. No brittle σ-phase precipitates were found in the same area. No enhanced corrosion at the steel/sealing interface (Continued )

TABLE 7.2 (Continued)

4 [45]

IC Alloy and Coating

Testing Conditions

Cell

Stack Degradation

Interconnect Behavior

Crofer 22 APU MCF (APS)

B6000 h 700 C 0.5 A cm2

Anode: NiYSZ (LNG 7.2 slm, H2 3.2 slm, H2O 15.2 slm) Cathode: LSCF (air)

0.3%/kh for 4.5 kh

G

Anode: NiYSZ (N2/H2) Cathode: LSC (air)

0.4%/kh

G G G

Strong MCF adhesion on Crofer 22 APU No cracks penetrating through the coating MCF coating retained chromium Melting of IC and sealing glass occurred in one cell because of a leakage in the sealant

Anode Supported 5a

AISI 441 MCO (WPS)

B5000 h 780 C 0.4 A cm2

G

G G

G G

6a

AISI 441 Spinel (WPS)

2600 h 760 C 124 thermal cycles

Anode: NiYSZ(N2/H2) Cathode: LSC(air)

-

G

G

G

At cathode side: thicker scale on the IC ribs (B5 μm) with respect to the valleys (B3 μm). Fe breakaway corrosion and Cr diffuses through MCO. Good adherence: no coating or scale spallation Densification of the MCO coating where the ribs are in contact with LSC. A possible reason is Fe diffusion into MCO from the steel substrate No enhanced corrosion due to dual atmosphere At anode side: no coating, oxide layer about 5 μm At cathode side: scale thickness ,3 μm where MIC not in contact with the cathode (valley), ca. 10 μm where the MIC was in contact with the cathode (rib). Fe breakaway corrosion into the scale and the spinel coating No sign of delamination at the steel/scale and scale/ protective coating interface despite the 124 thermal cycles

7 [46]

AISI 441 Ce-MCO (Slurry) on LSM // Al2O3 where no electrical contact needed

6000 h 800 C 0.3 A cm2

Anode: Ni-YSZ (H2,N2(1:1) 1 3% H2O) Cathode: LSMYSZ(air)

B1.7-4.5%/kh

G

G

G

G

At AISI 441/Ni-current collector interface corrosion is enhanced if Ni wires have diameter ,100 μm At cathode side: Cr depletion below the scale (16 at.% instead of 20 at.%) No Cr traces in LSM. No spallation at scale/IC interface Typical Si, Ti accumulation below the scale, but no continuous layer

Electrolyte Supported 8 [17,47]

CFY LSM()

B40,000 h 900 C 0.25 A cm2

Anode: Ni/CGO (CPO 1 reformed natural gas) Cathode: LSM(air)



G

G

G

G

Considering the IC ribs geometry, faster scale grows at the IC/cathode interface than in the IC valley. This happens for both cathode and anode. Higher average scale thickness at anode with respect to cathode Different CFY behavior at anode side: at the inlet valley, Cr2N compound is found at the center, the interaction of the oxide layer with the Ni-mesh is more intense; at the outlet valley, more pores in scale compared to the scale under the ribs MIC-cell thermal mismatch lead to local stress peaks which in turn boost crack propagation (Continued )

TABLE 7.2 (Continued) IC Alloy and Coating

Testing Conditions

Cell

Stack Degradation

Interconnect Behavior

G

Increment in ohmic losses due to corrosion interlayer in between metal substrate and infiltrated CGO-Ni

G

STN:FeCr half-cell showed better corrosion resistance than reference YSZ:FeCr reference cell STN:FeCr (50:50) composition demonstrated a better oxidation resistance than STN:FeCr (70:30)

Metal Supported 9 [48]

FeCr alloy powders 

1100 h 650 C 0.25 A cm2

Anode: CGO-Ni (H2/3% H2O) Cathode: LSCF/CGO (air)

(single cell) B5%/kh

10b [49]

FeCr powders Ni-CGO

500 h 850 C 

Anode: STN/FeCr (pH2O/ pH2 5 9) Half-cellb



a

G

Courtesy of SOLIDpower s.p.a. The corrosion test has been done on half cells composed of FeCr/STN:FeCr infiltrated by Ni-CGO. STN: Nb doped SrTiO3

b

Lifetime Issues for Solid Oxide Fuel Cell Interconnects Chapter | 7

141

CONCLUSION This chapter provided an overview of the main lifetime issues of SOFC IC materials and sealants, in particular corrosion and mechanical stress. The first causes an increase in stack ohmic resistances, while the second can trigger structural rupture and failures in electrical contacts. The discussion was focused on the currently widely used FSS materials for ICs. Influence of air/fuel environment (pO2, pH2, pH2O) on the FSS aging process is not completely understood yet, even if there is a tendency to consider the corrosion process independent of pO2 and accelerated by pH2O. Separate experiments showed higher degradation kinetics of uncoated FSS in dual atmosphere condition, but observations on MICs tested in commerciallike stacks (run in dual atmosphere too) did not show any dramatic corrosion. SOFC-customized steel compositions and protective coatings have been developed to ensure the goal of at least 40,000 h stack life. An overview of these solutions was given, with the conclusion that high-density coating techniques such as PVD or APS are the options leading to best performance. In the final part, a table summarizes lifetime results on stacks with different cell types tested in operating conditions for medium to extended durations. The reported posttest analyses of these stacks/cells indicate that the materials appear more resistant to aging than expected from simulation and from separate tests performed in out-of-stack conditions. This illustrates on the one hand that models can overestimate the materials degradation, and on the other hand, that current solutions to extend lifetime are already reasonably effective.

ACKNOWLEDGEMENTS The authors would like to thank the FCH-JU Programme and the Scored project partners (Grant Agreement 325331) for electrical resistance and materials microstructure data, the European Fuel Cell Forum board for the permission to republish images, the SOFC companies SOLIDPower and Hexis AG for stack and other experimental data, the EPFL Center for Electron Microscopy for the support in microscope imaging, and M.Sc. Ste´phane Poitel (EPFL) for his contribution to the sealing materials paragraph.

REFERENCES [1] Tietz F. Thermal expansion of SOFC materials. Ionics 1999;5(1):12939. [2] Zhu WZ, Deevi SC. Development of interconnect materials for solid oxide fuel cells. Mater Sci Eng A 2003;348(12):22743. [3] Yokokawa H. Perovskite oxide for solid oxide fuel cells. USA: Springer; 2009 [Chapter 2]. [4] Wu J, Liu X. Recent Development of SOFC Metallic Interconnect. J Mater Sci Technol 2010;26(4):293305. [5] Terner MR, Andreas Schuler J, Mai A, Penner D. On the conductivity degradation and phase stability of solid oxide fuel cell (SOFC) zirconia electrolytes analysed via XRD. Solid State Ionics 2014;263:1809.

142

Solid Oxide Fuel Cell Lifetime and Reliability

[6] Pappis J, Kingery WD. Electrical properties of single-crystal and polycrystalline alumina at high temperatures. J Am Ceram Soc 1961;44(9):45964. [7] Stanislowski M, Wessel E, Hilpert K, Markus T, Singheiser L. Chromium vaporization from high-temperature alloys. J Electrochem Soc 2007;154(4):A295306. [8] Yang Z, Weil KS, Paxton DM, Stevenson JW. Selection and evaluation of heat-resistant alloys for SOFC interconnect applications. J Electrochem Soc 2003;150(9):A1188201. [9] Plansee, CFY factsheet. https://www.plansee.com/en/materials/chromium.html; last visit 12.2016. [10] Tallgren J, Bianco M, Mikkola J, Himanen O, Rautanen M, Kiviaho J, et al. Comparison of different manganese-cobalt-iron spinel protective coatings for SOFC interconnects, XII EFCF proceedings, 2016, B06, pp. 114124. [11] VDM Metals, Crofer 22 APU datasheet. , http://www.vdm-metals.com/fileadmin/user_ upload/Downloads/Data_Sheets/Data_Sheet_VDM_Crofer_22_APU.pdf . ; last visit 12.2016. [12] Stanislowski M, Froitzheim J, Niewolak L, Quadakkers WJ, Hilpert K, Markus T, et al. Reduction of chromium vaporization from SOFC interconnects by highly effective coatings. J Power Sources 2007;164:57889. [13] Shreir LL, editor. Corrosion. Oxford: Butterworth-Heinemann; 1994 [Chapter 1]. [14] Wagner C. Theoretical analysis of the diffusion process determining the oxidation rate of alloys. J Electrochem Soc 1952;99(10):36980. [15] Brylewski T, Tanko M, Maruyama T, Przybylski K. Application of Fe-16Cr ferritic alloy to interconnector for a solid oxide fuel cell. Solid State Ionics 2001;143(2):13150. [16] Palcut M, Mikkelsen L, Neufeld K, Chen M, Knibbe R, Hendriksen PV. Corrosion stability of ferritic stainless steels for solid oxide electrolyser cell interconnects. Corros Sci 2010;52(10):330920. [17] Linder M, Hocker T, Holzer L, Friedrich KA, Iwanschitz B, Mai A, et al. Cr2O3 scale growth rates on metallic interconnects derived from 40,000 h solid oxide fuel cell stack operation. J Power Sources 2013;243:50818. [18] Fontana S, Chevalier S, Caboche G. Metallic interconnects for solid oxide fuel cell: performance of reactive element oxide coating during 10, 20 and 30 months exposure. Oxid Met 2012;78(5):30728. [19] Linder M, Hocker T, Holzer L, Friedrich KA, Iwanschitz B, Mai A, et al. Model-based prediction of the ohmic resistance of metallic interconnects from oxide scale growth based on scanning electron microscopy. J Power Sources 2014;272:595605. [20] Nagai H, Ishikawa S, Amano N, Shoji K. Electrical conductivity of sintered Cr2O3 simultaneously doped with NiO and rare earth oxide at low oxygen potential. Trans. Japan Inst Met 1985;26(10):75362. [21] Linder M, Hocker T, Holzer L, Pecho O, Friedrich KA, Morawietz T, et al. Ohmic resistance of nickel infiltrated chromium oxide scales in solid oxide fuel cell metallic interconnects. Solid State Ionics 2015;283:3851. [22] Polman EA, Fransen T, Gellings PJ. The reactive element effect; ionic processes of grainboundary segregation and diffusion in chromium oxide scales. J Phys: Condens Matter 1989;1(28):4497510. [23] Alnegren P, Sattari M, Froitzheim J, Svensson J-E. Corrosion stability of ferritic stainless steels for solid oxide electrolyser cell interconnects. Corros Sci 2016;110:20012. [24] Saunders SRJ, Monteiro M, Rizzo F. The oxidation behaviour of metals and alloys at high temperatures in atmospheres containing water vapour: a review. Prog Mater Sci 2008;53 (5):775837.

Lifetime Issues for Solid Oxide Fuel Cell Interconnects Chapter | 7

143

[25] Yang Z, Xia G, Singh P, Stevenson JW. Effects of water vapor on oxidation behavior of ferritic stainless steels under solid oxide fuel cell interconnect exposure conditions. Solid State Ionics 2005;176(1718):1495503. [26] Alnegren P, Sattari M, Svensson J, Froitzheim J. Severe dual atmosphere effect at 600 C for stainless steel 441. J Power Sources 2016;301:1708. [27] Haanappel VAC, Shemet V, Gross SM, Koppitz Th, Menzler NH, Zahid M, et al. Behaviour of various glass-ceramic sealants with ferritic steels under simulated SOFC stack conditions. J Power Sources 2004;150:86100. [28] Batfalsky P, Haanappel VAC, Malzbender J, Menzler NH, Shemet V, Vinke IC, et al. Chemical interaction between glass-ceramic sealants and interconnect steels in SOFC stacks. J Power Sources 2006;155(2):12837. [29] Poitel S, Antonnetti Y, Wuillemin Z, Van herle J, He´bert C. The effect of polarization on SOFC seal ageing, XII EFCF proceedings, 2016; B05:8897. [30] Wiener F, Bram M, Buchkremer H-P, Sebold D. Chemical interaction between Crofer 22 APU and mica-based gaskets under simulated SOFC conditions. J Mater Sci 2007;42 (8):264351. [31] Menzler NH, Batfalsky P, Blum L, Bram M, Groß SM, Haanappel VAC, et al. Studies of material interaction after long-term stack operation. Fuel Cells 2007;7(5):35663. [32] Thomann O, Rautanen M, Himanen O, Tallgren J, Kiviaho J. Post-experimental analysis of a solid oxide fuel cell stack using hybrid seals. J Power Sources 2015;274:100915. [33] Liu WN, Sun X, Stephens E, Khaleel MA. Life prediction of coated and uncoated metallic interconnect for solid oxide fuel cell applications. J Power Sources 2009;189 (2):104450. [34] Greco F, Nakajo A, Wuillelmin Z, Van herle J. Thermo-mechanical reliability of SOFC stacks during combined long-term operation and thermal cycling. ECS Trans 2015;68 (1):192131. [35] Nakajo A, Wuillemin Z, Van herle J, Favrat D. Simulation of thermal stresses in anodesupported solid oxide fuel cell stacks. Part II: Loss of gas-tightness, electrical contact and thermal buckling. J Power Sources 2009;193(1):21626. [36] Menzler NH, Batfalsky P, Beez A, Blum L, Gross-Barsnick S, Niewolak L, et al. Post-test analysis of a solid oxide fuel cell stack operated for 35,000 h, XII EFCF proceedings, 2016;A11:2907. [37] Lang E. The role of active elements in the oxidation behaviour of high temperature metals and alloys. Netherlands: Elsevier; 1989. [38] Chevalier S. What did we learn on the reactive element effect in chromia scale since Pfeil’s patent?. Mater Corros 2014;65(2):10913. [39] Bianco M, Auchlin M, Diethelm S, Van herle J. Evolution of oxidation of SOFC interconnect alloys in dry and wet air, XII EFCF proceedings, 2016;B05:98107. [40] Report on critical raw materials for the EU, 2014. [41] Froitzheim J, Meier GH, Niewolak L, Ennis PJ, Hattendorf H, Singheiser L, et al. Development of high strength ferritic steel for interconnect application is SOFCs. J Power Sources 2008;178(1):16373. [42] Zhu S, Yang M, Song XL, Tang S, Xiang ZD. Characterization of Laves phase precipitation and its correlation to creep rupture strength of ferritic steels. Mater Charac 2014;98:605. [43] Menzler NH, Batfalsky P, Groß S, Shemet V, Tietz F. Post-test characterization of an SOFC short stack after 17,000 hours of steady operation. ECS Trans 2011;35 (1):195206.

144

Solid Oxide Fuel Cell Lifetime and Reliability

[44] Malzbender J, Batfalsky P, Vaßen R, Shemet V, Tietz F. Component interactions after long-term operation of an SOFC stack with LSM cathode. J Power Sources 2012;201:196203. [45] Fang Q, Blum L, Batfalsky P, Menzler NH, Packbier U, Stolten D. Durability test and degradation behavior of a 2.5 kW SOFC stack with internal reforming of LNG. Int J Hydrogen Energy 2013;38(36):1634453. [46] Chou Y-S, Stevenson JW, Choi J-P. Long-term evaluation of solid oxide fuel cell candidate materials in a 3-cell generic stack test fixture, part III: Stability and microstructure of Ce-(Mn, Co)-spinel coating, AISI441 interconnect, alumina coating, cathode and anode. J Power Sources 2014;257:44453. [47] Fleischhauer F, Tiefenauer A, Graule T, Danzer R, Mai A, Kuebler J. Failure analysis of electrolyte-supported solid oxide fuel cells. J Power Sources 2014;258:38290. [48] Blennow P, Hjelm J, Klemensø T, Ramousse S, Kromp A, Leonide A, et al. Manufacturing and characterization of metal-supported solid oxide fuel cells. J Power Sources 2011;196(17):711725. [49] Blennow P, Persson AH, Nielsen J, Sudireddy BR, Klemenso T. Infiltrated SrTiO3: Fe-Cr based anodes for metal-supported SOFC, X EFCF proceedings, 2012;A09:7283.

Chapter 8

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Joongmyeon Bae Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea

Chapter Outline Introduction to Fuel Processing in Solid Oxide Fuel Cells Fuel Processing Stages of Fuel Processing Components of Fuel Processors Lifetime of Fuel Processors Catalyst Degradation in Reformers Deactivation Mechanisms of Catalyst Metals Carbon Formation on Reforming Catalysts Liquid Fuel Processor Designs to Enhance Reliability Component of a Liquid Fuel Processor for Solid Oxide Fuel Cells Design Factors for Fuel Processor Fuel Delivery Design of Liquid Fuel Processing

146 146 147 148 149 149 149 150 154

154 154 155

kW-Class Reformer for Reliable Solid Oxide Fuel Cell System 158 Postprocessing in Reforming to Enhance the Lifetime of Solid Oxide Fuel Cells 159 Concept of Postreforming 159 Desulfurizer for Heavy Hydrocarbons 161 Catalysts for Desulfurization 163 Lifetime Estimation of Fuel Processors 164 Engineering Issues (BOPs) of Fuel Processors on Durability 164 Durability Test Method for Fuel Processors 166 Practical Example of Durability Test 167 Lifetime Extension of Fuel Processors 167 Conclusion 169 References 169

Solid Oxide Fuel Cell Lifetime and Reliability. DOI: http://dx.doi.org/10.1016/B978-0-08-101102-7.00008-8 Copyright © 2017 Elsevier Ltd. All rights reserved.

145

146

Solid Oxide Fuel Cell Lifetime and Reliability

INTRODUCTION TO FUEL PROCESSING IN SOLID OXIDE FUEL CELLS Solid oxide fuel cells (SOFCs) are different from low temperature fuel cells in that they can use fuels other than hydrogen. SOFCs have fuel flexibility, which means they are capable of utilizing carbon monoxide due to the high operating temperature (5001000 C). Moreover, SOFCs can use some lower hydrocarbons such as methane as a fuel through an internal reforming process. Therefore SOFCs can easily use hydrogen-rich synthetic gases from liquid reforming processes, which include a certain amount of carbon monoxide and methane as shown in Fig. 8.1.

Fuel Processing Hydrogen is an eco-friendly alternative energy source that is widely available on Earth. However, most hydrogen atoms exist in the form of water [2]. The ultimate solution for producing hydrogen is decomposition of water. However, water decomposition consumes much energy during the production of hydrogen. The most practical way to obtain hydrogen is fuel processing, which currently includes catalytic hydrocarbon fuel reforming. Fig. 8.2 shows a schematic diagram for liquid fuel processing for SOFC systems. The fuel processing is a series of processes for converting hydrocarbons to “fuel cell-friendly” fuels, mainly hydrogen. Most hydrocarbons are converted into hydrogen-rich fuels in this process. In addition, there are several additional processes, including desulfurization and postreforming. These postprocesses play significant roles in stable SOFC operation and will be briefly discussed in this chapter.

FIGURE 8.1 Complexity of the fuel processors by several types of fuel cells [1].

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Chapter | 8

147

FIGURE 8.2 Fuel processing system.

Stages of Fuel Processing Catalytic Fuel Reforming Process Catalytic fuel reforming is a type of incomplete combustion reaction of hydrocarbon fuel. In fuel reforming reactions, the fuel reacts with oxygen, water, or both as oxidizer(s) at certain temperatures that are lower than the combustion temperature (6001000 C). The products of the fuel reforming reaction are mainly hydrogen and carbon monoxide. Most fuel reforming reactions need catalysts that accelerate the desired reaction. Catalytic fuel reforming process can be divided according to the oxidizer in the fuel reforming process—steam reforming (SR), partial oxidation (POX), and autothermal reforming (ATR). POX : Cn Hm 1 aO2 -bH2 1 cCO 1 dCO2 1 eH2 O ðΔH , 0Þ

ð8:1Þ

SR : Cn Hm 1 aH2 O-bH2 1 cCO 1 dCO2 1 eH2 O ðΔH . 0Þ

ð8:2Þ

ATR : Cn Hm 1 aO2 1 bH2 O-cH2 1 dCO 1 eCO2 1 f H2 O ðΔHB0Þ ð8:3Þ SR is the most common reforming method. SR offers a higher hydrogen concentration, but it requires additional heating equipment due to the endothermic reaction. POX produces less hydrogen than SR; however, the reaction is rapid and does not need heat sources. ATR is a combined reaction of SR and POX. Detailed characteristics of each type are summarized in Table 8.1.

148

Solid Oxide Fuel Cell Lifetime and Reliability

TABLE 8.1 Types of Catalytic Fuel Reforming Process SR

POX

ATR

Oxidizer

Steam

Oxygen (air)

Steam and oxygen

Hydrogen yield

High

Low

High

Start-up time

Slow

Quick

Quick

Reaction heat

Endothermic

Exothermic

Thermally neutral

Desulfurization Removing sulfur compounds, i.e., desulfurization, is certainly a necessary process because commercial hydrocarbon fuels normally contain a small amount of sulfur contents. The sulfur-containing species will harmfully affect SOFC anodes. Fuel cell performance can be degraded even at hydrogen sulfide levels of 5 ppm [3]. Therefore the desulfurization process plays an important role in fuel processing to achieve stable SOFC operation. There are several methods of desulfurization such as hydrodesulfurization (HDS), chemical or physical adsorption, and catalytic sieves [36]. Desulfurization will be discussed in detail in the Desulfurizer for Heavy Hydrocarbons section. Postreforming Process Yoon and Bae suggested postreforming to achieve stable long-term SOFC operation [7]. They asserted that a certain quantity of light hydrocarbons (C2C4 hydrocarbons, mainly C2H4) accelerates the degradation of the reforming performance by forming coke on the reforming catalyst [8]. In particular, even a small amount (approximately up to 0.1 vol.%) of C2H4 induces severe carbon deposition on Ni-based SOFC anodes. In this case, the electrical performance of the SOFC is critically degraded by the carbon deposition [9]. Postreforming is introduced to convert the light hydrocarbons in reformate off-gases into H2, CO, CO2, and CH4, and it will be discussed in the Concept of Postreforming section.

Components of Fuel Processors A fuel processor is a type of chemical reactor composed of the following parts: fuel injectors, catalysts, an outer container, and the balance of plants (BOPs). The fuel injectors deliver the fuel and oxidizers (i.e., water and oxygen) to the catalysts. Particularly for liquid hydrocarbon fuel, such as gasoline and diesel, fuel atomization and evaporation is a critical issue to ensure homogeneous mixing of reactants for the reliability of fuel processors. Catalysts are the main part of fuel processors and are composed of active

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Chapter | 8

149

materials and the support material. The catalysts are commonly contained in the reactor in a pelletized form. However, for mobile applications, structured catalysts such as honeycomb have recently been applied due to their high mechanical stability. The outer container is usually made of stainless steel. However, due to high temperature and humidity, other outer container materials have been considered to resist oxidation, corrosion, and carburization. The BOPs of the fuel processor contain mechanical and electrical components, such as fuel pumps, gas controller, heat exchanger, air compressors, and blowers.

Lifetime of Fuel Processors The lifetime of a fuel processor mainly depends on the reliability of the catalysts. The fuel processing catalysts degrade due to several causes. The main catalytic degradation of fuel processors is due to carbon deposition on the surface of the catalysts. Carbon deposits on the active catalyst materials and blocks the fuel reforming reaction. Sulfur compounds affect not only the anode of SOFCs but also the fuel processor catalysts. The sulfur compounds in liquid fuels convert the catalysts to metal sulfides and decrease the catalytic activity. Thermal degradation, such as agglomeration and metallic evaporation, is also a significant issue in the catalysts. Moreover, reactor corrosion and reliability of BOPs affect the lifetime of a fuel processor. These causes will be discussed in detail later.

CATALYST DEGRADATION IN REFORMERS Deactivation Mechanisms of Catalyst Metals Reforming catalysts can be deactivated during reaction processes. Degradation causes a loss of catalytic surface and support area. Typically this process is referred to as “sintering.” The sintering process generally takes place at high reaction temperatures ( . 500 C) and is generally accelerated by the presence of water vapor in reducing environments. Sintering can influence the other catalytic degradation processes. For example, the carbon deposition, which is one of the main factors of the catalytic degradation, can be controlled by the nickel particle size. Because of the increased nickel particle size from the sintering process, the catalyst activity is also degraded [10]. For this reason, sintering gives high impact during the reforming process with the high temperatures and pressures of the steam. Experimental and theoretical studies on the sintering of metallic catalysts have been reviewed extensively [1114]. In these reviews, two mechanisms for the metal particle growth have been proposed: (1) particle migration, where entire crystallites migrate over the support after coalescence and (2) Ostwald ripening (atom migration or vapor transport), where metal

150

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 8.3 A series of transmission electron microscope images of a Ni/MgAl2O4 reforming catalyst. Sequence (A) illustrates the particle migration and coalescence sintering mechanisms. Sequence (B) illustrates the Ostwald ripening mechanism. The conditions were 706.8 C and 3.1 mbar H2. Relative times were 0 s (a1), 44.5 s (a2), 65 s (a3), 0 s (b1), 41 s (b2), and 44.5 s (b3) [15].

species segregated from one crystallite migrate over the support or via the (gas) phase and are captured by another crystallite. These two fundamental sintering mechanisms are illustrated by the sequences of the electron microscope images in Fig. 8.3. Meanwhile, deactivation by the loss of metallic catalysts due to vaporization is also critical. Although direct vaporization is generally an insignificant route to catalyst deactivation, the loss of metallic catalyst through the formation of volatile compounds can be significant over a wide range of conditions, including relatively mild conditions. The typical volatile compounds generated in reforming are listed in Table 8.2. The conditions under which the volatile oxides are formed considerably vary according to the metallic catalyst. For example, RuO3 can be formed at room temperature, while PtO2 is formed at measurable rates only at temperatures exceeding B500 C. A generalized mechanism of deactivation via the formation of volatile metal compounds is described in Figs. 8.4 and 8.5.

Carbon Formation on Reforming Catalysts Carbon formation is the most detrimental deactivation factor, and it is frequently generated in the reforming process. It can increase the pressure drop, crush the catalyst pellets, block the active nickel surface, and even form inside

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Chapter | 8

151

TABLE 8.2 Types and Examples of Volatile Compounds Formed in Catalytic Reactions [14,16] Gaseous Environment

Compound Type

Compound Example

CO, NO

Carbonyls and nitrosyl carbonyls

Ni(CO)4, Fe(CO)5 (0300 C)

O2

Oxides

RuO3 (25 C), PbO ( . 850 C), PtO2 ( . 700 C)

H2O(steam)

Hydroxyls

Ni(OH)2

FIGURE 8.4 Generalized mechanisms and kinetics for deactivation by metal loss [14].

FIGURE 8.5 Formation of volatile tetra-nickel carbonyl at the surface of a nickel crystallite in a CO atmosphere [14].

the reforming tubes to block gas flow. A number of books and reviews discuss the formation of carbons on catalysts and the deactivation of the catalysts [1719]. Carbon deposition can reduce the activity of the catalyst in several ways, and eventually the catalyst will be fully covered by carbon. Three types

152

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 8.6 Transmission electron microscope images of pyrolytic carbon on a MgAl2O4 carrier (A), encapsulating carbon (B), and whisker carbon (C) on Ni/MgAl2O4 reforming catalysts [15].

TABLE 8.3 Carbon Species Formed During SR Hydrocarbons on Nickel Catalysts [14] Encapsulating Film

Whisker

Pyrolytic Carbon

Formation

Slow polymerization of CnHm radicals on a Ni surface into an encapsulating film

Diffusion of C through Ni crystal, nucleation, and whisker growth with Ni crystal at top

Thermal cracking hydrocarbons: deposition of C precursors on the catalyst

Effects

Progressive deactivation

No deactivation of Ni surface: breakdown of catalyst and increasing ΔP

Encapsulation of catalyst particles: deactivation and increasing ΔP

Temperature range

,500 C

.450 C

.600 C

Critical parameters

Low temperature Low H2O/CnHm Low H2/CnHm Aromatic feed

High temperature Low H2O/CnHm No enhanced H2O adsorption Low activity Aromatic feed

High temperature High void fraction Low H2O/CnHm High pressure Acidic catalyst

of carbon have been observed in a reformer: (1) encapsulating, (2) whisker carbon, and (3) pyrolytic as imaged by electron microscopy in Fig. 8.6 and described in Table 8.3. Among them, whisker carbon is the most destructive form of carbon formed in reforming on the catalysts, especially nickel. The mechanisms of carbon deposition and coke formation on metal catalysts from hydrocarbons [1719] are illustrated in Fig. 8.7. For example, CO dissociates on metals to form Cα, an adsorbed atomic carbon. Cα can react to form Cβ, a

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Chapter | 8

153

FIGURE 8.7 Formation and transformation of coke on metal surfaces [14].

polymeric carbon film. The more reactive, amorphous forms of carbons formed at low temperatures (e.g., Cα and Cβ) are converted at high temperatures over a period of time to less reactive, graphitic forms. The carbon formation in the reforming reaction directly affects catalyst activity. It is critical for the long-term performance of reforming systems to prevent catalysts from coking. There have been a number of studies to improve resistance against carbon formation (carbon resistance) by adding promoters and changing the active phase. In particular, in this chapter, some studies for preventing the formation of whisker carbon are described. For several years, CeO2, which has come under intense scrutiny as a structural promoter, has been added to Al2O3, and the CeO2Al2O3 has been proven to work effectively in catalysts. By storing and releasing oxygen reversibly, CeO2 increases carbon oxidization electronically. It may also improve dispersion of the active phase physically. High carbon oxidization activity prevents the formation of whisker carbon. However, CeO2 addition is not always beneficial. Ni/CeO2 has shown low activity in catalysts, so it is necessary to control the optimum content of CeO2 [20]. In addition, in Ce0.8Gd0.2O2x, oxygen ion vacancies are generated, which improve reforming activity. It has been reported that the interaction between Ce0.8Gd0.2O2x and Pt makes highly active O2, which begins to increase the reforming reaction rate on the catalyst surface via a synergetic effect with the supporter-metal active phase [21]. Another way to decrease carbon deposition that has been shown in the literature is by adding alkali promoters [22,23], such as calcium (CaOAl2O3) and potassium (K2Al2O4). It is well known that carbon formation is affected by the acidity of the surface. Positively charged acidic sites promote carbon deposition. Alkali-doped alumina supports reduce acidic sites by forming

154

Solid Oxide Fuel Cell Lifetime and Reliability

hydroxide, which increases the rate of the carbon gasification reaction. The steam ratio (H2O/CnHm) can be reduced by B16% compared to an undoped alumina [22]. This result implies that a sufficient steam ratio can reduce the whisker carbon formation, which is the most destructive form of carbon in SR over a nickel catalyst. However, a potassium promoter can decrease the catalyst’s activity, corroding the catalyst due to the desorption of potassium [23].

LIQUID FUEL PROCESSOR DESIGNS TO ENHANCE RELIABILITY Component of a Liquid Fuel Processor for Solid Oxide Fuel Cells A liquid fuel processor consists of the following sequences: (1) fuel delivery, (2) fuel reforming, and (3) reformate cleaning. (1) Fuel delivery refers to the overall process of fuel atomization, evaporation, and mixing with other oxidants before reaching the reforming catalyst. In contrast to gas fuel processors, it is important for a liquid fuel processor to make a proper mixture of fuel and oxidant before contacting the catalysts to ensure a high long-term reforming performance. The detailed information of fuel delivery is contained in the Fuel Delivery Design of Liquid Fuel Processing section. (2) In the fuel reforming process, the hydrocarbons of the liquid fuels are converted into hydrogen-rich gases as described in the Stages of Fuel Processing section. (3) The reformate cleaning process is necessary to convert or remove not only sulfur compounds but also light hydrocarbons in the reformate. Thus, the postreforming process should also be introduced to remove impurities before SOFCs. The detailed information of this part is described in the Postprocessing in Reforming to Enhance the Lifetime of Solid Oxide Fuel Cells section. Only when these parts are well-integrated, a complete liquid fuel processor can fully function, which makes it possible to drive SOFCs. An example of an integrated liquid fuel processor for SOFCs is shown in Fig. 8.8 [24]. This type of a fuel processor was introduced for the stable operation of SOFCs when diesel fuel is used. This fuel processor specifically contains the already mentioned sequences: (1) fuel delivery part, (2) fuel reforming part (especially ATR method), and (3) two stages of reformate cleaning parts that consist of adsorptive desulfurization and postreforming. From this fuel processor, H2S and light hydrocarbons were completely removed, and H2, CO, and CH4 that could be used for SOFC were successfully produced.

Design Factors for Fuel Processor In the design of the reformer, there are several variables that affect the performance of the reformer. In order to increase the reaction rate of the reformer, the mixing ratio of the reactants can be adjusted, or the reaction speed can be controlled using a high-performance catalyst.

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Chapter | 8

155

FIGURE 8.8 Schematic diagram of the integrated diesel fuel processor [24].

Improving the catalyst to have high activity and H2 selectivity should be the first step to improve the reforming performance. Increasing the yield of hydrogen is also important to minimize the carbon deposition that occurs during reforming. Carbon deposition will occur frequently not only during steady state operation but even during the reactor start-up. Therefore, in order to suppress the carbon deposition, the reforming catalyst must have high resistance. Reformer design parameters are presented in Table 8.4. The durability of an SOFC is favored by the high hydrogen partial pressure of the reformate gas. Therefore the reformer should be operated at conditions for increasing the hydrogen yield.

Fuel Delivery Design of Liquid Fuel Processing For successful liquid fuel processing, it is crucial to properly design the fuel delivery because the degree of mixing between the reforming reactants (liquid fuel and gaseous oxidants, such as air and steam) has a critical role on the downstream reforming performance. Incompletely mixed reactants can cause detrimental deposit formation and/or has a high risk for auto-ignition of the mixture that can significantly damage the reforming catalyst. Therefore it is important to achieve sufficiently complete mixing upstream of the reforming catalyst. However, liquid fuels, which consist of various mixture of hydrocarbons, have a wide range of boiling points, making evaporation and mixing of the fuel difficult challenges.

156

Solid Oxide Fuel Cell Lifetime and Reliability

TABLE 8.4 Design Factors of the Reformer Factor

Durability in SOFC

Note

Temperature

Increases at higher temperature

Thermodynamically, H2 and CO are produced above 600 C

H2O/C

Increases at higher H2O/C

H2O suppresses carbon deposition in the catalyst and increases the hydrogen production

O2/C

Decreases at higher O2/C

Oxygen effectively decomposes the aromatic compounds; however, high O2/C value reduces the hydrogen production

Gas hourly space velocity (GHSV)

Decreases at higher GHSV

GHSV is the flow rate of the volume of the catalyst. It is related to the residence time of the gas

Pressure

Depends on kinetics and thermodynamic principle

When increasing the pressure, the methanation reaction is thermodynamically increased. In addition, it is known to increase the kinetics of the catalysis

FIGURE 8.9 Diesel injection by UI [25].

To achieve a high degree of mixing, a nozzle is widely used because atomizing liquid fuel can significantly improve the mixing quality of reactants. Consequently, many studies about various nozzles were conducted to enhance the degree of mixing for their purpose. The ultrasonic injector (UI) was first suggested by Kang et al. for atomizing liquid fuel, especially in a diesel fuel processor [25]. The details of the UI are provided in Fig. 8.9 and Table 8.5. The UI can atomize liquid fuel into small droplets (B40 μm) using a piezo-electric transducer, which allows fuel droplets

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Chapter | 8

157

TABLE 8.5 Specifications of the UI Range of frequency

20120 kHz

Power consumption (max.)

22 W intermittent 15 W continuous

Droplet size

40 μm (average)

FIGURE 8.10 Schematic of twin-fluid nozzle.

to effectively penetrate into gas-phase oxidants. It was reported that the reforming efficiency (i.e., the efficiency based on the lower heating value between the input fuel and output H2 and CO) was increased by about 20% when compared with the case without the UI under the same operating conditions. A pressure-swirl nozzle was also introduced for atomizing liquid fuel [26]. This type of nozzle has a merit of robust characteristics to high temperature as well as a simple design consideration. This is because this type of nozzle does not require an additional electrical component such as a piezo-electric transducer or an additional supply of carrier gas. However, an appropriate condition of supply pressure should be met for acquiring fine droplet size. A twin-fluid nozzle can be applied in atomizing liquid fuel. The schematic of a twin-fluid nozzle is shown in Fig. 8.10. This nozzle can atomize the liquid fuel into small droplets in the range of 1050 μm using the force of the flow rate of supplied gases. When using this type of nozzle, there is no need for a high-pressure condition to drive the pressure nozzle and/or additional electrical power consumption. It was reported that a kWe-class fuel processor was successfully operated for B1000 h, even with commercial kerosene or diesel fuels with using a twin-fluid nozzle [24,27].

158

Solid Oxide Fuel Cell Lifetime and Reliability

In addition to nozzle studies, there are several methods to enhance the degree of mixing. For example, shape or structure-related mixer design is used for enhancing turbulent or recirculation flows, which can enhance the degree of mixing [26,28]. Furthermore, preheating the reactants is also applicable for accelerating the evaporation of liquid fuels [29]. To sum up, optimization of fuel delivery design has to be comprehensively considered to enhance the degree of mixing of reactant, which makes it possible to guarantee the stable operation of the entire liquid fuel processor.

kW-Class Reformer for Reliable Solid Oxide Fuel Cell System For development of SOFC systems operated with liquid fuel, a kW-class reformer is required. However, the design of the kW-class reformer causes some technical issues, which do not occur in a small reformer system. In kWclass systems, pressure drop is an important issue. Generally, for small reformers, pellet-type catalysts are loaded into the reformer. However, for large systems, pellet-type catalysts are not appropriate because of a high flow rate and the resultant pressure drop. This pressure difference inside the reformer can collapse the catalyst zone, which is one of the reasons for reformer failure. In kW-class systems, temperature distribution is also an issue. To manufacture kW-class reformers, a lab-scale microreactor cannot be used, and this size-up of the reactor causes a thermal distribution issue. If a microreactor is used, it can be assumed that the temperature distribution of the catalysts bed is uniform. However, in the case of large reactors for the kW-class reformer, a temperature difference is observed along diameter. For example, if the reforming reaction is endothermic, the temperature is higher near the center of the reactor. In contrast, if the reforming reaction is exothermic, the temperature is lower near the center. This nonuniform temperature distribution can cause carbon deposition because of extremely low or high temperature caused by the nonuniformity. In addition, for mobile applications such as transportation and portable systems, pellet-type reforming catalysts cannot be used because of vibration. Catalytic active sites on the catalysts’ surface can be reduced, and the catalysts themselves can even be broken down by constant vibration. This causes serious degradation of the catalysts’ performance. To improve the durability of kW-class reformers, monolith-type reforming reactors have to be introduced for large SOFC systems. In the case of monolith-type reformers, catalysts are coated on the surface of monolith channels as shown in Fig. 8.11. Therefore the input gas can flow along the channels without differential pressure, and the gas can diffuse into the thin catalyst layer easily. Furthermore, monolith-type reformers have a continuous heat conduction path, which is better for a uniform temperature distribution. In mobile applications, the monolith absorbs vibrations, which damage reforming catalysts and systems. Therefore, monolith-type reformers have to be used for highly durable reformer systems.

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Chapter | 8

159

FIGURE 8.11 Reformer with a monolith-type reactor and a cross-section SEM image of a catalyst-coated monolith channel [30].

FIGURE 8.12 The world’s first 1-kW class diesel reformer (RA-10DS, H&Power, Inc.) [30].

The world’s first commercialized kW-class reformer was developed by H&Power, Inc. as shown in Fig. 8.12. Diesel reformers were operated for 2500 h of long-term durability testing as shown in Fig. 8.13.

POSTPROCESSING IN REFORMING TO ENHANCE THE LIFETIME OF SOLID OXIDE FUEL CELLS Concept of Postreforming After the reforming reaction process occurs, there might be light hydrocarbons from the reforming reaction of heavy hydrocarbons. The usual forms of light hydrocarbons observed are carbon numbers 13. It is claimed that a small existence of light hydrocarbons (C2C4 hydrocarbons) could accelerate the degradation of the catalyst in the reformer due to coke formation.

160

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 8.13 Stable operation of a diesel reformer for 2500 h [30]

From this coke formation, the catalytic reaction could be damaged and the temperature profile in the reformer could be unstable due to the nonuniform temperature distribution. As a result, the performance of the reformer could drop as shown in Fig. 8.14 [8]. Especially in SOFC applications, ethylene, one of the light hydrocarbons present after reforming, could not only induce severe damage in the reformer catalyst, but it could also act as a source of carbon deposition on Ni-based SOFC anodes. This phenomenon occurs even when a small amount (B0.1 vol.%) is present in the reformate gas entering the SOFC anode. From this carbon deposition on anode surfaces, the electrical performance of the SOFC system could be critically degraded. Since the concentration of ethylene was increasing incrementally with the operation time of the ATR reformer, it seems like a special treatment is needed to reduce the amount of ethylene. After conducting experiments in this area, Professor Bae’s research group at the Korea Advanced Institute of Science and technology (KAIST) has found a way to suppress the production of light hydrocarbon complexes using an UI and the enhanced mixer design [25,31]. However, after longterm operation of the reformer, light hydrocarbons have been found on the samples. Therefore, it is difficult to suppress the formation of light hydrocarbons in the reformer, so the concept of postreforming is introduced to convert light hydrocarbons into reformate off-gases such as H2, CO, CO2, and CH4 [9] (Fig. 8.15).

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Chapter | 8

161

FIGURE 8.14 Lower hydrocarbon distribution and efficiency vs. operating time during ATR of synthetic diesel [8].

FIGURE 8.15 Diesel reforming with a postreformer.

The advantages of postreforming rely on the fact that it can decompose light hydrocarbons into mono-carbon products and hydrogen. The problem of carbon coking in SOFCs, which decreases cell performance due to the suppression of the catalyst reaction by carbon, is remedied because of the postreforming process that can decompose the light hydrocarbons. Moreover, this could be considered as a reforming process with light hydrocarbons in order to make fuel elements for operating SOFCs [7]. The resulting gases from the postreforming process such as H2, CO, and CH4 can then act as fuels in SOFCs. Thus, the fuel utilization could be increased because more fuel is produced in the postreforming part. In conclusion, postreforming will decompose light hydrocarbons, and this will lead to lower carbon coke formation in SOFC anodes, increase fuel utilization, and long-term durability.

Desulfurizer for Heavy Hydrocarbons Sulfur compounds are among the most common impurities in crude oil. Nowadays, oil refineries remove the sulfur by extracting it out of the liquid phase since this process is difficult to achieve in small systems.

162

Solid Oxide Fuel Cell Lifetime and Reliability

Sulfur removal is a crucial process not only because of the environment but also because its presence in liquid hydrocarbons causes negative effects in Pt catalysts in ATR reactors.

Sulfur-Containing Hydrocarbon Source Production of clean hydrocarbon fuels has seen an increase in demand in recent years. Therefore there is strong pressure on regulations from many national and international protocols, e.g., the European Emission Standards and the California Air Resources Board (CARB) Standards. Sulfur in hydrocarbons is present throughout the refining crude oil process, starting from its extraction to final fuel product. [32]. The average sulfur content varies from 0.03 to 7.89% by mass in crude oil [33], and it has become a requirement for refineries to reduce sulfur contents or to produce an almost sulfur-free petrochemical product [32]. Either way, diesel or gasoline must at the end comply with the fuel quality set by each country’s standards: Environmental Protection Agency (EPA) in the USA, European Standards Organization (CEN) in the EU. Currently, many countries are following the leading model from the European Standards (Table 8.6).

TABLE 8.6 CEN Standards. Three Standards Covered Automotive Fuel Quality: EN 590 for Diesel, EN 228 for Gasoline, and EN 589 for Automotive LPG [34] EU Fuel Specifications for Sulfur Content Name

CEN Standard

Implementation Date

Sulfur Limit (ppm)

n/a

EN 590: 1993,

1994

2000



1996

500 (diesel)

EN 590: 1999,

2000

350 (diesel),

EN 228: 1993 Euro 2 Euro 3

EN 228: 1999 Euro 4

EN 590: 2004,

150 (gasoline) 2005

50a

EN 228: 2004 Euro 5

EN 590: 2009

2009

10,10b

Euro 6

EN 590: 2014,

2014

10

EN 228: 2014 a

10 ppm fuel must be available. Nonroad fuel limit.

b

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Chapter | 8

163

Some of the fuels that nowadays have a concerning presence of sulfur are heavy fuel oil used in international shipping with an average of 1000 ppm [35] and jet fuels with an average of 400800 ppm [36]. This is pushing for the need of alternatives. Currently, fuels are going in the direction of ultra-low sulfur (ULSD) due to the previously mentioned strict regulations, but this also means a decrease in 12% lower fuel economy and a more expensive oil process [37]. Nevertheless, ULSD with a sulfur content of 15 ppm or less will decrease soot in engines [38]. The use of “clean diesel” will definitely shift current market, and countries saving millions of dollars in barrels of oil per day while at the same time allowing lower harmful exhaust emissions (Fig. 8.16).

FIGURE 8.16 Desulfurization timelines showing national regulations [39].

Catalysts for Desulfurization HDS process is one of the ways to remove sulfur compounds inside the hydrocarbon fuels. Molybdenum and tungsten with nickel or cobalt

164

Solid Oxide Fuel Cell Lifetime and Reliability

promoters were discovered as HDS catalysts in the 1920s and the 1930s in Germany [40]. The HDS catalysts are heterogeneous materials that are commonly referred as “promoted Mo or W catalysts,” e.g., the most commonly employed HDS catalyst is a “cobalt-promoted” molybdenum catalyst (CoMo), which comprises a mixture of MoS2 and Co9S8 supported on alumina (Al2O3) [34]. Chemical or physical adsorption of sulfur components is also used for desulfurization of the reformate, especially in gaseous state. The sulfur compounds inside the reformate were converted into H2S during the reforming process due to the high reaction temperature. Adsorbents such as ZnO catalysts are used after the reforming process to remove residual sulfur inside the reformate [24]. Eq. (8.4) shows the desulfurization process over ZnO catalysts. The ZnO catalyst bed can be introduced into small-sized reforming reactors and enhance the durability of the SOFC stack. H2 S 1 ZnO-H2 O 1 ZnS

ð8:4Þ

LIFETIME ESTIMATION OF FUEL PROCESSORS Fuel processors consist of various catalytic reactors (fuel reformer, postreformer, desulfurizer, etc.) and BOPs (pump, heat exchanger, etc.). In previous sections, several mechanisms that can explain why the catalysts degrade during operation are described. In addition, a correlation between the BOPs and durability of the fuel reformer has been explained briefly. This section contains details of widely used lifetime evaluation methods and lifetime extension methods to secure the long-term stability of fuel processors.

Engineering Issues (BOPs) of Fuel Processors on Durability As mentioned in the Introduction to Fuel Processing in Solid Oxide Fuel Cells section, BOPs affect the lifetime of a fuel processor. Especially, failure of mechanical BOPs may cause severe damage on a fuel processor by leading to uneven temperature, fuel, H2O/C, and O2/C distribution in the catalytic reaction region. As already explained, temperature, fuel, H2O/C, and O2/C are the major operating conditions of fuel processor. Therefore these uneven distributions of operating conditions induce catalyst degradation such as carbon formation and mechanical failure by a local temperature overshoot. Fig. 8.17 shows the mechanical failure by the local temperature overshoot due to an uneven fuel distribution [41]. In addition, irregular operation of mechanical BOPs also has effects on the durability of fuel reformer. Fluid machinery such as liquid pumps, gas blower, and fuel injectors inevitably has pulsating motion, and this is directly related to the fluctuation of H2O/C, O2/C, and fuel ratio in the catalytic

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Chapter | 8

165

FIGURE 8.17 Mechanical failure of a monolith-type liquid fuel reforming catalyst [41].

FIGURE 8.18 Fluctuation of O2/C due to pulsating flow of fuel and air [42].

reaction region [42]. Fig. 8.18 shows the fluctuation of O2/C due to the pulsating flow of fuel and air in a gasoline fuel processor. To moderate the fluctuation of operating conditions by the pulsating motion of fluid machinery, a flow straightener or attenuator components are required. Ji and Bae [42] took advantage of a monolith catalyst and a catalytic reaction chamber as a flow straightener and an attenuator, respectively, to suppress soot generation during the start-up process. Fig. 8.19 shows the suppression of soot generation by moderating fluctuation. (A) shows the strong soot generation by igniting fluctuating fuel, and (B) shows the weak soot generation by igniting fuel after installing the flow straightener and attenuator. Therefore, to enhance the durability of fuel processor, consideration of irregular operation of BOPs is always important.

166

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 8.19 Soot generation on the water surface when the glow-plug position was (A) at the front or (B) at the rear of the ATR catalyst [43].

Durability Test Method for Fuel Processors Degradation of catalysts has a significant impact on the conversion rate of catalytic reactors, and the rate decreases along with the passage of time. Fuel reformers are also catalytic reactors, and catalysts degrade during operation. Therefore, to evaluate the durability of the catalyst, the conversion rate reduction by long-term or accelerated degradation tests needs to be determined. Definitions of fuel and H2S conversion rates are expressed in Eqs. (8.5) and (8.6), respectively [44,45]. Fuel conversion ð%Þ 5 atomic carbon concentration of CO; CO2 ; and CH4 in the reformate 3 100 atomic carbon concentration in the fuel ð8:5Þ H2 S conversion ð%Þ 5 moles ofH2 S in inlet gas 2 moles ofH2 S in outlet gas 3 100 moles ofH2 S in inlet gas ð8:6Þ It is possible to calculate fuel conversion using gas chromatography (GC) analysis, and long-term testing can accumulate fuel conversion data to evaluate the fuel conversion reduction over time. As a result, the degradation rate of the fuel processor is expressed in Eq. (8.7). Degradation rate ð%=hÞ 5

ðinitial conversionÞ 2 ðfinal conversionÞ operation time

ð8:7Þ

Fig. 8.20 shows accelerated degradation test results according to changes in catalyst composition, and the results are listed in Table 8.7. As a result, by calculating the degradation rate, it is possible to sort which catalyst can be used for fuel reformers [46].

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Chapter | 8

167

FIGURE 8.20 The stabilities of the various Ni/CGO compositions (20.0, 40.0, and 60.0 wt.%) and C11PR catalysts during accelerated degradation tests [46].

TABLE 8.7 Degradation Rate of Various Ni/CGO Compositions (20.0, 40.0, and 60.0 wt.%) and C11PR Catalysts During Accelerated Degradation Tests Catalyst

C11PR

10 wt.% Ni

20 wt.% Ni

40 wt.% Ni

60 wt.% Ni

Degradation rate (%/h)

25.2

25.8

1.4

2.9

1.9

Practical Example of Durability Test The Agency for Defense Development of South Korea (ADD) and H&Power Inc. have developed fuel processors for the auxiliary power unit of commercial vehicles. Commercial gasoline was used as the fuel, and the composition of the reformate gas was analyzed by gas chromatography. The operating conditions were H2O/C 5 2.7 and O2/C 5 0.5. For analyzing the durability during onoff operation, the compositions of the reformate gas were calculated after repeating the onoff situation. As shown in Fig. 8.21, the degradation rate was calculated from the decrease in the reforming efficiency, and the degradation rate was 1.7% after 100 onoff cycles.

Lifetime Extension of Fuel Processors For stable operation and cost reduction, the lifetime of fuel processors is very important. As mentioned earlier, reforming catalysts are deactivated by coke formation on their surface. The catalytic activity of the catalyst can be

168

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 8.21 Durability test of a gasoline reformer (ADD and H&Power Inc., South Korea)

FIGURE 8.22 Long-term test of Ni-Ru/CGO (diesel SR, reactor restarted at 1320 h, KAIST) [46].

restored by burning off the coke with an oxygen/nitrogen mixture [47], and this process is called “regeneration.” Lee et al. [46] conducted long-term tests on diesel SR for 2000 h. In Fig. 8.22, the hydrogen yield rate gradually decreased after 1000 h, and this phenomenon was caused by coke formation on the catalyst surface. At 1350 h, the catalyst was regenerated at 800 C, and the regenerated catalyst showed B700 h of stable operation.

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Chapter | 8

169

Including the reforming catalyst, fuel processors consist of various components. If the fuel processor is fabricated in a single module, the fuel processor must be disassembled to replace the dysfunctional parts when there are any failures or malfunctions. To avoid this problem, modular systems can be applied to the fuel processor. By integrating each module into a single process, the dysfunctional parts can be easily replaced, and maintenance for stable operation can be performed simply.

CONCLUSION In this chapter, fuel processing and its design factors for SOFC application were illustrated. The quality of the reforming process affects the durability and reliability in SOFCs. To enhance the performance of the fuel processing, extensive studies on reforming catalysts, reactor design, and fuel delivery methods were conducted. Also, postprocessing steps were introduced to remove undesired products that are problematic for SOFCs. Postreforming and desulfurization were implemented after the reforming process to achieve long-term performances of SOFCs. In addition, engineering issues such as fluctuations of fluid machinery also limit the reliability of SOFCs. Therefore, to maintain a longer lifetime of SOFCs, performance and durability need to be ensured even from the fuel processing steps.

REFERENCES [1] Steele Brian CH, Heinzel Angelika. Materials for fuel-cell technologies. Nature 2001;414:34552. [2] Holladay Jamelyn D, Hu Jianli, King David L, Wang Yong. An overview of hydrogen production technologies. Catal Today 2009;139(4):24460. [3] Yang Ralph T, Herna´ndez-Maldonado Arturo J, Yang Frances H. Desulfurization of transportation fuels with zeolites under ambient conditions. Science 2003;301(5629):7981. [4] Chunshan Song. An overview of new approaches to deep desulfurization for ultra-clean gasoline, diesel fuel and jet fuel. Catal Today 2003;86(1-4):21163. [5] Ma Xiaoliang, Sprague Michael J, Song Chunshan. Deep desulfurization of gasoline by selective adsorption over nickel-based adsorbent for fuel cell applications. Ind Eng Chem Res 2005;44(15):576875. [6] Herna´ndez-Maldonado Arturo J, Yang Ralph T. Desulfurization of transportation fuels by adsorption. Catal Rev 2004;46(2):11150. [7] Yoon Sangho, Bae Joongmyeon. A diesel fuel processor for stable operation of solid oxide fuel cells system: I. Introduction to post-reforming for the diesel fuel processor. Catal Today 2010;156(1-2):4957. [8] Yoon Sangho, Kang Inyong, Bae Joongmyeon. Effects of ethylene on carbon formation in diesel autothermal reforming. Int J Hydrogen Energy 2008;33(18):47808. [9] Yoon Sangho, Kim Yongmin, Kim Sunyoung, Bae Joongmyeon. Effects of low hydrocarbons on the solid oxide fuel cell anode. J Solid State Electrochem 2010;14:1793800. [10] Rostrup-Nielsen JR, Sehested J, Nørskov JK. Hydrogen and synthesis gas by steam and CO2 reforming. Adv Catal, 47. 2002. p. 65139.

170

Solid Oxide Fuel Cell Lifetime and Reliability

[11] Wanke SE, Flynn PC. The sintering of supported metal catalysts. Catal Rev 1975;12: 93135. [12] Ruckenstein E, Pulvermacher B. Kinetics of crystallite sintering during heat treatment of supported metal catalysts. AIChE J 1973;19:35664. [13] Wynblatt P, Gjostein NA. Supported metal crystallites. Prog Solid State Chem. 1975;9: 2158. [14] Calvin H. Bartholomew, mechanisms of catalyst deactivation. Appl Catal Gen 2012;212: 1760. [15] Sehested J. Four challenges for nickel steam-reforming catalysts. Catal Today 2006;111: 10310. [16] Jiao Zhenjun, Takagi Norikazu, Shikazono Naoki, Kasagi Nobuhide. Study on local morphological changes of nickel in solid oxide fuel cell anode using porous Ni pellet electrode. J Power Sources 2011;196:101929. [17] Boldrin Paul, Ruiz-Trejo Enrique, Mermelstein Joshua, Menendez Jose´Miguel Bermudez, Reina Tomas Ramırez, Brandon Nigel P. Strategies for carbon and sulfur tolerant solid oxide fuel cell materials, incorporating lessons from heterogeneous catalysis. Chem Rev 2016;116:1363384. [18] David L. Trimm, The Formation and Removal of Coke from Nickel Catalyst, Catalysis Reviews 1977;16(1):15589. [19] Calvin H. Bartholomew, Carbon Deposition in Steam Reforming and Methanation, Catalysis Reviews 1982;24(1):67112. [20] Wang Shaobin, Lu GQM. Role of CeO2 in Ni/CeO2Al2O3 catalysts for carbon dioxide reforming of methane. Appl Catal B: Environ 1998;19:26777. [21] Kang I. Study on performance of diesel autothermal reformer for solid oxide fuel cell. PhD dissertation 2006. KAIST. [22] Carlsson M. Carbon formation in steam reforming and effect of potassium promotion. Johnson Matthey Tecnol Rev 2015;59(4):31318. [23] Stołecki K, Dmytrzyk J, Kotarba A. Studies of potassium-promoted nickel catalysts for methane steam reforming: effect of surface potassium location. Appl Surf Sci 2014;300:191200. [24] Yoon S, Lee S, Bae J. Development of a self-sustaining kWe-class integrated diesel fuel processing system for solid oxide fuel cells. Int J Hydrogen Energy 2011;36:1030210. [25] Kang I, Bae J, Yoon S, Yoo Y. Performance improvement of diesel autothermal reformer by applying ultrasonic injector for effective fuel delivery. J Power Sources 2007;172: 84552. [26] Porˇs Z, Pasel J, Tschauder A, Dahl R, Peters R, Stolten D. Optimised mixture formation for diesel fuel processing. Fuel Cells 2008;8:12937. [27] Yoon S, Bae Y, Kim S, Yoo Y. Self-sustained operation of a kW(e)-class kerosenereforming processor for solid oxide fuel cells. J Power Sources 2009;192:3606. [28] Lindstrom B, Karlsson J, Ekdunge P, Verdier L, Haggendal B, Dawody J, et al. Diesel fuel reformer for automotive fuel cell application. Int J Hydrogen Energy 2009;34: 336781. [29] Pasel J, Meibner J, Porˇs Z, Samsun R, Tschauder A, Peters R. Autothermal reforming of commercial Jet A-1 on a 5kWe scale. Int J Hydrogen Energy 2007;32:484758. [30] H&Power Inc. Products: Fuel Reformer [Internet]. Available from: http://www.hnpower. co.kr/ [31] Kim S, Dean Anthony M, Bae J. Coupled transport and kinetics in the mixing region for hydrocarbon autothermal reforming. Int J Hydrogen Energy 2013;38:1614051.

Fuel Processor Lifetime and Reliability in Solid Oxide Fuel Cells Chapter | 8

171

[32] Javadli E, de Klerk A. Desulfurization of heavy oil. Appl Petrochem Res 2012;1:319. [33] Mehran S, Amarjeet B, Argyrios M. Biodesulfurization of refractory organic sulphur compounds in fossil fuels. Appl Petrochem Res 2007;25:57096. [34] Chambliss S, Miller J, Fac¸anha C, Minjares T, Blumberg K. The impact of stringent fuel and vehicle standards on premature mortality and emissions, 3. Washington (DC): International Council on Clean Transportation; 2013. p. 12. [35] United States Environmental Protection Agency. EPA guidance on ECA marine fuel. EPA-420-B-14-097. United States: U.S. Environmental Protection Agency; 2014. [36] Gilmore Christopher K, Barrett Steven RH, Yim Steve HL, Murray Lee T, Kuhn Stephen R, Tai Amos PK, et al. PARTNER-COE-2011-006. Project: environmental costbenefit analysis of ultra low sulfur jet fuel. United States: Partnership for AiR Transportation Noise and Emissions Reduction; 2011 [37] U.S. Department of Energy. Ultra-low sulfur diesel. https://www.fueleconomy.gov/feg/ lowsulfurdiesel.shtml (accessed 2016). [38] Miller JD, Fac¸anha C. The state of clean transport policy: a 2014 synthesis of vehicle and fuel policy developments. Washington (DC): International Council on Clean Transportation; 2014. [39] Omidvarborna H, Kumar A, Kim D-S. Characterization of particulate matter emitted from transit buses fueled with B20 in idle modes. Appl Petrochem Res 2014;2:233542. [40] Donath EE. History of catalysis in coal liquefaction, catalysis science and technology, vol. 3. Berlin: Springer; 1982. p. 138. [41] Unpublished result from Argonne National Lab (ANL). [42] Ji H, Bae J. Start-up and operation of gasoline fuel processor for isolated fuel cell system. J Energy Eng 2016;25:7685. [43] Ji H, Bae J, Cho S, Kang I. Start-up strategy and operational tests of gasoline fuel processor for auxiliary power unit. Int J Hydrogen Energy 2015;40:410110. [44] Yoon S, Bae J, Lee S, Pham TV, Katikaneni SP. A diesel fuel processor for stable operation of solid oxide fuel cells system: II. Integrated diesel fuel processor for the operation of solid oxide fuel cells. Int J Hydrogen Energy 2012;37:922836. [45] Kang I, Bae J, Bae G. Performance comparison of autothermal reforming for liquid hydrocarbons, gasoline and diesel for fuel cell applications. J Power Sources 2006;163:53846. [46] Lee S, Bae M, Bae J, Katikaneni SP. NiMe/Ce0.9Gd0.1O22x (Me: Rh, Pt and Ru) catalysts for diesel pre-reforming. Int J Hydrogen Energy 2015;40:320716. [47] Ren XH, Bertmer M, Stapf S, Demco DE, Blumich B, Kern C, et al. Deactivation and regeneration of a naphtha reforming catalyst. Appl Catal A: Gen 2002;228:3952.

Chapter 9

Life and Reliability of Solid Oxide Fuel Cell-Based Products: A Review Subhasish Mukerjee1, Rob Leah1, Mark Selby1, Graham Stevenson2 and Nigel P. Brandon2 1

Ceres Power Ltd, Horsham, United Kingdom, 2Imperial College London, London, United Kingdom

Chapter Outline Introduction SOFC Technology Generations and Applications Generic Durability/Reliability Issues for SOFC Durability and Reliability Strategies Adopted for Solid Oxide Fuel Cell Development

173 173 176

The Japanese Programs—NEDO and ENE-Farm Ceres Power’s Stack Performance Verification Program LG Fuel Cell Systems Summary and Conclusions References

179 185 186 187 188

179

INTRODUCTION SOFC Technology Generations and Applications Solid oxide fuel cell (SOFC) products are being developed worldwide for multiple high-efficiency power generation applications with different requirements, primarily based on the use of natural gas as a fuel. The generic requirements for different SOFC applications are presented in Table 9.1. SOFC technology has been under development since the 1960s. Early SOFC concepts such as Westinghouse tubular cells, demonstrated excellent durability at the expense of cost and low volumetric power density. Some developers (notably Mitsubishi-Hitachi Power Systems and LG Rolls-Royce) are still developing segmented-in-series (SIS) tubular cells, mostly because they are easy to pressurize for integration with a gas turbine cycle. More recently most SOFC developers have moved to planar cell/stack designs, Solid Oxide Fuel Cell Lifetime and Reliability. DOI: http://dx.doi.org/10.1016/B978-0-08-101102-7.00009-X Copyright © 2017 Elsevier Ltd. All rights reserved.

173

174

Solid Oxide Fuel Cell Lifetime and Reliability

TABLE 9.1 Typical Applications for SOFC and Life Requirements Application Type

Power Requirement

Steady-State Life (h)

On Off Thermal Cycling (Number)

MicroCHP [1]

1 kW class

90,000

.100

Commercial buildings [2]

5 kW class

90,000

.100

Large power plant [1]

100 kW—MW class

.40,000

50

APU and range extenders [3]

5 30 kW class

.10,000

.1000

which offer much higher volumetric power density than tubular designs and the potential for lower cost if the majority of the stack can be made of metallic rather than ceramic components. It has however proved challenging to make planar SOFC stacks with the durability and reliability required for many practical applications due to issues with corrosion of metallic components, volatilization of chromium species and the requirement to seal ceramic cells to metallic interconnect plates. Planar SOFC designs fall into three technology generations, with the general tendency being toward lower operating temperatures with each successive generation, as this tends to facilitate the use of lower cost materials in the stack and system. Developers are still working on all three generations of planar SOFC technology with products based on all of them being demonstrated. The three generations of planar SOFC technology are: G G G

Electrolyte-supported cells (ESCs) Anode-supported cells (ASCs) Metal-supported cells (MSCs)

Table 9.2 presents some of the key features of these different generations of SOFC technology and their attributes. The different planar SOFC technologies ESC, ASC, and MSC have evolved over time and the older ESC technology has matured the most regarding reliability and durability, followed by ASC, which is operated at a lower temperature than ESC and then MSC, which is often termed as the “next generation” SOFC. Currently, ESCs have been demonstrated for more than 40,000 h while some ASCs are approaching 40,000 h, and MSCs have been demonstrated for tens of thousands of hours. Even at this early stage of development, MSCs have been demonstrated to be more robust to thermal cycling or on off cycles and promise to have an inherent cost advantage over other technologies.

TABLE 9.2 Types of SOFC and Their Key Characteristics Type of SOFC

Key Feature

Operating Temperature ( C)

Life Capability (Steady State Operation Demonstrated)

On Off Cycling Capability (Thermal Cycling)

Cost Projected

Tubular

Stacks made up of bundles of ceramic tubes

900 1000

High

Low, very long start-up times

High

ESC

Ceramic electrolyte (Zirconia based) as the main support

850 1000

High

Low

Medium to high

ASC

Ceramic anode (nickel zirconia) as the main support

700 800

Medium high

Medium

Medium to low

MSC

Metal substrate as support for thin ceramic layers

500 800

Low medium

High

Medium to low

176

Solid Oxide Fuel Cell Lifetime and Reliability

Generic Durability/Reliability Issues for SOFC SOFC reliability and durability issues can broadly be divided into three categories, intrinsic degradation, extrinsic degradation, and cyclic degradation, with common failure modes across all SOFC types [4]. For stationary applications the former two are most important, whereas for other applications such as auxiliary power units (APUs) cyclic degradation is critical. Examples of failure modes associated with each category of performance degradation are listed in Table 9.3. The table summarizes the key mechanisms associated with aging of different components within SOFCs that are currently being researched globally. A review of the scientific understanding of these mechanisms shows that significant progress has been made. However, using mechanistic understanding for predictive analysis for reliability and product life requires more research in many areas. As an example, considerable research effort has been put in understanding the mechanism of chromium-based poisoning of cathodes and potential mitigations using coatings of metallic parts. By contrast significantly less development work has been done to understand and mitigate the long-term effects of other poisons in the cathode such as SO2 or silica. Table 9.3 summarizes the authors’ assessment of the areas that are currently mature, and the areas that need more development in mechanistic understanding to be useful for reliability development of SOFC products. This includes strontium segregation and other microstructural changes in cathodes, phase changes in electrolyte, prediction of sintering and microstructural changes in anode over time, poisoning of cathodes, redox-based failure over life and reliability issues related to manufacturing defects. A concerted focus on these areas that need development by the wider SOFC development community will allow for faster advancements in reliability of SOFC-based products. In addition focus on translating the knowledge of mechanisms to predictive accelerated testing and analysis will allow for shortening of the development cycle, for improvement in reliability and for validating new materials and designs. Until now the most common strategy adopted by most SOFC developers is to attempt to understand the relative contribution of different failure modes to their overall degradation, and then attempt to mitigate them sufficiently that the required lifetime and/or number of thermal cycles can be achieved. Intrinsic degradation mechanisms can be mitigated by modification of cell materials and/or microstructure. Extrinsic degradation mechanisms can be mitigated by modifications to the system upstream of the stack, for example, by aluminizing upstream components to reduce chromium evaporation into the air. Cyclic degradation mechanisms are often hard to mitigate, other than by placing limits on the number and rate of cycles stacks are intended to withstand over their lives. With life requirements of 5 10 years (40,000 80,000 h) for stationary applications [25], different companies have taken different approaches to

TABLE 9.3 Summary of SOFC Degradation Mechanisms and Failure Modes, and Scientific Progress Toward Life Prediction-Based Upon Them Type of Degradation

Component

Degradation Description

Progress Toward Detailed Mechanistic Understanding

Progress Toward Predictive Analysis of Life

Reference

Intrinsic degradation

Cathode

Strontium segregation in cathode materials in real life environment

Medium

Low

[5]

Densification of cathode in real life environment

Medium

Low

[6,7]

Manganese segregation

High

Medium

[6,7]

Kinetic de-mixing of cathode materials in real life environment

Low

Low

[5,8]

Electrolyte

Phase change in electrolyte

Medium

Medium

[9,10]

Manganese diffusion into YSZ electrolyte

High

Medium

[11]

Anode

Sintering of metallic phase in anode in real life environment

Medium

Medium

[10,12]

All cell layers

Reaction between materials in cell (like SrZrO3)

High

Low

[10,13]

Failure due to cracking induced by thermal stresses during steady-state operation

High

High

[14]

Oxide scale growth in metallic stack components

High

High

[15,16]

Creep of metallic components

High

Medium

[14]

Stack

(Continued )

TABLE 9.3 (Continued) Type of Degradation

Component

Degradation Description

Progress Toward Detailed Mechanistic Understanding

Progress Toward Predictive Analysis of Life

Reference

Extrinsic degradation

Anode

Poisoning of anodes by fuel contaminants (like sulfur)

High

Medium

[17]

Cathode

Poisoning of cathodes by air contaminants (like sulfur dioxide, silica, steam)

Medium

Low

[18,19]

Poisoning of cathodes by chromium from upstream BoP

High

Medium

[20]

Poisoning of cathodes from chromium from metallic stack components through contact

High

High

[20]

Stack seal

Failure of stack sealing due to thermal stresses

High

High

[21]

Cell

Failure of ceramic cell components due to thermal stresses during thermal cycling

High

High

[14]

Failure of cells due to redox cycling (expansion/contraction of Ni/NiO)

High

Medium

[22,23]

Manufacturing reliability, quality control, cell performance variability, etc.

Medium

Medium

[24]

Cyclic degradation

Other reliability issues

Cell, stack, system

Life and Reliability of Solid Oxide Fuel Cell-Based Products Chapter | 9

179

validate their products. The typical approach has been to test for as long periods of time in steady state in the laboratory or in field trials as possible before a launch. However, recent development has focussed on developing accelerated tests and “new validation strategies” to predict life based on “lean” testing and using mechanistic understanding of degradation and failure modes to predict life. In the following sections, we will discuss how companies and institutions are carrying out these kinds of validation strategies. The Japanese New Energy and Industrial Development Organization (NEDO) program, Ceres Power’s stack verification program, and work reported by LG Rolls-Royce under the US SECA program will be discussed in detail, other manufacturers employ similar approaches. Table 9.4 presents a summary of the most recent published data (as of August 2016) on SOFC reliability and durability from a range of developers and reflects the state-of-the-art. As far as possible the data are quoted based on full-scale stacks; a number of developers have quoted longer operating times and/or numbers of thermal cycles on short-stack or single cell tests, which may not be representative of real operating conditions. It is worth noting based on the data that although steady-state degradation rates consistent with stack operating lives of 40 90 kh have been achieved by some developers, there is no published data for .40,000 h operation so this currently remains an extrapolation.

DURABILITY AND RELIABILITY STRATEGIES ADOPTED FOR SOLID OXIDE FUEL CELL DEVELOPMENT The Japanese Programs—NEDO and ENE-Farm Overview The NEDO is a conglomeration operating under Japan’s Ministry of Trade and Industry (METI) spanning multiple companies and academic institutions. The aim of the NEDO program is the large-scale deployment of fuel cell (Proton exchange membrane fuel cell also known as PEMFC and SOFC) and hydrogen technology for distributed power generation. NEDO has been active since 1989 [1,13]. NEDO has led the Japanese ENE-FARM fuel cell microCHP demonstrator program. NEDO also runs more research-oriented programs under the heading “Technology development for promoting SOFC commercialization.” A subproject of this program “Fundamental Study for Rapid Evaluation Method of SOFC durability” involves the development of methodologies to enable the identification of degradation mechanisms and lifetime prediction of SOFC cells and stacks. Multiple manufacturers have provided cells and stacks to this project for evaluation. The aim of this project [39] is to demonstrate the feasibility of an SOFC operating life of 90,000 h (B10 years) based on extrapolation of relatively

TABLE 9.4 A Summary of Published Commercial Fuel Cell Specifications Company

SOFC Type

Stack/ System

Power Level (kW)

Steady-State Number of Hours Reported

Steady-State Degradation Rate (%/kh)

Max Number of Thermal Cycles Reported

Max Number of Redox Cycles Reported

Reference

Ceres Power

Planar MSC

Stack/ system

1 25

10,000

0.34

. 1500

. 60

[26,27]

Bloom Energy

Planar ESC

System

200 300

No data

No data

No data

No data

[28]

Hexis

Planar ESC

System

1

40,000

0.2

50

60

[29 31]

Kyocera/Osaka Gas

Flattened tubular ASC

System

1

10,000

0.2

No data

No data

[9]

Mitsubishi-Hitachi Power Systems

SIS tubular

System

250

16,000

0.1

No data

No data

[9]

Fuel Cell Energy/ Versa Power

Planar ASC

Stack/ system

50

15,000

, 0.5

No data

No data

[32,33]

LG/Rolls-Royce

Flattened tubular SIS

System

250 1000

18,000

0.78

No data

5

[6,7]

Delphi

Planar ASC

System

5

15,000

0.4

170

No data

[34,35]

TOTO

Microtubular

Stack

1

7000

0.5

No data

No data

[9]

NGK

Flattened tubular SIS

Stack

1

8000

0.5

No data

No data

[9]

Murata

Planar ASC

Stack

1

3000

0.6

No data

No data

[9]

SOLID Power

Planar ASC

Stack/ system

2.5

8000

0.32

120

No data

[36]

Sunfire

Planar ESC

Stack

3-4

20,000

0.8

20

No data

[37]

Fraunhofer IKTS

Planar ESC

Stack/ system

0.3 1.2

20,000

0.7

120

No data

[38]

Life and Reliability of Solid Oxide Fuel Cell-Based Products Chapter | 9

181

short-duration tests and a fundamental understanding of the underlying degradation mechanisms, allowing reliable lifetime extrapolation by the development of a rapid evaluation method. The 90,000 h lifetime requirement implies a target cell voltage degradation rate of 0.11%/kh [25]. The rapid evaluation method project has two major work streams; stack testing by the Central Research Institute of Electric Power Industry (CRIEPI) [18] and predictive model development by a consortium of universities (The University of Tokyo, Kyoto University, Tohoku University and the National Institute of Advanced Industrial Science and Technology (AIST)) [25].

Summary of Stack Testing Results at CRIEPI CRIEPI has undertaken steady-state tests of between 3000 h and 16,000 h on stacks from six suppliers with very limited thermal cycling. Electrode polarization models [40,41] have been developed, which enable the breakdown of cell internal resistance into components such as the Nernst loss (ηne), the anode polarization (ηa), cathode polarization (ηc), and ohmic losses (ηIR), and thus to evaluate the rate of change of each component of the cell resistance over time as the cells degrade [18]. Table 9.5 presents the results of CRIEPI testing as of April 2016. It can be seen from Table 9.5 that based on CRIEPI testing, only the SIS tubular cells from MHPS meet the 90,000 h degradation target currently; although the Kyocera flattened tubular cells are close. Note that results from two technology generations are included in the table, as all of the developers continuously improve their cell/stack technology, releasing a new technology generation to the NEDO program every year. The overall average degradation rates and the total operating time are quoted from the older of the two generations as a greater number of testing hours are available on this generation. The resistance breakdown into anode, cathode, and ohmic contributions are from the newer (generally improved) generation. This means that the overall average voltage degradation rate is not necessarily the sum of the individual components. Also note that some component degradation rates have a positive value; this means that component actually improved on average over the duration of the test. Of the stacks that showed higher degradation, the NTK planar cells were operated at a much higher current density than the others, although the performance factor analysis to ascertain the breakdown of degradation was undertaken at 250 mA cm22 to make it comparable with the other stacks. The NGK flattened-tube SIS cells showed rapid voltage drop for the first 2000 h of the test after which the degradation rate reduced to 0.2%/kh. The rapid voltage drop was mostly as a result of an increase in cathode overpotential, possibly as a result of sulfur dioxide contamination of the cathode air feed, as this behavior was apparently not observed when these cells were tested at a different laboratories.

TABLE 9.5 Summary of Results of Stack Degradation Testing at CRIEPI Under the NEDO Program [9] (Degradation rates and operating time shown in bold) SIS Tubular (MHPS)

Flattened Tubular (Kyocera)

Microtubular (TOTO)

Planar (NTK)

Flattened Tubular SIS (NGK)

Single-Step Cofired Planar (Murata)

Temperature ( C)

900

750

700

700

750

750

Current density (mA cm 2)

150

200

200

520

230

350

Average voltage degradation rate (%/kh)

0.1

0.2

0.5

0.6

0.5

2.6

Cathode contribution (%/kh)

0.1

0.1

0.4

0.2

0.4

2.1

0.3

0.2

0.2

Ohmic contribution (%/kh)

0

0.1

Anode contribution (%/kh)

0

0

Operating time (h)

16,000

10,000

0 0.1

0.1

10,000

10,000

0.1

0

8000

3000

Life and Reliability of Solid Oxide Fuel Cell-Based Products Chapter | 9

183

The very rapid degradation of the Murata cofired planar cells was also attributed largely as a result of cathode degradation, again largely as a result of SO2 poisoning of the cathode. As a result of these tests CRIEPI is evaluating filtering the stack air supplies to reduce the SO2 content [18].

Thermal Cycle Testing at Tokyo Gas While steady-state degradation rates are assessed by CRIEPI, stack thermal cycling tolerance in the NEDO program is assessed by Tokyo Gas [42]. It is acknowledged that evaluating thermal cycle tolerance at the end of the nominal 90,000 h stack life is very difficult, as degraded cells may be more susceptible to thermal cycling than fresh ones. In order to attempt to perform an accelerated thermal cycle degradation test, the main degradation mechanism for each cell type was first determined from the testing at CRIEPI. The degradation mechanism was then deliberately accelerated (for example by placing a chromium source in the air stream upstream of the stack or adding SO2 to the air to poison the cathodes) to simulate 10 years of normal operation. Having done this the stacks were then thermal cycled to ensure that they were not more prone to failure after poisoning than they were initially. Due to the difficulty in accelerating many aging mechanisms, accelerated degradation due to poisoning was the only mechanism evaluated in this study, although in both cases poisoning had been identified as the dominant degradation mechanism in nonaccelerated testing. Degradation Mechanisms Identified and Mitigated in the NEDO Program Reliability and Stability of Cathode Electrolyte Barrier Layers Cobaltite cathodes such as lanthanum strontium cobalt ferrite (LSCF) require a doped-ceria barrier layer between the cathode and zirconia electrolyte to prevent the formation of an insulating SrZrO3 (strontium zirconate) phase between the cathode and electrolyte [13]. Cells made by Kyocera incorporate these layers as this has been a significant degradation mechanism in the past. However, Kyocera has developed an accelerated test (operating at 1000 C) for SrZrO3 formation to demonstrate adequate stability for the design cell life. Chromium Poisoning of Manganite-Based Cathodes MHPS’s type V SIS cells operated with a Cr-vapor source upstream of the stack, and showed very little degradation for 20,000 h [25]. This has been attributed to the use of a doped-ceria interlayer between the cathode and electrolyte to improve the cathode performance. This inhibits the main chromium poisoning mechanism of manganite cathodes (Cr2O3 deposition at the cathode electrolyte triple-phase boundary (TPB) through the

184

Solid Oxide Fuel Cell Lifetime and Reliability

electrochemical reduction of Cr(VI) vapor species). This is due to the dopedceria emitting water vapor [43] from the electrochemically active sites. Where Cr(VI) reacts with manganite cathodes, it tends to dissolve into the perovskite lattice on the B-site rather than react with strontium [43]. Chromium Poisoning of LSCF-Based Cathodes LSCF-based cathodes, although much more active that manganite cathodes, are inherently chromium intolerant due to the tendency of vapor-phase Cr (VI) species to react with strontium on the surface to form SrCrO4. This is a chemical rather than electrochemical process and occurs over the whole surface rather than just the TPB area [20]. For this reason it is critical to minimize the partial pressure of chromium vapor in the air, by appropriate surface coating of chromium-containing alloys upstream of and within the stack. Operation at lower temperatures also allows for mitigating the effects of Cr poisoning. Secondary ion mass spectroscopy (SIMS) analyses of cathodes from 2012FY (cells from the 2012 financial year technology generation) Kyocera cells showed lower Cr levels than 2008FY or 2011FY cells, correlating with a lower degradation rate [25]. Similarly TOTO cells with La1 xSrxGa1 yMgyO3 electrolytes are susceptible to Cr accumulation in both the anode and electrolyte [25,42]. Due to design changes to the system, SIMS analysis of 2013FY TOTO cells by AIST showed essentially no chromium. Sulfur Poisoning of Cathodes The presence of sulfur in the cathodes of tested stacks was first noted in disk-type cells by the Mitsubishi Materials Corp [25]. There are multiple potential sulfur sources, most notably SO2 in the air. Cobaltite cathodes (SSC, LSCF, etc.) are susceptible to sulfur poisoning through the formation of SrSO4 on the surface, so as with chromium, the avoidance of gaseous sulfur species in the stack air feed is important. Anode Poisoning Sasaki et al. [17] at Kyushu University have been systematically analyzing SOFC poisoning mechanisms for over a decade, using techniques such as Field emission scanning electron microscopy and energy dispersive X-ray (FESEM -EDX), and microsampling by focussed ion beam milling followed by high-resolution scanning transmission electron microscopy with energy dispersive X-ray (STEM-EDX) elemental analysis. This work is described in detail in Chapter 3, The Impact of Fuels on Solid Oxide Fuel Cell Anode Lifetime: The Relationship Between Fuel Composition, Fuel Impurities, and Anode Lifetime and Reliability of this book.

Life and Reliability of Solid Oxide Fuel Cell-Based Products Chapter | 9

185

Ceres Power’s Stack Performance Verification Program Overview Ceres Power is a UK-based SOFC developer with a unique metal-supported SOFC technology (branded “Steel Cell”) based on ceria rather than zirconia electrolytes. This allows a lower operation temperature (500 600 C) than virtually all other SOFCs. Key advantages for this technology are low cost and excellent thermal and redox cycle tolerance relative to other SOFCs, whilst maintaining high efficiency, low degradation, and fuel flexibility [44 46]. Ceres Stack Performance Verification Process As part of its commercialization strategy, Ceres has reported a process to verify the cell and stack technology, and validate its operation within a selfcontained fuel cell power system (FCPS), which is a product prototype [26]. The first goal of the verification process was to establish the degradation rate and variation therein for the cell and stack technology released to customers in 2015. The second goal was to understand the response in degradation rate to changes in key operating condition parameters. In addition to these goals, there was a desire to evaluate performance across test platforms and validate the operation of the fuel cell stacks within the Ceres Power prototype CHP system. Twelve nominally 1 kWe stacks were built and tested with a mixture of steady-state (galvanostatic) running with some shut down events. The degradation rate in the standard operating condition (. 900 W gross, 700 W net, and 45% net electrical efficiency) was determined for all units after a given operating time. In a second phase of testing the operating conditions were varied and the change in degradation rate was analyzed to determine the sensitivity of stack degradation to the selected operating parameters. Of the 12 stacks, half were operated in a stack test module (STM), where the stack is placed directly in an insulated box and temperature is controlled through the heating of the anode and cathode gas streams. These tests were operated on a simulated reformate mixture and air from the local environment. The other half were operated within FCPS’s and thus received steam reformed natural gas and air from the flue located outside of the building. The test plan was split into two phases. In the first phase, each of the 12 units were operated under the same standard conditions, with stack air inlet and outlet temperatures fixed, fixed current and fuel flow. Each unit was operated for B1500 h with the steady-state operation interrupted by a small number of planned stops (including E-stops where fuel and air supplies are cut to the stack). In the event of unplanned stops on a particular unit the number of planned stops was reduced to ensure that each unit performed a similar test.

186

Solid Oxide Fuel Cell Lifetime and Reliability

In the second phase the operating conditions were moved for further steady-state operation. Three operating parameters were varied: air outlet temperature, fuel utilization, and current. The total planned test duration was 3000 h. Degradation rates were analyzed as the rate of change of stack voltage through a linear model. In the second phase of the test the key output metric was the change in degradation rate with respect to the nominal operating condition for that unit. For the first test phase 11 of the 12 tested stacks met the specification of ,0.5% galvanostatic voltage degradation/kh. The 12th specimen was excluded from the test program due to faulty metering system. There were no significant differences in degradation rate between stack (STM) and system (FCPS) tests. The results of the second phase analysis showed that the most significant factor affecting degradation was fuel utilization (with higher utilization leading to higher degradation), followed by temperature (higher temperature leading to higher degradation). The effect of current was not statistically significant. In addition to galvanostatic degradation testing, Ceres also reported accelerated thermal cycle and emergency stop (redox cycle) tests on 1 kWe class stacks [27]. Over 1500 thermal cycles, and 60 E-stops have been reported without major degradation. Comparison with the results quoted from other developers in Table 9.4 (where the greatest number of thermal and E-stop cycles reported are 170 and 60, respectively) shows the intrinsic robustness of MSC technology to cyclic operation.

LG Fuel Cell Systems Overview LG fuel cell systems (LGFCSs) are based in Canton, OH, USA. They use SIS flattened tubular SOFC technology developed by Rolls-Royce in the UK, for integration into pressurized MW-scale SOFC-gas turbine hybrid systems giving electrical efficiencies of .60% [47]. Their work is partly funded by the US Department of Energy. Identification and Reduction of Cell Degradation Mechanisms Extensive testing of LGFCS cells at the Pentacell level (single tube with five cells in series) coupled with AC impedance spectroscopy to deconvolute cell resistance contributions has enabled the main causes of cell performance degradation to be identified. The degradation of LGFCS cells is dominated by the cathodes [6,7], with smaller contributions from the anode and primary interconnect (the conductors connecting the cells together along each tube).

Life and Reliability of Solid Oxide Fuel Cell-Based Products Chapter | 9

187

The primary causes of cathode degradation are the decomposition of the lanthanum strontium manganite (LSM) cathode leading to localized accumulation of manganese oxide, densification of the cathode during operation and chromium poisoning from the balance of plant (BoP). Manganese segregation is more severe at lower (800 C) operating temperatures, whereas cathode densification is more severe at higher (900 C) operating temperatures. Chromium poisoning results from chromium evaporation from the upstream BoP, and can be mitigated by appropriate coatings. Cathode densification can be reduced by using a cathode material with a second element doped onto the B-site of the LSM perovskite. To both improve cell performance and eliminate the manganese segregation issue nickelate cathodes are currently under evaluation as an alternative. Anode degradation was shown to be dominated by the migration of nickel from the interface between the anode current collector and the active anode toward the electrolyte, demonstrated by tomographic reconstruction of the anode microstructure. This effect has been partially mitigated by replacing the bilayer anode with a single layer anode. The anodes have limited redox stability (being able to withstand up to five redox cycles) so the system requires a protection gas system to maintain a reducing atmosphere over the anodes during shutdowns. The degradation of the primary interconnects has been mitigated by the introduction of a barrier layer on the anode side between the primary interconnect and the substrate. Incorporation of all these improvements into the latest generation of cells has resulted in performance degradation rates of 0.1 0.15%/kh using small lab-scale cells. On a larger bundle test degradation rates of the order of 0.3%/kh have been demonstrated.

SUMMARY AND CONCLUSIONS It is clear from the results and the examples of work mentioned in this chapter that SOFC durability and reliability issues are starting to be well understood mechanistically, and mitigation strategies to address them are rapidly maturing to the point where the lifetimes required for commercial applications can be achieved. A good example of this is the development of protective coatings on metallic components to prevent corrosion and chromium poisoning of cathodes, which have largely mitigated what was historically a severely life-limiting failure mode for planar SOFC stacks with metallic interconnects. The main challenge for SOFC technology generally is being able to achieve performance and reliability at a cost consistent with widespread commercialization. The older SOFC technology generations (particularly tubular variants) are inevitably the most mature in terms of proven durability, but have the highest intrinsic cost. Conversely newer variants (particularly

188

Solid Oxide Fuel Cell Lifetime and Reliability

planar MSC technology) are less mature but have the greatest potential for low intrinsic cost. The remaining issues with each technology generation can be summarized as follows: G

G

G

G

Tubular SOFC: High intrinsic cost, low intrinsic volumetric density, and poor dynamic behavior. Planar electrolyte-supported SOFC: High intrinsic cost, poor thermal cycle resistance due to sealing problems. Planar anode-supported SOFC: Poor resistance to redox cycling, potential problems with thermal cycle resistance. Planar metal-supported SOFC: Relatively immature technology (still evolving rapidly) so maximum number of operating hours limited by comparison with earlier technology generations; risk of as yet unknown failure modes. Some developers have reported issues with metal substrate corrosion.

While durability can ultimately only be proven by running very longterm (80 90 kh) tests, this is not a viable strategy for developers as it would imply 10-year development cycles. There is clearly therefore a requirement to be able to predict SOFC life based on extrapolation of much shorter test durations, and then use this information to develop mitigation strategies. In this chapter, we have highlighted some of the areas of development that could help accelerate this maturity of understanding of degradation mechanisms that allows for improvements in reliability and life prediction. Of the information available publicly, the NEDO program in Japan is closest to achieving lifetime prediction, but even there the techniques required to do this reliably are still very much under development. The latest generation of metal-supported SOFC cells (MSCs) offer the possibility of lower cost and greatly enhanced robustness. The development of accelerated testing strategies and validation will allow for this “next generation” technology to mature faster and reach its potential in the next few years.

REFERENCES [1] Horiuchi K. Current status of national SOFC projects in Japan. ECS Trans 2013;57 (1):3 10. [2] Braun RJ, Kazempoor P. Chapter 12: Application of SOFCs in combined heat, cooling and power systems. In: Ni M, Zhao TS, editors. Solid oxide fuel cells, from materials to system modelling. London: Royal Society of Chemistry; 2013. RSC Energy and Environment Series No. 7. [3] Rechberger J, Kaupert A, Hagerskans J, Blum L. Demonstration of the first European SOFC APU on a heavy duty truck. Transport Res Proc 2016;14:3676 85. [4] Gerdes K, Williams MC, Gemmen R, White B. A global framework for examination of degradation in SOFC. ECS Trans 2013;57(1):289 97. [5] Finsterbusch M, Lussier A, Schaefer JA, Idzerda YU. Electrochemically driven cation segregation in the mixed conductor La0.6Sr0.4Co0.2Fe0.8O3-δ. Solid State Ionics 2012;212:77 80.

Life and Reliability of Solid Oxide Fuel Cell-Based Products Chapter | 9

189

[6] Liu Z., DeBellis C.L.G. Fuel cell systems program and technology update. 16th annual solid state energy conversion alliance worskhop; 2015. Available from: http://www.netl. doe.gov/File%20Library/Publications/Proceedings/2015sofc/Liu-and-DeBellis.pdf. [7] Lee I.S., Babcock A., L.G. Fuel Cell Systems program and technology update. 17th annual solid solid state energy conversion alliance worskhop; 2016. Available from: http://www.netl.doe.gov/File%20Library/Events/2016/sofc/Lee-Babcock.pdf. [8] Heneka MJ, Ivers-Tiffee E. Accelerated life tests for fuel cells. ECS Trans 2006;1 (8):377 84. [9] Yokokawa H., Current status of NEDO durability project with emphasis on correlation between cathode overpotential and ohmic loss. Proceedings of 12th European fuel cell forum, Lucerne, Switzerland, Chapter 2, A0602; 2016. p. 124 33. [10] Yokokawa H, Tu H, Iwanschitz B, Mai A. Fundamental mechanisms limiting solid oxide fuel cell durability. J Power Sources 2008;182:400 12. [11] Menzler N.H., Batfalsky P., Beez A., Blum L., Groß-Barsnick S., Niewolak L., et al., Post-test analysis of a solid oxide fuel cell stack operated for 35,000 h. Proceedings of 12th European fuel cell forum, Lucerne, Switzerland, Chapter 6, A1101; 2016. p. 290 297. [12] Wang X, Atkinson A. Simulation and prediction of 3-D microstructure evolution and long term performance of Ni-YSZ anode. ECS Trans 2015;68(1):2867 73. [13] Yokokawa H. Report of five-year NEDO project on durability/reliability of SOFC stacks. ECS Trans 2013;57(1):299 308. [14] Nakajo A, Mueller F, Brouwer J, Van herle J, Favrat D. Mechanical reliability and durability of SOFC stacks. Part I. Modelling the effect of operating conditions and design alternatives on the reliability. Int J Hydrogen Energy 2012;37:9249 68. [15] Linder M, Hocker T, Holzer L, Friedrich KA, Iwanschitz B, Mai A, et al. Model-based prediction of the ohmic resistance of metallic interconnects from oxide scale growth based on scanning electron microscopy. J Power Sources 2014;272:595 605. [16] Falk-Windisch H, Svensson JE, Froitzheim J. The effect of temperature on chromium vaporization and oxide scale growth on interconnect steels for solid oxide fuel cells. J Power Sources 2015;287:25 35. [17] Sasaki K, Yoshizumi T, Haga K, Yoshitomi H, Hosoi T, Shiratori Y, et al. Chemical degradation of SOFCs: external impurity poisoning and internal diffusion-related phenomena. ECS Trans 2013;57(1):315 23. [18] Yoshikawa M, Yamamoto T, Asano K, Yasumoto K, Mugikura Y. Performance degradation analysis of different type SOFCs. ECS Trans 2015;68(1):2199 208. [19] Schuler JA, Yokokawa H, Calderone CF, Jeangros Q, Wuillemin Z, Hessler-Wyser A, et al. Combined Cr and S poisoning in solid oxide fuel cell cathodes. J Power Sources 2012;201:112 20. [20] Tucker MC, Kurokawa H, Jacobson CP, De Jonghe LC, Visco SJ. A fundamental study of chromium deposition on solid oxide fuel cell cathode materials. J Power Sources 2006;160(1):130 8. [21] Blum L, Groß SM, Malzbender J, Pabst U, Peksen M, Peters R, et al. Investigations of solid oxide fuel cell sealing behavior under stack relevant conditions at Forschungszentrum Ju¨lich. J Power Sources 2011;196:7175 81. [22] Waldbillig D, Wood A, Ivey DG. Electrochemical and microstructural characterization of the redox tolerance of solid oxide fuel cell anodes. J Power Sources 2005;145:206 15. [23] Young JL, Birss VI. Crack severity in relation to non-homogeneous Ni oxidation in anode-supported solid oxide fuel cells. J Power Sources 2011;196:7126 35.

190

Solid Oxide Fuel Cell Lifetime and Reliability

[24] Fernandes N., SOFC Quality Control and the role of manufacturing defects on stack longevity. 17th annual solid state energy conversion alliance worskhop; 2016. Available from: http://www.netl.doe.gov/File%20Library/Events/2016/sofc/Fernandes.pdf [25] Yokokawa H. Current status of rapid evaluation of durability of six SOFC stacks within NEDO project. ECS Trans 2015;68(1):1827 36. [26] Bone A., Postlethwaite O., Leah R., Mukerjee S., Selby M., Validation results and methodology from a ceres power cell technology platform. Proceedings of 12th European fuel cell forum, Lucerne, Switzerland, Chapter 9, B0902; 2016. p. 9 19. [27] Leah R., Bone A., Lankin M., Rahman M., Hammer E., Selcuk A., et al., Development status of ceres power steel cell technology: further improvements in manufacturability, durability and performance. Proceedings of 12th European fuel cell forum, Lucerne, Switzerland, Chapter 2, A0303; 2016. p. 25 34. [28] ES-5700 Energy Server Data Sheet: BloomEnergy; 2015. Available from: http://www. bloomenergy.com/fuel-cell/es-5700-data-sheet/. [29] Mai A, Iwanschitz B, Schuler JA, Denzler R, Nerlich V, Schuler A. Hexis’ SOFC system Galileo 1000 N lab and field test experiences. ECS Trans 2013;57(1):73 80. [30] Mai A, Schuler JA, Fleischhauer F, Nerlich V, Schuler A. Hexis and the SOFC system Galileo 1000 N: experiences from lab and field testing. ECS Trans 2015;68(1):109 16. [31] Mai A., Fleishhauer F., Schuler J.A., Denzler R. Nerlich V. Schuler A., Advances in HEXIS’s SOFC development. Proceedings of 12th European fuel cell forum, Lucerne, Switzerland, Chapter 2, A0301; 2016. p. 8 16. [32] Borglum BP, Ghezel-Ayagh H. Development of solid oxide fuel cells at versa power systems and FuelCell energy. ECS Trans 2015;68(1):89 94. [33] Borglum A., Ghezel-Ayagh H., Solid oxide fuel cell developemnt at versa power systems and FuelCell energy. Proceedings of 12th European fuel cell forum, Lucerne, Switzerland, Chapter 2, A0302; 2016. p. 17 24. [34] Kerr R., Solid oxide fuel cell power system development DE-FE0001179. 15th annual solid state energy conversion alliance worskhop; 2014. Available from: http://www.netl. doe.gov/File%20Library/Events/2014/2014%20SECA%20workshop/Rick-Kerr.pdf. [35] Kerr R., Solid oxide fuel cell power system development DE-FE0001179. 16th annual solid state energy conversion alliance worskhop; 2015. Available from: http://www.netl. doe.gov/File%20Library/Publications/Proceedings/2015sofc/Kerr.pdf. [36] Bertoldi M., Bucheli O., Ravagnia A.V., High-efficiency cogenerators from SOLIDpower SpA. Proceedings of 12th European fuel cell forum, Lucerne, Switzerland, Chapter 2, A0304; 2016. p. 35 44. [37] Walter C., Strohbach T., Meisel P., Herbrig K., Schimanke D., Posdzeich O., 25 kW stack module development-status at sunfire. Proceedings of 12th European fuel cell forum, Lucerne, Switzerland, Chapter 2, A0305; 2016. p. 45 58. [38] Kusnezoff M., Megel S., Jahn M., Pfeifer T., Baade J., Status of SOFC/SOEC stack and system development and commercialization activities at Fraunhofer IKTS. Proceedings of 12th European fuel cell forum, Lucerne, Switzerland, Chapter 2, A0601; 2016. p. 114 23. [39] Kadowaki M. Current status of national SOFC projects in Japan. ECS Trans 2015;68(1):15 22. [40] Mugikura Y, Yasumoto K, Morita H, Yoshikawa M, Yamamoto T. Performance evaluation technology for long term durability and reliability of SOFCs. ECS Trans 2013;57(1):649 56. [41] Yamamoto T, Yasumoto K, Yoshikawa M, Hiroshi M, Mugikura Y. Performance evaluations for long term durability and reliability of segment-in-series tubular type SOFCs. ECS Trans 2013;57(1):763 70.

Life and Reliability of Solid Oxide Fuel Cell-Based Products Chapter | 9

191

[42] Hatae T, Sato K, Somekawa T, Matsuzaki Y, Amaha S, Yoshikawa M, et al. Durability assessment of SOFC stacks with several types of structures for thermal cycles during their lifetimes on residential use. ECS Trans 2015;68(1):2209 16. [43] Yokokawa H, Horita T, Yamaji K, Kishimoto H, Yamamoto T, Yoshikawa M, et al. Chromium poisoning of LaMnO3-based cathode within generalized approach. Fuel Cells 2013;13(4):526 35. [44] Leah RT, Bone A, Selcuk A, Corcoran D, Lankin M, Dehaney-Steven Z, et al. Development of highly robust, volume-manufacturable metal-supported SOFCs for operation below 600 C. ECS Trans 2011;35(1):351 67. [45] Leah R., Bone A., Selcuk A., Lankin M., Pierce R., Rees L., et al., Robust, low-cost, efficient steel cell stack development at ceres power. Proceedings of 11th European fuel cell forum, Lucerne, Switzerland, Chapter 4, A0605; 2014. P. 39 46. [46] Leah RT, Bone A, Lankin M, Selcuk A, Rahman M, Clare A, et al. Ceres power steel cell technology: rapid progress towards a truly commercially viable SOFC. ECS Trans 2015;68(1):95 107. [47] LG Fuel Cell Systems: Overview Presentation. Hydrogen and Fuel Cell Technical Advisory Committee. LG Fuel Cell Systems Inc.; 2013. Available from: https://www. hydrogen.energy.gov/pdfs/htac_apr13_6_fleiner.pdf.

Chapter 10

New Materials for Improved Durability and Robustness in Solid Oxide Fuel Cell Mark Cassidy1, Dragos Neagu1, Cristian Savaniu1 and Paul Boldrin2 1

University of St Andrews, St Andrews, United Kingdom, 2Imperial College London, London, United Kingdom

Chapter Outline Introduction Solid Oxide Fuel Cell Electrolytes Anodes Cathodes

193 194 197 202

Stack Materials Accelerated Testing Summary References

206 209 210 211

INTRODUCTION It is often argued that durability is as least as important as absolute performance in solid oxide fuel cells (SOFCs). However, a significant amount of SOFC research into new materials is focused on improving performance of the cell in terms of maximizing current and power density. Although this absolute performance does influence durability and lifetime, in so far as the target operational current density may require a smaller voltage drop allowing the cell to be run at a higher voltage improving efficiency and possibly degradation rates [1]. Secondly a higher performing but degrading system will take longer to reach a nominal end of life performance point than a similarly degrading system with lower initial performance. However, this must be balanced with the fact that a more active (and potentially reactive) material may degrade more rapidly creating a breakeven point where the performance drops below that of a relatively low-performance material which does not degrade at the same rate, as illustrated in Fig. 10.1. Many standard SOFC materials, Ni/yttria-stabilized zirconia (YSZ) anode cermets, lanthanum strontium manganite (LSM) cathodes, YSZ electrolytes, have been shown to exhibit good stability with adequate performance Solid Oxide Fuel Cell Lifetime and Reliability. DOI: http://dx.doi.org/10.1016/B978-0-08-101102-7.00010-6 Copyright © 2017 Elsevier Ltd. All rights reserved.

193

194

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 10.1 Schematic representation of various cell degradation behaviors with higher and lower initial performances.

demonstrated over many thousands of hours [2,3]. Even though in some of these cases the tests have been run under favorable conditions with optimized, pure fuels, low power densities or hand crafted short stacks, it shows that under the right conditions standard materials have shown good levels of fundamental long-term stability. Therefore many durability challenges revolve around interactions of adjacent materials or the robustness of the interfaces between them. Such challenges are tolerance to carbonaceous fuels and impurities, deactivation of the electrode function due to particle coarsening and redox tolerance in the anode. Cathode issues often involve reactions with adjacent materials forming resistive phases or segregation of cations within the structure leading to changes in activity or conductivity. The nature of many of these degradation effects is considered to be thermally driven therefore reduction of operational temperature is a strong theme across much new materials development. However, in some cases, such as effects of impurities it is not so straightforward and some aspects can get worse in lower temperature regimes. Where these effects occur at interfaces, they are exacerbated as cells are formed into larger stacks where the number of interfaces becomes greatly magnified and the scale up aspects demand that assembly processes are increasingly automated. In this chapter we look at where new materials are impacting the search for improved durability and study some promising directions for the main cell components and also where these interface to the larger stack. We also consider some materials whose initial performance is encouraging but have not yet been studied in terms of long-term behavior and discuss which direction could be taken with these to investigate promise for longer term operation.

SOLID OXIDE FUEL CELL ELECTROLYTES The electrolyte is the central part of the SOFC, separating the two electrodes and has to fulfill several conditions for suitable operation: must have sufficient ionic conductivity while exhibiting low or negligible electronic conductivity, be chemically and mechanically stable at elevated operating

New Materials for Improved Durability Chapter | 10

195

temperatures and under both oxidizing and reducing conditions and be gas impermeable to prevent gaseous fuel-oxidizer cross over. The most common oxygen-ion conductor material that is widely employed as an SOFC electrolyte material is doped zirconia, and this material is discussed in detail in Chapter 2, Solid Oxide Fuel Cell Electrolytes— Factors Influencing Lifetime. As has already been discussed it is highly desirable to decrease the SOFC operation temperature in the 750 800 C range in order to try to reduce the overall cost and degradation, therefore alternative materials with higher ionic conductivity than 8YSZ and/or a reduction in electrolyte thickness to B10 μm are required to maintain an acceptable ohmic loss across this component. With a thin electrolyte, usually the anode or a metallic structure takes the role of the cell support; however, this is a geometric rather than materials approach and is not discussed here. With respect to alternative higher conductivity compositions utilizing the well-studied zirconia base, doping with Sc31 (scandia-stabilized zirconia, SSZ) offers better ionic size match to Zr41 than Y31, leading to smaller defect association energy that increases oxygen mobility and consequently the conductivity, with an optimum doping level between 7 and 11 mol%. However, the phase assemblage in this system is more complicated with slow cation diffusion leading to the formation of secondary phases, which can degrade performance and stability. To minimize this and increase general phase stability over time, an additional dopant can be introduced: these are often Ce, Y, Yb, or Sm, with 1 mol% Ce being a common additive and Zr0.89Ce0.01Sc0.10O2 x (10Sc1CeSZ) showing the best overall conductivity. Although the overall conductivity is lower than undoped SSZ, it is more stable and relates to the argument over initial performance vs long-term stability illustrated in Fig. 10.1 [4 6]. The ionic conductivity of zirconias tends to decrease in time, at high temperatures, most likely due to phase evolution into lower conductive phases [7]. Zirconia-based electrolytes also tend to chemically react with typical Mn- or Co-based cathode materials such as LSMs or lanthanum strontium cobaltites (LSCs), at temperatures as low as 800 C, leading to formation of resistive La2Zr2O7/SrZrO3 phases. In addition the difference in thermal expansion coefficient between these materials causes deterioration of long-term performance and thermal cycling stability of the cell. These interfacial problems are usually overcome introducing a thin, dense, or porous ceria interlayer between YSZ and cathode [8]. Alternative fluorite-type conductors that are used as electrolytes are the doped CeO2 and Bi2O3. Typical dopants include Y31, Gd31, or Sm31 for ceria, with a maximum ionic conductivity occurring at 10 20 mol% dopant, while higher levels of Y31 or Dy31 are required to stabilize the highly conductive δ-phase of Bi2O3 at room temperature [9,10]. Another derivative is the BIMEVOX materials (substituted Bi4V2O11) [11] having Aurivillius structure. These solid solutions have conductivity values which are significantly higher than that of 8YSZ, particularly at lower temperatures (500 700 C). Despite this, an electronic contribution to the conductivity is observed at low oxygen

196

Solid Oxide Fuel Cell Lifetime and Reliability

partial pressures, characteristic to the anode compartment, above 500 C, accompanied by mechanical instability for ceria materials or even decomposition for Bi2O3 ones. Numerous attempts have been made to suppress the electronic conduction in these materials, by optimizing the doping level in Bi2O3 [12], using a bi-layered electrolyte structure [13], employing a composite electrolyte or a proton conductor material as anode component [14]. To date, one of the highest SOFC electrochemical performances was achieved at 650 C (B2 W cm22) using a thin bi-layer of gadolinium-doped ceria (GDC)-erbia stabilised bismuth oxide (ESB) electrolyte concept [15]. Pyrochlore structured oxides of A2B2O7 formula have also been explored as electrolytes due to their high intrinsic oxygen-ion conductivity and phase stability. The structure is related to the fluorite structure, exhibiting additional ordering of cation and anion sublattices, and typical examples include Ln2Zr2O7 and Ln2Ti2O7 (Ln 5 Y or Gd) [16]. The oxides with perovskite structure have attracted a lot of interest as SOFC electrolytes in recent years, especially the doped-lanthanum gallates. La1 xSrxGa1 yMgyO3 (LSGM) that presents superior ionic conductivity to 8YSZ at the same temperature and in a wide range of oxygen partial pressures [17]. Co-doped LSGM (LSGMCo) has even higher ionic conductivity while maintaining electronic insulation so is also used. As one of the more established candidate electrolytes, issues of manufacturing, stability, and lifetime have begun to be addressed in lanthanum gallate-based cells. One of the key issues is chemical compatibility during fabrication. LSGM shows strong reactivity with Ni in the anode, with the Ni doping into the LSGM and inducing electronic conductivity [18]. Instead, porous LSGM scaffolds can be infiltrated with Ni as the lower temperatures reduce the rate of Ni migration [19]. A further improvement is to use a barrier layer of Gdor Sm-doped ceria; however, this can result in the La migrating into the barrier layer, causing the formation of insulating phases [20]. Thus the use of La-doped ceria is favored, although this increases overall resistance as its conductivity is lower than Sm- or Gd-doped ceria [21]. On the cathode side, LSGM can react with cathode materials such as LSM during fabrication, or lose La to non-La-containing cathodes, again forming insulating phases, and again this can be combatted by use of infiltration techniques or barrier layers [22]. Despite these noted problems Mitsubishi and KEPCO have performed several 3000 5000 h tests on 1 10 kW stacks using electrolyte-supported cells of anode electrolyte cathode Ni/Sm-doped ceria LSGMCo Sm0.5Sr0.5CoO3 and found reasonable degradation rates of as low as 0.5%/kh [23], although development was stopped in 2012 due to cracking in the electrolyte, possibly caused by the variable oxidation state of the Co in LSGMCo [24]. Several perovskite oxides present protonic conduction in hydrogen containing atmosphere and the most known compositions are alkaline earth

New Materials for Improved Durability Chapter | 10

197

zirconates or cerates, such as BaZrO3 or BaCeO3 (co)doped with trivalent cations such as Y31, Gd31, or Yb31. Other examples include BaThO3, BaPrO3 [25] as well as rare-earth doped ortho-niobates and tantalates [26]. The main advantage of using a proton conducting electrolyte resides in the facile water management in the SOFC as steam forms on the air electrode side as opposed to the fuel one, minimizing the fuel dilution. Their conductivity is higher than YSZ or CeO2 below 750 C, but their chemical stability is an issue, as CO2 and moisture affect their mechanical properties. A solution to combine high conductivity and stability is offered by the Ba Sr and Zr Ce partial substitution as well as favorable co-doping on the Zr/Ce-site [27,28]. The high temperatures required for densification—also leading to BaO loss from the structure—are real obstacles in the development of these materials. Alternative materials proposed as electrolyte materials include lanthanum molybdates, La2Mo2O9 [29], oxides with Brownmillerite structure (A2B2O5), such as Ba2Ln2O5 and Ba2In2O5, also displaying proton conduction at low temperatures [30] and rare-earth apatite materials such as lanthanum silicates (i.e., La10Si6O27) or germanium counterparts [31]. The ionic conduction of apatites involves interstitial oxide ions as opposed to the oxygen vacancies mechanism that is present for fluorite and perovskite compounds. Stability and compatibility issues often limit their practical applications. To summarize, there are well-established solid electrolyte compositions with high oxide ion conductivity and their choice is driven by the SOFC operation temperature, electrode materials, and cost. The further development of electrolyte materials should encompass the search for new materials with enhanced ionic conductivity and the optimization of conventional and new compositions by suitable doping or the use of new processing routes. An understanding of the local structure, role of various defects, conduction mechanism, and the science behind the stability issues are key factors here. A significant amount of information could come from in operando studies on the cells and stacks and the mathematical modeling could offer excellent complementarity. The development of thin film fabrication technologies opened up various opportunities in reducing the SOFC operation temperature and beneficial microstructural control of the electrolyte layers. In this direction the exploration of recently reported thin nanostructured multilayers or heterostructures [32,33] is very promising as a very efficient way to achieve very high conductivity values by tailoring the ionic transport along the interfaces, where the lattice strain plays a vital role.

ANODES The anode underpins the performance, lifetime, and reliability of the SOFC due to the complex structural and functional roles it must fulfill. Ideally the anode consists of a mechanically and chemically robust porous structure, which supports three essential functionalities: ionic conduction, electronic conduction, and high-electrocatalytic activity toward fuel oxidation [34].

198

Solid Oxide Fuel Cell Lifetime and Reliability

FIGURE 10.2 Concepts and microstructures for SOFC anodes. Schematics of (A) cermet anode and electrochemical reactions. (B) MIEC decorated with catalyst particles and electrochemical reactions. (C) Particle backbone interface formed by exsolution and infiltration.

Electrode reactions occur at discreet locations, where these functionalities converge in the presence of reactants and as such constitute the active sites within the electrode structure (Fig. 10.2A). Typically the region containing the active sites only extends a few microns from the electrolyte anode interface and is often considered the “true electrode” with the remainder of the electrode having mechanical, electron, and gas transport functions [35]. The active electrode region typically consists of intricate structures on the nanoscale and largely governs performance of the cell and is where degradation and ageing are primarily manifested [27]. The components, structure, and evolution of this interface are central to many of the recent advances in the development and understanding of SOC device operation and it is advances in this area that will continue to drive this technology forward [27]. For a long time, cermets consisting of Ni metal and YSZ have been regarded as the state-of-the-art electrodes (Fig. 10.2A) [36]. Cermets offer high performance in various scenarios, but are susceptible to degradation due to various reasons including poor redox stability, coking, and sulfur poisoning [29]. To this end there has been considerable effort in improving bulk cermet materials specifically to make them more robust and tolerant to various forms of challenging anode conditions, such as redox and carbon deposition [37,38]. Application of impregnation techniques is often discussed with respect to improving ceramic anodes and have also been utilized in attempts to improve tolerance of Ni-based cermets to carbonaceous fuels [35]. The addition of ceria to Ni-based cermets reduced polarization resistance and increased tolerance to both carbon and H2S [30,39]. Also impregnation of proton conductors such as BaO or barium zirconium cerates have shown potential to improve carbon tolerance due to the proton conducting phase catalyzing the reverse water gas shift reaction and protecting the nickel [40,41]. While a number of these approaches have shown promise, there seems to be a consensus that alternative electrode concepts, structures, and materials are required in order to advance the technology. One approach has been the investigation of all ceramic-based anodes using various oxide materials. The native surfaces of many of the ceramic anodes do not have as good a catalytic activity for the reduction of hydrogen as Ni. While this can be advantageous for resistance to coking as the

New Materials for Improved Durability Chapter | 10

199

propensity to crack hydrocarbon fuels is also reduced, the overall performance of the anode can suffer. Consequently, electrode designs have gradually evolved toward microstructures consisting of a porous metal oxide backbone decorated with metallic particles [27,42] (Fig. 10.2B) and sometimes auxiliary active phases, such that final structure embodies the key functionalities highlighted above. The metal oxide backbone is typically a perovskite due to the versatility and diversity of the class [43] and may have a pure support and current collecting function, or may be a mixed ionic electronic conductor (MIEC) [27]. Examples of perovskite oxides include titanates, chromates, manganites, etc. [28,44 49]. The metallic nanoparticles typically include Ni, Fe, Ru, Pd, while the auxiliary active phases are typically MIEC oxides such as oxide and related compounds. These complex structures are typically assembled through procedures such as physical deposition [50] or chemical infiltration [35]. One such perovskite is based on the LaxSr1 xCryMn1 yO3 (LSCM) system. Derived from the LSC interconnect material that shows good stability in both oxidizing and reducing atmospheres, the addition of the Mn on the B-site enhances the anode catalytic activity through the Mn41 to Mn31 transition on reduction. The conductivity of this material decreases in a reducing environment, typically around 1 S cm21 under typical anode pO2; however, this can be compensated for by intelligent anode design where thinner active electrode layers are backed up by higher conductivity current collector materials minimizing any ohmic contributions. This material has been integrated successfully into commercially relevant designs where it has shown good retention of performance on redox cycling and good performance in highsteam environments, both of which suggest that this material may be best applied as a robust anode for applications in high fuel utilization environments where nickel-based cermets would become unstable due to higher oxygen partial pressures and electrochemical reoxidation [51]. Tests have shown stability over a few hundred hours but longer tests are required. Further development of the LSCM material has been via impregnation of this perovskite into porous YSZ skeletons via nitrate precursors, with the LSCM showing good activity and coking resistance when running on dry methane, subsequent infiltration of small amounts of CeO2 and Pd (5 and 0.5 wt.%, respectively) further enhanced catalytic activity [52]. Impregnation of metal nanocatalysts such as Pd, Cu, and Ru into LSCM-based systems has also been shown to improve performance in carbonaceous fuels while minimizing carbon formation [53,54]. Another family of materials that has received considerable interest is lanthanum strontium titanate (LST)-based perovskites. Here increased levels of oxygen vacancies and electronic conductivity are facilitated by the Ti41/Ti31 transition [55], anodic properties can be further enhanced by additional doping of cations onto the B-site [56 58]. As well as the specific perovskite mentioned here many others have been investigated as potential anodes [59 61]; however, none of these have yet found as widespread study or application as

200

Solid Oxide Fuel Cell Lifetime and Reliability

the LSCM- or LST-based materials. In the case of LST the perovskite is a porous skeleton, which forms an electrically conducting backbone into which catalyst precursors are infiltrated (often as solutions of metal salts), which form the nano-scale catalyst particles when decomposed then reduced. Encouraging results have been obtained using such architectures, for example, a full system test of nominally 1 kW was run under relevant conditions (reformed natural gas, 850 C), which to date is the largest scale test of this kind using alternative fuel-electrode materials and impregnated electrodes [62]. Performance of this stack is shown in Fig. 10.3A, and degradation in the power is obvious. This degradation is not so apparent when smaller five-cell stacks are tested over longer periods, shown in Fig. 10.3B, in the latter example the infiltrated structures show good durability over longer periods up to 1000 h and provide an encouraging indication of their longer term durability. Of more concern is the difference in degradation behavior when these cells are placed into a larger system. This difference in behavior has been attributed to localized hotter areas forming in certain locations within the stack due to the more difficult thermal management in the larger system, in turn resulting in sintering of the impregnated metal phase and associated reduction in performance. This suggests a thermal sensitivity in these materials which may require study and stabilization. It also points to the effects regarding issues of scale up, in this case more difficult thermal management in the larger systems. The most likely form of degradation in these structures may be particle coalescence however coking in a hydrocarbon environment, especially when dispersed Ni particles are employed should also be considered. Both these degradation mechanisms seem to be intimately linked to the particle backbone interaction. Certain auxiliary phases, such as Ce1 xGdxO2 x/2 (CGO), have been noted to improve this interaction, leading to better anchored and thus more stable particles. Although the possible increased interaction can lead to reduced catalytic activity, this has not been seen; however, the mechanism of the observed synergistic benefits of having the

FIGURE 10.3 (A) Hexis Galileo System test with Ni/CGO impregnated LSCT anode tested on reformed natural gas at 850 C. (B) Five-cell stack test based on Galileo repeat unit with Ni/CGO impregnated LSCT anode running on reformed natural gas at 850 C [55].

New Materials for Improved Durability Chapter | 10

201

two phases present has not yet been fully explained. Other metals such as Ru form more stable interfaces with the perovskite oxide backbones when compared with Ni [63], possibly due to better crystallographic coherence between the two phases. On the other hand, a recent study seems to point out that the method by which these structures are prepared, that is, the infiltration technique, poses some fundamental limitations on the degree of particle anchorage one can achieve [64]. The particle backbone interface produced by infiltration is shallow and less likely to lead to crystallographic epitaxy when compared with those produced by an alternative method referred to as redox exsolution (Fig. 10.2C). In exsolution the catalytically active metal is embedded in the crystal lattice of the backbone in oxidizing conditions, forming a solid solution, and released (exsolved) on the surface as metal particles upon exposure to a reducing atmosphere (e.g., H2) [65,66,60]. Because the particles emerge from the backbone metal oxide, they share close crystallographic links across the interface and are about one-third immersed in the surface of the host lattice [43] as shown in the micrographs in Fig. 10.4. Due to this confinement, exsolved particles are seemingly better anchored when compared with the unconstrained particles produced by infiltration and thus are expected to exhibit considerably different physical and chemical properties, some of which are discussed below. Not surprisingly, exsolved particles possess enhanced thermal stability and display low tendency to coalesce, even when spaced by less than one particle diameter [36]. Such enhanced stability is also observed in SOFC [67] or other catalytic applications [36,38]. Another notable consequence of particle anchorage is reflected in their remarkable coking resistance while maintaining activity for desirable reactions [36,68]. Ni particles prepared by infiltration are well known to catalyze the formation of carbon fibers at the interface with the support, resulting in particle uplifting and leading to irreversible catalyst damage in critical processes such as syngas production by methane steam reforming or SOFCs [69]. Ni particles prepared by exsolution

FIGURE 10.4 Scanning electron microscope (SEM) micrographs of exsolved Ni particles after reductions of (La0.52Sr0.28)(Ni0.06Ti0.94)O3 for 12 h in 5%H2/Ar at 920 C (A) before and (B) after etching in HNO3 showing the anchoring sockets into the substrates. Inserts show histograms of particle and socket sizes base on scale bar (200 nm). (C) AFM of the sockets equivalent to those in (B). Modified from li X, Zhou H, Zhou X, Xu N, Xie Z, Chen N. Electrical conductivity and structural stability of La-doped SrTiO3 with A-site deficiency as anode materials for solid oxide fuel cells. Int J Hydrogen Energy 2010;35(15):7913 18.

202

Solid Oxide Fuel Cell Lifetime and Reliability

were found to be exceptionally resistant to uplifting and subsequent carbon fiber growth (i.e., coking) in hydrocarbon catalysis. Another noteworthy functional aspect of redox exsolution is that in certain scenarios, it has been shown to be reversible [37]. Particles may be redissolved into the host lattice on oxidation and subsequently reexsolved upon reduction. Even though this does not appear to be universally applicable or fully reversible [70], it may serve as proof of concept that these microstructures and interfaces may be rejuvenated by carrying out controlled redox cycles, thus improving the longevity of certain systems. Aside from the exciting functional aspects that exsolved systems exhibit, they bring additional benefits worth mentioning. First, exsolution can be triggered under operationally relevant conditions, that is under H2, or in electrolysis mode [48]. Because of this, exsolution can be used to deploy active particles exactly at the active region within the electrode, saving materials and precursors. Importantly from a viewpoint of lifetime and stability the exsolved structures offer important advantages over the less well pinned and therefore more mobile infiltrated structures and could be a promising method for blending the improvements in performance observed with nano-catalyst structures while maintaining a high level of stability over the longer term. This could be significant as maintaining activity and stability over these length scales is often difficult to balance.

CATHODES As mentioned in the introduction, there has long been an association between reducing the operating temperature of SOFCs and reducing degradation as many of the associated processes are thermally driven. Therefore a significant driver in SOFC cathode development has been around new materials with higher electrode performance. Traditional cathode materials such as LSM have been shown to be very stable under the correct conditions; however, they exhibit an exponential growth in polarization resistance as the temperature drops making them unsuitable for operation temperatures of below around 750 C [71]. To some extent this can be compensated for by the use of composite structures of ionic and electronic conducting phases, such as YSZ/LSM composites [72]. The intimate mixing increases the triplephase boundary length by extending the ionic conducting phase into the body of the cathode increasing the area for reaction and so reducing polarization resistance, however, they can still struggle to deliver adequate performance below 700 750 C. This led to searches for materials with lower activation energies, which would perform better at lower temperatures, in particular materials exhibiting MIEC were of great interest and have formed the main basis of potential new materials for SOFC. The main advantage of MIECs is the extension of the triple-phase boundary from a line between the ionic and electronic conducting phases in single

New Materials for Improved Durability Chapter | 10

203

conductors to a much wider area taking in the surface of the mixed conductor where exposed to the gas phase [73]. Allied to this is the fact that many MIECs also have lower activation energies related to better oxygen exchange rates and surface diffusion coefficients than materials such as LSM. These properties have been of considerable interest in the search for new and better cathodes and have been largely studied through the application of O18 tracer diffusion measurements with secondary ion mass spectroscopy [74]. Popular materials here were the (La1 xSrx)CoO3 and (La1 xSrx)(Co1 yFey)O3 perovskites with the latter still widely used as high-performance cathode materials [75,76]. The disadvantages of these families of materials were that although they showed high catalytic activity they also exhibited high reactivity and rapidly formed insulating pyrochlore phases when in contact with zirconia-based electrolytes, even at operational temperatures. The mobility of a number of both the A-site and B-site cations leads to local nonstoichiometries creating increased resistivities in the case of Sr migration and segregation at the interconnect interfaces, or localized La excess at the electrolyte interface leading to lanthanum zirconate formation at the triple-phase boundary. The use of A-site deficient materials has become a well-used method of avoiding any La excess so minimizing zirconate formation. Another common approach to mitigate against any reaction between cathode and a zirconia-based electrolyte has been the application of a ceria-based barrier layer, such as that shown in Fig. 10.5, as the ceriabased materials show far lower levels of reactivity with the perovskites [77]. This has been discussed in more detail in the chapter on cathodes and while it has been shown to be effective, the potential manufacturing cost implications must be weighed against the improvements in cell durability.

FIGURE 10.5 Polished cross section of thin film interlayer (dense) on a more porous electrolyte both around 1 μm deposited by PVD. SEM image taken posttest with electrolyte porosity induced during operation [70].

204

Solid Oxide Fuel Cell Lifetime and Reliability

In a similar vein to the infiltrated anode materials discussed above infiltration has also been applied to cathode processing with promising results. In this case a porous backbone of electrolyte materials often YSZ or GDC is infiltrated with an appropriate mixture of nitrate precursors that when thermally decomposed will form the required perovskite cathode phase [76]. The advantage of this technique for durability and robustness in cells is that it avoids high-temperature processing of the reactive perovskites with phase formation taking place well below 1000 C, often as low as 600 C. As well as avoiding deleterious reactions, it also maintains a highly dispersed high surface area catalyst for oxygen reduction. As these structures are often intended for intermediate temperature (IT)-SOFC operation (600 C or below) and the catalyst particles are in an oxide form (rather than the metallic often seen in anode catalysts) particle coalescence may not be as rapid and stable performances have been observed over 1500 h of operation [78,79]. However, particle coarsening does take place and not surprisingly this is closely related to both the activity of the infiltrated phase and the target temperature of operation, with lower temperatures resulting in lower degradation [80]. This again reinforces the fact that these techniques may be best employed to enable IT-SOFC. Where particle sintering is observed as an issue, stabilization has been proposed by creating a core shell type structure, where the shell is a porous zirconia that prevents sintering of the more active catalyst but still allows oxygen diffusion through to the catalyst for oxygen reduction. Stable performance over 4000 h of operation has been reported for these structures [81]. Another approach was the use of praseodymium-based perovskites, where the La cation on the A-site was replaced or substituted with Pr [82]. In this instance any zirconate formed was the Pr-based system, which was still catalytically active for the oxygen reduction reaction and so should reduce any harmful effects of interlayer formation. Although some promising results were observed this approach did not achieve widespread adoption; however, research is still ongoing and recent work has shown a renewed interest in Pr-based materials within layered perovskite structures such as those based around Pr4Ni3O101δ and PrBa0.5Sr0.5Co2O51δ [83,84]. Recent studies of these crystallographically layered materials have shown promise for high activity cathodes with good catalytic and oxygen transport properties to allow lower temperature operation [85]. Often based on modified perovskite structures such as K2NF4, Ruddlesden-Popper and Brownmillerite phases [86 88], the layered structure can provide many opportunities for doping the material and provide pathways for rapid oxygen transport or electronic conduction via distortions and charge compensations in doped materials (see Fig. 10.6). The inherent structural and compositional flexibility in these materials has resulted in studies of many variations on the theme utilizing cobaltites, ferrites, cuprates amongst other and lanthanum, barium, samarium, and cerium on the A-site. Many of the studies focus on understanding the fundamental structure and catalytic activity of the materials or materials series in question as all electrode materials must meet the basic requirements of good

New Materials for Improved Durability Chapter | 10

205

FIGURE 10.6 Schematic of Ruddlesden-Popper structure such as in lanthanum nickelate cathodes (Lan11NinO3n11, where n 5 1, 2, and 3) [81].

catalytic activity, good match of physical properties such as coefficient of thermal expansion (CTE), reasonable process windows and minimal deleterious chemical interaction with adjacent materials. As yet few if any studies have begun to look at the longer term stability and compatibility of these materials over longer term SOFC operation with studies only extending toward a few hundred hours. Many of these are unstable with YSZ and somewhat better with CGO and in some cases LSGM. In these latter cases, there may be stable operational windows for IT-SOFC operation or with the use of a CGO interlayer with YSZ as described above. However, there needs to be further consideration of new doping strategies within these materials and this should be focused around optimizing the long-term behavior, stability and compatibility rather than the raw performance. While encouraging in performance more work is required in longer term testing to grow confidence as to the stability over longer term operation [66,85]. One significant issue with many common cathode materials is susceptibility to chromium poisoning. This is exacerbated by the use of chromium rich or chromia forming interconnect materials leading to volatile chromia species being released into the cathode air supply [89]. This leads to the deposition of solid chromium-based compounds at the triple-phase boundaries [90]. These block the electrochemical reactions leading to an increase in polarization resistance and so reduction in cell performance [91]. The precise mechanisms of both the deposition and the degradation have been the subject of much research and discussion and in depth reviews of the chemistry are available [92]. There have been several proposed and which one is dominant will be influenced by cathode material composition, the cathode microstructure and the level of polarization under which the cell is operating.

206

Solid Oxide Fuel Cell Lifetime and Reliability

However, one potential underlying mechanism proposed is that segregation of cation species on the surface of the perovskite leads to a chemical deposition of the Cr species via a nucleation process [82]. The main active cations associated with this are manganese and strontium, with obvious implications for popular SOFC cathode compositions. In some cases the application of composite structures can reduce the sensitivity to Cr but some level of degradation was still apparent. Therefore continued work has investigated materials with no Mn or Sr in the composition. Again some promising examples are based around the layered structures such as K2NiF4 structures, for example, Nd2NiO4 [93] or LaFe1 xNixO3 δ [94]. An alternative approach has been substitution of the susceptible cations for less susceptible cations, such as introduction of Ba in such materials as (La0.24Sr0.16Ba0.6) (Co0.5Fe0.44Nb0.06)O3 δ [95]. However, the traditional materials still dominate and engineering-based approaches such as suppression of evaporation of Cr from the interconnect has been a popular approach to this issue.

STACK MATERIALS As well as the materials within the cells themselves a significant degradation and durability challenge occurs when the cells are stacked together. This is unavoidable as cells must be formed into stacks to provide a realistic voltage output. However, this creates a number of further interfaces which can lead to issues around contact and sealing which in turn can exhibit themselves as difficult thermal management, individual material degradation and undesired materials interaction, a simple illustration of this is shown in Fig. 10.7 and shows how quickly the design complexity can increase as the stack scales up. One important consideration is electrical contact between the cell and the interconnect. In planar systems flatness can be a significant issue, especially camber or bowing caused by differential sintering of the various cell layers. This can lead to incomplete contact with the adjacent interconnect surface, which can be further exacerbated if the interconnect also has a bow or camber. If this is out of phase with that of the cell then one may end up with only a few percent of the available interfaces actually in contact with one another as depicted in Fig. 10.8. This can lead to current concentration around these points with local current densities far in excess of that which has been nominally applied to the stack. This then leads to overheating and failure of the weakest contacts which pushes further burden on those remaining which can then fail and so on, and could lead to the steadily increasing degradation rates often seen in stack failure. Attaining the high degrees of flatness required to maximize the contact between cell and interconnect is a tricky proposition in a multilayer device and the level of quality assurance to achieve this would have a detrimental effect both on processing costs and part rejection rates. Therefore some method of accommodating these surface irregularities is a preferable solution.

New Materials for Improved Durability Chapter | 10

207

FIGURE 10.7 Schematic representation of SOFC stack in cross section. Illustrating the increasing numbers and complexity of contacts and interfaces that must be considered over single cell tests [96].

FIGURE 10.8 Exaggerated schematic showing how nonuniformity in cell and I-C Flatness can lead to reduced contact, and in worst case point contact points between the cell and interconnect.

Metallic meshes and foams, expanded grids and felts are often used as the inherent ductility in the metal can provide the required compliance [97 99]. Nickel-based materials are often used on the anode side; however, the cathode side proves a little more problematic due to the oxidizing atmosphere present here and the relatively high surface area of the metals leading to rapid oxide formation. In these cases some have tried coating the meshes with a layer of ceramic-based material, often a conductive perovskite based

208

Solid Oxide Fuel Cell Lifetime and Reliability

on the cathode itself. However, this reintroduces the existing issue around the brittle nature of the ceramic that can lead to cracking and spalling of the later due to expansion mismatch on thermal cycling or mechanical stresses leading to brittle type failure of the layer [100]. Another common approach is to use a wet paste of the cathode material, where the viscous nature of the paste will take up any irregularities during the assembly of the stack and provide a continuous contact across the interface of the two components [101]. However, once this has been to operational temperatures the organic portion will have burnt out leaving a brittle natured ceramic with all of the associated problems of cracking or delamination during future operation. Similarly during assembly the amount of contact paste must be carefully metered as any excess can block gas flow channels leading to mass transport losses or even flow down manifold channels short circuiting adjacent stack layers. Another approach which has been suggested is to contain a percolating network of conducting particles within a glass matrix [102]. The object here is to have the glass transition temperature (Tg) of the matrix below that of the stack operational temperature thereby retaining compliance and viscoelestic behavior. Even though this system can crack on thermal cycling, once above Tg these cracks could self heal. One challenge of such a system is minimizing any tendency for crystal phases to nucleate in the glass matrix, which will reduce the viscoelastic nature of the matrix and adversely affect its compliant nature. One compliant system in commercial use is the SOFConnex system developed by HTCeramix and currently in use in the “SOLIDpower” stacks [103]. However, the exact nature of the mechanisms employed here are unclear due to the proprietary nature of this development. The sensitivity of the cathode to chromium poisoning has already been pointed out and while there is research into new cathode materials with improved chromium tolerance, an alternative approach is to stop the chromium at source by preventing the evaporation from the interconnect. One method is to apply a protective coating directly onto the surface of the interconnect to block the chromium migration paths [104,105]. This coating is often a conductive perovskite such as LSM or other similar cathode material. As well as the use of perovskites, systems based on (MnCo)3O4 spinels have shown promise and small levels of iron substitution on the cobalt site have also been used to fine tune properties such as CTE. These and the methods of deposition are discussed in more detail in Chapter 7, Lifetime Issues for Solid Oxide Fuel Cell Interconnects. A more integrated and one could argue more elegant approach is to modify the composition of the interconnect such that the protective film it grows minimizes the chromium loss. This film must be both dense and conductive, and alloys such as Crofer 22 APU have proven promising, growing films with a Cr/Mn spinel composition [106]. However, the film grows on top of an initial chromia subscale, which can still cause issues around chromium volatilization

New Materials for Improved Durability Chapter | 10

209

during early stages and depending on the specific oxidation environment, the interactions between these two films can reduce the blocking effectiveness of the spinel film. Therefore work is still ongoing investigating further protective films to complement the natural spinel films from the Crofer 22 APU such as the (MnCo)3O4 mentioned above [107,108]. Copper-containing spinel films have also received attention in this role such as CuMn1.8O4 and more recently CuFe2O4, which show promise in combining good conductivity and adherence of the film with good Cr vapor suppression characteristics. Ongoing work in this area shows that Cr evaporation and its effects on long-term performance continues to be a concern for SOFC developers and the search continues for ever improved solutions to this issue [109,110].

ACCELERATED TESTING One of the significant barriers to the adoption and integration of new materials into commercial or precommercial development systems is confidence in the long-term behavior of these materials. The materials used by the industrial sector have been studied over years of operation and while these tests may be under optimized conditions they still provide valuable insight into the true behavior of the materials over long-term operation. It is imperative that focus of new materials development not be solely focused on pure performance and that more effort be expended on pushing the performance to longer time spans, even better if these can also begin to replicate more realistic conditions such as with more complex fuels. Unfortunately 40,000 h, the bench mark for SOFC operation is 5 years, which is an unsustainable iteration loop for learning about the long-term behavior of new materials and there is an urgent need to be able to accelerate the testing cycle. However, this is an incredibly difficult proposition as SOFC degradation is a complex interaction of many mechanisms and by increasing one mode of stress (for example, temperature) to accelerate the degradation, there is a significant chance that one may change the primary degradation mechanism leading to erroneous results and addressing noncritical issues [111]. Several groups are now beginning to address this challenge by studying the effects of several life stress factors and the investigating the complex deconvolution of AC impedance response over time or comparing single cell and stack response during various oscillating dynamic duty cycles to try to elucidate critical factors in the degradation behavior, which would allow reliable life testing to be established [112,113]. It is still far from certain if any acceptable method of accelerating SOFC materials testing will be forthcoming. It may initially revolve around individual component testing, however, at some stage it will need to be integrated to the stack level at some point, where the reality is rather more complex and the degradation behavior may be dependent on processing method or interaction with other stack components such as seals or the design of the interconnect flow field [3,114].

210

Solid Oxide Fuel Cell Lifetime and Reliability

Although incredibly difficult and of highly uncertain outcome, if a reliable method of accelerating SOFC durability testing was developed it would definitely be one of the most significant breakthroughs in the field.

SUMMARY While it has been shown that in certain conditions the basic SOFC materials can exhibit good operational lifetimes, issues can still occur when operating in more demanding and arguably realistic conditions. Reduction of operational temperature is still a significant driver here, mainly to minimize degradation mechanisms such as chemical interaction, diffusion, phase stability in electrolytes and coarsening of electrode materials. The use of well-studied materials in new morphologies attained through techniques such as infiltration and exsolution is proving very promising here with significant potential to address some of these issues. With potential advantages for both anode and cathode in terms of reducing operational temperatures and allowing materials that otherwise would not be suitable for SOFC applications to be considered due to reduced processing temperatures. The sensitivity of SOFC to certain fuels and other poisons is well documented, here improving the robustness of the cell components to exposure to these can gain significant advantage both in stack and system design flexibility and cost, again ceramic-based anodes augmented by infiltrated or exsolved catalyst phases are showing encouraging results. Another vital aspect is bringing the individual components together in the stack. Issues here manifest themselves both in physical and chemical issues through consistent contacting, sealing, or chemical interaction. New materials are still being developed here to counter these issues, especially Cr resistant cathodes or new coatings on interconnects; however, the issues are also being tackled though engineering approaches on existing materials such as the investigation of compliance or improving application methods for existing coatings and sealants. While new materials will continue to be investigated and considered for SOFC application more attention needs to be paid to the long-term behavior; however, currently these aspects can only be investigated over long timescales, which is proving to be a significant bottleneck in the adoption of new materials and some form of reliable accelerated testing regime would be a significant step forward; however, this is arguably the most difficult aspect of cell and stack characterization and continues to be elusive. While this chapter has focused on potential new materials for improving durability and robustness in SOFC, many improvements will also be gained from improved control and engineering of existing materials, whether from improved morphology and microstructure in the materials themselves or the constancy or robustness of the stack manufacturing processes and these should not be underestimated in the materials development cycle.

New Materials for Improved Durability Chapter | 10

211

REFERENCES [1] Koch S, Hendriksen PV, Mogensen M, Liu Y-L, Dekker N, Rietveld B, et al. Solid oxide fuel cell performance under severe operating conditions. Fuel Cells 2006;6:130 6. [2] Huang K, Singhal SC. Cathode-supported tubular solid oxide fuel cell technology: a critical review. J Power Sources 2013;237:84 97. [3] Blum L, de Haart LGJ, Malzbender J, Margaritis N, Menzler NH. Anode-supported solid oxide fuel cell achieves 70,000 hours of continuous operation. Energy Technol 2016;4:939 42. [4] Lee DS, Kim WS, Choi SH, Kim J, Lee HW, Lee JH. Characterization of ZrO2 co-doped with Sc2O3 and CeO2 electrolyte for the application of intermediate temperature SOFCs. Solid State Ionics 2005;176:33 9. [5] Omar S, Najib WB, Bonanos N. Conductivity ageing studies on 1M10ScSZ (M4 1 5 Ce, Hf). Solid State Ionics 2011;189:100 6. [6] Politova TI, Irvine JTS. Investigation of scandia yttria zirconia system as an electrolyte material for intermediate temperature fuel cells—influence of yttria content in system (Y2O3)x(Sc2O3)(112x)(ZrO2)89. Solid State Ionics 2004;168:153 65. [7] Haering C, Roosen A, Schichl H. Degradation of the electrical conductivity in stabilised zirconia systems Parts I and II. Solid State Ionics 2005;176 253 259 and 261 268 [8] Jung HY, Hong K-S, Jung H-G, Kim H, Kim H-R, Son J-W, et al. SOFCs with Sc-doped zirconia electrolyte and co-containing perovskite cathodes. J Electrochem Soc 2007;154: B480. [9] Kharton VV, Naumovich EN, Yaremchenko AA, Marques FMB. Research on the electrochemistry of oxygen ion conductors in the former Soviet Union. J Solid State Electrochem 2001;5:160. [10] Jiang N, Wachsman ED, Jung S. ”A higher conductivity Bi2O3-based electrolyte. Solid State Ionics 2002;150:347. [11] Boivin JC, Mairesse G. Recent material developments in fast oxide ion conductors. Chem Mater 1998;10:2870. [12] Jung DW, Duncan KL, Wachsman ED. Effect of total dopant concentration and dopant ratio on conductivity of (DyO1.5)x (WO3)y (BiO1.5)12x2y. Acta Mater 2010;58:355. [13] Zhang X, Robertson M, Deces-Petit C, Xie Y, Hui R, Qu W, et al. Solid oxide fuel cells with bi-layered electrolyte structure. J Power Sources 2008;175:800. [14] Liu M, Ding D, Bai Y, He T, Liu M. An efficient SOFC based on samaria-doped ceria (SDC) electrolyte. J Electrochem Soc 2012;159(6):B661 5. [15] Wachsman ED, Lee KT. Lowering the temperature of solid oxide fuel cells. Science 2011;334:935. [16] Wuensch BJ, Eberman KW, Heremans C, Ku EM, Onnerud P, Yeo EME, et al. Connection between oxygen-ion conductivity of pyrochlore fuel-cell materials and structural change with composition and temperature. Solid State Ionics 2000;129:111. [17] Ishihara T, Matsuda H, Takata Y. Doped LaGaO3 perovskite type oxide as a new oxide ionic conductor. J Am Chem Soc 1994;116:3801. [18] Lin Y, Barnett SA. Co-firing of anode-supported SOFCs with thin La0.9Sr0.1Ga0.8Mg0.2O3 2 δ electrolytes. Electrochem Solid-State Lett 2006;9(6):A285 8. [19] Gao Z, Miller EC, Barnett SA. A high power density intermediate-temperature solid oxide fuel cell with thin (La0.9Sr0.1)0.98(Ga0.8Mg0.2)O3-δ electrolyte and nano-scale anode. Adv Funct Mater 2014;24:5703 9. [20] Huang K, Wan J-H, Goodenough JB. Increasing power density of LSGM-based solid oxide fuel cells using new anode materials. J Electrochem Soc 2001;148(7):A788 94.

212

Solid Oxide Fuel Cell Lifetime and Reliability

[21] Yuanbo L, Barnett SA. Co-firing of anode-supported SOFCs with thin La0.9Sr0.1Ga0.8Mg0.2O3 2 δ electrolytes. Electrochem Solid-State Lett 2006;9(6):A285 8. [22] Zheng F, Chen Y. ”High temperature electrode reactions of Sr and Mg doped LaGaO3 perovskite. J Mater Sci 2008;43:2058 65. [23] Nishi J, Inagaki T, Nishiwaki F, Kano J, Onishi Y, Yamasaki S, et al. Development of ITSOFC based on lanthanum gallate electrolyte. ECS Trans 2009;17(1):37 51. [24] Yokokawa H. Towards comprehensive description of stack durability/reliability behavior. Fuel Cells 2015;15(5):652 68. [25] Furøy KA, Haugsrud R, Ha€nsel M, Magraso A, Norby T. Role of protons in the electrical conductivity of acceptor-doped BaPrO3, BaTbO3, and BaThO3. Solid State Ionics 2007;178:461 7. [26] Haugsrud R, Norby T. Proton conduction in rare-earth ortho-niobates and ortho-tantalates. Nat Mater 2006;5:193 6. [27] Ryu KH, Haile S. Chemical stability and proton conductivity of doped BaCeO3 BaZrO3 solid solutions. Solid State Ionics 1999;125(1 4):355 67. [28] Kannan R, Singh K, Gill S, Furstenhaupt T, Thangadurai V. Chemically stable proton conducting doped BaCeO3 -no more fear to SOFC wastes. Sci Rep 2013;3:2138. [29] Marrero-Lopez D, Pena-Martinez J, Ruiz-Morales JC, Perez-Coll D, Martin-Sedeno MC, Nunez P. Applicability of La2Mo22yWyO9 materials as solid electrolyte for SOFCs. Solid State Ionics 2007;178:1366 78. [30] Zhang GB, Smyth DM. Protonic conduction in Ba2In2O5. Solid State Ionics 1995;82(34):153 60. [31] Kendrick E, Islam MS, Slater PR. Developing apatites for solid oxide fuel cells: insight into structural, transport and doping properties. J Mater Chem 2007;17:3104 11. [32] Kilner JA. Ionic conductors: feel the strain. Nat Mater 2008;7:838. [33] Yang SM, Lee S, Jian J, Zhang W, Lu P, Jia Q, et al. Strongly enhanced oxygen ion transport through samarium-doped CeO2 nanopillars in nanocomposite films. Nat Commun 2015;6:8588. Available from: http://dx.doi.org/10.1038/ncomms9588 [34] Irvine JTS, et al. Evolution of the electrochemical interface in high-temperature fuel cells and electrolysers. Nat Energy 2016;1:15014. [35] Atkinson A, Barnett S, Gorte RJ, Irvine JTS, McEvoy AJ, Mogensen M, et al. Advanced anodes for high-temperature fuel cells. Nat Mater 2004;3:17 27. [36] Irvine JTS, Connor P. In: Irvine JTS, Connor P, editors. Solid oxide fuels cells: facts and figures. London: Springer; 2013. p. 163 80. [37] Sarantaridis D, Atkinson A. Redox cycling of Ni-based solid oxide fuel cell anodes: a review. Fuel Cells 2007;07(3):246 58. [38] Murray EP, Tsai T, Barnett SA. A direct-methane fuel cell with a ceria-based anode. Nature 1999;400:649 51. [39] Jiang SP, Zhang S, Zhen YD, Wang W. Fabrication and performance of impregnated Ni anodes of solid oxide fuel cells. J Am Ceram Soc 2005;88(7):1779 85. [40] Liu L, Sun K, Zhou X, Li X, Zhang M, Zhang N. Sulfur tolerance improvement of NiYSZ anode by alkaline earth metal oxide BaO for solid oxide fuel cells. Electrochem Comm 2012;19:63 6. [41] Ma J, Jiang C, Connor PA, Cassidy M, Irvine JTS. Highly efficient, coking-resistant SOFCs for energy conversion using biogas fuels. J Mater Chem A 2015;3:19068. [42] Jiang SP. A review of wet impregnation—an alternative method for the fabrication of high performance and nano-structured electrodes of solid oxide fuel cells. Mater Sci Eng A 2006;418:199 210.

New Materials for Improved Durability Chapter | 10

213

[43] Verbraeken MC, et al. Modified strontium titanates: from defect chemistry to SOFC anodes. RSC Adv 2014;5:1168 80. [44] Kobsiriphat W, et al. Nickel- and ruthenium-doped lanthanum chromite anodes: effects of nanoscale metal precipitation on solid oxide fuel cell performance. J Electrochem Soc 2010;157:B279 84. [45] Ruiz-Morales JC, et al. Disruption of extended defects in solid oxide fuel cell anodes for methane oxidation. Nature 2006;439:568 71. [46] Savaniu C-D, Miller DN, Irvine JTS. Scale up and anode development for La-doped SrTiO3 anode-supported SOFCs. J Am Ceram Soc 2013;96:1718 23. [47] Boulfrad S, Nechache A, Cassidy M, Traversa E, Irvine JTS. Electrochemical impedance spectroscopy investigation of the anodic functionalities and processes in LSCM CGO Ni systems. ECS Trans 2015;68:2011 18. [48] Tao S, Irvine JTS. A redox-stable efficient anode for solid-oxide fuel cells. Nat Mater 2003;2:320 3. [49] Sengodan S, et al. Layered oxygen-deficient double perovskite as an efficient and stable anode for direct hydrocarbon solid oxide fuel cells. Nat Mater 2014;14:205 9. [50] Jung W, Gu KL, Choi Y, Haile SM. Robust nanostructures with exceptionally high electrochemical reaction activity for high temperature fuel cell electrodes. Energy Environ Sci 2014. Available from: http://dx.doi.org/10.1039/c3ee43546f [51] Cassidy M, Boulfrad S, Irvine J, Chung C, Jorger M, Munnings C, et al. Integration of oxide anodes into the Rolls-Royce IP-SOFC concept. Fuel Cells 2009;9:891 8. [52] Corre G, Kim G, Cassidy M, Vohs JM, Gorte RJ, Irvine JTS. Activation and ripening of impregnated manganese containing perovskite SOFC electrodes under redox cycling. Chem Mater 2009;21:1077 84. [53] Jiang SP, He YM, He TM, Ho SB. Nanostructured palladium La0.75Sr0.25Cr0.5Mn0.5O3/ Y2O3 ZrO2 composite anodes for direct methane and ethanol solid oxide fuel cells. J Power Sources 2008;185(1):179 82. [54] Barison S, Fabrizio M, Mortalo C, Antponucci P, Modafferi V, Gerbasi R. Novel Ru/ La0.75Sr0.25Cr0.5Mn0.5O3-δ catalysts for propane reforming in IT-SOFCs. Solid State Ionics 2010;181:285 91. [55] Marina OA, Canfield NL, Stevenson JW. Thermal, electrical, and electrocatalytical properties of lanthanum-doped strontium titanate. Solid State Ionics 2002;149:21. [56] Slater PR, Fagg DP, Irvine JTS. Synthesis and electrical characterisation of doped perovskite titanates as potential anode materials for solid oxide fuel cells. J Mater Chem 1997;7:2495 8. [57] li X, Zhou H, Zhou X, Xu N, Xie Z, Chen N. Electrical conductivity and structural stability of La-doped SrTiO3 with A-site deficiency as anode materials for solid oxide fuel cells. Int J Hydrogen Energy 2010;35(15):7913 18. [58] Canales-Vazquez J, Ruiz-Morales JC, Irvine JTS, Zhou W. Sc-substituted oxygen excess titanates as fuel electrodes for SOFCs. J Electrochem Soc 2005;152(7):1458 65. [59] Hui SQ, Petric A. Conductivity and stability of SrVO3 and mixed perovskites at low oxygen partial pressures. Solid State Ionics 2001;143:275 83. [60] Huang Y-H, Dass RI, Xing Z-L, Goodenough JB. Double perovskites as anode materials for solid-oxide fuel cells. Science 2006;312:254 7. [61] Shin TH, Ida S, Ishihara T. Doped CeO2 LaFeO3 composite oxide as an active anode for direct hydrocarbon-type solid oxide fuel cells. J Am Chem Soc 2011;133:19399 407. [62] Verbraeken MC, et al. Short stack and full system test using a ceramic A-site deficient strontium titanate anode. Fuel Cells 2015. Available from: http://dx.doi.org/10.1002/ fuce.201400183

214

Solid Oxide Fuel Cell Lifetime and Reliability

[63] Ramos T, et al. Effect of Ru/CGO versus Ni/CGO co-infiltration on the performance and stability of STN-based SOFCs. Fuel Cells 2014;14:1062 5. [64] Neagu D, et al. Nano-socketed nickel particles with enhanced coking resistance grown in situ by redox exsolution. Nat Commun 2015;6. [65] Nishihata Y, et al. Self-regeneration of a Pd-perovskite catalyst for automotive emissions control. Nature 2002;418:164 7. [66] Neagu D, Tsekouras G, Miller DN, Menard H, Irvine JTS. In situ growth of nanoparticles through control of non-stoichiometry. Nat Chem 2013;5:916 23. [67] Graves C, Ebbesen SD, Jensen SH, Simonsen SB, Mogensen MB. Eliminating degradation in solid oxide electrochemical cells by reversible operation. Nat Mater 2015;14:239 44. [68] Shiozaki R, et al. Partial oxidation of methane over a Ni/BaTiO3 catalyst prepared by solid phasecrystallization. J Chem Soc Faraday Trans 1997;93:3235 42. [69] Abild-Pedersen F, Nørskov JK, Rostrup-Nielsen JR, Sehested J, Helveg S. Mechanisms for catalytic carbon nanofiber growth studied by ab initio density functional theory calculations. Phys Rev B 2006;73:115419. [70] Katz MB, et al. Reversible precipitation/dissolution of precious-metal clusters in perovskite-based catalyst materials: bulk versus surface re-dispersion. J Catal 2012;293:145 8. [71] Juhl M, Primdahl S, Manon C, Mogensen M. Performance/structure correlation for composite SOFC cathodes. J Power Sources 1996;61:173. [72] Barbucci A, Viviani M, Panizza M, Delucchi M, Cerisola G. Analysis of the oxygen reduction process on SOFC composite electrodes. J Appl Electrochem 2005;35:399. [73] Kilner JA, Burriel M. Materials for intermediate-temperature solid-oxide fuel cells. Annu Rev Mater Res 2014;44:365 93. [74] Kilner JA, Skinner SJ, Brongersma HH. The isotope exchange depth profiling (IEDP) technique using SIMS and LEIS. J Solid State Electrochem 2011;15:861 76. [75] Huang Y, Ahn K, Vohs JM, Gorte RJ. Characterization of Sr-doped LaCoO3-YSZ composites prepared by impregnation methods. J Electrochem Soc 2004;151(10):A1592 7. [76] Vohs JM, Gorte RJ. High-performance SOFC cathodes prepared by infiltration. Adv Mater 2009;21:943 56. [77] Nedelec R, Uhlenbruck S, Sebold D, Haanappel VAC, Buchkremer H-P, Stover D. Dense yttria-stabilised zirconia electrolyte layers for SOFC by reactive magnetron sputtering. J Power Sources 2012;205:157 63. [78] Samson AJ, Søgaard M, Hjalmarsson P, Hjelm J, Bonanos N, Foghmoes SPV, et al. Durability and performance of high performance infiltration cathodes. Fuel Cells 2013;13 (No. 4):511 19. [79] Ding D, Li X, Lai SY, Gerdes K, Liu M. Enhancing SOFC cathode performance by surface modification through infiltration. Energy Environ Sci 2014;7:552. [80] Shah M, Voorhees PW, Barnett SA. Time-dependent performance changes in LSCF-infiltrated SOFC cathodes: the role of nano-particle coarsening. Solid State Ionics 2011;187:64 7. [81] Gong Y, Palacio D, Song X, Patel RL, Liang X, Zhao X, et al. Stabilizing nanostructured solid oxide fuel cell cathode with atomic layer deposition. Nano Lett 2013;13:4340 5. [82] Ralph JM, Rossignol C, Kumar R. Cathode materials for reduced-temperature SOFCs. J Electrochem Soc 2003;150(11):A1518 22. [83] Vibhu V, Rougier A, Nicollet C, Flura A, Fourcade S, Penin N, et al. Pr4Ni3O101δ: a new promising oxygen electrode material for solid oxide fuel cells. J Power Sources 2016;317:184 93.

New Materials for Improved Durability Chapter | 10

215

[84] Choi S, Park S, Shin J, Kim G. The effect of calcium doping on the improvement of performance and durability in a layered perovskite cathode for intermediate-temperature solid oxide fuel cells. J Mater Chem A 2015;3:6088. [85] Tarancon A, Skinner SJ, Chater RJ, Hernandez-Ramırez F, Kilner JA. Layered perovskites as promising cathodes for intermediate temperature solid oxide fuel cells. J Mater Chem 2007;17(3175 81):3175. [86] Cassidy M, Irvine J. Characterization of cuprate based cathode structures by AC impedance. ECS Trans 2009;25(2):2689 98. [87] Lee SJ, Yong S-M, Kim DS, Kim DK. Cobalt-free composite cathode for SOFCs: Brownmillerite-type calcium ferrite and gadolinium-doped ceria. Int J Hydrogen Energy 2012;37:17217 24. [88] Amow G, Skinner SJ. Recent developments in Ruddlesden Popper nickelate systems for solid oxide fuel cell cathodes. J Solid State Electrochem 2006;10:538 46. [89] Hilpert K, Das D, Miller M, Peck DH, Weiss R. Chromium vapor species over solid oxide fuel cell interconnect materials and their potential for degradation processes. J Electrochem Soc 1996;143:642 7. [90] Jiang SP, Zhang JP, Apateanu L, Foger K. Deposition of chromium species at Sr-doped LaMnO3 electrodes in solid oxide fuel cells. I. mechanism and kinetics. J Electrochem Soc 2000;147(11):4013 22. [91] Matsuzaki Y, Yasuda I. Dependence of SOFC cathode degradation by chromium-containing alloy on compositions of electrodes and electrolytes. J Electrochem Soc 2001;148 (2):A126 31. [92] Jiang SP, Chen X. Chromium deposition and poisoning of cathodes of solid oxide fuel cells - a review. Int J Hydrogen Energy 2014;39:505 31. [93] Yang M, Bucher E, Sitte W. Effects of chromium poisoning on the long-term oxygen exchange kinetics of the solid oxide fuel cell cathode materials La0.6Sr0.4CoO3 and Nd2NiO4. J Power Sources 2011;196:7313 17. [94] Tsipis EV, Kiselev EA, KolotyginVA, Waerenborgh JC, Cherepanov VA, Kharton VV. Mixed conductivity, Mo¨ssbauer spectra and thermal expansion of (La,Sr)(Fe,Ni)O3 2 δ perovskites. Solid State Ionics 2008;179:2170. [95] Chen XB, Jiang SP. Highly active and stable (La0.24Sr0.16Ba0.6)(Co0.5Fe0.44Nb0.06)O32δ (LSBCFN) cathodes for solid oxide fuel cells prepared by a novel mixing synthesis method. J Mater Chem A 2013;1:4871 8. [96] Lessing PA. A review of sealing technologies applicable to solid oxide electrolysis cells. J Mater Sci 2007;42:3465 76. [97] Canavar M, Kaplan Y. Effects of mesh and interconnector design on solid oxide fuel cell performance. Int J Hydrogen Energy 2015;40:7829 34. [98] Draper R., Antol; R.F., Zafred P.R., US Patent Number 6,379,831 (Assignee-Siemens Wesitnghouse Power Corp.) [99] Szabo P, Costa R, Park M, Kim B, Lee I. Investigation of advanced cathode contacting solutions in SOFC. Proc. 12th European SOFC & SOE forum, Lucerne, Switzerland, B0629; July 2016, p. 1 11. [100] Moran-Ruiz A, Vidal K, LarraBnaga A, Laguna-Bercero MA, Porras-Vazquez JM, Slater PR, et al. J Power Sources 2014;269:509 19. [101] Yang Z, Xia G, Singh P, Stevenson JW. Electrical contacts between cathodes and metallic interconnects in solid oxide fuel cells. J Power Sources 2006;155:246 52. [102] Couse S, Cassidy M. US Patent Number 7,951,509 (Assignee- Bloom Energy Corporation).

216

Solid Oxide Fuel Cell Lifetime and Reliability

[103] Diethelm S, Van herle J, Wuillemin Z, Nakajo A, Autissier N, Molinelli M. Impact of materials and design on solid oxide fuel cell stack operation. J Fuel Cell Sci Technol 2008;5(3):031003 6. [104] Le S, Chu C-L, Tsai M-J, Lee J. High temperature oxidation behavior of interconnect coated with LSCF and LSM for solid oxide fuel cell by screen printing. J. Appl Surf Sci 2010;256:1817 24. [105] Fu Q, Tietz F, Sebold D, WesseL E, Buchkremer H-P. Magnetron-sputtered cobalt-based protective coatings on ferritic steels for solid oxide fuel cell interconnect applications. Corros Sci 2012;54:68 76. [106] Froitzheim J, Meier GH, Niewolak L, Ennis PJ, Hattendorf H, Singheiser L, et al. Development of high strength ferritic steel for interconnect application in SOFCs. J Power Sources 2008;178:163 73. [107] Yang Z, Xia G, Li X, Stevenson JW. (Mn,Co)3O4 spinel coatings on ferritic stainless steels for SOFC interconnect applications. Int J Hydrogen Energy 2007;32:3648 54. [108] Smeacetto F, De Miranda A, Polo SC, Molin S, Boccaccini D, Salvo M, et al. Electrophoretic deposition of Mn1.5Co1.5O4 on metallic interconnect and interaction with glass-ceramic sealant for solid oxide fuel cells application. J Power Sources 2015;280:379 86. [109] Huang W, Gopalan S, Pal UB, Basu SN. Evaluation of electrophoretically deposited CuMn1.8O4 spinel coatings on Crofer 22 APU for solid oxide fuel cell interconnects. J Electrochem Soc 2008;155:B1161 7. [110] Hosseini SN, Karimzadeh F, Enayati MH, Sammes NM. Oxidation and electrical behavior of CuFe2O4 spinel coated Crofer 22 APU stainless steel for SOFC interconnect application. Solid State Ionics 2016;289:95 105. [111] Heneka MJ, Ivers-Tiffee E. Accelerated life tests for fuel cells. ECS Trans 2006;1 (8):377 84. [112] Weber A, Szasz J, Dierickx S, Endler-Schuck C, Ivers-Tiffee E. Accelerated lifetime tests for SOFCs. ECS Trans 2015;68(1):1953 60. [113] Hagen A, Høgh JVT, Barfod R. Accelerated testing of solid oxide fuel cell stacks for micro combined heat and power application. J Power Sources 2015;300:223 8. [114] Steinberger-Wilckens R, Blum L, Buchkremer H-P, de Haart B, Malzbender J, Pap M. Recent results in solid oxide fuel cell development at Forschungszentrum Juelich. ECS Trans 2011;35(1):53 60.

Index Note: Page numbers followed by “f ” and “t” refer to figures and tables, respectively.

A Activation losses, 4 Advanced visualization of SOFCs, 46f Air stream, impurities in, 107 109 carbon dioxide, degradation in the presence of, 108 sulfur dioxide, degradation in the presence of, 109 water vapor, degradation in the presence of, 107 108 Alcohol-based fuels, 39 40 Alcohols, 37 40 Alkali-doped alumina, 153 154 Anode functional layer (AFL), 52, 56 57 Anode lifetime, impact of fuels on, 37, 43 46 C H O ternary diagrams, 38 39, 39f fuel compositions, 38 39 fuel impurities, 41 43 power generation characteristics for fuels, 39 41 reliability, 46 48 Anode poisoning, 184 Anodes for improved durability, 197 202 Anode-supported cells (ASCs), 52 54, 56 59, 174 Anode-supported SOFC manufacture and microstructure, 53 54 Aqueous gel-casting, 10 Area-specific resistance (ASR), 8, 19, 125 126 Atmospheric corrosion, 124 128 Atmospheric plasma spray (APS), 133 Autothermal reforming (ATR) process, 147 Auxiliary power units (APUs), 176

B Balance of plant (BoP), 7, 15, 148 149 subsystems, 103 104

Barium calcium aluminosilicate (BCAS)based glasses, 128 129 Barrier layer of doped ceria, 22 23 BIMEVOX materials, 195 196 Biogas, 37 38, 40 41 Bloom Energy, 12 Butler Volmer equation, 4

C Cahn Hilliard evolution equation, 91 93 CALculation of Phase Diagram (CALPHAD) approach, 108 Carbon and sulfur poisoning, 81 Carbon dioxide, degradation in the presence of, 108 Carbon formation on reforming catalysts, 150 154 Catalyst degradation in reformers, 149 154 carbon formation on reforming catalysts, 150 154 deactivation mechanisms of catalyst metals, 149 150 Catalysts, in fuel processors, 148 149 Catalytic fuel reforming process, 147 types of, 148t Cathode degradation, from airborne contaminants, 101 approaches for the mitigation of chromiumassisted cathode degradation, 112 113 cathode materials, 104 106 degradation in solid oxide fuel cell systems, 103 104 evaporation of chromium impurities under SOFC system conditions, 110 111 degradation in the presence of chromium vapor, 110 111 “real world” air atmosphere, degradation in, 111 impurities in the air stream, 107 109

217

218

Index

Cathode degradation, from airborne contaminants (Continued) carbon dioxide, degradation in the presence of, 108 sulfur dioxide, degradation in the presence of, 109 water vapor, degradation in the presence of, 107 108 Cathode materials, 104 106 Cathode electrolyte barrier layers, reliability and stability of, 183 Cathodes for improved durability, 202 206 Cathode-supported cells, 52 53 Ce1 xGdxO2 x/2 (CGO), 200 201 Ce1 xGdxO2 x/2 yttria-stabilized zirconia interdiffusion in bilayer electrolytes, 22 24 Ceramic anodes, 198 investigation of, 198 199 Ceramic materials, 7 Ceres Power’s stack performance verification program, 185 186 process, 185 186 Ceria, 22 23, 27 CFY, 123 CGO barrier layers, 23 24 Chemical degradation, 14 15 Chemically induced stresses, 27 Chlorine poisoning, 43 Chromium getter, 112 113 Chromium impurities, evaporation of, 110 111 chromium vapor, degradation in the presence of, 110 111 real world” air atmosphere, degradation in, 111 Chromium poisoning, 110 111, 111f, 113f of LSCF-based cathodes, 184 of manganite-based cathodes, 183 184 Chromium tolerance, 208 Chromium-assisted cathode degradation approaches for the mitigation of, 112 113 Clean diesel, 163 Coal gas, 37 38, 41 42 Cobaltite-based cathodes, 105 106 Co-doped LSGM (LSGMCo), 196 Coefficient of thermal expansion (CTE), 24, 102 103, 123, 204 205 Coke oven gas, 37 38 Commercially available SOFC current technology status, 11 12 Concentration losses, 4

Conductivity degradation, 20 21 Creep processes, 31 32 Creep strain, 31 32 Crofer 22 APU, 123, 208 209 Crossover losses, 5

D Degradation, 12 15, 47, 103 chemical, 14 15 during cycle durability tests, 48f in SOFC systems, 103 104 physical, 13 14 Degradation models, 91, 92f mechanical models, 93 94 mesoscale models, 93 microscale models, 91 93 Delamination, 14, 83 84 Deposition techniques, 8 Design of solid oxide fuel cell, 5 7 Desulfurization, 148, 164 Direct imaging, 84 Distribution of relaxation times (DRTs), 83 Dopants, 195 196 Doped zirconia, 195 Doped-lanthanum gallates, 196 Dual-beam focused ion beam SEM instrument, 62 Durability and reliability strategies adopted for solid oxide fuel cell development, 179 187 Ceres Power’s stack performance verification program, 185 186 Japanese Programs, 179 184 NEDO program, 183 184 stack testing results at CRIEPI, 181 183 Tokyo Gas, thermal cycle testing at, 183 LG fuel cell systems, 186 187 Durability and robustness improvement, new materials for accelerated testing, 209 210 anodes, 197 202 cathodes, 202 206 solid oxide fuel cell electrolytes, 194 197 stack materials, 206 209 Durability of SOFCs, 42 43

E Electrochemical impedance spectroscopy (EIS), 81 84, 107 108 Electrode fabrication techniques, 10

Index Electrode imaging and quantification to measure degradation, 84 91 3D imaging applied to measuring solid oxide fuel cell electrode degradation, 86 91 changes caused by poisoning, 91 changes in volume and redox cycling, 88 91 coarsening, sintering, and dewetting, 86 88 introduction to approaches, 84 86 Electrode microstructure, 79 80 Electrolyte, 19, 194 197 chemical interactions, 21 24 Ce1 xGdxO2 x/2 yttria-stabilized zirconia interdiffusion in bilayer electrolytes, 22 24 La1 xSrxMnO3/Yttria-stabilized zirconia interactions, 21 22 mechanical degradation, 24 33 chemically induced stresses, 27 creep, 31 32 mechanical failure, 27 29 slow crack growth, 29 31 thermal and redox cycling, 33 thermal stresses, 24 26 structural stability of, 20 21 Electrolyte-supported cells (ESCs), 52 53, 174 Electron diffraction, 62 63 Electron energy loss spectroscopy (EELS), 62 63 Emergency stop” (E-stop), 13 ENE-FARM project, in Japan, 12, 179 184 Environmental SEM (ESEM), 65 Environmental TEM (ETEM), 62 63 Equivalent circuit modeling, 83 Exsolution, 201 202

F Finite element modeling (FEM) approaches, 26 Finite elements analysis (FEA) model, 131 132 Fluorite-type conductors, 195 196 Focused ion beam (FIB) technique, 45 46 4D” tomography, 84 Fracture toughness, 27 28 Fuel cell power system (FCPS), 185 Fuel cells, defined, 1 Fuel delivery, 154 155

219

Fuel flexibility, 44 Fuel impurities, 41 43, 41f Fuel injectors, 148 149 Fuel processing, 146 149 fuel processors. See Fuel processors stages of, 147 148 catalytic fuel reforming process, 147 desulfurization, 148 postreforming process, 148 Fuel processors component of liquid fuel processor for solid oxide fuel cells, 154 components of, 148 149 design factors for, 154 155 lifetime estimation of, 164 169 durability test method for fuel processors, 166 engineering issues (BOPs) of fuel processors on durability, 164 165 lifetime extension of fuel processors, 167 169 practical example of durability test, 167 lifetime of, 149 Fuel reforming, 147, 154 FuelCell Energy, 12 Fundamentals of SOFC, 3 5 activation losses, 4 concentration losses, 4 crossover losses, 5 Ohmic losses, 4

G Gadolinium (Gd)-doped ceria-based interlayer, 106 107 Gadolinium-doped ceria (GDC)-ESB electrolyte concept, 195 196 Gas tightness, 13 Gasoline, 37 38, 167 Generic durability/reliability issues for SOFC, 176 179 Glass transition temperature, 208 Goldschmidt tolerance factor, 105

H H2S poisoning, 42 High surface area (HSA) electrode materials, 105 High temperature SOFCs (HT-SOFCs), 7 History of solid oxide fuel cells, 1 3 Hydrocarbons, 146, 148, 154

220

Index

Hydrocarbons (Continued) desulfurizer for heavy hydrocarbons, 161 163 Hydrodesulfurization (HDS), 148, 163 164

I IC degradation, 122 solutions to decrease, 132 136 Impurity Concentration Threshold, 45t Infiltration techniques, 196, 201, 204, 210 Interconnects (ICs), 5 6, 121 Interdiffusion, 23 Intermediate temperature SOFCs (IT-SOFCs), 7 8, 10, 102 103, 204 I V characteristics of an SOFC, 39 41, 40f

J Japanese programs, 179 184 degradation mechanisms identified and mitigated in the NEDO program, 183 184 anode poisoning, 184 chromium poisoning of LSCF-based cathodes, 184 chromium poisoning of manganite-based cathodes, 183 184 reliability and stability of cathode electrolyte barrier layers, 183 sulfur poisoning of cathodes, 184 stack testing results at CRIEPI, 181 183 thermal cycle testing at Tokyo Gas, 183

K Kerosene, 37 38 Korea Advanced Institute of Science and technology (KAIST), 160 KW-class reformer for reliable solid oxide fuel cell system, 158 159

L La1 xSrxCo1 yFeyO3 (LSCF), 22 24 La1 xSrxCo1 yFeyO3 δ (LSCF), 102 103, 105 106 La1 xSrxCoO3 δ (LSC), 102 103 La1 xSrxFeO3 δ (LSF), 102 103, 105 106 La1 xSrxGa1 yMgyO3 (LSGM), 196 La1 xSrxMnO3/yttria-stabilized zirconia (LSM/YSZ) interactions, 21 22

La1 xSrxMnO3 δ (LSM), 102 103, 106 107 Lanthanum ferrite (LaFeO3)-based cathodes, 105 106 Lanthanum manganite (LaMnO3) cathodes, 105 106 Lanthanum nickelate cathodes, 205f Lanthanum strontium cobalt ferrite (LSCF), 183 -based cathodes, chromium poisoning of, 184 triple-phase boundary (TPB), 183 184 Lanthanum strontium cobaltites (LSCs), 195 Lanthanum strontium manganite (LSM) cathodes, 193 194 Lanthanum strontium titanate (LST)-based perovskites, 199 200 Layered perovskite, 204 LG fuel cell systems (LGFCSs), 186 187 identification and reduction of cell degradation mechanisms, 186 187 Life and reliability of SOFC-based products durability and reliability strategies, 179 187 Ceres Power’s stack performance verification program, 185 186 Japanese Programs—NEDO and ENEFARM, 179 184 LG fuel cell systems, 186 187 generic durability/reliability issues for SOFC, 176 179 SOFC technology generations and applications, 173 175 Lifetime issues for SOFC interconnects, 121 lifetime behavior of stacks and cells tested in operating conditions, 136 140 metal interconnects, 122 136 degradation, 124 132 IC degradation, solutions to decrease, 132 136 Liquefied petroleum gas, 37 38 Liquid fuel processor designs to enhance reliability, 154 159 component of liquid fuel processor for solid oxide fuel cells, 154 design factors for fuel processor, 154 155 fuel delivery design of liquid fuel processing, 155 158 kW-class reformer for reliable solid oxide fuel cell system, 158 159

Index

M M factor, 66 67 Manganite-based cathodes, chromium poisoning of, 183 184 Materials selection, 8 9 Mechanical stress, 130 132 Metal ICs (MICs), 122, 125 Metal interconnects, 122 136 degradation, 124 132 atmospheric corrosion, 124 128 mechanical stress, 130 132 sealing corrosion, 128 130 IC degradation, solutions to decrease, 132 136 Metal-supported cells (MSCs), 174 MicroCHP, 179 Microstructural degradation, 79 electrode imaging and quantification to measure, 84 91 3D imaging applied to measuring SOFC electrode degradation, 86 91 introduction to approaches, 84 86 impedance for identifying changes in microstructure, 81 84 mechanisms, 80 81 microstructural design strategies, 94 modeling of, 91 94, 92f Microstructural design, 9 11 Microstructure, 10, 13 evolution, 13 Mitsubishi Heavy Industries, 12 Mixed conductivity, 202 203 Mixed ionic electronic conductor (MIEC), 9, 102 103, 198 199, 202 203 Mode of loading, 27 28 Molybdenum, 163 164 Monolith-type reformers, 158, 159f

N Naphtha, 37 38 National Energy Technology Laboratory (NETL), 12 13 Natural gas, 37 38 NEDO program, 179, 183 184 anode poisoning, 184 cathode electrolyte barrier layers, reliability and stability of, 183 chromium poisoning of LSCF-based cathodes, 184 chromium poisoning of manganite-based cathodes, 183 184

221

sulfur poisoning of cathodes, 184 Nernst equation, 3 4 Nernst Glower, 2 Nernst Mass, 2 Nernst potential, 3 5 New Energy and Industrial Technology Development Organization (NEDO), 12 Ni particles, preparation, 201 202 Ni/YSZ cermets, 82f Ni/yttria-stabilized zirconia (YSZ) anode cermets, 193 194 Nickel, 51 52 Nickel microstructural evolution, 80 81 Nickel oxidation, 47 48, 55 59 Nickel oxide, 53 54 Nickel/NiO interfaces, 64

O Obtrusive YSZ network, 69 70 Ohmic losses, 4, 11 Ohmic resistance, 81 82 Operating temperature and materials, 7 11 materials selection, 8 9 microstructural design, 9 11 Ostwald ripening mechanism, 47 48 Outer container, 148 149 Oxidation process, 51 52 Oxygen reduction reaction (ORR), 102 103

P Partial oxidation (POX) process, 147 Particle size distributions (PSDs), 10 11 Perovskite oxides, 196 197 Perovskites, 104 105 Physical degradation, 13 14 Physical vapor deposition (PVD), 133 Plansee ITM, 123 Poisoning, 14 Poisson’s ratio, 24, 56 Postprocessing in reforming to enhance lifetime of solid oxide fuel cells, 159 164 catalysts for desulfurization, 163 164 concept of postreforming, 159 161 desulfurizer for heavy hydrocarbons, 161 163 sulfur-containing hydrocarbon source, 162 163

222

Index

Postreforming process, 148 Power generation characteristics for fuels, 39 41 Praseodymium-based perovskites, 204 Pressure-swirl nozzle, 157 Project Gemini, 1

R Ratcheting, 33 Real world” air atmosphere, degradation in, 111 Redox cycle degradation, temperature effects on, 71f Redox cycling, 33, 51 anode-supported solid oxide fuel cell manufacture and microstructure, 53 54 electrochemical performance, impact on, 57 60 kinetics of, 54 56 mechanical considerations, 56 57 microstructural changes, 60 70 solutions to redox cycle degradation, 70 73 Redox exsolution, 201 202 Redox reactions, 9 Reduction process, 51 52 Reformate cleaning, 154 Residual stresses, 24 25

S Samaria-doped ceria, 51 Sandvik Sanergy HT, 123 Screen-printing, 10 Sealing corrosion, 128 130 Secondary ion mass spectroscopy (SIMS), 184 Segmented-in-series (SIS), 173 174 Siemens-Westinghouse tubular design, 59 60 Siloxane, 41 42 Sintering, 149 SOFConnex system, 208 Sol gel thermolysis, 10 Solid oxide electrolysis cell (SOEC) testing, 73 Solid oxide fuel cell (SOFC) anodes, 51 53, 56, 71 73, 102, 102f Solid State Energy Conversion Alliance (SECA), 12 Sr-doped lanthanum manganite (LaMnO3) cathodes, 105 106 Stack materials, 206 209

Stack testing results at CRIEPI, 181 183 Stacks and cells tested in operating conditions, lifetime behavior of, 136 140 Stacks/cells based on different structural technology, long-term tests of, 137t Steam reforming (SR) process, 147 Strength-based assessment, 28 29 Strength robability time (SPT) relationship, 30 Substrate-supported cells, 52 Sulfur dioxide, 109 degradation in the presence of, 109 Sulfur poisoning, 42 43, 81 of cathodes, 184

T Thermal cycle testing at Tokyo Gas, 183 Thermal cycling, 33 Thermal degradation, 149 Thermal spray drying, 10 Thermal stresses, 24 26 Thermochemical stability of fuels, 38 Three-dimensional (3D) imaging, for measuring SOFC electrode degradation, 84, 86 91 changes caused by poisoning, 91 changes in volume and redox cycling, 88 91 coarsening, sintering, and dewetting, 86 88 Tokyo Gas, thermal cycle testing at, 183 Tomography, 84, 95 Transmission electron microscopy (TEM) samples, 60 62 Triple phase boundaries (TPBs), 9, 10f, 51, 79 81, 85, 88, 90 91, 102 Tungsten, 163 164 Twin-fluid nozzle, 157, 157f

U Ultrasonic injector (UI), 156 157, 156f specifications of, 157t

W Water vapor, degradation in the presence of, 107 108 Wet powder spray (WPS), 133 134 Whisker carbon, 150 153

Index

X

8 mol% Y2O3 (8YSZ), 20 21 YSZ lanthanum strontium cobaltite ferrite (LSCF) cell, 9

X-ray diffraction (XRD), 69

Y Young’s modulus, 56 Yttria-stabilized zirconia (YSZ), 7 9, 51, 53 56 3YSZ, 20 21, 29

223

Z Zirconia electrolyte, 20 23, 25, 195 ionic conductivity of, 20f