<%BANNER%>

A Multi-length Scale Approach To Correlating Solid Oxide Fuel Cell Porous Cathode Microstructure To Electrochemical Perf...

Permanent Link: http://ufdc.ufl.edu/UFE0024987/00001

Material Information

Title: A Multi-length Scale Approach To Correlating Solid Oxide Fuel Cell Porous Cathode Microstructure To Electrochemical Performance
Physical Description: 1 online resource (186 p.)
Language: english
Creator: Gostovic, Danijel
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: 3, 3d, 3dap, cathode, cell, connectivity, eis, electrochemical, electrochemistry, electrolyte, fib, fuel, impedance, leap, mazo, nano, orientation, porous, reconstruction, sem, sofc, solid, spectroscopy, tem, three, tomography, topological, tortuosity
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Solid oxide fuel cells (SOFCs) are a fuel and application flexible technology which may help the United States on its path to energy independence. The commercialization of intermediate-temperature SOFCs (IT-SOFCs) is limited by cathode polarization resistance. In an effort to understand the effect that sintering and doping processing conditions have on cathodic polarization a multiple length scale characterization approach has been developed to study SOFCs. The technique utilizes a focused ion beam / scanning electron microscope (FIB/SEM), transmission electron microscope (TEM) and a local electron atom probe (LEAP). The electrochemically active region of a LSCF cathode was for the first time reconstructed in three dimensions using a FIB/SEM. Various microstructural properties were measured including overall porosity, closed porosity, graded porosity, surface area, tortuosity, and pore size. Electrochemical Impedance Spectroscopy (EIS) data was correlated to microstructure. Microstructure and chemical segregation of composite cathodes were also analyzed using the FIB/SEM technique. The cathode/electrolyte interface was characterized with TEM-Energy Dispersive Spectroscopy (TEM-EDS) where calcium doping was found to have a significant effect ton cathode microstructure. Higher calcium sample had a coarser composite cathode microstructure. The cathode/electrolyte composite cathode interface was quantified via TEM-EDS line and point scans. The most significant difference seen was the coarser sample having a higher calcium content in both electrolyte and cathode phases by 1 and 1.4% respectively. The active regions of composite cathodes were further analyzed and for the first time the complex 3-D network topological connectivity has been measured. Such novel connectivity quantification allows for an advanced understanding of the transport processes in composite materials. An atomic resolution, LEAP microscope has been used for the first time to characterize SOFC materials. Two separately processed Siemens Energy Incorporated composite cathodes were analyzed. The scandium doped zirconia (SSZ) / calcium doped lanthanum manganate (LCM) buried interface was micro-machined into an atom probe tips and analyzed with the LEAP. Magnesium was found to segregate beyond the 10 nm wide interface another 10+ nanometers. Interfacial voids were found to be hydrogen enriched. Two-dimensional concentration profiles offer a glimpse at atomic segregation near electrochemically active triple phase boundaries (TPBs).
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Danijel Gostovic.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Jones, Kevin S.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024987:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024987/00001

Material Information

Title: A Multi-length Scale Approach To Correlating Solid Oxide Fuel Cell Porous Cathode Microstructure To Electrochemical Performance
Physical Description: 1 online resource (186 p.)
Language: english
Creator: Gostovic, Danijel
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: 3, 3d, 3dap, cathode, cell, connectivity, eis, electrochemical, electrochemistry, electrolyte, fib, fuel, impedance, leap, mazo, nano, orientation, porous, reconstruction, sem, sofc, solid, spectroscopy, tem, three, tomography, topological, tortuosity
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Solid oxide fuel cells (SOFCs) are a fuel and application flexible technology which may help the United States on its path to energy independence. The commercialization of intermediate-temperature SOFCs (IT-SOFCs) is limited by cathode polarization resistance. In an effort to understand the effect that sintering and doping processing conditions have on cathodic polarization a multiple length scale characterization approach has been developed to study SOFCs. The technique utilizes a focused ion beam / scanning electron microscope (FIB/SEM), transmission electron microscope (TEM) and a local electron atom probe (LEAP). The electrochemically active region of a LSCF cathode was for the first time reconstructed in three dimensions using a FIB/SEM. Various microstructural properties were measured including overall porosity, closed porosity, graded porosity, surface area, tortuosity, and pore size. Electrochemical Impedance Spectroscopy (EIS) data was correlated to microstructure. Microstructure and chemical segregation of composite cathodes were also analyzed using the FIB/SEM technique. The cathode/electrolyte interface was characterized with TEM-Energy Dispersive Spectroscopy (TEM-EDS) where calcium doping was found to have a significant effect ton cathode microstructure. Higher calcium sample had a coarser composite cathode microstructure. The cathode/electrolyte composite cathode interface was quantified via TEM-EDS line and point scans. The most significant difference seen was the coarser sample having a higher calcium content in both electrolyte and cathode phases by 1 and 1.4% respectively. The active regions of composite cathodes were further analyzed and for the first time the complex 3-D network topological connectivity has been measured. Such novel connectivity quantification allows for an advanced understanding of the transport processes in composite materials. An atomic resolution, LEAP microscope has been used for the first time to characterize SOFC materials. Two separately processed Siemens Energy Incorporated composite cathodes were analyzed. The scandium doped zirconia (SSZ) / calcium doped lanthanum manganate (LCM) buried interface was micro-machined into an atom probe tips and analyzed with the LEAP. Magnesium was found to segregate beyond the 10 nm wide interface another 10+ nanometers. Interfacial voids were found to be hydrogen enriched. Two-dimensional concentration profiles offer a glimpse at atomic segregation near electrochemically active triple phase boundaries (TPBs).
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Danijel Gostovic.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Jones, Kevin S.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024987:00001


This item has the following downloads:


Full Text

PAGE 1

1 A MULTI-LENGTH SCALE APPROACH TO CORRELATING SOLID OXIDE FUEL CELL POROUS CATHODE MICROSTRUCTURE TO ELECTROCHEMICAL PERFORMANCE By DANIJEL GOSTOVI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Danijel Gostovi

PAGE 3

3 To my parents Marija na and Dragan Gostovi (Mama i Tata)

PAGE 4

4 ACKNOWLEDGMENTS Being co-advised by Professors Kevin S. Jones and Eric D. Wachsm an has been a truly rewarding experience. Without their unrelenting support this work would not have been possible. I would like to thank my committee for their time and dedication. Further, I would like to thank my group members in both the Wachsm an and Jones groups for their help with the preparation of this dissertation. Special thanks to Lorenz Ho lzer, David Kundinger, Jeremiah Smith, Keith Duncan, Jerry Bourne, Rich Marten s, Jennifer Tucker, Katie OHara, Dong Jo Oh, Eric Macam, Eric Armstrong and Nick Vito. I would also like to thank the University of Florida Alumni Fellowship, the Florida Institute for Sust ainable Energy, the United States Department of Energy and Siemens Energy Incorporated for fu nding and providing us with the samples to complete this study. In addition we woul d like to thank the Major Analytical and Instrumentation Center at the University of Flor ida, and the Central Anal ytical Facility at the University of Alabama for use of their facilities.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES ...........................................................................................................................7LIST OF FIGURES .........................................................................................................................8ABSTRACT ...................................................................................................................... .............13 CHAPTER 1 GENERAL INTRODUCTION .............................................................................................. 152 BACKGROUND ....................................................................................................................182.1Cathode Background .....................................................................................................182.2Mixed Conductor SOFC Cathode .................................................................................192.3Electrolyte Background ................................................................................................. 192.4Research Method ...........................................................................................................212.5Electrochemical Impedance Spectroscopy .................................................................... 222.6Microstructural Analysis ...............................................................................................262.7Chemical Analysis ........................................................................................................273 COMPREHENSIVE QUANTIFICATION OF POROUS LSCF CATHODE MICROSTRUCTURE AND ELECTRO CHEMICAL IMPEDANCE .................................. 333.1Introduction .................................................................................................................. .333.2Experimental ................................................................................................................. 343.3Results and Discussion .................................................................................................. 353.3.1Microstructural Quantification ..........................................................................353.3.2Connectivity Quantification .............................................................................. 383.3.3Electrochemical Impedance Spectroscopy ........................................................433.3.4Microstructure-polari zation Relationships ........................................................ 463.4Conclusion .................................................................................................................... 494 THREE-DIMENSIONALRECONSTRUCTIO N OF COMP OSITE CATHODES .............. 724.1FIB/SEM Microstructure Quantifi cation of Composite Cathode .................................724.1.1Experimental ..................................................................................................... 724.1.2Results and Discussion ...................................................................................... 744.1.3Microstructure Summary ..................................................................................824.2TEM-EDS Characterization of LCM/YSZ Interface in Composite Cathode ............... 834.2.1Experimental ..................................................................................................... 834.2.2Results and Discussion ...................................................................................... 834.2.3TEM Summary ..................................................................................................84

PAGE 6

6 4.3Topological Connectivity ..............................................................................................844.3.1Introduction ....................................................................................................... 844.3.2Results and Discussion ...................................................................................... 864.3.2.1Connectivity quantification ................................................................864.3.2.2Measured average z-orientation .......................................................... 884.4Conclusions ...................................................................................................................905 LOCAL ELECTRON ATOM PROBE RE CONS TRUCTION OF BURRIED CATHODE/ELECTROLYTE SOLID OXI DE FUEL CELL INTERFACES ....................1255.1Introduction ................................................................................................................. 1255.2Experimental Procedure ..............................................................................................1275.3Results and Discussion ................................................................................................ 1325.4Conclusions .................................................................................................................1416 CONCLUSION .................................................................................................................... .1746.1Conclusion .................................................................................................................. 1746.2Future Work ................................................................................................................1766.2.13-D micro texture characterization of the composite cathode using FIB/SEM and EDX/EBSD ..............................................................................1766.2.2LEAP standards ............................................................................................... 1776.2.3LEAP bias study ..............................................................................................177REFERENCES .................................................................................................................... ........180BIOGRAPHICAL SKETCH .......................................................................................................186

PAGE 7

7 LIST OF TABLES Table page 3-1 Microstructure quantification ............................................................................................. 50 3-2 Connectivity quantification ................................................................................................50 4-1 Particle size ............................................................................................................. ...........92 4-2 Surface area .............................................................................................................. ..........92 4-3 Two phase surface area ......................................................................................................93 4-4 Tortuosity ...........................................................................................................................93 4-5 TEM-EDS point scan quantification ..................................................................................94 4-6 Connectivity quantification ................................................................................................94

PAGE 8

8 LIST OF FIGURES Figure page 1-1 Schematic Diagram of Solid Oxide Fuel Cell .................................................................... 17 1-2 Practical system volumes. ................................................................................................. .17 2-1 Perovskite crystal structure ................................................................................................29 2-2 Common strategies for SOFC cathodes .............................................................................29 2-3 FIB/SEM ................................................................................................................... .........30 2-4 FIB/SEM phase contrast ....................................................................................................30 2-5 Various TLD detector settings ...........................................................................................31 2-6 Schematic depicting transformation of 2-D im ages into a 3-D reconstruction and graded porosity...................................................................................................................32 3-1 Multiple oxygen reduction/conduction pathways in MIEC cathode. ................................ 51 3-2 Epoxy impregnated samples ..............................................................................................51 3-3 Six symmetric cathode samples mounted in cross section. ............................................... 52 3-4 Six reconstructions of various sintered cathode/electro lyte in terfaces. ............................. 52 3-5 Tortuosity, porosity and the porosity to tortuosity ratio plotted as a function of sintering temperature. ........................................................................................................ 53 3-6 Cathode volume normalized surface area per unit volume and trip le phase boundary length per unit area plotted vers us sintering temperature. .................................................54 3-7 Open pore and cathode particle diameters plotted versus sinter ing temperature. .............. 55 3-8 Skeletonization in 2-D .................................................................................................... ...56 3-9 Topological connectivity .................................................................................................. .57 3-10 Topological length ....................................................................................................... ......58 3-11 MAZO angle schematic ..................................................................................................... 59 3-12 Skeleton network orientation versus sinter temperature ....................................................60 3-13 Nyquist plot of LSCF polarizat ion measured at 700 degrees C ......................................... 61

PAGE 9

9 3-14 Bode plot of imaginary impedance versus frequency, collected in air at 700 degrees C. ............................................................................................................................ ............62 3-15 The partial pressure dependence on polarization resistance. ............................................. 62 3-16 Bode plot of the 950 C sintered sample .............................................................................63 3-17 Bode plot which displays the deconvoluti on process using Voigt elements in series. ......63 3-18 Polarization resistance sintering temperature dependence. ............................................... 64 3-19 Partial oxygen pressure dependence of various processes for 950 C sintered sample m easured at 700 C. ............................................................................................................ .65 3-20 Measurement temperature depe ndence and activa tion energies ........................................ 66 3-21 Concentration polarization dependenc on inverse of effectiv e diffusion term at measur ment temperatures. .................................................................................................67 3-22 Concentration polarization dependen ce on MAZO angle of open pore phase at various measurem ent temperatures in air ........................................................................68 3-23 Ohmic polarization dependence on MAZO angle.at measurem ent temperatures. ............ 69 3-24 Process 3 polarization is realed to bot h cathode diameter and cathode connectivity. ....... 70 4-1 FIB/SEM cross section of cathode suppor ted SOFC sam ple imaged with charge contrast TLD detector. ....................................................................................................... 95 4-2 Epoxy sample preparation ..................................................................................................95 4-3 C-trench preparation ...................................................................................................... ....96 4-4 Phase contrast ............................................................................................................ .........96 4-5 Background EDS scan of Siemens-Westinghouse electro lyte/cathode interface cross section. ...................................................................................................................... .........97 4-6 16.7 m long line scan across cathod e/electroly te interface. ............................................. 97 4-7 EDS selected element profiles of linescan displayed in Figure 4-3 ................................... 98 4-8 FEG SEM SE detector image of cro ss section region scanned for background EDS scan of Figure 4-1 ............................................................................................................ ..99 4-9 Reconstructed volume of Siemens SOFC electrolyte, composite cathode, and cathode support........................................................................................................................ ......100 4-10 Reconstructed volume ..................................................................................................... .101

PAGE 10

10 4-11 Phase fraction is plotted against the di stance away from the LSM/YSZ interface .......... 102 4-12 Sample 3 3-D reconstruction ...........................................................................................103 4-13 Sample 4 3-D reconstruction. ..........................................................................................104 4-14 Graded porosity plots .......................................................................................................105 4-15 Sample 3 particle size. .....................................................................................................106 4-16 Sample 4 particle size. .....................................................................................................107 4-17 Volume normalized surface area measurements for sample 3 and sample 4. .................108 4-18 Two phase surface area as fr action of total surface area ................................................. 109 4-19 Surface area packing factor. ............................................................................................. 110 4-20 Triple phase boundaries in 2-D image of the Siemens Sa mple 3 composite cathode .....110 4-21 Comparative graded tortuosities ......................................................................................111 4-22 TEM foil preparation .......................................................................................................112 4-23 TEM sample 3 ..................................................................................................................112 4-24 TEM sample 4 ..................................................................................................................113 4-25 Sample 3 SEM image and corresponding TEM image of composite cathode. ............... 113 4-26 Sample 3 sum EDS spectrum of cathode/electrolyte interface. ....................................... 114 4-27 TEM image and corresponding line scan profiles across the cathode/electrolyte interface of sample 3. ....................................................................................................... 115 4-28 Sample 3 TEM EDS point scan .......................................................................................116 4-29 Sample 4 SEM image and corresponding TEM image of composite cathode. ............... 116 4-30 Sample 4 sum EDS spectrum of cathode/electrolyte interface. ....................................... 117 4-31 TEM image and corresponding line scan profiles across the cathode/electrolyte interface of sample 4. ....................................................................................................... 118 4-32 Sample 4 TEM EDS point scan .......................................................................................119 4-33 TEM-EDS elemental composition ................................................................................... 120

PAGE 11

11 4-34 Calcium composition comparison between el ectrolyte and cathode phases for sampl e 3, and sample 4. .............................................................................................................. .121 4-35 3-D skeletonization ...................................................................................................... ....122 4-36 Skeletonization in 2-D ................................................................................................... ..122 4-37 Schematic showing the transition from SE M im age which is used to generate a 3-D reconstruction model which is then re duced to a mass centroid skeleton ....................... 123 4-38 Schematic deptiction of Measured Av erage Z-Orientation (MAZO) angle. ................... 123 4-39 MAZO angle frequency di stribution for samp le 3 and sample 4 composite pore networks. ..................................................................................................................... .....124 5-1 Experimental LEAP setup................................................................................................ 143 5-2 Schematic of the dual beam FIB/ SEM assisted LE AP preparation. ................................144 5-3 A schematic of the three-step mi cromanipulator welding process .................................. 145 5-4 SEM image of an Imago Scient ific flat top Si post array ................................................ 145 5-5 Depiction of sample liftout and orientation of mi cromanipulator and GIS platinum needle with respect to sample trench ...............................................................................146 5-6 Low current FIB micrographs .......................................................................................... 146 5-7 Focused ion beam top view of Si post with sample atop ................................................. 147 5-8 SEM side views of the FIB annular beam tip sharpening using 30kV beam to form the cone and then an 8kV beam to remove gallium damage and form the final tip. ....... 148 5-9 LEAP reconstruction of Sample 1 ................................................................................... 149 5-10 LEAP reconstruction of Sample 2 ................................................................................... 150 5-11 Mass spectrum plots of counts versus mass-to-charge ratios for Samples 1& 2 ............. 151 5-12 Hydrogen atomic reconstructions al ong with the hydrogen mass spectrum .................... 152 5-13 Scandium atomic reconstructions along with the ma ss spectra for Samples 1 and 2. ..... 153 5-14 Oxygen atomic reconstructions along w ith the ma ss spectra for Samples 1 and 2. ........ 154 5-15 Oxygen atomic reconstructions are divided into three dis tinct peaks labeled in blue, peach and purple. ............................................................................................................. 155 5-16 Oxygen isotope 17 atomic reconstruction with mass spectra for Sample 1 and 2. .......... 156

PAGE 12

12 5-17 Calcium atomic reconstruction with ma ss spectra for Samples 1 and 2. ......................... 157 5-18 Manganese atomic reconstruction with ma ss spectra for Samples 1 and 2. .................... 158 5-19 Lanthanum atomic reconstruction with ma ss spectra for Samples 1 and 2. .................... 159 5-20 Voltage history plot for the analysis of Sam ple 1 ............................................................ 160 5-21 The reconstruction of Sample 1 .......................................................................................161 5-22 Sample 1's oxygen-16 and calcium mass-to-to charge ratio peaks .................................. 162 5-23 Zirconium oxide atomic reconstruction with ma ss spectra peaks for Samples 1 and 2 .. 163 5-24 Sample 1 atomic reconstruction is shown with analysis direction .................................. 164 5-25 Sample 1 atomic reconstruction and 1-D at omic concentration versus distance profile 165 5-26 Sample 1 high magnification atomic reconstruction and 1-D atomic concentration versus distance profile...................................................................................................... 166 5-27 Sample 2 atomic reconstruction with a 1-D atomic concentration versus distance profile. ...................................................................................................................... ........167 5-28 High magnification of Sample 2 atom ic reconstruction with a 1-D atomic concentration versus distance profile. .............................................................................. 168 5-29 Sample 2 atom probe tip with a void shown by arrow imaged with SEM ...................... 168 5-30 Sample 2-3294's atomic reconstruction is shown above its 1-D concentration profile. .. 169 5-31 Sample 2-3294's hydrogen atomic reconstruction and 1-D atomic concentration profile. ...................................................................................................................... ........170 5-32 Sample 2-3384 atomic reconstruction t op view, side view, and 1-D concentration profile ....................................................................................................................... ........171 5-33 Sample 2-3401 LEAP reconstruction .............................................................................. 171 5-34 Sample 2-3401 ............................................................................................................ .....172 6-1 Change in measured bias under constant applied current of LSCF/YSZ symmetric cell. ......................................................................................................................... ..........179 6-2 Effect of applied 12 hour bi as on LSCF/YSZ pol arization..............................................179

PAGE 13

13 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy A MULTI-LENGTH SCALE APPROACH TO CORRELATING SOLID OXIDE FUEL CELL POROUS CATHODE MICROSTRUCTURE TO ELECTROCHEMICAL PERFORMANCE By Danijel Gostovi December 2009 Chair: Kevin S. Jones Major: Materials Science & Engineering Solid oxide fuel cells (SOFCs) are a fuel and application flexible technology which may help the United States on its path to en ergy independence. The commercialization of intermediate-temperature SOFCs (IT-SOFCs) is limited by cathode and electrolyte polarization resistance. In an effort to understand the e ffect that sintering and doping processing conditions have on cathodic polarization, a multiple lengt h scale characterizati on approach has been developed to study SOFCs. The technique ut ilizes a focused ion beam / scanning electron microscope (FIB/SEM), transmission electron microscope (TEM) and a local electron atom probe (LEAP). The electrochemically active re gion of a strontium iron doped lanthanum cobaltite (LSCF) cathode was for the first time reconstructed in three dimensions using a FIB/SEM. Various microstructural properties were measured incl uding overall porosity, closed porosity, graded porosity, surface area, tortuosity, and pore size. Electrochemical Impedance Spectroscopy (EIS) data was correlated to microstructure. Microstructure and chemical segregation of composite cathodes were also analyzed using the FIB/SEM technique. The cathode/electrolyte interface was characte rized with TEM-Energy Dispersive Spectroscopy (TEM-EDS) where calci um doping was found to have a significant

PAGE 14

14 effect on cathode microstructure. Higher calcium concentration sa mples had a coarser composite cathode microstructure. The cathode/electrolyte composite cathode interface was characterized via TEM-EDS line and point scans. The most si gnificant difference seen was the coarser sample having a higher calcium content in both el ectrolyte and cathode phases by 1 and 1.4% respectively. The active regions of composite cathodes were furt her analyzed and for the first time the complex 3-D network topological c onnectivity has been measured. Such novel connectivity quantification allows for an advan ced understanding of the transport processes in composite materials. An atomic resolution, LEAP microscope has be en used for the first time to characterize SOFC materials. Two separa tely processed Siemens Energy Incorporated composite cathodes were analyzed. The scandium doped zirconia (SSZ) / calcium doped lanthanum manganate (LCM) buried interface was micro-m achined into an atom probe tips and analyzed with the LEAP. Magnesium was found to segregate beyond the 10 nm wide interface another 10+ nanometers. Interfacial voids were found to be hydrogen enriched. Two-dimensional concentration profiles offer a glimpse at atomic segregation near electrochemically active triple phase boundaries (TPBs).

PAGE 15

15 CHAPTER 1 GENERAL INTRODUCTION Solid oxide fuel cells, often re ferred to as SOFCs are effici ent, environmentally friendly, and fuel flexible electrochem ical devices which generate electrical power, heat and water. They consist of three basic layers: ca thode, electrolyte, a nd anode. The functiona l requirement for the cathode is that it is a porous catalyst for the reduction of O2. The anode layer is a porous catalyst for the oxidation of fuel (H2). Sandwiched in between the cat hode and anode is the electrolyte which needs to be sufficiently dense to prevent mixing of the H2 and O2, an ion conductor and an insulator to electrical current A representative schematic of an Oxygen ion conducting SOFC can be seen in Figure 1-11. The deregulation of power comp anies in re cent decades, combined with growth in consumer electronics has brought intermediate-t emperature SOFCs (IT-SOFCs) to the forefront of portable power 1,2. Everything from laptops, to cell phones, to military radios, to automobile and aviation air conditioning systems require a du rable, high-power density mobile power source to run for long periods of time (5,000 hours 3). Fuel cells have a power density and size advantage over the current state of the art Li -ion batteries. They can potentially provide ten times the power density (3000Watthours per liter versus 300 Wh/l, ( Figure 1-22), of such batterie s. IT-SOFCs are feasible for mobile a pplications due to their sub-800 C operating temperatures which accommodate consumer electro nics thermal cycling re quirements better than traditional high temperature SOFCs. Intermediate temperature operation allows the material requirements on the interconnect and balance of pl ant materials to be less stringent, and less expensive. This cost reduction associated with the IT-SOFCs ba lance of plant materials makes them more commercially viable 4 than traditional high temperature SOFC systems. A disadvantage of reduced operating temperatures is increased cathodic polarization 4. This

PAGE 16

16 polarization loss may be reduced with material se lection of fast kinetics materials such as La1xSrxCo1-yFeyO3(LSCF), and optimization of microstruc ture for low temperature operation. But the effect on long term durability must also be better understood when using such fast kinetics materials often with nano-sized microstructures. Stability of IT-SOFCs is a key requirement for successful commercialization 5. In accordance with these issues and require ments, the proposed studys focus is the correlation of microstructure a nd electrochemical performance of various SOFC cathodes. The study will be three pronged: (a) processing of various microstructu re single phase, and composite cathodes, (b) performance testing of these materials using electrochemical impedance spectroscopy (EIS) and (c ) characterization of el ectrochemically active cathode regions using a dual-beam focused ion beam/scanning electron micr oscope (FIB/SEM), a transmission electron microscope (TEM), and a loca l electron atom probe (LEAP).

PAGE 17

17 Figure 1-1. Schematic Diagram of Solid Oxide Fuel Cell Figure 1-2. Practical system volum es. Reprinted with permissi on from Elsevier Dyer, C.K. "Fuel cells for portable applications." J. Power Sources 2002;106:31. Copyright 2002, Elsevier.

PAGE 18

18 CHAPTER 2 BACKGROUND 2.1 Cathode Background The basic function of an SOFC cathode is to reduce oxygen species using the electrons comi ng from the external load connection. The basic chemical form ula of this process is shown in Equation 2-1. X O OOeVO 22 2/ 1 (2-1) Due to the high operation temperatures of SOFCs (600-1000 C) only noble metals and electronic conducting oxides are fe asible as cathode. Initially the most widely studied noble metals were platinum and silver. The problem w ith noble metals was their high cost and lack of long term stability due to coar sening, cracking and evaporation 6. The first SOFC cathode suitable perovskite metal oxide was La1-xSrxCoO3(LSC) which was first published in 1966 7. In 1973 the favored perovskite SOFC cathode became La1xSrxMnO3 (LSM) 8 thanks to its high electr onic conductivity. Its crystal structure is based on the ABO3 perovskite structure 9 and it conducts electrons via a small polaron conducting mechanism which results in electroni c conductivities in excess of 100 S cm-1 at 950 C 10. LSM is limited for IT-SOFC application by its low ionic conductivity of 10-7 S/cm (measured at 900 C) 11.A SOFC with a purely electronic conducti on cathode relies on the two-dimensional (2D) triple phase boundary (TPB) to reduce oxygen. The triple phase boundary (TPB) is the site between cathode, air an d electrolyte where O2 can be directly reduced and incorporated into the oxygen vacancy, VO rich electrolyte. The regime in wh ich this triple phase boundary exists is referred to as the active region. For cathodes such as LSM the typical TPB length for a cathode fired at 1100 C is on the order of ~1 m-1 12. This is a result of TPB distance divided by active area.

PAGE 19

19 2.2 Mixed Conductor SOFC Cathode The reduced operating temperature of IT-SOF Cs makes highly electrocatalytic cathodes desirable. Use of a perovskite (Figure 2-1) mixed ion and elect ron conducting cathode (MIEC) boosts perform a nce by extending the active laye r depth thus improvi ng the kinetics at intermediate temperatures (< 800 C) regime 9,13 MIEC cathodes such as La1-xSrxCo1-yFeyO3(LSCF) are effective in extending this active region because the reduction of oxygen can occur anywhere on the surface of the cat hode, and is not limited to a TPB 14. A schematic comparison of the purely electronic cathode (LSM) and the MIEC cathode (LSCF) activ e regions can be seen in Figure 2-2. LSCF of typical composition L a0.6Sr0.4Co0.2Fe0.8O3has an ionic conduction is 0.2 S cm-1 15, which is orders of magnitude higher than LSM 9, despite LSMs high electronic conduction. The indicates oxygen non-stoichiometry, cause d by crystallographic point defects. The use of such a MIEC with an extended act ive region and high ionic conduction is that the activation polarization of the cathode is decreased thus effectively increasing the potential power output of the IT-SOFC 14. 2.3 Electrolyte Background The modern electrolyte owes its origins to physicist Walther Nernst who initially proposed stabilized zirconia as a light bulb filament for incandescent light in the 1890s. But the considerable start up time required by zirconia to reach the 600C operational threshold made it inferior to tungsten for this application 16, 17. These original startup time problems still plague SOFCs over a hundred years later. The electrolyte must be stable in both oxi dizing and reducing environments and have sufficiently high ion conductivity paired w ith low electronic conduc tivity in the SOFC operational temperature window. Th e first such feasible electro lyte was yttrium stabilized zirconia (YSZ). The initial 15 wt% Y2O3 doping of ZrO2 was accomplished by Nernst in the

PAGE 20

20 1890s, it was first used as a SOFC electrolyte by Baur and Preis in 1937 6. Throughout the years fluorite-structured and perovskite type materials have been applied as SOFC electrolytes. The main issues with the perovskite type electrolyt es have been reaction wi th electrodes and higher cost than fluorite type electroly tes. Bismuth oxide, the highes t oxygen ion conductor to date has been omitted from the study due to intermediate temperature degradation issues arising from a reduction around 600 C 18,19. One of YSZs disadvantages is its reaction with LSCF above 1100 C to form insulating phases. This study accordingly utilized a La0.8 Sr0.2 Co0.2 Fe0.8 O3composition to further reduce the reaction with YSZ than the standard La0.6Sr0.4Co0.2Fe0.8O3composition. Zirconia possesses a fluorite structure at SO FC operating temperatures. Mobile oxygen vacancies are created by substituting Zr4+ with lower valence rare earth ions such as yttrium as shown in Equation 2-2. X O OZr ZrOOVY OY 3 2' 322 (2-2) From the electroneutrality condition we can surmise the oxygen vacancy concentration from the doping level, [' ZrY] = 2[OV]. This implies that the vacancy concentration is linearly related to doping with the trivalent cation. The ionic conductivity can thus be expressed as Equation 2-3 20. = zen (2-3) Here z is the charge number, e, the unit char ge, n the number of charge carriers, [e], and the mobility of the vacancies. In the case of ion conductors like YSZ, analyzing their conduction using Arrhenius plots at various temperatures allows for the extraction of an activation energy. As seen in Equation 2-4 20 where [Vo ] is the site fraction of m obile oxygen vacancies and 1-

PAGE 21

21 [Vo ] the site fraction of unoccupied oxygen vacancie s. To conduct ions the electrolyte has to have sufficient unoccupied oxygen vacancies in clos e coordination to the occupied ones so that = A/T[Vo](1-[Vo])exp(-E/RT) (2-4) where A is the pre-exponential constant (which includes the number of equivalent sites per unit volume, temperature, particle charge, the Boltz man constant, lattice constant, and vibrational frequency), E is the activation energy, R, the gas constant and T, the temperature 20. The total conductivity of zirconia depends on the dopant selection and doping concentration. The conductivity is maximized at specific concentrations which have been experimentally observed 21,22 to occur at lower dopant concen trations than those predicted by Equation 2-4. Possible explanati ons for this are associated defect pairs formed between the dopant cation and the oxygen vacancy whic h effectively reduce vacancy mobility 20. 2.4 Research Method The field of materials science and engin eering exploits the re lationship between a materials microstructure, processi ng, properties, and performance. In this work we have studied the interfacial chemical segrega tion and microstructure of SOFC cathodes, and how they change with sintering temperature, doping and composition. Much IT-SOFC work has been done on im proving power density and reducing operating temperature,17-21 but not much has been done to correlate performance to actual 3-D microstructural or chemical properties. To achieve this, both composite and single phase cathodes were characterized in 3-D. Two different cathode groups have been charact erized: single phase cathodes compromised of sub-micron powders and micron sized compos ite cathodes. The singl e phase cathode has been deposited on micron grain sized, 100 micron th ick YSZ substrates using a screen printing method which has been proven as a high power density IT-SOFC electrolyte deposition

PAGE 22

22 technique 23. To eliminate variations in the cathode a commercially available LSCF powder (Praxair), a texanol-based vehi cle (Electro Science Products) and a Thinky planetary centrifugal mixer, were used with thorough, uniform mixing of the screen printing ink 24. The symmetric cell cathodes where then sintered at an appropriate temperature (800-1200 C) to yield a uniform film. Composite cathodes were processed by Si emens Energy, Inc. Similar composite cathodes have been previously studied by various groups 19-23. 2.5 Electrochemical Impedance Spectroscopy The single phase cathodes were analyzed in a symmetric cell setup using electrochemical impedance spectroscopy (EIS). EIS is commonl y used to gauge the various polarizations occurring in an SOFC. A polarizat ion is a loss of voltage theoretically available at a particular current density. This term is also known as an overvoltage and is commonly represented by the symbol There are three types of pol arizations found in SOFCs: (a ) ohmic; (b) concentration; and (c) activation. Ohmic polarization stems from the fact that most materials exert a resistance to the flow of charged particles, which can ofte n be described Ohms law. In an SOFC each of the electrodes and the electrolyte exhibit resistance to the flow of the respective charge d particles, be they vacancies, electrons or holes. When all of the contributions are combined the ohmic overpotential, ohm is represented by Equation 2-5. iRl l lcontact anode anode cathode cathode eelectrolyteelectrolyt Ohm (2-5) Here represents the specific layers porosity corrected resistivity, and l represents the layer thickness. Any contact resistan ce in the system is represented by Rcontact 17. This polarization contribution can be modeled by a resistor since the response time is very fast.

PAGE 23

23 Ohmic polarization would thus be affected by micr ostructural parameters such as film thickness, and by the composition of the electrode wh ich is a materials selection issue. Concentration polarization stems from the flow of gaseous reactants through the electrodes to the electrochemically active areas and the resultant outflow of reacted gases. On the cathode side it is linked to the binary diffusion of O2 and N2, DO2-N2, the cathode microstructure, partial oxygen pressures, and the specific cu rrent density. On the anode si de it is linked to the binary diffusion (in the simple case) of H2-H2O, the anode microstructure partial hydrogen and water pressures and the specific cu rrent density. The cathode c oncentration ov erpotential is represented as Equation 2-6 17. cs cath conci i F RT 1ln 4 (2-6) Here R is the gas constant, T, temperature, F, Faradays constant, i, specific current density, and ics the cathode limiting current density 25, beyond which mass flow starts to limit electrochemical performance. The limiting current density term includes the partial pressure, effective diffusivity, temperature, and microstructural parameters 17: cath cath O effcath cath O csRTl p pp DFp i 2 2)(4 (2-7a) Where F is Faradays constant, cath Op2 the partial oxygen pressure on the cathode side, R, the gas constant, T, temperature, and cathl cathode thickness, and )( effcathD, the effective cathode diffusivity is defined as seen in Equation 2-7b 26. 22)(NOv effcathDV D (2-7b)

PAGE 24

24 whereNOD2is binary diffusivity of O2-N2, vV is volume porosity fraction, and is tortuosity 26. Thus from Equations 7a-7b we can see the direct connection between cathodic concentration polarization and speci fic microstructural parameters. From Equation 7 we also see that concentration overpotential increases non-linearly with current density. This polariza tion is commonly modeled by an equivalent circuit known as a Warburg element which is represented by a variety of resistors and capacitors 27. Here the nonzero capacitance capacitors ensure a non-zero res ponse time, unlike the ohmic model circuit. The time dependence can not be described by simple first order kinetics, but a characteristic time is defined based on the cathode thickness, lcathode and cathodic effective diffusivity, Dcathode(eff) (Equation 2-8). Typically the characteristic tim e for a cathode is in the millisecond range 17. tcharacteristic ~ lcath 2/ Dcath(eff) (2-8) SOFC electrode concentration polarization is dominated by the cathodic contribution for two main reasons: (a) binary diffusion of H2-H2O (common SOFC fuel) on the anode side is upwards of five times fa ster than that of O2-N2 on the cathode side mainly due to the low molecular weight of hydrogen; a nd (b) partial pressure of hydrogen on the fuel side is greater than the partial pressure of oxygen on the air side In short, we want to optimize and study the cathode, because it has a bigger effect on cell performance than the anode. Activation polarization comes from the transfer of charge within the cathode, anode, or electrolyte phase which converts neutral specie s to charged ones and vice versa. For the reduction of oxygen there is a multiplicity of pathwa ys for the reactions to occur and this is especially the case in an MIEC cathode which can conduct both electronic and ionic charges 28. This thermally activated process is in the simplest view limited by the slowest reaction step. It is directly related to the cathodic cu rrent flow, which has an associat ed loss in voltage known as the

PAGE 25

25 activation overpotential. This overpotential is a function of material properties, microstructure, temperature, atmosphere and current density. The Butler-Volmer equation gives the current density as a non-linear function of activation polarization which for the cathode side is shown in Equation 2-9 17. RT zF RT zF iicath act cath act cath 1 exp exp0 (2-9) where cathi0is the cathodic exchange current density, is the transfer coefficient, z is electric charge, F is Faradays constant, cath actis the cathode activation polar ization. Note that this relationship is an implicit, nonlinear relations hip between the activation polarization and the current density. Thus we can not explicitly solve for cath actas a function of i unless we evaluate it at the limits; namely at the exchange current density (low i, RT zFcath act<< 1) and the limiting current density (high i, RT zFcath act>>1). For the low i case the relationship simplifies to 17: i zFi RTcath cath act 0 (2-10) For the high i case the relationship can be approximated as the Tafel equation 17, Equation 2-11. ibai zF RT i zF RTcath cath actln ln ln0 (2-11) Note here that the cathzFi RT0 term is in units of cm2 and is known as the charge transfer resistance. Activation polarizations ther mal dependence tends to dominate IT-SOFC polarization at reduced operational temperature17,4 but to date a definiti ve relationship has not

PAGE 26

26 been developed between the charge transfer resistance and microstructural parameters for MIECs. Hence, this research has an excelle nt opportunity to deve lop this relationship. All three overpoten tials sum to yield total cathodic pol arization. The cathodes with the lowest symmetric cell total polarization under acceler ated testing conditions will be promoted as candidates for durability testing. The IT-SOFC polarization may then be correlated to 3D microstructural and chemical changes. 2.6 Microstructural Analysis Since cathode microstructure and morphology have a strong effect on cathode polarization 8,29,30, dual beam FIB/SEM will utilized to reconstruct actual 3-D models of the single phase and composite cathodes, and their electrolyte interfaces. This high resolution, 3-D technique allows for the quantification of critic al microstructural properties su ch as surface area, tortuosity, interfacial porosity, and conn ectivity and correlation to el ectrochemical polarization. The semiconductor industry has used th e FIB since the 1980s to deposit, etch, micromachine, and image specimens during different stages of circuit processing 31, 32. This technology was brought forward to reconstruct 3D, geometrically complex sub-micrometer structures 33, 34, 35, 36, 37. With the advent of 3-D modeling so ftware nano-tomography utilizing the dual beam FIB/SEM technique was used to quantify nano-ceramic suspended powders 36,37, 38. This technique was recently applied to SO FC cermet anodes to qua ntify microstructural properties such as porosity, TPB length, and degree of anisotropy via tortuosity 39. Subsequently we have developed a similar technique to rec onstruct a cathode and th e cathode/electrolyte interface 40, and correlated EIS measured polarization to selected microstructural properties 40, 41. The automated sectioning and imaging w ill be conducted with a FEI Strata DB 235 FIB/SEM dual beam system. A schematic of the dual beam orientation is shown in Figure 2-3. The ion and electron pole pieces ar e orien t ed at 52. The system has an in-situ liquid metal-

PAGE 27

27 organic ion source (LMIS), for th e deposition of protective plati num layers. Rastering the FIB across the region of interest physically sputters aw ay material, which is redeposited locally. The field emission gun (FEG) SEM is utilized to image the freshly milled surface with a ThroughLens-Detector (TLD) down to a maximum resolution of 3 nm. The TLD deep hole contrast was used to distinguish between the electrolyte and cathode phases ( Figure 2-4). To distinguish between the different detectors on the FIB/SEM please see Figure 2-5. This process of slicing and im aging was auto m ated with the use of Auto Slice and View software package (FEI Company). Af ter the images are acquired AmiraTM-ResolveRTTM version 4.0 (Mercury Computer Systems Inc.) was utilized to align, segment, and concatenate the serial images into a 3-D triangular mesh (movie:http://hitec.mse.ufl.edu/ ( Figure 2-6.6). Amira was used to quan tify the su rface area, volume porosity, graded porosity 40, and connectivity of the various phases 42. Various groups have studied such phenomena as SOFC break-in during which IT-SOFC performance stabilizes 3,43,44, which last from 100-300 hours, but the microstructure analysis has been limited to 2-D. The combination of the EIS and the FIB/SEM technique allows us to reconstruct cathode microstructure so see what microstructural ch anges are occurr ing as a result of processing differences. This imaging was done at room temperature, under a 1 x 10-7 torr pressure. 2.7 Chemical Analysis Cathodes have been studied with transmission-electron-mi croscope/ energy-dispersivespectroscopy (TEM/EDS) analysis to track chem ical composition changes across the interface. When the FIB/SEM images are collected a TEM foil may be created using common liftout techniques 45. This chemical analysis will elucidate the presence of interfacial cross-diffusion during various stages of long term testing. Th e local electron atom probe (LEAP) technique has

PAGE 28

28 allowed us to quantify the atomic segregation across cathode/electrolyte interfaces. A more in depth discussion of this is presented in Chapter 5.

PAGE 29

29 Figure 2-1. Perovskite crystal structure. LSCF constituents are color labeled. Figure 2-2. Common strategies for SOFC cathodes. (a) porous single-phase electronically conductive oxide such as (La,Sr)MnO3 (LSM); (b) porous single-phase mixed conductor; (c) porous two-phase composite. Reproduced in part with permission from Adler S.B. "Factors governing oxygen reduction in solid oxide fuel cell cathodes. Chem. Rev. 2004;104:4791 Copyright 2004, American Chemical Society.

PAGE 30

30 Figure 2-3. FIB/SEM. (a) FEI St rata DB 235 Focused Ion Beam / Scanning Electron Microscope (FIB/SEM) (b) Beam orient ation incident upon sample. Figure 2-4. FIB/SEM phase contra st. Using the dual beam FIB/SEM it is possible to distinguish between electronic and ionic conducting phase s as seen in the th is SEM image of a polished cross section of ruthenium d oped bismuth oxide and bismuth oxide composite cathode.

PAGE 31

31 Figure 2-5. Various TLD detector settings. (a) scanning, (b) backscatter -150V (c) charging, (d) deep hole and (e) back scatter -250V

PAGE 32

32 Figure 2-6. Schematic depicting transformation of 2-D images into a 3-D reconstruction and graded porosity.

PAGE 33

33 COMPREHENSIVE QUANTIFICATION OF POROUS LSCF CATHODE MICROSTRUCTURE AND ELE CT ROCHEMICAL IMPEDANCE 3.1 Introduction Current SOFC performance is limited by cat hode polarization, which increases with decreasing operational temperature 29 8. The SOFC cathode is a por ous, conductive catalyst for the reduction of oxygen: hOVOX O O2 2/12 (3-1) Cathode microstructure and morphology have a strong effect on this polarization 29,8,30. In an MIEC such as LSCF there are three pr imary polarizations in addition to the ohmic and concentration contributions. They are: charge transfer (which occurs when an oxygen ion is incorporated into the electrolyte), adsorption (where diatomic oxygen is dissociated into oxygen species on the cathode surface) and ionic conduc tion (where an oxygen ion is conducted through the cathode bulk). Possible pathways for th e three additional processes are shown in Figure 331. W e see that one pathway is oxygen diffusion to the TPB boundary via open pores. Another possible pathway is adsorption of oxygen onto th e cathode surface where it di ssociates and either is incorporated into the LSCF phase to be conducted by vacancy mechanism or is conducted along the LSCF surface en route to the electrolyt e/cathode interface. Another possible route is grain boundary conduction of reduced oxygen which is reduced at the cathode/air surface. It is important to note that there is a multitude of possible pathways 28, and we are not limited to just the three modes shown in Figure 3-31. In this study a dual beam focused ion beam /scanning electron m icroscope (FIB/SEM) was utilized to reconstruct actual three-dimensional (3-D) models of La0.8Sr0.2Co0.2Fe0.8O3(LSCF) cathodes, and their interface with a dense yttrium-stabilized-zirco nia (YSZ) electrolyte. This high resolution, three-dimensional technique advances the understa nding of the cathode

PAGE 34

34 microstructures effect on performance. The identification of critical microstructural properties such as surface area, tortuosity, interfacial porosity, and topological connectivity were correlated to the ionic, electronic, and catalytic processes for a better fundamental understanding of electrochemical performance. 3.2 Experimental Square LSCF symmetric cell cathodes (8 X 8 mm) were screen printed using premixed LSCF ink (NexTech Materials, Inc) on both sides of a 100 m thick polycrystalline yttriumstabilized-zirconia (YSZ) electr olyte (Marketech Internationa l, Inc.) using common screen printing techniques 46,47. After low temperature drying to eliminate the organic vehicle, the samples were sintered at various temperatures for one hour with a 5 degree per minute ramp up and down. The study incorporated sintering temperatures every 50 degrees between 850 and 1100 C for a total of six sintering conditions. The resulting porous cathode films were approximately 20 m thick. Potentiostatic impedance measurements were collected at various at mospheres (300 ppm to 20.9% oxygen) and at various temp eratures ranging from 550 to 700 C at every 50 C interval using a Solartron 1260 frequency-response analyzer under a potentiostatic modulation of 50mV. The frequency range was 10 MHz to 10 mHz with ten points per decade. Electrode contacts were made using a platinum mesh pressure contact and leads. The automated sectioning and imaging was conducted with a FEI Strata DB 235 FIB/SEM dual beam system. The field emission gun (FEG) SEM was utilized to image the freshly milled surface with a Through-lens-detector (TLD) down to a maximum lateral resolution of 3 nm. The TLDs charging contrast was used to distinguis h between the electrolyte, cathode and open pore phases. The use of epoxy impregnation increased the open pore phase contrast and allowed for a 2-D image free of 3-D artifacts in the slice and viewing process ( Figure 3-42). After

PAGE 35

35 impregnation, the sample stub was polished down to a 1 m roughness to planarize the surface and for assurance of the top surface being orthogonal to the interface. Epoxy mounting also allowed for mounting of all samples on one specime n holder, and four repeats were collected for each sintering condition, on both si des of the symmetric cathode ( Figure 3-53). This FIB method was used to create a trench ar ound the region of interest (ROI). The slicing (z-axis) resolution was 50 nm which is optim al z-axis resolution for the >250 nm particle diameters. This process of slicing and imaging was expedited with the Auto Slice and View software package (FEI Company). Ion images were captured at regular intervals during the automated slicing, so as to corroborate the z-direction slice thickness. 20,000X was the im aging magnification during the slice and view for optimal microstruc ture resolution and quantification. AmiraTM-ResolveRTTM version 4.0 (Mercury Computer Systems Inc.) was utilized to align, segment, and concatenate the serial images into a 3-D triangular mesh. The reconstruction volumes were cropped to 125 m3 for each sample, with four (total 500 m3) samples per sintering condition. 3.3 Results and Discussion 3.3.1 Microstructural Quantification Six typical reconstructions of the blue/dark electrolyte phase and orange/light cathode phase ( Figure 3-64) allow for qualitative observation of m i crostructure coarsening, as sintering temperature is increased. The microstructura l properties of porosity, surface area, TPB length, particle diameter, tortuosity, were quantified for all six samples. Subsequent 3-D skeleton networks were quantified for connectivity, average topological length and orientation in the form of the measured average z-orientation (MAZO) angle. The AmiraTM tissue-statistics module was used to cal culate the volume pore fraction. This module is able to count up all of the voxels of a phase in order to determine the volume pore

PAGE 36

36 fraction. The porosity volume fractions (Vv) were tabulated in Table 3-1 and plotted in Figure 375. The volume porosity decrease s with increasing sintering tem perature. The 850 C sample had a volume fraction of 0.56 which deviated sligh tly from the trend, but was identical to the 950 C sample. The volume porosity versus sinter ing temperature trend agrees with coarsening theory whereupon as sintering temperature is increased, volume porosity decreases due to particle sintering. Tortuosity () was measured using AmiraTMs center of mass calculations. Tortuosity was quantified by tracking the center of mass of each pore as it goes from the surface of the cathode to the YSZ interface. The total length of this path was divided by the Euclidian distance between the air/cathode interface and cathode/electrolyte interface. Error bars were calculated from the standard deviation of the data for each sinteri ng condition. There were four repeats for each sintering condition. Figure 3-75 (Table 3-1) displays the tortuosity versus sintering temperature. Maxim u m tortuosity is observed at the 950 C processing condition. This occurs in the transition region between fine, loosely connected low sinterin g temperature and coarser high sintering temperature cathode pa rticles. The intermediate we ll interconnected cathode may exhibit the highest tortuosity, as a result of the high surface area, high TPB length and connectivity. The concentration polarization is related to the effective diffusivity of air through the porous cathode. The volume fraction of porosity divided by the tortuosity is related to the effective diffusion coefficient, D (eff) 26 as shown in Equation 3-2. D (eff) = D Vv / (3-2) D is the diffusivity of O2 through an unconstrained volume of air. A reduction in Vv / decreases the effective O2 diffusion through the cathode resulting in an increase in any cathodic

PAGE 37

37 polarization attributable to gas diffusion. It is evident that as the cathode is sintered the Vv / term decreases (Figure 3-5) due to coarsening of the microstructure at higher sintering temperature. Thus we would expe ct the concentration polarization to increase at higher sintering temperatures. The relationship between these (a nd other) microstructural features and measured impedance are shown in a subsequent section. AmiraTM tissue statistics module a llowed for measurement of the surface area between the cathode, and open pores. The measured surface area ( Table 3-1, Figure 3-86) is within the standard deviation of the previous studies 48 and shows a maximum va lue near the 950 C sintering condition. The volume normalized surf ace area (SA/Volume) results indicate that as the cathode sinters, the separated particles start to form necks, increasing surface area then with sufficient thermal energy the neck s thicken and the cathode coarsens. Another indication of microstructural coarsening is the triple phase boundary (TPB) length per unit area. The cross-sectional area normalized TPB length trend ( Table 3-1, Figure 3-86) is sim ilar to th e surface area trend and shows a maxi mum value at the 950 C processing condition. Just as neck formation increases surface area, it also increases the TPB length at the cathode/electrolyte interface. The initial disper sed particle TPB length is low and it increases with neck formation attributable to the coarse ning process, which is common in covalently bonded solids such as this perovskite cathode. Such solids prefer to coarsen over sintering due to relatively high vapor pressures, which are cau sed by high vapor/cathode interfacial energy 49,11. A common technique for particle/pore size cal culations is the use of the BET formula d=6V/S 36,50. Measurement of both open pore and cathode particle diameters were conducted. The average cathode particle diameters (Table 3-1, Figure 3-97) exhibit the expected 49 parabolic dependence on sintering temperature. It increases from 350 nm at the 850 C processing

PAGE 38

38 condition to ~500 nm for the 1100 C condition. Th e lowest sintering condition appears to be an outlier (in diameter terms), perhaps because it is below the synthesis temperature for the precursor powders. This high diameter is also affected by the relatively low porosity for this sample. We see that the 850 C sample appears to be an outlier in multiple microstructural properties, but we can not be sure without an even lower sintering condi tion, such as 800 C, which was not studied due to poor adhesion to YSZ substrate. The average open pore diameters ( Table 3-1, Figure 3-97) decrease as sintering tem p erature increases through the necking regime, but then increases at high temperatures as sintering starts to occur. The latter affect being due to the formation of fewer but larger pore structures. 3.3.2 Connectivity Quantification The connectivity was quantified using a newly developed skeletoniza tion technique with the aid of the Amira software. Skeletonization ta kes the cathode or pore 3-D reconstruction, and through a number of iterations, thin s the microstructure until only the structur al center of mass is left ( Figure 3-108). This center of mass network di splays th e interconnect ed nature of each phase, and allows us for the first tim e, to quant ify topological connectivity. For a more in depth background of topological connectivity please refer to Gostovic et al 51, 52. The pertinent connectivity values that were qua ntified were N, the number of ve rtices or nodes; E, the number of edges; , the mean degree or connectivity ; L, the average topol ogical network edge length; and kmax the maximum degree of a network ( Table 3-2). The open pore phase number of vertices for each sintering condition is shown in Table 3-2. The number of open pore vertices tends to decr ea se (within a fixed volume) as sintering temperature is increased. This indicates that the numb er of pores is decr easing. The cathode phase number of vertices for each sintering condition is also shown in Table 3-2. The number of

PAGE 39

39 cathode vertices tends to decrease (within a fixed volume) as sintering temperature is increased. This indicates that the number of cathode particles is also decreasing. The open pore phase number of edges for each sintering condition, shown in Table 3-2, indicates that there pore channe ls are formed as the cathode is sintered. The cathode phase number of edges for each sinter ing condition (Table 3-2) indicat es that there are less interparticle connections (or number of grain boundaries) as the cat hode is sintered. The only exception is the 850 C processing condition, which has less nodes and edges than the 900 C condition. This is possibly caused by insuffici ent thermal energy relative to the LSCF powder synthesis (because at 850 C, all that has happene d is that the screen printing vehicle has been burned off) and what remains is a loosely connected and well dispersed volume of particles. The number of vertices and edges al so indicates that we have suff icient numbers of particles and pores for proper counting statistics. The mean degree or topological connectivity of both open pore and cathode phases are shown in Figure 3-119 and tabulated in Table 3-2. We can see that the connectivity is m a ximum at the 950 C processing condition, for both the cathode and pore phases. This may indicate that the highest surface area, and highest TPB length is synonymous with highest connectivity microstructure. The fact that both phases exhi bit similar connectivity magnitudes confirms that this is in fact a volume and size independent pr operty. The difference in the profiles for the cathode and open pore phases may be attributable to the fact that the cathode phase starts to dominate the volume fraction at higher sintering temperatures, due to the occurr ence of partial sintering in the cathode s contraction. The open pore and cathode phase average t opological lengths (L) are plotted in Figure 31210 (Table 3-2) and both trend up w ith increasing sintering tem perature. The trends agree with

PAGE 40

40 the observation that the microstructure is coarsening (with the exception of 850 C sample). Furthermore this corroborates particle diameter trends. It also shows that utilizing the BET formula for the pore particle size has its limita tions because it uses global porosity and surface area measurements which do not distinguish between agglomerates and well dispersed particles. The network topology method separa tes the cathode and pore phase s into individual particles which allows for a more precise measurement of particle size (topol ogical length between adjacent particle/pore centers of mass). The network topological length is roughly half the magnitude of the cathode and open pore diameters due to the fact that two large particles are rarely going to be nearest neighbors in a hard shell, close packed system with a normal particle size distribution. The topologica l length distribution is skewed toward the short distance end which results in more short than long edge lengths. Furthermore, the cathode average topological length trend in Figure 3-1210 is greater than the po re phase because the pore network is m ore diffi cult to alter because is it not dependa nt on the transfer of mass which occurs during coarsening. Thus the pore phase network is more robust during the increased sintering conditions. A way of quantifying the maxi mum mean degree within each network is to list the kmax (Table 3-2). This property also helps to iden tify volumes where particle agglomerates may be present. This indicates that the 900 C sintering condition had a massive vertex where 43 different pores were connected. Table 3-2 also indicates that the 1050 C sintering condition had a massive vertex where 31 different cathode particle s were connected. It is important to note that the kmax was observed in just one volume, but ther e were three other volumes which may not have had such a high kmax. This means that although the kmax was measured for all four repeats,

PAGE 41

41 the kmax was highest in one repeat. The lack of a trend in kmax across sintering conditions indicates that agglomerates are randomly distributed. A method for quantifying particle or pore orientation has al so been developed. The measured average z-orientation (MAZO) of the sk eleton network, is calculated with respect to the z-axis which is orthogonal to the cathode/electrolyte interface ( Figure 3-1311). As the angle increases from 0 to 90 the path becom e s longer and deviates from the z axis which is normal to the electrolyte/cathode interface. Thus, should increase with increasing MAZO and one would expect a MAZO angle of 45 for a completely random cathode network. Figure 3-1412a-d shows the trends in orientat ion quantif ication which consists of: (a) the skeleton network orientation quan tification MAZO angle (the aver age angle between all of the connecting edges in the skeleton); (b) MAZO SD (the standard de viation of all MAZO angles in the range 0 to 90 ); (c) Not In Plane (NIP) angle, (which is identical to MAZO angle except that it omits edges which are in the Z-plane thus runs parallel to the cathode /electrolyte interface); and (d) percent In Z-Plane (% IZP) (quantificatio n of the percentage of total edges which are in the Z-plane {parallel to the cathode/electrolyte interface}). Figure 3-1412a ( Table 3-2) plots the MAZO angles fo r both cathode and open pore phases. The open pore phase MAZO angles tr end indicates that the o pen skel eton orientation is stable at the lower temperatures and starts to increase under high sintering conditions. The cathode phase MAZO angle trend indicates that the cathode skeleton orientati on trends up with sintering tem perature. For an in depth discussion of the definition of MAZO please refer to 51,52. In Figure 3-1412a we see that the MAZO angle tends to inc r ease with coarsening starting close to the 54.7 angle that exists between adjacent, close packed atoms in the (110) plane and the <100> direction in a BCC close packed hard sphe re model. The increasing MAZO angle is in

PAGE 42

42 agreement with the changes that oc cur during sintering attributable to cathode shrinkage. We see that both pore and cathode phase exhibit the same trend caused by cathode coarsening and partial sintering. Figure 3-1412b plots the MAZO standard deviation (SD) and is an estim ate of how wide the MAZO angle Gaussian prof ile is when inco rporating the whole volum e. Additionally the MAZO SD may have a slight d ecline due to the declining nu mber of nodes as sintering temperature is increased. Figure 3-1412c is a plot of NI P angle, which is th e MAZO angle m inus the edges in the plane parallel to the cathode/el ectrolyte (Z-plane). A rising trend in NIP angle is observed (similar to MAZO angle), but the low end starts around 45 which is expected when you have a 50 % volume porous microstructure. This is proof that the rising trend in the MAZO angle is not limited by the slice thickness of the reconstruction If the NIP angle trend was different from that of the MAZO angle then th ere would be concern that the cathode network is limited by the z-axis slicing thickness. Thus this NIP angle just considers edges which are not perpendicular to the z-axis. These lateral network connections in the z-axis effectively increase the MAZO angle of the sample as a whole even though they are not responsible for vertical flow of charge carriers from the air/cathode surface down to the cathode/elect rolyte interface. Figure 3-1412d is the percent of edges in-z-pla ne (%IZP). T he %IZP plot indicates that roughly 20% of the total edges in a typical cathode (or pore) networ k volum e are in the Z-plane. The lack of a trend in the %IZP may be indicative of a relatively isotropic network. The %IZP indicates how many of the total network connections allow for la teral flow of air and charge carriers. This property may be important for studying the critical dimensions of current

PAGE 43

43 collectors on top of cathodes. T hus if 20% of the total surface of a cathode is laterally connected this would aid researchers se lecting current collectors of appropriate mesh size. 3.3.3 Electrochemical Impedance Spectroscopy Prior to serial sectioning, electrochemical impedance spectrometry (EIS) was performed on all six symmetric cathode sample s at various temperatures and oxygen partial pressures. The study utilized EIS to characterize the cathode performance across the various sintering conditions. A typical Nyqui st plot of the impedance scans collect ed at 700 C in air is shown in ( Figure 3-1513). The x-axis inter cept is used to calculate the cathode ohm i c resistance, once the electrolyte contribution is subtra cted out. We notice that with high sintering temperatures the polarization resistance increases dramatically All of the samples imaginary and real polarizations increase with sint ering temperature, except for at the lowest end, where the 850 C sample appears to have higher real impedance than the 900 C processed sample, thus bucking the trend. This may be a result of the poorly connected microstructure, once again isolating that sample relative to the rest. If we look at the Bode plot of the same data ( Figure 3-1614) we notice that the low sintering tem p erature sam ples have very low im aginary resistance but that all of the processes become more convoluted due to less frequency separation. J.R. Smiths dissertation47 and recent publication discusse s the process of deconvoluting the different contributions to the polarization resistance in LSM, a predominantly electronic conductor 46. As in the case of LSM, the highest and lowest frequency impedance is caused by ohmic and concentration polarization, respectivel y. In the LSM we measured two additional impedances caused by charge transfer polarization and adsorption polarization46. However, here we show the results of impedance deconvolution and relationship to microstructure for LSCF, which is an MIEC, and thus has potential additional polarization contri butions due to ionic transport11.

PAGE 44

44 In order to deconvolute the gas diffusion pro cess, low partial pressures of oxygen were used to resolve the different processes which cont ribute to concentration po larization. A plot of the 950 C sintered sample measured at 700 C with varying oxygen concentrations is shown in Figure 3-1715. Above 1% oxygen concentrat ion two significant processes are readily discernable, one high frequency a nd one interm e diate frequency. However, below 1% additional intermediate and low frequency processes appe ar. If we reduce the oxygen partial pressure further ( Figure 3-1816) we enter a regime (below 0.32%) where bulk gas diffusion polarization resis t ance becomes larger than both intermed iate frequency processes. Process 1 (whose polarization resistance is R1) is the highest frequency compon ent and is commonly understood to be the charge transfer process contribution. The intermediate processes 2 and 3 (whose polarizations are referred to as R2 and R3, re spectively) may be due to adsorption and ionic conduction (since this is an MIEC). Process 4 is dependant on pO2 and is commonly regarded as gas diffusion polarization. The process of deconvolution utilized herein is the series Voigt element approach where each process is modeled by a single resistance variable constant phase element, (R-CPE) in series as shown in Figure 3-1917. This process is de scribed by J.R. Sm ith et al. 46,47. However, it is important to note (as point ed out in our previous work 46,47) that the process is somewhat subjective, may introduce some deconvolution error, and, moreover, a nested circuit model may be more appropriate. Unfortunately, the nested a pproach is more complex and we were not able to do it with the current data set. However, the overall results remain quite useful in understanding what effect microstructure has on the different polarization contributions. The result of the deconvolution is shown in a plot of individua l process polarization resistance versus sintering temperature ( Figure 3-2018). In air (Figure 3-2018a) we see that the

PAGE 45

45 ohmic contribution has a slight negative dependence on the sintering temperature perhaps caused by the previously observed decrease in porosity. Process 1 and 2 both incr ease significantly with sintering temperature, and th e concentration process increases slightly with sintering temperature. Process 1 and 2 appear to have minima around the sintering condition near the sample with the highest TPB length ( Figure 3-86). At pO2 900ppm (Figure 3-18b), we are able to deconvolute an additional component, process 3 (R3) ( Figure 3-1816). In Figure 3-2018b we see that this additional process (R3) has a slight negative dependence on the sintering tem perature. This m ay be caused by an improvement in the path length of the cat hode phase network. The bulk gas diffusion (concentration) process has a highe r overall impedance than in ai r but seems to be independent of sintering condition, which is most likely due to the high porosities observed in these cathodes (~50 vol. %). With the reduced oxygen concen tration, there is less dependence on the microstructure because the pores have excess pathway volume. The O2 partial pressure dependence is shown in Figure 3-2119. This confirms the assignment of the lowest frequency com ponent to concentration polarization with a strong negative dependence on partial pressure. The oh mic and process 3 contributions have slight positive dependence on partial pressure, which can be attributed to the p-type conductivity of this material. Processes 1 and 2 have unclear dependencies throughout this range. The measurement temperature dependence is shown in Figure 3-2220. In air ( Figure 32220a) the ohm i c, R1 and R2 activation energies ar e -0.5, -1.7, -1.4 eV, respectively. The ohmic polarization activation energy for this 50% porous cathode is twi ce as larger as previously measured ohmic activation energy (-0.08-0.29 eV) of dense cathodes53. When the partial pressure is reduced to 900 ppm ( Figure 3-2220b) the ohmic, R1, R2, R3 and concentration

PAGE 46

46 activation energies are -0.4,-1.6,-1.6,-1.5, 0.4 eV, respectively. We see that the ohmic activation energy is decreased slightly, al ong with the R1 activation ener gy. The R2 activation energy increases at the reduced partial pressure. R3 has an activation energy in between that of R1 and R2. Concentration polarization activation energy is positive, and opposite the trend of all other processes confirming its not a thermally dependant process. 3.3.4 Microstructure-polarization Relationships Increases in sintering temperature cause th e concentration polarization resistance to increase due to the coarsening and eventu al sintering of th e microstructure (Figure 3-2018). This polarization resistance is affect ed by the effective diffusion coe fficient of the cathode. When the negative of 1-Vv / is plotted versus con centration polarization ( Figure 3-2321) a logarithmic relationsh ip is seen with concentration polarization as Vi rkar shows for other SOFC concentration polarizations 14. This (1/Vv) effective diffusion term increases with the concentration polarization which means that it is more difficult fo r oxygen molecules to arrive at the electrochemically active region of the cat hode as tortuosity increases and/or porosity decreases. Thus as the cathode is coarsened and then sintered the gas transport starts to become limited. The effect of this inverse Vv / term on concentration polari zation is not significantly affected by the measurement temperature. Similarly there is a rela tionship between the open pore phase MAZO angle and concentration polarization ( Figure 3-2422). Concentration pola rization increases with the pore phase MAZO angle. Thus as the MAZO a ngle increases the oxygen channel path becom e s longer due to the increased averag e angle, thus increasi ng bulk transport resistance. We see that as the temperature is reduced this effect is magnified, thus the MAZO angle has great implications for IT-SOFC cathode concentration polarization.

PAGE 47

47 If we compare the ohmic polarization measur ed in air at 600 and 700 C to the cathode phase MAZO angle we also see a direct relationship ( Figure 3-2523) which indicates that as the cathode phases MAZO angle increa ses so does the cathodes ohm i c re sistance. This means that as the path deviates from being parallel to th e z-axis, the ohmic polarization increases as would be expected for the increased path length. The process 3 (R3) polarization resistance de creases with increasing cathode particle diameter and cathode connectivity (Figure 3-2624). This indicates that process 3 m a y be related to ionic resistance. This ioni c resistance/conduction is dependant on the both the cross sectional area (cathode diameter) and the number of possible paths availa ble (connectivity). Thus a porous cathode with a higher connectivity and a la rger particle cross se ctional area will have lower resistance to ionic conduction. If we plot the R1 and R2 polarizations versus surface area and LTPB in air and at 900 ppm we see that both processes (highe r R values) trend better with LTPB than with surface area ( Figure 3-2725). This seems to agree with previous LSM studies 46 where both charge transfer and adsorption processes had a strong dependence on LTPB. In that study by Smith et al. the slope of the series equivalent circuit fit of RCT as a function of LTPB was -3.5 and for the fit of Rads as a function of LTPB was -2.8. Additionally Smith showed that the slope of the series equivalent circuit fit of Rads as a function of surface area was -1.8. In our study the slope of the series equivalent circuit fit fo r R1 as a function of LTPB was -13.9 in air and -10.3 at 900 ppm O2.. The slope of the series equi valent circuit fit for R2 as a function of LTPB was -10.7 in air and -11 at 900 ppm O2.. As a function of surface area the R1 slopes were -30 and -21 for the air and 900 ppm conditions, respectively. The R2 slopes, as a function of surface area, were -23.5 and -20.1 for the air and 900 ppm conditions, respectively. Although the power slopes are very high

PAGE 48

48 compared to LSM this is most likely due to th e limitation of using a series Voigt model as opposed to a nested model. The general trend in the polarization versus surface area plot is similar to previous LSM studies of adsorption versus surface area 46., but the power fit slopes in the LSCF system are much higher than that of th e LSM, indicative of increased LSCF sensitivity to changes in surface area and LTPB. In the LSCF systems LTPB dependence trend, R1 has a larger polarization resistance magnitude in air, but when the partial pressure is reduced to 900 ppm, R2s polarization resistance magnitude beco mes larger than R1s. When comparing R1 and R2 polarization resistances with surface area, R1s resi stance is larger in air, but R2 is larger at the reduced partial pressure (900 ppm). This may be a sign that R2 is more dependant on partial pressure of oxygen which may indicate th at R2 is the adsorption polarization related process. R2 is a lower frequency process than R1 (Figure 3-16) which indicates that it may be attributable to adsorption which is typically a sl ower process than activat ion polarization.. R2 has a lower residual for all plots ( Figure 3-2725a-d) which makes it difficult to conclude that R2 is only attribu t able to the adsorption (surface area de pendant) process. Likewise it is difficult to conclude that R1 is only relate d to the charge transfer proces s. With LSCF being an MIEC cathode the reduction reaction is not limited to th e TPB site but it can occur anywhere on the LSCF surface. Thus the LSCF charge transf er process may exhibit a dependence on both the LTPB and the surface area. The surf ace area and TPB length relationshi ps with R1 and R2 are not conclusive, and further work needs to be done to separate out these two processes, which appear to be heavily intermingled in the LSCF porous cathode. One method to improve on this understanding may be to utilize a nested circuit model instead of a series circuit model when deconvoluting the EIS spectra.

PAGE 49

49 3.4 Conclusion In this study we successfully reconstructed in 3-D twenty four actual LSCF cathodes. We quantified the porous cathode micr ostructures at the sub-micron level. This technique has allowed us to quantify microstructu ral properties as a function of si ntering temperature. Further, this technique has allowed us to develop the direct links between th e microstructure and performance relationship of five different polarization contributions (i ncluding Ohmic) in the LSCF/YSZ system. The concentration polari zation process was related to the effective diffusivity which is a function of volume porosity and tortuosity. An intermediate frequency process attributable to the mixed ionic elec tronic conduction in LSCF correlates well with cathode diameter and cathode t opological connectivity. Two higher frequency processes attributable to activation and adsorption polarization had simila r correlation to triple phase boundary length and volume normalized surface area. Although the power fits indicate that future nested circuit modeling is required fo r the MIEC LSCF system to fully understand the relationship between these two high frequency pro cesses and microstructure. The triple phase boundary dependence should have diminished si gnificance in an MIEC because the reduction process is not limited to the cathode/electrolyte interface. This reduc tion reaction may occur across the whole surface of the MI EC cathode. Despite this need for future study of the high frequency processes, we have de veloped direct links to microstr ucture using the series Voigt model for the ohmic, concentration and inte rmediate processes of the LSCF cathode.

PAGE 50

50 Table 3-1. Microstruc ture quantification Table 3-2. Connectiv ity quantification

PAGE 51

51 Figure 3-3. Multiple oxygen reduction/c onduction pathways in MIEC cathode. Figure 3-4. Epoxy impregnated samples increa sed the analysis speed. Platinum cap was deposited over active cathode (a), tren ch was made around ROI (b), and phase contrast between the electrolyte, po re and cathode were excellent (c).

PAGE 52

52 Figure 3-5. Six symmetric cathode samp les mounted in cross section. Figure 3-6. Six reconstructions of various sintered cathode/el ectrolyte interfaces.

PAGE 53

53 Figure 3-7. Tortuosity, porosity and the porosity to tortuosity ra tio plotted as a function of sintering temperature.

PAGE 54

54 Figure 3-8. Cathode volume normalized surface area per unit volume and triple phase boundary length per unit area plotted vers us sintering temperature.

PAGE 55

55 Figure 3-9. Open pore and cathode particle diam eters plotted versus si ntering temperature.

PAGE 56

56 Figure 3-10. Skeletonization in 2-D. Depicti on of skeleton generation of single 2-D slice of pores in black and cathode in white (a), through multiple thinning iterations (b-d), and the final skeleton (e) with a higher ma gnification view of one vertex, (f).

PAGE 57

57 Figure 3-11. Topological connectivity of cathode (open square) and open pore (solid circle) phases plotted versus si ntering temperature.

PAGE 58

58 Figure 3-12. Topological length of open pore (solid circle) and cathode (open square) phases as a function of sintering temperature.

PAGE 59

59 Figure 3-13. MAZO angle schematic

PAGE 60

60 Figure 3-14. Skeleton network orientation versus sinter temperature. Cathode phase (filled squares), open pore phase (open squares) (a) MAZO angle, (b) MAZO standard deviation, (c) NIP angle and (d) %IZP.

PAGE 61

61 Figure 3-15. Nyquist plot of LSCF polarization measured at 700 degrees C. Ohmic resistance is at x-axis intercept.

PAGE 62

62 Figure 3-16. Bode plot of imaginary impedance vers us frequency, collected in air at 700 degrees C. Figure 3-17. The partial pressure dependen ce on polarization resist ance (Bode plot).

PAGE 63

63 Figure 3-18. Bode plot of the 950 C sintered sa mple, collected at 700 C with varying low pO2. Figure 3-19. Bode plot which disp lays the deconvolution process using Voigt elements in series.

PAGE 64

64 Figure 3-20. Polarization resistance sintering te mperature dependence (a) in air and (b) at 900 ppm oxygen measured at 700 C.

PAGE 65

65 Figure 3-21. Partial oxygen pressu re dependence of various proce sses for 950 C sintered sample measured at 700 C.

PAGE 66

66 Figure 3-22. Measurement temperature dependence and activation energies (a) in air and (b) 900ppm of 950 C sintered sample.

PAGE 67

67 Figure 3-23. Concentration polarization depende nc on inverse of eff ective diffusion term at measurment temperatures.

PAGE 68

68 Figure 3-24. Concentration polarization dependence on MAZO angle of open pore phase at various measurement temperatures in air

PAGE 69

69 Figure 3-25. Ohmic polarization dependence on MAZO angle.at measurement temperatures.

PAGE 70

70 Figure 3-26. Process 3 polarization is realed to both cathode diameter and cathode connectivity. Measured at 700 C and 900 ppm oxygen.

PAGE 71

71 Figure 3-27. Relationship between R1, R2 polarization resistance and TPB length at 900ppm (a) and in air (b), and surface area at 900pp m (c) and in air (d). Fits are power.

PAGE 72

72 CHAPTER 4 THREE-DIMENSIONALRECONSTRUCTION OF COMPOSITE CATHODES 4.1 FIB/SEM Microstructure Quantification of Composite Cathode 4.1.1 Experimental An initial LCM/YSZ cathode supported sample was provided by Siemens Energy Inc., for a initial study to show the FIB/SEM 3-D analysis cap abilities. We will call this Sample 1. This sample was analyzed utilizing SEM-EDS a nd FIB/SEM techniques, in order to gauge segregation at the LCM/YSZ inte rface. After the results on this sample were finished, two additional, different cathode s upported SOFC samples were provi ded by Siemens Power, Inc. Sample 3 is the standard Siem ens doped lanthanum manganate composition (WPC3) and Sample 4 is their new cathode material, similar to Sa mple 3 but with higher levels of calcium. All three samples consisted of a com posite cathode phase approximately 20 m thick sandwiched between a coarse cathode support and dense electrolyte phase. A sample 1 cross section may be seen in Figure 4-281. A Struers EpoVac system was used to both im pregnate the open pores of the sam ples and to mount them in cross section ( Figure 4-292). Impregnation of open pores was conducted on sam ples 3 and 4 in order to allow for identification of closed pores. This resulted in better contrast over sam ple 1. Grinding/polishing was conducted down to 1 m roughness to expose the cathode-electrolyte interface and to planarize the surface. An FEI Co. Strata dual beam FIB/SEM was us ed to deposit a protective platinum layer over the area of interest, and then a C-tren ch was milled around it to reduce redeposition curtaining ( Figure 4-303). The focused (Gallium) ion beam (FIB) serial milling and SEM im aging was autom ated using Auto Slice and View software (FEI Co.) with a slice thickness of 150 nm for sample 1 and 500 nm for sample 3 and sample 4. The sample 1 slice dimensions were found to be overkill for the 2-5 m average cathode particle diameters. In order to increase

PAGE 73

73 efficiency, and maintain a resolu tion of around five slices per pa rticle the slice thickness was increased for samples 3 and 4. Furthermore the slicing orientation for sample 1 was orthogonal to the cathode/electrolyte interface, which allowed a ll three phases to be imaged in each slice ( Figure 4-281). In order to be able to quantify tortuosity the slicing direction was m a de parallel to the cat hode/electrolyte interface for samples 3 and 4 ( Figure 4-314). Am ira software was used to stack and segment the images yielding a three-dimensional (3D) model and to quantify the microstructure. Surface area and porosity were measured using Amiras Tissue statistics module. Average particle size was measured using the BET formula which stipulates that the average particle diamet er is equal to six times the volume of a phase divided by the surface area of that phase (d= 6V /S). Tortuosity was calculated using Amiras Moment of Inertia module along with the 3-D form ula for Euclidian distance. Amira was used to track the center of the open pore phase th rough the respective volume and then the 3-D Euclidean distance was calculated for all of th e pores from the air-cathode free surface down toward the electrolyte-cathode in terface. Tortuosity was measur ed as Euclidean distance sum over the thickness of the cathode. Sample 1 was sliced over 160 times at 150 nm intervals to allow for a three-dimensional reconstruction. The voxel dimensions of th e reconstruction were 118 x 153 x 150 nm. The reconstructed volume of 1250 m3 had outer dimensions of approximately 22 x 50 x 10 um. Sample 3 was auto slice and viewed as seen in Figure 4-3 with a total of 160 slices 0.5 m thick. Sample 4 was sliced into 147 similar slices. SEM imaging was conducted using a ThroughLens-Detector (TLD), which yielded excellent imaging contrast between open pore, cathode and electrolyte phases ( Figure 4-314).

PAGE 74

74 The TLD solid state detector is positioned within the SEM pole piece and was used on the deephole setting in order to get a better signal to nois e ratio by attracting more stray electrons from the imaging interaction volume surface. The TLD allowed us to utilize the low kV increased Zcontrast in combination with ch arging contrast of the low el ectronic conductiv ity electrolyte phase to distinguish between the bright, electr olyte, mid-level, cathode, and dark, epoxy/pore phases. This was an extremely important contri bution to the SOFC reconstruction field, as we were the first group to succeed distinguishing th e electrolyte and cathode phases in a composite cathode. This work was submitted to Siemens Po wer Inc. in a 2007 preliminary report. The vertical striations observed are stray ion channels, which were air-brushed or cleaned in the reconstruction process. 4.1.2 Results and Discussion Initial cross sections of Sample 1 indicated that the actual chemical interface was difficult to see using the JOEL high reso lution FEG (field emission gun) SEM, due to insufficient contrast. Due to sample geometry complications backscattered detector images were not collected, which would have also been useful in identifying the true chemical interface. Thus SEM-EDS background scan was conduc ted to identify all of the el ements present in the cross section ( Figure 4-325). The main constituents observed were yttrium zirconium, lanthanum, manganese, and calcium. A 16.7 m long linescan was then colle cted with over 10,000 total counts, to allow for suffici ent counting statistics ( Figure 4-336). It was difficult (due to electroly t e charging) to identify the cathode/ele ctrolyte interface without doing the EDS linescan. The linescan profiles ( Figure 4-347) show that all the elemen tal profiles were relatively abrupt with the ex ception of the calcium which seemed to be more segregated about the interface. This allowed us to identify the true cathode/elect rolyte interface despite the electrolyte charging artifact. The interface seems to be li ned with a collection of small pores ( Figure 4-358).

PAGE 75

75 The 150 nm thick stack of images was used to create 3-D reconstruction seen in Figure 4369 with Am ira software (Mercury Com puter Systems, Inc.). The electrolyte is shown in orange and has some small closed pores 150-1100 nm in diameter. The blue composite cathode is 20 m above the dense electrolyte ( Figure 4-3710a-c). This compos ite cathode is characterized by pores larger than those found in the electrolyte but sm aller th an those found in the cathode support region as can be seen in the graded pore structure of Figure 4-3710d. The composite cathode pore diam eter ranges between 1400-3300 nm The cathode support pore diameter range is 2400-7500 nm. To gauge the progression of the microstruc ture a phase fraction versus distance from electrolyte plot is shown in Figure 4-3811. We see that the YSZ is over 95% dense on the far left. The co mposite ca thode thickness of 20 m is further corroborated with this plot. The YSZ phase fraction starts to decrease at 11 m and drops to zero at the 33 m x axis mark. Beyond this region is the support cathode. The porosit y of the composite cathode and the cathode support has an increasing trend between 10 volum e % and 20 volume % moving away from the dense electrolyte. The low porosity was simila r to the composite cathode porosity of 12-13 volume % seen in samples 3 and 4. Sample 3 images were stacked, segmented a nd reconstructed to yi eld a total volume of 78246 m3 ( Figure 4-3912 and Table 4-31). The correspon ding volume for sam ple 4 (Figure 44013 and Table 4-31) was 66527 m3. The composite cathode was ~20 m thick for both samples, similar to sample 1. Graded porosity plot was created to visualize differences in phase volume fractions through the reconstructed volume wi th the X-axis representing distance from electro lyte and the Y-axis representing phase fraction ( Figure 4-4114). The graded porosity plot allows for

PAGE 76

76 quantification of the differences in phase fraction between Sample 3 and Sample 4. The main difference appears to be the profile of the cathode and electrolyte phases in the composite cathode. In Sample 3 the crossover is half wa y through the composite cathode region (denoted by gray arrow in Figure 4-4114a), where as in Sample 4 the cross over is close to the bulk electroly t e. The graded porosity plot also indica tes that the average elec trolyte phase fraction in the composite region of Sample 3 is 43% a nd the average cathode phase fraction of the composite region is 44%, indicative of a 48:52 volume ratio. The corresponding values for Sample 4 are 39% electrolyte and 48% cathode, a 41:59 volume ratio in the composite region. Additionally the Sample 4 graded porosity plot ap pears to have a large pore (denoted by a black arrow in Figure 4-4114) at the bulk electrolyte interf ace, which aversely affects the su rface area within th e composite region. In order to further quantify the compos ite cathode and cathode support the Sample 3 reconstructed volume was split up in to ten divisions as seen in Figure 4-4215. The divisions through the com posite cathode and the s upport cathode divisions were 5 and 10 m thick, respectively. Sample 4 was similarly divided into el even divisions ( Figure 4-4316) with an extra division m a de through the bulk electrolyte. The di visions were used to quantify volume, surface area, particle/pore average diameters, porosity two phase surface area, surface area packing factor and tortuosity, Tables 4-1 to 4-4. Cat hode, electrolyte and pore di ameters as a function of distance from electrolyte are shown in Figure 4-4215 for Sample 3 and Figure 4-4316 for Sample 4. The com pos ite cathode of Sample 3 average open pore diameter, average cathode diameter, average electrolyte diameter, and averag e closed pore diameter were 0.95, 2.91, 2.46, and 1.81 m, respectively, with an average volume porosity of 12.33 1.22 %. The cathode

PAGE 77

77 support of Sample 3 average open pore diameter average cathode diameter, average electrolyte diameter, and average closed pore diameter were 3.11, 7.50, 4.24 and 2.53 m, respectively, with an average volume porosity of 28.66 2.36 %. The composite cathode of Sample 4 average open pore diameter, average cathode diameter, aver age electrolyte diameter, and average closed pore diameter were 1.09, 4.10, 2.23 and 2.03 m, respectively, with an average volume porosity of 13.01 2.69 %. The cathode support of Sa mple 4 average open pore diameter, average cathode diameter, average electrol yte diameter, and average clos ed pore diameter were 3.18, 7.84, 2.92 and 2.54 m, respectively, with an average vol ume porosity of 28.11 4.07 %. The Sample 4 composite cathode exhibi ts a coarser cathode particle size than Sample 3 (4.10 versus 2.91 m, respectively). Sample 4 cathode support also has a slightly higher average porosity and standard deviation when compared to Sample 3 ( Table 4-31), 13.01.69, 12.33.22 % by volum e, respectiv ely. This is evident in the graded porosity profiles ( Figure 4-4114b) and most likely caused by the presence of the aforem entione d large interfacial pore. The cathode support particle d iameters for the two samples are all very similar with the exception of average electrolyte particle diameter, but this is most likely due to the small number of impurity particles seen in the cathode support, thus no t allowing for good comparative statistics The coarser composite cathode particle size of Sample 4 causes the cathode and electrolyte surface areas of Sample 4 to be sm aller than those of Sample 3. Plots of the volume normalized surface area can be seen in Figure 4-4417a for Sample 3 and Figure 4-4417b for Sample 4. The plot in cludes surface area m easurements th rough each volume for all four phases. Table 4-42 has the valu es for the no rmalized surface area. The composite cathode surface area of Sample 3 has a volume normalized surface area of 10.95 m-1. It has an average open pore surface area, average cathode surface area, average electrolyte surface area, and average closed pore surface

PAGE 78

78 area of 0.78, 0.96, 0.96, and 0.04 m-1, respectively. The cathode support surface area of Sample 3 has a volume normalized surface area of 5.84 m-1. It has an average open pore surface area, average cathode surface area, average electrolyte surface area, and average closed pore surface area of 0.56, 0.56, 0.04, and 0.01 m-1, respectively. The composite cathode surface area of Sample 4 has a volume normalized surface area of 9.38 m-1. It has an average open pore surface area, average cathode surface area, average electrolyte surface area, and average closed pore surface area of 0.72, 0.91, 0.67, and 0.04 m-1, respectively. The cathode support surface area of Sample 4 has a volume normalized surface area of 5.60 m-1. It has an average open pore surface area, average cathode surface area, average electrolyte surface area, and average closed pore surface area of 0.54, 0.54, 0.03, and 0.01 m-1, respectively. The biggest difference between the two samples lies in the cathode a nd electrolyte surface areas of the composite cathode which may be due to the aforementioned volume phase fraction differences and the finer nature of Sample 3s microstructure. The higher total volume surface area in Sample 3s composite cathode agrees with particle size tren ds. The cathode supports have very similar surface areas for all phases. Further quantification of surface area was conducted through the measurement of two phase surface area between all four phases: open pore, cathode, electrolyte, closed pore. Figure 4-4518 shows the individual values while Table 4-53 sums them for the composite cathode and cathode support regions. These values show the relative dom inance of the various two phase boundaries in the m icrostructure. The important side note in Table 4-53is that the balance (9% of total for S a mple 3 and 14% for Sample 4) of the surface area fraction is found in the electrolyte bulk which was excluded from the calculations.

PAGE 79

79 The composite cathode of Sample 3 is dominated by the cathode-open pore, electrolyteopen pore and electrolyte-cathode two phase inte rfaces with sums of 0.13, 0.13, and 0.19, respectively. The cathode support of Sample 3 is dominated by the cathode-open pore surface area with a fraction sum of 0.42. The compos ite cathode of Sample 4 is dominated by the cathode-open pore, electrolyte-open pore and elec trolyte-cathode two phase interfaces with average fractions of 0.19, 0.10, and 0.17, respectively. The cathode support of Sample 4 is dominated by the cathode-open por e surface area with a fraction su m of 0.42. A comparison of Sample 3 and Sample 4 two phase surface area shows that Sample 3 has equal cathode-open pore and electrolyte-open pore surface area fractions (0.13 and 0.13) which are both less than the electrolyte-cathode fraction (0.19). But in Sample 4 we see that the dominant surface area fractions are cathode-open pore a nd electrolyte-cathode (0.19 a nd 0.17) as compared to the electrolyte-open pore surface area fraction (0.10). This difference is possibly due to the lower volume fraction of electrolyte in the composite cathode of Sample 4. Both composite cathodes have an equal fraction of total surface area (0.47). This is indicative of the high concentration of surface area present in the composite cathode when compared to the bulk electrolyte and cathode support. For comparison of surface area between the tw o samples a surface area packing factor was devised, to compare the relative concentration of su rface area through the rec onstructions. This consists of dividing the surface area of each divi sion by the total reconstruction surface area, and dividing that quantity by the volume of each sli ce divided by the total reconstruction volume. These plots are show in Figure 4-4619. The main difference th at we notice is that Sample 3 has a higher m axim um surface area packing factor whic h abruptly drops approaching the electrolyte interface in the composite cathode. This indicates that Sample 3 has a more abrupt interface than

PAGE 80

80 Sample 4 (as corroborated by Figure 4-4114) perhaps indicative of weaker interfacial adhesion between the com posite cathode and the bulk electro lyte. Although Sa mple 4 has a large pore at the dense electrolyte interface whic h would give weaker adhesion. Triple phase boundary (TPB) length was also ca lculated by summing all of the points in each composite cathode where the cathode, electroly te, and open pore phases intersected in a triple point ( Figure 4-4720). All of these points were tu rned into segm ents through each slice, and the total length was divided by the total com posite cathode reconstructed volum e to yield the TPBl of 0.037 m/ m3 for Sample 3, and 0.046 m/ m3 for Sample 4. These values are comparable to each other and about an order of magnitude less than the LSCF/YSZ studies from Chapter 3. The reason for the increased TPBl in Sample 4 is most likely due to the increased interfacial porosity as observed in the graded porosity ( Figure 4-4114). Tortuosity was also m e asured for the two samples ( Figure 4-4821 and Table 4-64). Sa m ple 3s tortuosity in the composite cathode is 1.58, and 1.40 in the cathode support. These are higher than the tort uosities of Sample 4s composite ca thode and cathode support of 1.44 and 1.31, respectively. Figure 4-4821 shows a plot of tortuos ity versus distance from de nse electrolyte/cathode interface. The overall tortuosit y is the y intercept value. The tortuosity is calculated by starting at the cat hode/air interface and tracking the inertial center of the open pore phase in each reconstructed sl ice as we progress through the cat hode support, then through the composite cathode and stop at the dense electrolyte. The Euclid ian distance is divided by the cathode thickness to calculate the tortuosity. This plot shows how that value updates with every slice, starting at 70 m and moving towards 0 m. It is evident that it takes at least ten slices for the tortuosity value to stabilize to wards the right end of the plot b ecause it is with respect to the initial slice. This is indicative of the length s cale of the pores in the cathode support, which are

PAGE 81

81 typically less than five m in diameter. It is also evident that the tortuosity profile is more torid in the cathode supports coarser microstruc ture than the composite cathodes finer microstructure. A more torid path is more diffi cult for gas to diffuse through then a less torid path. This difference in roughness between the support cathode and co mposite cathode exists because each support cathode pore has a larger eff ect on the overall tortuosity than a smaller pore within the composite cathode. This account s for the higher roughness in the support cathode region of the graded tortuosity plot. In additi on the slice thickness of 500 nm more effectively captures the larger cathode support pores, which al so contributes to the higher apparent graded tortuosity plot roughness. The slope of the graded tortuosity gives us insight about the uniformity of the pore channels as the oxygen molecules traverse th e cathode support on through the composite cathode. In the composite cathode region, where the reduction reaction o ccurs in the proximity of triple phase boundaries, Figure 4-4821 shows that the graded porosity slope is 68% greater in Sa m ple 4s composite cathode than in Sample 3 s (slope = -0.007). Th is indicates that the tortuosity has a faster rate of increase through the composite cathode in Sample 4 than in Sample 3. But Sample 4 has lower overall tortuosit y for an oxygen molecule traveling from the cathode/air interface towards the bulk electrolyte. Although no electrochemical data is present to fully characterize the effect that this slope di fference has on electrochemi cal polarization, it is plausible that the gas phase diffusion resistance (concentration polarization) into and out of the composite cathode of Sample 4 is lower than for Sample 3. This increased slope combined with the lower magnitude may also be a side effect of having a low porosity bottleneck half way through the composite cathode in Sample 4. If we look in Figure 4-4114 we can see that Sample 3 has a composite cathode porosity with lower standard deviation ( 12.3 1.2 % by volum e),

PAGE 82

82 where as Sample 4 has a higher composite catho de porosity, and a highe r standard deviation (13.0 2.7 % by volume). The increased standard deviation in Sample 4 is a result of low (sub 10 % by volume) porosity in the middle of th e composite cathode be ing sandwiched by high porosity a the electrolyte in terface (approaching 20 % by volume) and high porosity at the composite cathode, support cathode interface (>15 % by volume). Thus it is this bottleneck present in the Sample 4 composite cathode which l eads to a sharp increase in the tortuosity slope while maintaining a lower tortuosity magnitude. 4.1.3 Microstructure Summary The FIB/SEM serially sectioned Sample1 reco nstruction shows the interconnected nature of the composite and support cathode pore networks. The cathode support has large, interconnected pores. The composite cathode has smaller, interconnected pores, and interconnected YSZ. The electrolyte has closed small pores in comparison to the cathode regions. In addition, the composite cathode is observed to be about 20 m thick. This initial work helped isolate critical mill dimensions, orie ntation charging effects, and created a need to quantify the interconnectivity of these networks, which was done in Samples 3 and 4. Cathode supports of both Sample 3 and Sa mple 4 had very sim ilar microstructural characteristics, when compared to one anot her. Cathode support surface area and average particle/pore particle sizes were relatively e qual. An impurity phase with similar imaging contrast to the electrolyte phase was seen but not chemically id entified. Composite cathode of Sample 3 had a 48:52 percent by volume electrolyte to cathode ratio, while Sample 4 had a 41:59 percent by volume electrolyte to cathode ratio. The coarser Sample 4 composite cathode microstructure had lower volume normalized surface area. Sample 4 had lower tortuosity throughout the cathode most likely cau sed by coarser microstructure. Sample 4 also exhibited a more drastic tortuosity increase through the composite cathode.

PAGE 83

83 4.2 TEM-EDS Characterization of LCM/ YSZ Interface in Composite Cathode 4.2.1 Experimental The electrolyte/cathode inte rface of both samples was located and a TEM foil was milled with the FIB. In situ lift out was conducted with a micromanipulator (Omniprobe, Inc.), ( Figure 4-4922) and the foils were attached to copper alloy Om niprobe grids,( Figure 4-5023 and Figure 4-5124). SEM im ages of the m ounted foils show th at the epoxy has penetr ated the smallest of open pores at the bulk electrolyte interface as pointed out with the white arrow in Figure 4-5023. The foils were thinned down to a thickness of ~100 nm to allow electron transm ission. Chemical analysis was conducted with a m odel 2010 JOEL high resolution TEM equipped with an Oxford Scientific Instruments EDS detector. 4.2.2 Results and Discussion An interface between the cathode and electr olyte phases in the composite cathode was located in Sample 3 ( Figure 4-5225). A sum spectrum was colle cted to identify all the elem ents present in th e area ( Figure 4-5326). The copper and gold peaks are artifacts from the TEM grid. An EDS line scan was co llected to visualiz e the element profiles across the interface ( Figure 45427). Manganese and lanthanum are the dom inan t EDS signal species in the cathode, and zirconium is the dominant species in the electrolyte. Calcium is present on both sides of the interface. A series of point scans were then collected across the same interface in order to quantify the weight percent elemental composition of each phase ( Figure 4-5528). Both m a nganese and calcium seem to have appreciably diffused across the interface. A similar interface between the cathode and electrolyte phases in the composite cathode was located in Sample 4 (Figure 4-5629). The sum spectrum was collected to identify all the elem ents present in the area ( Figure 4-5730). The copper and gold peaks are artifacts from the TEM grid. An EDS line scan was co llected to visualize the element pr ofiles across the interface

PAGE 84

84 ( Figure 4-5831). Manganese and lanthanum are the dom inant EDS signal species in the cathode, and zirconiu m is the dominant species in the elec trolyte. Calcium is present on both sides of the interface. A series of point scans were then collected across the same interface in order to quantify the weight percent elemental composition of each phase ( Figure 4-5932). Similar to Sam p le 3 manganese and calcium seem to have diffused across the interface. The Sample 3 electrolyte phase ( Figure 4-6033a) has a higher average lanthanum and oxygen content, while S ample 4 has higher calc ium, manganese and zirconium content. When comparing the cathode phase chemical composition ( Figure 4-6033b) it is evident that Sam ple 3 has a higher average lanthanum and oxygen content, which correlates to the same differences in the electrolyte phase. Sample 4 has higher average composition for all other elements, which is in agreemen t with the electrolyte phase ( Table 4-75). A closer lo ok into th e calcium content of both samples (Figure 4-6134) reveals that Sample 3 has a lower calcium concentration in both electrol yte and cathode phases by 1 and 1.4 percent, respectively. 4.2.3 TEM Summary TEM line and point scans were used to quantify chemical composition across the cathode/electrolyte interface for both samples. Th e most substantial difference between the two samples was the calcium content which may explain the coarser microstructure seen in the Sample 4 composite cathode because calcium is known to act as a sintering aid. 4.3 Topological Connectivity 4.3.1 Introduction Topological connectivity is a volume indepe ndent intrinsic property which affects transport in any composite or porous material. Understanding the connectiv ity of a material is essential in understanding transport in three dimensions (3-D). In this work we have quantified in 3-D the connectivity of two separately pro cessed composite solid oxide fuel cell (SOFC)

PAGE 85

85 cathodes. This microstructural pr operty is related to all of the transport processes which occur within the fuel cell. For ex ample, cathode connectivity affects ion/electron conductivity. The connectivity also plays a role in the transfer of mechanical stre sses associated with thermal and or mechanical loads. While previous rese archers have quantified connectivity in 2-D 14 this is the first work in which 3-D topologi cal connectivity quan tification may be re lated to quantified microstructural properties. Ot her researchers have quantified volume connectivity percentages 54 in cathodes, but this is a volume sensitive measur ement which is sample size dependent. Some have been able to quantify 3-D characteristic length scales 55, but none have been able to quantify the interconnected nature of the topology. A solid oxide fuel cell (SOFC) efficiently converts fuel and air into electricity and heat 6. In order to increase performance (by decreasing polarization resistance) the electrochemically active triple phase boundaries (T PBs) of SOFC cathodes are extended by mixing the primarily electron conducting cathode phase and the primarily ion conducting electrolyte phases in a composite cathode. Our previous studies have reconstructed SOFC electrodes 46, 48. using the dual beam focused-ion-beam/scanning-electron-m icroscopy (FIB/SEM) approach, but were not able to gauge the topological connectivity of the complex 3-D network. In addition to characterizing microstructural properties, such as volume porosity, particle size, TPB length, surface area and tortuosity, the mean degree or connectivity of the electrochemically active composite cathode has been quantified for the first time. The quantification of topological connectivity paired with 3-D microstructural qua ntification has far reaching impacts on fields beyond SOFCs such as tissue scaffolds, vascular sy stems, particulate filters, structural ceramics, and oil field exploration just to name a few. Quantifying the connectivity allows for intelligent engineering and modeling of complex 3-D networks.

PAGE 86

86 4.3.2 Results and Discussion 4.3.2.1 Connectivity quantification Thus far, a variety of SOFC microstructural parameters have been quantified with the aid of three-dimensional FIB/SEM reconstructions, such as volume porosity, particle size, surface area, TPB length, and tortuosity 39, 48 Here, we present the microstructural parameter of topologic/structural 3-D connect ivity, which has for the first time been quantified for a manmade 3-D material network. Connectivity has b een quantified in two dimensions by various 2-D methods 56, 57 The advent of 3-D reconstruction soft ware paired with confocal microscopy allowed Wagner and coworkers 42 to quantify corrosion cast, rabbit kidney capillaries by visualizing and counting the number of edges and vertices of a single specimen in 2006. More recently Perna and coworkers utilized x-ray tomography to reconstruct termite nests in order to quantify the 3-D networks topological efficiency 58. Such a topological connectivity is an intrinsic microstruc tural property which may be related to all of the transport processes which occur w ithin the fuel cell. For example, cathode connectivity affects ion/electron co nductivity, pore connec tivity affects gas transport to and from TPB boundaries, and electrolyte connectivity affects ion conductivity. In addition, the connectivity of both the cathode and el ectrolyte play a role in the tr ansfer of mechanical stresses associated with thermal expansion/co ntraction during thermal cycling. The SOFC cathode connectivity was quantif ied with the Skeletonize module of Amira software, which examines each phase and through a number of iterations, p eels the outside layer off until all that exists is the mass centroi d, or skeleton of that phase in 3-D ( Figure 4-6235). The sam e m ethod is show in 2-D for Figure 4-6336. This 2-D skeleton ization represen tation allows us to visua l ize vertices with two, three, and f our edges. After the skeleton network has been created in 3-D we use the chamfer map tool to calculate the distances between the nodes where

PAGE 87

87 more than one branch intersect. This 3-D netw ork for a composite cathode may subsequently be visualized. A schematic of the progression from SEM images, to a 3-D reconstruction and the final 3-D skeleton is shown in Figure 4-6437. The Amira skeletoni zation by blocks for large 3-D datasets m e thod was developed in 2004 by Fouard et al. 59. It was first utilized to study Indian ink injected human brain microvascularization st ructures imaged with a confocal microscope. Generating the skeleton block by block ensures that homotopy, thinness and medialness are all maintained. Homotopy requires that the skeleton be topologically identical to the original 3-D structure. Thinness ensures that the skeleton is one point wide at all points except at junctions where it may be thicker. This thinness is adjusted in our work because we have a finite thickness to the skeleton which is scaled to the original st ructures thickness. Medialness requires that the skeleton be centrally located (mass centroid) within the reconstructed object 59. The skeleton of each phase consists of a series of vertices ( ) or nodes, which are connected to one another via edges, E, of vari ous topological length, L. Within each network, there is a node with the most edges. This is the kmax value for the network. It is valuable to know the highest degree of nodal connec tivity present within a 3-D structure when comparing different types of systems (biological versus inorgani c), because there may be a basic structural differences between nature grown biological system s and man made systems. This type of 3-D complex network quantification has prev iously been conducted on termite mounds 58 and that method is expanded below. The degree of a single node is defined by: iiiNEk / (4-1) Where Ei and Ni are local quantities. When all of the nodes in the network are analyzed, the mean degree may be defined as:

PAGE 88

88 Nkki/ (4-2) Where N is the total number of vertices for the whole network. The av erage topological path length may likewise be calculated by summing the e dge distances (d) for the whole volume (V): Vvv ijjid N L,1 (4-3) The 3-D reconstruction for each sample is first split into equal volumes of composite cathode, followed by a series of thin ning Skeletonizati on iterations ( Figure 4-6235, Figure 46336) to yield a view of the enhanced skeletons of m ass centroids of all three com posite cathode phases. Then each phase is evaluated for the various network topology parameters, as can be seen in Table 4-86. Sa m ple 3 has more nodes and edges in both the composite and support cathode regions. This is indicative of Sample 4 having coarser mi crostructure than Sample 3. Sample 3 has an overall lower mean degree (or connectivity) for vi rtually all phases, thus having lower overall connectivity than Sample 4. The kmax values do not follow a defi nite trend. However, the maximum connectivity in both samples was 11. Sample 4 has a longe r average topological length, which supports the coarser microstructure as observed in th e particle size, surface area and tortuosity values. The connectivity trends indicate that the sample with higher calcium doping has increased composite cathode connectivity. 4.3.2.2 Measured average z-orientation In addition to network topology parameters su ch as connectivity we have implemented a python based program to calculate the Measured Average Z-Orientation (MAZO) of all of the edges in each specific skeleton. MAZO is the an gle with respect to the z-axis between endpoints of edges (node to node) ( Figure 4-6538). This property allows f o r the quantification of how well

PAGE 89

89 aligned the available transport paths are with the Z-axis which is or thogonal to the porous composite cathode/dense electrolyte interface. T hus a MAZO angle of 90 would mean that the edge is parallel to the cathode /electrolyte interface, where a MAZO angle of zero degrees would indicate that the edge is orthogona l to the interface. This qu antification is useful in better understanding the transport path for oxygen from the air/cathode interface down to the electrolyte. The angular orie ntation of the gas phase affect s the path efficiency for oxygen transport. Similarly solid phase angular or ientation helps us bette r understand the conduction path of charge carriers through the solid phases. Thus the MAZO angle will be helpful in understanding latera l current and gas flows. From Table 4-86 we can see that Sample 3 and 4 have identical com posite cathode MAZO valu es of 55 degrees, indicative of sim ilar orientation. The distribution of angular orientations are shown in Figure 4-6639 for the com posite pore networks of Sa mple 3 (blue) and Sample 4 (yellow). The Sample 3 MAZO distribution is Gaussian, while the Sample 4 MA ZO distribution has a no n-Gaussian profile as shown by excess frequency of angles in the 80 degr ee range. As a result the Sample 4 composite pore angle is higher than that of Sample 3 ( Table 4-86). This may be caused by increased sintering, and the aforementione d large interfacial porosity obse rved in Sa m ple 4s graded porosity ( Figure 4-4114). The MAZO distribution (Figure 4-39) m a y also be an indicator of gradient uniformity, with a Gaussian MAZO angle distribution being indicative of a steady gradient. Such a steady gradient is observed in the Sample 3 graded porosity profile, as we transition from cathode to electrolyt e in a smooth fashion. In the case of Sample 4 the gradient is much more abrupt, and this seems to be mimicked in the skewed MAZO angle distribution. The broad distribution of MAZO angles with the larg e standard deviation is typical of isotropic microstructures. As the topological network characteristic leng th (L) increases so does the

PAGE 90

90 MAZO angle indicating that the la rger particle stacki ng in the composite cathode is further apart in the x-y plane than that of the finer particle network. 4.4 Conclusions Microstructures of three Siemens-Westinghouse cathode-supported solid oxide fuel cell (SOFC) samples were analyzed. The cathode/electro lyte interface was characterized with Transmission Electron Microscope-Energy Dispersive Spectro scopy (TEM-EDS). Sample 3 was the initial study which aided in identification of the electrolyte/cathode in terface, and the estimation of proper slice and view dimensions. The othe r two samples were designated by Siemens as Sample 3 and Sample 4. Sample 4 had higher calcium content in both electrolyte and cathode phases than Sample 3. This calcium difference may be the cause of the coarser microstructure seen in Sample 4 because calcium is known to act as a sintering aid 60. The cathode microstructure was quantified with the ai d of a Focused Ion Beam/Scanning Electron Microscope (FIB/SEM). All three samples cons isted of a composite cathode on the order of 20 m, sandwiched between the porous cathode support and dense electrolyte. Surface area, porosity, particle size, tortuosit y and, for the first time, 3-D topological connec tivity were all quantified with the aid of Amira software. Sample 4 had a coarser composite cathode microstructure with 2.5 % higher composite cathode connectivity. This was the first time that true 3-D topological connectivity was quantified for an SOFC structure. In the future, connectivity may be related to all transport processes that occur throughout the SOFC. This advancement in connectivity quantification can further the understanding of many other fields including biol ogical systems connectivity, for example, by aiding researchers in better understanding the 3-D lattices used in synthetic tissue harvesting. Moreover, the MAZO quantification technique helps in unders tanding 3-D load distribution in structural

PAGE 91

91 materials. The combined use of FIB/SEM and skeletonization technology allows for more intelligent design of porous micro and nano functional materials.

PAGE 92

92 Table 4-3. Particle size Table 4-4. Surface area

PAGE 93

93 Table 4-5. Two phase surface area Table 4-6. Tortuosity

PAGE 94

94 Table 4-7. TEM-EDS point scan quantification Table 4-8. Connectiv ity quantification

PAGE 95

95 Figure 4-28. FIB/SEM cross sec tion of cathode supported SOFC sa mple imaged with charge contrast TLD detector. Figure 4-29. Epoxy sample preparation. (a) EpoVac chamber used for impregnation. (b) Planarized/polished samples mounted in cross section

PAGE 96

96 Figure 4-30. C-trench preparation. (a) Initial polished sample 3 cross section is (b) deposited with platinum layer. (c) C-trench is made and then the (d) cat hode is automatically milled and imaged. Figure 4-31. Phase contrast. Sample slices of sample 3 microstructure: (a) Support cathode with arrow pointing to suspected electrolyte impurity (b) composite cathode and (c) electrolyte. Redeposition cu rtaining is seen in the bo ttom portion of (b) and (c).

PAGE 97

97 Figure 4-32. Background EDS scan of Siemens-We stinghouse electrolyte/cathode interface cross section. Figure 4-33. 16.7 m long line scan across cat hode/electrolyte interface.

PAGE 98

98 Figure 4-34. EDS selected element profiles of linescan displayed in Figure 4-3. The numbers next to the element name indicates the Y axis maximum value in counts. Notice the broad nature of the blue Calcium peak in contrast to all other elements. Gallium profile is flat because the element is em bedded across whole profile evenly from FIB/SEM milling.

PAGE 99

99 Figure 4-35. FEG SEM SE detector image of cross section region scanned for background EDS scan of Figure 4-1. There is a disparity between apparent interface as imaged by SED detector and true atomic interface revealed by EDS analysis.

PAGE 100

100 Figure 4-36. Reconstructed volume of Siemen s SOFC electrolyte, composite cathode, and cathode support.

PAGE 101

101 Figure 4-37. (a) Reconstructed vo lume of YSZ electrolyte including YSZ in composite cathode; (b) reconstructed volume of YSZ and LSM cat hode; (c) transparen t reconstruction of electrolyte and electrode with pore network visible within the LSM cathode; (d) pore network of entire reconstructed volume.

PAGE 102

102 Figure 4-38. Phase fraction (YSZ, LSM, Pore) in y axis is plotted against the distance away from the LSM/YSZ interface. A transparent image of the microstructure shows the orange dense electrolyte, 20 m thick cathode, and cathode support.

PAGE 103

103 Figure 4-39. Sample 3 3-D Reconstruction. La beled with phases and approximate dimensions. Orange arrow points to phase with similar imaging contrast to the electrolyte phase.

PAGE 104

104 Figure 4-40. Sample 4 3-D Reconstruction. La beled with phases and approximate dimensions. Orange arrow points to phase with similar imaging contrast to the electrolyte phase.

PAGE 105

105 Figure 4-41. Graded porosity plots of (a) sample 3 and (b) sample 4. Gray arrow shows crossover of cathode and electrolyte phases in sample 3. Black arrow denotes large interfacial pore in sample 4.

PAGE 106

106 Figure 4-42. Sample 3 particle size. Image of sample 3 divisions used for microstructural quantification (a). Particle size versus distance from electrolyte chart for sample 3 (b).

PAGE 107

107 Figure 4-43. Sample 4 particle size. Image of sample 4 divisions used for microstructural quantification (a). Particle size versus distance from electrolyte chart for sample 4 (b).

PAGE 108

108 Figure 4-44. Volume normalized surface area measurements for sample 3 and sample 4.

PAGE 109

109 Figure 4-45. Two phase surface area as fraction of total surface area for sample 3 (a) and sample 4 (b).

PAGE 110

110 Figure 4-46. Surface area packing factor. Comp arison between sample 3 (open diamonds) and sample 4 (closed squares) Figure 4-47. Triple phase boundaries in 2-D image of the Siemens Sample 3 composite cathode

PAGE 111

111 Figure 4-48. Comparative graded tortuosities of sample 3 (open square) and sample 4 (closed diamond).

PAGE 112

112 Figure 4-49. TEM foil preparation. Sample 3' s cathode/electrolyte inte rface (a) was deposited with platinum protective layer (b). TEM cross section was milled (c) and lifted out with a micro-manipulator (d). Figure 4-50. TEM sample 3. Back, top and front views of 100 nm thick sample 3 TEM foil mounted on copper grid. Epoxy filled pore at electrolyte interface (white arrow).

PAGE 113

113 Figure 4-51. TEM sample 4. Back, top and front views of 100 nm thick sample 3 TEM foil mounted on copper grid. Figure 4-52. Sample 3 SEM image and corre sponding TEM image of composite cathode.

PAGE 114

114 Figure 4-53. Sample 3 sum EDS spectr um of cathode/elect rolyte interface.

PAGE 115

115 Figure 4-54. TEM image and corr esponding line scan profiles across the cathode/electrolyte interface of sample 3.

PAGE 116

116 Figure 4-55. Sample 3 TEM EDS point scan lo cations (a) and elemental weight percent quantities (b). Figure 4-56. Sample 4 SEM image and corre sponding TEM image of composite cathode.

PAGE 117

117 Figure 4-57. Sample 4 sum EDS spectr um of cathode/elect rolyte interface.

PAGE 118

118 Figure 4-58. TEM image and corr esponding line scan profiles across the cathode/electrolyte interface of sample 4.

PAGE 119

119 Figure 4-59. Sample 4 TEM EDS point scan lo cations (a) and elemental weight percent quantities (b).

PAGE 120

120 Figure 4-60. TEM-EDS elemental co mposition. Point scan elemen tal comparison of electrolyte phase for sample 3 and sample 4 (a). Point scan elemental comparison of cathode phase for sample 3 and sample 4 (b).

PAGE 121

121 Figure 4-61. Calcium composition comparison be tween electrolyte and cathode phases for sample 3, and sample 4.

PAGE 122

122 Figure 4-62. 3-D Skeletonization. Depiction of skeleton genera tion from original volume (a), through multiple thinning iterations (b-d), and the final skeleton with a higher magnification view of one vertex, (f). Figure 4-63. Skeletonization in 2-D. Depicti on of skeleton generation of single 2-D slice of pores in black and cathode in white (a), through multiple thinning iterations (b-d), and the final skeleton (e) with a higher ma gnification view of one vertex, (f).

PAGE 123

123 Figure 4-64. Schematic showing the transition from SEM image (top left), a stack of which is used to generate a 3-D reconstruction mode l (top right), which is then reduced to a mass centroid skeleton (bottom center) of each phase. Figure 4-65. Schematic deptiction of Measured Average Z-Orientation (MAZO) angle.

PAGE 124

124 Figure 4-66. MAZO angle frequency distributi on for sample 3 and sample 4 composite pore networks.

PAGE 125

125 CHAPTER 5 LOCAL ELECTRON ATOM PROBE RE C ONS TRUCTION OF BURRIED CATHODE/ELECTROLYTE SOLID OXI DE FUEL CELL INTERFACES 5.1 Introduction Solid oxide fuel cells (SOFCs) offer society a means of increasing our energy independence through increased en ergy conversion efficiency. The technical advantage of SOFCs is the ability to convert both liquid and gaseous fuels into electrical power and useful heat through electrochemical conversion. The main barrier to SOFC commercialization is the cost/power ratio, the reduction of which is a main concern for both commercial and institutional researchers. As researchers seek to reduce cost and prol ong stability they have made changes to the electrode/electrolyte compositions and structures. These compositional changes sometimes have deleterious effects on electrode/el ectrolyte interfaces. Interfacia l stability evaluation is often limited to long term durability studies where sufficient amounts of tertiary phases must grow to be measured by scanning electron microscopy (SEM) energy dispersive spectroscopy (EDS) or transmission electron microscopy (TEM) EDS methods. The long term stability effect of compositional changes of the cathode/electrolyte in terface may now be evaluated faster than ever before thanks to the recent breakthrough of three-dimensional atom probe (3DAP) microscopy. 3DAP allows for the 3-D visualization and analysis of chemically resolved atoms. The small radius of curvature ( ~ 25 nm) at the top of the sample tip allows for a magnified field ionization 61, which with the aid of a pico second pulsed laser ( Figure 5-671) 62 evaporates the tip surface monolayer in the direction of the local electrode 63, 64. The laser assisted thermalexcitation evaporation mechanism works by increasi ng the local tip temperature which allows for low-conductivity materials to be evaporated, sinc e field evaporation is directly proportional to temperature 62,65,. It is important to note that the sample temperature should not be allowed to

PAGE 126

126 exceed 200K, because this is the temperatur e below which surface species migration is essentially frozen 63. To prevent this the chamber temp erature is kept well below 100 K for routine experiments. This allows for an accurate 3-D atomic reconstruction. Once the ions pass the local/counter electrode, having attained energy equal to the tip voltage, they enter a field neutral drift region, where given enough distance (~ 1 meter) they separate according to their atomic mass and then collide with a solid state X-Y spatially sensitive detector 66 which determines where on the tip the atoms originated. Due to the detectors geometry approximately 50 percent of the atoms are det ected and counted. As subseque nt pulses evaporate subsequent atomic layers from the probe tip a 3-D image of the probe tip may be reconstructed with corrections applied to adjust for the tip curvat ure. The time-of-flight (TOF) between the pico second laser pulses allows for 10,000 nano-second wide mass-to-charge (M/C) window within each laser pulse which is able to resolve individual atoms, their isotopes and clusters or complex atoms/ions. The formula which is utilized to calculate the M/C is based on the relationship between potential energy and kine tic energy (Equation 5-1), where v=d/t (d is distance in meters, t is time in seconds), m is the mass of the ion, n is the number of charge carriers, e is the charge of one electron, and V is the applied voltage. neVmv 2 2 1 (5-1) The time it takes for the ion to travel the 90 mm from the local electrode to the detector is shown in Equation 5-2, where m is in units of kilograms, d in units of meters, and the voltage in units of volts. neV md t 22 (5-2)

PAGE 127

127 Equation 5-3 is useful when trying to deal with M/C peak drift which is caused by the increasing voltage required to evaporate an atom probe tip with a growi ng radius of curvature (caused by the evaporation of each successive m onolayer off of the cone shaped tip). To calculate the actual M/C value or m/n value we would rearrange to yield Equation 5-3, which is in units of atomic mass units (amu) or Daltons (Da). The conversion from a kilogram based mass to the amu requires the use of Avogadros number, AN. 22000 v eVN n mA (5-3) When a single atom completes its time of flight and hits the X-Y det ector, both a unique X and Y coordinate may be identified, and this ev ent it is called a golden hit. When there are clusters of atoms that are ioni zed together, they may hit the detector in two adjacent spots, preventing a unique X and Y position to be detected subsequently detracting from the golden hit percentage. These composite ion evaporations are h eavier than the single ions and thus will be at higher M/C values due to longer flight times. 5.2 Experimental Procedure Two separately processed Siemens Energy, In c. samples were cross-sectionally mounted and impregnated with epoxy using a Struers EpoV ac chamber. The samples were designated as Sample 3 and Sample 4, for which 3-D FIB/SEM microstructure quantific ation is discussed in Chapter 4. They were subse quently ground and polished to a 1 m roughness to planarize the surface. Three layers of carbon coating were sputter deposited to prevent charging. The following technique is an adapted version of th e site specific atom pr obe preparation methods discussed by Miller 67, Thompson 68, Gorman 69 and Moore 70. Using an FEI Co. Strata DB235 dual-beam focused ion beam / scanning electron microscope (FIB/SEM) equipped with a metalorganic gas injection system, pla tinum was deposited in a 50 X 2 X 2 m pattern over the

PAGE 128

128 composite cathode region, parallel to th e electrolyte/composite cathode interface ( Figure 5682a). Trenches were m a de on either side of the platinum strip, followed by cleaning cuts to reveal a 50 m wide by 2 m thick foil ( Figure 5-682b-d). The Through-Lens-Detector (TLD) was utilized in scanning mode to differentiat e between the electrolyte, cathode and open pore (epoxy filled) phases ( Figure 5-682e). Subsequent 2 m wide fiduciary m arkings were made in the protective platinum cap to identify over a dozen foil regions with cathode/electrolyte interfaces ( Figure 5-682f-g). The 2 m f iduciary thickness was required for the samples to fit on the leap array posts, which is described below. The foil was then undercut at a 22 tilt, then rotated 180, and undercut agai n at the same 22 tilt ( Figure 5-682h). This re sulted in a cham fer on the bottom of the foil, which is a common prac tice to ensure good contact between the sample and the leap array post, once mounted, as discussed below. An Omniprobe, Inc. AutoProbeTM 200 micromanipulator was then made to touch the foil. A three step platinum weld was made on the top side of the sample, which allowed for the whole 50 m long by 7.5 m tall sample foil to be in-situ lifted up and out of the bulk sample ( Figure 5-693). The first step in this m i cromanipulator welding process was the boring of a small X m square hole into the region where the microman ipulator and foil were in contact using a 50 picoamp (pA) current aperture. The next step was to backfill that square hole with platinum using a 10 pA aperture. This reduced current a llows for a more dense platinum deposition, thus increasing the adhesion between the micromanipulat or and sample. This increased adhesion is necessary when dealing with such a large foil (relative to microman ipulator contact area), which is over 15 times longer than the micromanipulator c ontact area is thick. The sample was then cut free and the micromanipulator was used to lift it up and out of the bulk sample ( Figure 5-682i).

PAGE 129

129 The micromanipulator was then positioned ove r an Imago Scientific Instruments, Inc. manufactured, 36-specimen grid/coupon of fl at top silicon microtips with a 2 m diameter and height of 100 m ( Figure 5-704). The stage being at a ze ro degree tilt, and the foil liftout happening at the sam e stage orientation allowed for a top (end-on) elect ron beam view of the sample which allows it to be centered over the Si post. A side ion beam view allows for monitoring of the distan ce between the sample foil and the Si post, as they are brought closer together ( Figure 5-715). It must be noted that the i on beam current used for im aging the sample foil from the side is much lower than the ion beam current used to shape the same foil, thus preventing drastic sample deformation while mounting. This also means that whenever the foil is imaged from the side it is being implanted with gallium ions. Thus using a lower pA aperture in viewing mode ensures that a sm aller dose of gallium is implante d into the sample. Alternating between both views allows for the sample foil to be properly positioned over the Si post so that the 2 m wide fiduciary marked region is aligned with the 2 m diameter flat top of the Si post. The micromanipulator is moved at a speed of 100 nm/second to bring the sample in physical contact with the post. Contact is indicated by a slight deflection of the sample foil. The platinum GIS needle is then used in a similar th ree step welding process to attach the sample foil to the Si post. Subsequent higher current ion beam cuts are made on either side of the Si post to separate the Si mounted sample from the originating, larger sample foil ( Figure 5-726). This process is repeated for all of the m a rked regions of the 50 m long sample foil. Once all of the samples are mounted atop the Si posts, similar three step welds are applied to the opposite side of the post to further ensure proper adhe sion. The stage is then tilted to 52 so that the ion beam is looking straight down (end-on) onto the samples on top of the Si posts, and the electron beam now has a side view. This allows us to sharpen the samples into

PAGE 130

130 atom probe tips with the FIB, while the side SEM view allows us to track the progress of the sharpening, without implanting ga llium directly from the side. The FIB annular milling is utilized at 20,000X magnification and an initial beam condition of 1000 pA at 30 kV. The FIB top view through the sharpening process may be seen in Figure 5-737 The annular mill causes the ion b eam to be rastered in a circle which is expanded in size, thus allowing for the m illing of a cone shaped atom probe tip 67,71,72, when looked at with the electron beam side view ( Figure 5-748). This 1000 pA milling is conducted until the wh ole sam ple forms a cone shape, and any micro-crowns further down on the Si post are eliminated ( Figure 5-737a-b). This step takes ~6 minut es for a sam pl e with dimensions of 2 X 2 X 2 m. After this step is complete the thr ough-hole weld is visi ble (white arrow in Figure 5-748), ind i cating that we are close to our pre-marked interface of interest (Figure 5-682f). If the region of interest were not close, the current would be dropped to 300 pA and the cone would be m illed further until th e tip was about a micrometer from the interface of interest. The reason for the lower aperture is to prevent milling through the interface prematurely. This method is much easier on a more modern FIB/SEM system equippe d with spy mode imaging where both electron and ion beam live scans may be viewed simultaneously. Once the interface of interest is within 600 nm from the tip top, the FIB is switched over to a low kV aperture ( Figure 5-737c-e) in order to accom p lish the final sharpening and removal of the outermost FIB damaged regions. The low kV aperture results in reduced resolution thus requiring increased contrast in order to focus on the tip of interest at sufficient magnification The low kV cleaning in th is case was conducted at 8 kV with a current of 65 pA. This cleaning is done using the FIB in live view with the fastest refresh rate possible (0.028 sec), and may last anywhere from a 2-20 minutes depending on which material is on top. The face centered cubi c, zirconium rich electrolyte material typically

PAGE 131

131 mills at a (2X) slower rate than the perovski te, lanthanum manganate based cathode material. This final sharpening does an ex cellent job of sharpening the tip to an off-center-axis angle of 15, maximizes the electric field at the tip 64, and thus allows for evaporation to turn-on or begin at electrode voltages around 2,000 volts. Once this final low kV sharpening has been done it is extremely important not to FIB image the sharpened atom probe tip vicinity with a high potential (30 kV) 70. A total of 51 tips were thusly ma de for this LEAP microscope analysis because only about half of th e tips ran successfully in the LEA P microscope. Additionally, of the 51 tips, only five (~10%) yielde d interfaces between ca thode and electrolyte. This was a vast improvement on previous attempts where the succe ss rate was closer to 1% with the previous 108 FIB micro-machined tips. This increased suc cess was attributed to three changes in atom probe preparation: (1) use of three step weld, prevented premature tip weld failure, (2) use of fiduciary marks allowed for easier location of ca thode/electrolyte interf aces and (3) use of low energy FIB cleaning formed tips with smaller ra dii of curvature, which improved evaporation turn-on rates. Once all of the samples have been sharpene d, they are ready to be analyzed on a LEAP microscope 73. If the time required to transport th e samples is over a couple hours the whole coupon is plasma cleaned for an hour to remove any water vapor or other contaminants which may have adsorbed onto the sample coupon during tr ansport. These samples were analyzed using a LEAP microscope located at the Central Analytical Facility at the University of Alabama in Tuscaloosa, AL. The specimens were pum ped down to a pressure of 4-5 X 10-7 torr before being inserted into the analysis chamber where each tip was brought within a couple millimeters of the local electrode. The chamber is equipped with a cold finger which allows samples to be analyzed at sub 100 K temperatures Both samples were analyzed at a temperature of 64.7K.

PAGE 132

132 The target evaporation rates for both samples were 0.4%, with a 250 kHz laser frequency. Sample 1 had a 35 nm starting tip diameter as measured with a FEG SEM TLD-Deep hole detector. It had a golden hits ratio of 61.5% and had a total of 5.6 million detected atoms. Sample 2 had a 49 nm starting tip diameter, a golden ratio of 62.9%, with 8.4 million total detected atoms. 5.3 Results and Discussion After reconstructing runs #3017 (called Samp le 1 from here on) and #3386 (called Sample 2 from here on) using the proprietary Ja va based application Interactive Visualization and Analysis Software ( IVAS) version 3.4.1 (Ima go Scientific Instruments.), it was evident that both had cathode/electrolyte interfaces in their re constructed volumes. Please note that the run # is indicative of order of analys is, with the larger number bein g analyzed in the LEAP system after the smaller number. Furthermore Sample 1 LEAP reconstruction originates from Siemens designated Sample 3 (WPC3), and all Sample 2 LEAP reconstructions originate from Siemens designated Sample 4, which had in creased calcium content than WP C3. Sample 1 consists of a 81 X 81 X150 nm reconstructed volume where the fi rst two dimensions are the x and y axes and the final dimension is the vertical z-axis ( Figure 5-759). The Sample 1 reconstruction consists of a cathode region on top and an elec trolyte region on the bottom. In between the two regions is the in terf ace, with a void on one si de indicated by an arrow. Th e void is visible in both the reconstruction and the SEM image of the tip befo re LEAP analysis. This void may represent an open pore, which would mean that we have reconstructed a trip le phase boundary (TPB). The Sample 2 reconstruction was smaller with dimensions of 55 X 55 X 72 nm with an orientation where the electrolyte is on t op and the cathode is on bottom ( Figure 5-7610). The void observed in Sam ple 2s reconstruction (black arrow) is not visible in the SE M im age. There was no

PAGE 133

133 particular reason for the two sa mples having opposite orientations, but this geometry made the mass spectrum analysis easier. The mass spectrum plots of both samples are shown in Figure 5-7711. These plots have m a ss-to-charge (M/C) ratio in un its of amu or Daltons on the x-axis while counts are on the yaxis The first observation in comparing the two M/C plots is that the smaller cross section of Sample 2 has more noise which is believed to be caused by increased local heating from the laser due to it having a 38% smaller cross-sectional area. This reduced cross-sectional area reduces Sample 2s ability to conduct heat away from the atom probe tip, thus increasing M/C noise. This noise is most likely not caused by differences in thermal conductivity between the cubicelectrolyte and perovskite-cathode phase, sinc e both phases have similar thermal conductivity values. In Sample 1 the electrolyte phase on the bottom is scandium-stabilized zirconia (SSZ), similar to yttrium-stabilized zirconia (YSZ) wh ich has a thermal conductivity of ~2.3 W/mK at room temperature 74. The origins of this low therma l conductivity is vacancy induced phonon scattering due to the doped zirconias highly defective structure which allows for oxygen ion transport at high temperatures 75. Perovskite, similar to the cal cium doped lanthanum manganate cathode here have similar ther mal conductivities of less than 2.2 W/mK, which is why both materials are viable as thermal barrier coatings 76, and well matched in terms of thermal expansion coefficients. Although both samples have a number of peaks, ranging from 1 to 160 Da., we will be focusing on only seven of the major atomic cons tituents which (listed in rising M/C order) include hydrogen, scandium, oxygen, calcium, manganese, lanthanum and zirconium oxide. These seven peaks all occur below 60 Da. The multitude of peaks, above 60 Da, is caused by the many complex ions which are evaporated during analysis. Please note that in subsequent

PAGE 134

134 discussion we are able to resolve dual peaks within each elements M/C range, which displays the sub amu resolution of this technique. The hydrogen peak occurs at one Da, and cons ists of a singly ionized hydrogen atom. It appears that hydrogen is present in high concentra tion at the edge of each interface as seen near the arrows of Figure 5-7812, where there appears to be a lack of other constitu ent atoms. Since these samples were impregnated with a hydrocar bon based epoxy it is possibl e that this epoxy is present at the interface in the form of hydrogen atoms. The scandium peak located at just under 15 Da, consists of triply ionized scandium-45 ( Figure 5-7913). Since scandium is utilized in the elec trolyte phase we can see that it is present in high concentration in the bottom of Sample 1 and in a much smaller concentration in the top of Sample 2, because there is not much of th e electrolyte phase present. This scandium distribution agrees with the phase orientations. The oxygen-16 M/C peak shows a geometry-heati ng effect which we have identified in this work ( Figure 5-8014). If we look at the M/C for Sa m pl e 1 we notice that there is one major peak with a considerable shoulde r coming off of it. But when we look at the M/C for Sample 2 we notice that the shoulder is in the middle of two separate p eaks which are less than 0.3 Da apart. We believe this is proof of heating induced M/C drift during the eight plus million run for Sample 2. If we separate this O-16 peak into three regimes ( Figure 5-8115): the left most peak in blue (A), the m i ddle shoulder in peach (B) an d the right peak in purple (C), we can see how the peak drifts during Sample 2s run We see as we start to evaporate from the top of the sample, it is very rich in this purple O-16, and th en as we progress through the run we see that the peach phase is present in higher concentration in the top electrolyte phase, but as we progress through the run the localized heating forces the ion flight time to reduce due to higher tip

PAGE 135

135 temperature which shifts the oxygen peak to the left and we see that in the end the blue phase of O-16 is only present in the bottom (cathode) regi on of the sample. But the existence of the purple peak, and its higher content at the top show s us that the oxygen rich phase for Sample 2 is in the top, as seen with this purple O-16 peak. For Sample 1 we notice that the highest c oncentration of the oxygen-16, blue peak is present in the electrolyte, as we would e xpect the electrolyte phase to have a higher concentration of oxygen than the cathode phase. The peach or shoulder background is present in all regions of the sample, while the purple regi on is present at the top of the tip and in the electrolyte with a reduc tion in the middle. In Sample 1 we see that background peach and purple oxygen are present basically in the whole recons truction, but that the blue region oxygen is present in the highest concentration in the el ectrolyte phase. Thus to have an accurate visualization of the oxygen peaks we should visu alize the purple region for Sample 2 and the blue region for Sample 1 with the peach region. When we look at the oxygen-17 isotope we notic e a similar appearance of peak spreading ( Figure 5-8216). This peak exhibits the same geom etry-heating effect seen for oxygen-16, except that h ere, for Sample 2, the three regimes e xhibit three distinct peaks, with similar patters to oxygen 16 which show that in Sample 1 the electrolyte rich phase is on the bottom and in Sample 2 the oxygen rich phase is on the top. The doubly ionized 40Ca+2 peak exhibits similar shif ts as observed in oxygen ( Figure 58317). In Sam ple 1 the red calcium peak is strong in the cathode phase which is calcium doped lanthanum manganate. As the analysis goes through the interface and into the electrolyte the calcium concentration drops off (gray peak). In Sample 2 the opposite material progression is present and the opposite trend is observed with th e calcium peak, where in the beginning the red

PAGE 136

136 peak is weak as the electrolyte phase is evaporat ed and then increases into the gray peak as we cross into the calcium rich cathode phase. Pleas e note, as mentioned before that both red and gray peaks are in fact calcium peaks, and the M/C values are relative charge. The same effect is observed in the doubly ionized 55Mn+2 peak ( Figure 5-8418), where two peaks are seen for Samp le 1 because of its orient ation with the manganese phase being at the top of the tip, and first to be analyzes. As we progress through the tip, the manganese peak shifts from red on the right to the yellow on the left due to localized heating. In Sample 2 there is only one dominant manganese peak (yellow), with a sli ght second (red) peak present on the right. Here we can also see the effect of the cross-s ectional area because the peak spread is higher for Sample 2 than Sample 1 as shown by the higher background in Sample 2. This happens because Sample 2 has a smaller cross-section and conducts the thermal energy away from the evaporation surface at a slower rate. The triply ionized 139La+3 peak exhibits peak broadening with the same peak shifts observed with calcium and manganese, where there are only two regions ( Figure 5-8519). The left region is labeled in pink and the right region in blue. In S a mple 1 the lanthanum rich region is the cathode located on top (pink peak), and in Sample 2 the cathode is located on the bottom (pink peak). Sample 2 seems to have much more of the blue/weaker peak in the cathode than does Sample 1. With all of these overlapping peaks, a way to confirm that the peak shift is occurring due to heating towards the end of the analysis wa s to take Sample 1 and separate the cathode reconstruction from the electrolyte reconstruc tion. The voltage history of Sample 1 ( Figure 58620) is plotted as a voltage versus ion sequence num ber. As you can see the turn on voltage, or the voltage at which the tip starte d to em it ions was just above 2000 V. This turn on voltage is a

PAGE 137

137 function of tip radius. As we progress to the right in the analysis, by removing monolayer after monolayer of atoms the tip radius of curvature increases and thus the vol tage must increase in order to keep the evaporation rate, or the dose which the detector is c ounting, constant. The curvature in the voltage history plot is evidence of a transition from one phase (cathode on the left) to another phase (electrolyte on the right). If we split the reconstruc tion into a cathode (top) and electrolyte (bottom) phase with respect to the analysis direction ( Figure 5-8721) we can isolate th e p osition of the M/C peaks for each ph ase. Such a mass spectra is plotted in Figure 58822. It is shown that regardless if the peak is strong or weak in the phase (due to atom ic concentratio n), the reason why they drift to the left is dependant on the collection sequence. The peaks on the right are collected before the peak s on the left. The cathode phase has a stronger calcium peak than the electroly te phase, as expected because the cathode is LCM. The electrolyte phase has a stronger oxygen-16 peak than the cathode, which is also expected because the electrolyte, SSZ has higher oxyg en content than the LCM. The significance of the peak drifting to the left indicates that the time of f light for the ionic species is decreasing which is most likely caused by sample heating on the order of 2-3 K. Zirconium does not show up as a single ion, but it does show up as ZrO2 +1, 16 Da to the right of where the zirconium peak should appear. In fact zirconi um has five prevalent isotopes of varying intensities, yieldi ng a zirconium hand-shaped fin ger print pattern on the mass spectra plot ( Figure 5-8923). The zirconium oxide peaks ar e represented by five peaks, four of which are unique, and one which overlaps with an ir on peak at 56 Da. W e can see in that this signature peak pattern is strong in Sample 1 which has a considerable volum e of electrolyte phase (SSZ). Sample 2 has only a couple of th e strongest peaks present because only a small portion of the electrolyte phase being present in Sample 1.

PAGE 138

138 To confirm that the green peaks shown in Figure 5-8923 was a ZrO2 peak and not an iron peak, a data pipe was taken trough th e Sample 1 reconstruction volume 77. A plot of atomic concentration versus distance ( Figure 5-9024) confirms that the green M/C peak signal shown in pink on the concentration profile in f act m imics the zirconium oxide peak perfectly. Thus it is concluded that the green peak in Figure 5-8923 was in fact zirc onium oxide and not iron. One nanom eter diameter data pipes were ut ilized to quantify the 1-D atomic percent concentration profiles of all seven constituent elements through both samples. The Sample 1 concentration profile may be seen in Figure 5-9125, with a magnified view of the cathode (left), electroly t e (right) interface in Figure 5-9226. We can see that in Sample 1, the manganese species h a s the furthest segregation into the electrolyte, zirconium profile. The manganese profile crosses the zirconium profile 4 nm furthe r than the calcium and lanthanum profiles. The calcium and lanthanum profiles appear to segreg ate less across the in terface boundary than the manganese. The Sample 2 concentration profile may be seen in Figure 5-9327, with a higher m a gnification view of the electrolyte (lef t), cathode (right) interface shown in ( Figure 5-9428). In Sam p le 2 the manganese still appears to segreg ate further than the calcium or lanthanum into the electrolyte phase on the left. Here the samp le dimensions do not allow for a measure of how far the manganese segregates into the electrolyt e. We can see that the calcium segregates a nanometer further than the lanthanum in this sa mple. This may be due to the fact that the calcium atomic concentration is 13 at. % in Sample 2, and only 12 at. % in Sample 1. In addition the lanthanum concentration in the Sample 1 cat hode is 10.5 at. % while it is only 9 at. % in Sample 2. Both the higher calcium concentration and the lower lanthanum concentration in

PAGE 139

139 Sample 2s cathode may be the cause of this segregation difference between Sample 1 and Sample 2. In order to better characterize the Sample 2 interface, another Sample 2 atom probe tip was successfully reconstructed ( Figure 5-9529). Our sample identification number for this reconstruction was 3294 (Sam ple 2-3294). The pre anal ysis tip can be seen in part (a), where there is a void present. The same void is present in part (b). If we rotate the atomic reconstruction 180 degrees in part (c) we can see the void and the presence of electrolyte at the tip marked with a letter E. The concentration profile of Sample 2-3294s atomic reconstruction is seen in Figure 5-9630. This sample is the sam e orientation as the previous Sample 2 reconstruction with the electrolyte atop the cathode. It is apparent that the manganese profile penetrates the electr olyte further than calcium a nd lanthanum. The manganese concentration in the cathode bulk is 30 at. %, the calcium and lanthanum cathode bulk concentrations are 18 and 9 at. %, respectively. The void once again appears to have a higher concentration of hydrogen as previously observed in Samples 1 and 2 ( Figure 5-9731). If we reconstruct just the hydrogen (H+1) we observe that there is a higher concentration in the electrolyte phase near the E marker than th e cathode. As we approach the void, the hydrogen profile jumps up four times in concentration magnitude (Figure 5-9731). This may be caused by the im pregnated epoxy, or a higher concentration of hydrogen or water at the electrolyte/cathode grain boundaries. In the previous three LEAP reconstructions, th e interfaces were orient ed perpendicular to the direction of analysis. A f ourth tip, Sample 2-3384 was fabric ated and reconstructed to yield an interface with a perpe ndicular orientation to th e analysis direction ( Figure 5-9832). This figure shows us that cathode bulk and electrolyte bulk are both pr esent due to the flattening of

PAGE 140

140 the concentration profiles. Manganese is th e dominant cathode species (omitting oxygen) and segregates further than the calcium species in the E labeled electrolyte phase, despite calcium and manganese having equal concentrations at xdistance of 8 nm. Furthermore, the manganese concentration plateaus around 8 at. %, 15 nm away from the interface, in the electrolyte phase. This seems to solidify the observation that manganese is the species that segregates the furthest in the cathode/electrolyte in terface in the LCM/SSZ system. The presence of hydrogen at the interfacial r econstruction voids made us wonder if other elements experienced similar segregation at cathode /electrolyte interfaces. To visualize this a 2D concentration profile was constructed for a nother Sample 2 reconstruction, Sample 2-3401 ( Figure 5-9933). In the SEM image of Figure 5-9933 the electrolyte phase is in the f r ont, and the cathode phase in the back. The interface between the two forms an upside down U shape, with what appears to be a void, which is shown with a black arrow. In the LEAP reconstruction the tip we see the appearance of the same void (bla ck arrow) in the electr olyte region (denoted by the letter E). This sample is or iented with the interface parallel to the analysis direction. In Figure 5-10034 the top view atomic reconstruction is shown in part (a) where the arrow indicates the sam e void seen in Figure 5-9933. Electrolyte phase is de noted with the letter E, and cathode phase is den o ted with the letter C. The side view is shown in part (b) wh ere the electrolyte is on the right and cathode on the left. The side vi ew accompanying 1-D concentration profile shows the cathode/electrolyte interface present at an x distance of 130 nm. Once again the manganese segregates the furthest into th e electrolyte. If only the zirconium and manganese atoms are visualized ( Figure 5-10034d) a linear interface emerges. If a 1 nm by 1 nm probe is applied to the reconstruction we are able to derive a 2-D concentration profile (a s view end on with the analysis direction going into the plane of the page) for each element in the volume (Figure 5-

PAGE 141

141 10135). The void observed in the atomic recons truction top view is seen to have a high concentration of hydrogen and oxygen. The scandium intensity is present in the top, electrolyte region, in low concentrations. The manganese concentration appears to have the furthest penetration across the length of th e interface into the electrolyte region, where as the calciums segregation is less, and the lanthanum is even less. The scandium, oxygen manganese and calcium concentrations seem to outline this void. Although no electroc hemical measurements are presented here it is possible to postulate th at the hydrogen rich void is an epoxy impregnated active triple phase boundary (TPB). If this were the case, this 2-D concentration analysis may indicate that under fuel cell operation hydrogen, oxygen, scan dium, calcium and manganese segregate towards the electrochemically active re gion. Conversely the zirconium and lanthanum species tend to segregate away from this region. 5.4 Conclusions This work represents the first LEAP microsc ope 3-D atomic reconstructions of a solid oxide fuel cell. The interface between the electrolyte and cath ode phases within the composite regions of two separately processed Siemens Energy Inc., SOFC cathodes were prepared into atom probe tips with tip diameters below 50 nm. The reconstructions showed what effect tip geometry has on mass-to-charge ratio graphs. The cathode phase originat ing manganese species appears to segregate the furthest into the s candium doped zirconium oxide, electrolyte phase. This distance appears to be over 20 nanometers. Calcium and lanthanum are found to segregate across the cathode/electrolyte boundary only around 10 nanometers, which is on the order of the interface thickness. The cathode phase calcium atomic concentration values varied by 1-2 atomic percent between Samples 1 and 2, (Sample 2 having a higher average calcium concentration of 20 atomic per cent) which is comparable to previous STEM-EDS studies 51, 52. This work also shows the advantage of the LEAP microscopes atomic probe size as compared to

PAGE 142

142 STEM techniques for interfacial chemical segregation studies. Many compound ions are evaporated and future work should focus on confirming their existence and position in the M/C range in order to yield more understanding to interfacial phases present at the cathode/electrolyte interface. In addition the effect of bias on inte rfacial segregation may now be investigated on the atomic level. The LEAP technique provides research ers with a tool for earl y detection of ternary phase formation at the various interfaces present in solid oxide fuel cells.

PAGE 143

143 Figure 5-67. Experimental LEAP setup. The atom probe experiments can be carried out using either HV or laser pulses. Reprinted w ith permission from E.A. Marquis and B. Gault, "Determination of the tip temperatur e in laser assisted atom-probe tomography using charge state distributions," J. Appl. Phys., 104:8 (2008). Copyright 2008, American Institute of Physics.

PAGE 144

144 Figure 5-68. Schematic of the dual beam FIB/SE M assisted (a) deposition of platinum on the composite cathode, (b,c) trenching around the platinum, (d-e) thinning of the sample and using channeling contrast to differentiate pore,cathode and electrolyte phases, (fg) marking the platinum with fiduciary mark s for areas of intere st, (h) undercutting, and (i) micromanipulator liftout of sample.

PAGE 145

145 Figure 5-69. A schematic of the three-step micromanipulator welding process where (a) the micromanipulator is made to touch the sample, (b) small through hole is made at interface and backfilled with platinum, and (c) larger area platinum is deposited for reinforcement. 36 Flat top Si posts Three fiduciary posts Figure 5-70. SEM image of an Imago Scientific fl at top Si post array. The posts with white bases have been mounted with atom probe samples. The discoloration is caused by FIB damage caused during sharpening.

PAGE 146

146 Figure 5-71. Depiction of (a) sample liftout and orientation of micromanipulator and GIS platinum needle with respect to sample tr ench, (b) top, SEM view of sample being positioned over a silicon flat top post, and (c) side, FIB view of sample being lowered down onto the silicon post. Figure 5-72. Low current FIB micrograph of (a) sa mple being made to to uch the silicon flat-top post, (b) platinum weld deposited at the interface between the sa mple and the post, and (c) use of higher current FIB to cut mounted sample free from rest of sample.

PAGE 147

147 1000pA Ga, 30kV, 2 min 1000pA Ga, 30kV, 2 min 1000pA Ga, 30kV, 6 min 1000pA Ga, 30kV, 6 min 65pA Ga, 8kV, 1500X 65pA Ga, 8kV, 1500X 65pA Ga, 8kV cleaning 65pA Ga, 8kV cleaning 35,000X 35,000X 65pA Ga, 8kV Low 65pA Ga, 8kV Low magnification, increase magnification, increase contrast contrastabc d e Figure 5-73. Focused ion beam top view of Si post with sample atop as high voltage (30kV) annular beam is used to form the tip (a-b), then low voltage (8kV) beam is utilized to conduct final tip shapin g and cleaning (c-f).

PAGE 148

148 As mounted As mounted 1000pA Ga, 30kV, 2 min 1000pA Ga, 30kV, 2 min 1000pA Ga, 30kV, 4 min 1000pA Ga, 30kV, 4 min 1000pA Ga, 30kV, 6 min 1000pA Ga, 30kV, 6 min 65pA Ga, 8kV cleaning 65pA Ga, 8kV cleaning 10 min 10 min Final 25 nm tip diameter Final 25 nm tip diameter 65pA Ga, 8kV cleaning 65pA Ga, 8kV cleaning 20 min 20 min Figure 5-74. SEM side views of the FIB annular beam tip sharpening using 30kV beam to form the cone and then an 8kV beam to remove gallium damage and form the final tip.

PAGE 149

149 Figure 5-75. LEAP reconstruction of Sample 1. Axes are in units of nanometers. Top region is cathode phase and botom region is electrolyte phase. The pre analysis SEM image of the tip is shown on the right. Arrows ar e pointing to the void regions on both.

PAGE 150

150 Figure 5-76. LEAP reconstruction of Sample 2. Axes are in units of nanometers. Electrolyte phase is at the top. Cathode phase is on the bottom. Top region is cathode phase and botom region is electrolyte phase. The pre analysis SEM image of the tip is shown on the right.

PAGE 151

151 Figure 5-77. Mass spectrum plots of counts ve rsus mass-to-charge ratios for Samples 1& 2

PAGE 152

152 Figure 5-78. Hydrogen atomic reconstructions along with the hydrogen mass spectrum. Arrows point to hydrogen enrichment seen near ca thode/electrolyte interface of both Samples 1 and 2.

PAGE 153

153 Figure 5-79. Scandium atomic reconstructions al ong with the mass spectra for Samples 1 and 2.

PAGE 154

154 Figure 5-80. Oxygen atomic recons tructions along with the mass spectra for Samples 1 and 2.

PAGE 155

155 Figure 5-81. Oxygen atomic reconstr uctions are divided into three di stinct peaks labeled in blue, peach and purple. We can see that as the sample analysis progresses the only peak present throughout is the middle peach peak (B). The left most, blue peak is present at the end of the analysis (A). The right most, purple peak is pr esent at the begining of the analysis (C). This is ev idence of mass-to-cha rge peak drift.

PAGE 156

156 Figure 5-82. Oxygen isotope 17 atomic reconstruc tion with mass spectra for Sample 1 and 2.

PAGE 157

157 Figure 5-83. Calcium atomic reconstruction with mass spectra for Samples 1 and 2.

PAGE 158

158 Figure 5-84. Manganese atomic reconstruction with mass spectra for Samples 1 and 2.

PAGE 159

159 Figure 5-85. Lanthanum atomic reconstruction with mass spectra for Samples 1 and 2.

PAGE 160

160 Figure 5-86.Voltage history plot for the analysis of Sample 1. Y axis is voltage, X axis is ion sequence number. The analysis starts at a low voltage on the left and progresses to the right with increasing voltage. The r econstruction image is shown with cathode phase on the left and electrolyte phase on the right.

PAGE 161

161 Figure 5-87.The reconstruction of Sample 1 is split up into a cathode (top) region and an electrolyte (bottom) region with re spect to the analysis direction.

PAGE 162

162 Figure 5-88. Sample 1's oxygen-16 and calcium mass-to-to charge ratio peaks both drift to the left as the LEAP analysis progresses from the top, cathode phase to the bottom, electrolyte phase.

PAGE 163

163 Figure 5-89. Zirconium oxide atomic reconstructi on with mass spectra peaks for Samples 1 and 2 highlighted in purple with an possible overl ap with an iron(56) peak highlighted in green.

PAGE 164

164 Figure 5-90. Sample 1 atomic recons truction is shown with analysis direction as a reference, next to a 1-D concentration profile of zirconium oxide (purple) and pink signal related to the possible iron peak. This plot confirms that the pink signal is zirconium based.

PAGE 165

165 Figure 5-91.Sample 1 atomic rec onstruction and 1-D atomic concentration versus distance (nm) profile. Analysis direction is from left to right.

PAGE 166

166 Figure 5-92. Sample 1 high magni fication atomic reconstruction and 1-D atomic concentration versus distance (nm) profile. Analysis direction is from left to right.

PAGE 167

167 Figure 5-93. Sample 2 atomic reconstruction with a 1-D atomic concentration versus distance (nm) profile.

PAGE 168

168 Figure 5-94. High magnification of Sample 2 atomic reconstruction with a 1-D atomic concentration versus distance (nm) profile. Figure 5-95. Another Sample 2 atom probe tip with a void shown by arrow imaged with SEM (a), with an atomic reconstruction w ith the same void present (b), and a reconstruction rotated 180 degrees which shows the electrolyte region at the top marked with and E (c).

PAGE 169

169 Figure 5-96. Sample 2-3294's atomic reconstr uction is shown above its 1-D concentration profile.

PAGE 170

170 Figure 5-97. Sample 2-3294's hydrogen atomic reconstruction and 1-D atomic concentration profile.

PAGE 171

171 Figure 5-98. Sample 2-3384 atomic reconstruction top view, side view, and 1-D concentration profile. The interface is parallel to the direction of analysis. Figure 5-99. Sample 2-3401 LEAP reconstruction is shown on the right (axes dimensions are shown in nanometers). A SEM image of th e same pre-LEAP analysis tip is shown on the left. The arrows in both images point towards visible voids. E indicates electrolyte phase. C indicates cathode phase.

PAGE 172

172 Figure 5-100. Sample 2-3401 is shown with (a) top view where all atoms are visible, (b) side view with all the atoms vi sible, (d) top view with only zirconium and manganese atoms visible and a 1-D concentration profile Arrows point to voids seen in top views. E stands for electrolyte pha se. C stands for cathode phase.

PAGE 173

173 Figure 5-101. Sample 2-3401 2-D concentration prof iles are shown. The concentration intensity is indicated by color, where red is the highest and blue is zero inte nsity. Probe size is 1 nm x 1 nm pixel .

PAGE 174

174 CHAPTER 6 CONCLUSION 6.1 Conclusion We have successfully reconstructed a 3-D model of an actual LSCF cathode. This provides the ability to quantify porous cathode microstructures at the sub-micron level. This technique allows researchers to quantify microstr uctural properties in be tween processing steps, and at specific sites. Further, this technique has allowed us to develop the direct link between the microstructure and performa nce relationship in SOFC cathodes 78, 41. Microstructures of two Siemens-Westinghouse cathode-SOFC samples were analyzed. The cathode/electro lyte interface was characte rized with TEM-EDS. Sa mple 2 had higher calcium content in both electrol yte and cathode phases. This calcium difference may be the cause of the coarser microstruc ture seen in sample 2 because calcium is known to act as a sintering aid 60. The cathode microstructure was quantif ied with the aid of a FIB/SEM. Both samples consisted of a composite cathode on the order of 20 m, sandwiched between the porous cathode support and dense electrolyte. Surface ar ea, porosity, particle size, tortuosity and, for the first time, 3-D topological c onnectivity were all qua ntified with the aid of Amira software. Sample 2 had a coarser composite cathode micros tructure with 2.5 % higher composite cathode connectivity. This was the first time that true 3-D topological connectivity was quantified for an SOFC structure. In the future, connectivity may be related to all transport processes that occur throughout the SOFC. This advancement in connectivity quantification can further the understanding of many other fields including biol ogical systems connectivity, for example, by aiding researchers in better understanding the 3-D lattices used in synthetic tissue harves ting. Moreover, the

PAGE 175

175 MAZO quantification technique helps in unders tanding 3-D load distribution in structural materials. The combined use of FIB/SEM and skeletonization technology allows for more intelligent design of porous micro and nano functional materials. This work represents the first LEAP microsc ope 3-D atomic reconstructions of a solid oxide fuel cell. The interface between the electrolyte and cath ode phases within the composite regions of two separately processed Siemens Energy Inc., SOFC cathodes were prepared into atom probe tips with tip diameters below 50 nm. The reconstructions showed what effect tip geometry has on mass-to-charge ratio graphs. The cathode phase originat ing manganese species appears to segregate the furthest into the s candium doped zirconium oxide, electrolyte phase. This distance appears to be over 20 nanometers. Calcium and lanthanum are found to segregate across the cathode/electrolyte boundary only around 10 nanometers, which is on the order of the interface thickness. The cathode phase calcium atomic concentration values varied by 1-2 atomic percent between samples 1 and 2, (sample 2 having a higher average calcium concentration of 20 atomic per cent) which is comparable to previous STEM-EDS studies 51, 52. This work also shows the advantage of the LEAP microscopes atomic probe size as compared to STEM techniques for interfacial chemical segregation studies. Many compound ions are evaporated and future work should focus on confirming their existence and position in the M/C range in order to yield more understanding to interfacial phases present at the cathode/electrolyte interface. In addition the effect of bias on interfacial segregation should, and now may be investigated on the atomic level. The LEAP techni que provides researchers with a tool for early detection of ternary phase formation at the various interfaces present in SOFCs.

PAGE 176

176 6.2 Future Work 6.2.1 3-D micro texture characte rization of the composite cathode using FIB/SEM and EDX/E BSD A 3-D-EBSD technique has been devel oped to simultaneously quantify morphology (FIB/SEM), crystallographic orientation (EBSD), and chemical composition (EDX) of a composite cathode. The system uses a joint highresolution field emission SEM-EBSD set-up in conjunction with a focused ion beam (FIB) system in the form of a 3-D crystal orientation microscope (3-D EBSD) 79. With this technique FIB sequentially mills slices of material by sputtering with a high energy Gallium ion beam operating in gracing incidence. A CCD camera takes an EBSD pattern for each slice. Then so ftware automatically indexes the EBSD pattern, and builds an orientation image map. Consequently 2-D orientation image maps that consist of grain boundary of two phases highli ghted in various colors repres ent different crystal structure orientation of each phase. The orientation image maps obtained in each serial section of the material are then used to reconstruct the composite microstructu re in three dimensions including all features that can be jointly de tected via EBSD and/or EDS mapping 79 80.. EDS can be conducted on the flat surfaces after FIB sectioni ng to collect composition analysis per slice. Therefore, 3D dimensions of the composite cathode, distribution of LSM or YSZ in the composite cathode, and chemical information are all characterized by inte gration of EDX/EBSD and FIB/SEM. The EBSD assisted reconstruc tion would allow us to image and reconstruct the grain boundaries of the composite cathode in three dimensions due to the randomly oriented nature of the electrolyte and cathode grains. With this expanded capability over the current 3-D FIB/SEM reconstructions, a more thorough understanding of th e electrochemical importance of grain boundaries would be developed. This understa nding could be combined with the herein

PAGE 177

177 developed topological connectivity, thus allowing us to truly m odel the cathode performance by fully understanding how many different grains, and or ientations the ions/el ectrons must pass in order to arrive at the electrolyt e. This may also be combined with the skeletonization nodes to further understand where the skeleton nodes de velop and if they have any preferred crystallographic orientation. This work should be conducted on symmetric cell composite cathodes such as LSCF/GDC so that electroc hemical performance ma y be related to the composite cathode structure as has been done here with the single phase structure. This would allow for a more thorough understanding of the fundamental relationship between electrochemical performance of SOFCs and their microstructure. 6.2.2 LEAP standards As we have seen in chapter five, the M/C spectrum is quite noisy for the composite oxides that make up cathodes and electrolytes, especia lly above 60 Da. In order to understand the multiple ion peaks, a study of various compositi on standards should be done. These standards should be made so as to confirm the peaks th at are unique to certain compounds such as an insulating phase of strontium zirconate which appears at LSCF/YSZ interfaces at high temperatures. By making standards of known tern ary phases it will facili tate the understanding of which peaks are a part of standard backgr ound LEAP analysis and which peaks may indicate actual ternary phases. One such standard should be zirconium oxide to confirm that zirconium is difficult to ionize by itself below 100 Da. A study of the compound standards would be useful to researchers in many fields, especially as the LEAP microscope starts to gain popularity across academic and industrial research labs. 6.2.3 LEAP bias study An initial study of the effect of bias was conducted on LSCF/YSZ symmetric cells sintered at 950 C for one hour. The purpose of the study was to i nvestigate the e ffect of break-

PAGE 178

178 in on these cathodes to see if we could figure out w hy it is that fuel cells tend to initially improve in performance before starting to degrade. This process typically happen s within the first couple hundred hours of operations. In our study we bi ased the symmetric cathode for 12 hours with a constant current of 250 mA. We can see in Figure 6-1 the current in orange and the m easured bias in b lue. The interruptions in the bias were due to regularly scheduled EIS scans. We can see that as the 12 hours draws to a close that the bias required to pull th e same 250 mA of current increases from -4.2 volts to -5.7 volts thus in creasing by 35% (1.5V) in just 12 hours. The pre and post application of bias EIS scans confir med that something was changing as seen in Figure 6-2. W e see a 10X incre ase in the total polarization resistance between the initial (prebias) scan and the scan after 12 hours of direct current flowing through the symmetric cell. We theorized that something was occurring at the LSCF/YSZ interface, and sought to find proof using LEAP reconstructions. Unfortunately for us, the nano LSCF microstructure do es not allow for easy micromachining of LEAP atom probe tips. Most of the tips that we made resulted in catastrophic tip failure. It may ha ve happened due to the fine part icle size, thus the same bias study should be conducted on coarser LSCF/YSZ sy mmetric cells to isolate what species are segregating at the electrochemically active interfaces. It may also be of interest to investigate the effect of bias on LSCF/GDC which is known to be a more stable cathode/electrolyte combination. This work could be very beneficial as part of a fast feedback loop in investigating interfacial phases or segregation which may have deleterious effects on SOFC performance. The fact that it has atomic level special and chemical resolution means that it would be applicable only on the nanometer scale interfaces of interest, not bulk properties.

PAGE 179

179 Figure 6-1. Change in measured bias under cons tant applied current of LSCF/YSZ symmetric cell. Figure 6-2. Effect of applied 12 hou r bias on LSCF/YSZ polarization.

PAGE 180

180 REFERENCES 1. Dyer, C. K. Replacing the ba ttery in portable electronics. Scientific Am erican 281, 88-93 (1999). 2. Dyer, C. K. Fuel cells for portable app lications. Journal of Power Sources 106, 31-34 (2002). 3. Tietz, F., Haanappel, V. A. C., Mai, A., Mertens, J. & Stover, D. Performance of LSCF cathodes in cell tests. Journal of Power Sources 156, 20-22 (2006). 4. Dusastre, V. & Kilner, J. A. Optimisation of composite cathodes for intermediate temperature SOFC applications. Solid State Ionics 126, 163-174 (1999). 5. Primdahl, S. & Mogensen, M. Durability and thermal cycling of Ni/YSZ cermet anodes for solid oxide fuel cells. Journal of Applied Electrochemistry 30, 247-257 (2000). 6. Minh, N. Q. Ceramic Fuel-Cells. Journal of the American Ceramic Society 76, 563-588 (1993). 7. Mobius, H. H. On the history of solid el ectrolyte fuel cells. Journal of Solid State Electrochemistry 1, 2-16 (1997). 8. Adler, S. B. Factors governing oxygen reduction in solid oxide fuel cell cathodes. Chemical Reviews 104, 4791-4843 (2004). 9. Jiang, S. P. Issues on development of (La,Sr)MnO3 cathode for solid oxide fuel cells. Journal of Power Sources 124, 390-402 (2003). 10. McEvoy, A. J. Materials for high-temperat ure oxygen reduction in solid oxide fuel cells. Journal of Materials Sc ience 36, 1087-1091 (2001). 11. Richter, J., Holtappels, P., Graule, T., Naka mura, T. & Gauckler, L. Materials design for perovskite SOFC cathodes. Monatshefte fr Chemie / Chemical Monthly (2009). 12. Fukunaga, H., Ihara, M., Sakaki, K. & Yamada, K. The relationship between overpotential and the three phase boundary length. Solid State Ionics 86-8, 1179-1185 (1996). 13. Skinner, S. J. Recent advances in Perovs kite-type materials for solid oxide fuel cell cathodes. International Journal of In organic Materials 3, 113-121 (2001). 14. Virkar, A. V., Chen, J., Tanner, C. W. & Kim, J. W. The role of electrode microstructure on activation and concentration pol arizations in solid oxide fuel cells. Solid State Ionics 131, 189-198 (2000).

PAGE 181

181 15. Teraoka, Y., Nobunaga, T., Okamoto, K., Miura, N. & Yamazoe, N. Influence of Constituent Metal-Cations in Substitute d Lacoo3 on Mixed Conductivity and Oxygen Permeability. Solid State Ionics 48, 207-212 (1991). 16. Mendelssohn, K. The World of Walther Nern st, the Rise and Fall of German Science, 1864-1941 (Macmillan, London, 1973). 17. Singhal, S. & Kendall, K. (eds.) Hi gh Temperature Solid Oxide Fuel Cells: Fundamentals, Design and Applications (Elsevier, Oxford, 2003). 18. Wachsman, E. D., Ball, G. R., Jiang, N. & St evenson, D. A. Structural and Defect Studies in Solid Oxide Electrolytes. Solid State Ionics 52, 213-218 (1992). 19. Jiang, N. X. & Wachsman, E. D. Structural stability and conductiv ity of phase-stabilized cubic bismuth oxides. Journal of the Amer ican Ceramic Society 82, 3057-3064 (1999). 20. Kilner, J. A. Fast oxygen transport in acceptor doped oxides. Solid State Ionics 129, 1323 (2000). 21. Faber, J., Geoffroy, C., Roux, A., Sylvestre, A. & Abelard, P. A Systematic Investigation of the Dc-Electrical Conductiv ity of Rare-Earth Doped Ceria. Applied Physics aMaterials Science & Pro cessing 49, 225-232 (1989). 22. Ralph, J. (University of London, London, 1998). 23. Zhang, Y. H. et al. A screen-printed Ce0.8Sm0.2O1.9 film solid oxide fuel cell with a Ba0.5Sr0.5Co0.8Fe0.2O3-delta cathode. Journal of Power Sources 160, 1217-1220 (2006). 24. Von Dollen, P. & Barnett, S. A study of screen printed yttria-stabili zed zirconia layers for solid oxide fuel cells. Journal of the Am erican Ceramic Society 88, 3361-3368 (2005). 25. Gellings, P. J. & Bouwmeester, H. J. M. (eds.) The CRC handbook of solid state electrochemistry (CRC Pr ess, Boca Raton, 1997). 26. Kim, J. W., Virkar, A. V., Fung, K. Z., Meht a, K. & Singhal, S. C. Polarization effects in intermediate temperature, anode-supported solid oxide fuel cells. Journal of the Electrochemical Society 146, 69-78 (1999). 27. MacDonald, J. R. Impedance Spectroscopy; Emphasizing Solid Materials and Systems (John Wiley and Sons, New York, 1991). 28. Nowotny, J., Bak, T., Nowotny, M. K. & Sorrell, C. C. Charge transfer at oxygen/zirconia interface at elevat ed temperatures Part 1: Basic properties and terms. Advances in Applied Ceramics 104, 147-153 (2005).

PAGE 182

182 29. Zhao, F., Jiang, Y., Lin, G. Y. & Virkar, A. V. in Solid Oxide Fuel Cell VII (eds. Yokokawa, H. & Singhal, S. C.) 501-510 (Elect rochemical Society Proceedings Series, 2001). 30. Lee, J. H. et al. Quantitative analysis of microstructure and its re lated electrical property of SOFC anode, Ni-YSZ cermet. Solid State Ionics 148, 15-26 (2002). 31. Kirk, E. C. G., McMahon, R. A., Cleaver, J. R. A. & Ahmed, H. Scanning Ion Microscopy and Microsectioning of Electron-Beam Recrysta llized Silicon on Insulator Devices. Journal of Vacuum Scien ce & Technology B 6, 1940-1943 (1988). 32. Orloff, J. High-Resolution Focused Ion-Beam s. Review of Scientific Instruments 64, 1105-1130 (1993). 33. Dunn, D. N. & Hull, R. Reconstruction of three-dimensional chemistry and geometry using focused ion beam microscopy. A pplied Physics Letters 75, 3414-3416 (1999). 34. Dunn, D. N., Shiflet, G. J. & Hull, R. Quantitative three-dimensional reconstruction of geometrically complex structures with na noscale resolution. Revi ew of Scientific Instruments 73, 330-334 (2002). 35. Bansal, R. K., Kubis, A., Hull, R. & Fitz-G erald, J. M. High-resolution three-dimensional reconstruction: A combined scanning electron microscope and focused ion-beam approach. Journal of Vacuum Science & Technology B 24, 554-561 (2006). 36. Holzer, L., Indutnyi, F., Gasser, P. H., Munch, B. & Wegmann, M. Three-dimensional analysis of porous BaTiO3 ceramics using FIB nanotomography. Journal of MicroscopyOxford 216, 84-95 (2004). 37. Holzer, L., Muench, B., Wegmann, M., Gass er, P. & Flatt, R. J. FIB-Nanotomography of particulate systems Part I: Particle shap e and topology of interfaces. Journal of the American Ceramic Society 89, 2577-2585 (2006). 38. Munch, B., Gasser, P., Holzer, L. & Flatt, R. FIB-nanotomography of particulate systems Part II: Particle re cognition and effect of boundary trunc ation. Journal of the American Ceramic Society 89, 2586-2595 (2006). 39. Wilson, J. R. et al. Three-dimensional r econstruction of a solid-oxide fuel-cell anode. Nature Materials 5, 541-544 (2006). 40. Gostovic, D., Smith, J. R., Kundinger, D. P., Jones, K. S. & Wachsman, E. D. Threedimensional reconstruction of porous LSCF cathodes. Electrochemical and Solid State Letters submitted (2007). 41. Smith, J. R. et al. Evaluation of the re lationship between cathode microstructure and electrochemical behavior in SOFCs. Solid State Ionics submitted (2007).

PAGE 183

183 42. Wagner, R. C., Czymmek, K. & Hossler, F. E. Confocal microscopy, computer modeling, and quantification of glomerular vascular corrosion casts. Microscopy and Microanalysis 12, 262-268 (2006). 43. Kim, S. D. et al. Nano-composite material s for high-performance and durability of solid oxide fuel cells. Journal of Po wer Sources 163, 392-397 (2006). 44. Sholklapper, T. Z., Radmil ovic, V., Jacobson, C. P., Visco, S. J. & De Jonghe, L. C. Synthesis and stability of a nanoparticle-i nfiltrated solid oxide fuel cell electrode. Electrochemical and Solid Stat e Letters 10, B74-B76 (2007). 45. Giannuzzi, L. A. & Stevie, F. A. (e ds.) Introduction to Focused Ion Beams: Instrumentation, Theory, Techniques and Practice (Springer, New York, 2005). 46. Smith, J. R. et al. Evaluation of the re lationship between cathode microstructure and electrochemical behavior in SOFCs. Solid State Ionics In Press (2008). 47. Smith, J. R. 140 (University of Florida, Gainesville, 2007). 48. Gostovic, D., Smith, J. R., Kundinger, D. P., Jones, K. S. & Wachsman, E. D. Threedimensional reconstruction of porous LSCF cathodes. Electrochemical and Solid State Letters 10, B214-B217 (2007). 49. Barsoum, M. W. Fundamentals of Ceramics (ed. Goringe, B. C. a. M. J.) (IOP Publishing Ltd, Philadelphia, 2003). 50. Brunauer, S., Emmett, P. H. & Teller, E. Adsorption of gases in multimolecular layers. Journal of the American Chemical Society 60, 309-319 (1938). 51. Gostovic, D., O'Hara, K. A., Vito, N. J., Wachsman, E. D. & Jones, K. S. in Pacific Rim Meeting on Electrochemical and Solid-State Science (Electrochemical Society, Honolulu, Hawaii, 2008). 52. Gostovic, D., Vito, N. J., O'Hara, K. A., Jones, K. S. & Wachsman, E. D. Microstructure and connectivity quantification of comple x composite electrode three-dimensional networks. Journal of the American Ceramic Society Submitted (2009). 53. Stevenson, J. W., Armstrong, T. R., Carnei m, R. D., Pederson, L. R. & Weber, W. J. Electrochemical properties of mixed conducting perovskites La(1-x)M(x)Co(1y)Fe(y)O(3-delta) (M=Sr,Ba,C a). Journal of the Electro chemical Society 143, 2722-2729 (1996). 54. Wilson, J. R. et al. Quantitative three-dime nsional microstructure of a solid oxide fuel cell cathode. Electrochemistry Co mmunications 11, 1052-1056 (2009). 55. Genau, A. L., Voorhees, P. W. & Thornton, K. The morphology of topologically complex interfaces. Scripta Materialia 60, 301-304 (2009).

PAGE 184

184 56. Cairns, S. S. Introductory Topology (The Ronald Press Co., New York, 1961). 57. Dehoff, R. T., Aigeltin.Eh & Craig, K. R. Experimental Determination of Topological Properties of 3-Dimensional Microstructure s. Journal of Microscopy-Oxford 95, 69-& (1972). 58. Perna, A. et al. Topological efficiency in three-dimensional gallery networks of termite nests. Physica a-Statistical Mechanic s and Its Applications 387, 6235-6244 (2008). 59. Fouard, C. et al. in International Sym posium on Biomedical Imaging: From Nano to Macro" (ISBI'04) 4 (Arlington, USA, 2004). 60. Falamaki, C., Naimi, M. & Aghaie, A. Dual behavior of CaCO3 as a porosifier and sintering aid in the manufacture of alumina membrane/catalyst supports. Journal of the European Ceramic Society 24, 3195-3201 (2004). 61. Hellman, O. C., Vandenbroucke, J. A., Rusing, J., Isheim, D. & Seidman, D. N. Analysis of three-dimensional atom-probe data by the proximity histogram. Microscopy and Microanalysis 6, 437-444 (2000). 62. Marquis, E. A. & Gault, B. Determination of the tip temperature in laser assisted atomprobe tomography using charge state distributions. Journal of Applied Physics 104 (2008). 63. Kellogg, G. L. & Tsong, T. T. Pulsed-Laser Atom-Probe Field-Ion Microscopy. Journal of Applied Physics 51, 1184-1193 (1980). 64. Mayama, N., Yamashita, C., Kaito, T., Nojima, M. & Owari, M. Stress of needle specimen on the three-dimensional atom probe (3DAP). Surface and Interface Analysis 40, 1610-1613 (2008). 65. Cerezo, A., Smith, G. D. W. & Clifton, P. H. Measurement of temperature rises in the femtosecond laser pulsed three-dimensiona l atom probe. Applied Physics Letters 88 (2006). 66. Cerezo, A., Godfrey, T. J. & Smith, G. D. W. Application of a Position-Sensitive Detector to Atom Probe Micr oanalysis. Review of Scientific Instruments 59, 862-866 (1988). 67. Miller, M. K., Russell, K. F. & Thompson, G. B. Strategies for fabricating atom probe specimens with a dual beam FIB. Ultramicroscopy 102, 287-298 (2005). 68. Thompson, K. et al. In situ site-spe cific specimen preparation for atom probe tomography. Ultramicroscopy 107, 131-139 (2007). 69. Gorman, B. P., Norman, A. G. & Yan, Y. Atom probe analysis of III-V and Si-based semiconductor photovoltaic structures. Mi croscopy and Microanalysis 13, 493-502 (2007).

PAGE 185

185 70. Moore, J. S., Jones, K. S., Kennel, H. & Corcoran, S. 3-D analysis of semiconductor dopant distributions in a patterned struct ure using LEAP. Ultramicroscopy 108, 536-539 (2008). 71. Larson, D. J. et al. Focused ion-beam milling for field-ion specimen preparation:: preliminary investigations. U ltramicroscopy 75, 147-159 (1998). 72. Larson, D. J. et al. Advances in atom probe specimen fabrication from planar multilayer thin film structures. Microscopy and Microanalysis 7, 24-31 (2001). 73. Kelly, T. E. et al. First data from a co mmercial local electrode atom probe (LEAP). Microscopy and Microanaly sis 10, 373-383 (2004). 74. Soyez, G. et al. Grain-size-dependent thermal conductivity of nanocrystalline yttriastabilized zirconia films gr own by metal-organic chemical vapor deposition. Applied Physics Letters 77, 1155-1157 (2000). 75. Hasselman, D. P. H. et al. Thermal-Diffusi vity and Conductivity of Dense Polycrystalline Zro2 Ceramics a Survey. American Ce ramic Society Bulletin 66, 799-806 (1987). 76. Ma, W. et al. New Generation Perovskite Th ermal Barrier Coating Materials. Journal of Thermal Spray Technology 17, 831-837 (2008). 77. Thompson, K., Larson, D. J., Ulfig, R. M., Bunton, J. H. & Kelly, T. F. Analyzing Sibased structures in 3D with a laser-pulsed local electr ode atom probe. Solid State Technology 49, 65-+ (2006). 78. Wachsman, E. D. (United States Department of Energy, 2006). 79. Zaefferer, S. Application of orientati on microscopy in SEM and TEM for the study of texture formation during recrysta llisation processes. Icotom 14: Textures of Materials, Pts 1and 2 495-497, 3-12 (2005). 80. Dingley, D. J. & Randle, V. Microtex ture Determination by Electron Back-Scatter Diffraction. Journal of Materi als Science 27, 4545-4566 (1992).

PAGE 186

186 BIOGRAPHICAL SKETCH Danijel Gos t ovi was born in Ogulin, Croatia in 1981, within the Lika province, also the birthplace of Nikola Tesla. Exposure to solar he aters and wind turbines on his grandmothers farm sparked his interest in technology. His family immigrated to the city of White Plains, NY in February of 1990, months before the outbreak of the Yugoslavian Civil War. He went on to graduate from the White Plains Senior High School in 2000 with an Honors Regents Diploma. Danijel enrolled in Northwestern University in Evanston, Illinois in the Fall of 2000. A chance conversation resulted in Danijel enrolling in a Scanning Electron Microscopy (SEM) freshman engineering seminar class taught by Professor Vinayak P. Dravid. Th is class turned Danijel onto the materials science and engi neering major through hands-on SE M experiments. In 2003 he accepted a co-operative education extended inte rnship position with Functional Coating Technology, LLC. (FCT) of Evanston, IL, while stil l enrolled as a full time student. Danijel graduated with a Bachelor of Science., a Busi ness Basics Certificate and a Co-operative Education Certificate in 2004. Upon graduation Da nijel accepted a full time research engineer position with FCT where he learned a lot about SOFC engineering, accounting, and business management. In the Fall of 2005 he moved to Ga inesville, Florida in pursuit of a Master of Science and Doctor of Philosophy de grees at the University of Fl orida.. During his time at UF, Danijel was a founding member of the Electrochemical Society Stude nt Chapter, and served as its treasurer for three consecutive terms. He enjoyed the interdisciplinary nature of a UF graduate education which allowed him to collabora te with researcher in stitutions from around the world including EMPA (Switzerland), RIS Nati onal Laboratory (Denmark), Universit di Catania (Italy), University of North Texas (Den ton, TX), University of Alabama (Tuscaloosa, AL), University of Illinois (Urbana-Champaign, IL) and Siemens Energy, Inc. (Pittsburg, PA).