SOLID OXIDE FUEL CELL CATHODE INFILTRATE PARTICLE SIZE CONTROL AND OXYGEN SURFACE EXCHANGE RESISTANCE DETERMINATION By Theodore E. Burye A THESIS Submitted to Michigan State University i n partial fulfillment of the requirements for the degree of Chemical Engineering Doctor of Philosophy 2015 ABSTRACT SOLID OXIDE FUEL CELL CATHODE INFILTRATE PARTICLE SIZE CONTROL AND OXYGEN SURFACE EXCHANGE R EISTANCE DETERMINATION By Theodore E. Burye Over the past decade, n ano - sized Mixed Ionic Electronic C onducting (MIEC) micro - sized Ionic C onducting (IC) c omposite cathodes produced by the infiltration method have received much attention in the literature [ 1 - 9 ] due to their low polarization resistance ( R P ) at intermediate (500 - 700°C) operating temperatures. S mall infiltrated MIEC oxide nano - particle size and low intrinsic MIEC oxygen surface exchange resistance ( R s ) have been two critical factors allowing these Nano - Micro - Composite Cathodes (NMCCs) to achieve high performance and/or low temperature operation. Unfortunately, previous studies have not f ound a reliable method to control or reduc e infiltrated nano - particle size . In addition, controversy exists on the best MIEC infiltrate composition because : 1) R s measurements on infiltrated MIEC particles are presently unavailable in the literature, and 2) bulk and thin film R s measurements on nominally identical MIEC compositions often vary by up to 3 orders of magnitude [ 10 ] . Here , t wo processing techniques , precursor nitrate so lution desiccation a nd ceria oxide pre - infiltration , were developed to syst ematic ally produce a reduction in the average La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF) infiltrated nano - particle size from 50 nm to 22 nm. This particle size reduction reduced the SOFC o perating temperature , (defined as the temperature where R P 2 ) from 650 °C to 540°C . In addition , R s values for infiltrated MIEC particles were determined for the first time through finite element iii modeling calculations on 3D Focused Ion Beam - Scanning Electron Microscope (FIB - SEM) reconstructions of electrochemically characterized infiltrated electrodes . iv ACKNOWLEDGEMENTS This work was supported by Nat ional Science Foundation (NSF) A ward No. CBET - 1254453 and a Michigan State University faculty start up grant to Dr. Jason D. Nicholas . Microscopy work was conducted at the Michigan State Composites Center, which is supported by the NSF Major Instrumentation Program and Michigan State University. Profilometry work was conducted at the W.M. Keck Microfab rication Facility supported by the W.M. Keck Foundation and Michigan State University. I would also like to thank my committee members for their support in my academic studies and research. First, I would like thank my advisor, Dr. Jason D. Nicholas, who has helped me numerous times throughout my st udies and helped me understand Material S cience. Next, I would like to thank Dr. Lunt, who has provided a vast amount of diffraction knowledge which helped me characterize my samples. My understanding of imped ance characterization has been greatly improved by Dr. Barton, and Dr. Duxbury has also provided a valuable critical external perspective . I would also like to thank p revious and current group members for their support and guidance through my doctoral pr ogram. Dr. Qing Yang, Lin Wang, Andrew Flegler, Peter Su, Vasiliy Sharikov - Bass, Tridip Das, Eric Straley, Hongjie Tang, and Yuxi Ma all contribute d to my work and provided me with helpful insig hts into my research and useful discussions . Finally, I would like to thank my family for supporting me through my time in the Ph.D program. v TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ............. ix LIST OF FIGURES ................................ ................................ ................................ ............ x KEY TO ABBREVIATIONS ................................ ................................ ......................... xvii CHAPTER 1: Thesis Motivation and Overview ................................ ................................ 1 1.1 The World Energy Problem ................................ ................................ ...................... 1 1.2 Solid Oxide Fuel Cells as a Solution to the World Energy Problem ........................ 1 1.3 Solid Oxide Fuel Cell Knowledge Gaps ................................ ................................ ... 3 1.4 Thesis Overview ................................ ................................ ................................ ....... 3 1.5 Summary ................................ ................................ ................................ ................... 4 CHAPTER 2: Literature Review ................................ ................................ ........................ 5 2.1 Solid Oxide Fuel Cell Overview ................................ ................................ ............... 5 2.1.1 Solid Oxide Fuel Cell Operating Principles ................................ ....................... 6 2.1.2 Traditional and Nano - Micro - Composite Cathodes ................................ ............ 9 2.1.3 Thin Film Cathodes ................................ ................................ .......................... 11 2.2 Planar Solid Oxide Fuel Cell Geometric Arrangements ................................ ......... 12 2.3 Commonly Used Solid Oxide Fuel Cell Materials ................................ ................. 13 2.3.1 Ionic Conductors ................................ ................................ .............................. 14 2.3.2 Electronic Conductors ................................ ................................ ...................... 14 2.3.3 Mixed Ionic Electronic Conductors ................................ ................................ . 14 2.4 Composite Solid Oxide Fuel Cell Cathode Geometries ................................ ......... 15 2.4.1 Electronic Conducting Scaffolds ................................ ................................ ..... 15 2.4.2 Mixed Ionic Electronic Conducting Scaffolds ................................ ................. 16 2.4.3 Ionic Conducting Scaffolds ................................ ................................ .............. 17 2.5 Models Identifying the Parameters Controlling Nano - Micro - Composite Cathode Performance ................................ ................................ ................................ .................. 18 2.5.1 The Surface Resistance Model ................................ ................................ ......... 18 2.5.2 The Simple Infiltration Microstructure Polarization Loss Estimation Model . 19 2.6 Infiltrate Particle Size is Difficult to Control According to the Literature ............. 22 2.6.1 Fabrication Techniques that Reduce Particle Size ................................ ........... 23 2.6.2 Fabrication Techniques that Have No Impact on Particle Size Reduction ...... 25 2.7 Literature Oxygen Surface Exchange Resistance Measurement Techniques ......... 25 2.7.1 Oxygen Surface Exchange Resistance from Thin Film Electrochemical Impedance Spectroscopy ................................ ................................ ........................... 26 2.7.3 Oxygen Surface Resistance from Chemical Rate Constant Measurements ..... 29 2.8 Large Uncertainties Exist in the R s Value of Even the Most Common Mixed Ionic Electronic Conducting Materials ................................ ................................ .................. 32 2.9 Summary ................................ ................................ ................................ ................. 33 vi CHAPTER 3: Experimental Methods ................................ ................................ ............... 35 3.1 Symmetrical Cathode Fabrication Processes ................................ .......................... 35 3.1.1 Cathode - Electrolyte - Cathode Symmetric Cell Production .............................. 35 3.1.2 Cathode Fabrication ................................ ................................ ......................... 36 3.1.3 Mixed Ionic and Electronic Conducting Precursor Solution Fabrication and Cathode Infiltration ................................ ................................ ................................ ... 37 3.1.4 Mixed Ionic and Electronic Conducting Oxide Phase Purity Analysis ........... 38 3.1.5 Current Collector Application ................................ ................................ .......... 39 3.2 Characterization Techniques ................................ ................................ ................... 39 3.2.1 Electrochemical Impedance Spectroscopy ................................ ....................... 39 3.2.2 Scanning Electron Microscopy ................................ ................................ ........ 41 3.2.3 Solid Oxide Fuel Cell Nano - Particle Coarsening Rate ................................ .... 42 3.2.4 X - ray Diffraction ................................ ................................ ............................. 42 3.2.5 Williamson - Hall Particle Size Determination from X - ray Diffraction ............ 43 3.2.6 ThermoGravimetric Analysis ................................ ................................ ........... 44 3.2.7 Profilometry ................................ ................................ ................................ ..... 45 3.2.8 Focused Ion Beam - Scanning Electron Microscopy 2D Serial Sectioning ...... 46 3.3 Finite Element Model ing of Cathode Microstructure and Performance ................. 47 3.3.1 Motivation for Using Finite Element Modeling ................................ .............. 47 3.3.2 Finite Element Modeling Performance Calculation Overview ........................ 48 3.3.3 Focused Ion Beam - Scanning Electron Microscopy Sample Preparation ........ 52 3.3.4 Cathode Microstructure 3D Reconstruction ................................ .................... 54 3.3.5 3D Cathode Reconstruction Volume Meshing ................................ ................ 55 3.3.6 Infiltrated Solid Oxide Fuel Cell Cathode Finite Element Modeling to Predict Polarization Resistance ................................ ................................ ............................. 56 3.4 Summary ................................ ................................ ................................ ................. 58 CHAPTER 4: The Impact of Precursor Nitrate Solution Desiccation on Infiltrated La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - Cathodes ................................ ................................ ....................... 60 4.1 Introduction ................................ ................................ ................................ ............. 60 4.2 Experimental Methods ................................ ................................ ............................ 61 4.2.1 Cathode - Electrolyte - Cathode Symmetric Cell Production .............................. 61 4.2.2 Symmetric Cell Impedance Measurements ................................ ..................... 63 4.2.3 X - ray Diffraction Measurements ................................ ................................ ..... 64 4.2.4 Scanning Electron Microscopy Measurements ................................ ................ 65 4.2.5 Nano - Micro - Composite Cathode Performance Modeling ............................... 65 4.3 Results ................................ ................................ ................................ ......................... 67 4.3.1: Desiccant Impacts on Infiltrate Particle Size ................................ .................. 67 4.3.2: Desiccant Impacts on Infiltrate Phase Purity ................................ .................. 69 4.3.3: Desiccant Impacts on Cathode Electrochemica l Performance ....................... 71 4.4 Summary ................................ ................................ ................................ ................. 78 CHAPTER 5: The Impact of Surfactants on Desiccated La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - Infiltrated Solid Oxide Fuel Cell Cathodes ................................ ................................ ........................ 79 5.1 Introduction ................................ ................................ ................................ ............. 79 vii 5.2 Experimental Methods ................................ ................................ ............................ 79 5.2.1 Cathode - Electrolyte - Cathode Symmetric Cell Production .............................. 79 5.2.2 Symmetric Cell Impedance Measurements ................................ ..................... 80 5.2.3 X - ray Diffraction Measurements ................................ ................................ ..... 81 5.2.4 Scanning Electron Microscopy Measurements ................................ ................ 81 5.2.5 Nano - Micro - Composite Cathode Performance Modeling ............................... 81 5.2.6 Williamson - Hall Particle Size and Strain Calculations ................................ ... 81 5.3 Results ................................ ................................ ................................ ..................... 82 5.3.1: Desiccant and Solution Additive Impacts on Infiltrate Particle Size .............. 82 5.3.2: Desiccant and Solution Additive Impacts on Infiltrate Phase Purity .............. 94 5.3.3 Impurity Impacts on LSCF Nano - Particle Coarsening Behavior .................. 102 5.3.4 Desiccant and Solution Additive Impacts on Performance and Stability ...... 105 5.4 Summary ................................ ................................ ................................ ............... 110 CHAPTER 6: The Impact of Nano - Ceria Pre - Infiltration on La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - Infiltrated Solid Oxide Fuel Cell Cathodes ................................ ................................ ..... 111 6.1 Introduction ................................ ................................ ................................ ........... 111 6.2 Experimental Methods ................................ ................................ .......................... 111 6.2.1 Cathode - Electrolyte - Cathode Symmetric Cell Production ............................ 111 6.2.2 Symmetrical Cell Impedance, X - ray Diffraction, and Scanning Electron Microscopy Measurements ................................ ................................ ..................... 114 6.2.3 ThermoGravimetric Analysis Measurements ................................ ................ 114 6.2.4 Nano - Micro - Composite Cathode Performance Modeling ............................. 114 6.3 Results ................................ ................................ ................................ ................... 115 6.3.1 Pre - Infiltration and Solution Additive Impacts on Infiltrate Particle Size ..... 115 6.3.2 Pre - Infiltration and Solution Additive Impact on Infiltrate Phase Purity ...... 130 6.3.3 Pre - Infiltration and Solution Additive Impact on Precursor Solution Decomposition Behavior ................................ ................................ ........................ 134 6.3.4 Pre - Infiltrati on and Solution Additive Impacts on Performance and Stability ................................ ................................ ................................ ................................ . 136 6.4 Summary ................................ ................................ ................................ ............... 139 CHAPTER 7: The I mpact of Precursor Solution Desiccation and Nano - Ceria Pre - Infiltration on La 0.6 Sr 0.4 Co 1 - x Fe x O 3 - ................................ 141 7.1 Introduction ................................ ................................ ................................ ........... 141 7.2 Experimental Methods ................................ ................................ .......................... 141 7.2.1 Cathode - Electrolyte - Cathode Symmetric Cell Production ............................ 141 7.2.2 Symmetrical Cell Impedance, X - ray Diffraction, and Scanning Electron Microscopy Measurements ................................ ................................ ..................... 143 7.2.3 Nano - Micro - Composite Cathode Performance Modeling ............................. 143 7.3 Results ................................ ................................ ................................ ................... 143 7.3.1 Pre - Infiltration, Desiccation and Infiltrated Composition Impacts on Infiltrate Particle Size and Performance ................................ ................................ ................ 143 7.3.2 Pre - Infiltration, Desiccation and Infiltrated Composition Impacts on Infiltrate Phase Purity ................................ ................................ ................................ ............ 149 7.4 Summary ................................ ................................ ................................ ............... 151 viii CHAPTER 8: Determination of Infiltrated Mixed Ionic and Electronic Conducting Nano - Particle Oxygen Surface Exchange Material Properties through Finite Element Modeling of 3D Reconstructed Microstructures ................................ ................................ ............. 153 8.1 Introduction ................................ ................................ ................................ ........... 153 8.2 Experimental Methods ................................ ................................ .......................... 154 8.2.1 Cathode - Electrolyte - Cathode - Symmetric Cell Production ............................ 154 8.2.2 Electrochemical Impedance Spectroscopy Measurements ............................ 154 8.2.3 Nano - Micro - Composite Cathode Finite Element Modeling of 3D Reconstructions ................................ ................................ ................................ ....... 154 8.3 Results ................................ ................................ ................................ ................... 155 8.3.1 A Comparison of Finite Element Modeling Mixed Ionic Electronic Conducting Materials Intrinsic Oxygen Surface Exchange Material Properties ........................ 155 8.3.2 Identifying the Materials Property Combinations Causing the Surface Resistance Limit and the Simple Infiltration Microstructure Polarization Loss Estimation Model to Breakdown ................................ ................................ ............ 158 8.4 Summary ................................ ................................ ................................ ............... 161 CHAPTER 9: Dissertation Conclusions ................................ ................................ ......... 162 APPENDICES ................................ ................................ ................................ ................ 165 Appendix 1: Sim ple Infiltrated Microstructure Polarization Loss Estimation (SIMPLE) Model Derivation ................................ ................................ ...................... 166 Appendix 2: Focused Ion Beam Scanning Electron Microscopy (FIB - SEM) 3D Reconstruction and Modeling Instructions ................................ ................................ . 195 BIBLIOGRAPHY ................................ ................................ ................................ ........... 210 ix LIST OF TABLES Table 1: Desiccated TXD LSCF - GDC NMCC Cathodes ................................ ................. 71 Table 2 : Desiccated PND, TXD and CAD LSCF - GDC NMCC Cathodes ...................... 83 Table 3 : Pre - Infiltrated PND, TXD and CAD LSCF - GDC NMCC Cathodes ................ 116 x LIST OF FIGURES Figure 1 .1: Gravimetric and Volumetric Energy Densities for Electricity Generating Devices ................................ ................................ ................................ ................................ . 2 Figure 2.1 : Traditional LSM - YSZ|YSZ|YSZ - Ni SOFC Component Overpotential Comparison ................................ ................................ ................................ ......................... 5 Figure 2 .2 : Traditional SOFC Cathode and Anode Chemical Reactions ............................ 6 Figure 2 .3 : Current - Voltage and Power Density Curves for SOFC Devices ....................... 7 Figure 2 .4 : Traditional SOFC Cell Efficiency ................................ ................................ ..... 9 Figure 2 .5 : Representative Micro - Composite Cathode Microstructure ............................. 10 Figure 2 .6 : Representative Nano - Micro - Composite Cathode Microstructure with Connected MIEC Nano - Particles ................................ ................................ ....................... 10 Figure 2 .7 : SOFC SIMPLE Model Approximated Geometry ................................ ........... 20 Figure 2.8 : Infiltrated MIEC Nano - Particle Size for SSC, LSCF, and LSFC ................... 23 Figure 2 .9 : SOFC LSCF R s Measurement using Electrochemical Impedance Spectroscopy ................................ ................................ ................................ ...................... 27 Figure 2.10 : Typical Nano - Micro - Composite Cathode Equivalent Circuit Model ........... 28 Figure 2.11: Oxygen Concentration and Thermodynamic Factor Values for LSF, LSFC, LSCF, LSC, SSC and BSCF in Air Reported in Literature Studies ................................ .. 30 Figure 2.1 2 : k chem and k o Values Reported in Literature for the Cathode MIEC Materials LSF, LSFC, LSCF, LSC, SSC, and BSCF ................................ ................................ ........ 31 Figure 2.13: R s , Values Reported in Literature for Cathode MIEC Materials LSF, LSFC, LSCF, LSC, SSC, and BSCF ................................ ................................ ............................. 33 Figure 3 .1 : Uniaxial Press Device ................................ ................................ ..................... 36 Figure 3 .2 : Screen Printer Device ................................ ................................ ...................... 37 Figure 3 .3 : Representative X - ray Diffraction Data for Phase Pure LSCF Oxide Powder ................................ ................................ ................................ ............................... 38 xi Figure 3.4: Nano - Micro - Composite Cathode Electrochemical Impedance Measurement Device ................................ ................................ ................................ ................................ 40 Figure 3.5 : Scanning Electron Microscopy Component Schematic ................................ .. 41 Figure 3.6 : X - ray Diffraction Component Schematic ................................ ........................ 43 Figure 3.7 : Thermo - G ravimetric Analysis Schematic ................................ ....................... 45 Figure 3.8 : Profilometer Schematic ................................ ................................ ................... 46 Figure 3.9: Epoxy Coated Porous GDC Scaffold on a Dense GDC Electrolyte Oriented Inside the FIB - SEM ................................ ................................ ................................ ........... 53 Figure 3 .10 : FIB - SEM Cathode Backscatter Image used for 3D Reconstruction ............. 54 Figure 3 .11 : FIB - SEM 3D Reconstruction for Cathode and Electrolyte ........................... 55 Figure 3 .12 : FIB - SEM 3D Reconstruction Volume Mesh of Cathode and Electrolyte ................................ ................................ ................................ ................................ ................. 56 Figure 3 .13 : FIB - SEM 3D Reconstruction Electrochemical Potential Gradient ............... 58 Figure 4 .1 : Scanning Electron Microscope Fracture Surface I mage ................................ . 65 Figure 4 .2 : Scanning Electron Micrographs for Desiccated LSCF Nano - Particles Produced by using Triton X - 100 ................................ ................................ ........................ 68 Figure 4 .3 : Bar - Graph of Desiccated Average LSCF Infiltrate Particle Sizes from Scanning Electron Microscopy Images ................................ ................................ ............. 69 Figure 4 .4 : XRD Scans of LSCF Nano - Particles Produced by Firing Desiccated Precursor Nitrate Solutions ................................ ................................ ................................ ................ 70 Figure 4 .5 : Desiccated LSCF - GDC NMCC R P EIS Nyquist Plot ................................ ..... 72 Figure 4 .6 : Desiccated LSCF - GDC NMCC R P Arrhenius Plots ................................ ....... 73 Figure 4 .7 : Arrhenius R P Plot for Desiccated LSCF - GDC NMCCs Tested under Different Atmospheres ................................ ................................ ................................ ...................... 75 Fi gure 4 .8 : Arrhenius Ohmic Resistivity Plots for LSCF - GDC NMCCs Tested in Air .... 76 Figure 5 .1 : Scanning Electron Micrographs of Desiccated LSCF Nano - Particles Produced by using Different Triton X - 100 Solution Volumes ................................ .......................... 82 xii Figure 5 .2 : Scanning Electron Micrographs of Desiccated LSCF Nano - Particles Produced using Different Solution Additives ................................ ................................ .................... 84 Figure 5.3: Scanning Electron Micrographs of Desiccated LSCF Nano - Particles Produced usi ng Different Solution Additives ................................ ................................ .................... 85 Figure 5.4: Desiccated Williamson - Hall Raw Data Plot for PND LSCF, TXD LSCF and CAD LSCF Nano Particles ................................ ................................ ................................ 86 Figure 5.5: Strain Plots for Desiccated PND LSCF, TXD LSCF and CAD LSCF Nano - Particles ................................ ................................ ................................ .............................. 87 Figure 5 .6 : Bar - Graph of Average Desiccated LSCF Infiltrate Particle Size Produced using Different Solution Additives ................................ ................................ .................... 89 Figure 5 . 7 : Raw Impedance Data Plots of Desiccated LSCF - GDC NMCCs Produce using Citric Acid and Triton X - 100 ................................ ................................ ............................. 91 Figure 5 . 8 : LSCF - GDC R P Arrhenius Plots for Desiccated NMCCs Produced using Different Solution Additives ................................ ................................ .............................. 92 Figure 5 . 9 : Arrhenius Ohmic Resistivity Plots for Desiccated LSCF - GDC NMCCs Tested in Air and produced using Different Solution Additives ................................ ....... 93 Figure 5 . 10 : XRD Scans of De siccated LSCF Nano - Particles Produced using Different Solution Additives ................................ ................................ ................................ ............. 94 Figure 5 . 11 : XRD Scans for CaCl2 - Desiccated PND LSCF Fired between 80°C and 800°C ................................ ................................ ................................ ................................ . 95 Figure 5.12: XRD Scans for Dry Air - Desiccated TXD LSCF Fired between 80°C and 800°C ................................ ................................ ................................ ................................ . 96 Figur e 5.13: XRD Scans for CaSO 4 - Desiccated TXD LSCF Fired between 80°C and 800°C ................................ ................................ ................................ ................................ . 97 Figure 5.14: XRD Scans for CaCl 2 - Desiccated TXD LSCF Fired between 80°C and 800°C ................................ ................................ ................................ ................................ . 98 Figure 5.15 : XRD Scans for Dry Air - Desiccated CAD LSCF Fired between 80°C and 800°C ................................ ................................ ................................ ................................ . 99 Figure 5 .1 6 : XRD Scans for CaSO 4 - Desiccated CAD LSCF Fired between 80°C and 800°C ................................ ................................ ................................ ............................... 100 xiii Figure 5 .1 7 : XRD Scans for CaCl 2 - Desiccated CAD LSCF Fired between 80°C and 800°C ................................ ................................ ................................ ............................... 101 Figure 5.18: Williamson - Hall Raw Data Plots for Coarsened Undesiccated TXD LSCF Na no - Particle Sizes Produced at 600°C, 700°C and 800°C ................................ ............ 103 Figure 5.19: Strain Plots for Coarsened Undesiccated TXD LSCF Nano - Particle Sizes Produced at 600°C, 700°C and 800°C ................................ ................................ ............. 104 Figure 5 .20 : Coarsened Undesiccated TXD LSCF Average Nano - Particle Sizes Pro duced at 600°C, 700°C and 800°C ................................ ................................ ............................. 105 Figure 5 . 2 1 : LSCF - GDC 500 hour R P hour Plot for NMCCs Produced using Desiccation and Different Solution Additives ................................ ................................ ..................... 106 Figure 5.22: Desiccated Scanning Electron Micrographs for CAD and TXD LSCF - GDC Symmetric Cells Tested for 500 hrs ................................ ................................ ................. 108 Figure 5 . 2 3 : Raw Impedance Data Plots of Desiccated LSCF - GDC NMCCs Produced using Different Solution Additives ................................ ................................ .................. 109 Figure 6 .1 : Pre - Infiltrated Cathode Nano - Particle Fabrication Diagram ......................... 113 Figure 6.2: Scanning Electron Micrographs for Pre - Infiltrated TXD LSCF Nano - Particles Produced using Different Solution Molarities ................................ ................................ . 115 Figure 6.3 : Scanning Electron Micrographs for Pre - Infiltrated LSCF Nano - Particles Produced using Different Solution Volumes ................................ ................................ ... 117 Figure 6 .4 : Scanning Electron Mi crographs for Pre - Infiltrated LSCF Nano - Particles Produced using Different Solution Additives ................................ ................................ .. 118 Figure 6.5: Williamson - Hall Raw Data Plots Pre - Infiltrated for PND LSCF, TXD LSCF and CAD LSCF Nano - Particles ................................ ................................ ....................... 119 Figure 6.6: Strain Plots Pre - Infiltrated for PND LSCF, TXD LSCF and CAD LSCF Nano - Particles ................................ ................................ ................................ .................. 120 Figure 6.7: Williamson - Hall Raw Data Plots for Pre - Infiltrated Nano - GDC Particle Sizes in TXD LSCF and CAD LSCF Infiltrated Cells ................................ .............................. 121 Figure 6.8: Strain Plots for Pre - Infiltrated Nano - GDC Particles in TXD LSCF and CAD LSCF Infiltrated Cells ................................ ................................ ................................ ...... 122 Figure 6 .9 : Bar - Graph of Average LSCF Infiltrate Particle Size Produced using Pre - Infiltration and Different Solution Additives ................................ ................................ ... 124 xiv Figure 6.10 : Bar - Graph of Average Nano - GDC Infiltrate Particle Size Produce d using Pre - Infiltration and Different Solution Additives ................................ ............................ 125 Fi gure 6 . 11 : Raw Impedance Data Plots of Pre - Infiltrated LSCF - GDC NMCCs Prodcued using Citric Acid and Triton X - 100 ................................ ................................ ................. 127 Figure 6 . 12 : LSCF - GDC R P Arrhenius Plots for NMCCs Produced using Pre - In filtration and Different Solution Additives ................................ ................................ ..................... 128 Figure 6 . 13 : Arrhenius Ohmic Resistivity Plots for Pre - Infiltrated LSCF - GDC NMCCs Tested in Air using Different Solution Additives ................................ ............................ 129 Figure 6 .1 4 : XRD Scans for Pre - Infiltrated LSCF and GDC Nano - Particles with Different Solution Additives ................................ ................................ ................................ ........... 130 Figure 6 .1 5 : XRD Scans for 7.4 vol% Pre - Infiltrated PND LSCF Fired between 80°C and 800°C ................................ ................................ ................................ ............................... 131 Figure 6.16: XRD Scans for 7.4 vol% Pre - Infiltrated TXD LSCF Fired between 80°C and 800°C ................................ ................................ ................................ ............................... 132 Figure 6 .1 7 : XRD Scans for 7.4 vol% Pre - Infiltrated CAD LSCF Fired between 80°C and 800°C ................................ ................................ ................................ ............................... 133 Figure 6 .1 8 : Pre - Infiltrated TGA Plot using Different Solution Additives between 25°C and 850°C ................................ ................................ ................................ ........................ 135 Figure 6 .19 : LSCF - GDC 500 hour R P hour Plot for NMCCs Produced using Pre - Infiltration and Different Solution Additives ................................ ................................ ... 137 Figure 6.20: Pre - Infiltrated Scanning Electron Micrographs for CAD and TXD LSCF - GDC Symmetric Cells Tested for 500 hrs ................................ ................................ ....... 138 Figure 6 .21 : Raw Impedance Data Plots of Pre - Infiltrated LSCF - GDC NMCCs Produced using Different Solution Additives ................................ ................................ .................. 139 Figure 7 .1: Raw Impedance Data Plots of Desiccated and Pre - Infiltrated La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - ................................ ................................ ....................... 144 Figure 7 .2: Desiccated or Pre - Infiltrated La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - P Arrhenius Plots for N MCCs Produced using Citric Acid ................................ ................................ . 145 Figure 7 .3: Desiccated or Pre - Infiltrated La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - - Particle Sizes Produced using Citric Acid ................................ ................................ .................... 147 xv Figure 7 .4: La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - Nano - Particle Sizes Determined from Scanning Electron Microscopy Images ................................ ................................ ........................... 148 Figure 7 .5: Arrhenius Ohmic Resistivity Plots for Desiccated or Pre - Infiltrated La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - Air ................................ ................................ ................................ ................................ .... 149 Figure 7 .6 : XRD Scans for CaCl 2 - Desiccated La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - using Citric Acid ................................ ................................ ................................ .............. 150 Figure 7 .7: XRD Scans for 7.4 vol% Pre - Infiltrated La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - Produced using Citric Acid ................................ ................................ .............................. 151 Figure 8 . 1 : R s , k chem and k o values Reported in Literature and Calculated using FIB - SEM FEM 3D Reconstructions for the Cathode MIEC Materials LSF, LSFC, LSCF, LSC and SSC ................................ ................................ ................................ ................................ .. 157 Figure 8 . 2 : R s , k chem and k o Values Calculated using the FIB - SEM FEM 3D Reconstruction for Cathode MIEC Mate rials LSF, LSFC, LSCF, LSC and SSC ........... 158 Figure 8. 3 : Calculated R P Values from the FIB - SEM FEM 3D Microstructure, SIMPLE Model and Surface Resistance Model Determined for Different R ct / Scaffold GDC Conductivity Ratio Values ................................ ................................ ............................... 160 Figure 9.1: Infiltrated Cell Comparison from Different Infiltration Groups ................... 163 Figure A1.1: An I dealized Representation of a Symmetric SOFC Cathode C ell ........... 167 Figure A1.2 : Oxygen Transport Proof ................................ ................................ ............ 171 Figure A1.3 : Electrode Polarization Resistance Proof ................................ ................... 174 Figure A1.4: Repeat Unit with Numbered Interfaces Across with Current Flow .......... 178 Figure A1.5 : ................................ ................................ . 178 Figure A1.6 : Determining ................................ ................................ ........................... 179 Figure A1.7 : Determining ................................ ................................ ........................... 181 Figure A1.8 : Determining ................................ ................................ ........................... 182 Figure A1.9 : Determining ................................ ................................ ........................... 183 Figure A1.10 : Applying the 1 st Boundary Condition Locally ................................ ......... 184 xvi Figure A1.11 : Applying the 2 nd Boundary Condition to Solve for ............................ 188 Figure A1.12 : Applying the 1 st Boundary Condition Across the Entire Cathode to Solve for ................................ ................................ ................................ ...................... 190 Figure A1.13 : Solving for the Cathode Polarization Resistance ................................ .... 192 Figure A1.14 : Altering the TFV Equation ................................ ................................ ...... 193 Figure A2.1 : Representative 2D FIB - SEM Serial Section Stacked Images used for 3D Reconstruction ................................ ................................ ................................ ................. 195 Figure A2.2 : 3 - Matic Imported 3D Microstructure ................................ ........................ 196 F igure A2.3 : 3 - Matic Cathode and Electrolyte Merged Microstructures ....................... 198 Figure A2.4 : COMSOL Volume M esh Import S creen ................................ ................... 200 Figure A2.5 : GDC Conductivity Assigned to Cathode and Electrolyte Microstructure Volume Mesh ................................ ................................ ................................ ................... 201 Figure A2.6 : Scaled R s Surface Impedance Assigned to the Cathode Surface Layer Mesh ................................ ................................ ................................ ................................ . 203 Figure A2.7 : 0V Reference P otential Applied to the Electrolyte Surface Layer Mesh ................................ ................................ ................................ ................................ . 204 Figure A2.8 : FIB - SEM 3D Reconstruction Electrochemical Potential Gradien t ........... 20 5 Figure A2.9 : FIB - S EM 3D Reconstruction Electrochemical Potential Gradient Multislice Plot ................................ ................................ ................................ ................................ ... 206 Figure A2.10 : Cutplane used to Integrate Current Density Across Electrolyte Surface Mesh ................................ ................................ ................................ ................................ . 208 Figure A2.11 : COMSOL Integration Area Calculation ................................ .................. 209 xvii KEY TO ABBREVIATIONS AC=Alternating Current BSCF= Barium Strontium Cobalt Iron Oxide ( Ba 0. 5 Sr 0. 5 Co 0.8 Fe 0.2 O 3 - ) CAD= Citric Acid Derived DC=Direct Current EIS=Alternating Current Electrochemical Impedance Spectroscopy EISA=Evaporation Induced Self Assembly FIB =Focused Ion Beam FEM= Finite Element Modeling GBCO=Gadolinium Barium Cobalt Oxide (GdBaCo 2 O ) GDC =Gadolinia Doped Ceria Oxide (Gd 0.1 Ce 0.9 O 1.95 ) IC=Ionic Conducting k chem = Chemical Rate Constant into the L attice k o =Chemical Rate Co nstant through the B ulk L C = Characteristic Thickness LNO=Lanthanum Nickel Oxide (La 2 NiO 4 ) LSC=Lanthanum Strontium Cobalt Oxide ( La 0.6 Sr 0.4 CoO 3 - ) LSCF= Lanthanum Strontium Cobalt Iron Oxide (La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - ) LSCF55=Lanthanum Strontium Cobalt Iron Oxide (La 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3 - ) LSF=Lanthanum Strontium Iron Oxide ( La 0.6 Sr 0.4 FeO 3 - ) LSFC=Lanthanum Strontium Iron Cobalt Oxide ( La 0.6 Sr 0.4 Co 0. 2 Fe 0. 8 O 3 - ) LSGM=Lanthanum Strontium Gallium Magnesium ((La,Sr)(Ga,Mg)O 3 ) xviii LSM= Lanthanum Strontium Manganese Oxide (La 0.6 Sr 0.4 MnO 3 - ) MIEC=Mixed Ionic Electronic Conducting MCC= Micro - Composite Cathode NMCC=Nano - Micro - Composite Cathode OCV=Open Circuit Voltage PND =Pure Nitrate Derived = Cathode Polarization Resistance R s = Oxygen Surface Exchange Resistance SDC=Samaria Doped Ceria Oxide (Sm 0.2 Ce 0.8 O 1.9 ) SEM=Scanning Electron Microscopy SIMPLE=Simple Infiltrated Microstructure Polarization Loss Estimation SOFC=Solid Oxide Fuel Cell SR=Surface Resistance SSC=Samarium Strontium Cobalt Oxide ( Sm 0. 5 Sr 0. 5 CoO 3 - ) SSCF=Samarium Strontium Cobalt Iron Oxide ( Sm 0. 5 Sr 0. 5 Co 0.8 Fe 0.2 O 3 - ) TEC=Thermal Expansion Coefficient TGA=Thermo - Gravimetric Analysis TPB=Triple Phase Boundary TXD= Triton X - 100 Derived XRD=X - ray Diffraction YSB=Yttria Stabilized Bismuth (Y 0.25 Bi 0.75 O 1.5 ) YSZ=8 mol% Yttria Stabilized Zirconia ( (Y 2 O 3 )0.08 ( ZrO 2 ) 0.92 ) 1 CHAPTER 1 : Thesis Motivation and Overview 1.1 The World Energy Problem According to the U.S. Census Bureau, the world population in 2012 was approximately 7 billion people and is projected to continue to increase by approximately 6.5 million peop le each month on average [ 11 ] . A ccording to the Energy Information Administration, world energy consumption for fossil fuels has in creased from 354 quadrillion BTU s in 1990 to 505 quadrillion BTU s in 2008 and is projected to increase to 770 q uadrillion BTU s by 2035 [ 12 ] . This increasing world ene rgy demand stimulate s the need for the development o f alternative energy sources and/or improve d energy conversion devices, such as Solid Oxide Fuel Cells (SOFCs). 1.2 Solid Oxide Fuel Cell s as a Solution to the World Energy Problem As demonstrated in Figu re 1.1 [ 13 - 15 ] , SOFCs have some of the highest gravimetric and volumetric power densities of any electricity generating technology . SOFCs also have the ability to operate on a variety of fuel types (hydrogen, ethanol, biofuel, gasoline, natural gas, syngas, landfill gas, jet - fuel, etc.) [ 15 , 16 ] . These benefits provide SOFCs with the capability to both reduce the environmental impact s hydrocarbon based eco nomy while simultaneously developing the infrastructure for a CO 2 - neutral economy utilizing biofuels, solar fuels or hydrogen. SOFCs can also be used for : 1) chemical separation, 2) chemical sensing, and 3) energy storage when operated in reverse as Solid Oxide Electrolysis Cells [ 17 - 19 ] . 2 SOFCs demonstrate high electrical conversion efficiencies greater than tra ditional electricity generating technologies . For example , typical coal - fired electricity generating plan ts have demonstrated electrical efficiencies around 46% [ 20 ] compared to 60% for SOFCs [ 21 , 22 ] . When both heat and electricity are valued , coal - fired power plants have demonstrated efficiencies near 60% [ 20 ] , compared to 90% for SOFCs [ 23 , 24 ] . SOFC efficiencies are also size independent, ( a feature not shared by many chemical to electrical conversion devices such as gas turbines ) , which allows SOFCs to rang e from 1 Watt to multi - Megawatt [ 25 , 26 ] . Commercial SOFC devices using traditional macro - porous cathodes typically operate at temperatures above 750°C. However, in the past two decades Nano - Micro - Composite Cathodes ( NMCCs ) have been shown to increase cath ode performance ( i.e. lower the polarization resistance, R P ) and/or reduce the SOFC operating temperature compared to traditional Micro - Composite cathodes . In addition, many studies have sought to experimentally improve [ 5 , 27 , 28 ] and /or mathematical ly model NMCC Figure 1 .1: Gravimetric and Volumetric Energy Densities for Electricity Generating Devices . Modified from [13 - 15 ]. 3 performance [ 5 , 29 ] . Unfortunately, two knowledge gaps ar e currently preventing additional NMCC performance improvements . 1.3 Solid Oxide Fuel Cell Knowledge Gaps The first knowledge gap is the inability to control the size of infiltrated M ixed I onic and E lectron C onducting (MIEC) nano - particles [ 5 , 7 , 30 ] . Small MIEC nano - particle diameter s allow for increased ox ygen exchange into the cathode microstructure due to an increased MIEC surface area to volume ratio. Unfortunately, most infiltrated NMCCs exhibit average particle sizes around 4 0 nm or greater [ 5 , 7 , 30 ] , and p revious literature studies have not identified a reliable method to further reduce infiltrated MIEC particle sizes . The second knowledge gap is an inability to accurate ly determine intrinsic NMCC MIEC oxygen surface exchange resistance ( R s ) values. Previous literature studies have only determined R s values on thin film and bulk MIEC materials and the surface structure and/or stress state of MIEC infiltrate could be very different . Further, the magnitude of the R s value reported in these thin film and bulk studies can vary by as much as 3 orders of magnitude f or the same material composition [ 10 ] . An accurate set of MIEC infiltrate R s numbers are needed to determine the best MIEC infiltrate material and to perform cathode microstructural optimization (such as that done in Song et al . [ 31 ] ). 1.4 Thesis Overview This thesis will address the se two knowledge gaps . Specifically, Chapter 2 will delve into a literature review of SOFC devices and explore the status of research today. Chapter 3 will address the experimental methods and characterization techniques used to conduct the ex periments in this thesis . Chapters 4 - 7 will illustrate two methods, 4 desiccation (aka precursor nitrate solution desiccation) and pre - infiltration (aka ceria oxide pre - infiltration) , that were developed to systematically control infiltrated MIEC nano - partic le size. T he effect of organic solution additives and MIEC composition are also explored in these chapters to examine their effect on desiccated and pre - infiltrated particle sizes. Chapter 8 will show infiltrated MIEC R s determinations made by performing Finite Element Modeling (FEM) on Focused Ion Beam Scanning Electron Microscopy (FIB - SEM) 3D reconstructed SOFC cathodes . Lastly, Chapter 9 will provide a set of overall conclusion s and restate the advances this work has provided to the scientific communi ty. 1.5 Summary In summary , SOFC s are a promising alternative e nergy conversion technology because they : 1) have one of the highest gravimetric and volumetric power densities of any electricity generating technology , and 2) have the capability to both reduce environmental impact s nomy while also one day serving as the infrastructure for a CO 2 - neutral economy utilizing biofuels, solar fuels or hydrogen . Com mercialization of this technology is held back by poor perfo rmanc e at low operating temperatures, especially on the cathode side. NMCCs help improve the performance at low operating temperatures but additional performance increases are being restrained by : 1) a lack of control of infiltrate nano - particle size , and 2) a l ack of accurate infiltrated MIEC R s values . The work presented in this th esis advances SOFC technology by addressing these two knowledge gaps. 5 CHAPTER 2 : Literature Review 2.1 Solid Oxide Fuel Cell Overview Even though SOFC performance improvements have occurred through the use of NMCCs [ 1 , 5 ] there still exist s the need to control infiltrated MIEC nano - particle size and understand the R s va lues of infiltrated MIEC materials . As pointed out in Section 1.3 addressing these two limitations will great ly improve SOFC devices by allowing the operating temperature to be l owered, which in turn will limit the performance degradation caused by nano - particle size coarsening . As shown in Figure 2.1 [ 32 ] , the cathode is the most resistive component of a traditional LSM - YSZ|YSZ|YSZ - Ni SOFC device (the resistance can be calculated from the slope of each curve and clearly the cathode has the greatest resistance at low and moderate currents) . The greatest source of resistance in the cathode comes from limitations in oxygen incorporation into the MIEC lat tice structure [ 5 , 33 ] ; hence the need for smaller MIEC n ano - particles . Further, r eliable measurements of the intrinsic oxygen surface exchange resistance process, R s , of the MIEC infiltrate are needed to Figure 2 .1 : Traditional LSM - YSZ|YSZ|YSZ - Ni SOFC Component Overpotential Comparison . Resistance values depicted as the s lope of each Polarization curve [32 ] . 6 accurately select the best MIEC material to infiltrate , and for SOFC m icrostructure optimization [ 31 ] . This chapter will provide an overview of SOFC operating principles and discuss how infiltrate particle size and infiltrate R s values play a role in determining SOFC performance. 2.1.1 Solid Oxide F uel Cell Operating Principles Figure 2 .2 [ 34 ] shows the three components of a n SOFC , which are the: 1) cathode, 2) electrolyte, and 3) anode. Figure 2.2 also demonstrates the chemical reactions and/or the transport of charged species that take place in each component [ 34 ] . The cell shown in F igure 2.2 yielded the performance curves shown in Figure 2 .1 . A t the cathode , oxygen gas is first incorporated into oxygen vacancies located in the MIEC latti ce structure and then transported to the anode through the electrolyte . At the anode, the fuel ( hydrogen , methane, etc. ) rips oxygen out of the anode MIEC c rystal structure to form oxygen vacancies, water and electrons . The resulting gradient in oxygen vacancies and electrons across the cell drives t he flux of these species . An external circuit conducts the electrons (thus creating electricity) Figure 2 .2 : Traditional SOFC Cathode and Anode Chemical Reactions . Modified from [34 ] . 7 from anode t o cathode . The electrolyte acts as a gas barrier and prevents fuel cross - over to the cathode or gaseous O 2 transport to the anode. Operational SOFCs have a thermodynamically determined open circuit voltage (OCV) (which is ~1.2 V at SOFC operating tempera tures) [ 35 ] and generate power through the flow of electrons shown in Figure 2.2. Note that to balance the reaction , for a given number of electrons to flow through an external circuit , half that number of oxygen vacancies must flow across the electrolyte. Therefore, high rates of oxygen ion transport are critical for the development of high power SOFCs. Figure 2.3 shows that t he SOFC operating voltage (V) can be plotted against the operating cur rent (I) to form an I - V plot , and the power can also be plotted against current to form a power density plot. Both of these plots are useful to help characterize SOFC performance. F igure 2 .3 : Current - Voltage and Power Density Curves for SOFC Devices . Region 1 is below 0.6 A/cm 2 , Region 2 is between 1.6 and 0.6 A/cm 2 and Region 3 is above 1.6 A/cm 2 [35 ] . 8 Figure 2.3 [ 35 ] shows the current - voltage (I - V) curve and power densit y curve for a typical SOFC which provides an idea of how SOFCs operate. The slope of the I - V curve is the sum of all three resistances (cathode, electrolyte and anode) . Region 1 is where the electrical losses are dominated by oxygen incorporation into or removal from the cathode or anode lattice structure , respectively. Region 2 is where electrical losses are dominated by ohmic losses (resistances originating from the electrolyte and electrode), and R egion 3 is where R P losses occur due to the current density being so large that the SOFC consumes more oxygen than can diffuse through the pores in the cathode microstructure or hydrogen than can diffuse through the pores in the anode microstructure; a condition referred to as gas - phase polarization. The slope in Region 1 is the sum of the cathode and anode polarization resistances. However, in common SOFCs , this slope reduces to the cathode polarizatio n resistance , R P , because the cathode performance is so much worse than the anode performance. Hence, Re gion 1 is where NMCC infiltrate nano - particle size control and an understanding of cathode MIEC R s valu es will have the largest impact. Figure 2.4 shows that SOFCs are not operated at the ir maxi mum power densities because as th e output power density increase s , the cell efficiency decrease s [ 36 ] . I nstead , SOFCs are operated at current densities to the left of the maximum power density value [ 36 ] . The cell efficiency is calculated by dividing the cell power by the enthalp y of combustion of the fuel [ 36 ] . 9 One possible approach to lower the cathode R P and increase SOFC pe rformance is to fabricate NMCCs . A lower R P value will decrease the initial slope of the I - V curve which also results in an increase in the maximum power obtainable at a given efficiency . 2.1.2 Traditional and Nano - Micro - Composite C athodes NMCCs are a type of SOFC cathode that combine materials with low R s values (MIEC materials) and materials with high ionic conductivity , such as Gd 0.1 Ce 0.9 O 1.95 ( GDC) [ 7 ] or Y 2 O 3 - ZrO 2 (YSZ) [ 16 ] . NMCCs are produced by first infiltrating metal nitrate so lutions into por ous ionic conducting ( IC ) scaffold s , and then firing the gelled solutions at elevated temperatures to form MIEC oxide nano - particles upon decomposition of the nitrate solution s [ 5 ] . The t echniques used to fabricate MIEC nano - particles are described in more detail in Chapter s 3 - 7 . Both t raditional Micro - Composite Cathodes and N MCC s have interconnected particles which allow electronic and ionic transport , but only NMCC s have nano - scale MIEC particles which allow for increased oxygen incorporation into the MIEC latt ice due to an increase in MIEC surface area ( shown in Figures 2.5 [ 37 ] and 2 . 6 [ 7 ] ) . Traditional Micro - C omposite C athodes have Figure 2 . 4 : Traditional SOFC Cell Efficiency . Cell efficiency decreases as power output increases. [36 ]. 10 MIEC and IC particles sizes that are tens of micrometers in size while NMCCs have particles that are tens of nanometers in size . This small MIEC particle size causes increased cata lytic activity [ 38 ] resulting in lower cathode operating temperatures ( defined here as the cathode operating temperature where R P values equal 2 [ 1 , 5 , 9 ] ) . By lowering the operating temperature of N MCC s , MIEC particle coarsening can also be limited . Particle c oarsening reduces SOFC performance by decreasing the n umber of active surface sites for oxygen incorporation into the MIEC lattice structure by decreasing the MIEC surface area to volume ratio. In a NMCC, t he low values of the Figure 2 .5 : Representative Micro - C omposite Cathode Microstructure . White components are ionic conducting material is the black component is MIEC material. [37 ]. F igure 2 .6 : Representative N ano - M icro - C omposite C athode Microstructure with Co nnected MIEC Nano - Particles . Nano - particles a re MIEC materials that have been infiltrated [7 ]. 11 MIEC infiltrate particles (which facilitates oxygen surface exchange), the high characteristic thickness ( ) of the MIEC infiltrate particles (which ensures that the oxygen exchange reaction takes place over the entire MIEC infiltrate particle surface), and the high MIEC infiltrate surface area (which provides many locations for the oxygen surface exchange reac tion) , work together with a high IC scaffold (which facilitates oxygen transport to the electrolyte) , and a high MIEC electronic conductivity (which facilitates electron transport to the current collectors) to produce cathodes reported to reach polarizatio n resistances of 2 at temperatures as low as 550 C [ 1 ] . 2.1.3 Thin Film Cathodes Thin film SOFC electrodes are another area of research . The se electrodes can be used as part of SOFCs or as model electrodes to better understand or measure SOFC material properties . Thin film electrode s are typically dense, single phase , substrate supported films , with thicknesses less than 100 nm. Thin film electrodes can be fabricated through many different approaches such as radio frequency s puttering, DC sputtering, vacuum plasma spraying, pulsed laser deposition [ 39 - 45 ] , etc . Using thin films , m aterial properties such as R s and the bulk oxygen conductivity can be measured, which has been shown numerous times in the literature using both cathode and anode materials [ 46 - 52 ] . Compared to traditional porous electrode s , t he geometry o f a thin film is less complex since only a single, dense layer on top of a substrate is present. This allows for easily known surface and cross - sectional areas and hence easy conversion of a measured resistance into area - corrected intrinsic materials properties [ 53 , 54 ] . For this reason many authors have performed MIEC R s measuremen ts using t hin film micro - electrodes [ 47 , 53 , 54 ] . These thin film R s 12 measurements are often performed using the Electrochemical Impedance Spectroscopy (EIS) characterization technique. However, since thin film measurements may not accurately describe the R s of infiltrated MIEC particles , (due to differences in surface structure , stress states , etc. ) further analysis of infiltrated MIEC R s values is needed . Unfortunately, no studies containing this information have ever been reported in the literature. 2.2 Plana r Solid Oxide Fuel Cell Geometric Arrangements T hree different types of SOFCs are described in the literature : 1) cathode supported, 2) anode support ed , and 3) electrolyte supported SOFCs. The mechanically supporting layer is typically the thickest (and typically the first to be fabricated) . In c athode supported cell s the cathode layer thickness is typically several hundred microns thick , while the anode and electrolyte layers are typically much thinner ( ~ 20 nanometers to 100 microns) [ 55 ] . Typically cathode supported cells are not used for SOFC devices because increasing the thicknes s of the cathode layer will not increase the performance of the SOFC due to the acti ve area of the cathode being set to less than ~ 50 microns by the typical MIEC infiltrate R s and ionic conductivity of the scaffold materials [ 51 ] . A node s upported cells have anodes that are much thicker than either the cathode or electrolyte . Like cathode supported cells, t hese cells have the advantage of using a thin ner electrolyte, which minimizes ohmic resistance performance losses [ 56 ] . In these cells , t he cathode is typica lly fabricated after the anode and electrolyte , which prevents the cathode materials from seeing the higher fabrication temperatures used in process ing the anode and electrolyte . This promotes less coarsen ing of the cathode MIEC nano - 13 particles [ 57 , 58 ] tha n cathode supported cells . Anode supported SOFCs are typically used for commercial SOFC devices because the anode R s value is typically much lower (an order of magnitude or lore lower) than cathode R s values [ 59 , 60 ] due to increased oxygen exchange in the anode material compared to the cathode material. This means that the electrochemically active region in the anode i s larger than the cathode. Since the optimal electrochemically active region is much larger than the cathode , the anode is often made the mechanical support. Electrolyte supported SOFCs have electrolytes as the thickest component , which are typically hundreds of microns thick [ 5 , 7 ] as opposed to the cathode or anode electrodes which are thinner [ 56 ] . I n these cells the electrolyte is typically fabricated first , with the result that any cathode and/or anode nano - particles present do not coarsen significantly in response to a high electrolytic firing temperature . Since the electrolyte is the thickest com ponent it also will have increased ohmic resistance performance losses , preventing their use in commercial SOFCs . However, e lectrolyt e supported SOFCs are often used in labora tory experiments when electrode performance, not total SPFC performance is of in terest [ 5 , 7 , 61 ] . 2.3 Commonly U sed Solid Oxide Fuel Cell Materials Chapter 1 pointed out that one key knowledge gap w as a lack of und erstanding of the infiltrated MIEC R s values for the cathode . In fact, because of this knowledge gap, the SOFC community has no clear consensus on which material makes the best MIEC cathode in filtrate (cathode microstructural effects make it difficult to judge MIEC performance from the cathode R P ) . Since this thesis focuses on improving the cathode 14 performance , the following sections discuss the various MIEC and IC materials found in literature and their roles in improving NMCC performance . 2.3.1 Ionic Conductors For the electrolyte there currently exist a number of materials used to promote bulk oxygen transport from the cathode to the anode, including : 1) Bi 2 V 0.9 Cu 0.1 O 5.35 (BICVOX) [ 62 ] , 2 ) Ce 0.9 Gd 0.1 O 1.95 (GDC) [ 7 ] , 3) La 0.9 Sr 0.1 Ga 0.8 Mg 0.2 O 2.85 (LSGM) [ 63 ] , 4) Sc 2 O 3 - ZrO 2 (ScSZ) [ 64 ] , and 5) (ZrO 2 ) 0.9 (Y 2 O 3 ) 0.1 (YSZ) [ 65 ] . Due to it s combination of high ionic conductivity and chemical compatibility with most MIEC materials, doped ceria is commonly used as an ionic conductor in SOFC cathodes [ 66 ] . 2.3.2 Electron ic Conductors Typically a layer is applied on top of the electrode , named t he current collection layer. This high electronic conductivity layer promote s a uniform flow of electrons across the entire electrode surface . Commonly used current collector materials include : La 0.6 Sr 0. 4 TiO 3 - x (LST) [ 67 , 68 ] , and La 0.8 Sr 0.2 MnO 3 - x (LSM) [ 5 ] . 2.3.3 Mixed Ionic Electronic Conductors Section 2.1.1 showed that oxygen in an SOFC is exchanged from the gas phase into the oxygen vacancies in the MIEC lattice structure in the cathode, and is then transported through the electrolyte to the anode. Currently a number o f different MIEC materials have been used as cathodes or infiltrated into cathode microstructures to promote oxygen incorporation. The most commonly used MIEC materials include : La 0. 6 Sr 0.4 FeO 3 - x (LSF), La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - x (LSFC) , La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - x (LSCF) , La 0.6 Sr 0.4 CoO 3 - x (LSC) , Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3 - x (BSCF) , and Sm 0.5 Sr 0.5 CoO 3 - x (SSC) [ 54 ] . 15 2.4 Composite Solid Oxide Fuel Cell Cathode Geometries Section 2.2 discussed that there are a number of SOFC geometries . However, there are also a number of cathode geometries as well . The followin g are different cathode geometries that exist in literature w hich utilize different MIEC and ionic conducting materials . The cathode scaffold (into which the infiltrate precursor solutions are infiltrated and to which the MIEC particles cling) c an be manufactured from either : 1 ) electronic conducting materials (suc h as La 0.6 Sr 0.4 MnO 3 - ) infiltrated with an ionic conducting material, 2 ) ionic conduc ting materials (such as Yttria Stabilized Z irconia) infiltrated with electronic conducting or mixed - conducting materials (which all have >99% of the conductivity resulting from the transfer of electronic, not ionic, species) , or 3 ) mixed - conducting materials (such as La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - ) and infiltrated with electronic, ioni c or mixed - conducting materials . A composite cathode allows for the combination of multiple mat erials that each do something well, so the entire cathode can benefit from both materials. A s shown in Figure 2.2, electrons, oxygen vacancies, and oxygen gas must all be transported to support the c athode reaction . For instance, i n the composite cathode of Figure 2.6, the MIEC material nano - particles transports electrons and incorporates oxygen gas into the MIEC lattice structure, while the GDC ionic conducting scaffold material transports the oxygen vac ancies through the scaffold . 2.4.1 Electronic Con ducting Scaffolds Cathodes fabricated using an electronic conducting scaffold typically utilize LSM screen printed onto an ionic conducting electrolyte pellet (such as YSZ) and fired to form a porous electrode [ 69 - 71 ] . I onic m aterials are then typically infiltrated into the porous 16 electrode to form nano - particles on the electronic conducting surface [ 69 - 71 ] . An advantage of an electronic conducting scaffold is that it facilitates oxygen incorporation into the ionic conductor at the triple - phase - boundary where the gas, electronic conductor and ionic conductors all meet, and has a high electronic conductivity. The infiltration of electronic or mixed - conducting nano - particles into the electronic scaffold is typically not perfor med because, as discussed in Section 2.3, both materials have low ionic conductivity [ 72 ] and hence no path would exist in the cathode for ionic conducti on from the cathode to the electrolyte . Another d isadvantages of using electronic cond ucting scaffolds is that t here is the problem of thermal expansion mismatch between the electronic conducting scaffold and the electrolyte material, which, if different enough, cause cracks to form and damage cell performance [ 28 ] . Since the scaffold material has a low R s value and high electronic conductivity, but a low ionic conductivity, these reasons make electronic conducting scaffolds a poor choice to use for the SOFC cathode geometry. 2.4.2 Mixed Ionic Electronic Conducting Scaffolds Cathodes with MIEC conducting scaffolds are typically made by screen printing an MIEC layer onto an ionic conducting electrolyte and firing the cell to form a porous MIEC electrode [ 73 , 74 ] . An electronic conducting, ionic conducting or mixed - conducting material is then infiltrated onto t he mixed - conducting scaffold surface. MIEC scaffold s have the advantage of high electronic conductivity, low R s values, but unfortunately have an ion ic conductivity much lower than IC scaffolds such as GDC [ 66 , 72 ] . 17 There are two d isadvantages to using a mixed - ionic electronic conducting scaffold geometry . The first is that there can be a thermal expansion mismatch between the MIEC scaffold and the IC electrolyte material, which, if large enough, can cause cracks to form and reduce the cell performance. MIEC thermal expansion coefficients (for temperatures between 30 - 1000°C) are between ~17.5*10 - 6 K - 1 (LSFC) [ 72 ] , and ~25*10 - 6 K - 1 (LSC) [ 72 ] , while the thermal expansion coefficient of IC scaffold materials (for t emperatures between 30 - 1000°C) are between ~10*10 - 6 K - 1 (GDC) [ 75 ] and ~13*10 - 6 K - 1 (GDC) . The second disadvantage is that the oxygen ion diffusion through MIEC material s is much lower than traditional IC materials such a s YSZ or GDC . This means that IC materials must be infiltrated into the MIEC scaffold to transport oxygen to the electrolyte. Unfortunately this reduces the MIEC surface area available for oxygen exchange. 2.4.3 Ionic Conducting Scaffolds Cathodes IC scaffolds are typically made by screen printing an ionic conducting layer onto an ionic conducting electrolyte and firing the cell to form a porous IC electrode [ 5 , 61 ] . An electronic conducting (such as La 0.6 Sr 0.4 MnO 3 - ) [ 76 ] or mix e d - conducting (such as La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - or Sm 0. 5 Sr 0. 5 CoO 3 - ) [ 5 , 61 ] material ar e then typically infiltrated into the ionic conducting scaffold structure . The advantage of infiltrating an MIEC into an IC scaffold is that the oxygen incorporation reaction into the lattice structure is spread over the entire MIEC surface and not just l imited to the triple phase boundary [ 7 ] . Another advantage to using an ionic conducting scaffold is that there is no thermal expansion mismatch between the scaffold and electrolyte (assuming 18 the same IC material is u sed for both) [ 28 ] . Hence MIEC infil trated IC scaffolds were examined in this these. 2.5 Models Identifying the Parameters Controlling N ano - M icro - C omposite C athode Performance Since oxygen surface exchange into the cathode typically limits overall performance [ 5 , 77 ] , b eing able to reduce infiltrate oxide nano - particle size is very important to ultimately achieve lower R P values at lower operating temperatures. In addition, b eing able to model the impact that different parameters (such as R s values , infiltrate nano - particle size , the scaffold ionic conductivity, the cathode thickness, the cathode porosity, etc. ) have on cathode performance is also important. Two of the most widely used MIEC on IC NMCC performa nce model s in the literature are the Surface Resistance (SR) model [ 5 , 7 , 61 ] , and Simple Infiltration Microstructure Polarization Loss Estimation (SIMPLE) model, [ 5 , 7 , 61 , 78 ] . B oth models quantify the performance increases possible by varying infiltrated nano - particle size and infiltrate MIEC R s values , but the SIMPLE model also partially account s for IC scaffold effects. 2.5 .1 The Surface Resistance Model T he Surface R esistanc e (SR) model predict s the SOFC cathode performance using the equation: [1 ] where R P is the polarization resistance of the cathode , R s is the intrinsic MIEC oxygen surface exchange resistance, A G is the geometric area (i.e. footprint) of the cathode , and A Inf is the surface area of the MIEC infiltrate particles [ 7 ] . The SR model ignores any 19 resistance caused by the ionic conducting scaffold , assuming instead that the R P is just linearly related to the MIEC R s and the MIEC surface area in a specific unit of cathode . Oftentimes this is a good assumption because : 1) the R S for oxygen incorporation into the IC scaff old is typically much higher than the R S of the MIEC material (meaning that oxygen exchange directly into the IC scaffold is limited), and 2) the resistance associated with bulk transport through the IC scaffold is typically low . That said, a s the operati ng temperature increases, the SR model deviates more from the experimental results because the IC ionic transport resistance starts to become a larger contributor to the R P value . Due to the fact that only the oxygen surface resistance is accounted for , l iterature studies have shown that this model is typically not as effective as the SIMPLE Model at accurately predicting cathode performance values [ 7 ] . 2.5 .2 The Simple Infiltration Microstructure Polarization Loss Estimation Model The SIMPLE model takes into account both the surface resistance of oxygen incorporation into the MIEC lattice and the bulk resistance of oxygen transport to the electrolyte material . The SIMPLE M odel is also simple in that it is an anal ytical expression that can be evaluated quickly (in contrast to the more rigorous finite elemen t modeling methods) , and can typically predict the R P values of cathodes made from within 33% of experimentally measured val ue s [ 7 ] . Figure 2.7 [ 7 ] depicts the approximate d SOFC N M CC microstructure geomet ry used to predict the overall R P value of N MCCs using the SIMPLE m odel. Th e actual 20 microstructure of a NMCC is complex, but here it is modeled as a symmetrical N M CC with repeating unit cell columns . S ymmetrical cells are N MCC s that have both scaffold microstructures fabricated as if they w ere cathodes on each side of a mechanically - supporting electrolyte. The benefit of experimentally t esting symmetrical cells, instead of a whole SOFC , is that the individual open - circuit electrode performance values, such as each cathode, can be obtained by dividing the total measured electrode response by 2 . In the SIMPLE model , t he cathode thickness i s L , the column within the unit cell is w , the unit cell width is r , and the electrolyte thickness is d , as labeled in Figure 2.7 . The SIMPLE mode assumes that t he surface of each IC column is heavily coated with an MIEC electrocatalyst infiltrate which h as an R s va lue associated with the MIEC material s ability to incorporate oxygen into its lattice structure . Having the MIEC heavily infiltrated onto the surface of the electrode also promotes electronic contact within the current collector , helping ensure that all the MIEC particles are electrochemically active. This, coupled with the high electronic conductivity of the MIEC infiltrate and the low current ( i.e. open - circuit) conditions, also place each infiltrate particle surface (i.e. eve ry MIEC - gas interface) at the same electrochemical Figure 2.7 : SOFC SIMPLE Model Approximated Geometry . SIMPLE model approximated geometry is for a symmetrical cell that contains two cathodes in this example [7 ]. 21 potential . The SIMPLE model is derived from the Tanner, Fung, and Virkar (TFV) m odel for Micro - Composite Cathodes. The difference is that the SIMPLE mode correctly accounts for the surface area ratio bet ween the MIEC and the IC; which is close to 1, and can therefore be ignored in a nano - composite cathode but often greater than 2 in a NMCC . The SIMPLE model does this by mathematically spreading the R s value across the entire IC scaf fold [ 33 ] . By us ing microstr uctural parameters from Figure 2 . 7 (such as cathode thickness, MIEC infiltrate nano - particle size, IC scaffold ionic conductivity, etc. ) , and intrinsic m aterial properties (MIEC infiltrate R s and IC scaffold ionic conductivity) , the SIMPLE m odel provides an estimate of the N M CC R p value using the equation: [2] where: , [3] [4] R s is the effective oxygen surface exchange resistance, A Sc is the surface area of the IC scaffold, A Inf is the surface area of the infiltrated MIEC material, r is the repeat unit 22 thickness, h is the height of the repeating unit cell, o2 - is the IC scaffold oxygen conductivity , and p is the non - infiltrated scaffold porosity. The SIMPLE Model has been compared to experimental data from NMCCs with varying MIEC infiltrate loading levels [ 7 , 8 ] , cathode thicknesses [ 7 ] and MIEC materials [ 5 ] . In all cases the results to date in d icate that the SIMPLE Model is capable of accurately predicting R P values for the N MCC microstructures and materials combinations used today to within an order of magnitude . A complete derivation of the SIMPLE model is included in Appendix 1 of this thesis . As shown in Equations 1 - 4 , both the SR limit and the SIMPLE model indicate that the cathode R P drops (i.e. cathode performance increases) as the MIEC surface area ( A inf ) increases and the R s value decreases . Unfortunately, as discussed in the next section, experimental methods to increase A inf by reducing MIEC nano - particle size and precisely determined infiltrated R s values are presently lacking in the literature. 2.6 Infiltrate Particle Size is Difficult to Control According to the Literature Figure 2.8 show s nano - particle sizes for a number of literature studies using SSC, LSCF , and LSFC [ 5 , 7 - 9 , 30 ] . Each set of literature studies f or a given material was proce ssed in exactly the same manner , yet reported a large range in average nano - particle size. The SSC nano - particles were produced using a 10°C/min firing ramp rate, an 800°C firing temperature, a 1 hr hold time at 800°C, a 12.0 vol% MIEC loading level, an d a 0.5 molar precursor solution. The LSCF and LSFC nano - particles were produced using a 5°C/min firing ramp rate, an 800°C firing temperature, a 1 - 2 hr hold time at 800°C, a 12.0 vol% MIEC loading level, and a ~ 0.5 molar precur sor solution. The 23 different literature studies also used different solution additives such as Triton X, Citric Acid, or a lack of solution additives (i.e. Pure Nitrate). As shown in Figure 2.8, there appears to be a lack of control in inf iltrated nano - particle size for particles 40 nm and larger, and an inability to produce MIEC infiltrate particles wit h average sizes less than 40 nm, which may be partly the result of different solution additives used in the literature studies. That being said, so me studies have reported MIEC particle sizes less than 15 nm [ 79 , 80 ] , but the process variables necessary to achieve this hav e not been clearly identified or understood. Figure 2.8 : Infiltrated MIEC Nano - Particle Size for SSC, LSC F, and LSFC . Colors indicate which solution additives were used in each literature study for easy comparison [5], [7 - 9 ], [30 ] . The following sections will discuss in greater detail the different fabrication techniques used in the literature to reduce MIEC infiltrate particle size and improve performance. 2.6.1 Fabrication Techniques that Reduce Particle Size The addition of an additive to the p recursor solution has been shown in literature to reduce infiltrate particle size and improve performance. Solution additives (Surfactants, Chelating agent, etc.) are typically used to prevent the precipitation of 24 individual nitrate cations from the precu rsor solution [ 5 , 7 ] . However, Nicholas et al. [ 5 ] have shown SSC nano - particle size also is reduced when using Triton X - 100 as compared to not using a solution additive. Nicholas et al . [ 5 ] have also shown that no t all solution additives reduce nano - particle size. For example, SSC made with Citric Acid had similar nano - particle size wh en compared to not using a solution additive. The mechanism behind these nano - particle size reductions using solution additives may be described by the synthesis method Evaporation - Induced Self - Assembly (EISA). The EISA synthesis method has been document ed in multiple literature studies and has been shown to result in the reduction of nano - particle sizes [ 81 - 86 ] . EISA was initially used to create micro and meso - porous silica nano - particles which were ordered [ 87 ] . The EISA method takes advantage of the fact that solution additives generate templates (such as micelles for surfactan ts) as the solution there are in becomes evaporated [ 88 ] . Thermal decomposition is then used to remove the additive template and to form continuous nano - oxide particles with a high level of porosity [ 89 ] . Alternatively, increasing the heating rate of the thermal decomp osition of the precursor solutions has been shown in literature [ 1 ] to decrease the average size of MIEC oxide particles formed. Zhao et al. [ 1 ] also showed that the performance of their SOFC cells increased as a result of reduced infiltrate nano - particle size as a result of the increased heating rate . Overall, the connection between solution additives , the water content in the gelled precursor solutions , and infiltrate MIEC nano - particle oxide size s suggested b y the literature [ 79 , 80 ] matches other studies using different material compositions demonstrating that 1) the EISA method wi th organic solution additives is capable of 25 control ling the nano - particle size , geometry, and template arrangement produced from thermally decomposition gels [ 85 , 89 ] , and 2) that the water content in precursor gels can have a large effect on the oxide nano - particle size resulting from this process [ 84 , 90 ] . 2.6 .2 Fabrication Techniques that H ave N o Impact on Particle Size Reduction The amount of MIEC precursor solution infiltrated into a porous IC scaffold can easily be adjusted by increasing or decreasing the volume of each infiltration. Literature studies have shown that increasing or decreasing the volume of MI EC prec ursor solution has little to no effect on the average M IEC infiltrate particle size and the resulting cathode performance [ 5 , 7 ] . The molarity of the MIEC precursor solution infiltra ted into the porous IC scaffold can also be changed by increasing or decreasing the amount of water used in the solution. Literature studies have also shown that the precursor solution molarity has little to no effect on the average MIEC oxide nano - particle size and the resulting cathode performance [ 5 ] . Therefore, the first objective of this thesis is to investigate how and why several processing variables can be used to reduce MIEC infiltrate particle sizes. The second objective of this thesis is to obtain MIEC infiltrate R s val ues for the first time. As background, t he next part of Chapter 2 will discuss common approach es used to measure R s values for diffe rent infiltrate MIEC materials. 2.7 Literature Oxygen Surface Exchange Resistance Measurement Techniques As discussed at the end of Section 2.5.2, measuring the R s values for various MIEC materials is crucial for gauging whether a material will be a good choice when used in NMCCs. A smaller R s value indicates that the material has a lower resistance for 26 exchanging oxygen into its lattice structure and would therefore lower overall SOFC resistance and/or SOFC operating temperatures if it were employed in a NMCC . The following measurement techniques are typically used to determine R s values found in the literature: 1) thin f ilm EIS measurements, 2) R s conversions from k o measurements, and 3) R s conversions from k chem measurements. R s values determined through these three techniques can then be input into the active models such as the SR and SIMPLE models to predict NMCC perf ormance . 2.7 .1 Oxygen Surface Exchange Res istance from Thin Film Electrochemical Impedance Spectroscopy The R s value is often measured in literature using the EIS characterization technique [ 47 , 54 ] . In this method a thin film ( described in S ection 2.1.3 ) of a MIEC material is typically d eposited onto a substrate (such as YSZ or GDC ), and an AC signal is passed through the t hin film at various frequencies ( typically between 10 5 and 10 - 1 Hz ). This process is repeated over a wide range of temperatures to understand how the R s value of each MIEC material varies as a function of temperature. R s value s are typically extracted from electroch emical impedance spectroscopy using Nyquist plot s produced by plotting the real (x - axis) and imaginary (y - axis) components of each frequency impedance data point . As shown in Figure 2.9, multiple arcs may be present in an impedance spectra, but once the arc related to MIEC oxygen surface exchange is identified , t he distance be tween the two x - axis intercepts m ultiplied by the electrode area can be used to ext ract the R s value [ 47 ] . 27 Figure 2.9: SOFC LSCF R s Measurement using Electrochemical Impedance Spectroscopy . The Electrochemical Imp edance plot is depicted as a Nyquist p lot in this example [47] . Oftentimes, identification and fitting of the AC arc corresponding to oxygen surface exchange is facilitated by equivalent circuit modeling [ 59 , 91 - 94 ] . Figure 2.10 [ 59 , 91 - 94 ] shows an equivalent circ uit commonly used in the literature to describe the physical processes involved in a NMCC. Figure 2.10 shows three resistive processes: 1) resistance through the bulk ( R b ), 2) resistance through an interfacial layer between the MIEC and the IC scaffold ( R i ) , and 3) the oxygen surface exchange resistance in the MIEC material ( R s ). Because they occur across interface s the R i and R s processes also have cap acitance values. Each process is assumed to be independent of each other, which is why the R b , R i , and R s RC elements are shown in series in Figure 2.10. 28 Figure 2 .10 : Typical N ano - M icro - C omposite C athode Equivalent Circuit Model . This example depicts three resistive processes in series, where two resistive processes have capacitance elements [59], [91 - 9 4 ] T he equivalent circuit has been used to determine R s values in Figure 2.10 [ 59 , 91 - 94 ] and can be depicted mathematically described by: [5] where Z is the total calculated impedance ( AC resistance ), R b is the resistance for the electroly te, R i is the resistance for either the ionic bulk transport through the IC scaffold or the interfacial resistance between the IC scaffold and the MIEC lattice, R s is the resistance for oxygen surface exchange into the MIEC lattice structure, C i is the capacitance for the ionic bulk transport through the IC scaffold or the interfacial capacitance between the IC scaffold and the MIEC lattice, C chem is capacitance for the oxy gen surface exchange into the MIEC lattice structure , is the frequency , and j is the imagina ry component of the equation . Fitting the equivalent circuit model to experimentally determined EIS results allows for the different resistance values to be dete rmined. In the literature, fitting the equivalent circuit is typically accomplished by 29 adjusting the R b , R i , and R s values (and the capacitance values too) in Equation 5 so the resulting impedance arcs match the experimental data. 2.7 .3 Oxygen Surface Res istance from Chemical Rate Constant Measurements R s values can be converted from chemical rate constant values, ( k o or k chem ) found in the literature. The difference between k o and k chem is that k o is only dependent on the diffusion of oxygen vacancy species through the material, while k chem is dependent on the combined ambipolar diffusion of oxygen vacancies and electrons through the material . The relationship between R s and k o is described as: [6 ] where k b is the Boltzmann constant, T is temperature (K), e is the elementary charge, C o is the concentration of oxygen in the MIEC lattice structure , and k o is the chemical rate constant through the bulk [ 10 ] . If the k chem value is provided in literature , then k o can be determined from k chem and R s can be determined using E quation 5 . The relationship between k che m and k o is described as: [7 ] dynamic factor (which is determined by dividing the derivative of the natural log oxygen partial pressure of the equilibrium gas by the derivative of the natural log of the oxide concentration in the MIEC material) [ 95 ] . 30 Figure 2.11 shows the lit erature - reported concentration of oxygen in the MIEC lattice structure [ 50 , 54 , 95 - 98 ] as a function of temperature for the MIEC materials LSF, LSFC, LSCF, LSC, SSC and BSCF. All these materials increase the amount of oxygen in the MIEC lattice structure as the temperature increases except for SSC, which stays constant. ( The fact that SSC has constant oxygen concentrations with respect to temperature may indicate that this material is o perating in the extrinsic regime ). Figure 2.11 also shows the thermodynamic factor for the different cathode MIEC materials as a function of temperature [ 48 , 95 , 97 - 99 ] . These C o values were all used in Equations 6 and 7 to calculate R s from k chem , and k o . Figure 2.11 : Oxygen Concentration and Thermodynamic Factor Values for LSF, LSFC, LSCF, LSC, SSC and BSCF in Air Reported in Literature Studies . The oxygen concentration values [50], [54], [95 - 98] are shown on the left and the thermodynamic factor values are shown on the right [48], [95], [97 - 99]. Due to literature sources typically only reporting k chem or k o values, C o an values from multiple literature studies were used to convert k chem or k o values into R s values. This may not be completely le gitimate since d ifferences in sample preparation , surface structure, grain size, stress s tate, etc. could lead to different k , C o between samples with the same composition . The uncertainty introduced by the need to 31 mix study results produces an additional incentive to directly measure the R s values of MIEC infiltrate particles. Figure 2.12 shows both k chem and k o values for some typical MIEC materials found in literature [ 48 - 51 , 54 , 95 - 102 ] . These values were determined from EIS Figure 2.12 : k chem and k o Values Report ed in Literature for the Cathode MIEC Materials LSF, LSFC, LSCF, LSC, SSC, and BSCF . Literature study authors are listed in the legend [48 - 51], [54],[95 - 102]. 32 measurements or were converted using Equations 6 and/or 7. The values shown in Figure 2.12 vary in magnitude between reports within a single material, for unknown reasons, which also promotes large variations in the R s value. 2.8 Large Uncertainties Exist in the R s Value of E ven the Most Common Mixed Ionic Electronic Conducting Materials Figure 2.13 shows R s values for several common MIEC materials (with a lower R s value promoting more oxygen incorporation ) [ 1 0 , 48 - 51 , 54 , 95 - 97 , 99 - 102 ] . The values shown in Figure 2.13 were determined using: 1) EIS, 2) R s conversion from k o values, or 3) R s conversion from k chem values. As shown in Figure 2.13, some MIEC material s have reported R s value s that range by as much as 3 orders in magnitude . This inconsistency in R s values within the same material composition demonstrates the need to understand where infiltrated MIEC R s values lie in comparison to the bulk and thin film R s literature values in Figure 2.13 . R s values determined from EIS measurements were from Baumann (LSF, LSFC, LSCF, LSC, SSC, and BSCF), Niu (LSF), Egger (LSC), and Xiong (LSCF). R s values converted from k chem or k o Elshof (LSF), Dalslet (LSFC), Steele (LSFC), Simrick (LSFC), Yeh (SSC), Girdauskaite (BSCF), Burriel (BSCF), and Bucher (BSCF) 33 Figure 2.13 : R s , Values Reported in Literature for Cathode MIEC Materials LSF, LSFC, LSCF, LSC, SSC, and BSCF . Literature study authors are listed in the legend [10], [48 - 51], [54], [95 - 97], [99 - 102] . 2.9 Summary In summary, previous literature studies have shown that SOFCs have had their performance improved through the use of NMCCs. Cathode, electrolyte and anode supported SOFCs are have bee n used previously in literature. The NMCC itself can also use different co mbinations of ionic conducting, electronic conducting, or mixed - conducting scaffolds and infil trate materials. Despite the performance advances in SOFC s made possible through the use of NMCCs , there exist two knowledge gaps that previous literature studies have not adequately addressed: 1) controlling NMCC infiltrate MIEC nano - particle size , and 2) accurately determining the R s values for MIEC infiltrate materials. The first knowledge gap (lack of infiltrate particle size control) needs to be filled to increase cathode performance at operating lower temperatures. The second 34 knowledge gap (a ccurate determination of infiltrate MIEC R s values) need to be filled to improve cathode performance at lower operating temperatures, and provide the scientific community useful knowledge which MIEC material is the best to use. 35 CHAPTER 3 : Experimental Methods 3.1 Symmetrical Cathode Fabrication Process es Since SOFC electrode performance and not entire SOFC performance was the objective of this thesis, f or the reasons described in Section 2.2, electrolyte supported cells were used here to explore the impact of processing conditions on NMCC performance and infiltrate particle size. Although specific sample fabrication details can be found in later chapters, this section provides an overview of the NMCC fabrication process. 3.1.1 Cathode - El ectrolyte - Cathode Symmetric Cell Production Dense SOFC electrolyte s can be fabricat ed through a number of approaches including dry pressing and firing [ 103 , 104 ] , chemical vapor deposition [ 105 , 106 ] , electrochemical vapor deposition [ 107 , 108 ] , spray pyrolysis [ 109 , 110 ] , pulsed laser deposition [ 111 , 112 ] , 6) sputtering [ 113 , 114 ] , etc. Here, e lectrolyte - supported cells were prepared first by pressing GDC powder in a steel die using a uniaxial press, shown in Figure 3.1 [ 115 ] , to form a porous cylindrical pellet. The resulting porous pellet was sintered at temperatures close to 1450°C and then cooled to room temperature to produce pellets with relative densities >95%. Lastl y, the top and bottom of these cylindrical pellets were sanded flat and parallel , first using 240 grit SiC sandpaper and then 600 grit SiC to obtain final thickness values ranging from 430 to 480 µm, which promotes a uniform electrode thickness. 36 3.1.2 Cathode Fabrication The cathode scaffold was fabricated using the screen printing technique. Porous well - nec ked GDC IC scaffolds were produced on both sides of the electrolyte cell . To achieve this , GDC powder was coarsened prior to being mixed with an electronic vehicle to form a GDC ink . Three layers of GDC ink were then scree n printed, using a screen printer shown in Figure 3.2 [ 116 ] , onto each side of the dense GDC pellet . Before the next ink layer was applied, each ink layer was allowed to flow across the pellet surface , and then was placed in a bake oven to extract the electronic vehicle solvent and increase the green strength ( i.e. mechanica l strength prior to sintering) . After screen printing, the samples were then sintered and cooled to room temperature . Sintered IC scaffold thickness and roughness measurements were made with a profilometer . Figure 3.1: Uniaxial Press Device . Uniaxial press used to produce green strength electrolyte samples [ 11 5 ]. 37 Figure 3.2: Screen Printer Device . The scree n printer is used to fabricate the porous cathode microstructure [ 11 6 ] . . 3.1.3 Mixed Ionic and Electronic Conducting Precursor Solution Fabrication and Cathode Infiltration MIEC lanthanum strontium cobalt iron oxide (La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - x , LSCF) metal nitrate precursor solutions were prepared by dissolving lanthanum nitrate, strontium nitrate, cobalt nitrate and iron nitrate in distilled water , with or without a solution additive ( such as Triton X - 100 or Citric A cid) . These LSCF precursor solutions were infiltrated into symmetrical cells fabricated using both the desiccation and pre - infiltration techniques (mentioned in Section 2.6.3) . GDC precursor solutions , used in the pre - infiltration technique, were prepared by dissolving gadolinium nitrate and cerium nitrate. GDC precursor solutions were only used with the pre - infiltration technique. After being infiltrated these solution were allowed to soak into the scaffold for 5 minutes and gelled at 80°C in a drying oven for 10 minutes . Desic cated cells were then sealed inside a desiccator with chemical desiccant , and then thermally decomposed after 38 being desi ccated. The two step infiltrate - gel - fire technique omits the desiccation step , and thermally decomposes the precursor solutions after being infiltrated and dried in the oven at 80°C . 3.1.4 Mixed Ionic and Electronic Conducting Oxide Phase Purity Analysis After the MIEC precursor solution was thermally decomposed , the phase purity of the resulting MIEC oxide powder was characterized usin g XRD. Precursor solution s were placed onto an alumina slide, gelled and then thermally decomposed. The resulting oxide powder was placed onto a low X - ray background fused silica slide and characterized using XRD to determine the resulting oxide powder p hase purity. The purpose of analyzing the LSCF oxide powder outside the scaffold was that XRD characterization of the LSCF powder inside the GDC scaffold was much more difficult due to XRD peak overlap between the LSCF and GDC [ 1 17 , 118 ] . T he phase purity of the resulting LSCF oxide powder was assumed to be the same inside the scaffold as outside the scaffold , since the processing conditions were the same using both fabrication techniques. A representative phase pure XRD scan for LSCF is sh own in Figure 3.3. Figure 3.3: Representative X - ray Diffraction Data for Phase Pure LSCF Oxide Powder . The PDF file was obtained from the JADE XRD computer program for LSCF. 39 3.1.5 Current Collector Application After complete NMCC fabrication , c urrent collector layers were applied to symmetrical cathodes by screen printing bilayer LSM Au layers onto the surface of each NMCC. The LSM ink was prepared by mixing LSM powders with a polymer formulation to form an ink with a 34% solids loading. A single layer of LSM ink was then screen printed onto each NMCC cathode , and the samples were sintere d and allowed to cool to room temperature. Finally Au paste was screen printed in an open grid pattern across the surface of each LSM current c ollector . 3.2 Characterization Techniques 3.2 .1 Electrochemical Impedance Spectroscopy Here, EIS measurements were used to measure cathode performance (i.e. cathode R P ) at open circuit as a function of temperature using the setup schematic shown in Figure 3 .4 [ 10 ] . EIS measurements were evaluated in both air, 20%O 2 - 80%He, and 20%O 2 - 80%N 2 . EIS measurements using the two - point technique were taken every 50°C between 400 °C and 700 o C. An unbiased AC signal, at frequencies typically between 1 MHz and 0.1 Hz [ 1 , 5 , 61 , 119 ] , was applied to the sample to collect impedance measurements . 4 - point EIS measurements are performed in literature , but were not performed for this thesis because lead and contact resistances for the EIS measurement are minimal compared to the cathode R P [ 5 , 61 ] . Samples were allowed to thermally equilibrate before each EIS measurement was collected. 40 Cathode R P values were extracted by measuring the distance between the high and low frequency x - axis intercepts on Nyquist plots , which had been multiplie d by the geometric cathode area and divided by tw o (since each symmetric cell has two cathodes). Ohmic resistivity values for each cell were determined by combining the measured distance between the origin and the high frequency x - intercept on the Nyquist plot , the geometric cathode area , and the measured electrolyte thicknesses using the definition of resistance: [ 8 ] Figure 3.4 : N ano - M icro - C omposite C athode Electrochemical Impedance Measurement Device . The depicted measurement device uses a 2 - point measureme nt technique instead of the 4 - point technique [10 ]. 41 3.2 .2 Scanning Electron Microscopy Figure 3.5 [ 120 ] shows a diagram of the SEM ch aracterization technique which can be used t o determine whether different fabrication techniques impact nano - part icle size . The SEM characterization technique was also useful for evaluating the morphology and overall size of powders and bulk samples. Ins ide a SEM electrons are generated, Figure 3.5: Scanning Electron Microscopy Component Schematic . The schematic depicts the Scanning Electron Microscopy components from the electron producing component to the individual detectors [12 0 ]. focused using condenser lenses, and directed onto the sample . These high velocity electrons interact with the sample and generate different signals which are: secondary electrons, backscatter electrons, diffracted backscatter electrons, etc. and are measured by variou s detectors. The secondary electrons and backscatter electrons are typically used in literature [ 1 , 3 , 5 , 7 , 9 ] for image processing and were used to characterize nano - particle 42 siz es and the cathode microstructure in this thesis. The SEM characterization technique has been used in many different literature studies to determine MIEC and IC particle sizes [ 1 , 5 , 7 , 61 ] . 3.2.3 Solid Oxide Fuel Cell Nano - Particle Coarsening Rate At the elevated temperatures at which SOFCs operate, an opportunity for MIEC nano - particle size growth via diffusion exists. Further, d uring the thermal decomposition of MIEC precursor solutions the precursor oxides that form also have an opportunity to c oa rsen in order to reduce the nano - particle surface area . Here nano - part icles coarsening were characterized through either SEM images or through the Williamson - Hall method using X - ray diffraction (XRD). Specifically, various precursor solutions were firs t thermally decomposed at the same starting conditions (such as molarity, solution composition, solution infiltration volume, heating rate, dwell time, etc. ), and the resulting oxide powders were characterized to determine the average particle size using S EM or the XRD Williamson - Hall method. Afte r being characterized, the same oxide powder s were again heated at an elevated temperature , cooled and characterized (using SEM or XRD Williamson - Hall) a second time to determine how much the average oxide particle size increased (if at all). 3.2 .4 X - ray Diffraction The XRD characterization technique was used to identify the crystal structure of the different powders. In this technique, X - rays are generated from a metal target by hitting the target with el ectrons generated from a sealed tung st en filament, shown in Figure 3.6 [ 121 ] . When the electrons hit the target they eject electrons from the target K or L shells, which are then filled by electrons in the shell above the lost electron (L or M 43 shells, respectively). The process of transferring electrons from an upper shell to a lower shell releases energy in the form of an X - ray. The most common X - rays are the K alpha (from the K shell) and K beta (from the L shell). X - rays are generated in all directions Figure 3.6: X - ray Dif fraction Component Schematic . Components depict filtered x - rays impacting sample crystal [1 2 1 ]. aro und the target , but a filter is applied in the direction of the sample to gene rate a monochrom atic, or nearly monochromatic, X - ray beam . The X - ray beam hits the crystal or powder sample and diffracts, acco rding to : [ 9 ] where n is an integer, is the X - ray wavelength (nm), d is the lattice plane spacing (nm) , and is the angle betw een the incident and scattered X - rays off the crystal or powder sample , before being measured in an X - ray detector . 3.2 .5 Williamson - Hall Particle Size Determination from X - ray Diffraction Nano - particle size characterization was not simply limited to the SEM characterization technique. Infiltrated MIEC o xide nano - particle sizes were calculated via the Williamson - Hall method [ 122 ] for powders fabricated inside and outside a porous 44 IC scaffold. The Williamson - Hall method was applied to XRD spectrum taken of the MIEC powder , assuming that peak broad ening was caused by the instrument, particle size , and nano - particle strain. The full - width half - max (FWHM) values for at least two peak s were utilized to obtain average particle size and strain . These XRD peak were carefully selected so only peaks corre sponding to a single miller index are used ; (i.e. in order to prevent peak broadening caused by additional diffraction from additional lattice planes ). Peak broadening contributions for the instrument were removed by: 1) values measured with the JADE computer program) for a number of particle sizes, and 2) subtracting the The y - axis linear intercept was then used to calculate the infiltrate MIEC nano - particle size, using the following equation : where D is the average infiltrate MIEC nano - particle size ( nm ) , K is shape factor ( typical value of is the X - ray wavelength (0.154 nm for Cu K X - rays). The particle strain was determined directly from the slope of the linear trend line determined from . 3.2 .6 ThermoG ravimetric Analysis The ThermoGravimetric Analysis (TGA) technique was also used to elucidate precursor solution decomposition processes . A TGA device , such as that shown in Figure 3.7 [ 123 ] , was used to measure mass changes during sample heating for a small sample (usually between 25 and 50 mg) placed inside a platinum or alumina pan . As the 45 sample was heate d , decomposition products were released until only oxide powder r emained , with the mass loss caused by sample decomposition measured by the TGA . Here, experiments were conducted in flowing nitrogen to determine the waters of hydration for each nitrate, and in air for precursor solution decomposition measurements. Figure 3.7 : Thermo - G ravimetric Analysis Schematic . The depicted Thermo - Gravimetric Analysis uses a hanging scale to measure mass changes. [ 1 2 3 ]. 3.2 .7 Profilometry Characterization of the cathode thickness was necessary to accurate ly determine the precursor solution volume s required for infiltration [ 33 ] . The profilometry characterizat ion technique, shown in Figure 3. 8 [ 124 ] , allows for careful measure me nt of both the cathode thickness and surface r oughness. The profilometer used a need le that was lowered and moved across the surface of the cathode . The need le was attached to a spring and as the needle was raised and lowered, due to height changes in th e cathode , and th e profilometer calculates cathode thickness values 46 Figure 3 .8 : Profilometer Schematic . The zoomed image of the stylus tip shows the difference in the measured and original surface profile. [ 12 4 ]. 3.2.8 Focused Ion Beam - Scanning Electron Microscopy 2D Serial Sectioning NMCC microstructure 2D serial sectioning was performed using Focused Ion Beam - Scanning Electron Microscopy (FIB - SEM). The 2D microstructural images were obtain ed using a scanning electron microscope ( Auriga Dual Column Focused Ion Beam - Scanni ng Electron Microscope) operated with an electron beam voltage of 20 kV, an 120 um aperture , a 40 second scan speed, a 5.0 mm working distance, and a 10,000x magnification. The FIB - SEM x - y spatial res olution was 1 nm and the z spacing b etween each 2D section was 20 nm [ 125 ] . Th e FIB was operated with a beam voltage of 2 kV and a 5.0 mm working distance. The FIB beam current used to initially clear around the desired target image area was 16 nA, while the beam current used to collect 2D serial section images was reduced to 4 nA. The reducti on in beam current allowed for more 47 detailed image to b e obtained for image processing . A detailed outline of FIB - SEM sample preparation and analysis technique is presented in Appendix 2. 3.3 Finite Element Modeling of Cathode Microstructure and Performance 3.3.1 Motivation for Using Finite Element Modeling In addition to experimentally determined EIS measured and SIMPLE Model predicted R P measurements , F inite E lement M odeling (FEM) was used to predict R P values on imported cathode 3D micros tructures taken from FIB - SEM images. Here, the R s values used in FEM R P calculations were adjusted until the FEM R P value matched the EIS measured R P values. In this manner, MIEC infiltrate R s values were obtained for the first time. Determination of R s values from FIB - SEM images is not a common technique found in the literature because sample preparation is a lengthy process, and construction of 3D models used in the FEM cal culations are not a trivial task. Since only R s values for thin film microstructures and bulk samples currently exist as a guide to select which MIEC cathode material s provided improved performance FEM allow s for a much more accurate determination of infiltrated MIEC R s values in NMCCs. T he assumptions used to determine R s from R P via FEM were that the 1) gas - phase concentration polarization resistance was minimal , 2) the interfacial resistances were minimal , 3) the electronic resistance losses in the oxygen surface exchange layer were minimal , and 4) t he ionic conductivity in the MIEC infiltrate was minimal due to the layer being 50 nm or less in thickness. These assumptions were valid for the following r easons. First, p ost - infiltration porosity measurements for all cells in this thesis were 20 vol% o r g reater ( Post - infiltration porosity measurements were determined by 48 subtracting infiltration volume % values from FIB - SEM porosity measurements on non - infi ltrated cathode microstructures [ 33 ] typically sufficient for percolation [ 7 ] suggests that the pore network in the reconstructed cathodes was percolated (and therefore gas - phase concentration polarization resistance was low). In addition, experimental tests performed in 2 0%O 2 - 80%He and 20%O 2 - 80%N 2 have shown similar R P results in air, which also indicates that gas - phase concentration polarization resistance is low [ 126 ] . Second, XRD scans taken on NMCCs only detect impurity phase fractions which are greater than 5% of th e main peak, which indicates that any p ossible interfacial resistan ce from these oxi de phases would be minimal. Literature studies have shown phase pure LSCF and GDC to be free of reaction products [ 5 ] . Third, e lectronic resistance losses were minimal because infiltrated MIEC nano - particles in the cathode microstructure have been shown to be interconnected at infiltration loading leve ls of 12 vol%, which was the case for all experimentally determined results in this thesis [ 5 ] . EIS ohmic offset NMCC values also indicate an abse nce of electronic losses in the open - circuit R P measurements taken here. Four th, t he characteristic thickness (ratio of the diffusion coefficient to the oxygen surface exchange coefficient) of LSCF has been measured to be greater than m [ 51 ] , indicating that the 50 nm or smaller infiltrate particles here were electrochemically active over their entire surface and their performance was dominated by their surface resistance. 3.3.2 Finite Element Modeling Performance Calculation Overview FEM op en - circuit R P modeling work was performed using the same set of microstructurally independent assumptions covered in Appendix 1 up to Equation A1.1. Like the SIMPLE model, this FEM open - circuit R P modeling did not require the use of 49 the Buttler - Volmer equ ation to solve the current density flux across the electrolyte because both models predict open circuit R P values . As done previously in the literature [ 127 ] , the first step in deriving the FEM open - circuit potential was to assume that diffusion - limited oxygen vacancy flux densit y through the IC scaffold was controlled by dilute thermodynamics and the refore amenable to modeling via the Nern st - Planck equation : [11 ] where J is the atomic flux, i is the sum of all the species charge carrier, z is the charge, is the mobility, F is the Faraday constant, is the electrical potential, D is the diffusivity of species, c is the concentration, is the convective velocity, is i was the index number. T he Nernst - Planck equation was then simplified by first assuming that the IC scaffold in the electrode and electro lyte had a constant composition ( a common assumption used in the SOFC modeling community [ 33 ] ) . This assumption was reasonable because : 1) the curren t densities seen in SOFC cathodes near operation are relatively low , and 2) the oxygen partial pressure inside the cathode pore was constant (which leads to low concentration polarization ). This first assumption allowed the second term of the Nerns t - Planc k eq uation to be removed and reduced the equation to the following form: [12 ] 50 The second assumption was that no convection in the c athode or electrolyte materials existed . This was a reasonable assumption since the MIEC material, cathode microstructure , and electrolyte had been made of solid oxide materials which would lack con vection. This assumption allowed the third term in the N ernst - Planck eq uation to be removed and reduced the equation to the following form: [13 ] Using these assumptions and the fact that GDC was a pure ionic conducting material (at least in air) [ 128 ] allowed the atomic flux in the IC material to be reduced to the following form: [14 ] where z was replaced with 2 because each oxygen ion has a 2 + oxidation state , the mobility was the mobility of oxygen vacancies ( ), F was Faradays constant , and the concentration was the oxygen vacancy concentration ( ). In order to solve for the steady state cathode R P the gradient of each side of Equation 14 was taken and the fact that at stead y state [15 ] 51 Hence in this special, open - circuit situation for a N MCC with an IC scaffold, FEM programs such as ABAQUS or in this case COMSOL equation) could be used to determine the potential distribu tion within the ionic conducting scaffold . Since the MIEC material was limited by the oxyg en exchange across the surface , (i.e. since they were well below the characteristic thickness there was no electrical potential drop across the bulk of each MIEC particle, only across its surface) the MIEC nano - particles were modeled as a thin surface resi stance layer ac ross the cathode microstructure . The surface resistance for this MIEC layer was applied to the ionic conducting scaffold as a surface material property. The final assumption was that an extremely small interfacial resistance between the M IEC surface impedance layer and the IC GDC cathode microstructure existed . This last assumption was realistic because Baumann et al. [ 47 ] have shown the interfacial resistance to be at least 2 orders of magnitude lower than the surface impedance resistance for MIEC - IC material combinations at SOFC operating temperatures . The assumptions up to this point are shown in the Appendix 1 derivation up to Equation A1.1. The potential distribution , discussed in Equation A1.1, was then solved for within the GDC . After the potential distribution thro ugh the GDC had been determined by using COMSOL to solve Equation 15, t he R P value was determined by dividing the potential difference , V across the reconstructed volume by the tota l oxygen vacancy current density flowing through the reconstructed volume (A/I ) (a derivation showing how , from Equation A1.1, can be determined from I/A can be found in Appendix 1, Figure A1.3 ) and subtracting out the ohmic resistance for the reconstructed electrolyte ( i.e. in Equation A1.1) . The potential difference was 52 defined as the difference between the 1V applied to the MIEC surface impedance layer and the 0V applied t o the bottom of the electrolyte using the equation: [16 ] where R P is the polarization resistance, is the applied reference potential to the MIEC surface, I is the integrated current across the electrolyte, A is the electrolyte area integrated across to obtain the current, t is the electrolyte thickness contained in the reconstructed volume , and IC is the IC of the electrolyte and scaffold materials. 3.3.3 Focused Ion Beam - Scanning Electron Microscopy Sample Preparation Samples for FIB - SEM analysis were prepared first by screen printing a porous GDC scaffold s on top of a dense, sanded electrolyte pellet s . Samples were then epoxy infiltrated to improve pore - scaffold contrast in later FIB - SEM measurements. Epo xy resin and hardener (EpoThin, Buehler) previously mixed in a 5:1.95 mass ratio and stirred for ~5 minutes , were then placed inside a chamber where a vacuum was applied to remove bubbles from the epoxy (the vacuum was - 20 inches Hg). While maintain the v acuum t he epoxy was poured around the sample until the sample was submerged in epoxy . The vacuum was maintained for at least 5 minutes to remove gas bubbles, and allowed the epoxy to soak into the scaffold pores. The vacuum was then released and the epoxy was allowed to harden arou nd the sample over 12 hrs. After the epoxy hardened the sample was cut to expose the scaffold surface, and was sanded using 1200 grit sandpaper which produced a mirror smooth surface. The sanded sample was bonded to an aluminum sample holder and had ~2 nm of tungsten sputtered on its surface. 53 Figure 3.9 shows the epoxy coated sample oriented in the FIB - SEM [ 129 ] . The sample was placed at a 54 0 angle so it was perpendicular with the FIB. The FIB was used to remove a section of material in front of the desired observation surface and the SEM was aligned to image the observation surface. After each serial section, t he FIB was used to remove 20 nm off the observation surface (in the direction of the arrow on Figure A2.1) and the SEM was used to take a 2D image of the newly exposed surface , shown in Figure 2.10 where black areas represent pores and the grey areas are electrode scaffold material . This pr ocess was repeated to create a set of 2D serial section images. Figure 3.9 : Epoxy Coated Porous GDC Scaffold on a Dense GDC Electrolyte Oriented Inside the FIB - SEM . The FIB is used to make serial sections in the direction of the arrow while the SEM images each newly exposed sample surface [129] . 54 Figure 3. 10 : FIB - SEM Cathode Backscatter Image used for 3D Reconstruction . The depicted black areas are pores while the depicted lighter regions are scaffold material. 3 .3.4 Cathode Microstructure 3D Reconstruction T hese 2D serial section images were then recombined using a program called MIMICS [ 130 ] to create a 3D reconstruction of the cathode and electrolyte microstructure . A step - by - step discussion of how 3D reconstruction s were cr eated is shown in A ppendix 2 . For the work in this thesis, 205 serial section images were i mported into the MIMICS program, and a threshold was applied to tell MIMICS what grey - scale val ues were IC scaffold and what gray - scale values were pores. From these images a 3D reconstruction was c reated for the cathode and electroly te microstructures . Figure 3.11 shows a 3D reconstruction created in MIMICS of the cathode and electrolyte, done separately . The 3D reconstruction can be generated using an optimum, high, medium , or low number of triangles. A larger number of triangles improve s the accuracy of the final measurement but increas es the calculation time required for 3D volume meshing and the R P calculation . For this thesis , due to the sample size being large (in the millions of triangles), a medium density of triangles was used. 55 Figure 3 .1 1 : FIB - SEM 3D Reconstruction for Cathode and Electrolyte . The electrolyte is shown on the left and the cathode is shown on the right. Both parts were reconstructed in the MIMICS computer program separately. 3.3.5 3D Cathode Reconstruction Volume Meshing Once a 3D representation of the electrolyte and cathode was created in MIMICS , the cathode and electrolyte were merged together and volume meshed using the 3 - Matic computer program before being FEM modeled with the COMSOL co mputer program . The dense electrolyte layer was added to each cathode to facilitate summation of the current flowing across the cathode. Step - by - step instructions for how to merge the 3D reconstructions and volume mesh the entire sample are sh own in A ppe ndix 2. Figure 3 .1 2 shows the volume mesh of the combined cathode and electrolyte microstructures produced using 3 - M atic. This volume mesh was then exported to COMSOL for performance calculations. 56 Figure 3 .1 2 : FIB - SEM 3D Reconstruction Volume Mesh of Cathode and Electrolyte . The electrolyte and cathode reconstructions from Figure 3.11 were merged and volume meshed using the 3 - Matic computer program. 3.3.6 Infiltrated Solid Oxide Fuel Cell Cathode Finite Element Modeling to Predict Polarization Resistanc e Figure 3.13 shows the electrochemical potential lines modeled in COMSOL using the 3 - Matic assembled volume mesh shown in Figure 3.12 . Inside COMSOL a n oxygen exchange surface resistance layer was applied to the surface layer of the volume mesh around the electrode to represent the contribution from the infiltrated MIEC oxide nano - particles to account for the non - flat nature of the MIEC infiltrate hemisphere . The boundary condition used by COMSOL for the applied surface resistance layer to the IC scaffold is described: [17] 57 where n is the normal vector, J is the electric current density vector , R surf is the surface resistance of the infiltrated MIEC material (defined in Equation 18) , V is the pote ntial through the IC GDC scaffold material, and is the 1V reference potential applied to t he porous IC scaffold surface. The oxygen exchange surface resist ance value was scaled by multiplying the intrinsic R s value by the surface area ratio between the total cathode surface area and the infiltrated nano - parti cle surface area on the cathode using the equation : [18 ] where R surf is the R s value spread over the surface of the IC scaffold, A Sc is the surface area of the IC scaffold, and A inf is the surface area of the infiltrate. The GDC conductivity was then assigned to the entire electrol yte and cathode IC scaffold volume mesh . Finally, a 1V bias was applied to the entire cathode microstructure surface (all four sides and the top) and a 0V bias was applied to the bottom of the electrolyte microstructure surface to s imulate an electrochemical potential through the microstructure. A pplying a potential to the edges of t he microstructure is unphysical because in real life the edges of the reconstructed microstructure would be connected to other GDC particles in the next r epeating unit and would not be covered by MIEC infiltrate. However, in large enough reconstruction s the se incorrect edge polarizations would not be expected to significantly contribute to the overall cathode results (as discussed in Chapter 8, reconstru ctions of various sizes all yielded identical R P values, indicating these edge effects could be safely ig nored) . The COMSOL FEM program was 58 used to solve ( Equation15 ) to calculate the electrochemical pote ntial lines shown in Figure 3.12 . The current generated from the electrochemical potential lines located at the 0V surface were then integrated (the I term in Equation 16 ) to determine R P values at different operating temperatu r es. Figure 3 .1 3 : FIB - SEM 3D Reconstruction Electrochemical Potential Gradient . The electrochemical potential gradient was generated using the COMSOL computer program. 3.4 Summary In summary, various processing and characterization techniques were used to analyze SOFC devices and better understand the underlying processes. The insights from these experimental techniques are discussed in the following chapters. The next chapters evaluate the effects desiccation and nano - ceria pre - infiltration have on infiltrated MIEC nano - particl e size, MIEC oxide phase purity , and ultimately SOFC symmetrical cell 59 performance. In addition, different solution additives and infiltrated MIEC compositions will be used with desiccation and pre - infiltration to evaluate if infiltrated MIEC nano - particle size can be controlled other parameters as well. The characterization techniques outlined in this chapter are used in the following chapters to evaluate the underlying processes that control MIEC nano - particle size, MIEC oxide phase purity and SOFC perfo rmance. 60 CHAPTER 4 : The Impact of Precursor Nitrate Solution Desiccation on Infiltrated La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - Cathodes 4.1 Introduction This chapter evaluates the desiccation fabrication approach as a means to systematically reduce infiltrated MIEC nano - particle size , and subsequently improve NMCC performance. Processing techniques, such as how desiccation was performed, will be discussed in detail in addition to the c haracterization techniques used . This chapter is intended to initially demonstrate that desiccation reduces nano - particle size . A n in - depth analysis of what promotes this particle size reduction will be pres ented in C hapter 5 . As mentioned in Section 2.1.2 , NMCCs are typically produced by the dissolution of MIEC precursor nitrates in water, the infiltration of these nitrate solutions in to porous IC scaffolds, and the therm al decomposition of the nitrates into MIEC oxide nanoparticles t hat cling to the IC scaffold . Unfortunately, it has been difficult to control the size of MIEC nanoparticles obtained via the infiltration method. For instance, a survey of recent literature indicates that (SSC) ceria NMCCs produced by infiltration, a 10 o C/min heating ramp , and firing at 800 o C for 1 hour after each infiltration step had average particle sizes ranging from 40 to 100 nm [ 5 , 7 , 30 ] . Similarly, LSCF ceria NMCCs produced by infiltration, a 10 o C/min heating ramp, and firing at 800 o C for 1 hour after each in filtration step had average particle sizes ranging from 50 to 60 nm [ 7 ] . Lastly, LSFC ceria NMCCs produced by infiltration, a 10 o C/min heating ramp , and firing at 800 o C for 2 hours after each infiltration step had average LSFC particle sizes ranging from 38 to 80 nm [ 8 , 9 ] . Since past studies have shown that average SSC or LSCF infiltrate particle 61 sizes remain constant with successive infiltration and firing steps [ 7 ] , the cause of these large average particle si ze variations remains a mystery. This chapter shows that the chemical and physical desiccation of LSCF precursor nitrate solutions infiltrated into porous GDC scaffolds , prior to thermal decomposition, can be used to control the average LSCF infiltrate particle size. 4.2 Experimental Methods 4.2.1 Cathode - Electrolyte - Cathode Symmetric Cell Production Cathode - supporting electrolytes were prepared in the following manner. First, 0.8 grams of GDC powder were pressed to 63 MPa in a steel die with a 19 mm di ameter bore . According to manufacturer specifications, t his ultra - high surface area powder (Rhodia; Cranbury, NJ, USA) had an agglomerate d 50 size of 240 nm, a specific surface area of 25.2 m 2 /g, and an average particle size of ~32 nm. The pressed porous pellets were heated to 1475 o C at 3°C/min, held at 1475 o C for ~10 hours, and then cooled to room temperature at a nominal cooling rate of 3°C/min to produce discs with relative densities >95%. T hese discs were sanded flat and parallel with SiC sandpaper u ntil they achieved thicknesses ranging from 432 to 457 µm. Porous, well - necked GDC IC scaffolds were then produced on both sides of these electrolyte pellets. To achieve this, some of the aforementioned Rhodia GDC powder was coarsened at 800°C for 4 ho urs and then mixed with a polymer (Herae us ; West Conshohocken, PA, USA) to form a GDC ink with a 34% solids loading. Three layers of GDC ink were then screen printed onto each side of the dense GDC electrolyte pellet using a patterned 80 mesh stainless st eel screen with a circular 0.5 cm 2 open area. Before the next ink layer was applied, each ink layer was allowed to flow for 5 minutes and then 62 was placed in a bake oven at 120°C for 5 minutes to extract the electronic vehicle solvent and increase the gree n strength. After screen printing, the samples were heated to 400°C at 3°C/min , held at 400°C for one hour, heated to 600°C at 3°C/min, held at 600°C for one hour, heated to 1100 o C at 5 o C/min, held at 1100 o C for 3 hours, and then cooled to room temperatur e at a nominal cooling rate of 10 o C/min. Sintered IC scaffold thickness and roughness measurements were then made with a Dektak 3 profilometer (Bruker; Tucson, AZ, USA ). 1.50 molar MIEC precursor solutions were infiltrated into these IC scaffolds, desicc ated, and fired to coat the IC scaffold surface with MIEC oxide nanoparticles. As done in past studies [ 7 ] , the infiltrated LSCF metal nitrate precursor solutions were prepared by dissolving 99.999% pure La(NO 3 ) 3 *6.3 H 2 O, 99.9965 % pure Sr(NO 3 ) 2 , 99.999% pure Co(NO 3 ) 2 *5.67 H 2 O , and 99.999% pure Fe(NO 3 ) 3 *9.42 H 2 O (Alfa Aesar; Ward Hill, MA, USA) in distilled water containing 3 wt% of pre - dissolved Triton - X 100 (weig ht Triton X - 100/weight nitrate), (These precise ni trate water of hydration contents were determined prior to nitrate weighing by measuring the mass loss that occurred inside a Q500 thermogravimetric analyzer (TA Instruments; New Castle, DE, USA) with heating under nitrogen up to 850 o C ) . These solutions w ere then pipetted into the porous GDC cathode scaffolds, allowed to soak into the scaffold for 5 minutes, and heated in an 80 o C drying oven for 10 min before being desiccated. Based on previous desiccation kinetics studies [ 131 ] , samples desiccated with dry air were desiccated for 30 minutes, sample s desiccated with CaSO 4 were desiccated for 56 hours, and samples desiccated with CaCl 2 were desiccated for 8 - 10 hours. After each desiccation, the samples were heated to 700°C at 10°C/min , held at 700°C for one hour, and then cooled to room 63 temperature a t a nominal cooling rate of 20°C/min. This infiltrate - fire process was repeated 3 times to achieve the desired MIEC loading level of 12.0 vol%. Lastly, symmetric cells were prepared for electrical measurements by screen printing bilayer ( LSM ) Au current collectors onto the surface of each NM CC. The LSM ink was prepared by mixing LSM powders (Praxair Specialty Ceramics; Woodinville, WA, USA) that had a d 50 agglomerate size of 1.1 µm, a specific surface area of 3.00 m 2 /g, a nd an average particle size of ~300 nm with V - 737 electronic vehicle to form an ink with a 34% solids loading. A single layer of LSM ink was then screen printed onto each NMCC using a 290 mesh stainless steel screen with a circular 0.5 cm 2 open area. The samples were then heated to 700 o C at 10°C/min, held at 700 o C for 1 hour, and then cooled to room temperature at a nominal cooling rate of 20°C/min. Herae us C5756 fritted Au paste was then screen printed in an open grid pattern across the surface of each L SM current collector using a patterned 290 mesh stainless steel screen. 4.2.2 Symmetric Cell Impedance Measurements NMCC performance as a function of temperature was evaluated in air , 20%O 2 - 80%He, or 20%O 2 - 80%N 2 under open circuit conditions using EIS . Measurements in air were performed in a static atmosphere, while controlled atmosphere experiments in 20%O 2 - 80%He and 20%O 2 - 80% N2 were performed with a 100 mL/min flow rate. Two - point EIS measurements were taken every 50 degrees between 400 and 700 o C using an IM6 impedance analyzer (Zahner Elektrik GmbH; Kronach, Ge rmany). At each temperature, a 100 mV AC amplitude was used to collect impedance measurements between 100 kHz to 0.1 Hz. All samples were allowed to equilibrate for a minimum of 30 minutes befo re an EIS spectrum was collected. Cathode polarization resistances were 64 extracted by measuring the distance between x - axis intercepts on Nyquist plots that had been multiplied by the geometric cathode area (0.5 cm 2 ) and divided by two (since each symmetric cell had two cathodes). Ohmic resistivity values for each cell were determined by combining the measured distance between the origin and the high frequency x - intercept on a Nyquist plot , the 0.5 cm 2 geometric cathode area , and t he measured electrolyte thicknesses using the definition of resistance: [19 ] 4.2.3 X - ray Diffraction Measurements MIEC infiltrate ph ase purity was evaluated using XRD . XRD analyses were conducted from 20 o 2 80 o with a 0.040 o step, a 1.00 second scan speed, and a copper filament using a Miniflex II (Rigaku Americas Corporation; The Woodlands, TX, USA) operated at 30kV and 15 mA. Due to overlap of the 33.0° LSCF and 33.0° GDC peak, the 47.3° LSCF peak and the 47.4° GDC peak, the 69.2° LSCF peak and the 69.3° GDC peak, and the 78.6° LSCF peak and the 78.9° GDC peak [ 117 , 118 ] , LSCF XRD analyses were conducted on precursor nitrate solutions fire d both inside a GDC scaffold and outside a GDC scaffold on an alumina plate. 65 4.2.4 Scanning Electron Microscopy Measurements NMCC microstructure was evaluated after EIS testing using a Auriga Dual Column Focused Ion Beam - Scanning Electron Microscope (FIB - SEM) (Carl Zeiss aperture, a 40 second scan speed, a 1.4 mm working distance, and magnifications bet ween 100,000 and 300,000 times. SEM samples were prepared by fracturing symmetric cells in half, bonding the remnants to an aluminum vertical sample holder, and sputtering 2.0 nm of tungsten on the surface. Figure 4. 1 [ 61 ] shows the infiltrate particle fracture surface measurements that were performed at the electrolyte - cathode interface. Particle size measurements were conduc ted using the Heyne linear intercept method [ 132 ] , 3 microgr aphs per sample, and at least 35 intercepts per micrograph. 4.2.5 Nano - Micro - Composite Cathode Performance Modeling NMCC per formance was modeled using the SIMPLE m odel [ 5 , 7 , 8 ] . As discussed [ 7 ] , the SIMPLE model accounts for NMCC resistances associated with oxygen exchange across the MIEC infiltrate surface and oxygen transport through a microstructurally idealized IC scaffold. As such, it predicts the lowest - possible NMCC Figure 4.1 : Scanning Electron Microscope Fracture Surface Image . The depicted parts are an u ndesiccated LSCF - GDC NMCC (le ft), and a FIB - SEM Serial Section Showing the Microstructure of a LSM Current Collector, a GDC Scaffold, and a GDC E lectrolyte (right) [61 ]. 66 , since additional sources of resistance (electronic tr ansport losses through the MIEC infiltrate, IC tortuosity induced resistances, gas concentration polarization resistances, etc.) are ignored . properties and easily - measured microstructura l properties . These materials properties include the intrinsic area specific resistance for oxygen surface exchange into the MIEC ( R s ) and the bulk oxygen vacancy conductivity of the IC scaffold ( . These microstructural parameters include th e IC scaffold surface area ( ), the MIEC infiltrate surface area ( ), the cathode thickness ( ), the IC scaffold column repeat unit width ( ), and the scaffold porosity ( is expressed in terms of an analytical expression: [20 ] where: [ 21 ] and: [ 22 ] that can be quickly solved within a spreadsheet or via an online calculator, as done on - demand for a variety of user - selected NMCC materials and geometries at https://www.egr.msu.edu/nicholasgroup/simple.php . Here, the procedures described in the literature [ 7 ] were used to combine the SEM measurements, previous GDC IC scaffold measurements [ 7 ] , literature LSCF measurements [ 51 ] , and literature GDC measurements [ 7 ] to produce SIMPLE m odel NMCC predictions. 67 NMCC perf ormance was al so modeled using the surface resistance (SR) model that assumed oxygen surface exchange was the only source of NMCC resistance [ 5 , 7 , 8 ] . Based on the definition of area specific resistance this model predicts using the following equ ation : [23 ] where is the geometric (i.e. footprint) area of the cathode and the other symbols have their previously defined meanings. 4.3 Results 4.3 . 1 : Desic cant Impacts on Infiltrate Particle Size Both the SIMPLE model and the SR Limit indicate that NMCC performance is influenced by infilt rate oxide nano - particle size. The results in this chapter will demonstrate that precursor gel desiccation systematically reduces infiltrated LSCF oxide nano - part icle size. Figure 4 .2 [ 131 , 133 ] shows a series of representative scanning electron micrographs indicating that the chemical or physical desiccation of precursor nitrate solution s results in the reduc tion of the average size of infiltrated LSCF particles produced by firing precursor nitrate solutions at 700 o C . In fact, precursor nitrate solutions that were infiltrated, air - dried , and subsequently fired at 700 o C produced 48 +/ - 15 nm diameter LSCF nano - particles while those infiltrated, desiccated with CaCl 2 , and subsequently fired at 700 o C, produced 22 + / - 5 nm diameter LSCF nano - particles . 68 Figure 4 .2 : Scanning Electron Micrographs for Desiccated LSCF Nano - Particles Produced by using Triton X - 100 . LSCF oxide particles were produced i nside GDC Scaffolds at 700 o C. Note: Desiccants are Listed in Order of Increasing E ffectiveness . [131,133] Each scale bar is 50 microns in width. Figure 4 .3 [ 131 , 133 ] shows the impact of desiccation and firing temperature on the average LSCF particle size. As indicated in the literature, calcium chloride was the strongest chemical desiccant, followed by calcium sulfate, and finally treatment in dry air [ 131 , 133 ] . Desiccant strength ranking is based on each chemical desiccant s thermodynamic driving force to lower the water partial pressure in the infiltrated precursor solution. At each firing temperature, increased amo unts of precursor nitrate solution desiccation produced smaller LSCF particles. For each desiccant, increased firing temperature resulted in larger LSCF particles. Given the high number of particles counted (>100 in each case) and the high resolution of th e SEM, the error bars mainly represent the actual infiltrate particle size distribution within each NMCC. 69 Figure 4 .3 : Bar - Graph of Desiccated Average LSCF Infiltrate Particle Sizes from Scanning Electron Microscopy Im ages . 4, 2 . Desiccants are listed in order of increasing effectiveness. [131,133] Error bars are +/ - a standard deviation calculated using the SEM - measured particle size distribution. 4.3 . 2 : Desicc ant Impacts on Infiltrate Phase Purity Figure 4. 4 [ 117 , 118 , 134 - 137 ] shows that LSCF infiltrate phase purity remained constant with desiccation an d firing at 700 o C. Although intense signal s from the large volume of GDC sc affold particles in the Figure 4 .4 a NMCCs XRD scans made the LSCF and impurity phase peaks difficult to resolve, those peaks we re well resolved in the Figure 4 .4 b XRD scans taken on loose LSCF powders produced outside a GDC scaffold. Regardless of the precursor nitrate solution desiccation conditions, LSCF was the dominant infiltrate phase; making up ~85% of the infiltrate material based on peak intensity ratios. Similar phase purity has been observed in other LSCF - GDC cathodes reported in the literature [ 7 , 138 , 139 ] . Although generally similar to the phase purity observed after firing at 700 o C, LSCF - GDC NMCCs fired at 600 o C (not shown) had a larger percentage of impurity phases, while those fired at 800 o C had a smaller percentage of impurity phases (not shown). NMCCs fired at 700 o C were chosen for their combination of high LSCF infiltrate phase purity and small LSCF infiltrate particle size. It is important to note that the identity and phase fraction of each impurity phase shown 70 in Figure 4.4 remained essentially constant as the precursor nitrate solution desiccation conditions were varied. Figure 4 .4 : XRD Scans of LSCF Nano - Particles Produced by Firing Desiccated Precursor Nitrate Solutions . Solutions were fired at 700°C : a) outside a porous GDC scaffold, and b) inside a porous GDC scaffold . 0.9 Gd 0.1 O 1.95 (JCPDS # 01 - 075 - 0161 ) [118] . =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 (JCPDS # 00 - 048 - 0124) [117] , =CoCo 2 O 4 (JCPDS # 01 - 080 - 1537) [134] , =La 2 O 3 (JCPDS # 00 - 040 - 1279) [135] , =La(Co 0.42 Fe 0.58 O 3 ) (JCPDS # 01 - 074 - 9369) [136] , =La 2 O 3 (JCPDS # 01 - 071 - 4953) [137] . 71 4. 3. 3 : Desiccant Impacts on Cathode Electrochemical Performance Table 1: Desiccated TXD LSCF - GDC NMCC Cathodes . Processing parameters are also shown. Figure 4 .5 [ 131 , 133 ] shows EIS data for the LSCF - GDC symmetric cells listed in Table 1. EIS measurements taken using both 50 mV and 100 mV AC amplitudes produced similar results, indicating that these EIS results represent open - circuit behavior. The observed EIS curv e shapes were similar to those observed previously for LSCF - GDC NMCCs [ 6 , 7 , 9 ] . As shown in Table 1 [ 131 , 133 ] , the only significant, systematic difference between these cells was the LSCF infiltrate particle size. This suggests that the systematic reduction in polarization r esistance with desiccation shown in Figure 4 . 5 resulted from the reduction in LSCF infiltrate particle size. Desiccant Type LSCF Loading Level (Vol %) Cathode Thickness (µm) Scaffold RMS Surface Roughness (µm) Electrolyte Thickness (µm) LSCF Diameter (nm) Total LSCF Surface Area (cm 2 ) Total GDC Surface Area (cm 2 ) LSCF Infiltrate Hemispherical Footprint Area/GDC Surface Area (%) Air - LSCF Surface Area/GDC Surface Area (%) Undesiccated 12.0 35.2 4.39 457 46 275 146 95 188 Undesiccated 12.0 36.1 4.05 457 46 282 150 95 188 Dry Air 12.0 32.1 4.11 432 30 385 133 145 289 Dry Air 12.0 35.8 3.91 432 30 429 148 145 290 CaSO 4 12.0 34.9 3.67 432 25 503 145 173 347 CaSO 4 12.0 36.7 4.64 432 25 529 152 173 348 CaCl 2 12.0 36.6 4.28 432 22 599 152 197 394 CaCl 2 12.0 36.2 4.36 432 22 592 150 197 395 Note, desiccants are listed in order of increasing effectiveness [118, 120 ] . All samples in this table were produced by infiltrating 1.50 M LSCF precursor solution into GDC scaffolds with a geometric area of 0.50 cm 2 , an initial porosity of 32%, and a scaffold column width of 120 nm. See the experimental methods section for additional details. 7 2 Figure 4 .5 : Desiccated LSCF - GDC NMCC R P EIS Nyquist Plot . 4, 2 . Desiccants are listed in order of increasing effectiveness [131,133] . 73 Figure 4. 6 (derived from the EIS measurements including those in Figure 4. 5) shows that increased amounts of precursor nitrate solution desiccation led to improved SOFC NMCC performance across the full 400 - 700 o C range. Further, the cathode operating temperature (defined here as the temperature at which 2 , after Steele and Heinzel [ 16 ] ) decreased from 640 o C for an undesiccated LSCF - GDC NMCC to 575 o C for a CaCl 2 - desiccated LSCF - GDC NMCC. The 1.0 - 1.1 eV 400 - 600 o C activation energie s displayed in Figure 4 . 6 are similar to the 1.1 eV 400 - 600 o C activation energy for LSCF reported in literature [ 51 ] . This suggests that poor oxygen exchange into the LSCF infiltrate par ticles was the dominant source of NMCC resistance between 400 and 600 o C. The 0.56 eV 650 - 700 o C activation energy displayed in Figure 4 . 6 is identical to the 0.56 eV activatio n energy for GDC reported [ 7 ] . Figure 4 .6 : Desiccated LSCF - GDC NMCC R P Arrhenius Plots . Plots p roduced by Desiccated Firing P rec ursor Nitrate Solutions at 700°C. R P predictions for the undesiccated (top, red) and CaCl 2 - de siccated (bottom, black) NMCCs are compared to the a) SIMPLE model, and b) SR model. =Undesiccated, 4, 2 . The so lid inclined lines are a guide to the eye linking the 2 performance target. 74 This suggests that poor ionic conduction through the GDC scaffold was the dominant source of NMCC resistance between 65 0 and 700 o C. Figure 4 . 6a also shows SIMPLE model [ 5 , 7 , 8 ] predictions for the undesiccated and CaCl 2 - desiccated LSCF - GDC NMCCs. These predictions were made using the average SEM - measured particle sizes shown in Figure 4 . 3, the intrinsic LSCF oxygen surface exchange resistance [ 51 ] , the GDC oxygen i on conductivity [ 7 ] , and the NMCC microstructural properties listed in Table 1. Given the microstructural simplicity of the SIMPLE model, it is perhaps unsurprising that the model predictions were not an exact fit to the experimentally measured data. However, it is interesting to note that the factor of two 400 - 600 o C difference predicted by the SIMPLE model when changing the average LSCF infiltrate particle size from 48 nm (the size observed for the undesiccated samples) to 22 nm (the size observed for the CaCl 2 - desiccated samples) is similar to the measured 400 - 600 o C difference between the CaCl 2 - desiccated and undesiccated samples. Figure 4 . 6b shows SR model [ 5 , 7 , 8 ] predictions for the undesiccated and CaCl 2 - desiccated LSCF - GDC NMCCs. Like the SIMPLE model, the SR model predicted a factor of two 400 - 600 o C difference when changing the average LSCF infiltrate particle size from 48 nm (the size observed for the undesiccated samples) to 22 nm (the size observed for the CaCl 2 - desiccated samples). This difference is close to the experimentally measured 400 - 600 o C difference between the undesiccated and CaCl 2 - desiccated NMCCs, sug gesting that the observed reductions were solely the result of desiccation - induced infiltrate particle size reductions. 75 Figure 4 . 20%O 2 - 80%He, and 20%O 2 - 80%N 2 . If gas concentration polarization were a significant source of resistance, the 3.5 times higher diffusivity of oxygen in a 20%O2 - 80%He Figure 4 .7 : Arrhenius R P Plot for Desiccated LSCF - GDC NMCCs Tested under Different Atmospheres . A ir ( ), 20%O 2 - 80%H e ( ), and 20%O 2 - 80%N 2 ( ). mixture compared to that in air or a 20%O2 - 80%N 2 mixture would be expected to lead a significantly lower , as has been the case for other SOFC cathodes [ 126 ] . However, the essentially identical Figure 4.7 values at 400 - 600°C indicate that concentration polarization resistance was not a major source of resistance in undesiccated NMCCs in this temperature range. Therefore, the desiccation - induced performance improvements shown in Figure 4 . 6 were not the result of a reduction in concentration polarization resistance. Temperature me asurements were not conducted in pure O 2 because an oxygen partial pressure greater than air will improve the oxygen diffusion through the cathode microstructure , but also lower the oxygen vacancy concentration in 76 the MIEC material [ 95 ] . All three gas mixtures listed in Figure 4.7 contain approximately the same O 2 concentration where only the O 2 diffusivity was altered. Figure 4 . 8 [ 7 ] shows that 400 - 700 o C ohmic resistivity of all the NMCCs tested in air. Identical results (not shown) were also observed for the undesiccated NMCCs tested in 20%O 2 - 80%He and 20%O 2 - 80%N 2 [ 7 ] demonstrated that electronic transfer losses within infiltrated electrodes increase both t he polarization resistance and the ohmic resistivity. The good agreement with the ohmic resistivity of each cell and that of pure Figure 4 .8 : Arrhenius Ohmic Resistivity Plots for LSCF - GDC NMCCs Tested in Air . Samples produced by firing desiccated precursor nitrate s olutions at 700°C. 4, and 2 . The inclined solid line is the resistivity of pure GDC from the literature [7] . GDC [ 7 ] in Figure 4 . 8 indicates that oxygen ion transport through the electrolyte was the only source of ohmic resistance and that electronic losses within the current collectors or through the network of MIEC nano - particles were insignificant in magnitude. This indicates tha t the desiccation - induced performance increases shown in Figure 4 . 6 were not the result of a reduction in electronic resistance. The fact that the Table 1 LSCF infiltrate hemispherical footprint area to GDC surface area ratios were much greater than 77 the 44% needed to ensure percolation between randomly deposited non - overlapping circles on a 2D plane [ 140 ] also support s this interpretation. The previously mentioned constancy of the infiltrate phase compositions and fractions with desiccation in Figure 4 . 4 suggests that the desiccation - induced performance increases s hown in Figure 4 . 6 were not related to changes in LSCF phase purity. Additional studies performed on phase pure LSCF infiltrate showing similar desiccation - induced and infiltrate particle size behavior [ 61 ] confirm th is. Taken together, the data in Figures 4 . 2 - 4 . 8 suggest that precursor nitrate solution desiccation led to a reduction in average LSCF infiltrate particle size that, in turn, lead to improved NMCC electrochemical performance. Although not previously recog nized as a method to tailor infiltrate particle size, desiccation may be the reason why Bansal and Wise [ 79 ] were able to use an unconventional, 15 hour, 300 o C precursor nitrate solution pre - heating treatment to produce 15 nm SSC particles, while other studies not employing a precursor nitrate solution pre - heating treatment [ 5 , 7 , 30 ] but using an identical maximum firing temperature of 800 o C for 1 hour produced SSC particles larger than 40 nm in diameter. In add ition, as mentioned in Section 2.6.1, EISA may be the mechanism which controls the desiccation process to reduce nano - particle size when using Triton X - 100. By desiccating the precursor solution prior to thermal decomposition this chapter has shown that d esiccation can be a useful tool for tailoring SOFC infiltrate particle size and infiltrated electrode electrochemical performance . Future chapters determine if desiccation has a similar effect on other infiltrate solution co mpositions, and provide a clear er understanding how desiccation a lters infiltrate particle size. 78 4.4 Summary This chapter represents the first time that chemical desiccants have been shown to impact the size of infiltrated SOFC nano - particles. For infiltrated LSCF in particular, precursor nitrate solution desiccation and firing at 700 o C lowered the average LSCF particle size from 48 to 22 nm. However, desiccation did not alter the infiltrate phase purity, the NMCC concentration polarization resistance, or the NMCC electronic resistance. These results, coupled with mathematical predictions made using the observed infiltrate particle sizes, indicate that the observed 65 o C drop in cathode operating temperature was solely the result of desiccation - induced infiltrate particle size reductions. The precursor nitrate solution desiccation technique explored in this chapter may be useful for tailoring the size of nano - particles used as catalyst s , fuel cell s , and other applications. 79 CHAPTER 5: The Impact of Surfactants on Desiccated L a 0.6 S r 0.4 C o 0.8 F e 0.2 O 3 - Infiltrated S olid O xide F uel C ell Cathodes 5.1 Introduction Many studies have use d organic solution additives (surfactants , such as Triton X - 100, and chelating agents , such as Citric Acid, etc. ) [ 5 , 7 ] to alter MIEC pre cursor solution behavior . Chapter 4 introduced precursor gel desiccation and showed that the desiccation of Triton X - 100 containing LSCF precursor nitrate solutions was capable of controlling the average LSCF oxide infiltrate nano - particle size. The objective of this chapter was to examine the effect desiccation has on infiltrate solution nano - particle sizes containing different organic solution additives. A deeper evaluation of the desiccation processes is also conducted. 5.2 Experimental Method s 5.2.1 Cathode - Electrolyte - Cathode Symmetric Cell Production Symmetrical infiltrate d cathode - electrolyte cell fabrication was performed in the s ame manner as in Section 4 .2.1 . Specifically , GDC IC scaffolds were screen printed and sintered the same way as in Section 4.2.1. Infiltra ted 1.50 molar Triton X - 100 derived (TXD) , 0 .50 molar Citric A cid derived (CAD) LSCF , and 1.50 molar Pure nitrate derived (PND) precursor solutions were infiltrated into sintered GDC IC scaffolds, desiccated, and fired in th e same manner as Section 4.2.1 . Infiltrated TXD LSCF precursor solution s were prepared in the same manner as Section 4.2.1 and PND LSCF precursor solutions are the same as TXD LSCF solutions without the addition of a solution additive . Infiltrated CAD LSCF precursor solutions were fabricated by first 80 dissolving 99.999% pure La(NO 3 ) 3 *6.3 H 2 O, 99.9965 % pure Sr(NO 3 ) 2 , 99.999% pure Co(NO 3 ) 2 *5.67 H 2 O , and 99.999% pure Fe(NO 3 ) 3 *9.42 H 2 O (Alfa Aesar; Ward Hill, MA, USA) in disti lled water containing 0.75 mol Citric A cid/mol metal nitrates. (These precise nitrate water of hydration contents were determined prior to nitrate weighing by measuring the mass loss that occurred inside a Q500 thermogravimetric analyzer (TA Instruments; New Castle, DE, USA) with heatin g under nitrogen up to 850 o C). These different molar solutions were then allowed to soak into the porous GDC scaffolds and gelled in a similar manner as Section 4.2.1. D esiccation time s were based on previous desiccation kinetics literature studies [ 131 ] . PND LSCF, CAD LSCF, and TXD LSCF infiltrated cells were all heated to 700°C at 10°C/min , held at 700°C for one hour, and then cooled to room temperature at a nominal cooling rate of 20°C/min after being desiccated. This infiltrate - gel - desiccate - fire routine was repeated up to 6 times to achieve the desired MIEC loading level of 12.0 vol%. Lastly (LSM) Au current collectors were screen print ed onto the surface of each NMCC. 5.2.2 Symmetric Cell Impedance Measurements Just like in Section 4.3.3 NMCC performance as a function of temperature was evaluated in air, 20%O 2 - 80%He, or 20%O 2 - 80%N 2 under open circuit conditions using EIS . Measurements in ambient air were performed in a static atmosphere, while controlled atmosphere experiments in 20%O 2 - 80%He and 20%O 2 - 80%N2 were performed with a 100 mL/m in flow rate. Since each atmosphere contained 20% O 2 the only difference was the diffusivity of O 2 in those mixtures . EIS measurements were taken in the same manner as Section 4.2.2. 81 5.2.3 X - ray Diffraction Measurements MIEC infiltrate phase purity was evaluated using XRD in the s ame manner as was described in Section 4.2.3 . 5.2.4 Scanning Electron Microscopy Measurements NMCC microstructure was evaluated after EIS testing using an Auriga Dual Column Focused Ion Beam - Scan ning Electron Microscope (FIB - SEM) (Carl Zeiss Microscopy GmbH; Jena, Germany) in the s ame manner as was described in Section 4 .2.4 . 5.2.5 Nano - Micro - Composite Cathode Performance Modeling The NMCC performance was mo deled using the SIMPLE m odel [ 5 , 7 , 8 ] and was performed in the same manner as was described in Section 4 .2.5 . 5.2.6 Williamson - Hall Particle Size and Strain Calculations LSCF nano - particle sizes were calculated via the XRD Williamson - Hall method [ 122 ] for powders fabricated outside a GDC scaffold. The full - width half - max (FWHM) values for the LSCF peaks located at ~22.7 0 , ~ 40.7 0 , and ~ 46.9 0 (2 - Theta) were used for TXD LSCF and CAD LSCF precursor solutions , but PND LSCF precursor solutions calculated particle sizes using peaks only at ~22 0 and ~40.7 0 due to the PND LSCF solutions being amorphous even at firing temperatures above 600°C . The Lorentzian profile shape function using a parabolic background was used to determine the FWHM values using JADE 9. eta) were then used to calculate nano - particle size and strain values in the same manner described in Section 82 3.2.4. All other peaks had multiple planes that contribu te to peak broadening and therefore could not be used to determine particle size from XRD peak broadening. 5.3 Results 5.3 . 1 : Desic cant and Solution Additive Impacts on Infiltrate Particle Size Table 2 shows the processing parameters used when fabricating symmetrical PND , TXD, or CAD L SCF NMCCs . Since the precursor solutions used different molarities, ( 0.5M and 1.5M ) , the infiltrate MIEC particle size was evaluated to see if it was influenced by precursor solution molarity. As shown in Table 2 n ano - particle size, at least for PND and TXD LSCF precursor solutions were not altered by the sol ution molarity. CAD LSCF precursor solution. Similarly the precursor solution infiltration volume was found to not influence the infiltrated oxide nano - particle size . These re sults indicate that it was legitimate to compare particle sizes between the 1.5 molar PND 1.5 molar TXD, and 0.5 molar CAD LSCF solutions. Figure 5.1 [ 131 , 133 ] shows that the infiltrated nano - particle size using PND LSCF and TXD LSCF d id not change when using different solution molarities. Figure 5 . 1 : Scanning Electron Micrographs of Desiccated LSCF Nano - Particles Produced by using Different Triton X - 100 Solution Volume s . 1.5 molar PND LSCF using a) undesiccated and b) desiccated precursor solutions, and 0.50 molar TXD LSCF c) unde si c cated and d) desiccated precursor solutions inside GDC scaffolds at 700 o C. Note, desiccants are listed in order of increasing effectiveness [131,133] . Each scale bar is 50 microns in width. 83 Table 2: Desiccated PND, TXD and CAD LSCF - GDC NMCC Cathodes . Processing parameters are also shown. Surfactant Desiccant Type Firing Temperature (°C) LSCF Loading Level (%) Number of LSCF Infiltrations LSCF Nitrate Solution Molarity Cathode Thickness (µm) Scaffold RMS Surface Roughness (µm) Volume of Solution per Infiltration Cathode Geometric Area (cm 2 ) Cathode Porosity (%) Scaffold Column Width (nm) LSCF Diameter (nm) Total LSCF Surface Area (cm 2 ) Total GDC Surface Area (cm 2 ) LSCF Surface Area/GDC Surface Area (%) PND Undesiccated 700 12.0 3 1.5 35.7 4.00 2.78 0.5 32 120 65 198 148 134 PND Undesiccated 700 12.0 3 1.5 36.3 4.05 2.82 0.5 32 120 65 202 151 134 PND Undesiccated 700 12.0 6 0.5 35.4 3.50 4.13 0.5 32 120 62 206 147 140 PND Undesiccated 700 12.0 6 0.5 36.4 3.41 4.25 0.5 32 120 62 212 151 140 PND Dry Air 700 12.0 3 1.5 34.3 3.50 2.67 0.5 32 120 73 170 142 120 PND Dry Air 700 12.0 3 1.5 35.2 3.90 2.74 0.5 32 120 73 174 146 120 PND CaSO 4 700 12.0 3 1.5 36.3 4.30 2.82 0.5 32 120 72 182 151 121 PND CaSO 4 700 12.0 3 1.5 36.0 4.42 2.80 0.5 32 120 72 180 149 121 PND CaCl 2 700 12.0 3 1.5 35.2 4.61 2.74 0.5 32 120 71 179 146 123 PND CaCl 2 700 12.0 3 1.5 35.7 4.04 2.78 0.5 32 120 71 182 148 123 PND CaCl 2 700 12.0 6 0.5 34.8 4.10 4.06 0.5 32 120 70 179 144 124 PND CaCl 2 700 12.0 6 0.5 35.8 3.95 4.18 0.5 32 120 70 185 148 125 CAD Undesiccated 700 12.0 6 0.5 34.2 3.74 3.98 0.5 32 120 50 246 142 173 CAD Undesiccated 700 12.0 6 0.5 34.5 3.54 4.02 0.5 32 120 50 249 143 174 CAD Dry Air 700 12.0 6 0.5 26.7 3.84 3.11 0.5 32 120 43 224 111 202 CAD Dry Air 700 12.0 6 0.5 31.8 4.67 8 3.70 0.5 32 120 43 266 132 202 CAD CaSO 4 700 12.0 6 0.5 27.7 4.62 3.23 0.5 32 120 42 237 115 207 CAD CaSO 4 700 12.0 6 0.5 29.5 4.35 3.44 0.5 32 120 42 253 122 207 CAD CaCl 2 700 12.0 6 0.5 32.1 3.50 3.74 0.5 32 120 41 282 133 212 CAD CaCl 2 700 12.0 6 0.5 34.1 3.55 3.97 0.5 32 120 41 300 141 213 TXD Undesiccated 700 12.0 3 1.5 35.2 4.39 2.74 0.5 32 120 48 275 146 188 TXD Undesiccated 700 12.0 3 1.5 36.1 4.05 2.81 0.5 32 120 48 282 150 188 TXD Undesiccated 700 12.0 3 0.5 36.0 3.50 8.40 0.5 32 120 49 254 149 170 TXD Undesiccated 700 12.0 9 0.5 36.3 3.41 2.82 0.5 32 120 51 256 151 170 TXD Dry Air 700 12.0 3 1.5 32.1 4.11 2.50 0.5 32 120 30 385 133 289 TXD Dry Air 700 12.0 3 1.5 35.8 3.91 2.78 0.5 32 120 30 429 148 290 TXD CaSO 4 700 12.0 3 1.5 34.9 3.67 2.71 0.5 32 120 25 503 145 347 TXD CaSO 4 700 12.0 3 1.5 36.7 4.64 2.85 0.5 32 120 25 529 152 348 TXD CaCl 2 700 12.0 3 1.5 36.6 4.28 2.84 0.5 32 120 22 599 152 394 TXD CaCl 2 700 12.0 3 1.5 36.2 4.36 2.81 0.5 32 120 22 592 150 395 TXD CaCl 2 700 12.0 3 0.5 35.2 4.30 8.20 0.5 32 120 23 551 146 377 TXD CaCl 2 700 12.0 9 0.5 36.6 4.21 2.85 0.5 32 120 23 575 152 378 84 Figure 5 . 2 [ 131 , 133 ] shows SEM images of undesiccated and Ca Cl 2 - desiccated symmetri cal cells infiltrated with 0.50 molar TXD LSCF precursor solutions using different infiltration solution volume s. Figure 5.2 shows that the infiltration solution volume did not impact the average infiltrated oxide nano - particle size . Figure 5 .2 : Scanning Electron Micrographs of Desiccated LSCF Nano - Particles Produced using Different Solution Additives . Samples used a) undesiccated 3 infiltrations, b) desiccated 3 infiltrations, c) undesiccated 9 infiltrations and d) desiccated 9 infiltrations, precursor solutions inside GDC scaffolds at 700 o C. Note, desiccants are listed in order of increasing effectiveness [131,133 ] . Each scale bar is 50 microns in width. Figure 5.3 [ 131 , 133 ] shows that desiccation reduced the nano - particle size of TXD LSCF and CAD LSCF, but not the PND LSCF nano - particle size. SEM images of these infiltrated NMCCs show that both TXD and CAD nano - particles decreased their average size as the s trength of the desiccant increased, while PND infiltrated oxide nano - particle sizes remained constant with increasing desiccant strength. Specifically, the TXD average infiltrated oxide nano - particle sizes were reduced from 50 nm to 22 nm, the CAD average infiltrated oxide nano - particle sizes were reduced from 48 nm to 41 nm and 85 the PND infiltrated oxide average nano - particle sizes remained constant with increasing chemical or physical desiccation strength at around 70 nm in diameter on average. Figure 5 .3 : Scanning Electron Micrographs of Desiccated LSCF Nano - Particles Produced using Differ ent Solution Additives . Pure nitrate - containing precursor nitrate solutions (a - d), 1.50 molar Triton X - containing precursor nitrate solutio ns (e - h) and 0.50 molar Citric A cid - containing precursor solutions (i - l) inside GDC scaffolds at 700 o C. Note, desiccants are listed in order of increasing effectiveness [131,133] . Each scale bar is 50 microns in width. Figure 5.4 shows the raw data used to calculate the XRD Will iamson - Hall particle sizes and strain s for the LSCF particles resulting from PND, TXD and CAD LSCF precursor solutions fired at 600°C, 700°C and 800°C outside a GDC scaffold . Due to the amorphous nature of the 600 and 700°C fired PND LSCF (as mentioned ab ove) Williamson - Hall particle sizes could only be determined at the 800°C firing temperature. Further, p article sizes for the 800°C PND LSCF were calculated using only two peak s (instead of the standard 3) due to the amorphous nature of the powder. In al l cases, a 86 - Hall technique could be applied to determine size and strain. Figure 5.4 : Desiccated Williamson - Hall Raw Data Plot for PND LSCF, TXD LSCF and CAD LSCF Nano Par t i cles . Data was collected from XRD scans from samples fired at 600°C, 700°C a nd 800°C. Figure 5.5 shows the XRD Williamson - Hall calculated strain % determined from Figure 5.4 for the desiccated LSCF nano powders outside a GDC scaffold using PND, TXD and CAD solution additives. Strain was calculated directly from the slope of each data set in Figure 5.4 using the error bars to calculate the standard deviation values shown in Figure 5.5. 87 Figure 5.5 : Strain Plots for Desiccated PND LSCF, TXD LSCF and CAD LSCF Nano - Particles . S train was calculated using undesiccated, dry air, CaSO 4 , and CaCl 2 desiccants fired at 600°C, 700°C and 800°C for 1 hr. Overall, t he strain for PND LSCF nano - particles, for all desiccation strengths, was much lo wer than CAD and TXD LSCF strain values, and was consistently similar in magnitude . Since the PND LSCF nano - particles also did not change size with desiccation strength it also makes sense the desiccation would not impact the PND LSCF nano - particle strain either. The TXD and CAD LSCF nano - particles showed similar trends where the strain was greatest at 600°C, was then reduced at 700°C, and stayed relatively constant at 800°C. The strain magnitude for TXD LSCF nano - particles at 600°C did show some overlap shown by the error bars. Overall t he effect of desiccation on strain for TXD LSCF and CAD LSCF nano - particles is not conclusive. Figu re 5.6 [ 131 , 133 ] shows the average oxide nano - particle sizes for PND, TXD and CAD LSCF using various desiccant strengths when fired at 600°C, 700°C or 800°C for 1 hr. Using the techniques described in Section 4.2.4 the particle sizes were calculated directly from th e SEM images shown in Figure 5.3 , and standard deviation values were also calcula ted from the same SEM images. In Figure 5.6 t he SEM calculated particle sizes are displayed as opaque colored bars in the foreground. XRD Williamson - Hall calculated particle sizes from LSCF powders were also shown in transparent colors and were calculated from the raw XRD Williamson - Hall data in Figure 88 5.4. XRD particle sizes were unable to be calculated for the PND LSCF powders fabricated at 600°C and 700°C due to the powders being amorphous. In contrast, both TXD and CAD LSCF oxide nano - particle sizes were reduced when using desiccation, at least when fabricated above 600°C. Figure 5.6 shows that PND LSCF nano - particle sizes were not reduced with desic cation at 600°C, 700°C or 800°C, ( t he average nano - particle size is depicted in Figure 5.6 as a solid horizontal black line to show that the average nano - particle size for each fabrication temperature ) was within the standard deviation for each temperature , and thus was not influenced by desiccation . The black horizontal bar in the 600°C TXD and CAD LSCF data indicates that desiccation had no effect on the particle size of TXD and CAD LSCF fired to 600°C. 89 Figure 5 .6 : Bar - Graph of Average Desiccated LSCF Infiltrate Particle Size Produced using Different Solution Ad d itives . Samples were: a) 1.50 molar pure nitr ate - containing, (b) 0.50 molar Citric A cid - containing and c) 1.50 molar Triton X - containing precursor nitrate solutions at various temperatures. PND: =Undesiccated, =Dry Air, =CaSO 4 , and =CaCl 2 CAD: =Undesiccated, =Dry Air, =CaSO 4, and CaCl 2 and TXD: =Undesiccated, =Dry Air, =CaSO 4 and =CaCl 2 . Particle size data collected from SEM images are shown with 100% opacity and are striped pointing to the left, while particle size data collected using the Williamson - Hall method are shown with 50 % opacity and are striped pointing to the right. Desiccants are listed in order of increasing effectiveness [131,133] . Error bars are +/ - a standard deviation calculated using the SEM - measured particle size distribution. 90 Figure 5.7 [ 131 , 133 ] shows the raw EIS data for the desiccat ed CAD LSCF (left) and TXD LSCF (right) symmetrical cells at 500°C, 600°C and 700°C operating temperatures. The outermost nested impedance data was taken at an operating temperature of 500°C while the innermost nested impedance data was taken at an operat ing temperature of 700°C. PND LSCF raw data was collected for both the undesiccated and CaCl 2 - desiccated cases , but show ed no change in performance when desiccated . On the other hand, both the CAD and TXD LSCF raw EIS data show that the cathode resistanc e decreases with both increasing temperature and increasing desiccation strength. Figure 5.8 shows the R P data for desiccated PND, CAD and TXD LSCF symmetrical cells determined from the raw EIS data of Figure 5.7 . Consistent with the collected particle si ze trends, t he R P data for CAD and TXD LSCF were both reduced with increasing desiccation strength, while PND did not change R P with desiccation. Desiccation lowered the operating temperature (the temperature at which a R P value of 2 is achieved ) from 650°C to 570°C with TXD, and from 700°C to 650°C with CAD. In Figure 5.8 SIMPLE model R P predictions are also included for the undesiccated and CaCl 2 - desiccated cases using LSCF nano - particle sizes from Figure 5.1. For TXD and CAD LSCF, t he SIMPLE model predictions show similar R P reductions compared to the experimental results suggesting that nano - particle size reduction causes the measured performance gains. The SIMPLE model predictions also do not show any change in R P for PND LSCF which makes sense since Figure 5.1 shows no change in particle size. The SIMPLE model in all cases predicts lower R P values than the experimental results because (as mentioned in Section 2.5.2) the SIMPLE model does not take into account 91 tortuosity effects in the sca ffold microstructure which can provide an additional source of resistance. Figure 5 . 7 : Raw Impedance Data Plots of Desiccated LSCF - GDC NMCCs Produce using Citric Acid and Triton X - 100 . Samples are: 0.50 molar CAD LSCF (left) and 1.50 molar TXD LSCF (right) precursor nitrate solutions fired at 700°C. The outermost data was taken at 500°C, the nested data was taken at 600°C and the double nested data was taken at 700°C. CAD: =Undesiccated, =Dry Air, =CaSO 4, and =CaCl 2. TXD: =Undesiccated, =Dry A ir, =CaSO 4 and =CaCl 2 . Desiccants are listed in order of increasing effectiveness [131,133] . 92 Figure 5. 8 : LSCF - GDC R P Arrhenius Plots for Desiccated NMCCs Produced using Different Solution Additives . Samples are: a) 1.50 molar PND LSCF, b) 0.50 molar CAD LSCF, and c) 1.50 molar TXD LSCF precurs or nitrate solutions fired at 700°C fo r 1 hour. PND : =Undesiccated , inclined lines are a guide to the eye linking the experimental data. The dashed inclined lines are SIMPLE model prediction s for the undesiccated and CaCl 2 cases. The dashed horizontal line is the 2 performance target. . 93 Figure 5.9 [ 7 ] shows the ohmic re sistivity of the GDC IC scaffold at the different operating temperatures. All the ohmic resistivity data for PND, CAD and TXD LSCF symmetrical cells match the resistivity of pure GDC, which indicated that differences in performance shown in Figure 5.8 wer e not caused by electronic losses in the scaffold. Figure 5. 9 : Arrhenius Ohmic Resistivity Plots for Desiccated LSCF - GDC NMCCs Tested in Air and produced using Different Solution Additives . Samples are fired at 700°C. PND: =Undesiccated, =Dry Air, =Dry Air, =CaSO4 and =CaCl2. The inclined solid line is the resistivity of pure GDC from literature [7 ]. 94 5.3 . 2 : Desicc ant and Solution Additive Impacts on Infiltrate Phase Purity Figure 5. 10 shows ex - sit u XRD data for PND, TXD and CAD LSCF oxide powder s fired at 700°C for 1 hr using d ifferent desiccants. PND LSCF oxide powder XRD scans are shown on the left column , TXD LSCF oxide powder XRD scans are show n on the middle column, and CAD LSCF oxide powder XRD scans are shown on the right column. Each column, going from top to bottom, depicts XRD scans of each powder using increasingly stronger desiccants. Comparison of t he measured XRD data with JCPDS reference spectrum [ 117 , 134 - 137 , 141 , 142 ] indicates that LSCF was the dominant phase by volume in all cases. Within the leftmost column the phase impurity of the PND LSCF powder decreases with increasing desiccant strength, while the secondary phase i mpurity of the TXD LSCF and CAD LSCF powders remains relatively constant wh en using different desiccants. The fact that desiccant strength does not change the Figu re 5.10: XRD Scans of Desiccated LSCF Nano - Particles Produced using Different Solution Add itives . Samples are: a) 1.50 molar PND LSCF , b) 0.50 molar CAD LSCF , and c) 1.50 molar TXD LSCF nitrate solutions fired at 700°C . 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 (JCPDS # 00 - 048 - 0124) [117] , =Co 3 O 4 (JCPDS # 01 - 074 - 2120) [141] , =CoCo 2 O 4 (JCPDS # 01 - 080 - 1537) [134] , =(La 0.38 Sr 0.62 ) 2 FeO 4 (JCPDS # 01 - 072 - 7576) [142] , =La 2 O 3 (JCPDS # 00 - 040 - 1279) [135] , =La(Co 0.42 Fe 0.58 O 3 ) (JCPDS # 01 - 074 - 9369) [136] , 2 O 3 (JCPDS # 01 - 071 - 4953) [137] . 95 secondary phase impurity fraction significantly for TXD and CAD LSCF powder suggests that LSCF phase purity changes were not responsible for the performance changes observed in Figure 5.8. Figure 5.11 [ 117 ] shows undesiccated and CaCl 2 - desiccated XRD data for PND LSCF precursor solutions fired at various temperatures between 80°C and 800°C. The phase purity of the undesiccated powder, compared to the CaCl 2 - desiccated powder , did not improve and actually became worse at temperatures 600°C and above. The PND LSCF phase became dominant at fabrication temperatures of 700°C or greater. Figure 5.11: XRD Scans for CaCl 2 - Desiccated PND LSCF Fired between 80°C and 800°C . La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (PDF #00 - 048 - 0124) [117]. 96 Figure 5.12 [ 117 ] shows undesicca ted and dry air - desiccated XRD data for TXD LSCF precursor solutions fired at various temperatures between 80°C and 800°C. The phase purity of the undesiccated powder compared to the desiccated powder was improved at 600°C, 500°C, 400°C and 80°C, while 30 0°C and 200°C did not show any significant difference. Figure 5.1 2 : XRD Scans for Dry Air - Desiccated TXD LSCF Fired between 80°C and 800°C . La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (PDF #00 - 048 - 0124 ) [117] . 97 Figure 5.13 [ 117 ] shows undesiccated and CaSO 4 - desiccated XRD data for TXD LSCF precursor solutions fired at various temperatures between 80°C and 800°C. The phase purity was similar to the previous plot (as far as the phases present and their relative percentage). Figure 5.1 3 : XRD Scans for CaSO 4 - Desiccated TXD LSCF Fired between 80°C and 800°C . La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (PDF #00 - 048 - 0124) [117] . 98 Figure 5.14 [ 117 ] shows undesiccated and CaCl 2 - desiccated XRD data for TXD LSCF pre cursor solutions fired at various temperatures between 80°C and 800°C and shows similar phase purity trends to those in Figure 5.13. Figure 5.1 4 : XRD Scans for CaCl 2 - Desiccated TXD LSCF Fired between 80°C and 800°C . La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (PDF #00 - 0 48 - 0124) [117] . 99 Figure 5.15 [ 117 ] shows undesiccated and dry air - desiccated XRD data for CAD LSCF precursor solutions fired at various temperatures between 80°C and 800°C . The phase purity of the undesiccated powder compared to the desiccated powder was shown to change with dry air desiccation at 500°C and below. In all cases, phase pure LSCF was obtained at 600°C and above. Figure 5.15 : XRD Scans for Dry Air - Desiccate d CAD LSCF Fired between 80°C and 800°C . La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (PDF #00 - 048 - 0124) [117]. 100 Figure 5.1 6 [ 117 ] shows undesiccated and CaSO 4 - desiccated XRD data for CAD LSCF precursor solutions fired at various temperatu res between 80°C and 800°C. The phase purity of the undesiccated powder compared to the desiccated powder was shown to be slight ly different at 400°C and 500°C, but in all cases phase pure LSCF was obtained at 600°C and above. Figure 5.16 : XRD Scans for CaSO 4 - Desiccated CAD LSCF Fired between 80°C and 800°C . La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (PDF #00 - 048 - 0124) [117]. 101 Figure 5.1 7 [ 117 ] shows undesiccated and CaCl 2 - desiccated X RD data for CAD LSCF precursor solutions fired at various temperatures between 80°C and 800°C. The phase purity of the undesiccated powder compared to the desiccated powder was different at processing temperatures below 600°C. According to Figures 5.1 5 through 5.1 7 CAD LSCF XRD phase purity was only aff ected at processing temperatures below 600°C, while retaining constant phase purity at temperatures 600°C and above. The reduction in impurity phases can possibly be attributed to reducing uncontrolled precipitation via cation - surfactant solution complexin g. Figure 5.17 : XRD Scans for CaCl 2 - Desiccated CAD LSCF Fired between 80°C and 800°C . La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (PDF #00 - 048 - 0124) [117] . 102 Overall, Figures 5.1 2 through 5.1 4 indicate that the TXD LSCF XRD phase purity was only affected at fabrication temperatures below 700°C, and retained similar phase purity with desiccation at temperatures 700°C and above. The TXD LSCF XRD scans shown in Figure 5.12 - 5.14 possibly have a higher percentage of impurities compared to CAD LSCF XRD scans shown in Figure 5.15 - 5.17 because Triton X - 100 only interacts with dissolved cations over a portion of its polymer chain , while Citric A cid can interact with the dissolved cations along its entire polymer chain which leads to an increased cation interaction to form phase pure LSCF. In addition, desiccation may lead to smaller nano - particle sizes by collapsing the Triton X - 100 and Citric A cid polymer network chain to produce smaller decomposed solution volumes. These smaller solution volumes would be expected to reduce th e diffusion distances the cations need to travel and hence increase phase purity. While CAD and TXD LSCF XRD scans showed desiccation did not have an impact on the LSCF secondary phase purity at 700°C and above, desiccation did influence the secondary phase purity at processing temperatures below 700°C. This difference in phase purity at lower processing temperatures, when using desiccation, may influence low temperature impurity phase content and hence LSCF coarsening rates. 5.3.3 Impurity Impacts on LSCF Nano - Particle Coarsening Behavior The impact impurity phases had on infiltrated MIEC particle size coarsening is the next area investigated in this chapter. Particle size reduction has been suggested to result in increased cell performance, which may be the result of coarsening caused by different oxide impurity phases shown in Figures 5.12 - 5.17. 103 Figure 5.18 shows the raw data used to calculate the XRD Williamson - Hall particle sizes for TXD LSCF precursor solutions. The 22 0 2 - Theta, ~40 0 2 - Theta and ~47 0 2 - Theta peak positions were used to determine particle sizes. The linearity of the - Hall method could accurately be used to extract LSCF size and strain values from this data. Figure 5.19 shows the Williamson - Hall calc ulated strain % from Figure 5.18 for the TXD LSCF nano - powders combined with different impurity oxides. Strain was calculated directly from the slope of each data set in Figure 5. 18 . Desiccated TXD LSCF strain showed a reduction fr om 600°C to 700°C and then stayed relatively constant from 700°C to 800°C. When impurit y phases were combined with LSCF , the resulting LSCF na no - particle strain decrease d from 6 00°C to 800°C in almost a linear fashion . The strain data at 600°C did overlap when taking into account the error bars, which could suggest that strain for all cases was similar at 600°C and desiccation strength did not impact strain at that temperature. The only except ion is phase pure LSCF , which decreased its strain from 600°C to 700°C and then remained relative ly constant at 800°C , possibly due to its smaller particle size . The add ition of iron oxide also reduced the LSCF nano - particle strain at 800° C but showed reduced particle size as well, which needs future F igure 5.18 : Williamson - Hall Raw Data Plots for Coarsened Undesiccated TXD LSCF Nano - Particle Sizes Produced at 600°C, 700°C and 800°C . Williamson - Hall data was obtain for LSCF nano - particles: a) without oxide additives, b) with La oxide added, c) with Sr oxide added , d) with Co oxide added, and e) with Fe oxide added. 104 analysis. These impurity effects were only conducted on TXD LSCF and these trends may also occur with PND and CAD LSCF, but is not know at this time. Figure 5.19 : Strain Plots for Coarsened Undesiccated TXD LSCF Nano - Particle Sizes Produced at 600°C, 700°C and 800°C . Strain data was obtain for LSCF nano - particles: a) without oxide additives, b) with La oxide added, c) with Sr oxide added, d) with Co oxide added, and e) with Fe oxide added. F igure 5 .20 shows the average particles of TXD LSCF made from 1.50 molar precursor solution s that were first fired at 700°C for 1 hr, combined with various impurity oxides , and finally coarsened at 600°C, 700°C or 800°C for 1hr. The data shows that : 1) TXD LSCF nano - particle size coarsens with increasing processing temperature for all impurity oxides, and 2) TXD LSCF nano - parti cle size coarsening rates of were much larger with the addition of lanthanum oxide, cobalt oxide, or strontium oxide then that of ph ase pure TXD LSCF oxide nano - particles. The effect of impurity phases on TXD LSCF nano - particle coarsening rates support the trends observed in the Figures 5.15 - 5.17 XRD data and the Figur e 5.4 nano - particle size data. A possible explanation why desiccate d CAD LSCF nano - particles are 105 Figure 5.20 : Coarsened Undesiccated TXD LSCF Average Nano - Particle Sizes Produced at 600°C, 700°C and 800°C . 600°C (blue, left), 700°C (green, middle) and 800°C (orange, right), each for 1 hour, with different oxide impurities. Particle size data collected using the Williamson - Hall method. larger than TXD LSCF nano - particles could be due to the fact that dif ferent impurity phases (La 2 O 3 , Co 3 O 4 , etc.) that form ed with different solution additives, in the process producing different LSCF particle sizes . A second explanation for particle size reduction, when using precursor solution additives is EISA, which was discussed in Section 2.6.1. Both TXD and CAD LSCF nano - particles were reduced using desiccation, while PND LSCF nano - particles were not affected. After being evaporated (i.e. desiccated) both the Triton X - 100 and Citric Acid solution additives could be forming into templates which produce, when thermally decomposed, reduced MIEC nano - particle sizes. 5.3 . 4 Desicc ant and Solution Additive Impacts on Performance and Stability Figure 5.21 shows 500 hr open - circuit R P data for undesiccated PND LSCF , and de siccated TXD and CAD LSCF symmetrical cells all taken at an operating temperature of 540°C. The degradation rates for CAD and TXD LSCF are similar but PND LSCF 106 had a significantly higher degradation rate . - fir st 100 hrs and after that period the cells developed a lower, more constant degradation rate . SOFC durability and degradation stuides in literature have also observed this - [ 143 - 145 ] . The degradation rates between 100 - 500 hrs were 9.8% /khr for PND LSCF, 1.7% /khr fo r TXD LSCF and 3.3% /khr for CAD LSCF. These degradation rates can also be related back to the phase purity. Both the CAD a nd TXD LSCF had either phase pure or nearly phase pure oxide powders and had the lower degradation rates, while the PND LSCF had much lower overall phase purity and had the largest degradation rate. Figure 5.21: LSCF - GDC 500 hour R P hour Plot for NMCCs Produced using Desiccation and Different Solution Additives . Samples are: a) 1.50 molar PND LSCF , b) CaCl 2 0.50 molar CAD LSCF , and 1.50 molar TXD LSCF precursor nitrate solutions fired at 700°C for 1 hour. 107 Figure 5.22 shows SEM images of the CAD and TXD LSCF nano - particle sizes before and after 500 hr at 540°C. Even with 500 hrs at 540°C, t he particle sizes remain ed constant and did not show any signs of coarsening. This can be related back to the low phas e impurity fraction seen in the CAD and TXD LSCF XRD data. Unfortunately, SEM images of an aged PND LSCF cell were not obtained for this thesis due to the long time needed to perform an additional 500 hr PND LSCF test . Particle size coarsening could expl ain the increased R P observed with the undesiccated PND LSCF. Particle sizes of ~100 nm would be needed to obtain the measured 500 hr R P values (calculated using the SIM PLE model calculator), and this size of particles is within the standard deviation for 0 hr PND LSCF nano - particles shown in Figure 5.6. In contrast, t he final 500 hr CAD and TXD LSCF nano - particle sizes required to obtain the observed 500 hr R P values would need to be between 60 - 65 nm, which is much larger than the standard deviation shown in Figure 5.6 and the 500 hr CAD and TXD LSCF particles sizes shown in Figure 5.22. In addition, literature [ 146 ] shows that performance degradation, due to particle size coarsening, would show a different performance degradation trajectory t han the one shown on Figure 5.21. This suggests that some mechanism other than MIEC particle size coarsening was responsible for the TXD and CAD LSCF degradation . 108 Figure 5. 2 2: Desiccated Scanning Electron Micrographs for CAD and TXD LSCF - GDC Symmetr ic Cells Tested for 500 hrs . Samples held at 540°C were imaged using the SEM for 0.5 molar CAD LSCF symmetric cell after a) 0 hrs and c) 500 hrs , and CaCl 2 - Desiccated 1.50 molar TXD LSCF symmetric cells tested at 540°C after b) 0 hrs and d) 500 hrs. Each scale bar is 50 microns in width. Figure 5.23 shows the raw EIS data for Figure 5.21 taken at an operating temperature of 540°C over 500 hrs of aging . Consistent with Figure 5.22, t he data shows that over time the R P increase d for PND, CAD and TXD L SCF NMCCs. However, Figure 5.23 also shows a significant increase in the ohmic offset with time . The increased ohmic offset , possibly caused by iron or cobalt doping the GDC over time, may help explain the performance degradation. H owever additional expe riments beyond the scope of the present thesis are needed to investigate the op erable degradation mechanisms. 109 Figure 5 . 2 3 : Raw Impedance Data Plots of Desiccated LSCF - GDC NMCCs Produced using Different Solution Additives . Samples are: a) 1.50 molar PND LSCF ( ), b) 0.50 molar CAD LSCF TXD LSCF ( ) precursor nitrate solutions fired at 700°C. R P LSCF - for 500 hours. 110 5.4 Summary In summary, desiccation was shown to reduce infiltrated LSCF nano - particle size usi ng different precursor solution additions, such as Triton X - 100 and Citric A cid . In contrast, d esiccation was shown to not have an effect on infiltrated LSCF nano - particle size when a precursor solution was not present , as with the PND case. The magnitud e of nano - particle size reduction was dependent on the precursor solution additive choice , as average TXD LSCF nano - particles were reduced from 48 nm to 22 nm, while the average CAD LSCF nano - particles were reduced from 50 nm to 41 nm. P erformance was also shown to increase when using increasingly stronger desiccation , for both TXD and CAD LSCF symmetric cells. XRD data showed that the use of desiccation lower ed the LSCF phase impurity fraction at temperatures below 700°C . Both Triton X - 100 and Citric Acid may increase the LSCF infiltrate phase purity by reducing the amount of nitrate precursor precipitation via cation - surfactant solution complexing. The improved phase purity of CAD LSCF compared to TXD LSCF may result from the fact that onl y a portion of amphiphlic surfactants (such as TXD) i nteract with cations in the solution while chel ating agents (such as CAD) deprotonate over the ir entire length and hence are typically better at preventing cation segreg ation during processing. Finally, the LSCF oxide nano - particle coarsening rate was shown to increase when impurity oxide phases were present at aging temperatures of 600°C and above. At 540°C LSCF coarsening did not occur, but LS CF - GDC cell performance degraded by some other mechanism. 111 CHAPTER 6: The Impact of Nano - Ceria Pre - Infiltration on La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - Infiltrated Solid Oxide Fuel Cell Cathodes 6.1 Introduction The previous two chapters evaluated the impact of desiccation and surfactant choice on infiltrated nano - particle si ze, NMCC performance, and NMCC performance stability . This chapter explains a second infiltration size variation technique : c eria oxide pre - infiltration . P re - i nfiltration has been demonstrated in the literature to lower infiltrate oxide nano - particle siz e [ 27 , 147 - 150 ] . H owever, these studies have not fully explored or explained the relationship between pre - infiltration and infiltrate particle size control . Therefore, t he o bjective of this chapter was to : 1) characterize the effects ceria oxide pre - infiltra tion have on performance and LSCF nano - particle size, 2) investigate why pre - infiltration lowers nano - particle size , and 3 ) chara cterize the effects solution additives have on NMCC performance and LSCF nano - particle size when using pre - infiltration. 6.2 Ex perimental Methods 6.2.1 Cathode - Electrolyte - Cathode Symmetric Cell Production Cathode - supporting electrolytes were prepared in the same manner as Section 4.2.1 . Next , GDC IC scaffolds were then screen printed on both sides of these dense GDC electrolyte pellets. To achieve this, Rhodia GDC powder was coarsened at 800°C for 4 hours prior to being mixed with a polymer (Herae us ; West Conshohocken, PA) to form a GDC ink with a 34% solids loading. Three layers of GDC ink were scree n printed onto each side of dense GDC pellet s using a patterned 80 mesh stainless steel screen with a circular 0.5 cm 2 open area. Before the next ink layer was applied, each ink layer was 112 allowed to flow across the pellet surface for 5 minutes and then was placed in a bake oven at 120 °C for 5 minutes to extract the electronic vehicle solvent and increase the green strength. After screen printing the GDC ink, the resulting scaffolds were heated to 400°C at 3°C/min , held at 400°C for one hour, heat ed to 600°C at 3°C/min, held at 600° C for one hour, heated to 105 0 o C at 5 o C/min, held at 1050 o C for 3 hours, and then returned to room temperature at a nominal cooling rate of 10 o C/min. Sintered IC scaffold thickness and roughness measurements were then made with a Dektak 3 profilometer (Br uker; Tucson, AZ ). GDC precursor solutions were then infiltrated into the IC scaffolds, gelled, and fired at 700°C for 1 hr to coat the IC scaffold surface with nano - GDC oxide particles. GDC precursor solutions were prepared by dissolving 99.99% pure Ce (NO 3 ) 3 * 6.0 H 2 O and 99.99 % pure Gd (NO 3 ) 3 *4.0 H 2 O (Alfa Aesar; Ward Hill, MA) in distilled water containing 3 wt% of pre - dissolved Triton - X 100 ( weight Triton X - 100/ weight nitrate). (T hese precise nitrate water of hydration contents were determined prior t o nitrate weighing by measuring the mass loss that occurred inside a Q500 thermogravimetric analyzer (TA Instruments; New Castle, DE) with heating under nitrogen up to 850 o C. )) Next, LSCF precursor solutions were infiltrated into the nano - GDC coated micro - GDC scaffold. Specifically, TXD, CAD and PND LSCF precursor solutions were prepared and infiltrated in the s ame manner as Section 5.2.1. This infiltrate - gel - f ire process was rep eated twice for both the nano - GDC nanoparticles and LSCF nano - particles to achieve both the desired nano - GDC loading levels between 0 vol % and 7.4 vol% and LSCF loading levels of 12.0 vol%. Lastly, symmetric cells were prepared for 113 electrical measurements by screen printing bil ayer LSM Au cu rrent collectors on each NMCC i n the same manner described in S ection 5.2.1. Figure 6.1 visually demonstrates the nano - GDC pre - infiltration and LSCF infiltration process es described above. Figure 6.1a shows a bare micro sized GDC scaffold w ith no infiltrated nano - GDC or LSCF oxide nano - particles sintered onto the scaffold microstructure . Figure 6.1b shows how subsequent nano - GDC infiltration produced oxide particles sintered onto the bare micro sized GDC oxide scaffold. Finally, Figure 6.1 c shows that LSCF infiltration produced nano - particles sintered onto both the bare micro sized GDC scaffold and nano - GDC oxide particles . Figure 6.1: Pre - Infiltrated Cathode Nano - Particle Fabrication Diagram . (a) without pre - infiltration of nano - GDC or LSCF, (b) with pre - infiltrated nano - GDC particles and without LSCF and (c) pre - infiltrated nano - GDC particles (small black circles) and LSCF nano - particles (small orange circles). This schematic is not drawn to scale. SEM images are used to illustrate th e diagram shown on the left for each stage in the infiltration process. Scale bars indicate a length of 50 nm. 114 6.2.2 Symmetrical Cell Impedance, X - ray Diffraction, and Scanning Electron Microscopy Measurements N M CC symmetrical cells w ere characterized using EIS, XRD , and SEM in the same manner described in S ection 5.2.2, 5.2.3, and 5.2.4 respectively . XRD Williamson - Hall particle size measurements were conducted in the same manner as Section 5.2.6 except that only two peaks (the ~23 0 and ~40 0 2 - Theta p eaks) were used for particle size characterization. Williamson - Hall measurements were restricted to two LSCF XRD peaks because the addition of GDC (in the form of nano - GDC particles) resulted in peak broadening of the ~47 0 2 - Theta peak, excluding it from LSCF particle size analysis. Strain data was calculated from the slope of the raw data in the same manner as Section 5.2.6. 6.2.3 ThermoGravimetric Analysis Measurements Thermo Gravimetric A nalysis (TGA) was performed by taking MIEC precursor solutions and evaluating their thermal decomposition behavior using a Q500 TGA (TA Instruments; New Castle, DE, USA) with a 0.01 mg mass resolution. Precursor solutions were gelled outside porous GDC scaffolds, placed in a platinum sample pan, and heated at 10°C/min in air up to 830°C with no hold time . For those samples containing nano - GDC, MIEC precursor solutions were infiltrated into loose GDC powders and then analyzed in the T GA. 6.2.4 Nano - Micro - Composite Cathode Performance Modeling NMCC performance was modeled using the SIMPLE model [ 5 , 7 , 8 ] in the same manner as was described in Section 5.2.5. 115 6.3 Results 6.3.1 Pre - Infiltration and Solution Additive Impacts on Infiltrate P article Size Table 3 shows the processing parameters used when fabricating symmetric al, GDC pre - infiltrated cathodes using PND, TXD or CAD LSCF. Table 3 shows that TXD and CAD LSCF oxide nano - particles were reduced in size as the loading level of na no - GDC was increased to 7.4 vol%. Table 3 also shows that PND LS CF o xide nano - particle sizes were not a ffected by the addition of nano - GDC . Both GDC pre - infiltration and desiccation show ed no particle size reduction when using PND LSCF. The 700°C TXD LSCF nano - particle size reduction was the same for desiccation and ceria pre - infiltration , where both methods reduce d the TXD LSCF nano - particles to ~ 22 nm. The 700°C CAD LSCF nano - particle reduction was , however, larger using ceria pre - infiltration than desiccation. Figure 6.2 shows scanning electron microscopy images of cer ia pre - infiltrated symmetric cells with 0.0 vol% and 7.4 vol% nano - GDC infiltrated with 1.50 molar and 0.50 molar PND and TXD LSCF precursor solutions. The particle sizes were not affected by molarity, as was seen in Section 5.3.1 for desiccated cells. Figure 6. 2 : Scanning Electron Micrographs for Pre - Infiltrated TXD LSCF Nano - Particles Produced using Different Solution Molarities . 1.5 molar PND LSCF a) 0.0 vol% nano - GDC and b) 7.4 vol% nano - GDC precursor sol utions, and 0.50 molar TXD LSCF c) 0.0 vol% n ano - GDC and d) 7.4 vol% nano - GDC precursor solutions inside GDC scaffolds at 700 o C. Each scale bar is 50 microns in width. 116 Table 3 : Pre - Infiltrated PND, TXD and CAD LSCF - GDC NMCC Cathodes . Processing parameters also shown. Surfactant GDC Loading (%) Firing Temperature (°C) LSCF Loading Level (%) Number of GDC Infiltrations Number of LSCF In filtrations GDC/ LSCF Nitrate Solution Molarity Cathode Thickness (µm) Scaffold RMS Surface Roughness (µm) Volume of Solution per Infiltration (µL) Cathode Geometric Area (cm 2 ) Cathode Porosity (%) Scaffold Column Width (nm) LSCF& GDC Diameter (nm) Total LSCF Surface Area (cm 2 ) Total GDC Surface Area (cm 2 ) LSCF Surface Area/GDC Surface Area (%) PND 0.0 700 12.0 0 2 1.5/1.5 35.2 3.42 4.10 0.5 41 120 66 192 146 132 PND 0.0 700 12.0 0 2 1.5/1.5/ 36.1 3.85 4.21 0.5 41 120 66 197 150 132 PND 0.0 700 12.0 0 6 1.5/0.5 35.5 4.01 4.14 0.5 41 120 62 206 147 140 PND 0.0 700 12.0 0 6 1.5/0.5 36.7 3.85 4.28 0.5 41 120 62 213 152 140 PND 5.0 700 12.0 2 2 1.5/1.5 34.2 4.10 3.99 0.5 41 120 72 171 142 121 PND 5.0 700 12.0 2 2 1.5/1.5 34.8 4.30 4.06 0.5 41 120 72 174 144 121 PND 6.4 700 12.0 2 2 1.5/1.5 35.5 3.95 4.14 0.5 41 120 77 166 147 113 PND 6.4 700 12.0 2 2 1.5/1.5 34.9 3.81 4.07 0.5 41 120 77 163 145 113 PND 7.4 700 12.0 2 2 1.5/1.5 36.7 3.75 4.28 0.5 41 120 77 172 152 113 PND 7.4 700 12.0 2 2 1.5/1.5 36.1 4.15 4.21 0.5 41 120 77 169 150 113 PND 7.4 700 12.0 2 6 1.5/0.5 36.5 4.05 4.26 0.5 41 120 70 188 151 124 PND 7.4 700 12.0 2 6 1.5/0.5 36.2 3.95 4.22 0.5 41 120 70 186 150 124 CAD 0.0 700 12.0 0 7 1.5/0.5 34.2 3.74 3.42 0.5 41 120 50 247 142 174 CAD 0.0 700 12.0 0 7 1.5/0.5 34.5 3.46 3.46 0.5 41 120 50 250 143 175 CAD 5.0 700 12.0 2 7 1.5/0.5 32.6 4.96 3.26 0.5 41 120 43 273 135 202 CAD 5.0 700 12.0 2 7 1.5/0.5 37.9 5.34 3.78 0.5 41 120 43 317 157 202 CAD 6.4 700 12.0 2 7 1.5/0.5 32.9 4.50 3.29 0.5 41 120 27 439 136 323 CAD 6.4 700 12.0 2 7 1.5/0.5 34.3 4.78 3.43 0.5 41 120 27 458 142 323 CAD 7.4 700 12.0 2 7 1.5/0.5 34.8 3.50 3.48 0.5 41 120 27 465 144 323 CAD 7.4 700 12.0 2 7 1.5/0.5 36.0 3.99 3.60 0.5 41 120 27 481 149 323 TXD 0.0 700 12.0 0 2 1.5/1.5 35.2 4.39 4.10 0.5 41 120 48 264 146 181 TXD 0.0 700 12.0 0 2 1.5/1.5 36.1 4.05 4.22 0.5 41 120 48 271 150 181 TXD 0.0 700 12.0 0 2 1.5/0.5 36.0 3.50 12.60 0.5 41 120 49 265 149 178 TXD 0.0 700 12.0 0 9 1.5/0.5 36.3 3.41 2.83 0.5 41 120 51 257 151 171 TXD 5.0 700 12.0 2 2 1.5/1.5 35.8 4.43 4.17 0.5 41 120 42 307 148 207 TXD 5.0 700 12.0 2 2 1.5/1.5 35.6 4.27 4.17 0.5 41 120 42 307 148 207 TXD 6.4 700 12.0 2 2 1.5/1.5 34.8 5.30 4.06 0.5 41 120 22 507 144 352 TXD 6.4 700 12.0 2 2 1.5/1.5 36.4 5.78 4.24 0.5 41 120 22 595 151 394 TXD 7.4 700 12.0 2 2 1.5/1.5 34.6 5.41 3.86 0.5 41 120 21 568 143 397 TXD 7.4 700 12.0 2 2 1.5/1.5 35.1 6.00 3.92 0.5 41 120 21 577 146 395 TXD 7.4 700 12.0 2 2 1.5/0.5 36.1 4.23 12.62 0.5 41 120 23 566 150 377 TXD 7.4 700 12.0 2 9 1.5/0.5 35.4 4.30 2.75 0.5 41 120 23 554 147 377 117 Figure 6.3 shows scanning electron microscopy images of pre - infiltrated symmetric cells with 0.0 vol% nano - GDC and 7.4 vol% nano - GDC infiltrated with 0.50 molar TXD LSCF precursor solutions using different infiltration volume amounts, as reported in Table 3. Similar to experiments on the desiccated cells in Section 5.3.1, the particle sizes of the ceria pre - infiltrate d cells shown in Figure 6.3 were not impacted by the LSCF precursor solution infiltration volume. Figure 6.3 : Scanning Electron Micrographs for Pre - Infiltrated LSCF Nano - Particles Produced using Different Solution Volumes . 1.50 molar TXD LSCF a) 0.0 vol% nano - GDC 2 infiltrations, b) 7.4 vol% nano - GDC 3 infiltrations, c) 0.0 vol% nano - GDC 9 infiltrations and d) 7.4 vol% nano - GDC 9 infiltrations, precursor solutions inside GDC scaffolds at 700 o C. Each scale bar is 50 microns in width. Figure 6 .4 shows SEM images for PND, CAD and TXD LSCF nano - particles formed with increasing amounts of pre - infiltrated nano - GDC ranging from 0.0 vol% to 7.4 vol%. The SEM images show that both the CAD and TXD LSCF nano - particles decreased their average size once the amount of nano - GDC increased above 5.0 vol%, while PND LSCF nano - particle sizes remained constant with increasing nano - GDC loading levels. The TXD LSCF average nano - particle sizes were reduced from 48 nm to 22 nm, CAD LSCF average nano - particle sizes were reduced from 50 nm to 27 nm, and PND LSCF average nano - particle sizes remained constant at around 70 nm. 118 Figure 6.5 s hows the raw data used to calculate the XRD Williamson - Hall particle sizes for PND, TXD and CAD LSCF precursor solutions fired at 600°C, 700°C and 800°C. Due to peak overlap between the nano - GDC and LSCF on the XRD only two peak locations (instead of the standard 3) were used to determine LSCF particle size. Figure 6. 4 : Scanning Electron Micrographs for Pre - Infiltrated LSCF Nano - Particles Produced using Different Solution Additives . 0.50 molar CAD LSCF pre - infiltrated nano - GDC and LSCF precursor gel decomposition (a - e) and after 1.50 molar TXD LSCF pre - infiltrated nano - GDC and LSCF precursor gel decomposition (f - j) a GDC scaffold. Scale bars indicate a length of 50 nm. 119 Figure 6.5: Williamson - Hall Raw Data Plots Pre - Infiltrated for PND LSCF, TXD LSCF and CAD LSCF Nano - Particles . Williamson - Hall data was obtained only for two LSCF XRD peaks due to peak overlap between the GDC and LSCF. Figure 6.6 shows the Williamson - Hall calculated strain % from Figure 6.5 for the pre - infiltrated LSCF nano powders using PND, TXD and CAD solution additives. Strain was calculated directly from the slope of each data set in Figu re 6.5. . 120 Figure 6. 6 : Strain Plots Pre - Infiltrated for PND LSCF, TXD LSCF and CAD LSCF Nano - Particles . LSCF nano - particle strain was calculated using 0.0 vol% nano - GDC, 5.0 vol% nano - GDC, 6.4 vol% nano - GDC, and 7.4 vol% nano - GDC fired at 600°C, 700°C and 800°C for 1 hr. The strain for PND LSCF nano - particles is very large, although is the smallest of any strains reported on Figure 6.6, and has a similar magnitude for each nano - GDC loading level, similar to Section 5.3.1. The PND LSCF nano - particles w ere again the largest of any precursor solution and have the smallest strain. The TXD LSCF and CAD LSCF nano - pa rticles showed similar trends for the st rain displayed in Section 5.3.1, but unlike the strain reported in Figure 5.5 at 600°C, the strain value s in Figure 6.6 at 600°C are different enough that the error bars do not completely overlap. The only exceptions are between the: 1) 5.0vol% and 6.4vol%, and 2) 0.0 vol% and 7.4 vol% nano - GDC strain values. Overall strain started at a maximum at 600°C, w as reduced at 700°C and then stayed relatively constant at 800°C, compared to the strain calculated at 700°C , which was also observed in Section 5.3.1 . Figure 6.7 shows the raw data used to calculate the nano - GDC (not LSCF) Williamson - Hall particle sizes shown in Figure 6.6 for PND, TXD and CAD LSCF precursor solutions fired at 600°C, 700°C and 800°C. Nano - particle strain was also determined from the slope of each data set, which is shown in the following figure. The 121 error for the ~28 2 - Theta and ~73 2 - Theta values used for determining the strain were much larger than previous error values primarily at 600°C and 700°C because there was a much more peak overlap at those 2 - Theta values, which made JADE have more difficulty di fferentiating between the correct values. These error values decreased at 800°C since the peaks were much more defined at higher temperatures. Without the error bars the slope would appear negative in Figure 6.5 , but with the error included the slope can be interpreted as 0 or positive in value . F igure 6.7: Williamson - Hall Raw Data Plots for Pre - Infiltrated Nano - GDC Particle Sizes in TXD LSCF and CAD LSCF Infiltrated Cells . Williamson - Hall data was collected for the ~28 2 - Theta, ~56 2 - Theta, and the ~7 6 2 - Theta values for nano - GDC. Figure 6.8 shows the XRD Williamson - Hall calculated strain % from Figure 6.7 for the pre - infiltrated nano - GDC nano powders using PND, TXD and CAD solution additives. Strain was calculated directly from the slope of each data set in Figure 6.7 . Since the slope could possibly be 0 or positive, a 0 slope from Figure 6.7 would appear as a horizontal line while a positive slope is similar in appearance to previous Williamson - 122 Hall plot already shown. The strain for the nano - G DC particles was much smaller than the LSCF nano - particles but still significant (reaching a maximum of 0.08%). The strain Figure 6.8: Strain Plots for Pre - Infiltrated Nano - GDC Particles in TXD LSCF and CAD LSCF Infiltrated Cells . Strain data was determined from Figure 6.7 for the ~28 2 - Theta, ~56 2 - Theta, and the ~76 2 - Theta values for nano - GDC. initially starts at a higher values for TXD and CAD cases at 600°C, then decreases at 700°C, and finally increases again at 800°C. This makes sense sinc e the nano - GDC particles were fabricated at 700°C and would be closer to a strain - free state at 700°C. Under a typical LSCF nano - particle atop a GDC nano - particle bilayer situation one would expect the nano - GDC to under significant strain since at 25°C the Thermal Expansion Coefficient (TEC) for LSCF [ 72 ] (~21*10 - 6 K - 1 between 30 - 1000°C) is larger than that of GDC [ 151 ] (~13*10 - 6 K - 1 between 27 - 827°C). Bilayer s train measurements (where calculations were based on a 23nm LSCF particle on top a 23 nm nano - GDC particle ) were also calculated and range d in m agnitude from 0.001% to 0.02% , which is large but not nearly as large as shown in Figure 6.8 . Strain measurements were [ 152 ] (150/GPa at 800°C, 110/GPa at [ 152 ] (~0.3 at 600 - 800°C), LSCF TEC (21.4*10 - 6 K - 1 ) [ 72 ] [ 153 ] (~200/GPa at 600 - 800°C), GDC 123 [ 153 ] (0.33 at 600 - 800°C), and GDC TEC [ 151 ] (~13*10 - 6 K - 1 ). These values considerably are considerably lower than those in Figure 6.8, but additional surface tension described by the Young - LaPlace equation co uld be an addition al source of stress in real - world GDC nano - particles. Figure 6 .9 shows the average oxide nano - particle sizes for PND, TXD and CAD LSCF using various nano - GDC loading levels when fired at 600°C, 700°C or 800°C for 1 hr. The particle sizes and standard deviation values were also calculate d from the SEM images shown in F igure 6 .2. The LSCF particle size s stay ed relatively constant at nano - GDC loading levels below 5.0 vol% , but were drastically reduced at nano - GDC loading levels above 5.0 vol% for TXD and CAD LSCF nano - particles fabricated at 700°C and 800°C. This is in sharp contrast to the gradual nano - particle reduction shown when using desiccation in Section 5 .3.1 . Figure 6.10 shows the average pre - infiltrated nano - GDC and LSCF particle sizes in the PND, TXD and CAD LSCF cells fired at 600°C, 700°C or 800°C for 1 hr. The opaque particle size bars were calculated using the SEM images o f Figure 6.2 . The columns that are slightly transparent in the back were calculated using the Williamson - Hall technique. Note that the Figure 6.10 SEM particle sizes are the average size for both the GDC and LSCF infiltrate particles (since it is impossi ble to distinguish between GDC and LSCF infiltrated in the SEM) , while the Williamson - Hall particle sizes are the average size of just the LSCF infiltrate. The fact that PND LSCF nano - particle size was not impacted by any nano - GDC loading level, but both TXD and CAD LSCF nano - particle size were, indicates that both a nano - GDC loading level above 5.0 vol% and a 124 solution additive are required for LSCF nano - particle sizes to be reduced. Possible reasons for this behavior are discussed at the end of this sect ion. F igure 6.9 : Bar - Graph of Average LSCF Infiltrate Particle Size Produced using Pr e - Infiltration and Different Sol ution Additives . a) 1.50 molar PND LSCF , b) 1.50 molar TXD LSCF and c) 0.50 molar CAD LSCF precursor nitrate solutions at various temperatures. PND: =0.0 vol%, =5.0 vol%, =6.4 vol%, and =5.0 vol%, =6.4 vol%, and =7.4 vol% and TXD: =0.0 vol%, =5.0 vol%, =6.4 vol% and =7.4 vol%. Particle size data collected from SEM images are shown with 100% opacity and are striped pointing to the left, while particle size data collected using the Williamson - Hall method are shown with 50% opacity and are striped pointing to the right. Error bars are +/ - a standard deviation calculated using the SEM - measured particle size distribution. 125 Figu re 6 .10 : Bar - Graph of Average Nano - GDC Infiltrate Particle Size Produced using Pre - Infiltration and Different Solution Ad d itives . a) 1. 50 molar PND, b) 1.50 molar TXD LSCF and c) 0.50 molar CAD LSCF precursor nitrate solutions at various temperatures. PND: =0.0 vol%, =5.0 vol%, =6.4 vol%, and =5.0 vol%, =6.4 vol%, and =7.4 vol% and TXD: =0.0 vol%, =5.0 vol%, =6.4 vol% and =7.4 vol%. Particle size data collected from SEM images are shown with 100% opacity and are striped pointing to the left, while particle size data collected using the Williamson - Hall method are shown with 50% opacity and are striped pointing to the right. Err or bars are +/ - a standard deviation calculated using the SEM - measured particle size distribution. 126 Figure 6 . 11 shows the raw EIS data for the pre - infiltrated CAD and TXD LSCF symmetrical cells tested at 500°C, 600°C and 700°C operating t emperatures. Consistent with the LSCF particle size trends, t he EIS data shows that the cathode resistance decreases with both increasing temperature and nano - GDC loading levels above 5.0 vol% . The raw EIS impedance data for CAD LSCF and TXD LSCF have both shown that pre - infiltration improves performance but the performance increase is larger when using TXD LSCF, which corresponds to the larger particle size reduction shown in Figure 6.2. The TXD LSCF performance using desiccation was also shown to be bet t er than its CAD LSCF counterpart as well, which was previously shown to be the result of reduced infiltrate LSCF nano - particle size. Figure 6.12 shows the R P data for desiccated PND, CAD and TXD LSCF - GDC symmetrical cells determined from the raw impedance data. The R P for CAD and TXD LSCF both decreased with increasing nano - GDC amount, while that for the PND LSCF did not change in response to pre - infiltrated nano - GDC. Since the LSCF particle size did not change either with PND LSCF, (just like desiccation), having the R P not change is 2 operating temperature decreased from ~650°C to ~545°C when using TXD LSCF and from ~700°C to ~600°C when using CAD LSCF. The operating temperature was not affected when using PND LSCF. 127 Figure 6. 11 : Raw Impe dance Data Plots of Pre - Infiltrated LSCF - GDC NMCCs Prodcued using Citric Acid and Triton X - 100 . The outermost data was taken at 500°C, the nested data was taken at 600°C and the double nested data was taken at 700°C. TXD: =0.0 vol%, =5.0 vol%, =6.4 vol% and =5.0 vol%, =6.4 vol%, and =7.4 vol%. All cathodes used 12.0 vol% LSCF. Differences in ohmic offset are due to thickness variations in the electrolyte between cells and not the result of electronic losses f rom GDC resistivity changes. 128 F igure 6.12 : LSCF - GDC R P Arrhenius Plots for NMCCs Produced using Pre - Infiltration and Different Solution Additives . a) 1.50 molar pure nitrate, b) 0.50 molar Citric A cid - con tain and c) 1.50 molar TXD LSCF precursor nitrate solutions at 700°C fpr 1 hour. PND: =0.0 vol% and =7.4 vol% and TXD: =0.0 vol% and =7.4 vol%. All cathodes used 12.0 vol% LSCF. Differences in ohmic offset are due to thickness variations in the electrolyte between cells and not the result of electronic losses from GDC resistivity changes. 129 SIMPLE model R P predictions were also made using n ano - particle sizes from Figure 6.2. The SIMPLE model predicts similar performance changes compa red to the experimental results. This suggests that LSCF nano - partic le size reductions alone were responsible for the observed cathode performance gains. Simi lar behavior was observed for the desiccated LSCF of Section 5.3.1 , which makes these results un surprising. Figure 6. 13 [ 7 ] shows the ohmic resistivity of the GDC IC scaffold for the different operating temperatures. All the ohmic resistivity data for PND, CAD and TXD LSCF symmetrical cells match the resistivity of pure GDC, which indicate that differences in performance shown in Figure 6. 12 were not caused by electronic losses in the scaffold. Figure 6. 13 : Arrhenius Ohmic Resistivity Plots for Pre - Infiltrated LSCF - GDC NMCCs Tested in Air using Different Solution Additives . PND: =0.0 vol%, =5.0 vol%, =6.4 vol%, and =5.0 vol%, =6.4 vol%, and =7.4 vol% and TXD: =0.0 vol%, =5.0 vol%, =6.4 vol% and =7.4 vol%. The inclined solid line is the resistivity of pure GDC from literature [7] . 130 6.3.2 Pre - Infiltration and Solution Additive Impact on Infiltrate Phase Purity Figure 6 .1 4 [ 117 , 118 , 134 - 137 , 141 ] shows ex - situ XRD data for P ND, TXD and CAD - containing LSCF oxide powder fired at 700°C for 1 hr using different loading level s of nano - GDC. Just like the desiccated LSCF results of Section 5 .3.2 , in Figure 6.14 the phase purity of the powder increased as the solution changed from P ND (left column) to CAD (right column) without the use of nano - GDC . After nano - GDC was added the phase purity could not be determined accurately due to peak overlap between the GDC and LSCF phase . Figure 6 .1 5 [ 117 , 118 ] shows 0.0 vol% nano - GDC and 7.4 vol% nano - GDC XRD data for LSCF 1.50 molar PND oxide powder fired at various temperatures between 80°C and 800°C. This data suggests that t he addition of 7.4 vol% nano - GDC made the 0.0 vol% LSCF powder , at all fabrication temperatures (except at 80 ° C), phase pure. E xact Figure 6 .14 : XRD Scans for Pre - Infiltr ated LSCF and GDC Nano - Particles with Different Solution Additives . a) 0.50 molar Citric A cid - containing pre - infiltrated nano - GDC precursor gels and b) 1.50 molar Triton X - 100 - containing pre - infiltrated nano - GDC precursor gels (b) outside GDC scaffolds at 700 o C. The pre - infiltrated nano - GDC \ LSCF vol% ratio for all samples is 50:12. 0.9 Gd 0.1 O 1.95 (JCPDS # 01 - 075 - 0161) [118] . 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 (JCPDS # 00 - 048 - 0124) [117] , =Co 3 O 4 (JCPDS # 01 - 074 - 2120) [141], =CoCo 2 O 4 (JCPDS # 01 - 080 - 1537) [134] , =La 2 O 3 (JCPDS # 00 - 040 - 1279) [135] , =La(Co 0.42 Fe 0.58 O 3 ) (JCPDS # 01 - 074 - 9369) [136] , 2 O 3 (JCPDS # 01 - 071 - 4953) [137] . 131 conclusions whether small impurity oxide phases are present are difficult to make due to the strong overlapping XRD signal of the nano - GDC . Figure 6.1 5 : XRD Scans for 7.4 vol% Pre - Infiltrated PND LSCF Fired between 80°C and 800°C . The LSCF vol% was held constant at 12.0 vol % in all cases. Ce 0.9 Gd 0.1 O 1.95 (JCPDS # 01 - 075 - 0161) [118] . La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 (JCPDS # 00 - 048 - 0124) [117] . 132 Figure 6.16 [ 117 , 118 ] shows 0.0 vol % nano - GDC and 7.4 vol% nano - GDC XRD data for LSCF 1.50 molar TXD - containing oxide powder fired at various temperatures between 80°C and 800°C. The phase purity of the LSCF is phase pure for all fabrication temperatures above 200°C when using 7.4 vol% nan o - GDC. Figure 6 .16 : XRD Scans for 7.4 vol% Pre - Infiltrated TXD LSCF Fired between 80°C and 800°C . The LSCF vol% was held constant at 12.0 vol % in all cases. Ce 0.9 Gd 0.1 O 1.95 (JCPDS # 01 - 075 - 0161) [118] . La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 (JCPDS # 00 - 048 - 0124) [117] . 133 Figure 6.1 7 [ 117 , 118 ] shows 0.0 vol% nano - GDC and 7.4 vol% nano - GDC XRD data for LSCF 0.50 molar CAD - containing oxide powder fired at various temperatures between 80°C and 800°C. The phase pu rity of the LSCF is phase pure fo r all fabrication temperatures when using 7.4 vol% nano - GDC. This is in contrast to the phase purity of CaCl 2 - desiccation CAD LSCF which still had phase impurities that formed at lower fabrication temperatures. Figure 6 .1 7 : XRD Scans for 7.4 vol% Pre - Infiltrated CAD LSCF Fired between 80°C and 800°C . The LSCF vol% was held constant at 12.0 vol % in all cases. Ce 0.9 Gd 0.1 O 1.95 (JCPDS # 01 - 075 - 0161) [118] . La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 (JCPDS # 00 - 048 - 0124) [117] . 134 The 7.4 vol% nano - GDC TXD LSCF had greater phase purity for all temperatures than the CaCl 2 - desiccated TXD LSCF data presented in Section 5.3.2. Phase purity, using desiccation, for temperatures below 600°C still had a noticeable quantity of impurity peaks while the data in Figure 6.17 showed no im purity phases at any fabrication temperature other than 80°C. Overall, the LSCF oxide phase when using TXD and CAD became the dominant phase around 60 0°C. However, t he addition of nano - GDC was hypothesized to act as a coarsening reducing agent. It is hyp othesized that at 600°C the addition of nano - GDC (at any loading level) did not reduce nano - particle size because the LSCF oxide particles have not had time to coarsen. In contrast a t 700°C and 80 0°C coarsening was assumed, based on the results in Figure 6.16 - 6.17 , which is why the TXD LSCF and CAD LSCF nano - particles show reduced particle size at nano - GDC loading levels greater than 5.0 vol%. It is further hypothesized that t he PND LSCF nano - particles were not influenced by the nano - GDC because there was no surfactant or chelating a gent to limit cation separation, preventing PND LSCF oxide particles from becoming the dominant phase until 700°C , by which time their pre - existing impurity phases had already coarsened. 6.3.3 Pre - Infiltration and Solution Additive Impact on Precursor Solution Decomposition Behavior Figure 6 .1 8 shows TGA datasets for PND, CAD and TXD LSCF - GDC cells infiltrated with different loading levels of nano - GDC. Since the nano - GDC does not allow for impurity phases to be observed in the XRD plots , TGA was used to observe the effects of nano - GDC. The pre - infiltrated PND LSCF and TXD LSCF TGA data indicates that the LSCF precursor thermal decomposition peaks (at low nano - GDC vol%) shift ed to lower temperatures and finally disappear ed a t higher nano - GDC loading levels. 135 Figure 6.1 8 : Pre - Infiltrated TGA Plot using Different Solution Additives between 25°C and 850°C . Samples are for : a) PND LSCF precur sor solutions, b) CAD LSCF precursor solutions, and c) TXD LSCF precursor solutions. TGA data was collected up to 850°C. 136 The CAD LSCF TGA data also shows that a large multistage decomposition event (between 200°C and 400°C) gradually shifted to lower temperatures at higher nano - GDC loading levels. Previous lit erature studies have shown that ceria catalyzes nitrate decomposition [ 27 ] , which maybe the reason why nano - GDC reduces the LSCF formation temperature. 6.3.4 Pre - Infiltration and Solution Additive Impact s on Performance and Stability Figure 6 .19 shows 500 hr open - circuit R P data for 0.0 vol% nano - GDC PND and 7.4 vol% nano - GDC TXD and CAD LSCF - GDC symmetrical cells. As in Chapter 5 , the initial p erformance of TXD LSCF cells were the highest, followed by the CAD LSCF cells, and finally the PND LSCF cells . Figure 6.19 shows that t he PND and CAD LSCF - GDC degradation rate appears to be similar in magnitude and much larger than the TXD LSCF - - for all PND, TXD and CAD are much higher that the later, more constant degradation rates, similar to Chapter 5 desiccation. The 100 - 500 hr break - in degradation rates for PND are 9.8 % / khr while TXD has degradation rates of 6.3 % / khr and CAD degradation rates are 12.0 % / khr. For unknown reasons the degradation break - in rates are clearly larger than desiccation. However, t he CAD and TXD LSCF nano - particle sizes required to obtain 500 hr R P values would need to be between 60 - 65 nm, which is much larger than the standa rd deviation shown in Figure 5.6, and the particle sizes shown in Figure 6.20. This suggests that like desiccation, MIEC nano - particle size coarsening was not an active degradation mechanism at 540°C. However, m ore work beyond the scope of this thesis is needed to understand the active degradation mechanisms in these pre - infiltrated cathodes. 137 Figure 6 .19 : LSCF - GDC 500 hour R P hour Plot for NMCCs Produced using Pre - Infiltration and Different Solution Additives . a) 1.50 molar pure nitrate , an d b) 0.50 molar Citric A cid - containing and 1.50 molar Triton X - containing precursor nitrate solutions at 700°C for 1 hour. Figure 6.20 shows SEM images of the CAD and TXD LSCF nano - particle sizes before and after 500 hr at 540°C. The particle sizes remain constant and do not show any signs of coarsening. This can be related back to the low phase impurity fraction seen in the CAD and TXD LSCF XRD data. As mentioned previously, SEM images of the PND LSCF cell were not obtained for this thesis due to time constraints for taking the PND LSCF 500 hr EIS measurement . Particle size coarsening could explain the increased R P observe d with the undesiccated PND LSCF, since p article sizes of ~100 nm were needed to produce the final 500 hr R P values (calculated using the SIMPLE model ), a 100 nm 138 was well within the standard deviation for PND LSCF nano - particles shown in Figure 6.9 . Just like in Section 5.3.4 literature studies [ 146 ] support our hypothesis that particle size coarsening was not responsible for the cathode performance degradation observed in Figure 6.19. Figure 6 . 20 : Pre - Infiltrated Scanning Electron Micrographs for CAD and TXD LSCF - GDC Symmetric Cells Tested for 500 hrs . Samples held at 540°C were imaged using the SEM for 0.5 molar CAD LSCF symmetric cells: a) 0 hrs and c) 500 hrs , and CaCl 2 - Desiccated 1.50 molar TXD LSCF symmetric cells tested at 540°C after b ) 0 hrs and d) 500 hrs. Each scale bar is 50 microns in width. Figure 6.21 shows the raw EIS data before and after 500 hr open - circuit stability testing. The data shows that similar trends to these in Section 5.3.4, with R P and ohmic offset suggesting that some mechanism other than infiltrate particle size. Further investigation into performance degradation is outside the scope of this thesis but should be investigated in the future. 139 6.4 Summary In summary, ceria pre - infiltration was shown to reduce infiltrated LSCF nano - particle size using different precursor solution additives , such as TXD and CAD. Pre - infiltration was shown to not have an effect on infiltrated LSCF nano - particle size when a precursor solution additive was n ot present as with the PND case. Similar behavior was also observed with desiccation. As with desiccation , the magnitude of the TXD and CAD Figure 6 .21 : Raw Impedance Data Plots of Pre - Infiltrated LSCF - GDC NMCCs Produced using Different Solution Additives . a)1.50 molar Pure Nitrate ( ), b) 0.50 molar Citric A cid - - 100 containing ( ) precursor nitrate solutions at 700°C. R P LSCF - s were taken for 500 hours. 140 LSCF nano - particle size reduction was d ependent on the precursor sol ution addition choice. With ceria pre - infiltration the average TXD LSCF nano - particle size was reduced from 48 nm to 22 nm, while the aver age CAD LSCF nano - particle was reduced from 50 nm to 27 nm. In contract to desiccation , the reduction of nano - particle size with ceria pre - infiltration was not gradual , but suddenly decreased once the loading level of nano - GDC was greater than 5. 0 vol% for both CAD and TXD LSCF . As with desiccation, ceria pre - infiltration included p erformanc e changes were found to be solely the result of infiltrate nano - particle size reductions. There were d ifferences observed between the trends using pre - infiltration and desiccation . First, the performance degradation rate over 500 hrs was significantly high er using pre - infiltration for both CAD and T XD when compared to desiccation . Second, the XRD data for pre - infiltration was inconclusive due to peak overlap between the GDC and LSCF phases which resulted in TGA data being used to observe changes during dec omposition. The TGA data suggests that when the loading level of nano - GDC is increased , the LSCF precursor solution decomposition events are shifted to lower temperatures or completely removed . The decrease in low temperature impurity content this produc es may limit particle size growth during manufacturing, resulting in reduced average LSCF particle size. 141 CHAPTER 7: The Impact of Precursor Solution Desiccation and Nano - Ceria Pre - Infiltration on La 0.6 Sr 0.4 Co 1 - x Fe x O 3 - 7.1 Introduction The previous chapters have shown that the average infiltrated TXD and CAD LSCF nano - p article size can be reduced and NMCC performance can be increased using desiccation and ceria oxide pre - infiltration. In addition, the previous chapters have shown that the performance and average nano - particle size of LSCF - GDC NMCCs both have a stro ng dependence on surfactant type . The objective of the work in this chapter was to evaluate whether the precursor gel desiccation and ceria oxide pre - infiltration fabrication techniques could also be used to control the average nano - particle size of other CAD MIEC compositions in the La 0.6 Sr 0.4 Co y Fe 1 - y O 3 - (y= 0 to 1) (LSF - LSC) system . 7.2 Experimental Methods 7.2.1 Cathode - Electrolyte - Cathode Symmetric Cell Produc tion Symmetrical cathode - supporting electrolytes were fabricated in the sam e manner that was described in C hapters 4 - 6. Specifically, p orous well - necked GDC IC scaffolds were produced on both sides of these electrolyte pellets. To achieve this, some of the aforementioned Rhodia GDC powder was coarsened at 800°C for 4 hours and then mixed with V - 737 electronic vehicle (Herae us ; West Consh ohocken, PA) to form a GDC ink with a 34% solids loading. Three layers of GDC ink were then screen printed onto each side of each dense GDC electrolyte pellet using a patterned 80 mesh stainless steel screen with a circular 0.5 cm 2 open area. Prior to th e next ink layer being applied, each ink layer 142 was allowed to flow across the pellet surface for 5 minutes and then was placed in a bake oven at 120°C for 5 minutes to extract the electronic vehicle solvent and increase the green strength. After screen pr inting, the samples were heated to 400°C at 3°C/min , held at 400°C for one hour, heated to 600°C at 3°C/min, held at 600°C for one hour, heated to either 1100 o C (desiccation) or 1050°C (pre - infiltration) at 5°C/min , held at either 1100 o C for 3 hours (desi ccation) or at 1050°C for 3 hours (pre - infiltration) , and then cooled to room temperature at a nominal cooling rate of 10 o C/min. Sintered IC scaffold thickness and roughness measurements were made with a Dektak 3 profilometer (Bruker; Tucson, AZ ). Three different fabrication techniques 1) a standard technique commonly used in the literature [ 5 , 7 ] that was neither desiccated or pre - infiltrated, 2) the desiccation technique described in Chapter 4, or 3) the ceria pre - infiltration technique described in Chapter 6 were used in conjunction with various precursor solution compositions in the (La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - ) (x =0 to 1) system. The standard technique had precursor solutions that were pipetted into the porous GDC cathode scaffolds, allo wed to soak into the scaffold fo r 5 minutes, and gelled at 80 o C fo r 10 min before being fired at 7 00°C for 1 hr to form nano - sized MIEC oxide particles on the scaffold surface. The desiccation technique is described in Section 4.2.1 , and the pre - infiltrat ion technique is described in Section 6 .2.1 . For all three techniques the MIEC precursor solutions were fabricated in t he same manner as described in Sections 4 .2.1 and 6 .2.1 . Desiccated cells used CaCl 2 as the desiccant and pre - infiltrated cells used a 7.4 vol% loading level of nano - GDC precursor solution. The c ell s made by all three techniques used a 12.0 vol% MIEC loading level for each composition. Lastly, symmetric cells were prepared for electrical 143 measurements by screen printing bilayer LSM Au cu rrent collectors onto each NMCC in exactly the same manner as previous chapters. 7.2.2 Symmetrical Cell Impedance, X - ray Diffraction, and Scanning Electron Microscopy Measurements N M CC symmetrical cells were characterized using EIS, XRD, and SEM in the sa me manner as described in Sect ions 4.2.2, 4.2.3, and 4.2.4 respectively . 7.2.3 Nano - Micro - Composite Cathode Performance Modeling The NMCC performance was modeled using the SIMPLE model [ 5 , 7 , 8 ] in the manner described in Section 4.2.1. 7.3 Results 7.3.1 Pre - Infiltration, Desiccation and Infiltrated Composition Impact s on Infilt rate Particle Size and Performance Figure 7 .1 shows 600°C raw electrochemical impedance spectroscopy data for symmetrical cells infiltrated with CAD L a 0.6 S r 0.4 F eO 3 (LSF) , CAD L a 0.6 S r 0.4 Co 0.2 Fe 0.8 O 3 (LSFC) , CAD L a 0.6 S r 0.4 C o 0.5 F e 0.5 O 3 (LSCF 55 ) , CAD L a 0.6 S r 0.4 C o 0.8 F e 0.2 O 3 (LSCF), or CAD L a 0.6 S r 0.4 C oO 3 (LSC) using the : a) standard, b) desiccation , or c) pre - infiltration processing techniques. The highest R P value was observed with CAD LSF using the standard technique while the lowest R P value was observed us ing CAD LSC with the pre - infiltration technique. Note the systematic behavior of R P vs. composition for all three processing methods. Another observation was that desiccation and pre - infiltration both improved performance to a greater extend with cobalti te compositions than ferrite compositions. 144 Figure 7.1: Figure 7.1: Raw Impedance Data Plots of Desiccated and Pre - Infiltrated La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - ) using Citric Acid . Standard: =La 0.6 Sr 0.4 FeO 3 - (LSF), =La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC), =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF), =La 0.6 Sr 0.4 CoO 3 - (LSC) ;Desiccated: =La 0.6 Sr 0.4 FeO 3 - (LSF), =La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC), =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF), =La 0.6 Sr 0.4 CoO 3 - (LSC);Nano - GDC: =La 0.6 Sr 0.4 FeO 3 - (LSF), = La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC), =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF), =La 0.6 Sr 0.4 CoO 3 - (LSC). All data was taken at 600°C. Figure 7.2 shows Arrhenius R P data for CAD LSF, CAD LSFC, CAD LSCF55, CAD LSCF, and CAD LSC infiltrated cells tested at operating temperatures between 400°C and 700°C for the standard, desiccation and pre - infiltration processing techniques. 145 Figure 7 .2 : Desiccated or Pre - Infiltrated La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - P Arrhenius Plots for NMCCs Produced using Citric Acid . The dashed lines are SIMPLE model predictions, where the top most (for each composition) is for the standard case, the middle most is for the desiccate d case and the lowest is for the pre - infiltrated nano - GDC case. Standard: =La 0.6 Sr 0.4 FeO 3 - (LSF), =La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC), =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF), =La 0.6 Sr 0.4 CoO 3 - (LSC) ;Desiccated: =La 0.6 Sr 0.4 FeO 3 - (LSF), =La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC), =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF), =La 0.6 Sr 0.4 CoO 3 - (LSC);Nano - GDC: =La 0.6 Sr 0.4 FeO 3 - (LSF), = La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC), =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF), =La 0.6 Sr 0.4 CoO 3 - (LSC). All data was taken at 600°C. 146 In Figure 7.2 b oth desiccation and pre - infiltration are shown to reduce R P for all material compositions but pre - infiltration reduced R P to a greater extent than desiccation. The operating temperature for the target R P 2 was reduced to a minimum temperature of ~55 0°C with CAD LSC using pre - infiltration. ( 2 behavior reported in Section 6.3.1 was for pre - infiltrated TXD LSCF, not CAD LSCF ) . As was observed with Figure 7.1 , the cobaltite MIEC materials performed bett er than the ferrite materials. Based on the R s values from the literature shown in Figure 2.13 , on average the cobaltite materials have a lower R s (LSC, LSCF, SSC, and BSCF) value than the ferrite compounds (LSF and LSFC), which would tend to lead to improved cathode performance. Figure 7.3 shows SEM images for : 1) standard (a - e) , 2) desiccated (f - j), and 3) pre - infiltrated (k - o) cells fabricated at 700°C. For each processing technique, t he average infiltrate particle size decrease d slightly as the La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - cobalt content increased ; with standard average particle sizes ranging from ~55 nm for CAD LSF to ~50 nm for C AD LSC ; desiccated average particle sizes ranging from ~46 nm for LSF to ~39 nm for LSC and pre - infiltrated average particle sizes ranging from ~31 nm for LSF to ~24 nm for LSC. Figure 7.4 shows the average particle sizes for the various MIEC compositions for the a) standard, b) desiccation and c) pre - infiltration techniques. The nano - particle sizes shown in this figure are taken directly from Figure 7.3. As was stated in Figure 7.3, the nano - particle size decreased within each processing technique , with ferrite materials having larger average nano - particle sizes than cobaltit e materials . Overall, desiccation and pre - infiltration both reduced the average nano - particle size with desiccation lowering 147 particle sizes from 55 nm to 42 nm and pre - infiltration l owering parti cle sizes from 55 nm to 26 nm. CAD LSC average nano - particle sizes were found to be the smallest of any Figure 7.3: Desiccated or Pre - Infiltrated La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - - Particle Sizes Produced using Citric Acid . S tandard (a - e) desiccated (f - j) and pre - infiltrated nano - GDC oxide particles (k - o). tested composition. This, coupled with the low R s of LSC om Figure 2.13 which correlates to the CAD LSC cells having the lowest R P values. The changes in particle size, when input into the SIMPLE model predicted R P results that were consistent with observed experimental results. Williamson - Hall particle size measurements were not 148 conducted due to the large number of samples analyzed , and th e reliability of the SEM - determined particle sizes in Chapters 5 and 6 . Figure 7 .4 : La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - Nano - Particle Sizes Determined from Scanning Electron Microscopy Images . a) Standard, b) desiccated and c) pre - infiltrated nano - GDC nano - particles. Both desiccation and pre - infiltrated nano - GDC cells lower LSCF nano - particle sizes when compared to the standard case. Error bars indicate the standard deviation of LSCF nano - part icle sizes calculated from the SEM images. Standard: =La 0.6 Sr 0.4 FeO 3 - (LSF), =La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC), =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF), =La 0.6 Sr 0.4 CoO 3 - (LSC) ;Desiccated: =La 0.6 Sr 0.4 FeO 3 - (LSF), =La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC), =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF), =La 0.6 Sr 0.4 CoO 3 - (LSC);Nano - GDC: =La 0.6 Sr 0.4 FeO 3 - (LSF), = La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC), =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF), =La 0.6 Sr 0.4 CoO 3 - (LSC). All data was taken at 700°C. Figure 7 .5 shows the ohmic resistivity of the GDC IC scaffold for the various CAD La 0.6 Sr 0.4 Co x Fe 1 - x O 3 in Figure 7.1 . All the ohmic resistivity data for tested sample symmetrical cells match ed the resistivity of pure GDC, 149 which indicated that the differences in performance shown in F igure 7. 2 , were not caused by electronic losses in the IC scaffold. Figure 7 .5: Arrhe nius Ohmic Resistivity Plots for Desiccated or Pre - Infiltrated La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - . Standard: =La 0.6 Sr 0.4 FeO 3 - (LSF), =La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC), =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF), =La 0.6 Sr 0.4 CoO 3 - (LSC) ;Desiccated: =La 0.6 Sr 0.4 FeO 3 - (LSF), =La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC), =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF), =La 0.6 Sr 0.4 CoO 3 - (LSC);Nano - GDC: =La 0.6 Sr 0.4 FeO 3 - (LSF), = La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC), =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF), =La 0.6 Sr 0.4 CoO 3 - (LSC). All data was taken at 600°C.The inclined solid line is the resistivity of pure GDC from literature [ 7 ] . 7.3.2 Pre - Infiltration, Desiccation and Infiltrated Composition Impact s on Infiltrate Phase Purity Figure 7 .6 [ 117 , 154 - 156 ] shows standard and desiccated XRD data for the different powder compositions fired at 700°C for 1 hour . The powder produced using the standard method is shown in the upper datase ts and the powder produced using the desiccation method is shown in the lower datasets for each material composition. The results in this figure show that desiccation did not affect CAD La 0.6 Sr 0.4 Co x Fe 1 - x O 3 phase purity. 150 Figure 7 .6 : XRD Scans fo r CaCl 2 - Desiccated La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - . Standard XRD spectra are stacked above the desiccated XRD spectra for each peak. Standard: =La 0.6 Sr 0.4 FeO 3 - (LSF)(JCPDS # 01 - 072 - 8133) [154] , =La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC) [155] , =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF) (JCPDS # 00 - 048 - 0124) [117] , =La 0.6 Sr 0.4 CoO 3 - (LSC)(JCPDS # 01 - 089 - 5718) [156] ;Desiccated: =La 0.6 Sr 0.4 FeO 3 - (LSF) [154] , =La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - (LSFC) [155] , =La 0.6 Sr 0.4 Co 0.5 Fe 0.5 O 3 - (LSCF55), =La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - (LSCF) (JCPDS # 00 - 048 - 0124) [117] , =La 0.6 Sr 0.4 CoO 3 - (LSC) [156] . Figure 7.7 [ 118 ] shows standard and pre - infiltrated XRD data for the different powder CAD La 0.6 Sr 0.4 Co x Fe 1 - x O 3 fired at 700°C for 1 hour. As observed previ ously for CAD LSCF in Figure 6.10 the 7.4 vol% nano - GDC pre - infiltration powder had such a large signal that small impurity oxide phases were not detectable, but large impurity oxide phases were still not observed . 151 Figure 7 .7 : XRD Scans for 7.4 vol% Pre - Infiltrated La 0.6 Sr 0.4 Co x Fe 1 - x O 3 - Acid . 0.1 Ce 0.9 O 1.95 (JCPDS 01 - 075 - 0161) [118] . 7.4 Summary In summary, t his chapter was intended to evaluate the impact desiccation and pre - infiltration have on infiltrated nano - particle size and NMCC performance when using MIEC compositions other than LSCF . For all tested CAD La 0.6 Sr 0.4 Co x Fe 1 - x O 3 compositions b oth nano - particle size and R P values were shown to be reduced using CaCl 2 - desiccation or 7.4 vol% nano - GDC pre - infiltration. The nano - particle size reduction was observed to be the reason for increased performance with all material com positions, just as was seen in C hapters 4 - 6. Desiccation reduced infiltrated nano - 152 particle size from 55 - 50 nm (LSF to LSC) to 45 - 39 nm (LSF to LSC), while pre - infiltration reduced nano - particle size to 30 - 24 nm (LSF to LSC). Clearly , pre - infiltration reduced infiltrated nano - particle size for all material compositions to a greater extent than desiccation. Material composition p hase purity was not impacted by either desiccation or pre - infiltration, at least at a fabrication temperature of 700°C. Chapters 4 - 7 have examined both the desiccation and pre - infiltration techniques and investigated their effects on infiltrated cathode na no - particle size and performance. Based on the results observed, a preliminary recommendation for the desiccation technique would be given as the best technique to use for commercial SOFC production. This recommendation is based on: 1) desiccation having a substantially lower degradation rate at 540°C for both CAD and TXD LSCF cells compared to pre - infiltrated CAD and TXD LSCF cells, 2) desiccation lowering the operating temperature by 75°C (from 650°C to 575°C) compared to the standard technique, and 3) the manufacturing costs for desiccation are low since CaCl 2 is inexpensive and desiccation has a low cost capital expenditure. The main downside of using this technique is an increased manufacturing time due to the desiccation process. However, t he de sic cation manufacturing time could possibly be exped ited through the use of alternative desicca nts possessing faster desiccation kinetics . 153 CHAPTER 8 : Determination of Infiltrated Mixed Ionic and Electronic Conducting Nano - Particle Oxygen Surface Exchange Material Properties through Finite Element Modeling of 3D Reconstructed Microstructures 8 .1 Introduction Section 2.8 highlighted that a lack of knowledge regarding the intrinsic oxygen surface exchange resistance , R s , of infiltrate MIEC materials was one o f the greatest obstacles limiting the development and improve ment of low temperature NMCCs and SOFCs. P as t literature studies have measured thin film R s values for LSF [ 52 , 54 ] , LSCF [ 51 , 54 ] , LSFC [ 54 , 95 ] , LSC [ 54 , 97 ] , SSC [ 54 , 98 ] , and BSCF [ 49 , 50 ] composition s . However, no infiltrate R s values have ever been measured . Further , as stated in Section 1 .3 , literature studies have shown that large MIEC R s discrepancies exist between current literature reports of these compositions [ 10 ] . Infiltrate R s measurements are needed because the di fferent stress and surface states of infiltrated MIEC particles could lead to R s values significantly different than those obtain from thin or bulk films. Further , these R s values would allow the SOFC community to determine the optimal MIEC infiltrate com positions and correctly model and optimize NMCC performance. Therefore, t he objective of the work in this chapter was to: 1) determine R s values for different infiltrated MIEC materials using FIB - SEM FEM 3D reconstructions, and 2) evaluate the R s /IC ratios under which the SR lim i t and SIMPLE models break down. Both these objectives were achieved through the use of finite element modeling. 154 8 .2 Experimental Methods 8 .2.1 Cathode - Electrolyte - Cathode - Symmetric Cell Production Cathode - supporting symmetrical cells were pre pared in the same manner as S ection 7.2. 8 .2.2 Electrochemical Impedance Spectroscopy Measurements EIS characterization was performed in the same manner as S ection 4 .2 . 8.2.3 Nano - Micro - Composite Cathode Finite Element Modeling of 3D Reconstr uction s As discussed in S ection 3.3 .3 , a set of 2D serial sections were obtained to create a 3D reconstruction using a computer program called MIMICS (Materialise Inc, Leuven Belgium) . This 3D reconstruction was then volume meshed using a second computer program called 3 - Matic (Materialise Inc, Leuven Belgium) , and finally a FEM computer program called COMSOL (COMSOL Inc, Palo Alto California) was used for perfo rmance calculations. Section 3.3 outlines the multiple steps involved in : 1 ) creating the FIB - SEM 3D reconstruction, 2) creating th e 3D microstructure volume mesh, and 3) modeling the 3D volume mesh in COMSOL. A step - by - step guide on this procedure is provided in Appendix 2. Here, t he volume mesh of both the cathode and electrolyte was created with a large enough number of tetrahedrons (calculation boxes used in FEM) to provide an accurate determination of the performance. The volume mesh size used in the performance calculations was slightly over 1 million tetrahedrons (using the dimensions p rovided in Figure 3.11). Performance calculations using a volume mesh size of over 5 million tetrahedrons also conducted using the 3D reco nstruction shown in Figure 3.11 ) 155 yielded performance values that were only different by ~4%, indicating the adequacy of the FEM mesh size. Next, e le ctrochemical potential lines were generated from the volume mesh using COMSOL . As discussed in Section 3.3.6 and Appendix 2, t he potential lines were calculated by applying a scaled R s value, and a 1V potential to the entir e cathode surface mesh , and a 0V potential to the bottom of the electrolyte microstructure. Cathode R P calculations were then conducted using the calculated electrochemical potential lines to mathematically determine the current crossing the bottom of the electrolyte and the use of Equation 16. R P values from 3D reconstructions m were compared and R P predictions from both reconstructions were very close to each other indicating that edge effects had no significant contribution to the final result s . Finally, R P values were calculated using C OMSOL for temperatures between 6 00°C - 700°C . These R P values were then compared against experimentally determined NMCC EIS R P values from: 1) Chapter 4 for LSCF, 2) Chapter 7 for LSF, LSFC, an d LSC, and 3) Nicholas et al. [ 5 ] for SSC . To calculate R s , the FEM inputted R s values were adjusted u ntil the FEM determined and experimentally determined R P values agreed. 8 .3 Results 8.3.1 A Comparison of Finite Element Modeling Mixed Ionic Electronic Conductin g Materials Intrinsic Oxygen Surface Exchange Material Properties Figure 8 . 1 [ 51 , 54 , 95 - 102 ] shows the reported literature R s , k chem , and k o values from Section 2.7.3 overlaid with R s values determined from FEM calculations of FIB - SEM 3D reconstructed microstructures. In Figure 8.1, t he activation energies of infiltrated CAD LSF (1.01 eV), CAD LSFC (1.18 eV), CAD LSCF (1.04 eV), CAD LSC 156 (1 .08 eV) and CAD SSC (1.3 8 eV) are lower than or equal to the reported literature data . In contrast , however, the 600 - 700°C magnitude of these FEM calculated values agree surprisingly well with the literature values (i.e. within 1 order of magnitude ) . F igure 8.2 shows all the FEM - determined R s, k chem and k o values for LSF, LSFC, LSCF, LSC and SSC on a single plot. The R s values for the different cathode MIEC materials consistently stay within 2 orders of magnitude of each other, but the k chem and k 0 val ues have a much larger variation. This larger variation in k chem and k o is caused from a difference in the C o shown in Figure 2.12. F rom a purely R s perspective, the best infiltrated MIEC material to use between 600 - 700°C is SSC because it has the lowest Figure 8.2 R s value of any MIEC composition and also has k chem and k o values that are nearly 2 orders of magnitude higher than the other MIE C compositions. These high - temperature results are not surprising since they match closely with what Baumann et al. [ 54 ] has shown in literature for thin film MIEC materials. However when these results are projected to lower operating temperatures, then LSF becomes a better MIEC materia l choice since it has the lowest activation energy, and thus (by extension) would have the lowest R s value. This is surprising since Baumann et al. [ 54 ] shows thin film LSF to have much higher activation energy (1.8 eV) compared to SSC (1.3 eV) over the 600 - 700°C temperature range. This disagreement clearly shows that the properties of MIEC infiltrate can be different than MIEC thin films, and justifies the need for FIB - SEM FEM calculations such as those presented here. 157 Figure 8 .1 : R s , k chem and k o values Reported in Literature and Calculated using FIB - SEM FEM 3D Reconstructions for the Cathode MIEC Materials LSF, LSFC, LSCF, LSC and SSC . Open symbols are for data from literature studies and closed symbols are for calculated values from FIB - SEM FEM 3D reconstructions [51], [54], [85 - 92]. 158 Figure 8 . 2 : R s , k chem and k o Values Calculated using the FIB - SEM FEM 3D Reconstruction for Cathode MIEC Materials LSF, LSFC, LSCF, LSC and SSC . The legend in the top - right corner of each subfigure shows the MIEC material for each symbol. Despite the importance of a low R s value, det ermination of the optimal MIEC composition for infiltration should also account for the ability to reduce the size of infiltrate particles made of that composition. In light of this, LSCF ( R ct is 5.06 at 600°C, 2.53 at 650°C, and 1.52 at 700°C) and LSC ( R ct is 5.46 at 600°C, 2.29 at 650°C, and 1.29 at 700°C) are probably the best MIEC compositions to use at the moment because unlike SSC and LSF both these compositions exhibit large nano - particle size reductions using both desiccation and pre - infiltration s hown in Chapters 5 - 7 . Of course, a low MIEC R s value alone is not enough to guarantee a low MIEC infiltrate cathode R P . The cathode microstructure also has an effect. 8.3.2 Identifying the Materials Property Combinations Causing the Surface Resistance Li mit and the Simple Infiltration Microstructure Polarization Loss Estimation Model to Breakdown Throughout this thesis three NMCC R P prediction models have been used : 1) the FIB - SEM FEM, 2) SR Model , and 3) the SIMPLE M odel. As mentioned , in Section 2.5.1 and 8.2.4 the FIB - SEM FEM and SIMPLE model calculations take into account the scaled R s values, scaffold bulk IC and ohmic scaffold effects. T he SR model only take s into account the scaled R s values. The FIB - SEM FEM calculations also ar e different 159 from the SIMPLE model in that they use a real scaffold microstructure , while the SIMPL E model uses an idealized scaffold microstructure. These differences between the three models could lead to vastly different R P values depending on the ratio of the Equation 1 surface scaled R s value ( R ct ) to the low scaffold IC low R ct would produce a high electrochemical potential gradient in the IC scaffold that would raise the impact of the IC scaffold geometry . Therefore, the objectiv e of this section was to determine the full range of the R ct values that could accurately be modeled by the SR model and the SIMPLE mode l . Figure 8.3 shows comparison s between the three models , where the calculated R P value is plotted against the scaled R s to scaffold IC ratio. The first thing to note is that at high ratio values (where the scaled R s dominates over the scaffold IC) the three models have very similar R P values. In contrast, l arge differences between t he three models start to occur when the scaled oxygen surface exchange resistance is only 100 times larger than the scaffold IC. At this point the scaffold IC and microstructure both start to become a significant component in the R P calculations. Since t he SR model does not account for performance losses in the IC scaffold it continues to predict R P values using the same activation energy at low and high ratio values. This results in the SR model predicting R P values that are completely incorrect (i.e. s ignificantly from the FIB - SEM FEM values) at low R ct values. In contrast to the SR model , the SIMPLE model increases its activation energy at ratio values lower than 100 because it does account for IC scaffold contributions. Unfortunately, the SIMPLE m odel R P calculations are still significantly different than the actual (i.e. FIB - SEM FEM values) at these low R ct In this regime, the FIB - 160 SEM FEM calculated R P values are higher than the SIMPLE model R P values possibly due to increased tortuosity in the real cathode microstructure (something the SIMPLE model assumes is minimal). For comparison experimentally measured LSCF - GDC R P and R ct 600 - 700°C and are also shown on Figure 8.3 . It is imp ortant to note that these experimental values are at or slightly above the 100 ratio mark , and this explains why all 3 models do a reasonably good job describing their per formance ( SIMPLE model studies in the literature [ 5 , 7 , 61 ] have been able to accurately predict LSCF - GDC NMCC performance to within 33%) . However, if improvemen ts are made in lowering the R s value of future MIEC materials , the n R P values should be predicted using FIB - SEM FEM 3D reconstructions instead of the SR and SIMPLE model s . Figure 8. 3 : Calculated R P Values from the FIB - SEM FEM 3D Microstructure, SIMPLE Model and Surface Resistance Model Determined for Different R ct / Scaffold GDC Conductivity Ratio Values . = FIB - SEM FEM R P values, = SIMPLE model R P values, = SR R P values, = FIB - SEM FEM dete rmined R P value at 600°C, = FIB - SEM FEM determined R P value at 650°C and FIB - SEM FEM determined R P value at 700°C . 161 8 .4 Summary For the first time NMCC R s values were determined for various MIEC infiltrate compositions using FIB - SEM FEM 3D reconstruction s . The calculated 600 - 700°C R s magnitudes were consistent with past literature studies, but had lower or identical activation energies, suggesting that R s value discrepancies between infiltrated R s values and literature values may ex ist at lower temperatures. SIMPLE model and SR limit calculations were compared to the 3D re construction values to show the MIEC/ IC scaffold ratios required for these models to accurately measure R P . Knowing accurate R s values for NMCCs will be useful fo r future NMCC microstructure optimization studies [ 31 ] aimed at improving SOFC performance . 162 CHAPTER 9 : Dissertation Conclusions In summary , t wo new method s to systematically control infiltrate MIEC nano - particle size , and one new method to accurately det ermine infiltrate d MIEC R s values were developed in this thesis, and t he following are conclusions that from this thesis were demonstrated . As shown in Chapters 4 - 7, precursor solution d esiccat ion and ceria pre - infiltration were shown to reduced infiltrated La 0.6 Sr 0.4 Fe x Co 1 - x O 3 - MIEC na no - particle size when precursor solution additives such as Triton X - 100 and Citric Acid were present (these methods were not effective in reducing the size of La 0.6 Sr 0.4 Fe x Co 1 - x O 3 - MIEC when solution additives wer e not present) . Both d esiccation and ceria pre - infiltration lowered the TXD LSCF average nano - particle sizes from 48 to 22 nm . In contrast, d esiccation lowered the CAD LSCF average nano - particle size from 50 to 41 nm, while ceria pre - infiltration lowered the CAD LSCF average nano - particle size from 50 to 25 nm. Across the entire LSF - LSC solid solution, pre - infiltration was shown to be more effective than desiccation at reducing the initial cathode polarization resistance. These nano - particle size reduct ions allowed the cathode operating temperature ( the temperature at which the cathode reached a target R P 2 ) to be reduced , with the largest operating temperature reduction of ~650°C to ~545°C occurring with pre - infiltrated TXD LSCF. Figure 9.1 shows that the infiltrated cathode s produced in this thesis were some of the best ever produced [ 1 , 3 - 5 , 27 , 28 , 71 , 76 , 77 , 157 - 160 ] . 163 Figure 9.1: Infiltrated Cell Comparison from Different Infiltration Groups . LSCF=La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 - , LSC=La 0.6 Sr 0.4 CoO 3 - , GDC=Gd 0.1 Ce 0.9 O 1.95 , SSC=Sm 0.5 Sr 0.5 CoO 3 - , SDC= Sm 0.2 Ce 0.8 O 1.9 , LSGM = (La,Sr)(Ga,Mg)O 3 , YSB= Y 0.25 Bi 0.75 O 1.5 , LSM=La 0.6 Sr 0.4 MnO 3 - , LSF=La 0.6 Sr 0.4 FeO 3 - , YSZ = (Y 2 O 3 ) 0.08 (ZrO 2 ) 0.92 , GBCO = GdBaCo 2 O , LSFC=La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 - , LNO= La 2 NiO 4 . Pre - infiltrated TXD LSCF, =Pre - infiltrated CAD LSC, [5], [27], [71], [76], .[3], [1], [28], [157], [158], [77], [159], [4], [160]. In addition to superior initial performance, the desired TXD and CAD LSCF cathodes produced here were also to have favorable 100 - 500 hr 540°C degradation rates of 1.7% /khr and 3.3% /kr, respectively , in comparison to the 9.8% /khr 100 - 500 hr 540°C degradation rates observed for PND LSCF - GDC cells . (P re - infiltrated TXD and CAD 164 cells were found to have 100 - 500 hr 540°C degradation rates of 6.3% /kh r and 12.0 % /khr, res pectively ) . In addition, nano - particle s ize was found to remain constant over 500 hrs at 540°C for all tested cathodes, indicating that nano - structured electrodes can survive intermediate SOFC operating temperatures. Even though pre - infiltrated cells exh ibited superior initial performance and took less time to manufacture (6 hours instead of 7 days), desiccated cells are recommended for use in commercial SOFC s because of their superior long - term stability (assuming that faster - acting desiccants can be fou nd and that additional tests performance under bias yield similar results to those open - circuit measurements performed here). As shown in Chapter 8, i nfiltrated MIEC R s measurements for phase pure (i.e. CAD) LSF, LSFC, L SCF, LSC, and SSC were also determin ed for the first time. These FEM determined R s values were found to have activation energies equal to or less than reported bulk and thin film R s values between 600 - 700°C . This suggests that the low temperature performance of MIEC nano - particles may be co nsiderably better than the same material in bulk or thin film form. From a purely R s perspective , S SC was shown to be the best - performing MIEC infiltrate at SOFC operating temperatures of 600 - 700 °C while LSF was shown to be the best MIEC infiltrate choice at lower operating temperatures. However, when both R s and an ability to control the infiltrate particle size were considered, LSC and LSCF were shown to be the best NMCC infiltrate choice. 165 APPENDI CES 166 Appendix 1: Simple Infiltrated Microstructure Polarization Loss Estimation (SIMPLE) Model Derivation The following derivation is fo r the SIMPLE model outlined in S ection 2.5.2. This derivation was put together by Dr. Jason D. Nicholas and has been available for electronic download fr om https://www.egr.msu.edu/nicholasgroup/simple.php since October of 2012. Motivation As discussed in Nicholas, J. D., L. Wang, et al. (2012). " Use of the Simple Infiltrated Microstructure P olarization Loss Estimation (SIMPLE) Model to Describe the Performance of Nano - Composite Solid Oxide Fuel Cell Cathodes ." Phys. Chem. Chem Phys ., the main usefulness of the SIMPLE model is that it provides a quick means of determining the lowest possible R P for Mixed Ionic Electronic Conductor (MIEC) - Ionic Conductor (IC) composite cathodes when : 1) bulk oxygen transport only happens through the ionic conducting scaffold, and 2) oxygen ion incorporation only occurs at the MIEC particles. These requirements a re commonly met f or nano - composite cathodes (NCC s) made via the infiltration of a range of MIEC materials (LSCF, SSC, other MIEC cobaltite oxygen surface exchange catalysts, etc.) into a range of ionic conducting scaffolds (ceria, zirconia, lanthanum stron tium gallium manganite, etc.). The SIMPLE model can also be 167 SIMPLE Model Derivation - composite particles atop an ionic - conductor scaffold. It is b ased to a large degree on the Tanner Fung Virkar (TFV) model which describes the performance of SOFC cathodes made of electronic conductors atop an ionic conductor scaffold. Therefore, much of the derivation below is an expansion and extension of that foun d in Tanner, C. W., K. - Z. Fung, et al. (1997). "The effect of porous composite electrode structure on solid oxide fuel cell performance." J. Electrochem. Soc. 144 (1): 21 - 30. Modeled Geometry Like the TFV model, the SIMPLE model assumes that the ionic cond ucting scaffold geometry can be represented as a series of columns Figure A1.1 : An Idealized Representation of a Symmetric SOFC Cathode Cell . (Not Drawn to Scale). Image courtesy of Lin Wang. Assuming a columnar cathode microstructure enables an analytical solution for NCC polarization resistance, i.e. the SIMPLE model, to be obtained Note that for th e idealized geometry in Figure A1.1 where is the volume fraction of pores in the cathode, is one - half the total scaffold column thick ness, and is the repeat unit thickness 168 As discussed in Nicholas, J. D., L. Wang, et al. (2012). " Use of the Simple Infiltrated Microstructure Polarization Loss Estimation (SIMPLE) Model to Describe the Performance of Nano - Composite Solid Oxide Fuel Cell Cathodes ." Phys. Chem. Chem Phys ., the major source of NCC electrical losses is the difficulty of incorporating oxygen into the MIEC, and electrical losses associated with the difficulty of transporting oxygen through the bulk of the ionic conducting scaf fold are comparatively small. (This can be observed by comparing the results of the SIMPLE model, which accounts for surface resistance and bulk transport losses, and a Surface Resistance (SR) Model, which only accounts for surface resistance). Because of matter that much. The close agreement between the measured and SIMPLE model predicted polarization resistances also speak to this fact. Model Assumptions: Discussed f urther in Nicholas, J. D., L. Wan g, et al. (2012). " Use of the Simple Infiltrated Microstructure Polarization Loss Estimation (SIMPLE) Model to Describe the Performance of Nano - Composite Solid Oxide Fuel Cell Cathodes ." Phys. Chem. Chem Phys . 1. The IC scaffold geometry can be idealized as a series of high aspect - ratio IC columns. P oorly - necked rea l - world scaffold particles will invalidate this assumption. However, the close agreement between SIMPLE model predictions and real - - necked scaffold particles , exchange resistances dominate the overall polarization resistance, and the microstructural details of the IC scaffold are relatively unimportant. 169 Corollaries a. The equipotential l ines within the IC scaffold are horizontal. Finite element modeling results on IC columns in Tanner, C. W., K. - Z. Fung, et al. (1997). "The effect of porous composite electrode structure on solid oxide fuel cell performance." J. Electrochem. Soc. 144(1): 2 1 - 30 showed that for large aspect ratio columns such as those in an approximated real - world cathode structure (a 250 nm thick column, 20 microns high), the equipotential lines are horizontal through the majority of the column. 2. The only significant resistan ces are MIEC oxygen surface exchange resistance and IC scaffold bulk oxygen transport. Corollaries a. Bulk oxygen transport only happens through the ionic conducting scaffold (a good assumption since the bulk oxygen conductivities of ionic conductors are many orders larger than that of MIEC materials) and oxygen ion incorporation only occurs at the MIEC pa rticles (a good b. Losses associated with electron transport through the MIEC can be ignored. Another way to say this is that the electronic conductivity of th e infiltrated MIEC is high enough, and the MIEC infiltrate is interconnected enough, that electronic transport losses can be ignored. Analysis on the performance of real cathodes indicated that this is the case for heavily infiltrated cathodes tested near open circuit. 170 i. This causes the potential on the surface of the MIEC, , to be the same on all MIECs surfaces throughout the cathode. c. Losses associated with oxygen transport through the MIEC can be ignored. Another way to state this assumption is to say t hat the MIEC nanoparticles are below their characteristic thickness. This is a good assumption since the characteristic thickness values for most MIEC materials are at least several microns from 400 - 700C, and the infiltrate particle thickness is typically below 25 nm. d. Losses associated with gas - phase diffusion can be ignored. Concentration polarization resulting from poor gas - phase diffusion is not expected to contribute to the near open circuit potential R P predicted by the SIMPLE model since under these c onditions the oxygen fluxes are small and the NCC pores are large. e. Losses associated with the IC - MIEC interface can be ignored. In the few cases where they have been measured, MIEC - ionic conductor interfacial resistances have been anywhere from 10 - 100 time s less than the ionic surface exchange resistance at sub - 700 o C temperatures. 3. The resistance caused by the movement of charged species is independent of driving force. Another way of saying this is that it is the electrochemical potential, , which drives charged species, and since , where is the chemical potential, zF is the charge on the charged species, and is the electrical potential, a polarization resistance calculated by assuming an electrical driving 171 force (as occurs in a symmetric cell impedance test) will be the same as that results from a concentration gradient (as occurs in a working fuel cell). 4. Oxy gen transport through the IC scaffold behaves ohmically. This is another way of saying that the ionic conductor behaves as a dilute, ideal solution which has a constant composition and structure throughout. This proof starts the relationship between the to tal current, i , and the current density, j . is the current density of species i is the Kroger - Vink oxidation state of species i is the atomic flux of species i is the charge carried by species i Applying the Nernst - Planck Equation in the absence of convection, where is the mobility of species i, is the concentration of species i, and is the gradient of the electro - chemical potential of species i Applying the definition of electrical conductivity, Applying the definition of the electrochemical potential, where is the chemical potential, zF is the charge on the charged species, and is the electrical potential. For ideal materials, Where is the mole fraction of species i. Since by definition where is the molar density, Since Taking the gradient of all terms Since is the reference chemical potential, Figure A1.2: Oxygen Transport Proof . Shows IC Scaffold Behave s Ohmically . 172 For materials of constant composition and structure that behave ideally, the Nernst - Planck equation says that This is a form of and Applying the rule Assuming that the composition of the IC is constant, . This is a intermediate temperatures because they are heavily doped materials where the oxygen vacancy concentration is extrinsically controlled. Also assuming that the structure of the IC is constant so that the molar density is con stant through the ionic conductor, , and Plugging this in, If species i is the dominant charge carrying species we can drop the subscripts Multiplying both sides by the cross - sectional area, A The current, , is related to the current density via The conductivity is defined as where is the resistivity The 1D gradient is where l is the path length Figure 173 This is the forms we know and love. The voltage is defined as The resistance of a dense block of material is defined as where R is the resistance Rearranging 5. Oxygen transport across the MIEC surface behaves ohmically. This is a good assumption because the more - complicated Butler - Volmer Equation predicts ohmic behavior at low overpotentials. Because of this assumption, the SIMPLE mod el can only be used to predict open - circui t NCC polarization resistances. 174 Figure A1.1 shows the modeled repeat unit. The resistance across the repeat unit, , is given as: where is the resistance of the cathode and is the resistance of the electrolyte Multiplying everything by the geometric cathode area, A, to turn these resistances into Area Specific Resistances Defining the cathode polarization resistance, , as the area specific resistance of the cathode Rearranging the expression Assuming a rectangular geometric cathode area such that and the columns shown in Figure 1 extend into and out of the page by a distance , so that Figure 1 shows a portion of the cathode in the r direction. Since the electrolyte is dense block o f material, the definition of resistance can be applied where is the electrolyte resistivity, l is the path length, and A is the cross - sectional area. Since, , and as shown in Figure 1, for the electrolyte, and , this equation becomes: where is the oxygen ion conductivity of the electrolyte Plugging this in, Solving for As stated in in assumptions 3 and 4, the SIMPLE model assumes that current flow across Applying the definition Figure A1.3: Electrode Polarization Resistance Proof . Depicts the General Equation. 175 the repeat unit behaves ohmically so of current density and A= r t , The voltage across the repeat unit given as If we have a symmetrical cathode setup, such as that depicted in Figure 1, and we apply to one cathode and to the other, because of symmetry, =0 The local current density at one spot along the bottom of the repeat unit in Figure 1 (where the origin is in the lower left corner of the repeat unit and the y direction is up Law acc ording to assumption 3 Since there is only a gradient in the y direction this reduces to: Since, the total current density coming out of the bottom of the repeat unit, which is equal to , is given as Plugging in Figure A1.3 (cont ) 176 Equation A1.1 : Plugging in this expression for Turns this into Plugging this in Rearranging This expression shows that the polarization resistance can be determined as long as the potential distribution at the center of the electrolyte is known. The equation is independent of geometry of the cathode. Because of that, it can be used as the basis for Finite Element Modeling studies on crazily complicated cathode geometries aimed at calculating the polarization resistance of those cathodes. It can also be solved analytically for idealized geometries. Regardless of the cathode geometry, the potent ial distribution within the cathode is solved at steady state Within any material, Where is the current density, is the charge density, and t is time This is a form of the charge conservation equation At steady state, Since the electrolyte is assumed to be ohmic, Figure A1.3 177 Equation A1.2 : Equation. Since the structure and composition of the electrolyte are assumed to be constant as per assumption 3, from the definition of conductivity, is position independent Figure A1.3 178 Figure A1.4 : Repeat Unit with N umbered Interfaces Across with Current F lows . Due to symmetry, no current flows across the unlabeled leftmost and rightmost interfaces. By assuming the columnar geometry to the right we can use the boundary conditions below to solve for, first, the potential distribution within the scaffold, and second, for the polariz ation resistance, . 1 st Boundary Condition: At steady state, the current within a volume element of scaffold must sum to zero. (this is the same 1. This can be applied locally to any volume of material wit hin the scaffold, or 2. This can be applied across the entire repeat unit scaffold so that the current across the top 3 facets must equal that across the bottom (i.e 4 th ) facet 2 nd Boundary Condition: The current across the surface of Facet 1 is equal to that immediately inside the scaffold Figure A1.5: Solving for . Idealized Geometry Boundary Conditions are used. 179 From assumption 4, we assume ohmic surface reactions so, The voltage across the surface is the potential difference on the surface, , minus that on the other side of the surface, i.e. just inside the scaffold. Put mathematically, What is , the resistance across the surface Facet 1? W ell if we were talking about the resistance of a bulk material we would turn to the definition of resistance, But what would the path length, , be across a surface? What would be the resistivity either of t hese terms we multiply both sides by A, and refer the grouped terms as the area specific resistance (ASR), denoted for a surface, we measure area specific resistances of surfaces. Based on the definition of ASR, where is the area specific surface resistance of the scaffold surface. From Figure A1.2 , the cross - sectional area of Facet 1 is given by . From the Figure A1.1 model geometry where is volume Figure A1.6 : Determining . Determined for Figure A1.4. 180 Plugging in this expression for , fraction of pores in the cathode, is one - half the total scaffold column thickness, and is the repeat unit thickness. Rearr anging this, Figure A1.6 (cont ) 181 Unlike the case for and , the potential along the left side of Facet 2 (i.e. inside the scaffold) changes along the facet. Therefore we must divide the cathode up into infinitesimally small pieces thick, analyze the currents in each of those pieces, and sum the effects together to get the entire current across Facet 2. cannot be evaluated until the potential distribution within the scaffold, is determined. We will have to come back to later. A small section of the scaffold column: Since we are assume ohmic surface reactions, Assuming horizontal potential lines, From the definition of ASR, where is the area specific surface resistance of the scaffold surface. Figure A1.7: Determining . Determined for Figure A1.4. 182 Just as with Facet 1, from assumption 4, we assume ohmic surface reactions so, The voltage across the surface is the potential difference on the surface, , minus that on the other side of the surface, i.e. just inside the scaffold. Put mathematically, Based on the definition of ASR, Plugging in this expression for , From Figure A1.2 , the cross - sectional area of Facet 1 is given by From the Figure A1.1 model geometry Multiplying both sides by r. Figure A1.8: Determining . Determined for Figure A1.4. 183 Note that this is not a surface. This facet is located in the middle of the electrolyte. From the definition of current density, From Figure A1.2 , the cross - sectional area of Facet 4 is given by . Since the scaffold behaves ohmically, Because of the assumption of horizontal potential lines, from the electrolyte looks like a dense block of material with a constant gradient in potential across it. Therefore evaluated at facet 4 is the same as evaluated between and Tanner, C. W., K. - Z. Fung, et al. (1997). " The effect of porous composite electrode structure on solid oxide fuel cell performance." J. Electrochem. Soc. 144 (1): 21 - 30 do finite difference modeling to show that the assumption of constant potential gradient in the electrolyte is a good one when h<