Mechanical Characterization of Coating-Interconnect Interfaces and Anode-

Electrolyte Interfaces for Solid Oxide Fuel Cells

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy

in the Graduate School of The Ohio State University

By

Sajedur Rahman Akanda, M.Sc

Graduate Program in Mechanical Engineering

The Ohio State University

2012

Dissertation Committee:

Professor Mark E. Walter, Advisor

Professor Noriko Katsube

Professor Brian Harper

Professor Daniel Mendelsohn

Copyright by

Sajedur Rahman Akanda

2012

ii

Abstract

A planar solid oxide fuel cell (SOFC) consists of multiple layers of dissimilar materials

with distinct physical, mechanical, and thermal properties. High operating temperatures

and mechanical loadings during service can significantly weaken the interfaces of

different components in an SOFC. The strength and integrity of various interfaces, for

example, coating-interconnect interfaces and electrode-electrolyte interfaces play an

important role in increased power density of an SOFC.

In the first part of the present investigation, the interfaces between oxide coatings and

interconnects are characterized. The repeating anode-electrolyte-cathode units in a planar

SOFC stack are physically separated by electrically conductive interconnects. With the

reduction of operating temperature to 800oC, it is possible to replace lanthanum based

ceramics with less expensive, more readily available chromium alloyed iron metals as

interconnects. However, when incorporating chromium-alloyed interconnects, steps must

be taken to inhibit chromium poisoning of cathodes. To prevent the chromium poisoning,

a dense manganese cobalt spinel oxide (MCO) coating is applied on the cathode side

surface of interconnect prior to its installation in the fuel cell. But highly ceramic brittle

nature of MCO coatings makes them susceptible to damage under mechanical loads and

thermal stresses developed during cooling down the fuel cell from operating temperature

to room temperature. A room temperature four-point bend experiment is designed to

iii

assess the quality of coatings and coating adhesion. Resulting tensile cracking patterns on

the coatings on the convex side of the bend specimen are used to quantify the interfacial

shear strength from a shear lag model. In addition, the onset strain of coating spallation is

incorporated in an energy based fracture mechanics model to obtain the interfacial

fracture energy. Images from scanning electron microscopy (SEM) of the tested coating

surfaces are processed to analyze the interface failure mechanisms, the crack spacing, and

the spalled areas at higher strains. The analysis obtained from the present investigation is

able show distinct differences between coatings processed with different parameters. In

addition, based on the results obtained from the bend experiments, coating lifetime is

predicted. Lifetime prediction of coatings will greatly assist in optimizing the coating

process parameters and assessing the reliability of coated interconnects.

In the second part of this dissertation, anode-electrolyte interfaces at which the important

electro-chemical hydrocarbon fuel reactions take place are investigated. Frequent

anticipated and unanticipated shut down and startup of fuel cells can cause delamination

and failure of the anode-electrolyte interfaces. Room temperature four-point bend

experiments are performed to obtain the interfacial fracture energy of the anode-

electrolyte interfaces. The notched bend test specimens consist of NiO-YSZ anode and

ScSZ electrolyte bi-layers are sandwiched between two steel stiffeners. A stable crack is

forced to propagate along the interfaces and is monitored with a long distance camera

lens. The constant load at which the stable crack propagates is recorded and utilized to

obtain the critical strain energy release rate of the interfaces. The cracked surfaces are

studied with SEM and energy dispersive spectroscopy (EDS).

iv

Dedication

To my parents

Abdur Razzak Akanda

Sabiha Sultana

v

Acknowledgments

The author would like to express his sincere and deepest gratitude to Professor Mark E.

Walter for his constant encouragements and guidance to conduct this research. The

author’s thanks are also extended to Matthew M. Seabaugh, Director, Nexceris for his

valuable advices time to time. The author would also like to mention the name of Neil J.

Kidner of NexTech Materials with thankfulness for his constant technical supports and

advices. The author also thanks all the employees and staffs of NexTech Materials Ltd.

The author appreciates the technical supports from Sean O Fallon of TestResources,

Cameroon Begg of CEOF and Chad Bivens of Machine Lab. The author acknowledges

herewith the moral supports received from Ryan Berke, Angel Suresh, Bodhayan Dev,

Andrew Davis and Eric Belknap of Experimental Mechanics of Materials laboratory

(EMML).

This project was financially supported by National Science Foundation (NSF) CMMI

GOALI Grant No. 082558. NSF’s contribution to this project is greatly acknowledged.

Special gratitude is expressed to author’s parents for their unconditional love and support.

Author also likes to mention the name of Nusrat Sharmin for her love. Finally author

acknowledges that it is the God’s blessings only which shone light upon the deepest dark

throughout the years.

vi

Vita

1999................................................................Notre Dame College, Dhaka, Bangladesh

2005................................................................B.Sc. Mechanical Engineering, Bangladesh

University of Engineering and Technology

(BUET), Dhaka, Bangladesh

2007................................................................M.Sc. Mechanical Engineering, Tuskegee

University, Tuskegee, AL, USA

2008 to 2012 .................................................Graduate Research Associate, Department

of Mechanical and Aerospace Engineering,

The Ohio State University, Columbus, OH,

USA

Publications

Akanda S.R., Walter M.E., Kidner N.J., Seabaug M.M, Mechanical characterization of

oxide coating–interconnect interfaces for solid oxide fuel cells, Journal of Power Sources

210 (2012) 254-262.

Zhou Y., Akanda S.R., Jeelani S., Lacy T.E., Nonlinear constitutive equation for vapor-

grown carbon nanofiber-reinforced SC-15 epoxy at different strain rate, Material Science

and Engineering A465 (2007) 238-246.

vii

Rahman M.A., Akanda S.R., Hossain M.A., Effect of Cross-section Geometry on the

Response of SAM Column, Journal of Intelligent Material Systems and Structures 19

(2008) 243-252.

Fields of Study

Major Field: Mechanical Engineering

viii

Table of Contents

Abstract…………………………………………………………………………………....ii

Dedication………………………………………………………………………………...iv

Acknowledgments…………………………………………………………………………v

Vita………………………………………………………………………………………..vi

List of Tables…………………………………………………………………………….xii

List of Figures…………………………………………………………………………...xiii

Chapter 1: Introduction……………………………………………………………………1

1.1 Background…............................................................................................................1

1.1.1 Basic Principles of an SOFC…………………………………………………...1

1.1.2 SOFC Materials ………………………………………………………………..2

1.1.3. Applications of SOFCs……………………………………………...................3

1.2 Problem Statement………………………………………………………………….4

1.3 Dissertation Structure ………………………………………………………………5

References………………………………………………………………………………9

Figures…………………………………………………………………………………10

Chapter 2: Adhesion of Reduced and Oxidized Spinel Coatings on Metallic Interconnects

……………………………………………………………………………………………12

Abstract………………………………………………………………………………..12

ix

2.1 Introduction………………………………………………………………………..13

2.1.1 Requirements of Interconnects………………………………………………..13

2.1.2 Drawback of Metallic Interconnects…………………………………………..15

2.1.3 Protective Coatings on Interconnects…………………………………………15

2.1.4 Experimental Techniques for Measuring Coating Adhesion…………………18

2.2 Experimental………………………………………………………………………20

2.2.1 Materials………………………………………………………………………20

2.2.2 Bend Experiments……………………………………………………………..21

2.3 Theory……………………………………………………………………………..22

2.3.1 Interfacial Shear Strength……………………………………………………..22

2.3.2 Interfacial Fracture Energy……………………………………………………25

2.4 Results……………………………………………………………………………..27

2.4.1 Cumulative AE………………………………………………………………..27

2.4.2 Interfacial Shear Strength……………………………………………………..29

2.4.3 Interfacial Fracture Energy……………………………………………………30

2.5 Discussions………………………………………………………………………...32

2.5.1 Effects of Reduction and Oxidation Heat Treatment…………………………32

2.5.2 Effects of Interconnect Compositions………………………………………...34

2.6 Conclusions………………………………………………………………………..34

References……………………………………………………………………………..36

Tables………………………………………………………………………………….38

Figures…………………………………………………………………………………39

x

Chapter 3: Lifetime of MCO Coatings on Metallic Interconnects………………………50

Abstract………………………………………………………………………………..50

3.1 Introduction………………………………………………………………………..51

3.2 Materials…………………………………………………………………………...54

3.3 Analytical Model for Residual Stress Distributions……………………………….55

3.3.1 Background……………………………………………………………………55

3.3.2 Interface Displacements……………………………………………………….56

3.3.3 Equilibrium Relations and Governing Equations……………………………..58

3.3.4 Boundary Conditions………………………………………………………….59

3.3.5 Analytical Interfacial Fracture Energy………………………………………..61

3.4 Results and Discussions…………………………………………………………...62

3.4.1 Experimental Results………………………………………………………….62

3.4.2 Analytical Results……………………………………………………………..64

3.4.3 Lifetime of MCO……………………………………………………………...65

3.5 Conclusions………………………………………………………………………..67

References……………………………………………………………………………..69

Tables………………………………………………………………………………….71

Figures…………………………………………………………………………………72

Chapter 4: Effects of reduction Heat Treatments on Coating Performances………….....86

Abstract………………………………………………………………………………..86

4.1 Introduction………………………………………………………………………..87

4.2 Experimental………………………………………………………………………88

xi

4.3 Results……………………………………………………………………………..88

4.4 Conclusions………………………………………………………………………..90

References……………………………………………………………………………..92

Tables………………………………………………………………………………….93

Figures…………………………………………………………………………………94

Chapter 5: Investigating Anode-Electrolyte Interface by a Steady-State Crack

Propagation………………………………………………………………………………99

Abstract………………………………………………………………………………..99

5.1 Introduction………………………………………………………………………100

5.2 Experimental……………………………………………………………………..101

5.2.1 Test Specimen………………………………………………………………..101

5.2.2. Strain Energy Release Rate…………………………………………………102

5.3 Results……………………………………………………………………………104

5.4 Conclusions………………………………………………………………………106

References……………………………………………………………………………108

Tables………………………………………………………………………………...109

Figures………………………………………………………………………………..110

Chapter 6: Conclusions and Future Works……………………………………………..118

6.1 Conclusions………………………………………………………………………118

6.2 Future Works……………………………………………………………………..121

Bibliography……………………………………………………………………………122

Appendix A: List of Symbols…………………………………………………………..128

xii

List of Tables

Table 2.1. Mechanical and thermal properties of MCO coating, native scale and

interconnect………………………………………………………………………………38

Table 2.2. Average values of critical tensile stress ( ), interfacial shear strength (

),

experimental interfacial fracture energy ( ), and % spall area for each type of test

specimen…………………………………………………………………………………38

Table 3.1. Average critical tensile stress of coating (σcct), Average saturated crack spacing

(l), interfacial shear strength (τsti1

), interfacial fracture energy (Gexpi1

), and % spall area

for each type of test specimen……………………………………………………………71

Table 4.1. Process conditions for standard and modified heat treatment………………..93

Table 4.2. Interfacial fracture energy of standard and modified fired test specimen……93

Table 5.1. Mechanical and thermal properties of anode and electrolyte………….........109

xiii

List of Figures

Figure 1.1. A schematic diagram of an SOFC with its components and basic principles….

……………………………………………………………………………………………10

Figure 1.2. (a) FlexCellTM

. (b) An SEM micrograph of an anode-electrolyte bi-layer…..11

Figure 2.1. Steps of chromium poisoning of a cathode………………………………….39

Figure 2.2. Steps of processing MCO coatings on interconnects……………………….40

Figure 2.3. SEM images of (a) Reduced MCO on Crofer (tested). (b) Oxidized MCO on

Crofer (tested)……………………………………………………………………………41

Figure 2.4. Experimental setup to characterize coating-interconnect interfaces………...42

Figure 2.5. (a) Schematic of failure mechanisms of a brittle coating during bend

experiments. (b) Shear stress distributions at interface and tensile stress distributions in a

coating segment………………………………………………………………………….43

Figure 2.6. (a) Experimental stress-strain curve synchronized with AE data. (b) An

enlarged view of region 1 of cumulative AE data……………………………………….44

Figure 2.7. Saturate parallel in-plane transverse cracks in (a) Reduced MCO on SS441.

(b) Oxidized MCO on Crofer……………………………………………………………45

Figure 2.8. Calculated values of interfacial shear strength for each type of test

specimen…………………………………………………………………………………46

xiv

Figure 2.9. (a) Interfacial fracture due to buckling of coating. (b) SEM image of MCO

buckling before spallation (reduced MCO -SS441)……………………………………..47

Figure 2.10. Representative SEM and processed images of MCO coating surfaces at 3%

strain (a) Reduced MCO-SS441. (b) Oxidized MCO-SS441. (c) Reduced MCO-Crofer.

(d) Oxidized MCO-Crofer……………………………………………….........................48

Figure 2.11. SEM and EDS analysis of spalled sections of (a) Reduced MCO on SS441.

(b) Oxidized MCO on Crofer…………………………………………………………….49

Figure 3.1. Integrated experimental-analytical methodology to predict lifetime of MCO

coatings on metallic interconnects……………………………………………………….72

Figure 3.2. SEM images of MCO coatings oxidized at 900oC for (a) 100 hours. (b) 600

hours. (c) 1000 hours…………………………………………………………………….73

Figure 3.3. Native scale growth kinetics at 900oC……………………………………….74

Figure 3.4. Normal forces, shear forces and bending moments generated in each layer due

to experiencing temperature differences.………………………………………………...75

Figure 3.5. Differential element of each layer subjected to normal force, shear force and

bending moment………………………………………………………………………….76

Figure 3.6. Steps to obtain analytical interfacial fracture energy………………………..77

Figure 3.7. Cumulative AE data synchronized with strain………………………………78

Figure 3.8. Saturated parallel in-plane transverse cracks in 900oC-1000 hour oxidized

MCO……………………………………………………………………………………..79

Figure 3.9. MCO coatings with native oxide scale in spalled area………………………80

xv

Figure 3.10. Representative SEM and processed images of MCO coating surfaces at 3%

strain (a) 900oC-100 hour. (b) 900

oC-600 hour. (c) 900

oC-1000 hour…………………..81

Figure 3.11. (a) Variation of normal compressive stress of MCO and curvature of

composite along x-direction. (b) Resultant stress distribution in MCO thru MCO

thickness. (hn = 6.14 μm)………………………………………………………………...82

Figure 3.12. (a) Residual shear stress distribution at MCO-native scale interface τri1

(hn =

6.14 μm). (b) Shear strength from experiments τsti1

and maximum shear stress from

analytical model τr,maxi1

as a function of native scale thickness………………………….83

Figure 3.13. (a) Determination of critical native scale thickness to initiate MCO spallation

during cooling from 750oC to room temperature. (b) Native scale growth kinetics at

750oC…………………………………………………………………………………….84

Figure 3.14. Projected lifetime of MCO coatings as a function of operating temperature

of a fuel cell……………………………………………………………………………...85

Figure 4.1. A schematic diagram of an ASR setup………………………………………94

Figure 4.2. SEM observation of microstructure of (a) Standard MCO. (b) Modicfied

MCO. ………………...………………………………………………………………….95

Figure 4.3. (a) Cumulative AE data synchronized with strain. (b) Post-test SEM image of

modified MCO.…………………………………………………………………………..96

Figure 4.4. Long-term ASR behavior of standard vs. non reduction firing MCO coated

interconnect………………………………………………………………………………97

Figure 4.5. Post-ASR SEM observations of cross section of (a) Standard MCO (6500

hours). (b) Modified MCO (2000 hours)………………………………………………...98

xvi

Figure 5.1. Notched four-point bend test specimen to propagate a steady-state crack along

the interface……………………………………………………………………………..110

Figure 5.2. High magnification camera image of an anode-electrolyte test

specimen………………………………………………………………………………..111

Figure 5.3. SEM micrograph of porous anode and dense electrolyte…………………..112

Figure 5.4. Steps to Steps to strengthen anode by inserting glue………………………113

Figure 5.5. Steady-state crack propagation along anode-electrolyte interface…………114

Figure 5.6. SEM and EDS analysis of (a) Electrolyte (b) Anode………………………115

Figure 5.7. Experimental load-displacement curve…………………………………….116

Figure 5.8. Steady-state energy release rate as a function of applied load……………..117

1

Chapter 1

Introduction

This dissertation is divided into 5 chapters from Chapter 2 to Chapter 6 that make up two

separate topics. The first topic, focusses on adhesion of coatings on metallic interconnects

of a solid oxide fuel cell (SOFC) stack. The coats are assessed in terms of interfacial

shear strength and interfacial fracture energy. In addition, lifetime of coated interconnects

is estimated. The second topic investigates the anode-electrolyte interface of an SOFC by

characterizing the propagation of a steady-state crack along the interface.

In Chapter 1, the general background and the organization of this dissertation are

discussed.

1.1 Background

1.1.1 Basic Principles of an SOFC

Solid oxide fuel cells (SOFCs) convert chemical energy of hydrocarbon fuels to electrical

energy at high temperatures generally ranging from 600 – 800oC. Figure 1.1 presents a

schematic diagram of an SOFC with its components and basic chemical reactions. The

main constituent of an SOFC is the gas-tight, dense ceramic electrolyte which is a good

oxygen ion conductor at high temperatures. The solid ceramic electrolyte is sandwiched

between two porous fuel and air electrodes known as the anode and cathode, respectively.

2

In the basic principle, the air or oxygen at the cathode combines with an electron and

forms an oxygen ion. The oxygen ion diffuses through the electrolyte to the anode. At the

anode, an electron is liberated by oxidation of hydrocarbon fuels, thus producing water.

The emitted electron is routed through an external circuit to the cathode and completes an

electrical, power-producing circuit [1, 2]. In practical applications, unit fuel cells are

connected in series to form a fuel cell stack. Interconnects are the devise that physically

separate but electrically connect the adjacent cells.

1.1.2 SOFC Materials

NexTech Materials Ltd. (NTM) has developed a family of electrolyte-supported planar

cells that have sufficient mechanical robustness and excellent electro-chemical

performance. As shown in Fig. 1.2(a), the FlexCellTM

is formed from a thin electrolyte

membrane (40 μm) sintered with a support layer consisting of thicker electrolyte material

(150 μm) that has hexagonal cut-outs. The thin regions within each hexagon are called

“active areas” where ion transport is most active. The electrolyte is typically composed of

either yttria (Y) stabilized or scandia (Sc) stabilized zirconia (Zr). The thin electrolytes

and 30 μm screen-printed electrodes minimize the component thickness and results in

mechanically flexible components that are compliant during stack assembly and

operation. Electrochemical performances of the FlexCellTM

are comparable to the best

anode supported cell designs and enable a wider window of material selection for the

electrodes [3]. This electrode material flexibility can enable lower operating temperatures

and is more tolerant of the use of sulfur contaminated fuels.

3

Over the years NTM has found that an LSM (lanthanum strontium manganese) based

cathode and Ni-GDC/Ni-YSZ anode, where GDC refers to gadolinium doped cerium,

provide the best performance and stability. The anode material is produced by controlled

mixing of NiO, GDC and YSZ (yttria stabilized zirconia) powders with well-defined

powder size and tape-casting the powder slurry into sheets. The sheets are subsequently

stacked and co-sintered with adjoining materials. Finally exposure to an H2 environment

converts NiO to Ni [2].

In Fig. 1.2(b) an SEM image of an anode-electrolyte bi-layer is shown. As can be seen in

the figure, the anode is separated into two distinct layers: the active anode layer and the

current collector layer. The active anode layer is 10 μm in thickness and current collector

layer is 25 μm in thickness.

Within an operating temperature of 600-800oC, it is preferable to use iron based metals as

interconnects because of their low cost and availability [4, 5]. To protect the metallic

interconnects from oxidation at high temperature, the cathode side interconnect surfaces

are coated by ceramic oxides [6].

1.1.3 Applications of SOFCs

Due to their high efficiencies and low emissions, SOFCs have the potential to radically

alter the distributions and productions of electrical energy. Although SOFCs operate at

high temperatures, they have several advantages over other fuel cell types. The principal

advantage is fuel flexibility. Because the ceramic membrane is an oxygen ion conductor,

oxygen partial pressure gradients create voltage that allows cell operation. Thus both H2

and CO can be consumed as fuel, allowing a range of reformed hydrocarbon fuels to be

4

considered. The high operating temperatures create advantages both by enabling catalysis

of the fuels without special, expensive materials and by paving the way to enhanced

efficiency within combined cycle systems. Another important advantage is the enhanced

tolerance of SOFCs to fuel impurities. CO, which poisons proton exchange membrane

(PEM) fuel cells, is a fuel for SOFCs. H2, another common contaminant in hydrocarbon

and some biomass-derived hydrogen fuels, is tolerated in SOFCs currently being

developed; competing fuel cells require large desulfurizers which reduces overall system

efficiency. In addition, as the SOFCs do not require combustion to generate energy, they

produce cleaner energy with lower emissions compared to the conventional power

systems [7].

With increasing demands of global power consumptions in the next decades, SOFCs may

be a promising source of power generation in stationary, mobile and military

applications. Capability of producing cleaner energy has made SOFCs an attractive

option from an environmental point of view. Because of high operating temperatures,

SOFCs are integrated with combined heat and power systems ranging from 1 KW to

several MW. The potential applications of SOFCs include individual households, large

residential units, industries and business perimeters, and hospital backup power.

1.2 Problems Statement

Degradation and failure of SOFC components while operating and after anticipated and

un-anticipated shutdowns and subsequent start-ups are largely a consequence of high

operating temperatures. With such a range of materials within stack monolith, challenges

with spatially and temporally non-homogenous high temperature environment are

5

tremendous and a direct result of (a) widely varying of thermal coefficient of expansion

(TCE) (b) non-equilibrium and low temperature phases (especially at interfaces) and (c)

ceramic components with low toughness and low thermal conductivity. In SOFCs,

thermal gradients associated with improper thermal balance between low temperature

inlet gas streams and the exothermic oxidation reactions can result in catastrophic stack

failure. The current approach to solving problems associated with thermal gradients and

TCE mismatches is to increase the mechanical strength of various components. To some

extent, this has been done; however, strengthening of different components has exposed

weakness at interfaces. Clear evidences from NTM and literature show that during

thermal cycling coating-interconnect interfaces and electrode-electrolyte interfaces

degrade and delamination occurs [8-15].

1.3 Dissertation Structure

In Chapter 2, a room temperature four-point bend experimental technique is applied in

close collaboration with NexTech Materials (NTM) for the mechanical characterization

of coating-interconnect interfaces in an SOFC stack. The experimental set up is designed

in such a way as to placing the brittle coatings under tensile stresses. The spacing

between the resulted saturated tensile cracks on the coating surface is incorporated in a

shear lag model to quantify the interfacial shear strength. At higher strains, coating

spallation occurs. Interfacial fracture energy is calculated from an energy based fracture

mechanics utilizing the strain at onset of spallation. Thus the coating adhesion is assessed

by both the interfacial shear strength and the interfacial fracture energy. The coatings are

subjected to two different heat treatments by NTM: (a) reduction heat treatment and (b)

6

oxidation heat treatment. In this chapter, the evolution of the coating adhesion from the

reduction heat treatment to the oxidation heat treatment is described. Scanning electron

microscopy (SEM) along with the energy dispersive spectroscopy (EDS) are performed

on the cracked and spalled coating to analyze the failure mechanisms of coatings. Two

types of interconnect materials are considered: SS441 (441-HP) and Crofer 22 APU and

their effects on the coating adhesion are also discussed.

In Chapter 3, the lifetime of coating at a particular stack operating temperature is

estimated by implementing an integrated experimental-analytical methodology. Although

coating is intended to act as a barrier to oxidation of interconnects, the formation of

native oxide scales (Cr2O3) between coatings and interconnects is still inevitable during

stack operations. With operating time, the native scales grow in thickness experiencing

growth stresses and as a consequence the thicker native scales alter the adhesion of the

coatings. In addition, when cooling down the fuel cell stack from operating temperature

to room temperature, thermal coefficient mismatch between the native scales and the

coatings causes coating spallation. Therefore, effective interconnect lifetime is mainly

limited by the cooling induced coating spallation. The following steps are considered in

Chapter 3 to predict the coating lifetime at a particular stack operating temperature:

Four-point bend experiments are performed on heat treated, coated interconnects

oxidized at 900oC for 100, 600 and 1000 hours. The reason for oxidizing the

coated interconnects at higher temperature than the operating temperature (600-

800oC) is to accelerate the oxidation driven failure mechanisms of the interface.

7

The experimental results describe the evolution of interfacial shear strength and

interfacial fracture energy with oxidation time, i.e. native scale thickness. The

oxidation time is related to the native scale thickness by native scale growth

kinetics.

An analytical model for a tri-layer coating-native scale-interconnect assembly is

implemented to obtain the cooling induced residual stress distributions in the

coating and shear stress distribution at the interface as a function of native scale

thickness. From the residual stress distributions in the coating, the analytical

interfacial fracture energy as a function of native scale thickness is also assessed.

Equating the experimental fracture energy with the analytical fracture energy, the

critical native scale thickness at which cooling induced coating spallation occurs

is obtained.

The critical native scale thickness is converted to the equivalent coating lifetime

by incorporating the critical native scale thickness to native scale growth kinetics

chart at a particular operating temperature.

In Chapter 4, whether reduction heat treatment has beneficial impacts on coating

performances is evaluated. In coating process, it is the reduction heat treatment that

contributes to a significant portion of coating coasts. From an economical point of view,

it would be preferable for NTM if the reduction heat treatment is removed from the

process flow. But it is anticipated that the absence of reduction heat treatment may

degrade the coating performances by producing a less dense and poorly adherent coating.

In addition to the four-point bend tests, in Chapter 4 electrical resistance tests are

8

performed on coated interconnects. Based on the results obtained in Chapter 4,

recommendation whether the reduction heat treatment is necessary is made.

In Chapter 5, the anode-electrolyte interfaces are characterized. Four-point bend tests are

conducted in controlled environment in the laboratory. The brittle bi-layer test specimens

are sandwiched between two steel stiffeners. A notch is created in the relatively thicker

electrolyte which acts as a crack initiator at higher load. The crack initiated from the

notch tip is forced to stop at the interface and then to propagate along the interface in a

stable manner upon further loading. Critical strain energy release rate of the interface is

calculated by applying the beam bending theory. SEM and EDS analysis are performed to

study each separate layer.

Finally in Chapter 6, overall conclusions are drawn and some recommendations for future

works are made.

Experimental results and analysis obtained from this research will be helpful for NTM to

modify the manufacturing processes in order to develop the interfacial strength of

coating-interconnect interfaces and anode-electrolyte interfaces of an SOFC stack. Also

the scientific knowledge gained through the present research will be useful to other

researchers seeking to enhance understanding and control of thermo-physical stresses in

similar interface systems.

9

References

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

2. Haile S.M., Fuel Cell Material and Components. Acta Materials 51 (2003) 5981-

6000.

3. Unveils Largest SOFC Tech Platform. Fuel Cell Bulletin (2009).

4. Wincewicz K.C., Cooper J.S., Taxonomies of SOFC material and manufacturing

alternatives.

Journal of Power Sources 140 (2005) 280-296.

5. Zhu W.Z., Deevi S.C., Development of Interconnect Materials for Solid Oxide Fuel

Cells. Materials Science and Engineering A348 (2003) 227-243.

6. Bateni R. M., Wei P., Deng X., Petric A., Spinel Coatings for UNS 430 Stainless Steel

Interconnects. Surface & Coatings Technology 201 (2007) 4677-4684.

7. Blum L., Buchkremer H.P., Gross S., Gubner A., de Haart L.G.J., Nabielek H.,

Quadakkers W.J., Reisgen U., Smith M.J., Steinberger-Wilckens R., Steinbrech R.W.,

Tietz F., Vinke I.C., Solid oxide fuel cell development at Forschungszentrum Juelich.

Fuel Cells 7 (2007) 204-210.

8. Evans H.E., Cracking and Spalling of Protective Oxide Layers. Materials Science and

Engineering 120 (1989) 139-146.

9. Christl W., Rahmel A., Schutze M., Application of Acoustic Emission Technique for

the Detection of Oxide Scale Cracking During Thermal Cycling. Material Science and

Engineering 87 (1987) 289-293.

10. Li W., Hasinska K., Seabaugh M., Swartz S., Lannutti J.J., Curvature in Solid Oxide

Fuel Cells. Journal of Power Sources 138 (2005) 145-155.

11. Walter M., GOALI: Electrode Interfaces Stresses, Degradation and Failure, NSF

Proposal 2008.

12. Recknagle K.P., Williford R.E., Chick L.A., Rector D.R., Khaleel M.A., Three-

dimensional thermofluid electrochemical modeling of planar SOFC stacks. Journal of

Power Sources 113 (2003) 109-114.

13. Malzbender J., Wessel E., Steinbrech R.W., Reduction and re-oxidation of anodes for

solid oxide fuel cells. Solid State Ionics 176 (2005) 2201-2203.

14. Zhang Y., Liu B., Tu B.F., Dong Y.L., Cheng M.J., Redox cycling of Ni-YSZ anode

investigated by TPR technique. Solid State Ionics 176 (2005) 2193-2199.

15. Fouquet D., Muller A.C., Weber A., Ivers-Tiffee E., Kinetics of oxidation and

reduction of Ni/YSZ cermets. Ionics 9 (2003) 103-108.

10

Figures

Figure 1.1. A schematic diagram of an SOFC with its components and basic principles.

11

(a)

(b)

Figure 1.2. (a) FlexCellTM

. (b) An SEM micrograph of an anode-electrolyte bi-layer.

12

Chapter 2

Adhesion of Reduced and Oxidized Spinel Coatings on Metallic

Interconnects

Abstract

This chapter reports on the mechanical characterization of coating-interconnect interfaces

for coating processed in two different conditions: reduction heat treatment and reduction

heat treatment followed by an oxidation heat treatment. The interconnect materials

considered are ferritic AL 441-HP (SS441) and Crofer 22 APU to explore the effects of

the alloying elements in interconnects on the coating adhesion. An acoustic emission

(AE) sensor is coupled with a four-point bend fixture to monitor the coating cracking and

spallation. A long-distance camera lens is also incorporated in the experimental setup to

accurately identify the acoustic signals correlated with the coating and interface fracture.

Resulting tensile cracking patterns on the coatings on the convex side of the bend

specimen are used to quantify the interfacial shear strength with a shear-lag model. Using

energy based fracture mechanics; interfacial fracture energy is calculated from the strain

at the onset of coating spallation. Post-test SEM is performed to analyze the cracked and

spalled surfaces and to gain the insight of coating failure mechanisms. The experimental

13

results presented in this chapter are able to distinguish between two different coating-

substrate systems and reduction heat treatment versus reduction-oxidation heat treatment.

2.1 Introduction

2.1.1 Requirements of Interconnects

It is well known that, to reach practical power levels, individual cells are repeated to form

a fuel cell stack. Interconnects are the components that physically separate individual

cells and provide the means to complete the electrical circuit. Thus interconnects

maintain the uniform fuel and air flow as well as play a critical role in increased

efficiency and power density of a fuel cell stack. Since opposite sides of interconnects are

exposed to reducing (fuel) and oxidizing (air) environments, the resulting chemical

potential gradients place severe constraints on selecting the most appropriate materials

for interconnects. The additional design requirements that an interconnect should possess

are excellent electrical conductivity, chemical and physical inertness in high temperature

corrosive environments, good thermal conductivity, high strength and creep resistance,

and low material and fabrication cost [1].

As the main purpose of an interconnect is to maintain an electrical connection between

anode and cathode of the adjacent cells, the most important property of interconnect is to

exhibit excellent electrical conductivity, preferably with 100% electron conduction.

Electrical conductivity of 1 S-cm-1

is the minimum acceptable conductivity for use in a

fuel cell stack [1]. The next important property of interconnect is the stability of

geometry and microstructure in high temperature, corrosive environments. The anode

side of the interconnect is exposed to a high temperature, reducing environment whereas

14

the cathode side is exposed to an oxidizing atmosphere. The different oxygen partial

pressures on the anode and the cathode sides builds up significant oxygen partial pressure

gradients in the interconnects. The resulting geometrical changes can yield considerable

amount of mechanical stresses in interconnects and can cause breakage of interconnects

as well as cracking of seals [1]. Phase or microstructural changes in interconnects due to

chemical potential gradients can significantly alter the electrical conductivity. Sufficient

lack of chemical inertness of interconnects can result in formation of an electrically

insulating oxide scales in high temperature oxidizing atmospheres. Thus instability of

geometry and phase of interconnects can deteriorate the performance of an SOFC stack.

Finally, closely matched thermal coefficients of expansion (TCE) of interconnects with

electrodes is desirable to minimize the generation of thermal stresses due to start up and

shut down cycles [1].

With the use of thin electrolytes or thin electrolyte regions, the maximum operating

temperature of an SOFC can be kept below 800oC. This allows replacing conventional

but expensive LaCrO3 ceramic interconnects with metal-based interconnects. There are

many advantages of using metallic interconnects over the ceramic ones such as high

electrical and thermal conductivity, low cost, availability, and workability. However the

main difficulties associated with metallic interconnect are low resistance to oxidizing and

corrosive environments, low strength at high temperatures and thermal and mechanical

mismatch with electrode materials. In light of the necessary requirements on

interconnects, chromium alloyed iron (Fe) based (stainless-steel) interconnects such as

Crofer 22 APU or Allegheny Ludlum (AL) 441-HP are the most promising interconnect

15

materials for intermediate temperature (600 – 800oC) SOFCs. The formation of chromia

scale (Cr2O3) gives the property of fairly moderate oxidation and corrosion resistance at

high temperatures. The most attractive feature of stainless steel interconnects is the

closely matched TCE with the electrodes. High ductility, high compliance, high strength

and creep resistance at high temperature are other advantageous features when

considering chromium alloyed ferritic metals as interconnects [1].

2.1.2 Drawback of Metallic Interconnects

An important drawback of chromium alloyed metallic interconnects is chromium

poisoning of cathodes. Figure 2.1 demonstrates the sequential steps of chromium

poisoning. Depending on the atmosphere at the cathode, chromium rich alloy forms CrO3

chromium trioxides or CrO2(OH)2 chromium hydroxides. Upon combining with oxygen

ions in the cathode active area, the chromium compounds reduce to Cr2O3 chromia

scales. Thus the chromia scale formation results in decreased cathode active area. In

addition, the electrically insulating nature of chromia scales increases the contact area

specific resistance (ASR) between the interconnect and cathode. As a consequence, the

performance and efficiency of a fuel cell stack is significantly degraded [2, 3, 4].

2.1.3 Protective Coatings on Interconnects

It is therefore critical to inhibit chromium migration to cathodes and oxygen inward

diffusion to interconnects. One way to achieve the necessary chemical inertness of

interconnects is to protect the surfaces with application of a thin, dense coating. The

protective coating layer acts as a diffusion barrier to chemical reactions between the

interconnects and the corrosive environment, thus inhibiting chromium volatility in the

16

cathode active area and slowing the rates of chromia scale formation. Either bulk alloy

modification or bulk surface modification can be performed to form coatings on

interconnects. In bulk alloy modification, manganese is added in interconnects to form a

unique scale comprised of (Mn,Cr)3O4 spinel at the top with a chromia or chromia-rich

sub-layer beneath the top layer. But research conducted at Pacific Northwest National

Laboratory (PNNL) indicates that the (Mn,Cr)3O4 spinel coatings from bulk alloy

modification reduce Cr poisoning by only a factor of 3 over unmodified alloy [5]. This

may still result in an unacceptable amount of chromium poisoning in stack operation.

In bulk surface modification, a dense, protective coating on the surface of the cathode-

side interconnect is applied externally. Generally pervoskite compositions with a

chemical structure of (AB)2O3, where A and B are metallic cations are used as protective

coatings. The most commonly used pervoskite coatings are lanthanum manganese oxide

and lanthanum cobalt/chromium oxide. One potential challenge associated with this type

of coatings is to manufacture a dense protective layer. In general, insufficient density of

pervoskite coatings makes them unable to prevent inward oxygen diffusion. As a result,

pervoskite coatings possess poor thermo-mechanical stability due to extensive oxide scale

growth between the coatings and interconnects. A promising alternative to pervoskite-

type coatings is the spinel structure materials of (AB)3O4 where A and B are again

metallic cations [6].

Several researchers have recently applied manganese cobalt spinel oxide (MCO) as a

protective layer on interconnects in an SOFC stack. Larring et al. observed increased

capability of (Mn,Co)3O4 to prevent chromium evaporation compared to that of

17

pervoskite composites [6]. In addition, with iron (Fe) doping of the spinel MCO, it is

possible to achieve excellent electrical conductivity and thermal compatibility with

interconnects. Yang et al. studied thermally grown (Mn,Co)3O4 with a nominal

composition of Mn1.5 Co1.5 O4 on Crofer22 APU substrate [2, 5]. The substrate was slurry

coated and then heat treated in reducing and oxidizing environments. Their experimental

results indicate decreased ASR between LSF cathodes and interconnects due to

significant inhibition of subscale growth. The spinel coating acted as an effective barrier

to both chromium migration and oxygen inward diffusion.

The ceramic MCO coating is extremely brittle. In practice, the mechanical integrity of

this protective coating layer can be severely affected by the mechanical stresses from

mechanical loading and the residual stresses originating from oxide scale growth and

thermal cycling. The complex thermo-chemical stresses in addition to mechanical

stresses can result in coating fracture and spallation. The coating-substrate interfaces play

an important role in the mechanisms of coating fracture and spallation. For weaker

coating-substrate interfaces, coatings first buckle under high compressive stresses and

then spall if through-thickness cracks develop. For relatively stronger interfaces, shear

cracks can form in the coatings, which cause shear sliding in the cracked segments and

finally spallation in the protective coatings [7]. For SOFCs, when the protective coating

layer spalls, uncoated interconnect metal is exposed to the corrosive environments more

severely. The resulting damage to interconnects can cause significant degradation of the

electrochemical performances of the SOFCs. Therefore it is important to characterize

18

coating interfaces/measuring coating adhesion, for reliability assessments of

interconnects in SOFC applications.

2.1.4 Experimental Techniques for Measuring Coating Adhesion

Several researchers have incorporated different experimental methods to measure the

adhesion of coatings to substrate. Some of the experimental techniques include tensile

pulling test, scratch test, laser induced or shock wave induced spallation, micro-

indentation test, bend test etc. Although some of them have simple experimental setups,

each of them has their own drawbacks. For example, in a tensile pull test a stud-pull is

attached to the specimen by means of applying an adhesive. Failure can occur between

the stud-pull and the organic adhesive if the interfacial strength between them is less than

that of coating-substrate interface. Moreover, if the coating is porous, adhesive can

penetrate through the porosities of the coating to the interface and can affect the test

results by producing scattered data [8].

In a scratch test, a stylus or indenter is traversed along the film with either step wise or

continuously applied load. The critical load at which some well-defined film failure

occurs is a measure of film adhesion. Depending on the load, depth of penetration and

mechanical properties of the coating and substrate, various failure modes may exist.

Indenter wear and tip radius, loading rate and scratch speed can scatter the experimental

results significantly [8, 9].

In a laser test, a compressive wave pulse which is generated from a high energy laser

source propagates through substrate normal to the coating-substrate interface. After

reflecting back from the free surface of the coating, the compressive wave converts to a

19

tensile wave and can break the bond at the interface. Although attractive, this is a

complicated and expensive method. Extensive calibration is required to obtain reliable

results. The mechanical and thermal properties of coating and substrate also limit the use

of this technique [8, 10].

Indentation tests involve creating micro-indents on the surfaces of coatings. Indentation

creates a plastically deformed zone beneath the surface. Residual stress provides the

driving force for lateral and radial cracking. If the interfacial toughness is less than that of

the coating and substrate, a crack will propagate through interface. Together with finite

element simulations, Sun et al. performed stair stepping indentation tests to quantify the

interfacial shear strength of coating-oxide-crofer tri-layer systems in SOFCs [11]. The

critical load at which the scale spallation occurred was used to quantify the interfacial

shear strength. However, the formation of a plastically deformed zone beneath the surface

limits the depth of indentation to be less than the coating thickness [12].

In the present research, a room temperature four-point bend experimental technique is

applied in close collaboration with NexTech Materials (NTM) for the mechanical

characterization of coating-interconnect interfaces in an SOFC stack. The coatings are

subjected to two different heat treatments by NTM: (a) reduction heat treatment (b)

reduction and oxidation heat treatment. The experimental set up is designed in such a

way as to placing the brittle coatings under tensile stresses. The spacing between the

resulted saturated tensile cracks in the coating surface is incorporated in a shear lag

model to quantify the interfacial shear strength. At higher strains, coating spallation

occurs. Interfacial fracture energy is calculated from an energy based fracture mechanics

20

utilizing the strain at onset of spallation. Thus the coating adhesion is assessed by both

the interfacial shear strength and the interfacial fracture energy. The evolution of the

coating adhesion from the reduction heat treatment to the oxidation heat treatment is

described. Scanning electron microscopy (SEM) along with the energy dispersive

spectroscopy (EDS) are performed on the cracked and spalled coating to analyze the

failure mechanisms of coatings. Two types of interconnect materials are considered:

SS441 (441-HP) and Crofer 22 APU and their effects on the coating adhesion are also

discussed.

2.2 Experimental

2.2.1 Materials

Interconnects coated with manganese cobalt spinel oxide (MCO) are provided by NTM.

The ferritic interconnect substrates are: AL 441-HP (SS441) and Crofer 22 APU. SS441

is mainly composed of 17.6% chromium (Cr), 0.33% manganese (Mn), 0.47% silicon

(Si), 0.46% niobium (Nb), 0.18% titanium (Ti), 0.2% nickel (Ni) by weight with minor

amount of alloying elements such as carbon (C), aluminum (Al), phosphorus (P), sulpher

(S) and with the balance being iron (Fe) [13]. Crofer 22 APU is an Fe–Cr–Mn steel

specifically developed for SOFC interconnect applications [14, 15]. It has 22.3% Cr with

0.45% Mn, other alloying elements such as Ti, C, P, S in minor amounts and a balance of

Fe. In addition to having higher amounts of Cr and Mn, Crofer has also trace amounts of

rare earth elements, for example, 0.06% of lanthanum (La).

NTM’s coating processing steps are shown in Fig. 2.2. In the first step of coating process,

an MCO suspension is applied through a cost-effective, commercially viable aerosol

21

spray deposition (ASD) on the metallic interconnects at room temperature [16]. The

spray suspension is prepared by mixing an appropriate binder, solvent and dispersants

with MCO powder having a nominal composition of Mn1.5Co1.5O4. The thickness of the

MCO ranges from 10 - 15 μm. The MCO coated interconnect is then heat treated in two

different stages to develop a dense, adherent MCO coating: a controlled reduction heat

treatment at high temperature followed by an oxidation heat treatment at elevated

temperature. The reduction heat treatment reduces the spinel into two distinct

components: metallic Co and lower valence MnO. The organic binder is burned out

during the reduction firing. During subsequent oxidation in air, the Co and MnO react

with oxygen to reform the spinel with enhanced densification via reaction-sintering [17]:

6Co + 6MnO + 5O2 = 4 Mn1.5Co1.5O4

Figure 2.3 (a) and 2.3 (b) shows two SEM images of a reduced and a dense oxidized

MCO coating respectively.

2.2.2 Bend Experiments

The four-point bend experiment is set up such that the coatings will experience tensile

stresses during bending. The loading configuration and dimensions of the test specimen

are shown in Fig. 2.4. The inner and outer loading spans are 20 and 60 mm, respectively.

The stresses at the outer surfaces of coating-interconnect test specimen are calculated

from the applied loads by the method of Timoshenko [18]. Surface strains are obtained

from a conventional resistant strain gage attached to the uncoated side of the test

specimen and the specimens were loaded in displacement control up to 3% strain. An

acoustic emission (AE) sensor that is coupled to a Vallen System AMSY-4 is placed on

22

the coated side of the specimen to monitor coating failure. The AE sensor detects the

transient surface waves generated from cracking and other failure and deformation

phenomena of coatings that release strain energy. After the AE signal is pre-amplified by

34 dB, the signal threshold is set to 40 dB to eliminate unwanted noises from the load

frame and the test environment. To accurately identify the acoustic emission events

associated with coating spallation, the coating is observed in situ with a high

magnification camera lens. The stress-strain data and the AE data are synchronized for

subsequent analysis of failure mechanisms.

2.3 Theory

2.3.1 Interfacial Shear Strength

The tensile loading failure mechanisms of a brittle coating on a ductile substrate are

schematically illustrated in Fig. 2.5(a). When the tensile stresses exceed the critical

tensile stress of the coating, initial thru-thickness tensile cracks are produced in the

coating. The initial cracks, also called primary cracks, are created by stresses generated

from the applied bending moment. After formation of the primary cracks, tensile stresses

are no longer transferred to the coating segment directly from the applied bending

moment. Instead, as a result of continuous loading, the tensile stresses are transferred

from the ductile substrate to the coating segment through interfacial shear stresses or

what is also called “shear-lag.” As a consequence of stress transfer through shear at the

interface, tensile stresses continue to develop in the coating segment, creating further

thru-thickness tensile crack and as a consequence new coating segments. The crack from

shear-lag is known as secondary crack as this is not generated from the applied bending

23

moment directly. Secondary crack density increases as the applied strain increases until at

one point, the formation of the tensile thru-thickness cracks saturates [19, 20].

The formation of thru-thickness tensile cracks is described as elastic stress relaxation and

can be modeled by a shear-lag theory [21]. For a coating segment, Fig. 2.5(b) shows

shear stress distributions at the interface and the resulting tensile stress distributions in

the coating segment along x-direction. Although bending creates a thru-thickness stress

gradient, the coating is 0.5% of the substrate thickness and the effects of the stress

gradient thru the coating thickness are therefore negligible. At both ends (free surfaces)

of the segment where the tensile stress in the coating segment is minimum, all the stresses

are supported by the substrate resulting in maximum interfacial shear stress at both ends

of the segment. The tensile stress is highest at the middle of the coating segment and the

shear stress is lowest at the midpoint of the interface. When the maximum tensile stress

exceeds the critical tensile stress of the coating, a new cracked segment is formed. For a

coating segment of length l in Fig. 2.5(b), the segment is infinitely long in z-direction

comparing the length, l, and hence the variation of stresses in z-direction is neglected.

The equilibrium equation along x-direction can be written as [22]

Here is the normal stress distribution in the coating segment along x-direction and

is the shear stress.

Integrating Eq. (2.1) over the coating thickness, hc and then dividing by the thickness

yields

24

(

∫

)

∫

Considering the thru-thickness stress gradient in the coating segment is negligible and at

x = hc, (free surface) and after doing some mathematical manipulations, equation

(2.2) finally yields

∫

Here is the interfacial shear stress distribution. Tien and Davidson [13] considered

purely elastic behavior of the interface with a linear shear stress distribution along the

interface and for this case the maximum shear stress at the interface, can be related

to the critical tensile stress of the coating, from Eq. (2.3) by the following relation:

For an ideal plastic interface where the shear stress distribution is assumed to be constant,

the Eq. (2.4) takes the form as [24]:

In general, the maximum interfacial shear stress is related to the critical tensile stress of

coating thru coating thickness and crack spacing by:

Here K is an integration constant which depends on assumed shear stress distribution and

its value is bound between 2 and 4 as shown in Fig. 2.5(b). In the present case, a

25

sinusoidal shear stress distribution is considered along the interface to allow the interface

to have limited plastic stress relaxation next to the thru-thickness cracks. In this case, K is

π which is between an ideal elastic and plastic behavior [25].

Once the shear strength of the interface has been exceeded, the shear stresses are relaxed

by non-elastic mechanisms such as interface slip next to the tensile cracks or by substrate

yielding at the base of the thru-thickness cracks. This phase is described as plastic stress

relaxation, and the secondary crack formation saturates during this phase [25]. Interface

slip is followed by an interface delamination. By replacing l with the saturated crack

spacing, it is possible to calculate the interfacial shear strength, from Eq. (2.6).

2.3.2 Interfacial Fracture Energy

At higher strains, coating spallation is observed. If it is assumed that the coating is

perfectly adhered to interconnect and only elastically strained during the experiment,

elastic strain energy will be stored in the coating. At some point, it is energetically

favorable to release the stored elastic energy as interfacial fracture, resulting in coating

spallation. The interfacial fracture energy, can be calculated from the relation [26]

where W is the stored elastic energy density in the coating and hc is the coating thickness.

The subscript ‘exp’ refers to the experimentally determined interfacial fracture energy. W

is a function of in-plane stress-strain evolution in the coating along both the longitudinal

(x-axis) and in-plane transverse (z-axis) axes. Despite the bend loading, stress

perpendicular to the interface would be very small and are thus assumed to be neglected

[25]. During the experiment, the coating deforms elastically whereas the metal

26

interconnect undergoes elastic and plastic deformation. For a continuous, un-cracked

coating, the tensile stresses developed in the coating during elastic interconnect

deformation; can be expressed as a function of applied tensile strain on the coating by the

following equation [26]:

where E is Young’s Modulus, v is Poisson ratio, ε is strain and σ is stress. The subscript

‘c’ stands for coating and the subscript ‘s’ stands for substrate. The superscript ‘x’

corresponds to the longitudinal axis. If the interconnect undergoes plastic deformation,

the tensile stress-strain relation in the coating takes the following form [26]:

[ (

)]

During elastic deformation of the interconnect, the compressive stresses generated in the

coating along the in-plane transverse axis can be calculated by [26]

here the superscript ‘z’ corresponds to the in-plane transverse axis. During plastic

deformation of the interconnect, compressive stresses can be described by the following

relation [26]:

[ (

)]

As mentioned in Section 2.2.2, the longitudinal compressive strains of the interconnect

surface on the uncoated side of the test specimen are obtained from a strain gage. From

the four-point bend theory and for a very thin coating (hc << hs), the tensile strains in the

un-cracked coating are equal in value to the compressive strains from the strain gage and

27

the compressive strains in the coating along the in-plane transverse axis are equal to the

in-plane transverse compressive strains of the surface of coated side interconnect [27].

Considering the area under the stress-strain curve of the MCO coating, the stored elastic

energy in the coating at the onset strain of spallation (εsp) can be calculated from the

following relation [26]

∫

If the first thru-thickness transverse crack and the onset of spallation occur at nearly the

same strain, both the tensile and compressive stress evolution would need to be

considered. However, if the coating spallation occurs at a comparatively higher strain, the

tensile stresses are relaxed by the formation of the cracked segments in coating, and only

the z-axis compressive stresses in coatings are involved in computing the fracture energy

[26].

2.4 Results

2.4.1 Cumulative AE

The flexural stress–strain curve of a MCO coating–interconnect specimen together with

cumulative AE data is presented in Fig. 2.6. Two types of interconnects, SS441 and

CroferTM

, are considered. For each type of interconnect, the effects of both the reduction

and oxidation heat treatment are investigated. Both SS441 and Crofer are ferritic stainless

steels, and the young’s modulus (∼250 GPa) and yield stress (∼350 MPa) were found to

be same for both the substrates. Since the MCO coating is extremely thin comparing to

28

the interconnect, it has no effect on the stress–strain curve. The stress–strain curve shown

in Fig. 2.6 is representative of all coating–substrate systems.

The AE data provides information on cracking events. The cumulative AE curve for each

type of test specimen is plotted by adding the individual AE hits cumulatively. In Fig.

2.6(a) each type of cumulative AE curve is representative for the respective specimen

type and reveals three distinct regions of AE activity. In the figure, the regions are

labeled for oxidized MCO-Crofer only. In the initial region of each curve which is shown

in an enlarged view in Fig. 2.6(b), the cumulative AE hits are observed to increase with

the applied strain. Each individual AE hit in each curve in this region represents the

formation of thru-thickness tensile cracks in the MCO during the elastic stress relaxation.

The first AE hit is considered to take place at the critical strain for the tensile crack

formation in the MCO. Assuming elastic deformation of the MCO, the critical tensile

stress of the MCO is calculated from the critical strain and the mechanical properties of

the MCO. The mechanical properties of MCO are tabulated in Table 2.1.

For each type of specimen, at least ten experiments were conducted. The critical tensile

stresses obtained were consistent for each type of test specimen. The average value of the

critical tensile stresses of MCO for each type of test specimen is tabulated in Table 2.2.

As is observed in Fig. 2.6(b) and Table 2.2, the oxidation heat treatment decreases the

critical tensile stress of MCO significantly. The more dense structure of oxidized MCO is

favorable for thru-thickness crack propagation generated from a pre-existing defects or

voids in the coatings. In Fig. 2.6(b), the higher slopes of the cumulative AE curves for the

29

oxidized test specimen indicate that the oxidized MCO has higher crack density than that

of the reduced MCO.

After the elastic stress relaxation, the saturation of tensile cracks during the plastic stress

relaxation is illustrated by a relatively flat Region 2 in the cumulative AE curves as

shown in Fig. 2.6(a). Presumably because of the 40 dB threshold setting, the plastic stress

relaxation was not detected by the AE sensor. Due to the statistically variable nature of

the interface strength, there may still be some thru-thickness crack formation in Region 2.

In other words, one section of the interface may still have sufficient strength to transfer

tensile stresses to the MCO whereas the shear stresses of the interface in other sections

are no longer sufficient to cause damage in the MCO.

Finally in the third region, a sharp increase in the slope of the cumulative AE curves is

observed at higher strains for each type of test specimen except the reduced MCO-Crofer.

The strain at which the slope begins to increase sharply is the onset strain of MCO

spallation. The onset strain of spallation was also identified with in-situ coating

observations by imaging with a long-distance camera lens. Post-experiment SEM

observations found no spallation for the reduced coated Crofer interconnects, which is

consistent with the lack of AE events in Region 3 for this specimen.

2.4.2 Interfacial Shear Strength

SEM images of the coating were taken after the bend experiments. Figure 2.7 shows

representative images of saturated in-plane transverse cracks in reduced and oxidized

MCO coating on SS441 and Crofer interconnect. The tensile stresses were applied

perpendicular to the in-plane transverse cracks. Crack spacing was measured with Matlab

30

image processing tools. Coating thickness was measured from back scattered SEM

images of the specimen cross-sections. After incorporating the values of saturated crack

spacing and thickness into Eq. (2.6), interfacial shear strength was calculated for each

type of test specimen. The calculated values of interfacial shear strength are plotted in

Fig. 2.8. As a result of interface defects as well as variability in bulk microstructure,

interfacial shear strength is going to be statistical in nature. There are a number of

stochastic variables in Eq. (2.6) that would provide statistical variability. For example,

the coating strength varies from point to point, the thickness is not constant, and crack

spacing varies throughout. Since crack spacing incorporates other physical phenomena, it

was decided to plot the interfacial shear strength of each specimen by measuring

saturated crack spacing in different locations in each specimen. The mean shear

strength was calculated from at least 50 measurements in each specimen type and is

provided in Table 2.2. It is observed in Fig. 2.8 that the interfacial shear strength has

higher statistical variability for reduced test specimen. Comparing Fig. 2.7(a) and Fig.

2.7(b), it is found that the oxidized MCO has more continuous in-plane transverse cracks

with more uniform crack spacing than the reduced MCO. Furthermore, in Fig. 2.8, the

interface is clearly weaker after oxidation heat treatment for both types of interconnects.

2.4.3 Interfacial Fracture Energy

The interfacial fracture is preceded by buckling of coating due to Poisson induced

compressive stresses developed in the coating [28, 29]. The bucking is assumed to initiate

from a pre-existing separation or crack at the interface. The pre-existing separation may

originate from the interface slip induced delamination during the plastic stress relaxation.

31

As shown in Fig. 2.9(a), the buckling of coating results in tensile stress generation near

the crack tip at the perimeter of the buckled area. The tensile stress across the interface

induces crack propagation along the interface. With increase of the applied strain, thru-

thickness cracks may develop at the location of the buckled coating where the

combination of stresses and defects in the coating exceed the fracture toughness of the

coating and finally spallation occurs. Figure 2.9(b) shows an example of the early coating

buckling before the spallation with thru-thickness cracks developed in the coating. Thus

the spallation indicates interface fracture. The interfacial fracture energy for each type of

test specimen was calculated from the onset strain of spallation obtained from the

acoustic emission data and the in situ observations with high magnification camera lens.

Table 2.2 provides the computed values of interfacial fracture energy for each specimen

type.

The SEM images of the cracked and spalled surfaces obtained at the end of the

experiment (3% strain) were processed in Matlab image processing tools to measure the

percentage spallation area (%SA). In Fig. 2.10, the SEM images and their corresponding

processed images are presented for each type of test specimen. In the processed images,

the white portions denote the spalled areas whereas the black portions indicate the un-

spalled coatings. The values of %SA are also tabulated in Table 2.2. As illustrated in Fig.

2.10, reduced MCO-Crofer specimens have almost no spallation. The little-to-no

spallation in reduced MCO-Crofer is also consistent with the AE results in Fig. 2.6(a)

where there is no increase of slope in Region 3 of reduced MCO-Crofer. The tensile yield

stress of the Crofer substrate is approximately 350 MPa [31]. The shear yield stress is

32

therefore 200 MPa and 175 MPa from the Von Mises criteria and the Tresca criteria,

respectively. In Table 2.2, the average interfacial shear strength for reduced MCO-Crofer

has been found to be 230 MPa. Since the interfacial shear strength exceeds the shear yield

stress for the reduced MCO-Crofer specimen, it is likely that yielding of the substrate

relaxes the interfacial shear stresses during the phase of plastic stress relaxation. The

absence of interface slip induced delamination inhibits the buckling of the coating. This

results in almost no spallation of the reduced MCO on Crofer within 3% strain. Since the

onset strain of spallation in the reduced MCO-Crofer specimens was not measureable

during the experiment, the interfacial fracture energy was not calculated for this specimen

type. Although it has higher interfacial fracture energy than the oxidized and reduced

MCO-SS44, the oxidized MCO-Crofer has the maximum %SA at 3% strain. The sizes of

the spalled sections are also comparatively larger. Therefore, the fracture energy

distribution of the oxidized MCO-Crofer interface is such uniform that, after attaining the

critical energy for fracture at approximately 2% strain, extensive coating spallation takes

place with a small increase of the applied strain. Oxidized coating has the lowest

adhesion with Crofer interconnect at 3% strain.

2.5 Discussions

2.5.1 Effects of Reduction and Oxidation Heat Treatment

As is listed in Table 2.2, the oxidation heat treatment decreases the tensile strength of

MCO and both the interfacial shear strength and the interfacial fracture energy. The

oxidation heat treatment yields a denser but more brittle MCO decreasing the critical

tensile stress of the coating. Energy dispersive spectroscopy (EDS) was performed on the

33

spalled areas. The EDS results are presented in Fig. 2.11. In case of the reduction heat

treatment, Fe element was detected in spalled sections as shown in Fig 2.11(a).

Therefore, the interface fracture occurs between the interconnect and the MCO in case of

reduced specimen. It is the metallic bond between the cobalt in reduced MCO and the

interconnect that strengthens the interfaces in case of reduced specimen.

Conversely, in case of oxidized specimens porous, chromium oxide scale is found in the

spalled sections as shown in Fig. 2.11(b). During the oxidation heat treatment, selective

oxidation of the alloying elements in the substrate forms a very thin layer (3 - 5 um) of

native oxide scales of interconnects between MCO coatings and interconnects. The dual

phase native scales are mainly composed of (Mn,Cr)3O4 on top of a chromia (Cr2O3) rich

sub-layer at the bottom. The top layer is relatively thin ( 1μm) and for this reason, the

term ‘native scales’ in the present dissertation refers to Cr2O3 only. As indicating by Fig.

2.11(b), for the oxidized specimens the fracture is observed to occur along the brittle

native scale-MCO interface instead of the native scale-interconnect interface. The

interfacial shear strength and the interfacial fracture energy in Table 2.2 for oxidized

specimen refer to the interface between the native scale and the MCO. And it is the

formation of the native scales by oxidation heat treatment that degrades the oxidized

MCO adhesion significantly. The native scale-MCO interface is a much weaker interface

than the ductile-brittle interconnect-native scale interface. In their research using nano-

indentation on a coating-oxide-Crofer tri-layer system, Liu et al. also report that

interfacial fracture occurs along the native scale-coating interface [32].

34

2.5.2 Effects of Interconnect Compositions

When only the reduction heat treatment is considered, CroferTM

has a better bonding with

MCO than SS441. The rare earth (RE) element, lanthanum (La) present in the Crofer

substrate improves the adhesion between the MCO and the interconnects [2].

Conversely, after oxidation heat treatments, the tensile strength of the coating and the

shear strength of the interface become lower for Crofer interconnects than for SS441. It is

interesting to observe in Table 2.2 that although the interfacial shear strength is lower for

oxidized MCO-Crofer than that of oxidized MCO-SS441, the interfacial fracture energy

is higher for oxidized MCO-Crofer. Therefore, it can be concluded that the interfacial

shear strength does not correlate directly with the coating spallation; rather shear strength

indicates the capability of stress transfer from the substrate to the coating. Shear stresses

exceeding the shear strength of the interface do not necessarily cause instant coating

spallation. This can be seen in Fig. 2.5(a) where the Region 2 in a cumulative AE curve

separates the crack saturation due to shear strength in Region 1 from spallation due to

interface fracture in Region 3. At 3% strain, the significant amount of spallation for

oxidized MCO-Crofer specimen reveals that the adhesion quality between oxidized MCO

and Crofer interconnects degrades at high strain.

2.6 Conclusions

Four-point bend experiments are performed to characterize the interfaces between MCO

coatings and SS441 or Crofer interconnects. The shear strength and the fracture energy at

the interfaces are determined. Effects of both the reduction and oxidation heat treatments

are investigated. Interfacial shear strength calculated by a shear-lag model, measures the

35

capability of the interfaces to transfer stresses from interconnects to the coatings.

Interfacial fracture energy is more related to the assessment of coating adhesion.

Oxidation heat treatment is a necessary step to perform in coating process in order to

inhibit chromium poisoning by producing denser protective coatings on interconnects.

However, the denser structure of the oxidized MCO coating decreases the tensile strength

by increasing the brittleness of the MCO. In addition, the chromium rich native oxide

scales formed by the oxidation degrade the interfaces and lower the adhesion of MCO

coatings. The weakest interface is found to be between the MCO coatings and the native

scales rather than the interface between the interconnects and the native oxide scales.

Furthermore, in case of oxidation heat treatment, the interfacial fracture energy is higher

for Crofer interconnects than that of SS441 but at 3% strain Crofer has the maximum

amount of oxidized coating spallation.

36

References

1. Zhu W.Z., Deevi S.C., Development of Interconnect Materials for Solid Oxide Fuel

Cells. Materials Science and Engineering A348 (2003) 227-243.

2. Yang Z., Xia G., Simner S.P., Stevenson, J.W., Thermal Growth and Performance of

Manganese Cobaltite Spinel Protection Layers on Ferritici Stainless Steel SOFC

Interconnects. Journal of Electrochemical Society 152 (2005) 1896-1901.

3. Bateni R. M., Wei P., Deng X., Petric A., Spinel Coatings for UNS 430 Stainless Steel

Interconnects. Surface & Coatings Technology 201 (2007) 4677-4684.

4. Puranen J., Lagerbom J., Hyva¨rinen L., Kylma¨lahti M., Himanen O., Mikko K. J.,

Vuoristo P., The Structure and Properties of Plasma Sprayed Iron Oxide Doped

Manganese Cobalt Oxide Spinel Coatings for SOFC Metallic Interconnectors.

Journal of Thermal Spray Technology (2010).

5. Yang Z, Xia G., Li X., Stevenson J.W., (Mn,Co)3O4 Spinel Coatings on Ferritic

Stainless Steels for SOFC Interconnect Applications. International Journal of

Hydrogen Energy 32 (2007) 3548-3554.

6. Larring Y., Norby T., Spinel and Perovskite Functional Layers Between Plansee

Metallic Interconnect (Cr-5 wt % Fe-1 wt % Y2O3) and Ceramic(La 0.85Sr0.15) 0.91

MnO3 Cathode Materials for Solid Oxide Fuel Cells. Journal of Electrochemical

Society 147 (2000) 3251-3256.

7. Strawbridge A., Evans H. E., Mechanical Failure of Thin Brittle Coatings.

Engineering Failure Analysis 2 (1995) 85-103.

8. Hou P.Y., Atkinson A. Methods of Measuring Adhesion for Thermally Grown Oxide

Scales. Materials at High Temperatures 12 (1994) 119 199-209.

9. Xie Y., Hawthorne H.M., A Model for Compressive Coating Stresses in the Scratch

Adhesion Test. Surface and Coating Technology 141 (2001) 15-25.

10. Youtsos A.G., Kiriakopoulos M. Tinke T. Experimental and Theoretical/Numerical

Investigation of Thin Film Bonding Strength, Theoretical and Applied Fracture

Mechanics 31 (1999) 47-59.

11. Sun X., Liu W.N., Stephens E., Khaleel M.A. Determination of Interfacial Adhesion

Strength Between Oxide Scale and Substrate for Metallic Interconnects. Journal of

Power Sources, 176 (2008) 167-174.

12. Akanda S.R., Walter M.E., Kidner N.J., Seabaugh M.M. Mechanical

Characterization of Oxide Coating-Interconnect Interfaces for Solid Oxide Fuel

Cells. Journal of Power Sources, 210 (2012) 254-262.

13. Yang Z., Xia G., Nie Z., Templeton J., Stevenson J.W., Ce-Modified (Mn,Co)3O4

Spinel Coatings on Ferritic Stainless Steels for SOFC Interconnect Applications.

Electrochemical and Solid-State Letters 11 (2008) B140-B143.

14. Simner S.P., Anderson M.D., Xia G.G., Yang Z., Pederson L.R., Stevenson J.W.,

SOFC Performacne with Fe-Cr-Mn Alloy. Journal of the Electrochemical Society 152

(2005) A740-A745.

15. Yang Z., Singh P., Stevenson J.W., Xia G. Investigation of Modified Ni-Cr-Mn Base

Alloys for SOFC Interconnect Applications. Journal of The Electrochemical Society

153 (2006) A1873-A1879.

37

16. Kidner N., Seabaugh M., Ibanez S., Chenault K., Day M., Thrun L., Lifetime

Predictions of Oxide Protective Coatings for Solid Oxide Fuel Cell Interconnects.

Nextech Materials Ltd 2011.

17. Yang Z., Xia G., Simner S. P., Stevenson, J. W., Thermal Growth and Performance

of Manganese Cobaltite Spinel Protection Layers on Ferritic Stainless Steel SOFC

Interconnects. Journal of the Electrochemical Society 152 (2005) A1896-A1901.

18. Timoshenko S., Strength of Materials, part 1, D. Van Nostran Company, NewYork,

1930.

19. Yanaka M., TsukaharaY., Nakaso N., Takeda N. Cracking Phenomena of Brittle

Films in Nanostructure Composites Analysed by a Modified Shear Lag Model with

Residual Strain. Journal of Materials Science 33 (1998) 2111-2119.

20. Wojciechowski P. H., Mendola M.S. Physics of thin films, Vol 16, academic press,

San diego, 1992.

21. Nagl M. M., Saunders S. R. J., Evans W.T., Hall D.J., The Tensile Failure of Nickel

Oxide Scales at Ambient and At Growth Temperatures. Corrosion Science 35 (1993)

965-977.

22. Hsueh C.H., Yanaka M., Mutiple Film Cracking in Film/Substrate Systems with

Residual Stresses and Unidirectional Loading. Journal of Material Science 38 (2003)

1809-1817.

23. Tien, J.K., Davidson, J.M. Stress Effects and the Oxidation of Metals. J.V. Cathcard

ed. AIME, 1974.

24. Strawbridge, A., and Evans, H. E., Mechanical Failure of Thin Brittle Coatings.

Engineering Failure Analysis 2 (1995) 85-103.

25. Nagl, M. M., Evans, W. T., Hall, D. J., Saunders, S. R. J., An in Situ Investigation of

the Tensile Failure of Oxide Scales. Oxidation of Metals 42 (1994) 431-449.

26. Chandra-Ambhorn S., Dherbey R.F., Toscan F., WoutersY., Galerie A., Dupeux M.

Determination of Mechanical Adhesion Energy of Thermal Oxide Scales on AISI

430Ti Alloy Using Tensile Test. Materials Science and Technology, 23 (2007) 497-

501.

27. Schutze M., Mechanical Properties of Oxide Scales. Oxidation of Metals 44 (1995)

29-61.

28. Evans A.G., Hutchinson J.W., On the Mechanics of Delamination and Spalling in

Compressed Films. Int. J. Solids Structures 20 (1984) 455-466.

29. Evans H.E., Cracking and Spalling of Protective Oxide Layers Materials Science and

Engineering A120 (1989) 139-146.

30. Crofer 22 APU Material Data Sheet, No. 4046, June 2008 Edition.

31. Liu W.N., Sun X., Stephens E., Khaleel M.A., Life Prediction of Coated and

Uncoated Metallic Interconnects for Solid Oxide Fuel Cell Applications. Journal of

Power Sources 189 (2009) 1044-1050.

38

Tables

Table 2.1. Mechanical and thermal properties of MCO coating, native scale and

interconnect [20].

Table 2.2. Average values of critical tensile stress ( ), interfacial shear strength (

),

experimental interfacial fracture energy ( ), and % spall area for each type of test

specimen.

Component Elasticity, E

(GPa)

Poisson ratio, ν Thermal

coefficient of

expansion, α

(C-1

)

MCO 124.7 0.36 9 -6

Native Scale 260 0.27 6.5 -6

Interconnect 250 0.3 12 -6

Interconnect Average

σcct

(MPa)

Average

τsti

(MPa)

Average

Gexpi

(J.m-2

)

% Spall Area

at 3% strain

Reduced Oxidized Reduced Oxidized Reduced Oxidized Reduced Oxidized

SS441 131.76 3.70 130.32 3.70 4.94 2.76 3.23 9.43

CroferTM

225.50 1.38 229.95 2.0 _ 10.02 No Spall 18.80

39

Figures

Figure 2.1. Steps of chromium poisoning of a cathode.

40

Figure 2.2. Steps of processing MCO coatings on interconnects.

41

(a)

(b)

Figure 2.3. SEM images of (a) Reduced MCO on Crofer (tested). (b) Oxidized MCO on

Crofer (tested).

42

Figure 2.4. Experimental setup to characterize coating-interconnect interfaces.

43

(a)

(b)

Figure 2.5. (a) Schematic of failure mechanisms of a brittle coating during bend

experiments. (b) Shear stress distributions at interface and tensile stress distributions in a

coating segment.

44

(a)

(b)

Figure 2.6. (a) Experimental stress-strain curve synchronized with AE data. (b) An

enlarged view of Region 1 of cumulative AE data.

45

(a)

(b)

Figure 2.7. Saturate parallel in-plane transverse cracks in (a) Reduced MCO on SS441.

(b) Oxidized MCO on Crofer.

46

Figure 2.8. Calculated values of interfacial shear strength for each type of test specimen.

47

(a)

(b)

Figure 2.9. (a) Interfacial fracture due to buckling of coating. (b) SEM image of MCO

buckling before spallation (reduced MCO-SS441).

48

(a)

(b)

(c)

(d)

Figure 2.10. Representative SEM and processed images of MCO coating surfaces at 3%

strain (a) Reduced MCO-SS441. (b) Oxidized MCO-SS441. (c) Reduced MCO-Crofer.

(d) Oxidized MCO-Crofer.

49

(a)

(b)

Figure 2.11. SEM and EDS analysis of spalled sections of (a) Reduced MCO on SS441.

(b) Oxidized MCO on Crofer.

50

Chapter 3

Lifetime of MCO Coatings on Metallic Interconnects

Abstract

Although MCO coatings are intended to act as a barrier to oxidation of interconnects;

formation of native oxide scales of interconnects (Cr2O3) is still inevitable during fuel

cell stack operations. With native scale growth during the operations, the strength of the

MCO-native scale interfaces will degrade. Besides, because of the temperature

coefficient of expansion (TCE) mismatch between the native scales and the MCO, for

thicker native scales, MCO spallation will most likely to occur when cooling down the

fuel cell from an operating temperature to room temperature. Therefore, it is the

spallation tendency of MCO that plays a critical role in lowering the service lifetime of

interconnects from an expected lifetime of 40,000 hours. In this chapter, the effects of the

native scale thickness on the coating adhesion are explored. Previously described room

temperature, four-point bend experiments are performed on coating-interconnect test

specimens oxidized at 900oC for 100 hours, 600 hours, and 1000 hours. Utilizing the

native scale growth kinetics at 900oC, the experimental results characterize the

degradation of the MCO-native scale interfaces as a function of native scale thickness.

Comparing the evolving interface properties with the cooling-induced interfacial fracture

51

energy obtained from an analytical model, the lifetime of MCO is estimated. The

estimated lifetime of MCO at 750oC operating temperature is approximately 34,700

hours.

3.1 Introduction

In Chapter 2, it is concluded that, the native scales (Cr2O3) of interconnects are formed by

oxidation heat treatment and the formation of native scales decrease the coating adhesion

significantly. Therefore, it is anticipated that, although interconnects are protected by

MCO coatings, the native scales will grow in thickness during the stack operations. As a

result of experiencing growth stresses, the native scales may degrade the interfaces

between the native scales and the coatings. Because of the degraded interfaces, TCE

mismatch between the native scales and MCO coatings can cause early coating spallation

when shut-down the fuel cell stack from high operating temperature to room temperature.

As a consequence, in addition to losing the protective properties of the coatings, it is also

expected that the damaged interfaces will result in lower electrical conductivity, thus

decreasing the power density of the SOFC stack.

An important source of growth stresses is epitaxial constraint. Differences between lattice

parameters of the oxide and substrate cause stresses to become maximum in oxide-metal

phase boundaries. The stresses fall off toward the oxide surface. Borie et al. employed X-

ray techniques to reveal that thin oxide films on copper are strained because of the

epitaxial relationship between the oxides and underlying material [1]. Epitaxial stresses

are only important for thin oxide scales as they are inversely related to the oxide scale

thickness.

52

Appleby et al. studied the effect of microstructural composition of oxide scales on growth

stresses. Their study revealed that the transition of an initially formed scale on the surface

of (Cr,Fe)2O3 to a scale with increasing Cr and decreasing Fe content caused tensile

stresses to develop. A decrease in atomic volume associated with the transition is the

apparent explanation for the tensile stress development [2].

The formation of fresh oxides inside the scales themselves can be an important source of

compressive stresses in oxide scales. Jaenicke et al. found that in the oxidation of copper,

micro-cracking induced by the growth stresses provide pathways for gas migration [3].

The availability of copper molecules results in the formation of fresh oxides in the scale.

Since the new oxides have higher volume than the cracked volume, significant

compressive stresses are developed and there is a further development of compressive

stresses.

Mismatch of the thermal coefficients of expansion (TCE) between oxide coatings and

substrates is arguably the most significant reason for residual stress generation in the

oxide coatings during cooling or heating [4]. Thermal stress is a function of TCEs and

temperature differences. In most cases, oxides have lower TCE values than that of metal

substrates. In that case, cooling to room temperature from elevated temperature generates

compressive stresses in the oxide coatings. Conversely, heating to high temperature

develops tensile stresses in the oxide coatings. Thus both the tensile and compressive

stresses are developed in the oxide coatings due to frequent thermal cycling during

services. Moreover with temperature changes, phase transformation in both oxide

coatings and substrates can result in a high stress development.

53

Different researchers have studied thermal cycling induced stresses in oxide coatings. For

example, Christl et al. incorporated acoustic emission techniques to detect thermal

cycling-induced oxide scale cracking on low alloy steel [5]. In a different work, Zhang et

al. monitored the scale cracking and spalling on Ni-30Cr alloys oxidized at 1000oC and

then cooled to room temperature either by furnace cooling or constant rate cooling [6].

In the present chapter, an integrated experimental-analytical methodology is implemented

to estimate the effective lifetime of MCO coatings. The key steps of the methodology are

shown in Fig. 3.1. The experimental part consists of conducting room temperature four-

point bend tests on specimens having native scales of various thicknesses. The

experimental results describe the evolution of the fracture energy of the MCO-native

scale interfaces with the growth of the native scale thickness. In the analytical part, the

residual stress distributions developed in MCO of an MCO-native scale-interconnect tri-

layer assembly due to cooling from an high operating temperature to room temperature

are obtained. From the residual stress distributions in MCO, the analytical fracture energy

of MCO-native scale interfaces as a function of native scale thickness is assessed.

Comparing the experimental fracture energy with the analytical fracture energy, the

critical native scale thickness at which the cooling induced MCO spallation occurs is

obtained. Therefore, the MCO lifetime at a particular operating temperature is equivalent

to the time the native scales require to grow at critical thickness at that operating

temperature. Thus the MCO lifetime is estimated by incorporating the critical native scale

thickness to the native scale growth kinetics chart at a particular operating temperature.

54

3.2 Materials

After performing the heat treatment on MCO coated interconnects as described in

Chapter 2, the coated specimens were run under isothermal oxidations at 900oC for 100

hours, 600 hours, and 1000 hours. Oxidation at higher temperature than the fuel cell

operating temperature generally ranging from 600-800oC was performed in order to

accelerate the oxidation driven failure mechanisms of the interfaces. Figure 3.2 presents

the evolution of the microstructure of MCO coatings with increasing oxidation time. The

microstructure of MCO oxidized for 600 hours are coarser than that oxidized for 100

hours but the microstructural differences are not as prominent as between 600-hour and

1000-hour oxidized MCO. The native scale thickness in each type of test specimen was

obtained from the native scale growth kinetics at 900oC. Figure 3.3 shows the native scale

growth kinetics of an MCO coated specimen at 900oC. To obtain the native scale growth

kinetics, NTM performs isothermal cyclic oxidation on coated interconnects up to 1000

hours at 900oC. Samples were removed periodically from the furnace and the weight gain

was measured. The weight gain was converted to the native scale thickness by using the

density of the native scales. The growth of the native scales follows a parabolic law,

, where is the native scale thickness, is the time in hours, is the parabolic

rate constant [7]. The native scale growth rate, at 900oC is 0.0189 μm

2/hr. From the

native scale growth chart in Fig. 3.3, the native scale thickness after 100, 600 and 1000

hours oxidation at 900oC are calculated as 1.95 μm, 4.77 μm and 6.14 μm respectively.

55

3.3 Analytical Model for Residual Stress Distributions

3.3.1 Background

The analytical model considered in the present analysis to obtain the cooling induced

residual stresses or thermal stresses in a tri-layer assembly is based on a ‘strength-of-

materials’ or ‘structural’ approach. The main advantage of the ‘strength-of-materials’

approach is the ability to obtain a closed form solutions with sufficient accuracy. The

easy-to-use practical formula can clearly differentiate the roles of different geometrical

and mechanical properties on resulting thermal stresses. Simple numerical methods such

as finite difference scheme can also be applied instead of an analytical technique in order

to solve the governing differential equations.

The ‘strength-of-materials’ approach was first proposed by S. Timoshenko. In his

pioneering paper at 1925, Timoshenko applied the ‘strength-of-materials’ method on a bi-

material thermostat to estimate the temperature difference induced residual stresses in

each layer of the thermostat [8]. Although he was able to predict the existence of shear

stresses at the interface developed near the edge due to the so called ‘edge effect’, he was

not able to mathematically calculate the shear stresses. Based on S. Timoshenko’s

method, later E. Suhir invented an innovative approach to this problem. Considering the

displacement compatibility at the interface, Suhir’s method is able to calculate the

stresses at the interface and in each layer with sufficient accuracy comparable to the

experimental and FEM results [9-15]. In general, the basic assumptions used in the

present model to obtain the residual stress distributions are as following [13, 16-19]:

56

1. The aspect ratio (thickness/length) of the layered structure is very small. Each

layer is considered as a thin elongated beam experiencing small deformations.

2. The classical Euler-Bernoulli beam theory is applicable in each layer such that the

vertical displacement is governed by the bending moment only.

3. The interfaces between components are not perfect rather they are weak

interfaces. The weak interfaces are characterized by interface compliances that

allow interfaces to slip. Normal stresses across the interfaces are considered to be

negligible.

4. The radius of curvature is constant through the thickness of the layered structure

but varies along the lateral direction.

3.3.2 Interface Displacements

In Fig. 3.4, the free body diagrams of each layer of a coating-native scale-substrate tri-

layer specimen are shown with the normal forces, shear forces and bending moments

generated due to experiencing temperature differences. In the figure, N denotes the

normal force per width, V denotes the shear force per width and M denotes the bending

moment per width. The temperature difference induced residual shear stress at the MCO-

native scale interface (i1) is denoted by and the native scale-interconnect interface

(i2) by .The displacement function in x-direction at i1 of coating can be written as

[13]:

∫

57

where is the thermal coefficient of expansion (TCE) of the coating, is the

temperature change, and

is the axial compliance of the coating,

is the interface compliance and w(x) is the y-direction displacement of each layer and is

constant thru y-direction but varies along x-direction. The first term in the right hand side

of the equation represents the unrestricted thermal displacement. The displacement in the

second term originates from the thermal induced normal force at the cross section and

this displacement is assumed to be same at all points of the cross section. Correction of

the second term for the interface is done by the third term where the interface

displacement is assumed to be higher than the inner points of the cross section and is

proportional to the interfacial shear stresses. Finally in the fourth term, the interface

displacement due to moment induced curvature is presented.

Similarly, the x-direction displacement function of i1 of native scale and i2 of native

scale and interconnect can be written respectively as,

∫

∫

∫

58

3.3.3 Equilibrium Relations and Governing Equations

To obtain equilibrium relations, differential elements are considered from each layer and

are shown in Fig. 3.5 [19]. The sum of the x-direction forces of each layer yields the

following relations:

Similarly from the moment equilibrium of each layer, the following relations are found:

(

)

… … … (3.2)

The total bending moment applied at the tri-layer assembly is the summation of the

bending moment at each layer,

From Eq. (3.3) and Eq. (3.2), it is found that

∫

Where D is the flexural rigidity of the composite beam and D = Dc + Dn+ Ds, where Dc

can be written as

and so on. From the equilibrium of forces in y

direction , the Eq. (3.4) reduces to

Considering the displacement compatibility at the i1 and i2,

and

utilizing Eq. (3.1), (3.2) and (3.5), the following differential equations are found [7]:

59

where, and

Finallly separating Nc and Ns from Eq. (3.6) and (3.7), the governing differential

equations are obtained as [13]:

where

√

, √

,

3.3.4 Boundary Conditions

Considering the condition of a free surface at the edge of each layer (x = -L, L), the

following boundary conditions (BC) of the governing Eq. (3.8) and (3.9) can be written:

60

… … … (3.10)

For the second boundary conditions, the beam bending theory is used such that:

From Eq. (3.11) it is found that

and finally

Now Eq. (3.12) yields,

Eq. (3.13) implies that

Incorporation of Eq. (3.14) in to Eq. (3.2) with the condition of free surface at each layer

where Vc(BC) = Vn(BC) = Vs(BC) = 0, implies that,

,

… … … (3.15)

Finally Eq. (3.15) with Eq. (3.1) gives the required second boundary conditions of the

governing differential equations (Eq. (3.8) and (3.9)) as

61

3.3.5 Analytical Interfacial Fracture Energy

Figure 3.6 shows the steps to obtain the interfacial fracture energy from the analytical

model. The inputs of the model are mechanical, thermal and geometrical properties of

each layer and the temperature differences that the tri-layer assembly experiences when

cooling from high temperature to room temperature. The governing differential equations

(Eq. (3.8) and (3.9)) can be solved with the boundary conditions in Eq. (3.10) and (3.16).

From the resulting solutions of Nc(x) and Ns(x), the interfacial shear stresses and

and the normal force per width at the native scale, Nn(x) are calculated from Eq. (3.1).

The normal stresses in each layer are then calculated from the respective N(x) by dividing

with the thickness of each layer. The curvature of the tri-layer assembly generated due to

temperature change is calculated from Eq. (3.5). From the curvature, the bending stresses

in each layer can also be assessed. Finally, adding the normal stresses with the bending

stresses, the resultant cooling induced residual stress distributions in each layer is

obtained.

In general, the coatings experience compressive residual stresses due to cooling from an

elevated temperature to room temperature. Compressive residual stresses can cause

buckling of the coatings from a pre-existing crack or delamination. After buckling, the

compressive stresses in the buckled region are reduced but there is a stress concentration

at the perimeter of the buckled region and associated stress intensity resulting in interface

fracture. Interface fracture is followed by coating spallation. The interfacial crack under

the buckle region propagates if the analytical fracture energy at the MCO-native scale

interface satisfies the following criteria:

62

Here is the experimental interfacial fracture energy. Now

can be calculated from

the following relation of fracture mechanics by

(

)

The interfacial fracture toughness can be calculated from the average residual

compressive stress in the coating, [20, 21]:

And the so-called ‘interfacial elasticity’ Ei1

can be calculated by [22],

(

)

3.4 Results and Discussions

3.4.1 Experimental Results

Figure 3.7 presents a representative plot of cumulative AE hits for each specimen type

synchronized with strain. The first AE hit represents the critical tensile strain of MCO

coatings. The critical tensile stress and the saturated parallel tensile crack spacing of

MCO are incorporated in the shear-lag model to calculate the interfacial shear strength. A

representative SEM image of saturated parallel tensile cracks on 900oC-1000 hour

oxidized MCO is shown in Fig. 3.8. The average values of critical tensile stress of MCO,

saturated crack spacing and interfacial shear strength for each type of test specimen are

tabulated in Table 3.1. The trend of critical tensile stress reduction with oxidation time is

consistent with the evolution of the microstructure of the MCO presented in Fig. 3.1. In

63

addition, Table 3.1 clearly indicates the increase of saturated crack spacing and the

resulting degradation of the interfacial shear strength with increasing oxidation time.

The onset strain of spallation obtained from the acoustic emission data shown in Fig. 3.7

is utilized to calculate the experimental interfacial fracture energy for each type of test

specimen by the method described in Chapter 2. Figure 3.9 shows an SEM image of

spalled MCO with the point EDS analysis of coating and spalled sections. The absence of

cobalt (Co) element and the presence of chromium (Cr) in the spalled section clearly

indicate that the interface fracture occurs between the MCO and the native scales. The

average fracture energy of the MCO-native scale interfaces obtained from the

experiments ( for each type of test specimen is tabulated in Table 3.1. Similar to the

case of interfacial shear strength, the interfacial fracture energy also decreases with the

longer oxidation time or the increase of the native scale thickness.

The SEM images of the post-test tensile surfaces were processed with Matlab image

processing tools to measure the percent spalled area (%SA). In Figure 3.10, the SEM

images and their corresponding processed images are presented for each type of test

specimen. In the processed images, the white portions are the spalled areas whereas the

black portions are the un-spalled MCO. The values of %SA are also tabulated in Table

3.1. Table 3.1 shows increase of %SA with increasing oxidation time. In other words, as

the native scales grow thicker with oxidation time, the MCO coatings lose adhesion and

become more susceptible to spall.

64

3.4.2 Analytical Results

The lifetime of MCO coatings is calculated considering that the fuel cell operating

temperature is 750oC. The mechanical and thermal properties from Table 2.1 are

incorporated in the analytical model to assess the interfacial fracture energy developed

from cooling induced residual stresses in MCO. The thickness of interconnect and MCO

is considered to be 200 μm and 10 μm respectively. The thickness of the native scale is

varied from 0.5 μm to 10 μm in order to explore the effects of the native scale thickness

on the interfacial strength. The length of the interconnect is assumed to be 100 mm in the

model. The governing differential equations (Eq. (3.8) and (3.9)) are solved numerically

with a finite difference technique having a step size of 0.01 mm.

In Fig. 3.11(a), the normal compressive stress distribution developed in MCO due to

cooling is plotted along x-direction. In addition, the variation of the cooling induced

curvature along x-direction is also shown in Fig 3.11(a). From the figure it is found that,

both the compressive stress in MCO and the curvature are constant along x-direction

except regions very near the edge. As a result, the total stress distribution in MCO which

is an addition of the normal stress and the curvature induced bending stress will not vary

along the x-direction. In Fig. 3.11(b), the total stress distribution of MCO is presented

thru MCO thickness at any point along x-direction. In the figure it is found that, the stress

distribution obtained from the present method is comparable with the distribution

obtained from the classical lamination theory of composites [23]. But with the classical

lamination theory, it is not possible to obtain the shear stress distribution at the interface.

65

In Fig. 3.12(a), the residual shear stress distribution at MCO-native scale interface,

developed due to cooling from 750

oC to room temperature is plotted along the

interface when the native scale thickness is 6.14 μm. In the figure, it is seen that, the

maximum shear stress is developed in a very small region of the interface near the edge

and the shear stresses are approximately zero along the rest of the interface. In Fig.

3.12(b), the maximum residual shear stress, obtained from the analytical model

with the experimentally obtained room temperature shear strength, of the MCO-native

scale interface, are plotted as a function of the native scale thickness. As the native

scale thickness increases, the maximum residual shear stress decreases. This is because,

with increase of the native scale thickness, the interface compliance increases. As a

consequence of the increase of interface compliance, the interface capability of bearing

shear stresses decreases. From the figure, it is also found that the maximum residual shear

stress developed due to cooling to room temperature is much higher than the room

temperature shear strength of the MCO-native scale interface. The shear stresses

exceeding the shear strength will cause slip induced interface delamination near the edge

by coalescence of the voids or defects at the interface. Interface delamination in turn

promotes buckling of coatings and finally spallation.

3.4.3 Lifetime of MCO

In Fig. 3.13(a), the experimental interfacial fracture energy of the MCO-native scale

interfaces and the analytical interfacial fracture energy of the same interfaces

are plotted as a function of the native scale thickness. In the figure it is found that, with

the native scale growth, both the experimental and the analytical interfacial fracture

66

energy decrease. With longer oxidation, the interfaces become more brittle by the

formation of the native scales and this is the reason for the degradation of the interfaces

described by the experimental results. On the other hand, the inverse relation between the

analytical fracture energy and the native scale thickness implies that with native scale

growth, MCO coatings become more susceptible to cooling induced spallation due to

TCE mismatch as expected. In Fig. 3.13(a), the critical native scale thickness to cause

MCO spalltion due to cooling from 750oC to room temperature is found to be 4.2 μm. To

estimate the effective MCO lifetime, the critical native scale thickness is incorporated in

the native scale growth kinetics chart of 750oC. The kinetics chart is presented Fig.

3.13(b). The parabolic rate constant, kp at 750oC is 0.000254 μm

2/hr which is lower by

two orders of magnitude than kp at 900oC mentioned in Section 3.2. From the native scale

growth kinetics at 750oC, the lifetime of MCO coating is approximately estimated to be

34,720 hours.

The projected MCO lifetimes as a function of operating temperatures are presented in

Fig. 3.14. As the operating temperature increases the native scale growth kinetics

increase and as a result the MCO lifetime decreases significantly as it is shown by the

figure. In Fig. 3.14, the lifetimes measured from an electrical stability testing by NTM are

also presented. In electrical stability testing, the area specific resistance (ASR) of coated

interconnects are measured at 900oC. The 1300 hours at 900

oC over which the ASR

shows stable performance is converted to the equivalent service lifetimes at lower

temperature. The ASR results are consistent with the mechanical results at 750oC but

there is some deviation between the two results at higher temperatures. Figure 3.14 can

67

provide valuable information to the fuel cell designers in reliability assessments of

metallic interconnects.

As indicated by the mechanical results in Fig. 3.14, when the fuel cell operating

temperature is 750oC or above, the projected MCO lifetime is not able to reach target

40,000 hours. To improve the MCO coating integrity, the native scale growth kinetics

must be further suppressed. Yang et al. reported that adding rare earth such as cerium

(Ce) to the spinel MCO can alter the growth kinetics of the native scale beneath the MCO

coatings and thus can improve the adhesion [24]. Besides, lanthanum (La) can be alloyed

to interconnects to increase the bonding of MCO [25]. In addition to reducing the native

scale growth kinetics, MCO lifetime can be enhanced by altering the physical parameters

such as the thickness of the interconnect or MCO. For example, decreasing the

interconnect thickness from 200 μm to 100 μm in the analytical model, shows a lifetime

improvement by 47%.

3.5 Conclusions

Room temperature, four-point bend experiments were performed to predict the effective

lifetime of MCO coatings on metallic interconnects. The experimental results indicate the

degradation of the shear strength and the fracture energy of the MCO-native scale

interfaces with oxidation time i.e. native scale thickness. An analytical model is

implemented to calculate the cooling induced interfacial shear stresses and interfacial

fracture energy. Exceeding the shear strength of the MCO-native scale interfaces, the

residual shear stresses do not cause instant coating spallation. However, the shear stresses

exceeding the shear strength cause interface slip induced delamination near the edge of

68

the interconnect. Interface delamination can in turn cause buckling and finally spallation

of compressively stressed coatings. Comparing the experimental fracture energy with the

analytical fracture energy, the critical native scale thickness is calculated. From the

critical native scale thickness, the MCO lifetime is estimated as 34,700 hours when the

fuel cell operating temperature is 750oC. The present integrated experimental-analytical

methodology can be implemented in reliability assessment of metallic interconnects for

solid oxide fuel cells (SOFCs).

69

References

1. Borie B., Sparks C.J., Cathcart J.V., Epitaxially Induced Starins in Cu2O Films on

Copper Single Crystals X-ray Diffraction Effects. Acta Materialia 10 (1962) 691-697.

2. Appleby W.K., Tylecote R.F., Stresses during Gaseous Oxidation of Metal. Corros.

Sci, 10 (1970) 325-341.

3. Jaenicke W., Leistikow W., Stadler A., Mechanical Stresses During the Oxidation of

Copper and Their Influence on Oxidation Kinetics. III. J. of Electrochem. Soc, 111

(1964) 1031-1037.

4. Evans H.E., Cracking and Spalling of Protective Oxide Layers. Materials Science and

Engineering 120 (1989) 139-146.

5. Christl W., Rahmel A., Schutze M., Application of Acoustic Emission Technique for

the Detection of Oxide Scale Cracking During Thermal Cycling. Material Science and

Engineering 87 (1987) 289-293.

6. Zhang Y., Leistikow W., Shores D.A., Study of Cracking and Spalling of Cr2O3

Scale Formed on Ni-30Cr Alloy. Oxidation of Metals 40 (1993) 529-553.

7. Tammann G., Über Anlauffarben von Metallen. Z. anorg. Chem 111 (1920) 78-85.

8. Timoshenko S., Analysis of Bi-Metal Thermostat, (1925) 233-255.

9. Suhir E., Stresses in Bi-Metal Thermostats. Journal of Applied Mechanics 53 (1986)

657-660.

10. Suhir, E., An Approximate Analysis of Stresses in Multilayered Elastic Thin Films.

Journal of Applied Mechanics 55 (1988) 143-148.

11. Suhir, E., Interfacial Stresses in Bimetal Thermostats. Journal of Applied Mechanics

56 (1989) 595-600.

12. Suhir E., Predicted Thermally Induced Stresses in, and the Bow, of a Circular

Substrate/thin-film Structure. Journal of Applied Physics 88 (2000) 2363-2370.

13. Suhir E., Analysis of Interfacial Thermal Stresses in a Trimaterial Assembly. Journal

of Applied Physics 89 (2001) 3685-3694.

14. Suhir E., Predictive Analytical Thermal Stress Modeling in Electronics and

Photonics. Applied Mechanics Reviews 62 (2009) 1-19.

15. Suhir E., Stresses in Bi-material GaN Assemblies. Journal of Applied Physics 110

(2011) 1-13.

16. Ghorbani H. R., Spelt, J. K., Interfacial Thermal Stresses in Trilayer Assemblies.

Transactions of the ASME 127 (2005) 314-323.

17. Liu D.Y., Chen W.Q., Thermal Stresses in Bilayer Systems with Weak Interface.

Mechanics Research Communications 37 (2010) 520-524.

18. Liu, D.Y., Chen,W.Q. Thermal Stress Analysis of a Trilayer Film/substrate System

with Weak Interfaces, Composites: Part B (2012).

19. Girhammar A, Gopu V. K. A., Composite Beam-Columns with Interlayer Slip-Exact

Analysis. Journal of Structural Engineering, 119 (1993) 1265-1282.

20. Zhang, Y., Shores D. A., Cracking and Spalling of Oxide Scale from 304 Stainless

Steel at High Temperatures. J. Electrochem. Soc, 141 (1994) 1255-1260.

21. Evans A.G., Cannon R.M. Mater. Sci. Forum, 43 (1989) 243.

70

22. Hutchinson J.W., Suo Z., Mixed Mode Cracking in Layered Structures. Advances in

Applied Mechanics. 29 (1992) 63-191.

23. Daniel I. M., Ishai O. Engineering Mechanics of Composite Materials. Second ed.

Oxford University Press, 2006.

24. Yang Z., Xia Z., Nie Z., Templeton J., Stevenson J.W. Ce-Modified (Mn,Co)3O4

Spinel Coatings on Ferritic Stainless Steels for SOFC Interconnect Applications.

Electrochemical and Solid State Letters 11 (2008) B140-143.

25. Akanda S.R., Walter M.E., Kidner N.J., Seabaugh M.M., Mechanical

Characterization of Oxide Coating-Interconnect Interfaces for Solid Oxide Fuel

Cells. Journal of Power Sources, 210 (2012) 254-262.

71

Tables

Table 3.1. Average critical tensile stress of coating (σcct), Average saturated crack spacing

(l), interfacial shear strength (τsti1

), interfacial fracture energy (Gexpi1

), and % spall area

for each type of test specimen.

Test

Specimen

oxidized at

900oC for

Native

scale

thickness

(μm)

Average σc

ct

(MPa)

Average

l

(μm)

Average

( τsti1

)

(MPa)

Average

( Gexpi1

)

(J.m-2

)

% Spall

Area

at 3%

strain

100 hours 1.95 2.36 35.22 2.31 5.12 11.54

600 hours 4.77 0.32 58.83 0.22 0.31 69.48

1000 hours 6.14 0.25 82.99 0.16 0.13 78.78

72

Figures

Figure 3.1. Integrated experimental-analytical methodology to predict lifetime of MCO

coatings on metallic interconnects.

73

(a)

(b)

(c)

Figure 3.2. SEM images of MCO coatings oxidized at 900oC for (a) 100 hours. (b) 600

hours. (c) 1000 hours.

74

Figure 3.3. Native scale growth kinetics at 900oC.

75

Figure 3.4. Normal forces, shear forces and bending moments generated in each layer due

to experiencing temperature differences.

76

Figure 3.5. Differential element of each layer subjected to normal force, shear force and

bending moment.

77

Figure 3.6. Steps to obtain analytical interfacial fracture energy.

78

Figure 3.7. Cumulative AE data synchronized with strain.

79

Figure 3.8. Saturated parallel in-plane transverse cracks in 900oC-1000 hour oxidized

MCO.

80

Figure 3.9. MCO coatings with native oxide scales in spalled area.

81

(a)

(b)

(c)

Figure 3.10. Representative SEM and processed images of MCO coating surfaces at 3%

strain (a) 900oC-100 hour. (b) 900

oC-600 hour. (c) 900

oC-1000 hour.

82

(a)

(b)

Figure 3.11. (a) Variation of normal compressive stress in MCO and curvature of

composite along x-direction. (b) Resultant stress distribution in MCO thru MCO

thickness. (hn = 6.14 μm).

83

(a)

(b)

Figure 3.12. (a) Residual shear stress distribution, at MCO-native scale interface τri1

(hn =

6.14 μm). (b) Shear strength from experiments, τsti1

and maximum shear stress from

analytical model τr,maxi1

as a function of native scale thickness.

84

(a)

(b)

Figure 3.13. (a) Determination of critical native scale thickness to initiate MCO spallation

during cooling from 750oC to room temperature. (b) Native scale growth kinetics at

750oC.

85

Figure 3.14. Projected lifetime of MCO coatings as a function of operating temperature

of a fuel cell.

86

Chapter 4

Effects of Reduction Heat Treatment on Coating Performances

Abstract

Typical MCO coating processes involve both a reduction and oxidation heat treatment to

produce dense and well adherent coatings on metallic interconnects. In a high volume

production, it is convenient to perform the reduction heat treatment in a continuous

manner rather than batch firing to achieve the required cycle-time and capacity

throughput. The continuous reduction heat treatment in furnaces requires significant

capital investments and also incurs high operating costs. Therefore, it is the reduction

heat treatment that contributes to a significant portion of overall coating costs. For

reduced cost and increased throughput, it would be preferable to oxidize the green

coatings directly and eliminate the reduction heat treatment. However, it is anticipated

that coatings without the reduction heat treatment would have poor adherence and density

which would result in reduced functionality. In the present chapter, mechanical bend tests

and electrical area specific resistance measurements are used to evaluate the

performances of MCO coatings processed with and without the reduction heat treatment.

Quantifying these performance changes allows for an assessment of the cost versus

87

performance benefit of the reduction firing. Based on the results, recommendations for

whether the reduction firing is necessary are outlined.

4.1 Introduction

In this chapter, the mechanical bend tests described in Chapter 2 and the electrical area

specific resistance (ASR) tests are performed to evaluate the impacts of reduction heat

treatments on the functional performances of the coatings. The ASR tests are performed

in NexTech Materials (NTM) facilities. The coated interconnect test specimens are

subjected to either NTM’s standard process (reduction and ex-situ oxidation heat

treatment) or modified process where the reduction heat treatment is removed. Table 4.1

shows the key process of the standard and modified coatings. The NTM’s standard

process has been described in brief in Section 2.2.1 of Chapter 2. In the modified process,

the test specimens are oxidized during heating up the stack (in-situ oxidation) at 900oC

for 2 hours and at 800oC for 1 hour. Previous work at NTM has demonstrated that the in-

situ and ex-situ oxidation heat treatments result in equivalent coating performances.

Therefore, comparing the standard and modified coatings will enable to evaluate the

effects of reduction heat treatment on coating performances.

The contact resistance of coated interconnects is characterized by area specific resistance

(ASR). The contact resistance is the product of the electrical resistivity and the thickness

of each layer. As the resistivity of the metal interconnect is very small compared to that

of the coating, ASR is approximately the product of coating’s resistivity and coating’s

thickness. Thus the ASR reflects both the resistivity and thickness of coating [1, 2].

88

4.2 Experimental

The details of bend experiments have been described in Chapter 2. Figure 4.1 shows a

schematic diagram of an ASR setup with a 4-terminal dc technique. For the ASR test, an

interconnect with MCO on both sides is provided along with conductive LSM (standard

cathode) pads for electrical contact. Platinum mesh and leads are bonded to the pads with

LSM inks and cured at 1000oC temperature for one hour. To ensure good contact with the

surface, adequate load was applied in thru-thickness direction. The resistance of the

coating is calculated according to Ohm’s law. In other words, the voltage measured

across the specimen is divided by the applied current density. The experiments were

performed in an atmosphere of humidified air of 800oC for 6500 hours with an applied

current density was 0.5 A/cm2.

4.3 Results

In Fig. 4.2, SEM images of both the standard and modified MCO microstructure are

presented. Both microstructures show a native scale (with high Cr content) between the

coating and the substrate. The modified coating shown in Fig. 4.2(b) has a less dense

MCO structure comparing to the standard coating in Fig. 4.2(a) except for a thin region

(approximately two microns thick) adjacent to the native scales. The modified process

also results in a thicker native scale. The thicker native scale is anticipated to degrade the

adhesion of the modified MCO. The cumulative AE curves obtained from the bend

experiments for the standard and modified MCO are shown in Fig. 4.3(a). As described

in Section 2.4.1 in Chapter 2, the lower slope of the cumulative AE curve in the elastic

region for the modified MCO demonstrates that the modified MCO has lesser tensile

89

cracks formed than the standard MCO. To form a crack, tensile stress needs to be

transferred from the substrate thru interface to the coating. Therefore, the lower crack

density of modified MCO indicates that this type of coating possesses a comparatively

weaker interface. In addition, the modified MCO has a lower onset strain of spallation

than the standard MCO.

The values of the interfacial fracture energy are calculated from the onset strain of

spallation and are tabulated in Table 4.2.The table implies that, the modified MCO has

75% lower interfacial fracture energy than the standard MCO. In addition, after the

experiments, extensive coating spallation (almost 100% Spall Area) was also observed

for the modified MCO. In Fig. 4.3(b), a representative SEM image of the spalled surface

of a modified MCO with a small region of coating is shown. The SEM observations and

the experimental results clearly demonstrate that not performing reduction heat treatment

in the processing of coating produces a less dense and poorly adherent MCO coating.

Figure 4.4 shows a comparison in ASR performance at 800oC of both the standard and

modified samples. The test specimens were subjected to thermal cycles from the test

temperature to room temperature during the ASR tests. Thermal cycling introduces

thermal stresses within the coating which highlight occurrence of spallation. In the figure

it is seen that, MCO processed with the modified firing treatment demonstrate elevated

ASR and premature failure compared to the standard treatment. Higher ASR indicates

coating spallation as coating spallation increases the electrical resistance. The ASR

results are consistent with the mechanical results.

90

The post-ASR SEM observations of the MCO microstructure are shown in Fig. 4.5. After

6,500 hours of oxidation, the microstructure of the standard MCO coating has not

changed significantly. The native scale has grown by less than two microns and the

coating still appears dense. In contrast, the thickness of the native scale under the

modified MCO coating has increased dramatically to nine microns after only 2000 hours.

The native scale growth kinetics is much faster in case of the modified MCO. As a result,

the modified MCO will have significantly lower lifetime comparing to the standard MCO

as described in Chapter 3.

4.4 Conclusions

The following conclusions are made from the work presented in this chapter.

To reduce the capital cost associated with furnaces, NTM’s standard coating

process is modified by eliminating the cost effective reduction heat treatment

from the process flow.

Without the reduction heat treatment, a less dense and poorly adherent MCO

coating is produced. Having permeable microstructure, MCO without reduction

heat treatment is not able to act as a good barrier to oxidation of interconnects and

increases the native scale growth kinetics. As a result, the MCO without reduction

heat treatment is anticipated to decrease the interconnect lifetime significantly.

In spite of saving the capital cost, not performing reduction heat treatment

severely affects the coating’s mechanical and electrical performances. It has been

estimated that elimination of the reduction heat treatment can save 21% of overall

coating cost. In contrast, coating adhesion is decreased by approximately 75%

91

when the reduction heat treatment is removed. The extremely poor performances

of the coating processed without reduction heat treatment justify the necessity of

performing the reduction heat treatment in the coating process.

92

References

1. Zhu W.Z., Deevi S.C., Opportunity of Metallic Interconnects for Solid Oxide Fuel

Cells: A Status on Contact Resistance, Materials Research Bulletin 38 (2003) 957-

972.

2. Huang K., Hou Y.P., Goodenough J.B., Characterization of Iron-based Alloy

Interconnects for Reduced Temperature Solid Oxide Fuel Cells, Solid State Ionics

129 (2000) 237-250.

93

Tables

Table 4.1. Process conditions for standard and modified coating.

Coating Reduction Oxidation

Standard Yes Yes (Ex-situ)

Modified No Yes (In-situ)

Table 4.2. Interfacial fracture energy of standard and modified coating.

Coating Interfacial fracture energy

(J.m-2

)

Standard 8.19 3.75

Modified 1.98 0.21

94

Figures

Figure 4.1. A schematic diagram of an ASR setup.

95

(a)

(b)

Figure 4.2. SEM observation of microstructure of (a) Standard MCO. (b) Modified MCO.

(Courtesy: NTM).

96

(a)

(b)

Figure 4.3 (a) Cumulative AE data synchronized with strain. (b) Post-test SEM image of

modified MCO.

97

Figure 4.4. Long-term ASR behavior of standard vs. non reduction firing MCO coated

interconnect

98

(a)

(b)

Figure 4.5. Post-ASR SEM observations of cross section of (a) Standard MCO

(6500hours). (b) Modified MCO (2000 hours). (Courtesy: NTM).

99

Chapter 5

Investigating Anode-Electrolyte Interfaces by Steady-State Crack

Propagation

Abstract

Insufficient strength of the anode-electrolyte interfaces in fuel cells can cause interfacial

delamination. The delaminated interfaces act as obstacles to oxygen ion conduction. As a

result, the oxidation of hydrocarbon fuels and subsequent production of electrons will be

hindered. Therefore, it is necessary to characterize and evaluate the anode-electrolyte

interfaces. In this chapter, four-point bend experiments are performed on notched bi-

material bar test specimens composed of solid oxide fuel cell anode and electrolyte

material. The bi-layer is notched such that a steady-state crack propagates along the

anode-electrolyte interface. After exceeding the fracture toughness of the notched layer, a

thru-thickness crack is generated from the notch tip and the crack is forced to stop at the

interface and upon further loading; the crack deflects symmetrically along the interface.

The brittle bilayer is sandwiched between two steel stiffeners to keep the crack on the

interface. From the four-point bend theory, as long as the crack front remains between the

inner loading points, the strain energy release rate becomes independent of the crack

length. The relatively constant load in the load-displacement curve is an indication of the

100

steady-state strain energy release rate for the interface fracture. In this chapter, the critical

strain energy release rate for the anode-electrolyte interface is found to be 30 J.m-2

.

5.1 Introduction

As described in Section 1.1 of Chapter 1, electrons are produced when oxygen ions

diffuse through the electrolytes to oxidize hydrocarbon fuels at the anode. If the anode-

electrolyte interface delaminates, the oxygen ion conduction will be severely affected.

Therefore, strong anode-electrolyte interfaces play a vital role in maintaining of power

density of an SOFC stack. Thermal cycling due to frequent start-up and shut-down may

cause delamination of the anode-electrolyte interface. It is important to characterize the

anode-electrolyte interfaces to better understand limitations during both steady-state and

cyclic operation.

Although, to the best knowledge of the author, there are no known characterizations of

anode-electrolyte interfaces available in the literature, interfaces between cathodes and

electrolytes have been investigated by some fuel cell researchers. For example, Delette

and his coworkers applied four-point bending experiments to determine the fracture

energy between the cathode and electrolyte in a planar SOFC [1]. Two steel stiffeners

were bonded on two sides of the test specimen. The resin to bond the steel stiffeners to

the cathode-electrolyte bend test specimen impregnated the porous cathode. The adhesion

between the adhesive and electrolyte was measured from a separate experiment.

Considering the adhesion between the adhesive and the electrolyte, Delette et al. were

able to extract the cathode-electrolyte interfacial fracture energy as 20 J.m-2

.

101

In a similar experimental technique, Malzbender et al. applied a notched anode-

electrolyte-cathode bend test specimen to determine the weakest interface of planar

SOFC cells [2]. Cells were glued with steel strips to act as stiffeners in the test. The

interface between the active cathode and current collector exhibited the weakest interface,

having interfacial fracture energy of approximately 13 J.m-2

. Taking advantage of crack

extension through the anode from the notch tip to the anode-electrolyte interface, the

fracture toughness of the anode was also calculated.

In the present chapter, four-point bend experimental technique is applied to an anode-

electrolyte bi-layer test specimen to measure the interfacial fracture energy. In the

‘experimental’ section of the chapter, the test specimen and loading configuration are

described. The equation for the steady-state strain energy release rate is also developed.

The critical strain energy release rate of the anode-electrolyte interface obtained from the

‘result’ section is compared with the interfacial fracture energy developed from cooling

induced residual stresses. This comparison enables us to determine whether the anode-

electrolyte interface is susceptible to delamination during cooling down from operating

temperature to room temperature.

5.2 Experimental

5.2.1 Test Specimen

Figure 5.1 shows the geometry and loading configuration for a bend test specimen that is

designed to propagate a steady state crack along an interface. Charalambides et al. first

introduced this test method to measure the interfacial fracture energy of a film-substrate

bi-layer system [3, 4]. Later Hofinger et al. modified the test method for a brittle bi-layer

102

specimen [5]. According to the modification proposed by Hofinger et al., two stiffening

layers are glued on both sides of the brittle bi-layer specimen. The stiffening layers are

used to suppress segmentation of the brittle layers. In addition, the stiffeners increase the

stored elastic energy of the system and thus create a more appropriate driving force for

delamination. In Fig. 5.1, each layer is indicated by layer number successively from layer

1 being the bottom stiffener to layer 4 being the upper stiffener. A notch is created in

layer 3 prior to the experiment. The notch acts as a crack initiator. During the experiment,

when the stored elastic strain energy exceeds the fracture toughness of layer 3, a crack

starts to propagate from the notch tip thru layer 3. If it is assumed that layer 2 is tough

enough to prevent the penetration of crack thru layer 2, the crack will stop at the interface

and assuming the interface is weak enough, upon further loading the crack will

symmetrically deflect along the interface between layer 2 and layer 3. The red lines in the

figure indicate the thru-thickness and interfacial crack propagation.

5.2.2. Strain Energy Release Rate

From the theory of four-point bend experiment for interface crack analysis, as long as the

crack front remains between the inner loading spans, it is subjected to constant bending

moment. There is always negligible strain energy in the beam above the crack and as a

result the total strain energy stored, USE can be calculated from the following relation:

∫

∫

where M = Ps/2 is the applied bending moment between the inner loading span, P is the

applied load, s is the half distance between inner and outer loading span and a is the half

103

crack length as shown in Fig. 5.1. is the stiffness of a composite beam of k number

of layers and can be calculated by:

∑

where

is the plane strain elasticity and I is the moment of inertia of cross

section about the neutral axis of the composite beam. After doing some mathematical

manipulations, the following relation is obtained from Eq. (5.1):

[

]

From fracture mechanics, the strain energy release rate, GSE is defined by:

where b is the width of the specimen. From this definition, the following final relation for

GSE is yielded:

[

]

As Eq. (5.2) is independent of the crack length, a, the strain energy release rate is steady-

state in nature as long as the crack front is subjected to a constant bending moment

between the inner loading spans. The steady-state energy release rate (GSS) is achieved

for the condition of the crack length being much larger than the distance from the

interface to the free surface [2]. Thus the steady-state energy release rate can be obtained

analytically under these steady-state circumstances by considering the strain energy

difference between the cracked and the un-cracked section of a beam as referred by Eq.

104

(5.2). During the steady-state crack propagation along the interface, the load becomes

independent of the displacement of the bend fixture and is indicated by a plateau in the

load-displacement curve [2-8].

5.3 Results

The bilayer anode-electrolyte test specimens of 70 mm length and 10 mm width are

manufactured by NTM. The thickness of the anode is 60 μm and the electrolyte is 300

μm. The brittle bi-layer test specimen was sandwiched between two steel stiffeners of

thickness 1 mm. As the electrolyte was relatively thicker, a “notch” was created by

scratching the electrolyte with a steel razor blade. Thus layers 1, 2, 3 and 4 in Fig. 5.1

correspond to a stiffener, the anode, the electrolyte and another stiffener, respectively.

The mechanical and thermal properties of anode and electrolyte are tabulated in Table

5.1. The inner span distance was kept as 40 mm and outer loading span distance was 60

mm.

During the four point bend experiments, a CCD camera (Infinity 2, 2.0 megapixels) with

a long distance camera lens was used to monitor the interface. Figure 5.2 shows an image

of the anode-electrolyte test specimen from this camera. As can be seen in Fig. 5.2, the

crack initiated from the notch tip in the electrolyte, passed through the weak, thin anode

and stopped at the anode-stiffener interface. The figure clearly indicates that the test was

unsuccessful in delaminating the anode-electrolyte interface. As shown in Fig 5.3, the

anode is very thin (60 μm) and has significant porosity. Conversely, the electrolyte is

relatively thick and dense. Therefore the anode’s low fracture toughness relative to the

electrolyte is believed to be insufficient for deflecting the crack to the interface.

105

To reduce the likelihood of the cracking of the anode and causing delamination of the

anode-stiffener interface, the anode-electrolyte interface is purposely weakened in the

vicinity of the notch. During manufacturing of the test specimens, NTM deposited carbon

layers at the anode-electrolyte interfaces where the debonding is intended to start. In

addition, the test specimens were strengthened by applying superglue to the middle

section of the anodes. The glued anodes were cured for 24 hours. Figure 5.4 shows a

schematic of inserting glue on the anode of a carbon deposited (black portion at the

interface) anode-electrolyte bilayer.

Figure 5.5 shows a camera image of the modified test specimen during the experiment.

The figure illustrates that the crack stopped at the anode-electrolyte interface and then

upon further loading propagated along the anode-electrolyte interface. After the

experiments, SEM and EDS analysis were performed on the surfaces of the both side of

the crack. In Fig. 5.6, the SEM images and the corresponding EDS results are shown, and

it is clear that the one side of the crack is the electrolyte (primarily Zr) and the other side

is the anode (primarily Ni). The SEM and EDS results in Fig 5.6 demonstrate the success

of specimen modifications to propagate a crack along the anode-electrolyte interface.

Figure 5.7 presents the experimental load-displacement curves obtained from two

different experiments performed on the modified test specimens. The sudden load drop

indicates crack propagation through the notched electrolyte and the plateau region

represents the steady-state crack propagation along the anode-electrolyte interface. Since

there is variability in the depth of the notch in the electrolyte from one experiment to

another, the peak load at which the crack is initiated from the notch tip will differ. This is

106

seen in the load-displacement curves presented in Fig. 5.7. The plateau region in the load-

displacement curve indicates that the load is independent of the displacement of the bend

fixture and represents the steady-state crack propagation along the interface.

Figure 5.8 gives a plot of Eq. (5.2) and provides the calculated steady state energy release

rate, GSS, as a function of applied load. Using the value of the constant load (80 N) from

Fig. 5.7 in Fig 5.8, the critical strain energy release rate of the anode-electrolyte interface

can be obtained. From Fig. 5.8, the critical strain energy release rate of the anode-

electrolyte interface is found to be 30 J.m-2

.

The residual stresses developed in the anode-electrolyte bilayer due to cooling from

800oC to room temperature are calculated by the method described in Section 3.3 of

Chapter 3. Using the thermal properties of the anode and electrolyte as tabulated in Table

5.1, the analytical fracture energy from the residual stresses is found to be 4.26 J.m-2

by

the method described in Section 3.3 of Chapter 3. Comparing the analytical fracture

energy with the experimental fracture energy obtained in this chapter, it is concluded that

NTM’s anode-electrolyte interface is strong enough to prevent cooling induced interfacial

fracture.

5.4 Conclusions

Four-point bend experiments are performed to characterize anode-electrolyte interfaces

by propagating a steady-state crack. A notch is created in the relatively thick electrolyte.

In the initial experiments the crack initiated from the notch tip penetrates through the

thin, weak anode and stops at the anode-stiffener interface instead of the anode-

electrolyte interface. To propagate cracks along the desired interface, the anode-

107

electrolyte interface is purposely weakened by depositing carbon at the interface vicinity

of the notch tip. The anode is also strengthened in the vicinity of the notch by inserting

super glue. With the modified test specimen, it is possible to propagate a steady-state

crack along the anode-electrolyte interface. The plateau in the load-displacement curve

represents the steady-state crack propagation. The critical strain energy release rate of the

anode-electrolyte interface is found to be 30 J.m-2

. The experimental result indicates that

the as-received anode-electrolyte interface is strong enough to prevent interfacial

delamination during shutdown (cooling) of the fuel cell.

108

References

1. Delette J., Laurencian J., Dupeux M., Doyer J.B. Measurement of the Fracture

Energy at the Interface between Porous Cathode Layer and Electrolyte in Planar

Solid Oxide Fuel Cells.

Scripta Materialia 59 (2008) 31-34.

2. Malzbender J., Steinbrech R.W., Singheiser L., Determination of the Interfacial

Fracture Energies of Cathodes and Glass Ceramic Sealants in a Planar Solid-oxide

Fuel Cell Design. J. Mater. Res. 18 (2003) 929-934.

3. Charalambides P.G, Lun, J., Evan, A.G., McMeeking R.M., A Test Specimen for

Determining the Fracture Resistance of Bimaterial Interfaces. Journal of Applied

Mechanics 56 (1989) 77-82.

4. Charalambides P.G., Cao H.C., Lund J., Evans A.G., Development of a Test Method

for Measuring the Mixed Mode Fracture Resistance of Bimaterial Interfaces.

Mechanics of Materials 8 (1990) 269-283.

5. Hofinger I., Oechsner M., Bahr H. A., Swain M. V., Modified Four-point Bending

Specimen for Determining the Interface Fracture Energy for Thin, Brittle Layers.

International Journal of Fracture 92 (1998) 213-220.

6. Yamazaki Y., Schmidt A., Schol, A., The Determination of the Delamination

Resistance in Thermal Barrier Coating System by Four-point Bending Tests. Surface

& Coatings Technology 201 (2006) 744-754.

7. Howard S.J., Tsui Y.C., Clyne T.W., The Effect of Residual Stresses on the

Debonding of Coatings-1. A Model for Delamination at a Bimaterial Interfaces. Acta.

Metall. Mater 42 (1994) 2823-2836.

8. Tsui Y.C., Howard S.J., Clyne T.W., The Effect of Residual Stresses on the

Debonding of Coatings-II. An Experimantal Study of a Thermally Sparayed System.

Acta. Metall. Mater 42 (1994) 2837-2844.

109

Tables

Table 5.1. Mechanical and thermal properties of anode and electrolyte.

Material Elasticity

(GPa)

Thermal coefficient of

expansion (/oC)

Anode 96 12.6×10-6

Electrolyte 205 10.9×10-6

110

Figures

Figure 5.1. Notched four-point bend test specimen to propagate a steady-state crack along

the interface.

111

Figure 5.2. High magnification camera image of an anode-electrolyte test specimen

during experiment.

112

Fig 5.3. SEM micrograph of a porous anode and a dense electrolyte.

113

Figure 5.4. Steps to strengthen anode by inserting glue.

114

Figure 5.5. Steady-state crack propagation along anode-electrolyte interface.

115

(a)

(b)

Figure 5.6. SEM and EDS analysis of (a) Electrolyte. (b) Anode.

116

Figure 5.7. Experimental load-displacement curve.

117

Figure 5.8. Steady-state energy release rate as a function of applied load.

118

Chapter 6

Conclusions and Future Works

6.1 Conclusions

In the first part of this dissertation, room temperature four-point bend experiments with

acoustic emission (AE) are employed to characterize coating-interconnect interfaces.

Compared to other experimental techniques that measure the adhesion of coatings, the

current experimental setup is less expensive and less complicated. In addition, with the

single experimental setup described in the first part of this dissertation, it is possible to

obtain two important parameters for mechanical characterization of interfaces: (a)

interfacial shear strength and (b) interfacial fracture energy. The experiments are

performed to evaluate the NexTech Materials (NTMs) coatings and coating process for

metallic interconnects. The coating process involves a reduction heat treatment at

elevated temperature followed by an oxidation heat treatment at high temperature. The

coating applied on metallic interconnect is manganese cobalt spinel oxide (MCO). The

following conclusions are drawn from the first part of this dissertation:

In Chapter 2, the adhesion of reduced and oxidized MCO is discussed. After the

reduction heat treatment, the MCO is separated into two distinct layers of Co and

119

MnO. The metallic bond between Co and the metallic interconnect creates strong

interface between the reduced MCO and interconnect.

The oxidation heat treatment produces a dense MCO coating by reaction

sintering of the reduced MCO. Dense, impermeable structure is a necessary

requirement for the coating’s functions as a barrier to chromium migration to the

cathode and inward oxygen diffusion to the interconnect. However the dense

structure is also favorable for thru-thickness crack propagation and as a result the

oxidized MCO has a significantly lower critical tensile stress comparing to the

reduced MCO.

The MCO is a barrier to oxidation but does not completely eliminate oxygen

transport. As a result, both during the oxidation heat treatment and during

operation, native oxide scales form on the interconnect. The formation of native

oxide scales degrades the adhesion of oxidized MCO significantly.

In Chapter 3, the MCO lifetime is estimated as a function of operating

temperature ranging from 750°C to 900°C. With the native scale growth during

fuel cell operation, the adhesion of the MCO is severely degraded. In addition,

the TCE mismatch between the thickening native scales and MCO increases the

tendency of cooling induced MCO spallation. As a consequence, the effective

service lifetime of interconnects is limited by the MCO spallation. An analytical

method is integrated with the four point bend tests to estimate the MCO lifetime.

By analyzing the obtained results, the projected MCO lifetime is not able to

achieve target 40,000 hours when the fuel cell operating temperature is above

120

750°C. Addition of rare earth elements such as Ce or La are expected to further

improve the MCO adhesion.

In Chapter 4, the beneficial impacts of the reduction heat treatment on coating

performances are studied. From a manufacturing cost point of view, it is

preferable to remove reduction heat treatment from the coating process flow.

However, the experimental results in Chapter 4 indicate that the process without

the reduction heat treatment decreases the coating performance significantly by

producing less dense, poorly adherent coatings. Based on the performance

evaluation, it is not recommended to eliminate the reduction heat treatment from

the process flow.

In the second part of this dissertation (Chapter 5), four point bend tests are performed on

notched bi-material anode-electrolyte test specimen to propagate a steady-state crack

along the anode-electrolyte interfaces. Two stiffeners are added on both sides of the

specimen to prevent segmentation of the brittle layers. The notch is created on the

relatively thicker electrolyte. To reduce the likelihood of the crack propagation thru the

thin, porous anode, the anode-electrolyte interface is purposely weakened by depositing

a carbon layer at the interface in the vicinity of the notch. Also the anode is strengthened

in the vicinity of the notch with super glue. The critical strain energy release rate of the

anode-electrolyte interface is found to be 30 J.m-2

.

121

6.2 Future Works

The following future work is recommended by the author:

High temperature bend experiments can be performed on coated interconnects to

describe the evolution of the interfacial strength as a function of temperature. The

obtained results would enable determination of the critical temperature at which

coating spallation may occur during cooling. The present analysis assumes that

coating spallation occurs at room temperature.

Thermal cycling with an integrated AE sensor can be performed to estimate

coating lifetime experimentally. These experimental results would be used to

validate the results of the lifetime modeling in Chapter 3.

With a four point bend tests, a steady-state crack can be propagated along the

coating-interconnect interface. With integrating an ASR setup, it is possible to

obtain electrical resistance as a function of crack length. This will be an

innovative experiment providing useful information to NTM about how the

electrical resistance changes with the interface delamination.

More steady-state interface cracking experiments can be conducted on reduced

and unreduced anode-electrolyte specimens. Effect of redox cycling (reduction

and oxidation) on interfacial strength can be determined.

122

Bibliography

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

2. Haile S.M., Fuel Cell Material and Components. Acta Materials 51 (2003) 5981-

6000.

3. Unveils Largest SOFC Tech Platform. Fuel Cell Bulletin (2009).

4. Zhu W.Z., Deevi S.C., Development of Interconnect Materials for Solid Oxide Fuel

Cells. Materials Science and Engineering A348 (2003) 227-243.

5. Yang Z., Xia G., Simner S.P., Stevenson, J.W., Thermal Growth and Performance of

Manganese Cobaltite Spinel Protection Layers on Ferritici Stainless Steel SOFC

Interconnects. Journal of Electrochemical Society 152 (2005) 1896-1901.

6. Bateni R. M., Wei P., Deng X., Petric A., Spinel Coatings for UNS 430 Stainless Steel

Interconnects. Surface & Coatings Technology 201 (2007) 4677-4684.

7. Puranen J., Lagerbom J., Hyva¨rinen L., Kylma¨lahti M., Himanen O., Mikko K. J.,

Vuoristo P., The Structure and Properties of Plasma Sprayed Iron Oxide Doped

Manganese Cobalt Oxide Spinel Coatings for SOFC Metallic Interconnectors.

Journal of Thermal Spray Technology (2010).

8. Yang Z, Xia G., Li X., Stevenson J.W., (Mn,Co)3O4 Spinel Coatings on Ferritic

Stainless Steels for SOFC Interconnect Applications. International Journal of

Hydrogen Energy 32 (2007) 3548-3554.

9. Larring Y., Norby T., Spinel and Perovskite Functional Layers Between Plansee

Metallic Interconnect (Cr-5 wt % Fe-1 wt % Y2O3) and Ceramic(La 0.85Sr0.15) 0.91

MnO3 Cathode Materials for Solid Oxide Fuel Cells. Journal of Electrochemical

Society 147 (2000) 3251-3256.

10. Borie B., Sparks C.J., Cathcart J.V., Epitaxially Induced Starins in Cu2O Films on

Copper Single Crystals X-ray Diffraction Effects. Acta Materialia 10 (1962) 691-697.

123

11. Appleby W.K., Tylecote R.F., Stresses during Gaseous Oxidation of Metal. Corros.

Sci, 10 (1970) 325-341.

12. Jaenicke W., Leistikow W., Stadler A., Mechanical Stresses During the Oxidation of

Copper and Their Influence on Oxidation Kinetics. III. J. of Electrochem. Soc, 111

(1964) 1031-1037.

13. Evans H.E., Cracking and Spalling of Protective Oxide Layers. Materials Science and

Engineering 120 (1989) 139-146.

14. Christl W., Rahmel A., Schutze M., Application of Acoustic Emission Technique for

the Detection of Oxide Scale Cracking During Thermal Cycling. Material Science and

Engineering 87 (1987) 289-293.

15. Zhang Y., Leistikow W., Shores D.A., Study of Cracking and Spalling of Cr2O3

Scale Formed on Ni-30Cr Alloy. Oxidation of Metals 40 (1993) 529-553.

16. Strawbridge A., Evans H. E., Mechanical Failure of Thin Brittle Coatings.

Engineering Failure Analysis 2 (1995) 85-103.

17. Walter M. GOALI: Electrode Interfaces Stresses, Degradation and Failure, NSF

Proposal 2008.

18. Hou P.Y., Atkinson A. Methods of Measuring Adhesion for Thermally Grown Oxide

Scales. Materials at High Temperatures 12 (1994) 119 199-209.

19. Xie Y., Hawthorne H.M., A Model for Compressive Coating Stresses in the Scratch

Adhesion Test. Surface and Coating Technology 141 (2001) 15-25.

20. Youtsos A.G., Kiriakopoulos M. Tinke T. Experimental and Theoretical/Numerical

Investigation of Thin Film Bonding Strength, Theoretical and Applied Fracture

Mechanics 31 (1999) 47-59.

21. Sun X., Liu W.N., Stephens E., Khaleel M.A. Determination of Interfacial Adhesion

Strength Between Oxide Scale and Substrate for Metallic Interconnects. Journal of

Power Sources, 176 (2008) 167-174.

22. Akanda S.R., Walter M.E., Kidner N.J., Seabaugh M.M. Mechanical

Characterization of Oxide Coating-Interconnect Interfaces for Solid Oxide Fuel

Cells. Journal of Power Sources, 210 (2012) 254-262.

23. Nagl M. M., Evans W. T., Hall D. J., Saunders S. R. J., An in Situ Investigation of the

Tensile Failure of Oxide Scales. Oxidation of Metals 42 (1994) 431-449.

124

24. Delette G., Laurencin J., Dupuex M., Doyer J.B., Measurement of the Fracture

Energy at the Interface betweenn porous cathode layer and electrolyte in planar solid

oxide fuel cells. Scripta Materelialia 59 (2008) 31-34.

25. Malzbender J., Steinbrech R.W., Singheiser, L., Determination of the Interfacial

Fracture Energies of Cathodes and Glass Ceramic Sealants in a Planar Solid-oxide

Fuel Cell Design. J. Mater. Res. 18 (2003) 929-934.

26. Yang Z., Xia G., Nie Z., Templeton J., Stevenson J.W., Ce-Modified (Mn,Co)3O4

Spinel Coatings on Ferritic Stainless Steels for SOFC Interconnect Applications.

Electrochemical and Solid-State Letters 11 (2008) B140-B143.

27. Simner S.P., Anderson M.D., Xia G.G., Yang Z., Pederson L.R., Stevenson J.W.,

SOFC Performacne with Fe-Cr-Mn Alloy. Journal of the Electrochemical Society 152

(2005) A740-A745.

28. Yang Z., Singh P., Stevenson J.W., Xia G. Investigation of Modified Ni-Cr-Mn Base

Alloys for SOFC Interconnect Applications. Journal of The Electrochemical Society

153 (2006) A1873-A1879.

29. Kidner N., Seabaugh M., Ibanez S., Chenault K., Day M., Thrun L., Lifetime

Predictions of Oxide Protective Coatings for Solid Oxide Fuel Cell Interconnects.

Nextech Materials Ltd 2011.

30. Yang Z., Xia G., Simner S. P., Stevenson, J. W., Thermal Growth and Performance

of Manganese Cobaltite Spinel Protection Layers on Ferritic Stainless Steel SOFC

Interconnects. Journal of the Electrochemical Society 152 (2005) A1896-A1901.

31. Timoshenko S., Strength of Materials, part 1, D. Van Nostran Company, NewYork,

1930.

32. Yanaka M., TsukaharaY., Nakaso N., Takeda N. Cracking Phenomena of Brittle

Films in Nanostructure Composites Analysed by a Modified Shear Lag Model with

Residual Strain. Journal of Materials Science 33 (1998) 2111-2119.

33. Wojciechowski P. H., Mendola M.S. Physics of thin films, Vol 16, academic press,

San diego, 1992.

34. Nagl M. M., Saunders S. R. J., Evans W.T., Hall D.J., The Tensile Failure of Nickel

Oxide Scales at Ambient and At Growth Temperatures. Corrosion Science 35 (1993)

965-977.

125

35. Hsueh C.H., Yanaka M., Mutiple Film Cracking in Film/Substrate Systems with

Residual Stresses and Unidirectional Loading. Journal of Material Science 38 (2003)

1809-1817.

36. Tien, J.K., Davidson, J.M. Stress Effects and the Oxidation of Metals. J.V. Cathcard

ed. AIME, 1974.

37. Chandra-Ambhorn S., Dherbey R.F., Toscan F., WoutersY., Galerie A., Dupeux M.

Determination of Mechanical Adhesion Energy of Thermal Oxide Scales on AISI

430Ti Alloy Using Tensile Test. Materials Science and Technology, 23 (2007) 497-

501.

38. Hancock P., Nicholls J.R., Application of Fracture Mechanics to Failure of Surface

Oxide Scales, Material Science and Technology, 4 (1988) 398-406.

39. Schutze M., Mechanical Properties of Oxide Scales. Oxidation of Metals 44 (1995)

29-61.

40. Evans A.G., Hutchinson J.W., On the Mechanics of Delamination and Spalling in

Compressed Films. Int. J. Solids Structures 20 (1984) 455-466.

41. Hou P.Y., Cannon R. M., Spallation Behavior of Thermally Grown Nickel Oxide on

Nickel. Oxidation of Metals 71 (2009) 237-256.

42. Crofer 22 APU Material Data Sheet, No. 4046, June 2008 Edition.

43. Tammann G., Über Anlauffarben von Metallen. Z. anorg. Chem 111 (1920) 78-85.

44. Timoshenko S., Analysis of Bi-Metal Thermostat, (1925) 233-255.

45. Suhir E., Stresses in Bi-Metal Thermostats. Journal of Applied Mechanics 53 (1986)

657-660.

46. Suhir, E., An Approximate Analysis of Stresses in Multilayered Elastic Thin Films.

Journal of Applied Mechanics 55 (1988) 143-148.

47. Suhir, E., Interfacial Stresses in Bimetal Thermostats. Journal of Applied Mechanics

56 (1989) 595-600.

48. Suhir E., Predicted Thermally Induced Stresses in, and the Bow, of a Circular

Substrate/thin-film Structure. Journal of Applied Physics 88 (2000) 2363-2370.

126

49. Suhir E., Analysis of Interfacial Thermal Stresses in a Trimaterial Assembly. Journal

of Applied Physics 89 (2001) 3685-3694.

50. Suhir E., Predictive Analytical Thermal Stress Modeling in Electronics and

Photonics. Applied Mechanics Reviews 62 (2009) 1-19.

51. Suhir E., Stresses in Bi-material GaN Assemblies. Journal of Applied Physics 110

(2011) 1-13.

52. Ghorbani H. R., Spelt, J. K., Interfacial Thermal Stresses in Trilayer Assemblies.

Transactions of the ASME 127 (2005) 314-323.

53. Liu D.Y., Chen W.Q., Thermal Stresses in Bilayer Systems with Weak Interface.

Mechanics Research Communications 37 (2010) 520-524.

54. Liu, D.Y., Chen,W.Q. Thermal Stress Analysis of a Trilayer Film/substrate System

with Weak Interfaces, Composites: Part B (2012).

55. Girhammar A, Gopu V. K. A., Composite Beam-Columns with Interlayer Slip-Exact

Analysis. Journal of Structural Engineering, 119 (1993) 1265-1282.

56. Zhang, Y., Shores D. A., Cracking and Spalling of Oxide Scale from 304 Stainless

Steel at High Temperatures. J. Electrochem. Soc, 141 (1994) 1255-1260.

57. Evans A.G., Cannon R.M. Mater. Sci. Forum, 43 (1989) 243.

58. Stoney G.G., The Tension of Metallic Films Deposited by Electrolysis, Proceedings of

the Royal Society of London 82 (1909) 172-175.

59. Hsueh C.H., Thermal Stresses in Elastic Multilayer Systems, Thin Solid Films 418

(2002) 182-188.

60. Hutchinson J.W., Suo Z., Mixed Mode Cracking in Layered Structures. Advances in

Applied Mechanics. 29 (1992) 63-191.

61. Daniel I. M., Ishai O. Engineering Mechanics of Composite Materials. Second ed.

Oxford University Press, 2006.

62. Yang Z., Xia Z., Nie Z., Templeton J., Stevenson J.W. Ce-Modified (Mn,Co)3O4

Spinel Coatings on Ferritic Stainless Steels for SOFC Interconnect Applications.

Electrochemical and Solid State Letters 11 (2008) B140-143.

63. Charalambides P.G, Lun, J., Evan, A.G., McMeeking R.M., A Test Specimen for

Determining the Fracture Resistance of Bimaterial Interfaces. Journal of Applied

Mechanics 56 (1989) 77-82.

127

64. Charalambides P.G., Cao H.C., Lund J., Evans A.G., Development of a Test Method

for Measuring the Mixed Mode Fracture Resistance of Bimaterial Interfaces.

Mechanics of Materials 8 (1990) 269-283.

65. Hofinger I., Oechsner M., Bahr H. A., Swain M. V., Modified Four-point Bending

Specimen for Determining the Interface Fracture Energy for Thin, Brittle Layers.

International Journal of Fracture 92 (1998) 213-220.

66. Yamazaki Y., Schmidt A., Schol, A., The Determination of the Delamination

Resistance in Thermal Barrier Coating System by Four-point Bending Tests. Surface

& Coatings Technology 201 (2006) 744-754.

67. Howard S.J., Tsui Y.C., Clyne T.W., The Effect of Residual Stresses on the

Debonding of Coatings-1. A Model for Delamination at a Bimaterial Interfaces. Acta.

Metall. Mater 42 (1994) 2823-2836.

68. Tsui Y.C., Howard S.J., Clyne T.W., The Effect of Residual Stresses on the

Debonding of Coatings-II. An Experimantal Study of a Thermally Sparayed System.

Acta. Metall. Mater 42 (1994) 2837-2844.

69. Suo Z., Hutchinson J.W., Interface Crack between Two Elastic Layers, International

Journal of Fracture, 43 (1990) 1-18.

70. Evans A.G, Hutchinson J.W., The Thermomechanical Integrity of Thin Films and

Multilayers, Acta Metall Mater, 43 (1995) 2507-2530.

128

Appendix A: List of Symbols

: Half-length of the test specimen.

: Width of the test specimen.

: Thickness of coating, native scale and substrate respectively.

: Elasticity of coating, native scale and substrate respectively.

: Poisson ratio of coating, native scale and substrate respectively.

: Thermal coefficient of expansion of coating, native scale and substrate

respectively.

: Interfacial shear compliance of coating, native scale and substrate respectively.

: Axial compliance of coating, native scale and substrate respectively.

: Flexural rigidity of coating, native scale and substrate respectively.

: Thermally induced normal force in coating, native scale and substrate

respectively.

: Thermally induced bending moment in coating, native scale and substrate

respectively.

: Thermally induced shear force in coating, native scale and substrate

respectively.

129

: Longitudinal displacement functions in coating, native scale and substrate

respectively.

: Vertical displacement function of tri-layer assembly due to cooling induced bending.

: Difference between elevated temperature and room temperature.

: MCO-native scale interface, native scale-interconnect interface.

: Residual shear stress distribution at MCO-native scale interface, native scale-

interconnect interface.

: Maximum residual shear stress at MCO-native scale interface.

: Residual stress distribution in coating.

: Cooling induced fracture toughness at MCO-native scale interface.

: Analytical fracture energy of MCO-native scale interface.

: Longitudinal, out of plane transverse strain of coating.

Onset strain of coating spallation during bend test.

: Longitudinal, out of plane transverse stress of coating.

: Critical tensile stress of coating.

: Half-length of a coating segment.

: Constant of shear lag model.

: Elastic strain energy stored in coating during bend test.

Maximum interfacial shear stress from a shear lag model.

Interfacial shear strength from a shear lag model.

: Experimental interfacial fracture energy.

: Parabolic rate constant of native scale growth.

130

= Time in hours.

: Half-length of steady state crack at the interface.

s = Half distance between inner loading span and outer loading span.

: Strain energy release rate.

: Steady-state strain energy release rate.