Perovskite-Type Oxide Material As Electro-Catalysts for Solid Oxide Fuel Cells

Home , Solid oxide fuel cell, Strontium sulfate

Perovskite-type oxide material as electro-catalysts 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

Hyunkyu Choi, B.S.

Graduate Program in Chemical Engineering

The Ohio State University

2012

Dissertation Committee:

Umit S. Ozkan, Advisor

Jeffrey Chalmers

Andre Palmer

Hyunkyu Choi

2012

ABSTRACT

Solid oxide fuel cells (SOFCs) are one of the most promising energy conversion devices for the next generation. This is mainly a result of the low emission of environmental hazardous species associated with SOFC’s, high fuel conversion efficiency, and fuel flexibility. However, current ‘state-of-the art’ catalysts used in SOFCs are suffering from several drawbacks that encumber a commercial application of SOFCs in the direct conversion of hydrocarbons. In this application, high operating temperatures are required due to low oxygen reduction reaction (ORR) activity at the cathode, which limits the use of less expensive materials. In addition, ‘state of the art’ anode catalysts are very susceptible to sulfur poisoning and carbon depositions when it is operated with carbonaceous fuels. Since SOFCs aim to use the relatively abundant fuel sources, such as coal-derived syngas and natural gas, high activity and stable catalysts are required. Therefore, the need in developing and testing new catalysts formulations for both cathode and anode catalysts are crucial.

The current research aims to find solutions for these problems through development of novel electro-catalysts based on perovskite materials. The objective is to develop perovskite materials which are capable of enhancing ORR activity as well as possessing a high tolerance to sulfur poisoning and carbon laydown. ii

New formulations of cerium-doped perovskite material were synthesized with varying concentrations of cerium. The bulk structure, oxygen mobility, and electro-catalytic performance of the catalysts were examined using in-situ X-ray diffraction (XRD), oxygen temperature-programmed desorption (O2-TPD), X-ray absorption fine structure (XAFS), CO2 temperature programmed oxidation (TPO) and button cell/impedance measurements.

Cerium-doped perovskites exhibit a cubic structure at room temperature and no apparent structural changes were observed with increasing temperature.

An additional CeO2 phase was observed when cerium concentration exceeded

15% in the A-site of perovskite. Thermal compatibility of the catalyst gets closer to that of galdolinia-doped ceria (GDC) electrolyte by addition of cerium.

Oxidation states of both Fe and Co were shown to be very close to their valence state with different cerium dopant levels at room temperature, indicating that the charge imbalance is compensated by the creation of oxygen vacancies. The oxygen vacancy was generated mainly from the reduction of Co where the Fe contribution was minimal. It was observed that the oxygen vacancy generation was inversely proportional to the dopant level of Ce. However, the electrocatalytic activity showed that the intermediate concentration of Ce doping has the best unit cell performance, which suggests that the secondary ceria phase at higher cerium dopant levels has a detrimental effect on the performance. The trend of the unit cell performance followed the CO2-TPO experiment results and found to be a good probe for button cell performance.

iii

On the anode side, the structural stability of the new formulations was examined under anodic conditions. As the temperature was raised the catalyst with lower cerium loadings changed from its initial cubic structure to a lower symmetry structure while the higher loading cerium samples remained in the original cubic structure. The catalytic activity of cerium-doped catalysts for the methane oxidation showed fairly matched to that of state of the art anode Ni-YSZ catalyst. The sulfur tolerance was significantly enhanced with cerium-doped perovskites showing no deactivation up to 10 hrs of operation while Ni-YSZ catalysts deactivated almost immediately upon introduction of H2S. Surface analysis using XPS on poisoned samples showed that the sulfur exists as strontium sulfate where no significant changes for the transition metals, which are critical for oxidation catalytic activity, were observed. The button cell performance of intermediate cerium-loading catalyst matched to that of Ni-YSZ, while the highest cerium-loading samples showed lower performance due to the presence of secondary phase (CeO2). Lowest cerium-loading showed instability issues during the performance test. Therefore, the catalyst with intermediate cerium-loading catalyst is suitable for the anode catalyst.

Dedication

This document is dedicated to my Wife and my parents.

ACKNOWLEDGEMENTS

I would like to thank my advisor, Dr. Umit Okzan for all the support and guidance she provided during the years. Having her as an advisor was my luck and she always was an example of how a researcher should be and how to set your goals and achieve them.

I was happy to be a part of the HCRG and thank all the group members that had spent times with me during my five years at Ohio State. I have learned and been inspired by each one of you even though some of you did not share the same project. I acknowledge the senior group members Nandita

Lakshiminaryanan, Elizabeth Biddinger, and Dieter von-Deak during my first and second years who shared the same office with me. I am not sure whether I could have learned many things without your answers for my questions.

Thanks to Anshuman Fuller who shared the same SOFC project with me.

During the three years of being a team it was an enjoyable time as well as productive, although I am sure it was a struggle from time to time. I also want to thank to Anne-Marie Alexander for her support and your knowledge as a Chemist broadened my view of research as an Engineer. Thanks to Ilgaz Soykal and

Deepika Singh for helping me with the XAFS experiments during long and fun at

APS. vi

I also acknowledge our financial support from Ohio Coal Development

Office.

Finally, I would like to thank my wife Jinhee and my family who was on my side with all the supports and always being there for me.

vii

VITA

May 4,1980 ...... Seoul, Republic of Korea

August 2006 ...... B.S. Chemical Engineering,

Chung-Ang University, Seoul,

Republic of Korea

August 2007 ...... Researcher,

KCC central research Institute, Yong-In,

Republic of Korea

September 2007 to present ...... Graduate Research Associate,

The Ohio State University

viii

PUBLICATIONS

Lakshminarayanan,N., Kuhn, J.N., Choi, H., Millet, J.M. and Ozkan U.S., ” Variation of structral properties of La1-xSrxCo0.2Fe0.8O3-δ with Sr content: Implications for Oxidation Activity”, Journal of Molecular Catalysis, 336, 23-33 (2011).

Lakshminarayanan,N., Choi, H., Kuhn, J.N.,and Ozkan U.S.,”Effect of additional B-site transition metal doping on oxygen transport and activation characteristics in

La0.6Sr0.4(Co0.18Fe0.72X0.1)O3-δ (where X= Zn, Ni or Cu) perovskite oxides,” Applied Catalysis B, 103, 318-325 (2011).

Choi, H., Fuller, A., Davis, J., Wielgus, C., and Ozkan U.S., “Ce-doped strontium cobalt ferrite perovskites as cathode catalysts for solid oxide fuel cells: Effect of dopant concentration”, Applied Catalysis B, 127, 336-341 (2012)

Choi, H., Fuller, A., Davis, J., and Ozkan U.S., “Perovskite as alternative catalysts for solid oxide fuel cell anodes: Effect of dopants”, in process

FIELDS OF STUDY

Major Field: Chemical Engineering

TABLE OF CONTENTS

Abstract…………………………………………..……………………………………..Ii

Dedication……………………………………………………………………………….v

Acknowledgements…………………………………………………………………..vi

Vita……………………………………………………………………………………..viii

List of Tables…………………………………………………………………………xvi

List of Figures……………………………………………………………………….xvii

CHAPTER 1 : Introduction…………………………………………………………...1

1.1 Introduction to Fuel cells…...………………………………...... …………1

1.2 Advantages of Solid Oxide Fuel Cells……………………………………3

1.3 Drawbacks of Solid Oxide Fuel Cells for commercial use…..………....4

CHAPTER 2 : Literature Reviews…………………………………………………...8

2.1 Historical backgrounds for SOFCs………………………………………..8

2.2 SOFC cathode…………………………..………………..…………………9

2.2.1 Perovskite Oxides as Cathode material……………………….……..11

2.3 SOFC anode………………..………………..……………………………17

2.3.1 Catalyst deactivation due to carbon coking…………………..18

2.3.2 Catalyst deactivation due to sulfur…………………………….18

2.3.3 Cermet as anode material……………………………………...20

2.3.4 Oxide based material as anode……………………………….22

2.4 Button cell fabrication………………..……………………………………23

CHAPTER 3 : Experimental Methods………………..……………………………25

3.1 Catalyst Preparation………………..………………..……………………25

3.2 Catalyst Characterization………………..………………..…………...…26

3.2.1 BET………………..………………..…………………………….26

3.2.2 X-ray diffraction………………..………………..……………....26

3.2.3 Thermogravimetric Analysis (TGA).…………………………..27

3.2.4 Temperature Programmed Desorption (TPD).………………28

3.2.5 Temperature Programmed Oxidation (TPO) with CO2……..29

3.2.6 X-ray Photoelectron Spectroscopy……………………………29

3.2.7 X-ray absorption fine structure (XAFS)……………………….30

3.3 Catalytic activity testing………………..…………………………………31

3.4 Electrochemical activity testing………………..…………………………33

3.4.1 Button cell prepartion………………..………………………….33

3.4.2 Current-Voltage and impedance measurements…………….34

3.4.3 Chronoamperometry………………..…………………………..34

xii

CHAPTER 4 : Ce-doped strontium cobalt ferrite perovskites as cathode catalysts for solid oxide fuel cells: Effect of dopant concentration………..36

4.1 Motivation………………..………………..……………………………….36

4.2 Results and Discussion………………..…………………………………38

4.2.1 Bulk structure properties………………..……………………...38

4.2.2 Temperature-programmed desorption of oxygen (O2-TPD)..49

4.2.3 Bulk properties using X-ray absorption fine structure

(XAFS…………………………………………………………………...55

4.2.3.1 X-ray absorption near-edge spectroscopy (XANES)

analysis…………………………………………………………………55

4.2.3.2 Extended X-ray absorption fine structure (EXAFS)

analysis…………………………………………………………………61

4.2.4 CO2-temperature programmed oxidation (CO2-TPO)……….67

4.2.5 Button cell performance…..………………..…………………..70

4.3 Conclusions………………..………………..……………………………..76

CHAPTER 5 : Effect of Ce doping on the performance and stability of strontium cobalt ferrite perovskites as SOFC anode catalysts …………….78

5.1 Motivation………………..………………..………………..……………...78

xiii

5.2 Results and discussion………………..………………………………….81

5.2.1 X-ray diffraction (XRD).………………..……………………….81

5.2.2 Bulk properties using X-ray absorption fine structure

(XAFS)………………………………………………………….……….86

5.2.2.1 X-ray absorption near-edge spectroscopy (XANES)

analysis…………………………………………………………………86

5.2.2.2 Extended X-ray absorption fine structure (EXAFS)

analysis…………………………………………………………………96

5.2.3 Steady-state methane oxidation………………..…………....100

5.2.4 Surface properties using XPS………………..………………107

5.2.5 Temperature-programmed desorption (TPD) ……………...117

5.2.6 Button cell performance………………..……………………..119

5.2.7 Long-term stability testing………………..…………………...123

5.3 Conclusion………………..………………………………………………126

CHAPTER 6 : Conclusions and recommendations…………………………..127

6.1 Conclusions………………..………………..……………………………127

6.1.1 Cerium-doped perovskite as cathode catalyst……………..127

6.1.2 Cerium-doped perovskite as anode catalyst………………..129

6.2 Recommendations………………..………………..……………………130

6.2.1 Utilization of hydrocarbon fuel in SOFCs……………………130

xiv

6.2.2 Coal-based fuel cells………………..………………………...131

6.2.3 Button cell development………………..………………….....133

6.2.4 Electronic conductivity testing………………..………………133

APPENDIX A : LIST OF ACRONYMS………………..…………………………..135

APPENDIX B : REACTION SYSTEM………………..……………………………137

APPENDIX C : ELECTROCHEMICAL SYSTEM………………..……………….138

APPENDIX D : SAMPLE CALCULATION………………………………………. 139

References………………..………………………………………………………….140

LIST OF TABLES

Table 4.1 : Notation used for cerium-doped perovskites………………………….43

Table 4.2 : Calculated tolerance factor for cerium-doped perovskites…………..45

Table 4.3 : Thermal expansion coefficient values calculated form in-situ XRD patterns…………………………………………………………………………………48

Table 4.4 : Total amount of CO formation during CO2-TPO………………………69

xvi

LIST OF FIGURES

Figure 1.1 : Schematic diagram of an operating SOFC…………………………….7

Figure 2.1 : Schematic of the ORR path from surface and bulk……………….…11

Figure 2.2 : Schematic of perovskite oxide material……………………………….16

Figure 4.1 : XRD patterns of cerium-doped perovskites…………………………..44

Figure 4.2 : Diffraction patterns of cerium-doped perovskites as a function of temperature in air………………………………………………………………………46

Figure 4.3 : Unit cell parameters for cerium-doped perovskites as a function of temperature in air………………………………………………………………………47

Figure 4.4 : Temperature-programmed oxygen evolution in inert (He) environment measure using TGA……………………………………………………53

Figure 4.5 : (a) Temperature-programed oxygen evolution in inert (He) environment monitored using mass spectrometry, (b) O2 desorbed calculated from the TPD profiles...... 54

Figure 4.6 : XANES analysis of cerium-doped perovskite samples at room temperature a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c)

First derivatives of Fe K-edge XANES spectra; d) First derivatives of Co K-edge

XANES spectra………………………………………………………………………...57

Figure 4.7 : XANES analysis of SCeCF10 with elevated temperature under inert a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c) First derivatives xvii of Fe K-edge XANES spectra; d) First derivatives of Co K-edge XANES spectra…………………………………………………………………………………..58

Figure 4.8 : XANES analysis of SCeCF15 with elevated temperature under inert a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c) First derivatives of Fe K-edge XANES spectra; d) First derivatives of Co K-edge XANES spectra...... 59

Figure 4.9 : XANES analysis of SCeCF20 with elevated temperature under inert a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c) First derivatives of Fe K-edge XANES spectra; d) First derivatives of Co K-edge XANES spectra…………………………………………………………………………………..60

Figure 4.10 : EXAFS analysis of cerium-doped samples at room temperature for k2-weighted Fourier Transform of EXAFS function in magnitude: a) Fe K-edge; b)

Co K-edge………………………………………………………………………………63

Figure 4.11 : EXAFS analysis of SCeCF10 at different temperatures for k2- weighted Fourier Transform of EXAFS function in magnitude: a) Fe K-edge; b)

Co K-edge………………………………………………………………………………64

Figure 4.12 : EXAFS analysis of SCeCF15 at different temperatures for k2- weighted Fourier Transform of EXAFS function in magnitude: a) Fe K-edge; b)

Co K-edge………………………………………………………………………………65

Figure 4.13 : EXAFS analysis of SCeCF20 at different temperatures for k2- weighted Fourier Transform of EXAFS function in magnitude: a) Fe K-edge; b)

Co K-edge………………………………………………………………………………66

xviii

Figure 4.14 : CO2-TPO profiles for LSCF6428; SCeCF10; SCeCF15;

SCeCF20……………………………………………………………………………….69

Figure 4.15 : Voltage and power density vs. current density at (a)700oC; (b)750 oC; (c)800 oC……………………………………………………………………………72

Figure 4.16 : OCV electrochemical impedance spectra at (a)700oC; (b)750oC;

Figure 5.1 : In-situ XRD during TPR in 5%H2/He for (a) SCF; (b) SCeCF10; (c)

SCeCF15; (d) SCeCF20……………………………………………………………...84

Figure 5.2 : XRD for Sr1-xCexCo0.2Fe0.8O3-δ (x= 0.0, 0.10, 0.15, and 0.20) catalysts after re-oxidizing at 800oC for 30 min in air ……………………………..85

Figure 5.3 : XANES analysis of SCF with elevated temperature under reducing environment a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c)

First derivatives of Fe K-edge XANES spectra; d) First derivatives of Co K-edge

XANES spectra………………………………………………………………………...89

Figure 5.4 : XANES analysis of SCeCF10 with elevated temperature under reducing environment a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c) First derivatives of Fe K-edge XANES spectra; d) First derivatives of

Co K-edge XANES spectra…………………………………………………………...90

Figure 5.5 :XANES analysis of SCeCF15 with elevated temperature under reducing environment a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c) First derivatives of Fe K-edge XANES spectra; d) First derivatives of

Co K-edge XANES spectra…………………………………………………………...91

xix

Figure 5.6 : XANES analysis of SCeCF20 with elevated temperature under reducing environment a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c) First derivatives of Fe K-edge XANES spectra; d) First derivatives of

Co K-edge XANES spectra…………………………………………………………...92

Figure 5.7 : Quantified oxidation state results from linear combination fittings for a) Fe; b) Co…………………………………………………………………………….93

Figure 5.8 : Result of linear combination XANES fitting at Fe K-edge with reference compound of a) SCF; c) SCeCF10; c) SCeCF15; d) SCeCF20……...94

Figure 5.9 : Result of linear combination XANES fitting at Co K-edge with reference compound of a) SCF; c) SCeCF10; c) SCeCF15; d) SCeCF20……...95

Figure 5.10 : EXAFS analysis over Fe K-edge at different temperatures for k2- weighted Fourier Transform of EXAFS function in magnitude: a) SCF; b)

SCeCF10; c) SCeCF15; d)SCeCF20………………………………………………..98

Figure 5.11 : EXAFS analysis over Co K-edge at different temperatures for k2- weighted Fourier Transform of EXAFS function in magnitude: a) SCF; b)

SCeCF10; c) SCeCF15; d) SCeCF20……………………………………………….99

Figure 5.12 : Steady-state methane oxidation over Sr1-xCexCo0.2Fe0.8O3-δ (x=

0.10, 0.15, and 0.20), LSCF6428 and Ni-YSZ catalysts with sub-stoichiometric oxygen (a) CH4 conversion; (b) O2 conversion ………………………….……….103

Figure 5.13 : Methane oxidation in the presence of 50ppm H2S over Sr1- xCexCo0.2Fe0.8O3-δ (x= 0.10, 0.15, and 0.20), LSCF6428 and Ni-YSZ catalysts with sub-stoichiometric oxygen (a) CH4 conversion; (b) O2 conversion …….....104

Figure 5.14 : Steady-state methane oxidation over Sr1-xCexCo0.2Fe0.8O3-δ (x=

0.10, 0.15, and 0.20), LSCF6428 and Ni-YSZ catalysts with stoichiometric O2 (a)

CH4 conversion; (b) O2 conversion ……………………………………………..…105

Figure 5.15 : Methane oxidation in the presence of 50ppm H2S over Sr1- xCexCo0.2Fe0.8O3-δ (x= 0.10, 0.15, and 0.20), LSCF6428 and Ni-YSZ catalysts with stoichiometric oxygen (a) CH4 conversion; (b) O2 conversion ………….…106

Figure 5.16 : X-ray photoelectron spectra of Ce 3d region for (a) SCeCF10; (b)

SCeCF15; (c) SCeCF20; and La 3d for (d) LSCF6428………………………….111

Figure 5.17 : X-ray photoelectron spectra of Sr3d region for (a) SCeCF10; (b)

SCeCF15; (c) SCeCF20; (d) LSCF6428…………………………………………..112

Figure 5.18 : X-ray photoelectron spectra of Fe 2p region for (a) SCeCF10; (b)

SCeCF15; (c) SCeCF20; (d) LSCF6428……………………………………..……113

Figure 5.19 : X-ray photoelectron spectra of O1S region for (a) SCeCF10; (b)

SCeCF15; (c) SCeCF20; (d) LSCF6428…………………………………………..114

Figure 5.20 : X-ray photoelectron spectra of S2p region for (a) SCeCF10; (b)

SCeCF15; (c) SCeCF20; (d) LSCF6428…………………………………..………115

Figure 5.21 : TPD profiles over Sr1-xCexCo0.2Fe0.8O3-δ (x= 0.1, 0.15, and 0.2) and

LSCF6428 catalysts showing (a) m/z =34; (b) m/z =48; (c) m/z =64; (d) m/z =80

……..…………………………………………………………………………………..117

Figure 5.22 : Button cell performance of Sr1-xCexCo0.2Fe0.8O3-δ (x= 0.10, 0.15 and

0.20) and Ni-YSZ catalysts at (a) 700 oC; (b) 750 oC; (c) 800 oC………….…...120

xxi

Figure 5.23 : Chronoamperometry over Sr1-xCexCo0.2Fe0.8O3-δ (x= 0.10, 0.15, and

0.20) and Ni-YSZ catalysts (a) in the presence of 50ppm H2S; (b) without

H2S……………………………………………………………………………………..124

Figure 6.1 : The schematic of the experimental setup for electrochemical testing………………………………………………………………………………….132

xxii

CHAPTER 1

INTRODUCTION

1.1 Introduction to Fuel Cells

The need to develop a cleaner, more efficient process for energy production has progressed rapidly over the last decade as a result of the increasing concern over the shortage of fossil fuel and its negative environmental impacts.

Renewable sources of energy such as geothermal and hydrothermal offer an alternative to conventional fossil fuels, but their development is impeded due to their very unpredictable nature. In addition to alternative energy sources, alternative energy conversion devices are also considered. Among these possible alternatives, fuel cells are seen to have high potential, for replacing combustion processes for both stationary and mobile power generation applications.

Fuel cells are electrochemical energy conversion devices that can directly convert the chemical energy of the fuel into electrical energy. Internal combustion engines also convert chemical energy into electrical energy, but via an intermediate mechanical energy step, which in turn limits their theoretical efficiency. In internal combustion engines, the theoretical efficiency is 1 thermodynamically limited and does not exceed 50% (actual efficiencies are closer to 20%) but since fuel cells are not restricted by the Carnot limitation, they are generally much more efficient with efficiencies reaching almost 60%[1, 2].

Fuel cells can be classified into several distinct groups based on their application, operating temperatures and the types of electrolyte used. The major classes of fuel cells are Polymer Electrolyte Membrane fuel cells (PEMFC),

Molten Carbonate fuel cells (MCFC) and Solid Oxide fuel cells (SOFC). Polymer

Electrolyte Membrane fuel cells, also known as Proton Exchange Membrane fuel cells (PEMFC) typically operate at around 80 oC where protons travel through the polymer electrolyte. Molten Carbonate fuel cells (MCFC) which use molten potassium lithium carbonate as the electrolyte, operate at 650 oC. Solid Oxide fuel cells (SOFC) use ceramic oxides as an electrolyte and operate between 800 oC and 1000 oC.

Despite having different operating parameters and materials, the basic operating principles of all three fuel cell classifications are the same. All fuel cells consist of three main components, an anode at which the fuel is oxidized into protons and electrons, a cathode where the oxidant is reduced to oxide ion species and an electrolyte which provides a pathway for the conduction of either protons or oxide ions depending on the type of electrolyte[3].

1.2 Advantages of Solid Oxide Fuel Cells

As described above, Solid Oxide fuel cells (SOFC) comprise of three main components namely the cathode, the anode and the electrolyte. Molecular oxygen is reduced to oxide ions at the cathode, which then diffuse through the electrolyte, which is typically an ionic conductor, towards the anode. The oxide ions then react at the anode with the fuel and generate electrons along with water as the by-product. These electrons travel via an external circuit to produce electricity.

The basic schematic of an SOFC is shown in Figure 1.1. Typically, the operating temperatures of SOFCs (in the range of 800 oC ~ 1000 oC) are higher than those of other types of fuel cells described earlier. Although there are several negative consequences of this high operating temperature, it allows the

SOFC to run directly on hydrocarbon fuels without the need for a complex and expensive external reformer, unlike PEM fuel cells. SOFCs have a lot of flexibility in terms of the fuel utilized and this, coupled with its design simplicity, increases the efficiency of the system and ultimately results in reduced operating costs[4].

Furthermore due to the high operating temperatures of SOFCs, carbon monoxide, which is generally treated as a poison for other types of fuel cells, can be electrochemically oxidized to produce CO2 at the anode and subsequently used as a fuel. Another advantage associated with the high temperature is that the heat released can be used as a by-product, either for production of more

3 electricity or for heating. This can possibly increase the overall efficiency of the

SOFC compared to other fuel cells that are operated at lower temperatures. The various chemical reactions that can be processed in a SOFC are outlined below:

Oxidation : 2! ! H2 +O " H2O + 2e

2! ! CH4 + 2O " CO2 + 2H2O + 4e

2! ! CO +O " CO2 + 2e

Steam reforming : CH4 + H2O ! CO + 3H2

Dry reforming : CH4 +CO2 ! 2CO + 2H2

Water gas shift : CO + H2O ! CO2 + H2

Reverse water gas shift : CO2 + H2 ! CO + H2O

1.3 Drawbacks of Solid Oxide Fuel Cells for commercial use.

Although it is clear that there are many potential advantages of SOFC power generation, a significant amount of development is still required to optimize this technology in order for it to be commercialized. One of the main issues that hinders commercialization is the slow kinetics of the oxygen reduction reaction

(ORR) at the cathode site. In order to drive the ORR forward, high temperatures are required. Cathode reaction is ultimately responsible for elevated operating temperatures associated with SOFC’s.

The high operating temperature of SOFCs also severely limits the types of materials that can be used within the system, such as interconnects.

Interconnects are used to connect individual cells within the fuel cell stack as well as to collect current from the system. Since the interconnects are exposed to very reducing or oxidizing environments at high temperatures, its stability is a pressing issue. To be cost effective, metal-based interconnects are most suitable, however ferritic stainless steel composites can only hold up to 750 oC.

This limits the choice of materials which are used for interconnects to ceramics such as lanthanum chromite (LaCoO3), which results in increased costs. In addition, conventional cathode materials such as strontium-doped lanthanum manganite (LSM) react with the zirconia-based electrolytes at elevated temperatures to form La2Zr2O7 and SrZrO3 impurity phases. These phases ultimately impede the oxygen transfer between electrode and electrolyte resulting in a poorer performance[2].

Another issue that prevents the commercialization of SOFCs is the thermal expansion mismatch between the anode, cathode and electrolyte components.

Existing anode and cathode catalysts show large differences in the thermal expansion coefficient (TEC) values compared to that of the electrolyte. In the case where there is a large discrepancy of TEC values between the electrodes and the electrolyte, accelerated degradation of the cell may be observed during cyclic heating and cooling operations. This degradation of the cell could pose significant problems particularly during long-term operation.

When hydrocarbons are used as a fuel, carbon deposition takes place over the catalyst surface via carbon cracking. The performance of an SOFC can be lowered due to the carbon deposition which blocks the active site on the anode, preventing the oxygen’s reaction with the fuel and subsequently decreasing the likelihood of the reactions to occur. In addition, the combination of low-sulfur-tolerant anode materials and the presence of sulfur impurities in natural gas and coal-derived hydrocarbons, leads to significant performance loss during the fuel cell operation.

The work presented in this dissertation focuses on the development of new perovskite-based electro-catalysts for both the cathode and anode sides. In terms of the cathode catalyst, its structure-property relationships are examined in addition to its ORR activity. This will be done in order to develop an optimized formulation for obtaining good catalytic activity as well to ensure stability under actual SOFC cathode conditions. With respect to the anode catalyst, the development of mixed ionic and electronic conducting (MIEC) oxide materials is investigated and they are examined for their sulfur and coking resistance properties.

Figure 1.1. Schematic diagram of an operating SOFC

CHAPTER 2

LITERATURE REVIEWS

2.1 Historical background of Solid Oxide Fuel Cells

Fuel cells are not a new technology; William Grove demonstrated the first concept of fuel cells in 1839. Since then, interest in fuel cells has fluctuated with historical events and scientific challenges. The use of mixed-oxide materials, which are utilized as the electrolyte in SOFC’s, was initially proposed by Schottky

(1935) and later Wagner (1943). It was suggested that the ion conduction could take place due to oxygen vacancies present within the sub-lattice. In 1937, Baur and Preis first demonstrated the use of SOFCs at elevated temperatures of

1000oC. However, due to the high operating temperatures which are employed during reactions, coupled with the reducing nature of the fuel at the anode side, deactivation of the oxide materials was observed. This ultimately hindered further development of SOFCs. Despite this hindrance, there has been renewed interest in SOFC type applications since the 1960s, due to advances in the preparations and production of ceramic materials. There has been a lot of research focused

8 on improving efficiencies, as well as further developing the materials which are currently in use[2].

2.2 SOFC Cathode

As mentioned in chapter 1, the components of the SOFC are subjected to elevated temperatures and severe reducing or oxidizing environments.

Therefore, it is important that each component be able to withstand these types of environments and also have multifunctional capabilities. The main function of the cathode is to have a sufficient catalytic activity of oxygen reduction reaction

(ORR) to generate oxide ions. The oxygen reduction reaction can be described by the reaction equation presented below:

- 2- O2 + 2e -> 2O

For this reaction to take in place, a triple-phase boundary (TPB) is required. The triple phase boundary acts as a catalytic active site in which the oxygen ion conductor, the electronic conductor and the gas phase oxygen come into contact with each other. The breakdown in any portion of the TPB will not allow the reaction to take place[5].

Common materials used for the cathode are perovskite-type oxides.

These materials provide mixed conductivity i.e. they allow the conduction of both

9 the oxide ions and electrons. By using this material, electrochemical ORR can occur at the electrode surface as well as within the bulk of the electrode. The schematic of the ORR reaction at the perovskite is shown in Figure 2.1.

In order for oxygen to react with an electron to produce an oxide ion, the cathode material needs to have adequate porosity to allow gaseous oxygen to diffuse through to the TPB. Generally, high electronic conductivity (>100 S cm-1) of the cathode material is required to provide a steady supply of electrons for the

ORR to proceed. The oxide ions generated from this process are subsequently transported to the anode active sites through the electrolyte to complete the anodic reaction. For this reason, a high ionic conductivity of the cathode material is also crucial. From a stability viewpoint, the cathode material should have a thermal expansion coefficient (TEC) that is compatible with that of the electrolyte in order to prevent thermal cracking during the operation of the SOFC. In addition, the cathode material requires chemical and mechanical stability under the high operation temperatures associated with SOFC’s (800oC~1000oC) and an oxidizing atmosphere [6].

Figure 2.1 Schematic of the ORR path from surface and bulk[6].

2.2.1 Perovskite Oxides as Cathode Material

Perovskites are oxide based materials which have an ABO3 structure, in which the A-site is commonly made up of rare earth metals and the B-site with transition metals. In general, the transition metal cation (i.e. B) has a 6 fold- coordination with oxygen anions to give an octahedral complex. The A cation has a 12-fold coordination, with respect to the oxygen anions that are located in between octahedral complexes of the structure. Figure 2.2 illustrates the general structure of perovskite material with the chemical formula ABO3. The red spheres depict A-site atoms, the blue spheres are B-site atoms, and the grey sphere is an oxygen atom. The A- and B-sites can be varied, because approximately 90% of metallic elements in the periodic table are stable in the perovskite structure.

Partial substitution of each site (i.e. a perovskite with the formula of A1-xA’xB1- yB’yO3) is also possible, as a multi-component perovskite structure has been shown to lead to better performance in SOFCs[2, 7, 8].

The current commercial catalyst is Strontium-doped Lanthanum

Manganite oxide (LaSrMnO3 or LSM). Divalent strontium and trivalent lanthanum on the A-site of the perovskite structure allow manganese to adopt both tri- and quadro-valent states within the perovskite structure. The presence of multi- valence manganese in the perovskite helps to promote electron transfer through the material via a ‘hopping’ mechanism. This ultimately results in high electronic conductivity. Additionally, a good thermal expansion match with the electrolyte, yttria-stabilized zirconia, has made this catalyst a popular choice [9].

Oxygen vacancies under operating conditions of SOFCs are an important factor in the oxygen reduction kinetics and mass transfer rates. If the oxygen mobility is not sufficient, oxygen cannot get fully reduced in the bulk until at the

TPB. Therefore, it is imperative to develop novel cathode materials that form oxygen vacancies at desired SOFC operating temperatures. The amount of oxygen vacancies formed in the cathode material is proportionally related to the chemical potential driving force of ionic mobility of the oxygen ion and surface reduction kinetics. Oxygen vacancies on surface allow adsorbed oxygen atoms to be incorporated into the bulk while being reduced to oxide ions. Also, oxygen vacancies in the bulk provide pathways for the oxygen ions to move from the surface to the electrolyte interface. Increased oxygen vacancies in the cathode can make molecular oxygen reduction to take place not only at the electrolyte- electrode interface, but also throughout the cathode and eventually cause an increase in the surface area to be used in the reaction by delocalizing the TPB.

For this reason, it is essential to synthesize catalysts which form more oxygen vacancies at lower temperatures. Forming composite cathodes with electrolytes has enhanced the oxygen transport capabilities of LSM, however, composite cathodes cannot support high current densities below 800oC [5, 10-12].

There has been a significant research effort to substitute existing LSM cathode with other perovskite materials by additional doping at the A- and B-site to improve ORR activity. A-site substitution in Fe-Co based perovskite materials with praseodymium (Pr), samarium (Sm), and gadolinium (Gd) was examined to identify the effect of the cation size on structural changes within the perovskite structure. The structure of perovskites is an important factor to determine the conductivity of the material; greater symmetry improves the random ‘hopping’ of the electron, subsequently leading to increased conductivity. As the cation size decreases from Pr to Gd, the structure of the perovskite changes from a cubic to an orthorhombic phase. Cubic structures were formed with La; orthorhombic with

Pr and Sm, and both cubic and orthorhombic phases were observed in Gd- substituted perovskites. These results indicate that La-substituted perovskites have the highest conductivity [13-15].

Perovskites of the formula Ln1-xAxM1-yMnyO3-δ (where A= Ca, Sr;

M=transition metal) were synthesized (PrSrMnO3 and PrCaMnO3) and showed promise as cathode materials, particularly as the electrical conductivity is comparable to that of LSM. Gd-based materials doped with strontium or calcium at the B-sites were found to have a better thermal expansion coefficient

13 compatibility than LSM. In the case of La(Ni,Al)FeO3, the peak electronic conductivity was established as being 3 times greater than that of LSM. An increase in the nickel content at the B-site of the perovskite positively influenced both the non-stoichiometry of the oxygen and the electrical conductivity. On the other hand, increasing the iron content had the opposite effect. Also, introducing aluminum to the system was found to raise the resistance of the Sr-doped material [16-22].

Other rare earth metals substituted at the A-site in place of lanthanum, such as barium-doped Fe-Co based perovskite, have shown increased performance, especially in IT-SOFCs (intermediate temperature- SOFCs) at temperatures as low as 650oC. High oxygen vacancy formation strongly influences relevant properties such as the oxygen bulk diffusion and the surface exchange kinetics as mentioned previously. However, these barium-doped materials had high thermal expansion coefficient mismatch with the electrolyte, which can result in cracking during the fuel cell operation[23-26].

Previous work done in our group quantified oxygen vacancy generation in

Co-doped La0.6Sr0.4FeO3 by varying the concentration of Co, as well as characterized its bulk and surface properties[27-29]. It was shown that a high synthesis temperature was required to form the perovskite structure for

La0.6Sr0.4FeO3 based perovskites via the solid-state method. It was also observed that a non-linear trend existed for the oxygen vacancy formation of the perovskite structure with respect to Co concentration. It was shown that the perovskites

14 undergo a transition from the rhombohedral structure to cubic structure with increasing temperature, and that the transition takes place at lower temperatures with increasing Co content[28]. Oxygen exchange kinetics and mass transfer properties were evaluated on the samples and it was found that the oxygen exchange constants increased with Co concentration whilst the oxygen diffusion coefficient was highest for the intermediate Co content samples[29]. The surface properties were also examined by XPS and using methanol as a probe molecule with adsorption and desorption experiments. The amount of Lewis sites and basic sites were determined and confirmed with IR spectroscopy[27]

Additional B-site doping with Zn, Ni, and Cu on La0.6Sr0.4Co0.2Fe0.8O3-δ to improve the cathode properties has been examined in our previous studies[30]. It was reported that additional B-site doping led to an increase in oxygen activation and vacancy generation properties over undoped samples with Zn doping showing the best performance. The materials showed good thermal compatibilities with YSZ electrolyte, which was confirmed with TEC values calculated from the in-situ XRD experiments. As a screening test for actual cathode material, CO2-TPO experiments were performed. Samples doped with

Zn as a second dopant in the A site were seen to have a lower temperature for the onset of the ORR activity due to increased accessibility of the lattice oxygen and higher oxygen mobility[30].

Figure 2.2. Schematic of perovskite oxide material.

2.3 SOFC anode

Anode materials have to meet strict requirements under the SOFC operating conditions. Similar to the cathode requirements discussed in the previous section, the anode must also possess sufficient ionic and electronic conductivity for the anodic reaction to occur. The anode material must have an acceptable TEC match with the electrolytes, in addition to being chemically compatible with other components. It is necessary for the anode to be catalytically active in order to oxidize the fuels with the electrochemically pumped oxygen being transported across the electrolyte. Due to the severe reducing environment and high temperature in the anode, the catalyst must also be chemically and thermally stable.

A Ni-YSZ cermet anode is considered to be a state of the art in SOFC because of its low cost, high temperature chemical stability under reducing atmospheres and good thermal compatibility with YSZ electrolyte[31]. The cermet structure ensures porosity and thermal compatibility with the electrolyte.

Ni serves as an excellent catalyst for the electrochemical oxidation of hydrogen, while the YSZ provides both ionic conductivity and compatibility with the electrolyte [32].

2.3.1 Catalyst deactivation due to carbon coking

Current Ni-YSZ anodes suffer from carbon deposition when SOFCs are run with carbonaceous fuels[31]. Carbon deposition onto Ni-YSZ can occur through the disproportionation of carbon monoxide (i.e., the Boudouard reaction) and the reduction of carbon monoxide by hydrogen or methane cracking, a result of an insufficient supply of oxide ions[33]. Because the carbon formation on Ni-

YSZ blocks the active sites for oxygen adsorption, SOFC performance can be lowered by incomplete oxidation of the fuel. A large amount of steam is usually required to remove carbon deposition, which is economically expensive[34]. Both anode and cathode developments will allow electrochemically pumped oxygen to arrive more quickly to the anode for oxidation of carbon deposits and will provide resistance to coking[35].

2.3.2 Catalyst deactivation due to Sulfur

In addition to the carbon deposition, Ni-YSZ suffers severe deactivation in the presence of fuels which contain low levels of H2S, because of the low sulfur tolerance of Ni. This performance loss can be explained by (1) Physisorption/ chemisorption of H2S at the catalyst active sites; (2) sulfidation of anode material due to reaction between sulfur and Ni/YSZ, which converts metallic species into

18 metal sulfides resulting in loss of catalytic activity, conductivity, and stability[36].

The reactions are shown below:

Physisorption: H2S(g)! HS(ads)+ S ! S(ads)+ H2

Chemisorption: Ni + H2S ! NiS + H2

Sulfidation: 3Ni + xH2S ! Ni3Sx + xH2

Under normal SOFC operating temperatures (800oC~1000oC), thermodynamic prediction shows that bulk nickel sulfides tend to form at low H2S concentrations, generally below about 100 ppm and are more favorable at temperatures below 600oC[37, 38]. High temperature operation (>900oC) reverses the effect of sulfur poisoning whereas temperatures between

700~850oC make poisoning irreversible. This suggests that as the temperature is increased, the concentration of sulfur that can be tolerated by the Ni/YSZ anode also increases. This ultimately indicates that a higher operating temperature for

SOFC’s might be a solution for sulfur poisoning in anode catalyst[39]. However, because of the advantages of SOFCs working under intermediate temperature are more economically feasible, it is desired to synthesize anode materials that are tolerant to sulfur poisoning.

Sulfur can be utilized as fuel via reaction (1) and (2) below:

2! ! Reaction 1: H2S + 3O " H2O + SO2 + 6e

2! ! Reaction 2: S + 2O " SO2 + 4e

From a sulfur poisoning point of view, reactions 1 and 2 are preferred over the physisorption, chemisorption and sulfidation reaction pathways described above due to the fact that sulfur dioxide is considered less harmful for SOFC anodes.

Therefore, sulfur poisoning can be mitigated by pumping excess O2- at TPB, which will subsequently lead to reaction 1 and 2 taking place[40].

2.3.3 Cermet as anode material

Several approaches have been investigated to replace the current Ni-YSZ cermet anode. Sulfur containing materials and non Ni-based alloys have been studied in an effort to improve the sulfur tolerance. Sulfur containing materials such as CuFe2S4, NiFe2S4, WS2, CuCo2S4 and CoS2 were tested in an H2S environment, but deactivation occurred when long term studies were performed

[36]. Co-Mo-S showed stable performance up to 5 days of testing with H2S in the

20 fuel, however, testing with CO led to a performance loss due to carbon deposition[41].

In addition, active research is currently in progress to develop cerium oxide-based anode materials[32]. Ceria (CeO2) is a mixed conductor in the reducing fuel environment, which does not limit the reaction zone to three-phase boundaries, but expands it. Ionic conductivity of ceria is higher than that of YSZ, which increases the transport of oxygen ion from the electrolyte to the anode.

Ceria is also known as a material that can store and transfer oxygen readily, increasing the oxygen storage capacity[42]. Cu/Ceria-based materials improved tolerance towards coke deposits and sulfur [43, 44].However, they are limited by poor thermal stability due to low sintering temperature of Cu and low power densities/fuel conversions as compared to ceramics[36]. Oxygen-ion conductivity of ceria can be increased significantly by doping with other rare-earth cations and it is reasonable to expect that ionic conductivity could effect oxygen storage capacity through transfer of oxygen from the bulk[45]. Cu/CeO2/YSZ and

Cu/Co/CeO2/YSZ showed good performance under CO fuel compared to H2.

Ni/YSZ is more effective for H2 environment but they do not effectively catalyze the oxidation of CO[46]. Also, doping CeO2 with Fe or Mn has a significant effect on the electrochemical oxidation of methane and showed activity for the anodic reaction when LSGMC is used as the electrolyte. Under a reducing environment,

Mn- and Fe-doped CeO2 showed high oxidation tolerance where Fe-doping improved electrical conductivity[47, 48].

2.3.4 Oxide based material as anode

Replacing metal-based anodes with metal oxides such as perovskite materials has shown promise as potential anode catalysts due to their high resistance to coking and sulfur tolerance, as well as high ionic conductivity [49,

50]. Increased ionic conductivity of these materials can increase the triple phase boundary (TPB), where the oxide ions, the fuel and the catalytic active sites come into contact.

The perovskite material, La0.75Sr0.25Cr0.5Mn0.5O3-d (LSCM) is active for direct and total oxidation of hydrocarbon fuels, but less active for reforming. It was also observed that the selectivity of oxidation of hydrocarbon was dictated by oxygen ion flux[51-53]. Transition metals such as Co, Fe, and Ti were incorporated into the B-site of perovskite structure to investigate the methane oxidation activity and Co showed the highest activity while Ti showed the lowest[54]. Ni substitution in LaCrO3 increased methane conversion due to the high methane scission activity [55] as well as the absence of carbon deposition under steam/methane ratio equal to 1 or less conditions[56]. Redox stability of perovskite structure was enhanced when Nb is doped into SrTiO3 but electrocatalytic activity is insufficient in an anode environment[57]. Cerium-based perovskites showed promising results for tolerance toward sulfur poisoning.

Ce0.9Sr0.1VO3 (CSV) was chemically stable even in the 0.5%H2S/CH4 environment and showed high activity towards the oxidation of H2S in the fuel

3+ feed, but not active toward H2 or CH4 oxidation[58, 59]. Doping Cr into CSV perovskite leads to an improved conductivity and admixing of NiO increased methane oxidation activity, but Ni was not stable under H2S-containing fuels[60].

Cerium has not only improved the sulfur tolerance, but also improved oxygen mobility within the perovskite structure therefore increasing the performance of

SOFC[61, 62].

Double perovskites (AA’BB’O6) are also gaining attention for their electrochemical activity under anodic conditions. Sr2XMoO6 (with X= Mg, Mn, Co,

Ni) was tested and it showed good catalytic activity, little carbon coking and acceptable sulfur tolerance upon using methane as fuel. However, degradation of the structure occurred while using methane as the fuel[63-65]. The research based on perovskite materials is still ongoing. Increased catalytic activity as well as improved sulfur/ coking tolerances are needed in order to make the SOFC commercially feasible.

2.4 Button cell fabrication

To test the performance of the catalyst, a button cell is fabricated for lab scale experiments. The button cell is composed of 3 layers: anode and cathode electrodes and an electrolyte in between. In fabricating a button cell, either an electrolyte-supported or anode-supported button cell is possible. Both have their advantages and disadvantages. For the electrolyte-supported button cell a thick

(~120µm) electrolyte supports a thin electrode layer (~ 30µm) on either surface.

This configuration shows a relatively strong structural support from a dense electrolyte and little chance of failure due to anode re-oxidation. However, the thickness of the electrolyte limits SOFCs operating temperature to about

900~1000oC due to the high-ohmic losses of the electrolyte[66]. Typically, the electrodes are printed on the electrolyte using either a brush or screen printer.

Slurries of the electrode materials are made with α-terpineol, toluene as dispersant, and ethyl cellulose as a binder so that the electrode can be brushed or screen-printed onto the electrolyte. All compositions should be optimized based on each electrode that is made in order to minimize the TEC mismatch between electrode and electrolyte, which could potentially cause the stress to each layer and lead to the cracking of the cell. Optimized temperature for curing the anode layer is typically in between 900oC and 1350oC for both layers [67, 68].

Electrode-supported SOFCs consist of thick electrode (~150µm) on either an anode or cathode electrode, which supports the thin electrolyte layer (~30µm).

The most common configuration is a thick anode supporting a thin electrolyte and the cathode layer. The main advantage of the anode-supported SOFCs is the relatively lower ohmic resistance of the electrolyte to lower the operation temperature by around 150~200oC. However, this method leads to a mass- transport limitation to the thick electrode. Also, to fabricate a dense and porous anode, a careful co-sintering process has to be adopted which could be a complex and time-consuming process[66, 69-71].

CHAPTER 3

EXPERIMENTAL METHODS

3.1 Catalysts Preparation

The Sr1-xCexCo0.2Fe0.8O3-δ (x= 0, 0.10, 0.15, and 0.20) and the

La0.6Sr0.4Co0.2Fe0.8O3-δ samples were prepared by the EDTA-Citrate complexing method. All samples used metal nitrates as precursors. Stoichiometric amounts of Sr(NO3)2 (Alfa Aesar), Co(NO3) •6H2O (Acros), Fe(NO3) •6H2O (Aldrich),

Ce(NO3)3•6H2O (Fisher Scientific), or La(NO3)3•6H2O (Aldrich) for

La0.6Sr0.4Co0.2Fe0.8O3-δ were dissolved in distilled water and EDTA was added to the mixed metal nitrate solution under stirring. An NH4OH solution was added to the mixed solution to adjust the pH value to 6 while the solution was heated up to

60oC. Citric acid was then introduced to the solution, which was heated up to

90oC and kept at the temperature during magnetic stirring, until a brown gel was formed. The mole ratio of metal ions to EDTA and to citric acid was 1:1:1.5. The gel was dried overnight at 150oC to obtain the dry the precursor. The dried precursor was then ground with a mortar and pestle and calcined under air at

1100oC for 5hrs to form the perovskite phase.

The NiO-YSZ samples were synthesized using solid-state methods. NiO

(Sigma) and Yittria-Stabilized Zirconia (Sigma). The YSZ was 92% zirconia and

8% yttria (8YSZ) on a molar basis. The precursors were mixed in a ceramic jar with 60% NiO and 40% YSZ on a mass basis. Zirconia grinding beads and distilled water were added to the jar and ball-milled for 24 hrs at 120 revolutions per minutes using a long roll jar mill (U.S Stoneware) to obtain a homogeneous mixture. The mixture was then dried and crushed with a mortal and pestle and calcined at 700oC for 5 hrs.

3.2 Catalyst Characterization

3.2.1 BET

Surface area measurements were made by physical adsorption of N2 at

77K on a Micromeritics ASAP 2014. Due to the low surface area of perovskite catalysts Kr was used as adsorption gas instead of N2. The surface area of the samples was measured by fitting the data to BET isotherm.

3.2.2 X-ray diffraction

In-situ X-ray powder diffraction patterns were recorded with a Bruker D8

Advance diffractometer. A Cu Kα1 (λ= 1.5046 Å, 40 kV and 50 mA) was used as

26 the radiation source. The instrument is equipped with an incident beam Ge (111) monochromator, incident beam Soller slits, and a Braun position-sensitive detector (PSD). The measurements were made with the HTK 1200 oven equipped with graphite windows. Samples were loaded on a 0.5 mm deep alumina holder and the data were collected for 2θ values from 20o to 90o at a step size of 0.0144 with a dwell time of 1 s. A heating rate of 10 oC/min and a 20 min hold time were used prior to data collection at each temperature. For the cathode experiment, the data acquisition took place under ambient atmosphere while 5%H2/He with a flow rate of 30 sccm/min was introduced to the chamber for an anode experiment.

3.2.3 Thermogravimetric Analysis (TGA)

Thermogravimetric analysis of the samples was performed using a

Setaram TG-DSC 111, which is capable of simultaneous microgravimetry and scanning calorimetry measurements. Prior to each set of experiments, flow balancing of the instrument in He at room temperature was performed. The samples (~45 mg) were loaded into a Pt crucible and the mass and heat were allowed to equilibrate at room temperature to obtain a steady signal before flow was started. The samples were heated to 750 oC at 5 oC/min in 30 ccm of He with 30 min of isothermal hold before cooling down to room temperature. The

o samples were pretreated under 10%O2/He at 750 C for 30 min before the TPD.

The oxygen vacancies were calculated from the mass change from the program.

3.2.4 Temperature Programmed Desorption (TPD)

The oxygen TPD was conducted on an Autochem II 2920 instrument connected to a Cirrus RGA-MS (MKS instruments). The mass signals were traced in selected ion mode with a Faraday cup detector. The catalyst samples

(100 mg) were loaded in a fused quartz reactor. Prior to TPD experiments, each

o sample was pre-treated with 30 sccm/min of 10%O2/He at 750 C and held for 30 min before it cooled down to room temperature. The flow was switched to 30 sccm/min of He after the reactor was cooled down to room temperature and purged for 1 h and additional 30 min to stabilize the MS signals. The temperature was then heated up to 900 oC followed by the isothermal hold for 30 min with a ramp rate of 5oC/min. The desorption products were monitored using the MS and peak integration/deconvolution was performed using Grams AI software.

For H2S TPD experiments, MS was also operated in selected ion mode

(12, 16, 18, 28, 32, 33, 34, 44, 48, 60, 64, 76 and 80 monitored). Samples were

o pretreated with 50 ppm H2S/N2 (50 sccm/min) at 800 C with an isothermal hold for 10hrs and then the samples were cooled down to 500oC under the same flow.

The flow was switched to He (50 sccm/min) at 500oC and the sample was cooled down to room temperature. After the pre-treatment, samples (100mg) were

28 loaded on top of quartz wool in a fused quartz reactor and the reactor was purged for an hour and additional 30 min to allow for the MS signals to stabilize.

The temperature was then increased from 50oC to 900oC under 50 sccm/ He with a ramp rate of 10oC/min followed by isothermal hold for 30 min. The desorption products were monitored using the MS.

3.2.5 Temperature Programmed Oxidation (TPO) with CO2

CO2-TPO experiments were performed after the samples were reduced

o with 30 ccm of 5%H2/He at 850 C for 30 min and subsequently cooled down to room temperature using an Autochem II 2920 instrument connected to a mass spectrometer (Cirrus RGA-MS, MKS intruments). Once the reactor cooled down, the gas was switched to He at 30 sccm/min and purged for an hour followed by an additional 30 min to stabilize the MS signals. The TPO experiment took place

o with 30 sccm/min of 10% CO2/He with a linear temperature ramp (10 C/min) up to 850 oC and isothermal hold for 30 min at that temperature. The amount of CO generated was quantified using Grams AI software.

3.2.6 X-ray Photoelectron Spectroscopy

X-ray photoelectron spectra (XPS) were collected on a Kratos Ultra Axis

Spectrometer using mono Al Kα radiation under high vacuum. Prior to collecting

29 the spectra, all the samples were exposed to 50 ppm H2S/N2 (50 sccm/min) at

800 oC for 10 hours after which the flow was switched to He (50 sccm/min) at 500 oC while the samples were cooled down to room temperature. Poisoned samples were ground into carbon tape for analysis. The spectrometer with a slot aperture was used and analysis was conducted in spectrum analyzer mode and hybrid lens mode. The charge neutralizer was set at a current of 2.1 A, a bias of 1.3 V, and a charge of 2.4 V during analysis. A survey scan (1 sweep/ 100 ms dwell) was collected between 1400 and 0 eV. Concurrent sweeps for the Sr 3d (8/ 150),

O1s (8/ 75), S 2p (8/ 500), C 1s (8/ 150), Ce 3d (8/ 900), Fe 2p (8/ 350), La 3d (8/

400), and Co 2p (8/ 900) regions were collected. Binding energies for all regions were corrected with the C 1s peak at 284.5 eV. The peaks were deconvoluted with Gaussian-Lorentzian curves using the XPS Peak 4.1 software.

3.2.7 X-ray absorption fine Structure (XAFS)

X-ray absorption fine structure (XAFS) spectroscopy data were acquired at the insertion device beamline (5-BM-D) of the DuPont-Northwestern-Dow

(DND) Collaboration Access Team (CAT) of the Advanced Photon Source,

Argonne National Laboratories. The XAFS spectra were obtained at the Co and

Fe K edge for perovskite catalysts. The samples were pretreated at 10%O2/He for an hour at 750oC. After the pretreatment, the samples were heated first to 400 oC, and then to 800 oC under He (100 sccm). After isothermal hold at each

30 temperature, the samples were cooled down to room temperature prior to collecting spectra.

For reduction experiment, the samples were pretreated the same way as described above. After the pretreatment, samples were reduced with 3% H2/He

(100 sccm) for 30 min at 400 oC, 600 oC, and 800 oC. After isothermal hold at each temperature, the flow was switched to He (100 sccm/min) and cooled down to room temperature, then spectra were collected. The samples reduced at 800 oC were re-oxidized in air (100 sccm/min) at 800 oC for 30 min and cooled down under same flow to collect the spectra to observe reoxidation. The analysis and fitting of the collected data were performed using WinXAS software [72]

3.3 Catalytic activity testing

The catalytic activity testing over perovskite materials were conducted on an equal surface area (0.1 m2) basis. The samples were loaded on top of quartz wool in a fixed-bed flow reactor. Prior to the experiment, perovskite samples

o were pretreated in 10%O2/He (50 sccm/min) at 750 C for 30 min while Ni-YSZ catalyst was pretreated in 10% H2/He (50 sccm/min) at 700°C for an hour. All catalysts were cooled down under the same flow. The feed percentages for partial methane oxidation experiment were CH4/O2/N2/H2 =4.5/ 4.5/ 9/ 82 and 4.5/

9/ 9/ 77.5 for total methane oxidation studies, respectively. Total flow of both experiments were set as 55 sccm/min. Nitrogen in the feed stream was used as

31 the internal standard to adjust volumetric flow rate change due to the stoichiometry of the reactions. For H2S poisoning studies, the N2 was replaced with 500 ppm H2S/N2 to feed 50 ppm of H2S out of total stream while maintaining all other feeds. The reactor effluent was analyzed using a Gas Chromatograph

(Shimadzu 2014) equipped with a pulse discharge helium ionization detector

(PDHID), a flame ionization detector (FID), and a flame photometric detector

(FPD). Columns were separated using He as a carrier gas with two columns:

Supelco Q Plot (30m x 053mm, fused silica capillary column) and CarboxenTM

1010 Plot (30m x 0.53mm, fused silica capillary column). Prior to data collection at each temperature, 30 min of hold took place to reach steady state.

Conversions and yields are defined as follows:

moles of CH4 converted %CH4conversion = ( )!100 moles of CH4 in the feed

moles of O2 converted %O2conversion = ( )!100 moles of O2 in the feed

moles of CO2 formed %CO2 yield = ( )!100 moles of CH4 in the feed

3.4 Electrochemical activity testing

3.4.1 Button cell preparation

A commercial ScZ tape (ESL ElectroScience) was used to make an electrolyte supported SOFC cell. A circular disk (1.4 cm dia.) was punched form the electrolyte tape and sintered at 1400oC followed by isothermal hold for 2 hrs for densification. For testing cerium-doped perovskite as a cathode, NiO-YSZ

(yittria-stabilized zirconia) from Nextech Materials was used as an anode. The powders were mixed with α-terpineol (SAIC), ethyl cellulose (Aldrich), toluene, and ethanol in order to prepare slurries. The slurries were stirred with a magnetic stirrer for 10 hrs to achieve a homogeneous mixture. Anode slurries were brush painted onto ScZ disk with a controlled area of ~0.15 cm2 and fired at 1200 oC for

5hrs. The cathode slurries were pasted after the anode side was sintered. For experiments where cerium-doped perovskite was used as an anode, LSM

(Nextech Materials) was used as a cathode. The slurries were made by mixing the power with α-terpineol (SAIC), ethyl cellulose (Aldrich), and diethylene glycol dibutyl ether (Aldrich). The anode slurries were first screen printed (Aremco

Products, Inc.) onto ScZ electrolyte and fired at 1200 oC for 5hrs. The cathode slurries were also screen printed after the anode side was sintered and were fired at the same conditions as the anode. The area of both anode and cathode

33 sides were controlled with ~0.15 cm2. Ag current collectors were pasted on both electrodes using Ag paste.

3.4.2 Current-Voltage and impedance measurements

The polarization curve was obtained at multiple temperatures (700 oC, 750 oC, and 800 oC) using the SP-150 Potentiostat/Galvanostat (Biologic Science

Instruments) with Electrochemical Impedance spectroscopy (EIS) capability.

Pure H2 (80 sccm/min) was introduced at the anode and air was used on the cathode side. Prior to the measurements at each temperature, stabilization of the polarization curve was performed by a fast scan of the voltage from the Open

Circuit Voltage (OCV) to 0.1 V at a scan rate of 40 mV/s. The data were then acquired from OCV to 0.1 V at a scan rate of 10 mV/s. The AC impedance spectra were collected under the same operating conditions as polarization curves using the same instrument over a frequency range of 0.1 Hz to 600 kHz at

OCV and 0.6V.

3.4.3 Chronoamperometry

Chronoamperometry for the prepared button cells were performed using the SP-150 Potentiostat/Galvanostat (BioLogic Science Instruments). For durability testing, 90% H2/N2 (total 80 sccm/min) was flowed to the anode side of

34 the button cells while air was introduced to the cathode side. The data was collected for 10 hrs at 0.6V. For H2S poisoning test, the flow was the same as durability testing for the first hour, then 50 ppm of H2S was introduced to the anode side for an additional 9 hours.

CHAPTER 4.

Ce-doped strontium cobalt ferrite perovskites as cathode catalysts

for solid oxide fuel cells: Effect of dopant concentration

4.1 Motivation

Perovskite-type materials are mixed metal oxides with the general formula

ABO3 where the A ions can be rare earth metals which occupy dodecahedral sites, while the B ions can be transition metals which occupy octahedral sites.

The broad diversity of the metallic elements that can form stable perovskite-type oxides can be applied in the design of new high performance catalysts for several applications. These materials are widely studied for their application as oxygen sensors and oxygen separation membranes, due to their ionic conductive properties[11, 25]. They have also been studied as automotive exhaust catalysts where both total and partial oxidation catalytic activity can be utilized[9, 73]. Their mixed ionic and electronic conductive properties have made perovskite oxides one of the most promising cathode materials for solid oxide fuel cells (SOFCs) [2,

6].

The principal role of the SOFC cathode is to form oxide ions by the reduction of oxygen. The generated oxide ions travel through the electrolyte to the anode, where they undergo oxidation to produce electricity and water.

Therefore, the performance of the cathode is limited by the surface oxygen exchange kinetics of the oxygen reduction reaction (ORR) as well as the bulk oxide ion mobility[74]. These properties are expected to depend upon the surface and bulk structure, where oxygen vacancies provide the active sites for the ORR and the pathway of oxide ions to travel by random ‘hopping’ through the vacancies. An oxygen vacancy is formed by a charge imbalance in the material.

Substituting the normally trivalent A-site with a divalent or tetravalent cation creates those charge imbalances in the perovskite. B-site transition metals also impact the oxygen vacancy formation. Co, Fe and Mn-based perovskites were studied and the oxygen vacancy formation was found to increase in the order of

Co, Fe and Mn, respectively[75-77].

Our previous studies have focused on the effect of A- and B-site dopant concentration in LaFeO3[28-30, 78, 79]. Nonlinear dependence of oxygen vacancy generation and oxygen exchange kinetics on Co loading was found due to an electronic structure transition at high Co concentrations. Additional doping of the B-site with aliovalent cations (e.g. Zn, Ni, and Cu) increased the oxygen vacancy creation in the order of Zn > Ni > Cu. Increased Sr loading enhanced oxygen vacancy generation, however the mismatch between the yttria-stabilized

37 zirconia (YSZ) electrolyte and the catalysts also increased due to the thermal expansion coefficient (TEC) incompatibility between the components.

The present work examines the effect of Ce doping in SrCo0.2Fe0.8O3-δ

(SCF) materials. Mixed conductive composites close to the brown-millerite stoichiometry (e.g. Sr(CoFe)O2.5) exhibit high oxygen permeability at high temperatures (T >800oC), but due to their structural changes at intermediate temperatures (650oC-750oC), the oxide ion conductivity decreases. However, incorporating a small amount of CeO2 into the framework could promote increased stability of the structure while maintaining high oxide ion mobility[62].

This work investigates the oxygen vacancy generation, oxygen mobility and thermal compatibility with the electrolyte as well as actual SOFC button cell performance with different Ce loadings in the A-site.

4.2 Results and Discussion

4.2.1 Bulk structure properties

The bulk structure of Sr1-xCexCo0.2Fe0.8O3-δ (x=0, 0.1, 0.15, 0.2 and 0.3) catalysts under ambient conditions was determined using X-ray diffraction. For simplicity, an abbreviation will be used for all cerium-doped samples and the notation is listed in Table 4.1. The X-ray diffraction patterns of the Ce-doped

samples are shown in Fig. 4.1. The diffraction pattern for Sr Co0.2Fe0.8O3-δ (SCF)

catalyst is also shown as a reference. As shown in Fig. 4.1, XRD patterns

confirm the presence of a single-phase perovskite structure in samples

containing cerium doping levels of 15%, and below, in the A-site. The diffraction

patterns show a single, sharp peak at 40o (111), which is the characteristic

diffraction line associated with a cubic perovskite structure. The tolerance factor,

“t”, is used to predict the perovskite structure formation with the range of 0.8-1,

where rA, rB, and rO are the ionic radii of A, B, and the oxide ion, respectively.

(r + r ) t = A O 2(r r ) B + O

As the value of “t” approaches 1, the crystal structure is expected to be ! closer to the ideal cubic symmetry[8]. The calculated tolerance factors for all the cerium-doped perovskites were close to unity, indicating the presence of the

cubic structure, which is also verified from the XRD data and is shown in Table

4.2. As the cerium doping level exceeds 15%, a small peak due to unreacted

CeO2 appears and its intensity increases with the addition of cerium. There have

been earlier reports where perovskites with LaMO3 (M: Co or Fe) formulas were

modified with cerium doping where the maximum amount of cerium that could be

doped into the A-site was reported to be between 5 and 10%[80-82]. In this case,

having strontium as the primary cation, which brings a different charge balance

and more similar ionic sizes, appears to allow a higher dopant level incorporation 39 in the A- site. Overall, the synthesis procedure was successful in introducing cerium into the catalyst bulk without affecting the perovskite structure.

The high temperature bulk structure of the perovskite was also investigated via in-situ XRD. The crystal structure is an important factor determining the oxygen activation and mobility in these materials. Our previous studies with cobalt-doped lanthanum ferrite samples, including those with a second transition metal such as Zn in the B-site, showed rhombohedral structure at room temperature and a phase transition from rhombohedral to cubic with an increase in temperature[28, 30]. Among these different phases, the cubic phase is known to have more symmetry than the rhombohedral phase, leading to an increased random distribution of the vacancies resulting in higher oxide ion mobility. The diffraction patterns for the main cubic Miller indices (111) are shown in Fig. 4.2. For all cerium-doped perovskites studied as well as SCF, no phase transition was observed and the cubic structure remained intact with a rise in temperature. This was confirmed based on the presence of a single sharp peak.

With increasing temperature, the peak in all the samples became sharper and shifted to lower 2-θ values indicating thermal expansion. The unit cell length is expected to increase linearly with temperature based on the concept of linear thermal expansion. Unit cell lengths were calculated using the Bragg’s formula for the Miller indices (110) from the XRD patterns shown in Fig. 4.3. Calculations were repeated using (200), (220) and (211) diffraction lines and the average percent error was found to be around 0.4%. As seen from the slopes in the

40 figure, thermal expansions of these materials appear to be constant over a wide temperature range.

Table 4.3 presents the Thermal Expansion Coefficient (TEC) of each sample and compares them to the TECs of potential SOFC electrolytes, which were calculated using the same technique. In view of the long-term operation of

SOFCs, thermal compatibility between the electrodes and the electrolyte is vital because a large difference will generate stresses within the cell, eventually leading to cracking or degradation of the electrode[5]. As seen from the table, the

TEC of cerium-free SCF sample is the highest and the TECs decrease with increasing cerium doping level and get closer to the TEC of the GDC electrolyte.

This trend may be attributed to the chemical expansion in addition to the thermal expansion of the samples. With an increase in temperature, transition metals in the B-site of perovskite oxides may be chemically reduced, ultimately generating oxygen vacancies [83]. The change in oxidation states of the transition metal (e.g.

+4 to +3) induces a change in the lattice volume while maintaining its phase. In the case of the cerium-doped samples, some of the B-site metals will be in +3 state, hence reducing the number of transition metal atoms with the +4 oxidation state. This, in turn, will decrease the chemical expansion that could take place due to the reduction of transition metals with +4 oxidation state at higher temperature.

The large thermal incompatibility observed between the electrolyte and

SCF led us to conclude that cerium-free SCF samples do not offer any potential

41 to be used as cathode materials and hence, the remaining part of the results focuses only on Ce-doped catalysts.

Table 4.1 Notation used for cerium-doped perovskites

Formula Abbreviation

Sr0.9Ce0.1Co0.2Fe0.8O3-δ SCeCF10

Sr0.85Ce0.15Co0.2Fe0.8O3-δ SCeCF15

Sr0.8Ce0.2Co0.2Fe0.8O3-δ SCeCF20

La0.6Sr0.4Co0.2Fe0.8O3-δ LSCF6428

!!#" '#()*!# !" $##" !!!" $!!" $$#" %!#" !" !"

!"#$%&#

Figure 4.1: XRD patterns of cerium-doped perovskites

Table 4.2: Calculated tolerance factor for cerium-doped perovskites

τ Structure

SCeCF10 1.02 Cubic

SCeCF15 1.019 Cubic

SCeCF20 1.018 Cubic

LSCF6428 0.975 Other type

of tilting

!"$''(%%%)' !"#"$%&''(%%%)' !"#"$%*''(%%%)' !"#"$+&''(%%%)'

-"' -"' -"' -"' -"' -"' -"' -"'

./''''''0&''''''0%''''''0+' ./''''''0&''''''0%''''''0+' ./''''''0&''''''0%''''''0+' ./''''''0&''''''0%''''''0+'

+,'(-)'

Figure 4.2. Diffraction patterns of cerium doped perovskite as a function of temperature in air

Figure 4.3: Unit cell parameters for cerium-doped perovskites as a function of temperature in air

Table 4.3. Thermal expansion coefficient values calculated from in situ XRD patterns

TEC (ppm/˚C)

SCF 25

SCeCF10 20

SCeCF15 18

SCeCF20 16

Ceria

(GDC) 13

Zirconia

(YSZ) 9

4.2.2Temperature-programmed desorption of oxygen (O2-TPD)

The generation of oxygen vacancies is an important parameter for determining the oxygen reduction reaction (ORR) activity of a catalyst. Thermo- gravimetric analysis, which was performed in an inert atmosphere, was used to investigate the evolution of oxygen from the prepared catalysts as a function of temperature (Fig. 4.4). All of the Ce-doped catalysts exhibit a higher oxygen vacancy content compared to that of the LSCF-6428. The oxygen vacancy generation was found to be the highest for the sample with the lowest Ce loading level (SCeCF10). This can be explained by the fact that the amount of tetravalent transition metals in the B site is highest for the SCeCF10 and decreases with increasing Ce loading levels. Since Co and Fe are less stable in the +4 oxidation state, at higher temperatures, they partially convert to a +3 oxidation state and the resulting charge imbalance is compensated by increased oxygen vacancies.

Out of the three cerium-doped samples, the amount of oxygen that evolved for

15%- and 20%- cerium-doped catalysts was observed to be very similar to each other. From the ex-situ XRD data, the maximum amount of cerium that could be doped in the catalysts was found to be between 15% and 20%. Cerium doping at higher levels cannot be incorporated into the A-site and leads to the formation of a separate CeO2 phase. If the same amount of cerium was doped in the A-site for both SCeCF15 and SCeCF20 perovskites, this would result in a similar level of oxygen vacancy generation for these two catalysts.

The temperature-programmed desorption of O2 was further examined using mass spectrometry. The O2-TPD profiles of the cerium-doped catalysts showed two distinct and temperature-dependent oxygen evolution peaks, a low temperature one at ~450oC and a high temperature one above 450oC, indicating that the oxygen species may exist in different matrices in the sample. Fig. 4.5a shows the O2-TPD profiles of the cerium-doped catalysts under an inert atmosphere. The experiments were performed with a mass spectrometer in selected ion mode to trace the relevant ions. After the appropriate pretreatment of the samples, the amount of oxygen evolved was quantified using the m/z= 32 signal and is shown in Fig. 4.5b. Generally, two types of oxygen evolving from perovskites have been reported in the literature. As outlined by Seiyama earlier

[84], the perovskite structures with A or B site substitution form many lattice defects and these defects are thought to play a role in increasing the oxygen

“absorption” characteristics of these materials. “Absorbed” oxygen in these materials, which is weakly bound to the cations behaves differently from oxygen which is part of the crystal lattice, and is more readily accessible. In the TPD profiles, the low temperature peak observed below 450oC (α-oxygen), is thought to be due to the evolution of absorbed oxygen present in the materials as well as surface-adsorbed oxygen. The SCeCF10 catalyst showed the highest amount of desorbed α-oxygen followed by SCeCF15 and SCeCF20, which show equal amounts of α-oxygen desorption. If the A-site doping determines the amount of

α-oxygen desorbed, this result is consistent with our earlier findings which

50 showed that the maximum amount of Ce that could be incorporated in the A site may be around 15%. If this is the case, SCeCF15 and SCeCF20 would both have the same level of A-site substitution. The amount of α-oxygen evolved far exceeded the amount that could be present in a surface monolayer (4.0 µmol-

-2 O2m for a monolayer), indicating that most of the evolved α-oxygen that desorbed was associated with the bulk.

As seen in Figure 4.5, the Ce-doped catalysts exhibited a broad temperature range for the high-temperature oxygen evolution. This feature denoted as β-oxygen in the literature is thought to be associated with the partial reduction of the transition metal in the B-site and is linked to the anionic mobility in these materials [85-87].

Earlier reports from the literature as well as some of our own work have

shown the La0.6Sr0.4CoxFe1-xO3-δ (x = 0.2, 0.4, 0.6) catalysts to display a single β- oxygen desorption peak above 450oC [28, 29, 85, 87, 88] during oxygen TPD in an inert environment. The cerium-doped samples examined in this study, however, showed the evolution of an additional β-oxygen desorption feature above 800oC. This feature could possibly be due to the presence of cerium in the

A-site partially stabilizing the transition metals in the B-site. It is also conceivable that the β1 and β2 features stem from the two different cations (i.e., Co and Fe) occupying the B-site and showing different levels of propensity towards reduction.

It is also possible that they correspond to the reduction of tetravalent and trivalent cations, respectively. The fact that the amount of oxygen desorbing with the β1 51 feature decreases with increasing cerium doping is consistent with this possibility since cerium doping is expected to decrease the abundance of tetravalent cations to preserve the charge neutrality. The results in this study, however, do not allow a definitive assignment to the source of the β1 and β2 oxygen features.

The LSCF-6428 sample that was included for comparison showed the lowest oxygen desorption among all the samples, both in total and in the individual α and β features.

Figure 4.4 Temperature-programmed oxygen evolution in inert(He) environment measured using TGA

!"#$%&'#

Fig 4.5. (a) Temperature-programmed oxygen evolution in inert(He) environment monitored using mass spectrometry, (b) O2 desorbed calculated from the TPD profiles

4.2.3 Bulk properties using X-ray absorption fine structure (XAFS)

4.2.3.1 X-ray absorption near-edge spectroscopy (XANES) analysis

XAFS studies were done on the cerium-doped samples at both the Fe and

Co K-edges. These studies were conducted at multiple temperatures under inert environment in order to understand the valence state of transition metals and oxygen coordination with increasing temperature. The most common application of the XANES region is to use the shift of the edge position to determine the valence state of the element. The position of the edge is expected to shift as the valence state of either Fe or Co increases due to the strong bind of electrons to the transition metal’s nucleus, therefore, resulting in a shift of the absorption edge to higher binding energies[89]. Fig. 4.6a and b show the XANES spectra for

Fe and Co K-edge for different loadings of cerium into perovskite structure at ambient condition. Under these conditions there is no obvious change in the edge position for either the Fe or the Co edges (< 1eV) and this is confirmed with position of the maxima from the first derivative of XANE spectra shown in Fig

4.6c and Fig 4.6d. This indicates that the oxidation states of all the samples are similar for both Fe and Co at room temperature. The oxidation state of both Fe and Co should increase, from +3 to +4, as the strontium concentration increases in the A-site due to the charge neutrality within the structure. However, the edge position from XANES spectra for all samples shows less than 1 eV difference.

This could be the case where the charge imbalance is compensated by the creation of oxygen vacancies rather than inducing a higher oxidation state of transition metals at the B-site. This can be also confirmed by the O2-TPD data in the previous sections that α-oxygen, which is thought to be associated with absorbed oxygen, shows the most prominent desorption feature in SCeCF10 followed by SCeCF15 and SCeCF20 while LSCF6428 had the smallest feature.

XANES spectra were also collected at elevated temperatures for

SCeCF10, SCeCF15, and SCeCF20 and they are presented in Fig. 4.7, 4.8, and

4.9 respectively. As seen in Fig 4.7a the position of the Fe K-edge changes gradually with increasing temperature and the edge shifts to a lower binding energy of approximately 1 eV, indicating that Fe is reduced to some extent. The

Co K-edge showed more pronounced edge shifting than iron and is shown in Fig

4.7b and d. This trend is expected since cobalt is less stable than Iron resulting in a larger shift in edge position. The edge position of Co K-edge has shifted c.a. 1 eV upon increasing the temperature from 30oC to 400 oC while an additional 2 eV shift is observed upon a further temperature increase to 800 oC. This result indicates that further reduction of Co+3 to Co+2 takes place between 400 oC and

800 oC while from room temperature to 400 oC, Co undergoes reduction from a

+4 to +3 valency. Similar trends were also seen for both SCeCF15 and

SCeCF20, in which shifts in both the Fe and Co edges are observed with increasing temperature.

Figure 4.6. XANES analysis of cerium-doped perovskite samples at room temperature a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c)

First derivatives of Fe K-edge XANES spectra; d) First derivatives of Co K- edge XANES spectra.

Figure 4.7. XANES analysis of SCeCF10 with elevated temperature under inert a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c) First derivatives of Fe K-edge XANES spectra; d) First derivatives of Co K-edge

XANES spectra.

Figure 4.8. XANES analysis of SCeCF15 with elevated temperature under inert a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c) First derivatives of Fe K-edge XANES spectra; d) First derivatives of Co K-edge

XANES spectra.

Figure 4.9. XANES analysis of SCeCF20 with elevated temperature under inert a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c) First derivatives of Fe K-edge XANES spectra; d) First derivatives of Co K-edge

XANES spectra.

4.2.3.2 Extended X-ray absorption Fine Structure (EXAFS) analysis

In the EXAFS region, neighboring atoms can be probed by analyzing the scattered photoelectron wave where it creates interferences between the outgoing and scattered parts of photoelectron. The Fourier transformation of the

χ(k) k2 functions in R-space for all investigated samples at room temperature is shown in Fig 4.10a and b. R is the uncorrected radius of the nearest neighbor atoms and the phase shift is typically 0.5 Å while the intensities of the peaks can be qualitatively related to coordination number[90]. The k2 Fourier transforms of

Fe K-edge shows an intense peak at 1.37 Å and less intense peaks at 2.45 Å and 3.12 Å in Fig 4.10a. The first intense peak can be assigned to the first oxygen shell showing equal amount of oxygen is surrounding the Fe, as evidenced by similar FT magnitudes for all samples. The second and third peaks are more difficult to assign because the scattering path distances of Fe-Sr, Fe-Ce and Fe-O-Fe are similar. At room temperature the magnitude of the second and third peaks are similar for all samples that were investigated. This indicates the ordering of the atoms surrounding the Fe atom is the same in all cases. Fig 4.10b shows an intense peak at 1.32 Å, which can be assigned as the first oxygen shell surrounding Co while less intense peaks at 2.37 Å and 3.02 Å are more difficult to assign due to multiple scattering around Co atom. The magnitude of Co K- edge for all samples shows the same at room temperature as discussed above in the case of the Fe edge. As the temperature was increased the intensities of the

61 first peak decreased for both Fe and Co edge indicating the oxygen is released from the structure and is shown in Fig 4.11. The Fe regions show a relatively small decrease in the intensity of the first peak while the decrease in intensity is much more apparent in the case of the Co regions, especially between 400oC and 800oC. Fourier transform data for SCeCF15 and SCeCF20 are also shown in

Fig. 4.12 and Fig. 4.13. The SCeCF15 and SCeCF20 follow a similar trend to that of SCeCF10 showing a small decrease in the intensity of the Fe edge while the decrease in the intensity is more pronounced in the Co edge. This suggests that the evolution of oxygen from the perovskite structure takes place at the Co site. This is consistent with the results presented in the previous sections, which showed a large reduction of Co in the XANES spectra. The intensity of the third peaks for both Fe- and Co- edge were also observed to decrease with increasing temperature. The most plausible assignment for these peaks could be Fe-O-Fe and Co-O-Co as their ionic distances are 2.64 Å and 2.57 Å, respectively. Due to the scattering phase-shift in EXAFS the peak typically shifts by 0.5 Å[89].

Correcting for it leads to the expected Fe-O-Fe distance of 3.12 Å and Co-O-O distance of 3.02 Å. Other, however, possible peak assignments cannot be excluded. As the temperature is increased, the evolution of oxygen could alter the angle of the Fe-O-Fe and Co-O-Co bond resulting in low intensities[90-93].

Figure 4.10. EXAFS analysis of cerium-doped samples at room temperature for k2-weighted Fourier Transform of EXAFS function in magnitude: a) Fe K- edge; b) Co K-edge. 63

Figure 4.11. EXAFS analysis of SCeCF10 at different temperatures for k2- weighted Fourier Transform of EXAFS function in magnitude: a) Fe K-edge; b) Co K-edge.

Figure 4.12. EXAFS analysis of SCeCF15 at different temperatures for k2- weighted Fourier Transform of EXAFS function in magnitude: a) Fe K-edge; b) Co K-edge.

Figure 4.13. EXAFS analysis of SCeCF20 at different temperatures for k2- weighted Fourier Transform of EXAFS function in magnitude: a) Fe K-edge; b) Co K-edge.

4.2.4 CO2-temperature programmed oxidation (CO2-TPO)

The ORR at an SOFC cathode involves multiple steps: 1) adsorption of diatomic oxygen species on to the surface active sites, 2) dissociation to produce monoatomic species and 3) interfacial transport of the surface oxygen to the catalyst bulk. CO2 temperature-programmed oxidation can be an indicator of the ORR activity of SOFC cathode catalysts since the steps involved are similar to those of the ORR, namely dissociative adsorption of CO2 followed by filling the surface oxygen vacancies and/or transportation of monoatomic oxygen into the bulk lattice. Therefore, the activity of the catalysts for CO2 adsorption and dissociation can provide information, at least qualitatively, of the ease with which the surface oxygen can move into the perovskite lattice

[94, 95]. Fig. 4.14 illustrates the CO (m/z= 28) signal from the CO2-TPO experiments for cerium-doped catalysts. Also included in the figure is the CO signal for LSCF-6428 catalyst for comparison. From Fig. 6, it is apparent that there are two distinct features in the CO2-TPO profile, a sharp peak at around

650oC, and a broad shoulder at higher temperatures over 850oC. These features seen in the CO evolution profile are possibly associated with different sites and vacancies on the catalyst surface as well as in the bulk. At high temperatures (T>

850oC), especially, the SCeCF10 catalyst showed a relatively smaller CO evolution when compared to other catalysts, indicating the presence of vacant sites that are harder to re-oxidize. A large amount of CO evolution is indicative of

67 an increased activity of CO2 dissociation along with an increase in the amount of surface oxygen transported to the bulk. Although SCeCF10 shows slower oxygen transportation at high temperature, the overall peak area in terms of the total amount of CO evolution shows the trend SCeCF15 > SCeCF10 > SCeCF20 >

LSCF-6428 and is presented in Table 4.2. These results indicate that the oxygen transport rate increases with increasing cerium concentration in the A-site of the catalysts. The reason that SCeCF20 had a slower oxygen transportation rate compared to the other cerium-doped catalysts could possibly be caused by the presence of a secondary phase (CeO2).

!"#$%&'#

Figure 4.14. CO2-TPO profiles for LSCF6428; SCeCF10; SCeCF15; SCeCF20

Table 2. Total amount of CO formation during CO2-TPO

µmol/g LSCF 6428 SCeCF 10 SCeCF 15 SCeCF 20

Total 1893 1917 2128 1902

4.2.5 Button cell performance

Electrolyte-supported button cells were fabricated using three different levels of cerium doping on the cathode, SCeCF10, SCeCF15, SCeCF20. The current (I) versus voltage (V) and power density profiles for these button cells

o o o operating with H2 at 700 C , 750 C, and 800 C are shown in Fig. 4.15. The data obtained from LSCF-6428 are also included for comparison. From Figure 4.15a it can be seen that at 700 oC, LSCF6428 has lower activity when compared to the cerium doped samples. It is also apparent at this temperature, that samples with lower concentrations of cerium show a better performance. This performance trend observed could be related to the α- and β1- oxygen evolution as discussed in the previous section. The amount of α- and β1- oxygen evolved from the cerium-doped catalysts decreases as the cerium concentration increases. Since the performance of the cell is closely linked to the ionic mobility of oxygen, the performance of the cell decreased with increasing cerium concentration.

However, as the temperature is raised to 750 oC, this trend is no longer observed. The V-I curves for the button cells, seen in Fig. 4.15 b and c, show that the best performance was achieved by the SCeCF15 sample, followed by

SCeCF10, SCeCF20, and LSCF6428. As seen from the CO2-TPO experiments,

SCeCF10 showed a slower rate of oxygen transport at higher temperatures.

This could explain the decreased performance of this catalyst compared to

SCeCF15, especially at those temperatures. Overall, SCeCF15 and SCeCF10

70 samples achieved the best performance during button cell tests. These are the samples that showed the highest oxygen vacancy formation and the highest CO2 reduction activities, as discussed in the O2-TPD and CO2-TPO results section above. A high propensity for oxygen vacancy creation signals a higher ability for oxide ion conduction. CO2-TPO provides an insight into the rates of oxygen dissociation and transportation of surface oxygen to the bulk. Although the samples with intermediate cerium loading levels perform the best (and significantly better than LSCF), the sample with a higher loading level (SCeCF20) exhibits the poorest performance. One possible explanation is the existence of a CeO2 phase, segregating out of the perovskite structure. This phase is not likely to contribute to the ORR activity. On the contrary, it is possible that the

CeO2 phase hinders ionic conduction in the material. The very low CO formation data from CO2-TPO experiment signals a slow bulk transport of oxygen in this sample.

The AC impedance spectra for each cell at the open circuit voltage (OCV) are shown in Fig. 4.16. The spectra show that the decrease in performance was followed by an increase in the area specific resistance (ASR) of the cell, which corresponds to the overall width of the arcs. As identified by X-ray diffraction, the presence of a secondary phase (CeO2) with low electrical and/or ionic conductivity, in SCeCF20 is seen to increase the overall impedance of the button cell and is consistent with the SCeCF20 exhibiting the poorest performance among all the catalysts tested

Figure 4.15. Voltage and power density vs. current density at (a)700oC;

(b)750oC; (c)800 oC.

Figure 4.16. OCV electrochemical impedance spectra at (a)700oC; (b)750oC;

4.3 Conclusions

The structural properties of cerium-doped perovskite-type oxides with the

formula Sr1-xCexCo0.2Fe0.8O3-δ for x=0.10, 0.15 and 0.20 have been examined under ambient conditions and at elevated temperatures. All cerium-doped perovskites showed a cubic structure with no observed structural changes/distortions at elevated temperatures. CeO2 was seen to form a separate phase for dopant concentrations greater than 15%. TEC values calculated from in-situ XRD data verified that the Ce-free SrCoFeO3 material to have a much higher thermal expansion coefficient compared to the most commonly used

SOFC electrolytes. With increasing cerium dopant concentration, however, thermal compatibility with GDC increased, reaching 16 ppm/°C for SCeCF20.

As shown in the XAFS data, the oxidation state of Co and Fe was found to be similar at room temperature for all the cerium-doped. The charge imbalance within the matrix increases with decrease in the Ce-loading. To compensate for this charge imbalance, oxygen vacancies are created and these initial vacancies can be occupied by weakly bound absorbed oxygen (α-oxygen). Since the number of vacancies created in SCeCF10 is the largest, it showed the largest evolution of α-oxygen followed by SCeCF15 and SCeCF20. The oxygen release at high temperature (> 400oC) is mostly caused by the reduction of Co rather than that of Fe, as was concluded from the EXAFS studies.

Although the lowest Ce doping led to the highest oxygen vacancy generation, as seen by the oxygen TPD experiments, this did not directly translate to better button cell performance. The best performance was seen in the intermediate doping levels (15%), which also showed a higher ability for CO2 dissociation and a higher intra-facial transport of oxygen, suggesting that CO2

TPO may serve as a good probe for SOFC performance.

The highest loading level studied (20%) gave the poorest button cell performance as well as the highest area specific resistance as determined by impedance measurements. The presence of a secondary phase with low electrical and/or ionic conductivity, CeO2, in SCeCF20 may be a factor leading to increased overall impedance and poor button cell performance. This study shows that SrCoFeO3 perovskites with intermediate levels of Ce loading may give optimum levels of oxygen vacancy generation and oxide ion transport while providing TECs comparable to those of the electrolytes and hence may have potential as SOFC cathode materials.

* Information in this chapter is adapted/taken from Choi, H., Fuller, A., Davis, J., Wielgus,

C., and Ozkan U.S., Applied Catalysis B, 127, 336-341 (2012)

CHAPTER 5.

Effect of Ce doping on the performance and stability of strontium cobalt

ferrite perovskites as SOFC anode catalysts

5.1 Motivation

Solid oxide fuel cells (SOFCs) are promising energy conversion devices with high prospects of replacing conventional combustion units for the generation of electric power. Their ability to use a variety of fuels including bio-gas opens up the path for power generation from renewable sources. SOFCs comprise of three main components, namely the cathode where the oxidant is reduced to generate oxide ions using electrons, an oxide ion-conducting electrolyte and the anode where the fuel combines with the oxide ions to get oxidized and release electrons, which travel through the external circuit to the cathode. Since the solid electrolyte is able to conduct oxide ions formed at the cathode towards the anode, the choice of fuel is extended beyond hydrogen to include hydrocarbons.

The high operating temperature (800oC ~1000oC) allows the use of exhaust gases to be used in the cogeneration of power and heat to help SOFCs achieve efficiencies of almost 80%[96]. In addition, eliminating complex and expensive 78 external reformers in favor of internal reformers, which are used to reform the hydrocarbon fuel to hydrogen can reduce operating costs [2, 4]. However, current

SOFCs suffer from various challenges, which hinder its commercialization.

Issues associated with the anode are among these challenges.

The anode electro-catalyst is exposed to a harsh reducing environment and very high temperatures. Therefore, it must be chemically stable to endure the anode conditions and have thermal and mechanical compatibility with the electrolyte so as not to separate from it or react with it at elevated temperatures.

The anode must also be electronically conductive to pump out the electrons that are generated via the electrochemical oxidation of hydrocarbon fuel. The current state of the art anode catalyst is Ni-YSZ (Yttria-stabilized Zirconia) cermet where

Ni provides the catalytic active sites for fuel oxidation and is also an electronic conductor while YSZ serves to provide thermal compatibility with the electrolyte and also act as an oxide ion conductor [97, 98]. When a carbonaceous fuel

(syngas derived from coal, biomass, natural gas, etc.) is used, it typically contains sulfur species and even ppm levels of H2S can cause significant performance loss and degradation during SOFC operation due to the high vulnerability of Ni towards coking and sulfur poisoning [99]. Typical deactivation is attributed to (1) physical adsorption of H2S on Ni; (2) dissociative chemisorption of sulfur; (3) and sulfidation of Ni to produce nickel sulfide [36, 39,

44]. Therefore, a major challenge is to develop effective and economic anode electro-catalysts which do not deactivate due to sulfur.

Oxide-based mixed ionic and electronic conductors (MIECs) have been receiving increasing interest for their potential use as SOFC anodes. MIECs have an expanded reaction area, which is not limited to just the triple-phase boundary

(TPB) but to the entire bulk. They also have good thermal and mechanical compatibility with the electrolyte and a relatively high sulfur tolerance as compared to the conventional Ni-cermet catalysts [56, 100, 101]. Perovskite-type oxides of the type ABO3 are one such class of mixed conducting materials that have the potential of solving the problems associated with Ni-based cermets. The

ABO3 structure (A= rare earth metal, B= transition metal) is such that it allows a diverse number of elements to form stable perovskites whose properties can be tailored so that they are catalytically active in oxidation reactions [9, 73, 102]. The

present work investigates the effect of cerium doping in SrCo0.2Fe0.8O3-δ materials

as potential anode catalysts. Sr(CoFe)O3-δ exhibits high oxygen mobility, but due to its instability at higher temperatures, it does not meet the SOFC anode requirements. However, incorporating a small amount of cerium into the structure

could increase the stability of Sr(CoFe)O3-δ and can effectively act as an H2S absorbent, as reported in the literature [32, 36, 62]. In addition, Sr generally acts as sulfur guard to form sulphates which could improve the anode resistance towards sulfur poisoning that usually plagues Ni-based cermet anodes [103]. The current work focuses on the stability of the cerium-doped perovskites Sr1-

xCexCo0.2Fe0.8O3-δ under SOFC anode conditions and investigates the effect of

H2S poisoning on its methane oxidation activity along with its performance in an actual button cell.

5.2 Results and Discussion

5.2.1 X-Ray Diffraction (XRD)

The in-situ high temperature XRD (HTXRD) experiments performed under reducing conditions were aimed to understand the structural changes that might occur in the Ce-doped catalysts under simulated SOFC anode operating conditions. The data shown in Fig. 5.1 confirms the presence of a single-phase cubic perovskite (space group Pm-3m) at room temperature for all four samples,

with minor CeO2 peaks being observed for Sr0.8Ce0.2Co0.2Fe0.8O3-δ (SCeCF20).

The diffraction pattern of SrCo0.2Fe0.8O3-δ (SCF) is included as a baseline comparison to understand the changes that might occur due to cerium substitution at the A-site. As the temperature was raised, peak splittings were observed for the (110) and (211) diffraction lines for SCF and

o o Sr0.9Ce0.1Co0.2Fe0.8O3-δ (SCeCF10) samples beginning at 300 C and 400 C, respectively. There was no peak splitting for SCeCF15 and SCeCF20 at those intermediate temperatures. The splitting of the (110) peak into a triplet at 300 and

400 oC for SCF and SCeCF10, respectively, could be an indication of the brownmillerite structure being present with a slight tetragonal distortion (also inferred from the asymmetric (211) peaks) [104, 105]. SCeCF15 and SCeCF20 81 maintained their cubic structure at those intermediate temperatures. The deviation from ideal cubic perovskite could be explained in terms of the oxygen deficient non-stoichiometry occurring as a result of the changes in temperature, the presence of a reducing atmosphere and additional cerium doping at the A- site [106]. From our previous O2-TPD experiments carried out in an inert He atmosphere [90], oxygen evolution from the cerium-doped perovskites begins at

300 oC. This phenomenon could be enhanced in a reducing environment (i.e. low oxygen partial pressure) and could explain why SCF and SCeCF10 begin to structurally transform at those temperatures as the oxygen evolution causes vacancies to form in the perovskite structure. This loss of oxygen manifests itself in the breakdown of the cubic structure observed in the respective diffraction patterns of SCF and SCeCF10. The maintained cubic perovskite structure of

SCeCF15 and SCeCF20 up to 800 oC suggests that increased Ce-loading at the

A-site is providing some level of stability against oxygen loss and the accompanying structural changes it brings [90, 104].

The presence of oxygen non-stoichiometry can often lead to vacancy ordering in the perovskite [106, 107]. It is likely that such ordering might be occurring in all samples causing the brownmillerite structure Sr2CoFeO5 to form, albeit at different temperatures, as can be inferred from the splitting of the (110), (111) and (211) peaks into doublets [25, 108]. The transition temperature into the brownmillerite structure increases with increasing Ce loading, with all three

82 samples (x=0.0, 0.10, 0.15) exhibiting well-defined doublets for (110) line at

800°C. For the SCeCF20 sample, the transition is still not complete at 800°C.

In the brownmillerite-type structure, alternate BO6 (B= transition metal) octahedra are missing their corner-shared oxygen [109] leading to the formation of a “supercell” that is bigger than its cubic counterpart[106]. This vacancy ordering causes reduced oxygen flux and may not be good for oxygen transport in the SOFC anode [110].

All the samples that were reduced at 800oC were re-oxidized at the same temperature in air. This experiment was conducted to examine the possible recovery of the cubic perovskite structure after it is exposed to anodic conditions.

As the XRD patterns show in Fig. 5.2, all the perovskite structures have recovered their original single-phase cubic structure, including SCF and

SCeCF10 samples, which showed the brownmillerite phase transition at even intermediate temperatures. This result suggests that if the oxygen flux from the cathode side is high enough to compensate for the loss of oxygen from the anode catalyst, the Ce-doped perovskites could possibly hold their original cubic structure and avert a phase transition. The presence of a separate ceria phase represented by (111), (220) and (311) diffraction lines, is apparent in SCeCF20 after going through the reduction-reoxidation cycle.

Figure 5.1. In-situ XRD during TPR in 5%H2/He for (a) SCF; (b) SCeCF10; (c)

SCeCF15; (d) SCeCF20

Figure 5.2. XRD for Sr1-xCexCo0.2Fe0.8O3-δ (x= 0.0, 0.10, 0.15, and 0.20) catalysts after re-oxidizing at 800oC for 30 min in air

5.2.2 Bulk properties using X-ray absorption fine structure (XAFS)

5.2.2.1 X-ray absorption near-edge spectroscopy (XANES) analysis

The XANES studies for Sr1-xCexCo0.2Fe0.8O3-δ (x= 0, 0.1, 0.15, and 0.2) were performed to investigate the bulk properties of Fe and Co at elevated temperatures and a reducing environment. The XANES spectra of the Fe and Co

K-edge, as well as the first derivatives of both edges for all the samples are shown in Fig 5.3-5.6. As can be seen from the figures, the general trends of all the samples are similar. From Fig 5.3.a and c, the position of the edge shifted to a lower binding energy, indicating that the Fe is being reduced with increasing temperature in a H2/He atmosphere. Fe was slightly reduced between room temperature and 400oC and a slight change was observed upon increasing the temperature to 600 oC, while a significant reduction of Fe was seen when the temperature was raised to 800oC. At 800 oC, an additional peak at a higher binding energy (7.121 keV) was observed in addition to the metallic Fe peak

(7.110 keV), indicating the possibility of a multivalent Fe being present. An interesting feature was observed when the reduced Fe was re-oxidized with air, which is closer to Fe3O4 (7.125 keV) indicating the sample was re-oxidized. An additional peak was also observed at a lower binding energy (7.123 keV) than the main peak (7.125 keV) showing the possibility of multivalent Fe being present in the sample. The XANES spectra of Co K-edge show that most of the Co was

86 reduced in between room temperature and 400 oC, while a minor reduction was observed at 600 oC. At 800 oC significant reduction of Co was seen, similar to that observed in the Fe K-edge region. As seen from Fe K-edge spectra, multiple peaks were observed in the Co spectra at 800 oC for the re-oxidized sample showing multi valent Co could be also present.

Linear combination (LC) fitting was performed for both Fe and Co K-edge spectra to quantify the oxidation state of Co and Fe for Sr1-xCexCo0.2Fe0.8O3-δ (x=

0, 0.1, 0.15, and 0.2) samples. The data is shown in Fig 5.7. Metallic Fe, FeO,

Fe3O4, and Fe2O3 standards were used to fit the data for Fe K-edge while metallic Co, CoO, and Co3O4 standards were used for Co K-edge. The quantified oxidation states of both Fe and Co are shown in Fig 5.7a and b, respectively.

Initially Fe was in +3 oxidation state at room temperature for all the samples and reduced to lower oxidation state on raising the temperature. At 400 oC, SCeCF15 and SCeCF20 samples showed higher oxidation states than SCF and SCeCF10 which showed an approximate oxidation state of +2.5. This result could explain the tetragonal structure formation of SCF and SCeCF10 at the intermediate temperature (300 oC – 400 oC) in the in-situ XRD patterns and also indicate the threshold point for the transition from cubic phase to tetragonal phase. Reduction of the transition metals release oxygen the octahedral coordinated to them, resulting in oxygen vacancies. This leads to the formation of the tetragonal perovskite as a result of the missing corner-shared oxide ions[106, 107, 109].

Therefore, the lower oxidation state of transition metals in SCF and SCeCF10

87 could generate more oxygen vacancies than SCeCF15 and SCeCF20 resulting in a tetragonal phase transition from the original cubic phase. This relatively higher oxidation state of Fe remains up to 600oC and at 800oC oxidation state of

Fe in all the samples showed the same values. Two oxidation states of Fe were observed at both 800 oC and after re-oxidation. At 800 oC metallic Fe and Fe+2 was seen while +3 and +2 valence Fe were seen after they were re-oxidized.

Co transitioned from a +4 oxidation state at room temperature to a reduced

+3 state at 600 oC. SCF sample showed relatively lower oxidation state of Co than cerium-doped samples at 400 oC and 600 oC which could further strengthen the in-situ XRD data that SCF showed tetragonal phase transition at lowest temperature (300 oC). Metallic Co and Co+2 were seen at 800 oC and the oxidation state returned to +2 and +3 when the sample was re-oxidized.

Linear combination fitting of XANES spectra for both Fe and Co K-edge at

800oC with reference compounds using WinXAS software is present in Fig 5.8 and Fig 5.9. The XANES region of the samples were fitted by Fe, FeO standards for Fe K-edge and Co, CoO standards for the Co K-edge. The results show that the concentration ratio between metallic Fe to FeO was 7:3 for all samples and were independent of the cerium concentration. This ratio also holds for the Co K- edge.

Figure 5.3. XANES analysis of SCF with elevated temperature under reducing environment a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c) First derivatives of Fe K-edge XANES spectra; d) First derivatives of Co K-edge XANES spectra.

Figure 5.4. XANES analysis of SCeCF10 with elevated temperature under reducing environment a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c) First derivatives of Fe K-edge XANES spectra; d) First derivatives of Co K-edge XANES spectra.

Figure 5.5. XANES analysis of SCeCF15 with elevated temperature under reducing environment a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c) First derivatives of Fe K-edge XANES spectra; d) First derivatives of Co K-edge XANES spectra.

Figure 5.6. XANES analysis of SCeCF20 with elevated temperature under reducing environment a) Fe K-edge XANES spectra; b) Co K-edge XANES spectra; c) First derivatives of Fe K-edge XANES spectra; d) First derivatives of Co K-edge XANES spectra.

92 a)

Figure 5.7. Quantified oxidation state results from linear combination fittings for a) Fe; b) Co. 93

Figure 5.8. Result of linear combination XANES fitting at Fe K-edge with reference compound of a) SCF; c) SCeCF10; c) SCeCF15; d) SCeCF20

Figure 5.9. Result of linear combination XANES fitting at Co K-edge with reference compound of a) SCF; c) SCeCF10; c) SCeCF15; d) SCeCF20

5.2.2.2 Extended X-ray Absorption Fine Structure (EXAFS) analysis

The magnitude of k2-weighted fourier transforms of the Fe and Co K-edge

EXAFS spectra collected over Sr1-xCexCo0.2Fe0.8O3-δ (x= 0, 0.1, 0.15, and 0,2) catalysts under reducing environment at the elevated temperature. The spectra is given in Fig 5.10 and 5.11, respectively. All samples at the Fe K-edge show a peak at 1.37 Å which can be assigned to Fe-O scattering. As the temperature was raised the Fe-O peak moved to a slightly higher R value indicating that the

Fe is reduced and the intensity of the peak also decreased. From the EXAFS data, the intensity change among the sample was too small. Therefore, the oxidation of the sample as well as the oxygen coordination differences between

SCeCF15 and SCeCF20 vs. SCF and SCeCF15, which was shown in XANES analysis, cannot be differentiated. When the samples were at 800 oC, the most intense peak shifted to 2.12 Å (metallic Fe) indicating most of the samples were reduced to metallic Fe. After the re-oxidation peak at 2.12 Å shifted back to lower

R values where the Fe-O bonding is present.

The Co K-edge region had a peak at 1.32 Å which can be assigned to Co-O scattering. As the temperature was increased the peaks shifted to higher R values indicating the catalysts were reduced to a lower valence state. A large shift was observed between room temperature and 400 oC where Co+4 were reduced to Co+3 from XANES analysis. The intensity of the peak between the temperature region shows the significant intensity decrease, indicating the

96 oxygen vacancies were generated most significantly in that region. The temperature between 400 oC and 600 oC show minimum change of the intensities as well as the peak shifting. As the temperature went up to 800 oC metallic Co shows the most intensities and the peak position has decreased back to lower R values indicating the Co has re-oxidized.

Figure 5.10. EXAFS analysis over Fe K-edge at different temperatures for k2-weighted Fourier Transform of EXAFS function in magnitude: a) SCF; b)

SCeCF10; c) SCeCF15; d)SCeCF20.

Figure 5.11. EXAFS analysis over Co K-edge at different temperatures for k2-weighted Fourier Transform of EXAFS function in magnitude: a) SCF; b)

SCeCF10; c) SCeCF15; d) SCeCF20.

5.2.3 Steady-state methane oxidation

Sub-stoichiometric, steady state methane oxidation reaction was conducted

on Sr1-xCexCo0.2Fe0.8O3-δ (x = 0.10, 0.15, and 0.20) samples to study their potential as anode catalysts and examine the effect of cerium doping. Sub- stoichiometric oxygen in the feed was chosen in an effort to simulate the actual

SOFC environment where oxygen flow from the cathode would be limited. Ni-

YSZ cermet, which is the state-of-the art SOFC anode catalyst and a

La0.6Sr0.4Co0.2Fe0.8O3-δ (LSCF6428) perovksite catalyst, which showed considerable activity as documented in our previous publications [79, 111], were also used for comparison. The CH4 and O2 conversion data are presented for nine different temperatures in the temperature range of 400 oC to 800 oC in Fig.

5.12a-b. CO2 and H2O were the primary products of methane oxidation over all the catalysts, in addition to CO, which had a yield of less than 1%. The methane oxidation activity takes off for all catalysts at around 400 oC. This “light-off” temperature is also observed in the O2 conversion data. It can be seen from Fig.

5.12b that the complete conversion of oxygen was attained at 600 °C over the

LSCF6428 catalyst and at 700 °C over SCeCF10 and SCeCF15 catalysts. A

o reaction temperature of 750 C was needed to reach 100% O2 conversion over

SCeCF20. The catalytic activity of perovskite materials for oxidation reactions is a strong function of the oxidation state of transition metals at the B-site, which

100 can be tailored by doping lower oxidation state elements such as Sr at the A-site to improve the overall catalytic activity of the perovskite. In our case, the formation of higher valence state Co/Fe ions due to an increased concentration of divalent Sr at the A-site, may improve the oxidation catalytic activities at lower temperatures. However, the suppression of activity increases at elevated temperature with increasing the cerium concentration at the A-site. Ni-YSZ did not achieve complete conversion of oxygen even at 800°C .

After steady-state was reached at 800°C, 50 ppm of H2S was introduced to the feed mixture and isothermal time-on-stream data were recorded for 10 hours

(Figure 5.13). For Ni-YSZ, immediately after the introduction of 50 ppm of H2S, methane and oxygen conversion dropped by 50% and continued to decrease gradually. Perovskite catalysts are reported to have a higher tolerance to sulfur poisoning as compared to Ni-YSZ cermets [36], however the LSCF6428 perovskite started to deactivate after 6 hours of poisoning as seen from its decreased capability to convert methane and oxygen. However, the rate of deactivation was much slower compared to Ni-YSZ. Interestingly, the cerium- doped perovskites showed no deactivation with H2S and maintained the same activity level for the duration of the experiment.

In order to study the catalytic activity of Sr1-xCexCo0.2Fe0.8O3-δ (x = 0.10, 0.15 and 0.20) further, stoichiometric methane oxidation experiments were conducted over the same temperature range of 400- 800 oC and the results are shown in

Fig 5.14a-b. In these experiments, stoichiometric amounts of methane and 101 oxygen (CH4:O2= 1:2) were supplied in the feed stream. As can be seen from Fig

5.14a-b, perovskite-based catalysts showed a higher activity than Ni-YSZ throughout the temperature range. The cerium-doped perovskites showed the highest activity above 700°C, approaching complete conversion at 800°C.

Effect of sulfur was examined in a similar time-on-stream experiment at

800°C in the presence of 50 ppm H2S. The detrimental effect of sulfur poisoning at 800 oC (Fig 5.15a-b) was more pronounced in the stoichiometric methane oxidation experiment such that Ni-YSZ saw an immediate loss of its activity within the first hour. LSCF6428 also gradually deactivated due to sulfur poisoning as its methane conversion activity went from 90% to 30% . The cerium-doped catalysts did not display any measureable deactivation in the 10 hours that the catalysts were kept on-line.

102 a)

Figure 5.12. Steady-state methane oxidation over Sr1-xCexCo0.2Fe0.8O3

(x=0.10, 0.15, and 0.20), LSCF6428, and Ni-YSZ catalysts with sub- stoichiometric oxygen (a) CH4 conversion; (b) O2 conversion

103 a)

Figure 5.13. Methane oxidation in the presence of 50ppm H2S over Sr1- xCexCo0.2Fe0.8O3-δ (x=0.10, 0.15, and 0.20), LSCF6428, and Ni-YSZ catalysts with sub-stoichiometric oxygen (a) CH4 conversion; (b) O2 conversion 104 a)

Figure 5.14 Steady-state methane oxidation over Sr1-xCexCo0.2Fe0.8O3

(x=0.10, 0.15, and 0.20), LSCF6428, and Ni-YSZ catalysts with stoichiometric O2 (a) CH4 conversion; (b) O2 conversion

105 a)

Figure 5.15. Methane oxidation in the presence of 50ppm H2S over Sr1- xCexCo0.2Fe0.8O3-δ (x=0.10, 0.15, and 0.20), LSCF6428, and Ni-YSZ catalysts with stoichiometric oxygen (a) CH4 conversion; (b) O2 conversion 106

5.2.4 Surface properties using X-ray Photoelectron Spectroscopy (XPS)

The surface analysis of the samples was performed using X-ray

Photoelectron Spectroscopy (XPS) for both pristine and samples that were exposed to H2S. The spectra for cerium (lanthanum for LSCF6428), strontium, cobalt, iron, oxygen and sulfur are shown in Fig 5.16-20, respectively. Figure

5.16a-c, shows the Ce 3d spectra acquired over the SCeCF catalysts.

9 1 9 2 The Ce 3d5/2 peaks were present at 881.9 eV (3d f ), 888.2 eV (3d f ) and

897.5 eV (3d9f0) as shown in Fig 7a-c. The peak at 897.5 eV (3d9f0) indicates that the cerium at the surface is present as Ce+4 and there is no feature that would correspond to another phase with a lower-oxidation state, such as Ce2O3

[112-114]. Ce 3d spectra did not show any differences with increasing cerium levels. There were also no discernible changes between the pristine samples and those that were exposed to H2S.

The La 3d spectra of LSCF6428 are shown in Figure 5.16 d. La existed in two different forms in the LSCF6428 sample where the low binding energy peak at 833.2 eV can be assigned to a trivalent oxide and the peak at 837.1 eV could possibly be due to hydroxyl formation [27, 115]. No changes were observed in the La 3d spectra after exposure to sulfur.

Figure 5.17 shows the Sr 3d spectra for pristine and sulfur-exposed samples

(LSCF6428 and Sr1-xCexCo0.2Fe0.8O3-δ (x = 0.0, 0.10, 0.15, and 0.20).) Spectra showed that in all four samples, Sr existed in two different matrices. The first one 107 is Sr in the perovskite phase with a 3d5/2 binding energy near 131.5 eV and a second one is Sr in a suboxide phase, with a binding energy near 132.7 eV [115].

The surface distribution of Sr between the perovskite and suboxide phases increases in favor of the perovskite phase with increasing cerium concentration at the A-site. This implies that more Sr exists in the perovskite phase in SCeCF samples compared to LSCF6428.

Spectra taken after 10 hours of H2S poisoning showed that the surface concentration of Sr in the perovskite phase decreased. A more striking difference was that the peak at 132.7 eV corresponding to the suboxide Sr phase was replaced by a new peak at a higher binding energy of 133.4 eV, corresponding to a SrSO4 phase[116]. This suggests a strong interaction between the Sr with H2S in both the suboxide and perovskite, leading to the formation of a strontium sulfate phase on the surface.

Figure 5.18 shows the Fe 2p spectra taken over the SCeCF and LSCF samples. For all samples Fe 2p3/2 peaks were observed at 710.2. eV. The location of the peak is close to the values reported for Fe2O3 indicating that the

Fe on the surface samples exists as Fe3+[117]. The satellite peaks were located at 717.6 eV for all samples. The satellite peak of Fe 2p3/2 for Fe2O3 is approximately 8 eV higher than the main Fe 2p3/2, confirming the presence of surface Fe3+ for all the samples [118]. No significant change in the peak positions was detected nor were there any new peaks after the samples were exposed to sulfur. 108

Due to the low concentration of Co in all the samples, Co 2p spectra were not well-resolved (data not shown). However, the most intense peak located at 780.4 eV was higher than the values reported for the trivalent Co species in Co2O3

4+ (~779.5 eV) [117] and LaCoO3 [88] indicating the tetravalent species (Co ) may have formed on the surface. After poisoning the samples, no discernible change in the Co 2p spectra was observed.

The O 1s region of the XPS spectra shown in Fig 5.19 exhibits two distinctive oxygen species. The low binding energy peak at 528.2 eV corresponds to lattice oxide ions in the perovskite structure while the high binding energy peak at 530.8 eV is attributed to absorbed oxygen. Absorbed oxygen in the perovskite, in general, is broadly dispersed resulting in a broad peak at those binding energies

[115, 119]. After the samples were exposed to H2S, this broad peak at 530.8 eV became sharper and shifted to an even higher binding energy of 531.2eV confirming the presence of a sulfate species on the surface. The peak corresponding to lattice oxygen did not change much post-poisoning [116, 117].

This is in agreement with our previous studies, which showed that the absorbed oxygen is more loosely bound in the structure as compared to the lattice oxygen.

This explains why the H2S has mostly reacted with the absorbed oxygen to generate sulfate species on the surface while most of the lattice oxygen remained intact[28].

The presence of S on the surfaces of samples exposed to H2S was clearly seen in the S 2p region of the spectra taken over the pristine and H2S-poisoned 109 samples (Fig 5.20). As expected, no sulfur species were detected in any of the pristine samples while a peak at 168.4 eV was seen for all samples exposed to

H2S. Sulfur binding energies near 168.4 eV have been reported for SrSO4 . This is consistent with the results that showed the most pronounced differences were seen in the Sr 3d spectra between the pristine and sulfur-exposed samples, supporting the assertion that major interaction of sulfur has been with strontium[116].

110

Figure 5.16. X-ray photoelectron spectra of Ce 3d region for (a) SCeCF10;

(b) SCeCF15; (c) SCeCF20; and La 3d for (d) LSCF6428

111

Figure 5.17. X-ray photoelectron spectra of Sr3d region for (a) SCeCF10; (b)

SCeCF15; (c) SCeCF20; (d) LSCF6428

112

Figure 5.18. X-ray photoelectron spectra of Fe 2p region for (a) SCeCF10;

(b) SCeCF15; (c) SCeCF20; (d) LSCF6428

113

Figure 5.19. X-ray photoelectron spectra of O1S region for (a) SCeCF10; (b)

SCeCF15; (c) SCeCF20; (d) LSCF6428

114

Figure 20. X-ray photoelectron spectra of S2p region for (a) SCeCF10; (b)

SCeCF15; (c) SCeCF20; (d) LSCF6428

115

5.2.5 Temperature-programmed desorption (TPD)

To examine the stability of sulfur species on the poisoned catalysts and their potential regeneration, TPD experiments were performed over samples exposed to H2S and the results are summarized in Fig 5.21. The sulfate species generated on the surface of the catalysts evolved in the form of SO2 (shown by m/z= 64 and m/z=48). No SO3 (m/z= 80) or H2S (m/z= 34) was detected from the

TPD experiments. Cerium-doped perovskites showed sulfur desorption in the form of SO2 while LSCF6428 did not. This may be partially explained by the lower Sr content of LSCF compared to Ce-containing samples.

Even among the Ce-doped perovskites, differences were seen in the evolution temperature of SO2. SCeCF10 showed desorption of SO2 beginning at

800 oC while for SCeC15 and SCeCF20, it began at 750oC. It is possible that increasing the cerium concentration at the A-site of perovskite structure may have decreased the binding energy between strontium and sulfate. Between

SCeCF15 and SCeCF20, the former showed a narrow range of SO2 desorption while the latter desorbed SO2 over a wider range of temperature. The trend observed in SCeCF20 could be due to the presence of a ceria phase as was shown in our previous publication[120]. The presence of ceria phase in SCeCF20 could also act as an H2S absorbent along with strontium and this could have resulted in the broad range of SO2 desorption in TPD experiment [36].

116

Figure 5.21. TPD profiles over Sr1-xCexCo0.2Fe0.8O3-δ (x= 0.1, 0.15, and 0.2) and LSCF6428 catalysts showing (a) m/z =34; (b) m/z =48; (c) m/z =64; (d) m/z =80

117

5.2.6 Button cell performance

The performance of the button cells using the Ce-doped perovskites as the anode and LSM as the cathode was tested as a function of temperature and the polarization curves acquired at 700 oC, 750 oC, and 800 oC are shown in Fig

5.22. The Electrochemical Impedance Spectroscopy (EIS) results acquired at 0.6

V for each temperature are also shown in the insets of Fig 5.22. At 700 oC, cells with SCeCF10 and SCeCF15 as the anode exhibited better performance, which was also in accordance with the observations from the steady-state complete methane oxidation experiments. In our previous study [120], it was observed that the initial oxygen vacancies, which can be related to the oxide ion mobility within the cell, were in greater abundance in SCeCF10 and SCeCF15 than they were in SCeCF20 and may have resulted in the higher performance at 700oC. In addition, the possibility of the ceria phase in SCeCF20 hindering ionic conduction cannot be excluded. As the temperature was elevated to 750 oC, the performance of Ni-YSZ showed a larger increase, which could be due to a higher degree of reduction for Ni species at higher temperatures. At 800 oC, the performance of SCeCF20 and SCeCF15 matched that of Ni-YSZ, while

SCeCF10 showed a significant drop in its performance. As it was seen in the in- situ XRD patterns, SCeCF10 displayed structural instability under a reducing environment and this instability can be adversely affecting the performance of the button cell at 800 oC.

118

The impedance spectra of the cerium-doped perovskites (insets of Fig. 5.22) show that the overall polarization resistance, Rp, of the catalysts (except

SCeCF10) decreases with increasing temperature. The impedance data also follow the trends seen in polarization curves. For example, the loss in the performance of SCeCF10 due to its structural deformation is also exemplified in the impedance spectra through the highest polarization resistance (Rp) it exhibits at 800°C among all the cells. At lower temperatures, SCeCF20 has the highest impedance. The presence of a secondary CeO2 phase in SCeCF20, which was seen is in the XRD patterns, may be a contributing factor to the overall impedance of the button cell due to the electronic insulating nature of ceria. This observation is in agreement with the poorest performance of SCeCF20 seen among all the catalysts in the same temperature range. Overall, the AC impedance spectra were found to be in accordance with the polarization data.

119 a)

120 c)

Figure 5.22. Button cell performance of Sr1-xCexCo0.2Fe0.8O3-δ (x= 0.10, 0.15 and 0.20) and Ni-YSZ catalysts at (a) 700 oC; (b) 750 oC; (c) 800 oC

121

5.2.7 Sulfur-tolerance testing

The current density vs. time (chronoamperometry) profiles for the electrolyte- supported cells with SCeCF15, SCeCF20 and Ni-YSZ as the anodes, operated in

o the presence of 50 ppm H2S at 800 C are shown in Fig. 5.23. The chronoamperometry profile for SCeCF10 is omitted here because of the cell’s instability during the testing period. The data are corrected to isolate the effect of sulfur poisoning by subtracting the activity loss during sulfur-free time-on-stream.

At the introduction of 50 ppm of H2S, the Ni-YSZ cell showed an immediate drop in current density. As compared to Ni-YSZ, the cerium-doped perovskites showed a moderate decrease in current density when H2S was introduced to the feed stream. After 9 hrs of poisoning, the current densities showed an asymptotic approach to a constant performance, with Ni-YSZ showing the largest activity loss, while SCeCF15 showed the best tolerance to sulfur poisoning. These results are consistent with the stability runs made in catalytic methane oxidation in the presence of H2S, which also showed Ni-YSZ to have the highest propensity for poisoning with sulfur.

122

Figure 5.23. Deactivation of Sr1-xCexCo0.2Fe0.8O3-δ (x= 0.15, and 0.20) and

Ni-YSZ catalysts obtained through chronoamperometry at 0.6 V and 800°C in the presence of 50ppm H2S.

123

5.6 Conclusion

The structural stability of the cerium-doped catalysts with the formula Sr1-

xCexCo0.2Fe0.8O3-δ (x=0, 0.10, 0.15, and 0.20) was examined under a reducing

environment as a function of temperature. All the Sr1-xCexCo0.2Fe0.8O3-δ samples showed a cubic symmetry at room temperature. The cubic phase remained stable at higher temperatures for samples with higher cerium loading at the A- site. Oxygen non-stoichiometry in the Ce-free SCF perovskite structure is likely to cause the structural breakdown, which was also apparent in the in-situ diffraction patterns. At higher temperatures, it is plausible that a brownmillerite structure with orthorhombic symmetry is formed due to vacancy ordering in the perovskites. However, all of the perovskites tested returned to their original cubic structure when they were re-oxidized, suggesting that, if the oxygen flux from the cathode side is high enough, the structural breakdown of the anode material is less likely to occur.

Linear combination fitting of both Fe and Co K-edge spectra showed the presence of Fe3+ and Co4+ in the initial samples. At 400 oC Fe existed as Fe3+ in

SCeCF15 and SCeCF20 samples and as Fe2.5+ in SCF and SCeCF10. The Co

K-edge spectra exhibited the presence of Co3+ in SCeCF10, SCeCF15, and

SCeCF20 while Co in SCF was in slight lower oxidation state. This result is well matched with tetragonal distortion of SCF and SCeCF10 at the intermediate temperature (300 oC – 400 oC) from the HTXRD patterns since the reduction of

124 both Fe and Co will increase the oxygen non-stoichiometry in perovskite matrix which leads to structure changes.

The catalytic activity of all the cerium-doped samples for the oxidation of methane showed that there was a propensity towards total oxidation even under sub-stoichiometric oxygen conditions in the feed and the activities matched or surpassed those of Ni-YSZ and LSCF6428 after 750 oC. The sulfur tolerance of

the Sr1-xCexCo0.2Fe0.8O3-δ (x=0.10, 0.15, and 0.20) samples showed no deactivation even up to 10 hours of operation while Ni-YSZ deactivated as soon as H2S was introduced. LSCF6428 showed a slower deactivation rate compared to Ni-YSZ, but was not as stable as the SCeCF catalysts. The effect of poisoning was accelerated under oxygen rich conditions for LSCF and Ni-YSZ, but cerium- doped samples showed no loss of activity. On examining the surface composition of the samples following H2S poisoning, it was found that most of the sulfur existed as strontium sulfate. SO2 desorbed from the samples at elevated temperatures and it was observed that increased cerium-doping decreased the

SO2 desorption temperature.

The button cell performance of the cerium-doped samples with intermediate Ce loading levels matches or surpasses that of Ni-YSZ. Catalysts with lower Ce loading levels (SCeCF10) suffered from structural instability at higher temperatures. Higher Ce loading levels (SCeCF20), showed higher impedance, possibly due to segregation of a ceria phase hindering ionic conductivity. During button cell operation, Ni-YSZ showed the lowest tolerance 125 to sulfur. Perovskite with an intermediate Ce loading showed the highest stability in the presence of H2S in chronoamperometry experiments. This study shows that perovskites with intermediate Ce loadings may have potential as SOFC anodes, with button cell and methane oxidation performances matching or surpassing those of Ni-YSZ. Their much higher tolerance to sulfur poisoning compared to Ni-YSZ make them especially promising for sulfur-containing fuels.

* Information in this chapter will be submitted to Journal of Applied Catalysis B

126

CHAPTER 6.

Conclusions and Recommendations

6.1 Conclusions

The structural, surface, and catalytic properties as well as electrochemical performance of cerium-doped perovskite-type oxides with the formula Sr1-

xCexCo0.2Fe0.8O3-δ with x=0.10, 0.15 and 0.20 were studied in order to examine the possibility of these formulations for use in both SOFCs cathode and anode material.

6.1.1. Cerium-doped perovskite as cathode catalyst

Synthesis of a cubic perovskite structure was successful with no structural change or distortions observed under the SOFCs operating temperature range.

However as loadings of cerium dopant was increased beyond 15% a separated

CeO2 phase was seen to form. It was observed from HTXRD studies that there was a decrease in TEC values with increasing Ce concentration, which indicates that there is good thermal compatibility with the GDC electrolyte.

127

TPD studies showed different types of oxygen desorbing from cerium- doped perovskites, namely absorbed oxygen, denoted as α-oxygen and lattice oxygen, denoted as β- oxygen. Samples with the lowest Ce doping led to the highest oxygen vacancies generated during the oxygen TPD experiments. All cerium-doped catalysts showed the same oxidation state of Co and Fe at room temperature, which was obtained from in-situ XAFS studies as well XPS experiments. Charge unbalance of the A- and B-sites of the perovskite structures was compensated by the generation of initial oxygen vacancies.

Vacancy generation manifested itself though α-oxygen desorption, amount of which increased with decreasing cerium concentration.

The best electrochemical performance in a button cell was seen in the

SCeCF15 sample where intra-facial transport of oxygen is higher which was validated through CO2-TPO experiments described in the previous sections.

SCeCF20 showed the poorest cathodic performance as well as the highest overall area specific resistance which may be caused by the presence of a secondary phase, such as CeO2. As a result, SCeCF15 may give optimum oxygen vacancy generation and oxide ion transport while providing TEC compatibility with the electrolyte and hence may have potential application as a

SOFC cathode material.

128

6.1.2. Cerium-doped perovskite as anode catalyst.

The structural stability of the cerium-doped samples has been examined under a reducing environment. The cubic symmetry of the samples remained stable at higher temperatures for samples with higher cerium concentration at the

A-site. The cerium-doped samples were seen to return to the cubic structure on re-oxidation under air, showing that the anode structure breakdown can be prevented by providing sufficient oxygen from the cathode.

Steady-state methane oxidation studies indicate that catalytic activity of all the cerium-doped catalysts is comparable to that of Ni-YSZ and LSCF6428, especially at higher temperatures. Time-on-stream experiments performed in the presence of H2S showed high sulfur tolerance for the cerium-doped perovskite, while Ni-YSZ and LSCF both showed deactivation. The deactivation rate was higher for Ni-YSZ. The catalyst surface interaction with sulfur was studied using

XPS and TPD studies. Sulfate species were seen to interact primarily with Sr on surface while the B-site transition metals showed no significant changes by sulfur. The sulfates which formed on the surface of the catalyst were seen to desorb from the samples at elevated temperatures. Sulfate-desorption temperature decreased with increasing cerium concentration in the pervoskite structure.

The electrochemical performance of the cerium-doped samples with intermediate Ce-loadings matched that of Ni-YSZ. The instability of SCeCF10

129 under anodic conditions resulted in poorer performance. Highest Ce loading catalyst (SCeCF20) showed highest impedance, possibly due to the presence of the secondary phase (CeO2). Ni-YSZ showed lowest tolerance to sulfur while intermediate Ce loading (SCeCF15) showed the highest stability in chronoamperometry experiments in the presence of H2S. This study shows that cerium doped perovskites have a high sulfur tolerance and with intermediate Ce loading levels, they offer promise as a potential anode material.

6.2. Recommendations

6.2.1. Utilization of hydrocarbon fuels in SOFCs

The button cells tested so far had been subjected to only pure hydrogen on the anode. As hydrocarbon species, typically syngas, are used as the energy source for SOFC, internal reforming can be achieved due to its high operating temperatures. Therefore, the system set-up can be simplified by eliminating the external reformers where it also lowers the efficiencies[49, 121, 122]. Perovskite type oxide materials showed higher carbon coking tolerance than current Ni-YSZ catalyst, which require high loading levels [32, 123]. Initially, the perovskite-type oxide catalysts developed from our lab can be tested using a sulfur-free syn-gas.

Later sulfur-based poisonous gases can be included to verify the potential for the catalysts to be used with direct hydrocarbon fuels.

130

6.2.2. Coal-based fuel cells

Coal is the most abundant natural resources on earth and most of the energy reserve of coal is still remain underused[124]. Therefore direct carbon fuel cell (DCFC) or coal-based fuel cells have several advantages; (1) it is an abundant resource (2) transportation and storage of solid fuels are easier (3)

CO2 sequestration is easier since the only product from the anode will be CO2

[125-127]. Due to the similarity of operating coal-based fuel cell and SOFC, the novel catalysts developed from our lab can be tested under coal-based fuel cell set up with a small modification of the current SOFC set up. The set-up of coal- based fuel cell is shown in Fig 6.1. A fundamental understanding of the chemistry of the reactions that can take place should be the initial step for any progress in the design of coal-based catalysts. When the anode catalyst is in contact with the solid fuels (e.g. coal and carbon), there may be many relevant reactions. .

Combined with the understanding of the reaction chemistry and the novel catalysts developed, electrochemical testing of button cell charged with coal/carbon can then performed for coal-based fuel cell operations.

131

Inert!gas!+! vapor! Gas!outlet!

GC/MS!

Char!powder! Poten ostat/ Anode! `! Galvanostat Current! Cathode! `! +EIS!

Air!

Figure 6.1. The schematic of the experimental setup for electrochemical testing.

132

6.2.3. Button cell development

In this work the development of a screen-printing method to prepare the button cells was examined and compared with more traditional methods such as brush painting method. The idea behind screen-printing was to improve the performance and reproducibility of the button cells. The initial button cell testing performed using the screen-printing method has been successful. However, several steps in the fabrication process can be further optimized and studied to improve the current set up. These processes can be achieved through trial and error as well as conducting various literature searches. Initial steps could be re- optimizing the sintering temperatures and modification of the slurry compositions to get better adhesion between electrodes and electrolyte while sintering. In addition, electrode supported cells could also provide a solution for improving the performance of the button cells.

6.2.3. Electronic conductivity testing

Several perovskite type oxides are mixed ionic and electronic conductors

(MIEC). Electronic onductivity is a crucial factor determining the performance of the SOFCs. These materials, however, exhibit several orders of magnitude differences in their electronic and ionic conductive properties depending on the nature and loading levels of various dopants as well as on the operating

133 temperatures[128, 129]. Therefore, further understanding of electronic conductivities of the perovskite oxide catalysts developed from our lab will help provide useful insights into factors governing the performance of the catalyst as well as to help quantify the various sources of electrochemical losses during the

SOFC operation.

134

APPENDIX A – LIST OF ACRONYMS

SOFC – Solid Oxide Fuel Cell

PEMFC – Polymer Electrolyte Membrane Fuel Cell

DMFC – Molten Carbonate Fuel Cell

DCFC – Direct Coal Fuel Cell

LSM – Strontium Doped Lanthanum Manganite

LSCF – Strontium and Cobalt doped lanthanum Ferrite

LSCM – Strontium and Chromium doped Lanthanum Manganite

SCeCF – Cerium and Cobalt doped Strontium Ferrite

SCF – Cobalt doped Strontium Ferrite

EDTA – Ethlyenediaminetetraacetic acid

GDC – Galdolinium Doped Ceria

YSZ – Yttiria-Stabilized Zirconia

ScZ – Scandia-Stabilized Zirconia

TEC – Thermal Expansion Coefficient

TPB – Triple-Phase Boundary

GC – Gas Chromatograph

MS – Mass Spectrometry

XRD – X-ray Diffraction

XPS – X-ray Photoelectron Spectroscopy

135

TGA – Thermogravimetric Analysis

TPO – Temperature Programmed Oxidation

TPD – Temperature Programmed Desorption

TPR – Temperature Programmed Reduction

WGS – Water-gas Shift

ORR – Oxygen Reduction Reaction

XAFS – X-ray Absorption Fine Structure

XANES – X-ray Absorption Near Edge Structure

EXAFS – Extended X-ray Absorption Fine Structure

136

APPENDIX B – REACTION SYSTEM

137

APPENDIX C – ELECTROCHEMICAL SYSTEM

138

APPENDIX D –SAMPLE CALCULATION

Gas Chromatograph (Shimadzu 2014) equipped with a PDHID, FID, and

FPD detectors were used to determine the gas stream concentrations.

The conversions and yields of the components were calculated using the following relation ships:

moles of component A converted % Conversion of component A = ( )!100 moles of component A in the feed

moles of component A formed % Yield of component A = ( )!100 Theoretical maximum

139

REFERENCES

1. Smith, J.M., H.C.V. Ness, and M.M. Abbott, Introduction to Chemical Engineering Thermodynamics 5th ed ed1996, New York: The McGraw-Hill Companies.

2. Ormerod, R.M., Solid oxide fuel cells. Chem. Soc. Review, 2003. 32.

3. Carrette, L., K.A. Friedrich, and U. Stimming, Fuel Cells - Fundamentals and Applications. Fuel Cells, 2001. no. 1: p. 5-39.

4. Gorte, R.J., Recent developments towards commercialization of solid oxide fuel cells. AIChE Journal, 2005. 51(9).

5. Adler, S.B., Factors Governing Oxygen Reduction in Solid Oxide Fuel Cell Cathodes. Chemical Reviews, 2004. 104: p. 4791.

6. Sun, C.W., R. Hui, and J. Roller, Cathode materials for solid oxide fuel cells: a review. Journal of Solid State Electrochemistry, 2010. 14(7): p. 1125-1144.

7. Williams, M.C., Solid oxide fuel cells: Fundamentals to systems. Fuel Cells, 2007. 7(1): p. 78-85.

8. Tejuca, L.G. and J.L.G. Fierro, eds. Properties and applications of perovskite- type oxides. 1993, Marcel Dekker, Inc.: New York.

9. Pena, M.A., and Fierro, J.L.G., Chemical Structures and Performance of Perovskite Oxides. Chemical Reviews, 2001. 101: p. 1981.

140

10. L.Reitz, T. and H. Xiao, Characterization of electrolyte–electrode interlayers in thin film solid oxide fuel cells. Journal of Power Sources, 2006. 161: p. 437- 443.

11. Bouwmeester, H.J.M., H. Kruidhof, and A.J. Burggraaf, Importance of the surface exchange kinetics as rate limiting step in oxygen permeation through mixed-conducting oxides. Solid State Ionics, 1994. 72: p. 185.

12. Steele, B.C.H., Behaviour of porous cathodes in high temperature fuel cells. Solid State Ionics, 1997. 94: p. 239.

13. Kammer, K., Studies of Fe–Co based perovskite cathodes with different A-site cations. Solid State Communications, 2005. 177: p. 1047-1051.

14. Tu, H.Y., et al., Ln 1 _,Sr,CoO, (Ln = Sm, Dy) for the electrode of solid oxide fuel cells Solid State Ionics, 1997. 100: p. 283-288.

15. Takeda, Y., et al., GdSrCoO, for the electrode of solid oxide fuel cells Solid State Ionics, 1996. 86-88: p. 1187-90.

16. Skinner, S.J., Recent advances in Perovskite-type materials for solid oxide fuel cell cathodes. International Journal of Inorganic Materials, 2001. 3: p. 113- 121.

17. Y. Takeda, et al., Stability of La1−xAxMnO3−z (A=Ca, Sr) as cathode materials for solid oxide fuel cells. Solid State Ionics, 1994. 72: p. 257-264.

18. Chiba, R., F. Yoshimura, and Y. Sakurai, Properties of LaSryNiFeO3 as a cathode material for a low-temperature operating SOFC. Solid State Ionics, 1999. 124: p. 281-288.

19. Kharton, V.V., et al., Mixed electronic and ionic conductivity of LaCo( M)O ( M =Ga, Cr, Fe or Ni ) Solid State Ionics, 1998. 110: p. 53-60.

141

20. Kharton, V.V., et al., Mixed electronic and ionic conductivity of LaCo( M)O ( M 5 Ga, Cr, Fe or Ni )III. Diffusion of oxygen through LaCo Fe Ni O ceramic. Solid State Ionics, 1998. 110(61-68).

21. Holc, J., et al., Electrical and microstructural characterisation of (La)(Fe,Al)O and (LaO)(Mn,Al)O as possible SOFC cathode materials Solid State Ionics, 1997. 95: p. 259-268.

22. Komatsu, T., et al., Long-Term Chemical Stability of LaNi(Fe)O3 as a Cathode Material in Solid Oxide Fuel Cells. Journal of The Electrochemical Society, 2007. 154(4): p. B379-B382.

23. Zhou, W., R. Ran, and Z. Shao, Progress in understanding and development of Ba0.5Sr0.5Co0.8Fe0.2O3-d based cathodes for intermediate-temperature solid-oxide fuel cells:A review. Journal of Power Sources, 2009. 192: p. 231-246.

24. Zhou, W., et al., Evaluation of A-site cation-deﬁcient (Ba0.5Sr0.5)1 −x Co0.8Fe0.2O3−ı (x > 0) perovskite as a solid-oxide fuel cell cathode. Journal of Power Sources, 2008. 182: p. 24-31.

25. McIntosh, S., Vente, Jaap F., Haije, Wim G., Blank, Dave H. A., Bouwmeester, Henny J. M., Structure and oxygen stoichiometry of SrCo0.8Fe0.2O3-d and Ba0.5Sr0.5Co0.8Fe0.2O3-d. Solid State Ionics, 2006. 177(19- 25): p. 1737-1742.

26. Qingshan, Z., Tongan, Jin, Yong, Wang, Thermal expansion behavior and chemical compatibility of BaxSr1-xCo1-yFeyO3-d with 8YSZ and 20GDC. Solid State Ionics, 2006. 177(13-14): p. 1199-1204.

27. Kuhn, J.N. and U.S. Ozkan, Surface properties of Sr- and Co-doped LaFeO3. Journal of Catalysis, 2008. 253(1): p. 200-211.

28. Kuhn, J.N. and U.S. Ozkan, Effect of Co Content Upon the Bulk Structure of Sr- and Co-doped LaFeO3. Catalysis Letters, 2008. 121(3-4): p. 179-188.

142

29. Kuhn, J.N., et al., Oxygen Exchange Kinetics over Sr- and Co-Doped LaFeO3. Journal of Physical Chemistry, 2008. 112: p. 12468-12472.

30. Lakshminarayanan, N., et al., Effect of additional B-site transition metal doping on oxygen transport and activation characteristics in La0.6Sr0.4(Co0.18Fe0.72X0.1)O3−ı (where X= Zn, Ni or Cu) perovskite oxides. Applied Catalysis B: Environmental, 2011. 103: p. 318-325.

31. Zhu, W.Z., Deevi, S. C., A review on the status of anode materials for solid oxide fuel cells. Materials Science and Engineering: A, 2003. 362(1-2): p. 228- 239.

32. McIntosh, S., Gorte, R.J.,, Direct Hydrocarbon Solid Oxide Fuel Cells. Chemical Reviews, 2004. 104: p. 4845-4865.

33. Assabumrungrata, S., et al., Thermodynamic analysis for a solid oxide fuel cell with direct internal reforming fueled by ethanol. Chemical Engineering Science 2004. 59: p. 6015 – 6020.

34. Finnerty, C.M. and R.M. Ormerod, Internal reforming over nickel/zirconia anodes in SOFCS oparating on methane: influence of anode formulation, pretreatment and operating conditions. Journal of Power Sources, 2000. 86(1-2): p. 390-394.

35. Weber, A., et al., Oxidation of H2, CO and methane in SOFCs with Ni/YSZ- cermet anodes. Solid State Ionics, 2002. 152-153: p. 543-550.

36. Gong, M., et al., Sulfur-tolerant anode materials for solid oxide fuel cell application. Journal of Power Sources, 2007. 168(2): p. 289-298.

37. Lohsoontorn, P., D.J.L. Brett, and N.P. Brandon, Thermodynamic predictions of the impact of fuel composition on the propensity of sulphur to interact with Ni and ceria-based anodes for solid oxide fuel cell. Journal of Power Sources, 2008. 175: p. 60-67.

143

38. Bartholomew, C.H., Mechanisms of catalyst deactivation. Applied Catalysis A: General 2001. 212: p. 17-60.

39. Sasaki, K., Susuki, K., Iyoshi, A., Uchimura, M., Imamura, N., Kusaba, H., Teraoka, Y., Fuchino, H., Tsujimoto, K., Uchida, Y., Jingo, N., H2S Poisoning of Solid Oxide Fuel Cells. Journal of The Electrochemical Society, 2006. 153(11): p. A2023-A2029.

40. Peterson, D.R. and J. Winnick, Utilization of Hydrogen Sulfide in an intermediate-Temperature Ceria-Based Solid Oxide Fuel Cell. journal of electrochemical society, 1998. 145(5): p. 1449-1454.

41. Xu, Z., J. Luo, and K.T. Chuang, CoS-Promoted MoS2 Catalysts for SOFC Using H2S Containing Hydrogen or Syngas as Fuel. Journal of The Electrochemical Society, 2007. 154(6): p. B523-B527.

42. He, H., J.M. Vohs, and R.J. Gorte, Carbonaceous deposits in direct utilization hydrocarbon SOFC anode. Journal of Power Sources, 2005. 144: p. 135–140.

43. He, H., Gorte, R.J., and Vohs, J.M., Highly Sulfur Tolerant Cu-Ceria Anodes for SOFCs. Electrochemical and Solid-State Letters, 2005. 8(6): p. A279.

44. Choi, Y.M., A mechanistic study of H2S decomposition on Ni- and Cu-based anode surfaces in a solid oxide fuel cell. Chem. Phys. Lett., 2006. 421: p. 179.

45. Lee, A.C., et al., Conversion of Solid Carbonaceous Fuels in a Fluidized Bed Fuel Cell Electrochemical and Solid-State Letters, 2008. 11(2): p. B20-B23.

46. Costa-Nunes, O., R.J. Gorte, and J.M. Vohs, Comparison of the performance of Cu–CeO2–YSZ and Ni–YSZ composite SOFC anodes with H2, CO, and syngas. Journal of Power Sources, 2005. 141: p. 241-249.

47. Pointon, K., et al., The development of a carbon–air semi fuel cell Journal of Power Sources, 2006. 162(2): p. 750-756. .

144

48. Merino, N.A., et al., La1-xCaxCoO3 perovskite-type oxides: Identification of the surface oxygen species by XPS. applied surface science, 2006. 253: p. 1489- 1493.

49. Gorte, R.J. and J.M. Vohs, Novel SOFC anodes for the direct electrochemical oxidation of hydrocarbons. Journal of Catalysis, 2003. 216(1-2): p. 477-486.

50. Tao, S., Irvine, J.T.S., Discovery and Characterization of Novel Oxide Anodes for Solid Oxide Fuel Cells. The Chemical Record, 2004. 4: p. 83-95.

51. Lv, H., et al., Synthesis and electrochemical behavior of Ce1−xFexO2−δ as a possible SOFC anode materials. Solid State Ionics, 2007. 177(3467-3472).

52. Bossche, M.v.d. and S. McIntosh, The rate and selectivity of methane oxidation over LaSrCrMnO as a function of lattice oxygen stoichiometry under solid oxide fuel cell andoe conditions. Journal of Catalysis, 2008. 225: p. 313-323.

53. Tao, S. and J.T.S. Irvine, Catalytic Properties of the Perovskite Oxide La0.75Sr0.25Cr0.5Fe0.5O3-δin Relation to Its Potential as a Solid Oxide Fuel Cell Anode Material. Chemistry of Materials, 2004. 16(21): p. 4116-4121.

54. Ishihara, T., T. Oishi, and S. Hamamoto, Praseodymium oxide doped with Bi for diesel soot oxidation at low temperature. Catalysis communications, 2009. 10: p. 1722-1724.

55. Sfeir, J., et al., Lanthanum Chromite Based Catalysts for Oxidation of Methane Directly on SOFC anodes. Journal of Catalysis, 2001. 202: p. 229-244.

56. Sauvet, A.L. and J.T.S. Irvine, Catalytic activity for steam methane reforming and physical characterisation of La1-xSrxCr1-yNiyO3-d. Solid State Ionics, 2004. 167(1-2): p. 1-8.

57. Blennow, P., et al., Electrochemical characterization and redox behavior of Nb- dope SrTiO3. Solid State Ionics, 2009. 180: p. 63-70.

145

58. Gorte, R.J., Ceria in Catalysis: From Automotive Applications to the Water– Gas Shift Reaction. AIChE Journal, 2010. 56(5): p. 1126-1135.

59. Danilovic, N., et al., Ce0.9Sr0.1VOx (x = 3, 4) as anode materials for H2S- containing CH4 fueled solid oxide fuel cells. Journal of Power Sources 2009. 192 p. 247–257.

60. Ishihara, T., et al., Ce0.6(Mn0.3Fe0.1)O2 as an Oxidation-Tolerant Ceramic Anode for SOFCs Using LaGaO3−δ Based Oxide Electrolyte. Electrochemical and Solid-State Letters, 2010. 12(8): p. B95-B97.

61. Lay, E., et al., Ce-substituted LSCM as new anode material for SOFC operating in dry methane. Solid State Ionics 2008. 179: p. 1562–1566.

62. Colomer, M.T., B.C.H. Steele, and J.A. Kilner, Structural and electrochemical

properties of the Sr0.8Ce0.1Fe0.7Co0.3O3-δ perovskite as cathode material for ITSOFCS. Solid State Ionics, 2002. 147: p. 41.

63. Huang, Y.-H., Double Perovskites as Anode Materials for Solid-Oxide Fuel Cells. Science, 2006. 312: p. 254.

64. Faro, M.L., et al., Electrochemical investigation of a propane-fed solid oxide fuel cell based on a composite Ni–perovskite anode catalyst. Applied Catalysis B, environmental, 2009. 89: p. 49-57.

65. Huang, Y.-H., Synthesis and Characterization of Sr2MgMoO6− : An Anode Material for the Solid Oxide Fuel Cell. Journal of the Electrochemistry Society, 2006. 153(7): p. A1266.

66. Joo, J.H. and G.M. Choi, Thick-film electrolyte (thickness <20m)-supported solid oxide fuel cells. Journal of Power Sources, 2008. 180: p. 195-198.

67. Hsueh, C.H., et al., Contrained sintering of electrolyte films on porous electrode. Electrochemical society proceedings, 2001. 6: p. 293-.

146

68. Hibiki, I., Sintering behavior and performance of anode materials for SOFC. Electrochemical society 1995: p. 639-.

69. Shi, Y., N. Cai, and C. Li, Numerical modeling of an anode-supported SOFC button cell considering anodic surface diffusion. Journal of Power Sources, 2007. 164: p. 639-648.

70. Huang, B., et al., Performance of Ni/ScSZ cermet anode modified by coating with Gd0.2Ce0.8O2 for a SOFC. Materials research bulletin, 2007. 42: p. 1705-1714.

71. Chena, K., et al., Performance of an anode-supported SOFC with anode functional layers. Electochimica Acta, 2008. 53: p. 7825-7830.

72. Ressler, T., WinXAS: a Program for X-ray Absorption Spectroscopy Data Analysis under MS-Windows. Journal of Synchrotron Radiation, 1998. 5(2): p. 118-122.

73. Voorhoeve, R.J.H., et al., Perovskite Oxides: Materials Science in Catalysis. Science, 1977. 195: p. 827-833.

74. Adler, S.B., Mechanism and kinetics of oxygen reduction on porous La1-

xSrxCo3-δ electrodes. Solid State Ionics, 1998. 111: p. 125.

75. Mizusaki, J., Mima Y., Yamauchi S., Fueki, K., and Tagawa, H.,

Nonstoichiometry of the Perovskite-Type Oxides La1-xSrxCoO3-δ. Journal of Solid State Chemistry, 1989. 80: p. 102.

76. Mizusaki, J., et al., Nonstoichiometry and Defect Structure of the Perovskite-

Type Oxides La1-xSrxFeO3-δ. Journal of Solid State Chemistry, 1985. 58: p. 257.

147

77. Mizusaki, J., Moir, N., Takai, H., Yonemura, Y., Minamiue, H., Tagawa, H., Dokiya, M., Inaba, H., Naraya, K., Sasamoto, T., and Hashimoto, T., Oxygen nonstoichiometry and defect equilibrium in the perovskite-type oxides La1-

xSrxMnO3+δ. Solid State Ionics, 2000. 129: p. 163.

78. Kuhn, J.N. and U.S. Ozkan, Surface properties of Sr- and Co-doped LaFeO3. Journal of Catalysis, 2007. 253: p. 200-211.

79. Lakshminarayanan, N., et al., Variation of structure and properties of La1−xSrxCo0.2Fe0.8O3−ı with Sr content: Implications for oxidation activity. Journal of Molecular Catalysis A: Chemical, 2011. 336: p. 23-33.

80. Leanza, R., et al., Perovskite catalysts for the catalytic flameless combustion of methane Preparation by flame-hydrolysis and characterisation by TPD–TPR- MS and EPR. Applied Catalysis B: Environmental, 2000: p. 55-64.

81. Nitadori, T. and M. Misono, Catalytic Properties of La1-xA'xFe03 (A’ = Sr,Ce) and La1-xCexCo03. Journal of Catalysis, 1984. 93: p. 459-466.

82. Forni, L., et al., FT-IR and EPR spectroscopic analysis of La1-xCexCoO3 perovskite-like catalysts for NO reduction by CO. Applied Catalysis B: Environmental, 1997. 13: p. 35-43.

83. Adler, S.B., Chemical Expansivity of Electrochemical Ceramics. Journal of the American Ceramic Society, 2001. 84(9): p. 2117.

84. Selyama, T., N. Yamazoe, and K. Eguchl, Characterization and Activity of Some Mixed Metal Oxide Catalysts. Industrial & Engineering Chemistry Product Research and Development, 1985. 24(1): p. 19-27.

85. Seiyama, T., Yamazoe, N., and Eguchi, K., Characterization and Activity of Some Mixed Metal Oxide Catalysts. Ind. Eng. Chem. Prod. Res. Dev., 1985. 24: p. 19.

148

86. Yamazoe, N., Y. Teraoka, and T. Seiyama, TPD and XPS Study on Thermal Behavior of Absorbed Oxygen in La1-xSrxCoO3. Chemistry Letters, 1981: p. 1767.

87. Teraoka, Y., H.-M. Zhang, and N. Yamazoe, Oxygen-sorptive properties of defect perovskite-type La1-xSrxCo1-yFeyO3-d. Chemistry Letters, 1985: p. 1367- 1370.

88. Tejuca, L.G., J.L.G. Fierro, and J.M. Tascon, Structure and reactivity of perovskite-type oxides. Advances in Catalysis, 1989. 36: p. 237.

89. Bunker, G., Introduction to XAFS: A Practical Guide to X-ray absorption Fine Structure Spectroscopy2010, New York: Cambridge University Press.

90. Haas, O., et al., The Fe K-edge X-ray absorption characteristics of La1- xSrxFeO3-d prepared by solid state reaction. Materials Research Bulletin, 2009. 44: p. 1397-1404.

91. Haas, O., R.P.W.J. Struis, and J.M. McBreen, Synchrotron X-ray absorption of LaCoO3 perovskite. Jounal of solid state chemistry, 2004. 177: p. 1000-1010.

92. Haas, O., et al., X-ray absorption and diffraction studies of La0.6Ca0.4CoO3 perovskite, a catalyst for bifunctional oxygen electrodes. Electrochemica Acta, 2002. 47(3211-3217).

93. Hueso, J.L., et al., Chemical and electronic characterization of cobalt in a lanthanum perovskite. Effects of strontium substitution. Jounal of solid state chemistry, 2010. 183: p. 27-32.

94. Huang, T.-J. and S.-Y. Jhao, Ni-Cu/samaria-doped ceria catalysts for steam reforming of methane in the presence of carbon dioxide. Applied Catalysis A: General 2006. 302: p. 325-332.

149

95. Huang, T.-J., X.-D. Shen, and C.-L. Chou, Characterization of Cu, Ag and Pt

added La0.6Sr0.4Co0.2Fe0.8O3-δ and gadolinia-doped ceria as solid oxide fuel cell electrodes by temperature-programmed techniques. Journal of Power Sources, 2009. 187(2): p. 348-355.

96. Yokokawa, H., et al., Fundamental mechanisms limiting solid oxide fuel cell durability. Journal of Power Sources. 182(2): p. 400-412.

97. Jacobson, A.J., Materials for Solid Oxide Fuel Cells. Chemistry of Materials, 2010. 22: p. 660–674.

98. Fergus, J.W., Oxide anode materials for solid oxide fuel cells. Solid State Ionics, 2006. 177 p. 1529–1541.

99. Matsuzaki, Y. and I. Yasuda, The poisoning effect of sulfur-containing impurity gas on a SOFC anode: Part I. Dependence on temperature, time, and impurity concentration. Solid State Ionics, 2000. 132(3-4): p. 261-269.

100. Fagg, D.P. and J.T.S. Irvine, Synthesis and electrical characterisation of doped perovskite titanates as potential anode materials for solid oxide fuel cells. Journal of Materials Chemistry, 1997. 7(12): p. 2495-2498.

101. Sauvet, A.-L., et al., Surface Properties and Physicochemical Charaterizations of a New type of Anode Material, LaSrCrRuO, for a Solid Oxide fuel cell under methane at intermediate temperature. Journal of Catalysis, 2002. 209: p. 25- 34.

102. Forni, L. and I. Rossetti, Catalytic combustion of hydrocarbons over perovskites. Applied Catalysis B: Environmental, 2002. 38: p. 29.

103. Rossetti, I., et al., Effect of sulphur poisoning on perovskite catalysts prepared by flame-pyrolysis. Applied Catalysis B: Environmental, 2009. 89: p. 283-390.

104. Vashuk, V.V., et al., Srl-xCexFel-yCoy03-d (x <_ 0.15, 0 < y < 1) Solid Solutions. Inorganic Materials, 2000. 36: p. 1173-1177. 150

105. Kokhanovskii, L.V., et al., Oxygen Stoichiometry and Phase Transitions of SrCo1 – xFexO3 –d. Inorganic Materials, 2001. 37: p. 730-736.

106. HAGENMULLER, P., M. POUCHARD, and J.C. GRENIER, Nonstoichiometry in the perovskite-type oxides: An evolution from the classical Schoti'ky-Wagner model to the recent high Tc superconductors. Solid State Ionics, 1990. 43: p. 7-18.

107. Kruidhof, H., et al., Influence of order-disorder transitions on oxygen permeability through selected nonstoichiometric perovskite-type oxides. Solid State Ionics, 1993. 63-65: p. 816-822.

108. Bezdicka, P., et al., Mossbauer Characteristics of the Sr2CoFeO6 Perovskite Obtained by Electrochemical Oxidation. Solid State Communications, 1994. 91: p. 501-505.

109. Mizusaki, J., et al., Nonstoichiometry and Phase Relationship of the SrFeO2.5- SrFeO. Jounal of solid state chemistry, 1992. 99: p. 166-172.

110. Švarcová, S., et al., Structural instability of cubic perovskite BaSr1 − xCo1 − yFeyO3 − δ. Solid State Ionics, 2008. 178: p. 1787-1791.

111. Lakshminarayanan, N., et al., Doped LaFeO3 as SOFC catalysts: Control of oxygen mobility and oxidation activity. Catalysis Today, 2010. 157: p. 446-450.

112. Kang, J.-S., et al., The valence state of Ce in electron-doped manganites: La0.7Ce0.3MnO3. JOURNAL OF PHYSICS: CONDENSED MATTER, 2001. 12: p. 3779-3789.

113. Zhang, F., et al., Cerium oxidation state in ceria nanoparticles studied with X- ray photoelectron spectroscopy and absorption near edge spectroscopy. Surface Science 563 (2004) 74–82, 2004. 563: p. 74-82.

151

114. Song, H. and U.S. Ozkan, The role of impregnation medium on the activity of ceria-supported cobalt catalysts for ethanol steam reforming. Journal of Molecular Catalysis A: Chemical, 2010(318): p. 21-29.

115. Machkova, M., et al., Surface behavior of Sr-doped lanthanide perovskites. Applied Surface Science, 1997. 119(1-2): p. 127-136.

116. Vasquez, R.P., SrSO4 by XPS. Surface Science Spectra, 1992. 1: p. 117-121. 117. Moulder, J.F., et al., Handbook of X-ray Photoelectron Spectroscopy. 2nd ed, ed. J. Chastain1992, Eden Prairie, MN: Perkin-Elmer Corporation.

118. Yamashita, T. and P. Hayes, Analysis of XPS spectra of Fe2+ and Fe3+ ions in oxide materials. Applied Surface Science, 2008. 254: p. 2441-2449.

119. Yamazoe, N., Teraoka, Y., and Seiyama, T., TPD and XPS Study on Thermal Behavior of Absorbed Oxygen in La1-xSrxCoO3. Chemistry Letters, 1981: p. 1767.

120. Choi, H., et al., Ce-doped strontium cobalt ferrite perovskites as cathode catalysts for solid oxide fuel cells: Effect of dopant concentration. Applied Catalysis B: Environmental, 2012. 127: p. 336-341.

121. Gross, M.D., J.M. Vohs, and R.J. Gorte, Recent progress in SOFC anodes for direct utilization of hydrocarbons. Journal of Materials Chemistry, 2007. 17(30): p. 3071-3077.

122. Lu, C., Worrell, W.L., Wang, C., Park, S., Kim, H., Vohs, J.M., Gorte, R.J.,, Development of solid oxide fuel cells for the direct oxidation of hydrocarbon fuels. Solid State Ionics, 2002. 152-153: p. 393-397.

123. Gorte, R.J., et al., Anodes for Direct Oxidation of Dry Hydrocarbons in a Solid-Oxide Fuel Cell. Advanced Materials, 2000. 12(19).

124. Liu, R., et al., A novel direct carbon fuelcell by approach of tubular solid oxide fuel cells. Journal of Power Sources, 2010. 195: p. 480–482. 152

125. Lia, C., Y. Shib, and N. Caia, Performance improvement of directcarbon fuelcell by introducing catalytic gasiﬁcation process. Journal of Power Sources, 2010. 195: p. 4660–4666

126. Singh, R., et al., Performance and Byproduct Analysis of Coal Gas Solid Oxide Fuel Cell. energy & fuels article, 2010. 24: p. 1176-1183.

127. Li, X., et al., Modiﬁcation of Coal as a Fuel for the Direct Carbon Fuel Cell. Journal of Physical Chemistry, 2010. 114: p. 3855–3862.

128. MIZUSAKI, J., et al., Electronic Conductivity, Seebeck Coefficient, and Defect Structure of Lal-xSrxFeO3 (x=O.l, 0.25). Journal of the American Ceramic Society, 1983. 66(4): p. 247-252.

129. Esquirol, A., et al., Electrochemical Characterization of La0.6Sr0.4Co0.2Fe0.8O3 Cathodes for Intermediate-Temperature SOFCs. Journal of The Electrochemical Society, 2004. 151(11): p. A1847-A1855.

153