1 ENHANCED IONIC CONDUCTIVITY OF CERIA-BASED COMPOUNDS FOR THE ELECTROLYTE APPLICATION IN SOFCS By SHOBIT OMAR A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORID A IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008
2 2008 Shobit Omar
3 To my loving family
4 ACKNOWLEDGMENTS First of all, I would like to acknowle dge my advisor, Dr. J uan C. Nino, for his support and guidance. His faith in me helped me through the tough times in my research and was a constant sour ce of inspiration. His vi rtues of critical assessment and raising the bar for achievement helped me expand my potential. I would like to thank Dr. Eric D. Wachsman fo r his invaluable suggestions and fruitful discussions. I would also like to thank my other co mmittee members (Dr. Wolfgang Sigmund, Dr. Simon R. Phillpot and Dr. Mark Orazem) for their time, guidance and constructive comments. Further, I like to thank Dr. Jac ob L. Jones for the X RD discussions and Dr. Scott Perry for coming as a substitute in my PhD defense and for providing feedback about my research. I would also like to convey my heartfe lt thanks to Dr. Juan Ninos former and current group members: Samantha Yates, Lu Cai, Wei Qiu, Peng Xu, Mohammed Elshennawy, Satyajit Phadke, Laurel Wucherer, Marta Giachino, Kevin Tierney, Donald Moore, Louis Perez, Jillian Saxon and David Paredes fo r providing me excellent research environment and being hel pful. I gratefully acknow ledge all the fuel cell group members (Dr. Eric Wachsmans group) es pecially Doh Won Jung, Sean Bishop, Dr. Tak-keun Oh, Dr. Heesung Yo on, and Dr. Keith Duncan fo r the helpful materials science discussion. Special thanks go to Jin Soo Ahn for being a wonderful colleague and a friend. I had really enjoyed work ing with him for the last one year. I also wish to acknowledge all my friends at the University of Florida (UF) with whom I had a wonderful time in Gainesville. Sankara Sa rma Tatiparti, Rakesh Kumar Behera, Mahesh Tanniru, Kr ishna Prakash Ganesan (Machaa), Abhijit Pramanick,
5 Pankaj Nerikar, and Priyank Shukla, all of them had provided me a great company here in UF. Also, I like to thank my elder brother Ritesh Omar who spent numerous sleepless night worrying about my welfare. My sist er in law Smita Omar whose encouraging words had kept my enthusiasm up during the final stages of my Ph.D. work. And, last but not least, I express my sincere gratitude to my mot her Rekha Omar and father Rajendra Omar for always supp orting my decisions. Their love and belief in me has shaped me into who I am today. I cannot imagine myself having any of my success without my familys strong support behind me.
6 TABLE OF CONTENTS page ACKNOWLEDG MENTS .................................................................................................. 4 LIST OF TABLES............................................................................................................ 9 LIST OF FIGURES ........................................................................................................ 10 ABSTRACT................................................................................................................... 15 CHA PTER 1 INTRODUCTION.................................................................................................... 17 1.1 Statement of Problem and Mo tivation ............................................................... 17 1.2 Scientific Approach ........................................................................................... 19 1.3 Organization of th e Dissertation........................................................................ 22 2 BACKGRO UND ...................................................................................................... 24 2.1 Solid State Electrochem istry ............................................................................. 24 2.2 Fuel Cells .......................................................................................................... 24 2.3 Introduction to Solid Ox ide Fuel Cells............................................................... 26 2.4 Solid Oxide Electrolyte ...................................................................................... 29 2.4.1 Ceria Bas ed Materi als ............................................................................. 31 2.4.2 Ionic Conducti vity .................................................................................... 34 22.214.171.124 Temper ature depend ence .............................................................. 35 126.96.36.199 Oxygen partial pressure dependence ............................................ 37 188.8.131.52 Dopant c oncentrati on dependenc e ................................................ 39 2.5 Association of Defects ...................................................................................... 40 2.6 Impedance S pectroscopy ................................................................................. 41 3 MATERIALS AND EXPERIMENTAL PROC EDURES ............................................ 46 3.1 Materials Selection ............................................................................................ 46 3.2 Experimental Proc edures .................................................................................. 46 3.2.1 Powder Processing.................................................................................. 48 184.108.40.206 Conventional solid stat e route ........................................................ 49 220.127.116.11 Co-precipi tation route ..................................................................... 52 3.2.2 Forming................................................................................................... 55 3.2.3 Solid Stat e Sinter ing ................................................................................ 56 18.104.22.168 Conventi onal sint ering .................................................................... 56 22.214.171.124 Microwav e sinter ing ....................................................................... 57 3.2.4 Sample Preparation for Characte rization ................................................. 58 126.96.36.199 Ionic conducti vity ............................................................................ 58 188.8.131.52 X-ray diffrac tion .............................................................................. 58
7 184.108.40.206 Scanning elec tron micr os copy....................................................... 58 220.127.116.11 Transmission electron mi croscopy ................................................. 59 3.2.5 Impedance Spectrosc opy Experiment al Setup ........................................ 60 4 IONIC CONDUCTIVITY AND DOPING STRATEGIES FOR IONIC CONDUCTIVITY EN HANCEMENT ........................................................................ 62 4.1 Ionic Conductivi ty .............................................................................................. 62 4.2 Co-doping Strategies ........................................................................................ 65 4.2.1 Critical Dopant Ionic Radius .................................................................... 65 18.104.22.168 Theory ............................................................................................ 66 22.214.171.124 Materials select ion ......................................................................... 67 126.96.36.199 Results and discu ssion .................................................................. 68 188.8.131.52 Summary and conclusion............................................................... 74 4.2.2 Density Functional Theory Calculation .................................................... 74 184.108.40.206 Theory ............................................................................................ 75 220.127.116.11 Materials select ion ......................................................................... 78 18.104.22.168 Results and discu ssion .................................................................. 79 22.214.171.124 Summary and conclusion............................................................... 95 4.3 Processing Effects on Ionic Conductivity .......................................................... 96 5 STRUCTURE-PROPERTY RELATIONSH IPS IN DOPED CERIA MATERIAL S AT HIGH TEM PERTATURES................................................................................. 99 5.1 Introdu ction ....................................................................................................... 99 5.2 Crystal Structure of Trivalent Acceptor Doped Ceria at High Temperature ..... 106 5.2.1 High Temperatur e X-Ray Diffraction...................................................... 106 5.2.2 Lattice Strain ......................................................................................... 108 126.96.36.199 Thermal st rain .............................................................................. 108 188.8.131.52 Chemic al strain ............................................................................ 114 5.3 Ionic Conductivity of Trivalent Acc eptor Doped Ceria at High Temperatures 122 5.3.1 Activation Energy .................................................................................. 128 5.3.2 Pre-Exponent ial Coe ffici ent ................................................................... 133 5.4 Revisiting Critical Dopant Ioni c Radi us ........................................................... 136 5.5 Summary and Conclusi ons ............................................................................. 140 6 PROCESSING EFFECTS ON THE IONIC CO NDUCTIVITY ............................... 143 6.1 Introdu ction ..................................................................................................... 143 6.2 Experimental Proc edures ................................................................................ 148 6.3 Results and Discu ssion ................................................................................... 150 6.3.1 Micros tructure ........................................................................................ 150 6.3.2 Ionic Conducti vity .................................................................................. 154 6.4 Summary and Conclusi ons ............................................................................. 158
8 7 BUTTON CELL PERFOR MANCE T ESTING ........................................................ 160 7.1 Introdu ction ..................................................................................................... 160 7.2 Experimental Procedure ................................................................................. 160 7.3 Results and Discu ssion ................................................................................... 166 7.3.1 Partic le Size .......................................................................................... 166 7.3.2 Microstruc tural A nalysi s ......................................................................... 168 7.3.3 Impedanc e Analysis .............................................................................. 169 7.3.4 Power Density ....................................................................................... 171 7.4 Summary and Conclu sion ............................................................................... 173 8 SUMMARY AND FU TURE WORK ....................................................................... 175 8.1 Su mmary ........................................................................................................ 175 8.2 Future Work .................................................................................................... 180 8.2.1 Doping Strategy ..................................................................................... 180 8.2.2 Ionic Conducti vity .................................................................................. 181 8.2.3 Proc essing ............................................................................................. 182 8.2.4 Perform ance Test ing ............................................................................. 183 A DEFECT CHEMISTRY AND THERMO DYNAMIC IN P OINT DEFECTS .............. 185 B IONIC CONDUCTIVITY AND DEFECT CO MPLEXES ......................................... 187 B.1 Oxygen I on Conducti vity ................................................................................. 187 B.2 Defect Complex es.......................................................................................... 188 B.2.1 Dimers................................................................................................... 189 B.2.2 Tr imers.................................................................................................. 190 C HIGH TEMPERATURE XRD PATTE RNS OF CERI A COMPOUNDS .................. 192 D EXTRAPOLATION METHOD TO CA LCULATE LATTICE PARAMETER ............ 195 LIST OF RE FERENCES ............................................................................................. 197 BIOGRAPHICAL SKETCH .......................................................................................... 207
9 LIST OF TABLES Table page 2-1 Typical characteristics of various F uel ce lls. ....................................................... 25 2-2 2OP of doped ceri a system s................................................................................ 38 4-1 Comparison of the H and log o for Gd0.10Ce0.90O2, Sm0.05Nd0.05Ce0.90O2. and Sm0.075Nd0.075Ce0.85O2measured below 550oC. Further, of these oxide materials at 550oC is also reported........................................................... 91 5-1 Activation energies for ZrO2 and CeO2 systems doped with Ca2+.................... 100 5-2 Fitted (331) XRD peak profile parameters for Gd0.10Ce0.90O2at 400oC, measured while the furnace is heating up and cooling down........................... 105 5-3 Thermal expansion coefficient cons tants for different doped ceria systems. .... 110 5-4 Thermal expansion coe fficient constants for Smx /2Nd x /2Ce1xO2. ................... 113 5-5 Relationship between lattice par ameter and dopant ionic radius for D0.10Ce0.90O2at different temperat ures........................................................... 117 5-6 Relationship between lattice parameter and dopant content ( x ) for Smx /2Ndx /2Ce1xO2system at higher temperatures.......................................... 120 5-7 Critical dopant ionic radius for trival ent acceptor dopant cation for host ceria, for different te mperatur es. ................................................................................ 138 6-1 Abbreviation of samples synthes ized using different tec hniques...................... 150 6-2 Comparison of the grai n ionic conductivity of Sm0.075Nd0.075Ce0.85O2and Gd0.10Ce0.90O2at 600oC.................................................................................. 157 7-1 Comparison between the total ASR obtained from I-V characteristic and impedance meas urement................................................................................. 171
10 LIST OF FIGURES Figure page 1-1 Schematic of a so lid ox ide fuel ce ll..................................................................... 18 1-2 Ionic conductivity of fl uorite structur e ox id es...................................................... 20 2-1 Cubic fluorite st ruc ture of ceria........................................................................... 32 2-2 Oxygen vacancy jump in doped ceri a system .................................................... 33 2-3 Ionic conductivity as a function of dopant concentration in doped zir conia systems.............................................................................................................. 39 2-4 Oxygen vacancy trapped in the tetrahedr al site.................................................. 40 2-5 Typical impedance response for a given relaxati on process.. ............................ 42 2-6 Typical impedance spectrum of an electroded polycrystalline ceramic material. T he equivalent circuit to fit the spectrum is also shown...................... 44 3-1 Flow chart for the powder synthesis using c onventional solid state reaction and different characteri zation tec hniques........................................................... 47 3-2 XRD patterns of Lu0.10Ce0.90O2calcined at different temperatures for 10 h...... 50 3-3 XRD patterns of different doped ceria systems calcined at 1450oC for 10 h...... 51 3-4 Particle size distribution of Sm0.075Nd0.075Ce0.85O2synthesized using solid oxide r oute.......................................................................................................... 52 3-5 XRD pattern of Sm0.075Nd0.075Ce0.85O2synthesized using co-precipitation techni que............................................................................................................ 54 3-6 Particle size distribution of Sm0.075Nd0.075Ce0.85O2synthesized using coprecipitation techni que........................................................................................ 55 3-7 Sample placed in the therma l pod for microw ave sint ering. ............................... 57 3-8 AC impedance spectrosc opy experiment al setup. .............................................. 60 4-1 Grain ionic (GI) conductivity comparison of D0.10Ce0.90O2at 400oC.................. 67 4-2 XRD profiles of LuxNdyCe1x-yO2and Gd0.056Y0.044Ce0.90O2taken at room temperat ure........................................................................................................ 69 4-3 XRD profiles of D0.10Ce0.90O2taken at room temperature................................. 70
11 4-4 Elastic strain as a func tion of dopant concentration for Smx /2Ndx /2Ce1xO2and NdxCe1xO2. ............................................................................................... 71 4-5 Arrhenius plot for the bulk ionic conductivity of Gd0.10Ce0.90O2, Lu0.10Ce0.90O2, and Lu0.054Nd0.046Ce0.90O2systems in air................................ 73 4-6 Total interaction energy between dopant cation and oxygen vacancy sitting in nearest neighbor (NN) and next to neares t neighb or (NNN) site (of dopant cations) for different trivalent cations.................................................................. 76 4-7 Cubic fluorite structure of doped ce ria, where oxygen vacancy reside in nearest neighbor (NN) site A) and next to nearest neigh bor (NNN) site B) of dopant ca tion...................................................................................................... 77 4-8 Scanning electron microgr aph of the surface of Sm0.05Nd0.05Ce0.90O2sintered pellet..................................................................................................... 79 4-9 XRD patterns of Smx /2Ndx /2Ce1xO2measured at room temperature................ 80 4-10 Shift in (111) and (200) XRD peaks position of Smx /2Ndx /2Ce1xO2with increase in dopant concentration (x)................................................................... 81 4-11 Lattice parameter as a functi on of dopant concentration (mol%) for Smx /2Ndx /2Ce1xO2and NdxCe1x O2................................................................. 82 4-12 Typical impedance spectrum of a polycrystalline Sm0.075Nd0.075Ce0.85O2ceramic measured in air at 300oC...................................................................... 84 4-13 Arrhenius plots for the gr ain ionic conductivit y of Smx /2Ndx /2Ce1xO2................ 85 4-14 Log o as a function of activation energy measured below 475oC for Smx /2Ndx /2Ce1xO2............................................................................................. 86 4-15 Grain ionic conductivity as a function of x, dopant concentration (mol%) in Smx /2Ndx /2Ce1xO2and GdxCe1xO2................................................................. 88 4-16 Arrhenius plot for the gr ain ionic c onductivity of Gd0.10Ce0.90O2, Sm0.075Nd0.075Ce0.85O2and Sm0.05Nd0.05Ce0.90O2............................................ 89 4-17 Grain ionic conductivity of Smx /2Ndx /2Ce1xO2as a function of x, dopant concentrati on (mol%).......................................................................................... 92 4-18 Activation energy for oxygen ion diffusion in Smx /2Ndx /2Ce1xO2measured below 475oC and above 475oC, as a function of x .............................................. 93 4-19 Grain ionic conductivity comparison of doped ceria systems at 600oC between our work and literat ure......................................................................... 97
12 5-1 The (331) peak in XRD profiles of Gd0.10Ce0.90O2measured at 400oC in different rampi ng condit ions.............................................................................. 105 5-2 XRD profiles of measured Sm0.10Ce0.90O2at different temperature in air....... 106 5-3 XRD profiles of measured Lu0.10Ce0.90O2at different temperature in air......... 107 5-4 Thermal expansion of D0.10Ce0.90O2ceramic systems (where D3+ = Lu3+, Yb3+, Er3+, Dy3+, Gd3+, Sm3+, Sm3+/Nd3+, Nd3+) in air........................................ 109 5-5 Thermal expansion coefficient (at 600oC) as a function of dopant ionic radius for D0.10Ce0.90O2systems............................................................................... 111 5-6 Thermal expansion of Smx /2Ndx /2Ce1xO2ceramic systems (where x = 0.01. 0.03, 0.05, 0.08, 0.10, 0.12, 0.15, and 0.20) in air............................................ 112 5-7 Thermal expansion coefficient ( ) at 600oC for Smx /2Nd x /2Ce1xO2as a function of dopant concentra tion....................................................................... 114 5-8 Lattice expansion of D0.10Ce0.90O2systems as a function of dopant ionic radius................................................................................................................ 116 5-9 Lattice expansion in Smx /2Ndx /2Ce1xO2system as a function of dopant concentration at differ ent temper atures............................................................ 119 5-10 Vegards slope as a func tion of temperature for Smx /2Ndx /2Ce1xO2system. 120 5-11 Grain ionic conductivity comparison of different doped ceria materials (with 10 mol% dopant content) .................................................................................. 125 5-12 Grain ionic conductivity as a function of dopant concentration and temperat ur e...................................................................................................... 127 5-13 Normalized grain ionic conductivity as a function of dopant concentration and temperat ur e...................................................................................................... 128 5-14 Activation energy for oxygen vacancy diffusion in doped ceria systems with 10 mol% dopant. ............................................................................................... 130 5-15 Migration and Association enthalpies for oxy gen vacancy diffusion determined from the Arrhenius plot of doped ceria systems............................. 131 5-16 Pre-exponential coeffi cient in the Arrhenius ionic conductivity plot of D0.10Ce0.90O2systems..................................................................................... 134 5-17 Lattice parameter mismatch between D0.10Ce0.90O2and pure ceria, as a function of dopant cation ionic radius at 500oC. The grain ionic conductivity of D0.10Ce0.90O2at 500oC is also shown.......................................................... 139
13 6-1 A) Continuous layer of an silic eo us phase in the grain boundary of Gd0.10Ce0.90O2B) Glassy phase formed at the three-grain junction in Sm0.05Nd0.05Ce0.90O2....................................................................................... 145 6-2 Comparison of the tem perature time profi le in conventional solid state sintering and micr owave sint ering.................................................................... 149 6-3 SEM images of Conventi onal sintering + Solid oxide route, and Mi crowave sintering + Co-precipitation te chnique.............................................................. 151 6-4 TEM images of the microwave sintered Sm0.075Nd0.075Ce0.85O2sample......... 151 6-5 Grain boundary in the polycrystalline ceramic synthesized using microwave sintering and conventional solid state sinter ing................................................ 152 6-6 Electron dispersive X-ray spectrum of SamCoSo sample................................... 153 6-7 Impedance spectrum of SamMiCo and SamCoSo samples at 350oC, in air.......... 154 6-8 Arrhenius plot for t he ionic conductivity of S m0.075Nd0.075Ce0.85O2and Gd0.10Ce0.90O2materials................................................................................. 156 7-1 Schematic of anode-supported prototype solid ox i de fuel cell.......................... 161 7-2 Green tape of NiO-GDC and the circular punched tape. .................................. 162 7-3 Flow chart for the fabrication of anode-supported pr ototype SOFC. ................. 164 7-4 Experimental setup for t he I-V characteristics measur ement of the test SOFC sample. ............................................................................................................. 165 7-5 Particle size distribution of Sm0.075Nd0.075Ce0.85O2synthesized using coprecipitation technique and La0.6Sr0.4Co0.2Fe0.8O3.......................................... 167 7-6 FE-SEM images of cross-section vi ew of electrodes and electrolyte, and surface of the electrolyte. ................................................................................. 169 7-7 Impedance spectrum of t he SOFC cell measured at di fferent temperatures. ... 169 7-8 Area specific resistance at differe nt te mperatur es............................................ 170 7-9 The I-V characteristics of the prototype SOFC sample with Sm0.075Nd0.075Ce0.85O2electrolyte at va rious tem peratures............................ 172 C-1 XRD profiles of pure ceria collected at high temperatures. ............................... 192 C-2 XRD profiles of Yb0.10Ce0.90O2collected at high temperatures....................... 192 C-3 XRD profiles of Y0.10Ce0.90O2collected at high temperatures......................... 193
14 C-4 XRD profiles of Dy0.10Ce0.90O2collected at high temperatures....................... 193 C-5 XRD profiles of Gd0.10Ce0.90O2collected at high temperatures....................... 194 C-6 XRD profiles of Sm0.10Ce0.90O2collected at high temperatures...................... 194 D-1 Plot of calculated lattice parameter, a with respect to cos2 for pure ceria at different tem peratures...................................................................................... 196
15 Abstract of Dissertation Pr esented to the Graduate School of the University of Florida in Partial Fulf illment of the Requirements for t he Degree of Doctor of Philosophy ENHANCED IONIC CONDUCTIVITY IN CERIA-BASED COMPOUNDS FOR THE ELECRTROLYTE APPLICATION IN SOFCS By Shobit Omar August 2008 Chair: Juan C. Nino Major: Materials Science and Engineering Higher ionically conductive solid oxide electrolytes operating at intermediate temperatures (400 to 700C) ar e critical for the application of portable power generation device based on solid oxide fuel cells (SOFCs ). Doped ceria materials show promise due to their higher ionic conductivity, and good thermodynamic stability in the intermediate temperatur e range. Among the doped ceria electrolytes, Gd0.10Ce0.90O2(GDC) is widely accepted to exhibit the high est ionic conductivity. To further enhance the ionic conductivity in ceria in this work, novel co-doping strategies are tested. Codoping based on Sm3+ and Nd3+ as dopant cations resulted in an increase in the ionic conductivity. It is shown that at 550oC, the ionic conductivity of Sm0.075Nd0.075Ce0.85O2is observed to be 30 % higher than that of GDC. From the literature, it is known that the ionic conducti vity depends upon the processing variables. Thus, to have a clear understanding of t he ionic conductivity trend as a function of dopant type, a consistent set of ionic conductivity data is developed for different doped ceria materials synthesized under similar processing conditions. It was observed that for single dopants, Nd3+ is the best dopant cation for host ceria.
16 The critical dopant ionic radius ( rc) was determined for trivalent doped ceria system at high temperatures. The elastic strain and th e ionic conductivity of doped ceria systems measured at 500oC, show that the elastic strain ionic conductivity relationship based on rc concept is not a successful strategy in doped ceria. Further, the effect of processing condition on Sm0.075Nd0.075Ce0.85O2materials is tested. Co-doped ceria is synthesized using di fferent processing routes. It is observed that the grain ionic conductivity of Sm0.075Nd0.075Ce0.85O2synthesized using coprecipitation technique and microwave sintering is around 45 % higher than that of GDC at 550oC. Finally, the performance of the anode-supported prototype SOFC using Sm0.075Nd0.075Ce0.85O2as an electrolyte is analyzed at intermediate temperatures. An exceptionally high powe r density of 1.38 W/cm2 is obtained at 650oC, using 90 cm3/minute of dry air and wet hydrogen in ca thode and anode sides, respectively. The power density of a test ed cell shows that the Sm0.075Nd0.075Ce0.85O2ceramic electrolyte can successfully generate high er performance in SOFCs working in the intermediate temperature range.
17 CHAPTER 1 INTRODUCTION 1.1 Statement of Problem and Motivation The continuous increase in demand for f uels, limited oil reso urces and increas ing global warming requires for the use of renewable sources of energy.1,2 These renewable sources include so lar, biomass, hydropower, geothermal, and wind energy.3 Research on the use of thes e sources mainly focuses on the development of new technologies (e.g., fuel cells, batteries, sola r cells, windmills etc.) that can efficiently create cleaner sustain able energy. Among them, fuel ce ll technology shows significant promise.4 Development and commercialization of this technology would be a major step forward towards the goal of sustainabl e energy production. Among various fuel cells, solid oxide fuel cell (SOFC) is consider ed by many to be the most desirable fuel cell for generating electricit y from hydrocarbon fuels.5 This is because they are highly efficient, environmentally friendly, corrosion resistant, and have the potential to operate directly on existing transportation fuel s ranging from gasoline to jet fuels.6 Figure 1-1 shows the main components of SOFC, i.e., cathode, electroly te and anode. The operation of the devic e will be described in detai l in chapter 2. Although SOFC offers a lot of promise in energy-conversion system, the current technology, based on a stabilized ZrO2 electrolyte, must operat e in the region of 1000oC to avoid unacceptable high ohmic losses.7 The high operating temper ature requires the use of selected materials to withstand such extreme conditions. If the o perating temperature can be lowered to 400-700oC, which is termed the intermedi ate temperature (IT) range, it will open up the possibility for a wide range of materials in SOFC. It will not only improve the reliability (i.e., hours of operation) of the device, but also lower the cost and
18 time it takes to heat up to the operating temper ature. A main goal of the Department of Energy is to accelerate the development of SOFCs and get them commercialized as quickly as possible making them affordable option for power generation. Extensive research has been performed through the Soli d State Energy Conversion Alliance (SECA) program which include an industria l team, a core technology program team, and federal government experts.8 The main target of this program is to place SOFCs into market by 2010 at a cost of nearly $400/kW. These f uel cells will be 3-10 kW in size, adaptable for various applications.8 Figure 1-1. Schematic of a solid oxide fuel cell. However, with the current st ate-of-the-art SOFC material s, lowering the operating temperature increases the ohmic losses and electrode polarizations. This has a detrimental effect on the performanc e and efficiency of the device. Thus, there is a need to develop materials for all the co mponents of SOFCs that show improved properties in the IT range. In the present wo rk, research is focused on developing solid
19 oxide electrolyte materials with enhanced ioni c conductivity for the IT application of SOFCs. 1.2 Scientific Approach A summary of the ionic conduc tivity of oxide materials are shown in Figure 1-2.9 All these ionically conductive materials exhibi t the cubic fluorite crystal structure. Among the oxygen ion conductors, bismuth ox ide based materials show the highest ionic conductivity in IT range. However, these materials undergo an orderdisorder transition at around 600oC, which limits their application as solid oxide electrolytes. Although doped zirconia materials are stable at intermediate temper atures, their ionic conductivity is comparatively lower than other candidate materials. In recent years, SOFC electrolyte research primarily has focused on doped ceria.10-12 Doped ceria materials exhibit higher ionic conductivity than doped zirconia, and show better thermodynamic stability than doped Bi2O3 in the IT range. Among doped ceria materials, Gd-doped ceri a exhibits the highest ionic conductivity.13 The main objective of this dissertation is to identify dopants and doping strategies that can further enhance the ionic conductivity of ceria. Numerous works have been performed in understanding the effect of different acceptor dopants on the ionic conductivity in ceria.10,14 Since processing variables affect the ionic conductivity, a wide range of ionic conducti vity data is reported for a given composition in doped ceria.7,15 Thus, a clear understanding of the ionic conductivity trend as a function of dopant type is still lacking. To overcome this, a consistent set of i onic conductivity data will be developed for different doped ceria materials synthesized under similar processing conditions. The
20 comparison of this consistent data will a llow selecting the potential acceptor dopants that can enhance the ionic c onductivity in ceria. 0.60.70.80.91.01.11.21.3 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 1312111098765 Bi2O3( H f O2)0 8 8( C a O )0 1 2( T h O2)0 9 3( Y2O3)0 .0 7( Z r O2)0 8 7( C a O )0 1 3( L a2O3)0 9 5( S r O )0 0 5( Z r O2)0 9 0( Y2O3)0 1 0( Z r O2)0 9 0( S c2O3)0 1C e0 8 0G d0 2 0O1 9Log(S.cm-1)1000/T (K-1)( B i2O3)0 7 5( Y2O3)0 2 5T/100 (oC) Figure 1-2. Ionic conductivity of fluorite structure oxides.9 In the literature, co-doping in ceria is well established to increase the ionic conductivity.16 However, in the previous work, no systematic rationales behind selecting acceptor dopants were provided.17-20 Experiment will be conducted to test the novel codoping strategies for a series of acceptor dopants, which will be selected based on their ionic radius. After identifying the co-dopants, the effect of dopant concentration on the ionic conductivity will be examined.
21 The structure (elastic strain) property (ionic conductivity) relationship in the literature for doped ceria shows Gd3+ as the ideal dopant since it create the minimum elastic lattice strain.21,22 It is difficult to compare both aspects as the ionic conductivity is measured at intermediate temperatures, and the elastic strain is determined at room temperature. Therefore, to perform a fair comparison between st ructure and property, for all the doped ceria materials, both the ioni c conductivity and elastic strain will be measured at high temperatures. After identifying the dopants and optimizing the dopant concen tration, the effect of different processing variables on the ionic conductivity will be performed. The higher ionically conductive material will be synthes ized using varying processing routes, to obtain different ceramic microstructures. The effect of fast firing using microwave sintering on the ionic conductivity of novel ceramic material will be studied. Finally, the potential of the novel co-doped ceria as an elec trolyte will be tested in a real SOFC environment. Anode-supported SOFC prot otypes based on the novel electrolyte will be fabr icated. The power density measurem ent of the SOFC samples will be performed using current-voltage characte ristics at intermediate temperatures. Different characterization techniques will be utilized to ac hieve all the above objectives. The ionic conductivity measur ement will be perform ed using ac-impedance spectroscopy technique inside the quartz reac tor at high temperatur es, in air. High temperature X-ray diffraction will be used to characterize the crystal structure parameters of doped ceria mate rials. Microstructural analysis will be performed using scanning electron microscopy and transmission electron microscopy.
22 1.3 Organization of the Dissertation Chapter 2 presents a brief introduction on the solid state electrochemistry, solid oxide fuel cells, solid oxide electrolytes, and the main requirements for the material to be used as an electrolyte in SOFC environment. This general crystall ography of fluorite structure, ionic conduction in doped fluorite oxides, and the ionic conductivity dependence on temperature, dopant concentration, and the oxyg en partial pressure are also covered. In addition, a brief overview of impedance spectroscopy technique is also presented. In Chapter 3, ma terials selection, details of the experimental procedures and the various characterization techniques used are covered. Chapter 4 discusses the doping strategies that are tested on ceria, in order to enhance the ionic conductivity of the materi als. This chapter presents the ionic conductivity results of novel ceramic electrolyte materials, and their comparison to Gd0.10Ce0.90O2electrolyte. In Chapter 5, a consistent set of ionic conductivity data at intermediate temperatures, for different doped ceria syst ems synthesized under similar experimental conditions, is presented. In addition, crystal lattice elastic strain for different doped ceria systems at higher temperatures and its comparison with the io nic conductivity at the corresponding temperature are de scribed. Further, the relati on between critical dopant ionic radius and the ionic conductivity of doped ceria system is revisited. After understanding the structure and property in doped ceria systems, the effects of processing on the ionic conductivity are also co vered in Chapter 6. Different processing routes are used to synthesize phase pure samp les of our novel ceramic electrolyte. The effects of processing variables on the microstructure, and ionic conductivity are described in this chapter.
23 Chapter 7 presents in detail the experim ental setup and the fabrication of anodesupported prototype of SOFC. Further, power density result s of SOFC device using the novel ceramic electrolyte are discussed. Chapter 8 presents the summary of the dissertation and the areas for future work in doped ceria materials. At the end of the thesis, four appendices present the basic information about the defect chemistry, and thermodynamic of point defec ts (Appendix A), ionic conducti vity and defect complexes (Appendix B), X-ray diffraction profile for various doped ceria materials collected at different temperatures in air (Appendix C), and ex trapolation method to estimate lattice parameter of cubic struct ure material (Appendix D).
24 CHAPTER 2 BACKGROUND 2.1 Solid State Electrochemistry Electrochemistry is the branch of chemis try that covers all the phenom ena in which a chemical change is the result of electrical force and, vice versa, where an electric force is generated as a result of chemical processes. When associated with the chemistry and physics of solid state compounds, it is called solid stat e electrochemistry. Solid state electrochemistry involves th e study of thermodynamics, kinetics and mechanism of electrochemical reactions in solid state materials. It is a well established field which finds lot of applications in various fields. Electrochemical gas sensors, fuel cells, batteries, and electrochromic devices ar e few areas where the knowledge of solid state electrochemistry is exploited.23 Further, it can be divided into two main subjects, ionics and electrodics. In ionics, the emphasis is on the properties of electrolyte materials, while electrodics involves underst anding electrode reactions. Both subjects require fundamental understanding of structural and defect chemistry, diffusion and transport in solids, conductivity and electrochemical reactions, and adsorption and reactions on solid surfaces.24 The basic information about the defect chemistry, thermodynamic barriers, and diffusion proce sses in solid materials are provided in Appendix A and elsewhwere.24 2.2 Fuel Cells The concept of generating electrical power from a simple electrochemical c ell was first demonstrated by Sir William R. Grove in 1839.25 At that time, the electrochemical device was known as Grove cell instead of f uel cell. The Grove cell provided nearly double the voltage of the Daniel l cell which was the major de vice in the early years of
25 batteries. It was believed that a power generation using hydrogen as a fuel could replace coal, the primary source of energy at that time. T he term fuel cell wasnt used until 1889, when scientists Ludwig Mond and Charles Langer attempted to build the first practical device using hydrogen to produce el ectricity. In 1959, Bacon was able to assemble a stack of 40 fuel cells wh ich exhibited a useful power density.26 Bacons cell, which was modified by Pratt and Whitney became the onboard power system for the Apollo space vehicle that enabled astronauts to land on the moon in 1969. Today, scientists and technologists all over the wo rld are working to advance the fuel cell technologies for use in everyday appliances, such as computers and cell phones as well as to power homes, office buildings, and even vehicles. Table 2-1. Typical characteri stics of various Fuel cells.6 Type of Fuel Cell Typical Electrolyte Charge Carrier Operating Temperatures (oC) Efficiency (%) SOFC Y Stabilized ZrO2, Gd Doped CeO2, Er Stabilized Bi2O3 O21000 600-800 600 60-70 PEMFC Perfluorosulfonic Acid H+ 50-100 35-40 AFC KOH (aqueous) OH70-100 60 MCFC Li2CO3-K2CO3 (molten) CO3 2650 50-60 PAFC H3PO4 H+ 200 40-45 A fuel cell can be described as an electr ochemical conversion device that uses oxygen and hydrogen or hydrocarbons to generat e electricity with t he only byproducts being water and heat (shown in Figure 1-1). Like a simple battery it consists of an electrolyte sandwiched between two el ectrodes (a porous anode and cathode).5 However, while a battery only stores electricit y, a fuel cell produces electricity and does
26 not need recharging. Theoretically, as long as fuels (i.e., hydrogen or hydrocarbons, and oxygen) are supplied, it will continuously generate electricity. Fuel cells are generally classified by the type of electrolytes incorporated within the fuel cell: e.g., alkaline fuel cell (AFC ), polymer electrolyte membrane fuel cell (PEMFC), phosphoric acid fuel cell (PAFC), molten carbon ate fuel cell (MCFC) and solid oxide fuel cell (SOFC).27 For PEMFCs and PAFCs, protons move through the electrolyte towards the cathode where t hey combine with oxygen and electrons, to produce water and heat. For AFCs, MCFCs and SOFCs, negative ions diffuse through the electrolyte toward the anode to oxid ize the hydrogen fuel to produce water and electrons. The electrons move through the anode to the external circuit and back to the cathode, providing a source of useful elec trical energy in an ex ternal circuit. The general characteristics of these fuel cells are given in Table 2-1.6 Further useful technological information and their current applic ation status is publis hed in a variety of books, review papers, and websites.28-31 Fuel cells offer the promise of higher e fficiency in energy conversion and electric power generation for many different applicat ions. However, as main incentive for investing in the fuel cell research and development is the substantially reduced environmental impacts if the long-term ener gy scenario based on renewable energy source, i.e., hydrogen can be realized. 2.3 Introduction to Solid Oxide Fuel Cells As the name implies, SOFCs uses an ion c onducting solid-oxide or ceramic as an electrolyte. The first ceramic solid-oxide electrolyte was discovered by Nernst in 1899.5,32 The first successful operation of a ceramic fuel cell at 1000oC was demonstrated by Baur and Preis in 1937.33 Since that time, SOFC technology has
27 made excellent technical progress. Fuel cells based on the state of the art yttria stabilized zirconia electrolyte, have been operated for thousand of hours showing excellent performance. In the last few decades, SOFC research and development received significant attention which shows t he widening interest in this technology. Among various types of fuel cell technolog ies, SOFC is considered to be one of the most desirable fuel cell for generating electricity from hydrocarbon fuels. This is because of their ability to use currently ava ilable fossil fuels, thus reducing production cost. SOFCs have so far been operated on methane, propane, butane, fermentation gas, gasified biomass and paint fumes with minimal fuel processing.5 Other fuel cell technologies require pure hydrogen as their f uel. This makes them highly fuel flexible and far superior to other technologies su ch as PEMFC, MCFC. In addition, SOFCs have the highest efficiencies of all fuel ce lls and, potentially, a long life expectancy.34 The typical efficiency when used separately is around 40-60%. However, this figure can reach 70 % when the hot exhaust of the cell is used in a hybrid combination with gas turbines. Further, th e high operating temperature of SOFCs (~1000oC) eliminates the need for expensive Pt catalysts, which is the case for PEMFC or ot her low temperature fuel cells. Thus, they do not get poisoned by carbon monoxide. In addition, their high operating temperature allows di rect use of natural gas, thus eliminating the need for an expensive, external reformer. Furthermore, because SOFC is a two phase gas/solid system, many of the problems associated with liquid electrolytes such as corrosion, flooding, electrolyte distribution, and maintain ing a stable triple p hase boundary region are eliminated. They are attractive as an energy source because they are clean, reliable and environmentally frie ndly. Emission of pollutants from fuel cells (such as
28 CO, NOx, SOx) are several orders of magnitude lower than those produced by conventional combustion engin es. Finally, the vibrat ion-free operati on of SOFCs eliminates the noise associ ated with conventional power gener ation systems. For these reasons, SOFCs may be considered to be The Future of Power Generation. Figure 1-1 shows a schematic of a ty pical SOFC. The dense solid oxide electrolyte is sandwic hed bet ween two porous electrodes ( anode and cathode). Each component serves several functions in the cell, and meets certain requirements (such as conductivity, chemical and structural stability, compatibi lity, and mechanical properties) for a particular condition.35 The operation of SOFCs involves oxidation of the fuel at the anode, and reduction of the oxidant at the cat hode. Air flows along the cathode side. When an oxygen molecule contacts the cathode/ electrolyte interface, it splits into two oxygen ions. These oxygen i ons diffuse into the electrolyte material and migrate to the other side of the cell where they encounter the anode. The oxygen ions encounter the fuel at the ano de/electrolyte interface and react, producing water, carbon dioxide, heat, and most importantly electrons. These el ectrons transport through the anode to the external circuit and back to the cathode, providing a source of useful electrical energy in the form of direct current in an external circuit. The open circuit voltage ( Eo) is given by the Nernst equation, anodepO cathode pO F RT Eo 2 2ln 4 (2-1) where R T F and the pO2s are the universal gas cons tant, absolute temperature, Faradays constant, and the oxygen partial pressures, respectively. Under cell operating conditions, i.e., when a current passes through it, the cell voltage ( E ) is given by,
29 CA oIREE (2-2) where I is the current passing through the cell, R is the resistance of the cell, and A and C are the voltage losses due to the polarization associated with anode and cathode, respectively. For a high power density SOFC the resistive losses and the electrode polarizations need to be minimized. The amount of power produced by a fuel cell depends upon several factors, such as fuel cell type, cell size the temperature at which it operates, and the pressure at which the gases are supplied to the cell.36 Still, a single fuel cell produces enough electricity for only the smallest applications. Therefore, in order to generate sufficient volt age, individual fuel cells are typicall y connected in stacks and electrically connected in series through in terconnects. A typical fuel cell stack may consist of hundreds of fuel cells. 2.4 Solid Oxide Electrolyte In 1899, W. Nernst32 suggested that ZrO2 is a solid conductor for oxygen ions. However, the explanation for the conduction mechanism in this mate rial was first given by C. Wagner37. Doping of ZrO2 with different acceptor metal cations results in oxygen vacancies which are the charge carriers in this material. The acceptor doped ZrO2 is typically used as a solid oxide electrolyt e in sensors for measuring oxygen partial pressures and in fuel cells and electrolyzers. In order to reduce the ohmic losses, these devices are typically used at higher temperatures (~1000oC). Thus, the main requirements for good solid oxide electrolytes are mainly fixed by the high operating temperature which dictates cons traints of different types, 1. High ionic conductivity (> 0.1 S.cm-1 at 1000oC). 2. Low electronic transference number (< 10-3).
30 3. Large electrolytic domain. 4. Chemically stable under extreme conditions i.e., lo w reducing atmosphere (10-22 atmosphere) and high temperatures (1000oC). 5. Good phase stability and a good match of thermal expansion coefficient with other cell components. 6. High mechanical strength i.e., fracture toughness (> 400 MPa). In an effort to lower the operating tem perature of SOFCs, numerous research groups have focused on developing materials wi th higher ionic conductivity than doped ZrO2. Higher ionically conduct ive electrolytes are not only important for SOFC applications but also play a vital role in chemical processing, oxygen and hydrogen generators, electrochemical gas sensors, and combustion control. High oxygen diffusivity has been observed in oxides which ex hibit large tolerance for atomic disorder. Several investigations have focused on cubic fluorite structured oxides due to their relatively open structure. The most succ essful oxide electrolytes have been those based on group IVB oxides (i.e., ZrO2, HfO2, CeO2 or ThO2) with additions of either an alkaline earth oxides (e.g. CaO or SrO), Sc2O3, Y2O3 or a rare earth oxides.38 Optimization of these materials has led to 8 mol% yttria-stabilized zirconia (YSZ) as the favored electrolyte for solid oxi de fuel cells (SOFCs) application.39 Although YSZ is purely an ionic conductor even at reduc ing atmospheres, its high operational temperature restricts its usage to multi-MW SOFC systems (e.g.: electrical generation plants). For smaller SOFC stacks (3-5 kW), ionic conductive materials are required that can operate in intermediate tem perature (IT) r ange i.e. 400-700oC without compromising the internal resist ance of the cell. Studies of new electrolytes led to the
31 discovery of other materials e.g., -Bi2O3 and doped CeO2. A summary of the ionic conductivity of different fluorite structured oxides is shown in Figure 1-2. It can be seen that -Bi2O3 and yttria-stabilized Bi2O3 exhibit the highest ionic conductivity among all oxygen ion conducting materials in IT range. The exceptional conductivity is due to the unique intrinsic defect fluorite st ructure of this material. Because of the stoichiometry of the material, 25% of the oxygen sites are vacant. Therefore, there are an excess number of equipotential sites implying a random distribution of oxygen ions at high temperatures. This results in high oxygen ion mobility. However, Bi2O3 based materials possess a number of disadvantages, including thermodynamic instability in reducing atmospheres, volatilization of Bi2O3 at moderate temperatures, a high corrosive activity and low mechanical strength.40,41 Hence, the applicabilit y of these oxides in electrochemical cell is considerably limited In the last few decades, ceria based electrolytes have attracted much attention as an alternative of YSZ in SOFCs. As shown in Figure 1-2, the ionic conductivity of gadolinium doped ceria is about one order of magnitude higher than that of YSZ. In addition, they exhibit good thermal stability which makes them superior to yttria-stabilized Bi2O3. 2.4.1 Ceria Based Materials In recent years ceria-based materials have attracted considerable attention especially in the application lik e solid oxi de electrolytes and envir onmental catalysis.9,42 In the automotive industry, ceria is used in catalytic converter inside the engine.43 It acts as a buffer, absorbing or releasing ox ygen, depending on the condition of the engine. It can effectively reduce harmful emissions like NOx to pure nitrogen and convert harmful carbon monoxide to the less harmful carbon dioxide.44,45 Further, it exhibits high absorption for ultraviolet rays and it is transparent to visible light which
32 makes it a prospective mate rial to replace ZnO and TiO2 in sunscreen applications.46 In the present work, doped ceria materials in the role of solid electrolytes for the intermediate temperature SOFCs will be discussed in detail. CeCe OO CeCe OO Figure 2-1. Cubic fluorite structure of ceria. Pure stoichiometric ceria possesses the c ubic fluorite structure which is stable from room temperature to the melting point. Figure 2-1 shows the cubic fluorite structure of ceria wher e Ce4+ and O2ions are located at 4a and 8c Wyckoff positions, respectively. The space group associated with this structure is m3Fm and the number of molecules of CeO2 present in the unit cell is four (Z = 4). The standard lattice parameter of pure ce ria at room tem perature is 5.4114 .47 According to the radius ratio scheme proposed by Pauling, coordinati on number is related to cation-anion radius
33 ratio (rc/ra). For the cation positioned at the center of the anion cube, rc/ra value should be higher than 0.732.24 In stoichiometric ceria, the Ce4+ cation establishes eight-fold coordination with O2anions, while the O2resides in the tetrahedral position. However, the theoretical rc/ra ratio in ceria is 0.702.48 This value is below the critical rc/ra which is a prerequisite for eight fold coordination. Thus, CeO2 does not follow radius ratio scheme which is well suited for other known structures. CeCe/ CeD OV OO CeCe/ CeD OV OO CeCe/ CeD OV OO Figure 2-2. Oxygen vacancy jump in doped ceria system. Ionic conduction in oxide fl uorites takes place via an oxygen vacancy diffusion mechanism. At high oxygen partial pressure s, cubic fluorite structure of undoped ceria do not possess any oxygen vacancies. Thus pure ceria is not a good ionic conductor. These oxygen vacancies are incorporated into the ceria structure dur ing the substitution
34 of Ce4+ by an acceptor dopant cation in the lattice, which can be represented by the following defect equation using Krger-Vink notation. OO Ce CeOVOD OD 32232 (2-3) Most of the oxygen ions tend to vibrate in the region of their tetrahedral sites and migrate from one regular tetrahedral site to t he other. This migration process is called discrete hopping. Normally, oxygen ions will not occupy the octahedral sites of Ce4+ ions in the face centered cubic lattice si nce these are the interstitial sites. Figure 2-2 shows the diffusion of oxygen vacancy from one tetrahedral site to another in doped ceria. 2.4.2 Ionic Conductivity Ionic conductivity (i) in the case of a pure oxygen ion conductor is given by, iioOiqNV ] [ (2-4) where No is the number of oxygen sites per unit volume, ] [ OV is the mole fraction of mobile oxygen vacancy present in the oxygen sublattice, and qi and i are the charge and mobility of oxygen ion, respectively. When electrons or electron holes are localized on ions in the lattice, as in CeO2x or Fe1xO, the semiconductivity arises from electrons or electron holes moving from one ion to another, whic h is called hopping-type semiconductivity.23 Thus, the total conductivity ( t) of a solid is the sum of the contributions due to ions ( i), electron (e), and hole ( h) as follows, heit (2-5) The electron and hole conductiviti es of the solid oxide elec trolyte must be kept as low as possible to prevent losses from leakage currents. In order to force the electrons liberated from the fuel into the external ci rcuit, where they can do useful work, there
35 must be a huge resistance to prevent them from going through the electr olyte. At high temperatures (>1000oC) and under reducing atmospheres, ceria shows a tendency to transform from Ce(IV) to Ce(III). / 22 2 1Ce O O CeCeOVOCe (2-6) This results in the increase in the n-type conductivity which pr omotes the formation of electron hopping paths inside the electrolyte material and deteriorates the overall performance of the cell. The ionic conductivity contribution to the total conductivity can be described in terms of the ionic transference number ti which is defined as, elec ion ion iont (2-7) As mentioned earlier, one of the main requirements for a good electrolyte is a high ionic transference number especially in ex treme conditions. Lower ionic transference not only degrades the associated cell efficiency but also expands the materials (due to the formation of additional oxygen vacancies) both of which have a del eterious effect on the mechanical properties.49 The yttria stabilized zirconi a, is pure ionic conductor at high temperatures (1000oC) and oxygen partial pressure (< 10-22 atmosphere) with an ionic transference number of 1. 184.108.40.206 Temperature dependence Ionic conduction in the oxygen ion conducto rs is a thermally activated process. The temperature dependence of electrical conduc tivity of doped fluorite oxides can be expressed as, kT E TA oexp (2-8)
36 where o is a pre-exponential constant, T is the absolute temperature, and EA is the activation energy for electrical conduction. Equation (2-9) is a detailed form of an Arrhenius relationship. kT H k S aNV k q Tm m ooO Vexp exp ][2 2 (2-9) Here No is the number of oxygen sites per unit volume, ] [OV is the fraction of free mobile oxygen vacancy present in the oxygen sublattice, qv is the charge oxygen vacancy, a is the ion jump distance, o is an appropriate lattice vibration, and Sm and Hm is the entropy and the ent halpy change during oxygen ion diffusion, respectively. For a complete derivation of this equation, refer to Appendix B. Equation (2-9) provides the clear view of the relationship between the conductivity and temperature. It is important to note that ] [OV in equation (2-9) is the conc entration of free mobile oxygen vacancies. In a distorted cubic struct ure of doped ceria system, oxygen vacancies tend to interact with acceptor dopant cations. Thes e interactions result in the localization of oxygen vacancies near the dopant cations. T he association of point defects will be discussed in more detail in a la ter section of this chapter. With the increase in thermal energy (temperature), the localized oxygen vacancies try to free themselves from dopant cations which results in the increase of mobile oxygen vacancy concentration. Thus, the pre-exponential constant ( o) in equation (2-9) depends upon the concentration of free oxygen vacancies wh ich itself is a complex function of temperature. Hence, the ionic conductivity of doped fluorit e oxides cannot be expressed by a singl e exponential equation.
37 220.127.116.11 Oxygen partial pressure dependence The rate of electrochemical reaction depends upon the driving force for the process to occur.24 The driving force is a measure of how far the system is from the equilibrium state. In the case of SOFCs, the driving force is the oxygen partial pressure difference between the cathode and anode sides The higher the partial pressure difference, the larger the power density t hat can be obtained from the cell. Thus, extreme opposite condition in both electrodes (i.e., high reducing atmo sphere in anode side and high oxidizing condit ion in cathode side) is nec essary to obtain reasonable power from the single ce ll. Normal atmospheric pressure of air is usually maintained in the cathode side of SOFC. Oxide ceramics are usually stable in such conditions but problem can arise at low oxy gen partial pressure. Doped ceria tends to partially reduce from Ce(IV) to Ce(III) in the reducing condition (< 10-15 atmospheres) at the anode side of the cell. For a material to be use as an electrolyte in SOFCs, it must have a high ionic transference number. The ionic tr ansference number starts decreasing under reducing atmosphere and generates conducti on electrons by the following defect reaction, / 2 O Oe2O 2 1 VO (2-10) Applying the law of mass action to the above reaction, the electron concentration at equilibrium can be written as, 4/1 2 2/1 2/1][ pOV TK nO (2-11) where K ( T ) is a equilibrium constant which depends solely on the temperature. The [ OV] can be assumed to be unchanged due to t he reduction of doped ceria in which a
38 large number of oxygen vacancies is already generated via doping. Thus, the ionic conductivity is independent of oxygen partial pressure and the eq uation can be written as, 4/1 2 2/1pO TK n (2-12) The ionic transference number (2Ot) which is described as, i e ei i Ot1 12 (2-13) As conductivity primarily depends upon the concentration of charge carrier, 2Ot can be written as, 1 4/12 2 21 O O OP P t (2-14) where 2OPis the parameter associated with K(T). From equation (214), it can be seen that 2OP at a given temperature corresponds to the oxygen partial pressure at which 2Ot equals to 0.5. Table 2-2. 2OP of doped ceria systems.50 2OP (atmosphere) Composition 500oC 700oC 850oC 1000oC Ce0.9Ca0.1O1.9 10-20 10-13 10-9 Ce0.905Ca0.095O1.95 10-26 10-17 10-13 10-10 Ce0.9Gd0.1O1.95 1.2 10-19 1.7 10-15 Ce0.5Gd0.5O1.75 1.5 10-16 6.5 10-14
39 Table 2-2 shows the value of 2OP for ceria based solid solutions at different temperatures. At 700oC, Gd0.10Ce0.90O1.95 shows the 2OP value of 10-19 atmospheres. With the increase in temperatur e, this value increases to 10-15 atmospheres. Thus, it is essential to develop electrolyt es which not only contain higher dopant concentration, but also exhibit higher ionic conductivity, so that they can operate in the lower IT range. 18.104.22.168 Dopant con centration dependence The effect of dopant concentration on ionic conductivity is illustrated in Figure 2-3, where total ionic c onductivities of doped zirconia systems are plotted as a function of dopant content.51 48121620 -3.2 -2.8 -2.4 -2.0 -1.6 -1.2 (Nd2O3)x(ZrO2)1-x (CaO)x(ZrO2)1-x (Y2O3)x(ZrO2)1-x (Gd2O3)x(ZrO2)1-x (Yb2O3)x(ZrO2)1-xLog Conductivity (S.cm-1)Mol% M2O3 or MO ( x ) Figure 2-3. Ionic conductivity as a func tion of dopant concentration in doped zirconia systems.51 In the dilute regime, the ionic conducti vity as a function of dopant cation is consistent with the Arrhenius relationship in equation (2-9 ). It increases with the
40 increase in the number of charge carriers i. e., oxygen vacancies. However, it is experimentally observed that the ionic conductivity reac hes a maximum at certain dopant concentration, and then decreases. This is the general behavior observed in all the other fluorite oxide systems. Further, the observed peak positions are not the same for different dopant cations. It has been argued that the observed decrease in conductivity is associated with dopant cation and oxygen vacancy interactions; this will be discussed in the subsequent section.9,52 2.5 Association of Defects Whenever charged defects (/ AD) are incorporated into the crystal structure, there must be defects with opposite charges to pres erve charge neutrality. These defects are prone to strongly associate with each other due to electrostatic attraction between the oppositely charged defects. Also there is a possibility for el astic interaction to relieve the local stresses associated with both defects.52 The end result is the formation of local defect structures such as / /A OADVD and OAVD/, which effectively reduces the number of mobile oxygen ions, thus leading to a substantial drop in the ionic conductivity of the electrolyt e at higher dopant content (as shown in Figure 2-3). CeCe/ CeD OV CeCe/ CeD OV Figure 2-4. Oxygen vacancy tr apped in the tetrahedral site.
41 Figure 2-4 shows the situation in wh ich an oxy gen vacancy is trapped in a tetrahedral site which is surrounded by op positely charged dopant cations and neutral host cations. The binding energy between t hese defects is mainly due to the coulombic attraction of the defects caused by the effective charges in the lattice However, it also includes terms due to the relaxation of the lattice around the defect which depend on the effective charge, polarizabil ity and ionic radius of the dopant cation. Thus, in the lower intermediate temperature range, total activation energy ( H) is the sum of migration enthalpy ( Hm) and association enthalpy ( Ha).9,53 With the increase in temperature, the thermal vi brations of the local defect structures become more important. The stage comes w hen the thermal energy overco mes the binding energy of these local defect structures and most of t he oxygen vacancies are set free to diffuse from one site to another. Above this te mperature, oxygen vacancies require only Hm to cross the energy barrier. 2.6 Impedance Spectroscopy Alternating current (ac) electrochemic al impedance spectroscopy (EIS) is a valuable technique for the characterizati on of various electrochemical processes.54,55 EIS has increased the und erstanding of the limiting processes for different materials as well as effects of processing and microstructure of materials in tests applying electrode pellets, and other electrochemical devices. It measures voltage (or current) as a current (or voltage) is applied on the test sample over a wide range of frequencies. Impedance at a given frequency () is determined using Ohms la w i.e., the ratio between ac voltage and ac current. Since the output impedance is a co mplex number, therefore at any given frequency, it contains real (Z/) and imaginary (Z//) components. The complex impedance data is typically represented as a Nyquist plot, where the real impedance
42 and negative imaginary impedan ce is plotted in the x and y coordinates, respectively. The sign convention for the imaginary impedance is typically used in the plot as the capacitive behavior often dominat es the processes in ceramic materials. With the choice of sign convention, most of the processes of interest for SOFCs occur in the first quadrant of the complex impedance plot. The frequency range exam ined is typically from 0.01 to 107 Hz. The upper limit on the test frequency is determined by the limitation of the instrument, while the lower limit is usua lly determined by the time it takes for data acquisition. However, these lim itations are of not much concern as the most of the phenomenon lie in the measurable frequency r ange. The response of the cell is usually modeled in terms of equivalent circuits, i.e., a group of electrical circuit elements (resistors, capacitors, inductors) t hat are connected in a way that would give the same impedance response as the cell. 0123456 0 1 2 3 4 5 6 -Z// (Ohms)Z/(Ohms) Resistor Capacitor Lossy Capacitor Figure 2-5. Typical impedance response for a given relaxation process. A parallel resistor-capacitor element is also s hown which represents a lossy capacitor.
43 One of the main advantages of using EIS is that it assists in identifying the relaxation time associated with the pr ocesses occurring at the atomic and microstructural level. Each distinct process has its own real and imaginary impedances, with a characteristic relaxation frequency. A typical impedance response for a given relaxation process, and it equivalent circuit is shown in Figure 2-5. A parallel RC element represents a lossy capa citor with a typical relaxa tion time which c orresponds to the process. Such behavio r could be characteristic, fo r example, of a double-layer capacitance (due to charge separation betw een electrode and electrolyte) in parallel with a resistance to charge transfer or a polariz ation resistance. Thus a semi-circle in a complex impedance plot is a repr esentation of a relaxation proc ess with a distinct time constant. The peak of the semicircle will occur at the characteristic frequency, o. The Nyquist plot of the electrochemic al cell may contain more than one semicircle, which is indicative of various relaxation processes occu rring in the cell. Often the time constant for two processes lie very close to each other. In th is case, semi-circles must be de-convoluted in order to determine each individual contribution. A typical complex impedance response of an electroded polycrystalline ceramic is shown in Figure 2-6. In the figure, three main feat ures can be observed: incomplete depressed arc at high frequency and two distinct arcs at low frequency. Th e high frequency arc was identified as the grain polarization while the rema ining two arcs at low frequencies correspond to grain boundary and electrode polar ization processes. In the ideal case, the center of the semi-circle should lie on the x-axis. However, it is usually the case, that the center of the semi-circle is depressed below the xaxis. The equivalent circuit for such response is similar to that shown in Figure 2-5; however, in place of capacitor a
44 constant phase element (CPE) is required to model the exper imental data. A constant phase element is equivalent to a distribution of capacitors in parallel whose phase angle is somewhat less than 90o. Further, a depressed semi-c ircle in the complex impedance plot has been described in several ways. For example, the depressed semicircle in the intra-grain polarization can be attributed to the microstructural inhomogeneities within the sample. In addition, the depressed elec trode polarization semi-circle in solid electrode is explained by the surf ace roughness of the electrode. 0.0 3.0x1036.0x1039.0x1031.2x104 0.0 3.0x1036.0x1039.0x1031.2x104 -Z// (Ohms)Z/(Ohms)Equivalnt Circuit5.9 MHz 1.7 kHz 100 Hz Grain Grain Boundary Electrode Impedance Spectrum of an Electroded Polycrystalline Ceramics Figure 2-6. Typical impedance spectrum of an electroded polycrystalline ceramic material. The equivalent circuit to fit t he spectrum is also shown. CPE and R in the equivalent circuit stands fo r constant phase element and resistor, respectively. It is important to note that in the present work, the main focus is to determine the grain ionic conductivity rather than the tota l ionic conductivity. Thus, the grain ionic resistance was calculated by fitting t he observed impedance spectra with the analog
45 circuit having three parallel pair of resist or-constant phase element in series. The presence of the inductive effect of the experimental setup severely affects the high frequency datasets at high temperat ures. To counter this e ffect, external inductor was also used in the equivalent circuit, while fi tting the data. The grain ionic conductivity was calculated using the following relationship, A L Rg grain1 (2-15) where Rg is the grain ionic resistance, L and A are the length and area of the crosssection of the sample, respectively.
46 CHAPTER 3 MATERIALS AND EXPERIMENTAL PROCEDURES 3.1 Materials Selection Higher ionically conductive materials are cr itical for the realizat ion of intermediate temperature range SOFCs. In recent years, doped ceria electrolytes have opened up the possibility for such intermediate temper atures SOFCs, due to their higher ionic conductivity and good thermodynamic stabili ty. Ionic conductivity in doped ceria systems depends upon various factors such as dopant charge va lence, dopant ionic radius, dopant concentration, temperature, and oxygen parti al pressure. Keeping the last two parameters constant, ionic condu ctivity primarily depen ds upon the type of dopants. According to Yahiro et al, the higher the charge valence of the acceptor dopant cation the better is the ionic conductivity.56,57 Thus, trivalent doped ceria systems show higher ionic conductivity compar ed to divalent doped ceria systems. This can be attributed to the lower electrostatic in teraction involved in t he case of trivalent doped system, between oxygen vacancies and dopant cation sites. These interactions hinder the flow of oxygen vacancies and lo wer the free oxygen vacancy concentration.52 Thus, in order to design higher conductive materials, differ ent trivalent acceptor dopant cations were selected for host ceria. 3.2 Experimental Procedures In this section, the details of the ex perimental procedures that were used to synthesize phase pure powders and ceramic sample s for different characterizations will be discussed. Figure 3-1 shows the block diagram of the typical powder sy nthesis, and different ceramic sample characterization techniques.
47 Figure 3-1. Flow chart for the powder synthesis using conventiona l solid state reaction and different characterization techniques. Reagent Grade Oxides Ball Milled, 24 h Drying at 120oC for 16 h Calcination at 1450oC for 10 h Cold Isostatic Pressing and Sintering at 1550oC for 10 h Polishing and Electroding using Pt paste Sintering at 900oC for 1 h Ionic Conductivity measurements Pt wires were attached Drying at 120oC for 16 h Ball milling for 24 h AC Impedance Spectroscopy Scanning Electron Microscop y Dispersant Ball Milling, 24 h Dispersant Binder X-Ray Diffraction (XRD) Transmission Electron Microscopy Particle Size Distribution X-Ray Diffraction (XRD)
48 3.2.1 Powder Processing Different processing routes were us ed to synthesize phase pure doped ceria powders. Conventional solid oxide route wa s primarily used to synthesize powder in bulk amount. However, the particle size achieved using this technique was usually large compare to other wet chemical routes.58 Large particle size results in lowering the surface reactivity of the powder which in turn lowers the sintering kinetics of the material.59 Thus, high sintering temperature and time are required for such green ceramic bodies to obtain high theoretic al density. Even though high sintering temperature and time increases the diffusion kinetics for sintering process, they are usually avoided for several reasons. Such high sintering temperatures may lead to unwanted interfacial reactions during co-sin tering of electrolyte and cathode or anode layers. In addition, very high-temperature sintering can produce mi cro-cracks; e.g. due to oxygen liberation as a result of reduction of CeO2 to Ce2O3.60 The interest in low temperature sintering is also driven by th e need to reduce material processing costs. Furthermore, it has been shown that there is a detrimental effect of sintering at high temperatures on the ionic conductivity.61,62 This will be discussed in detail in Chapter 6. In the present work, all the doped ceria materials were synthesized using conventional solid state sinter ing under same experimental conditions. These materials were further processed for electrical charac terization at high temperatures. Among all the doped ceria systems, the material whic h shows the highest conductivity was then synthesized using wet chemical route (i.e., co-precipitation technique). Using this process, high purity, homogenou s and ultrafine powder can be obtained by calcining at temperatures lower than 1000oC.58,63,64
49 22.214.171.124 Conventional solid state route Phase pure powders of DxCe1xO2(where D = trivalent dopant cation, x = dopant content) were synthesized by conventional solid state reaction method, starting from stoichiometric mixtures of D2O3 and CeO2 powders (all with 99. 99% purity from Alfa Aesar). Loss on ignition was determined for all the starting powders. Accordingly, the stoichiometric amount of powders were we ighed and mixed with de -ionized water and 1 wt% dispersant (Ammonium Polyacrylate) to fo rm 60 vol% solids slurry. The role of dispersant is to keep the dispersed ceramic par ticles in suspension in the water. Mixing was performed using wet ball milling for 24 h. One of the main di sadvantages of using ball milling is that the wear of the grinding media can be fairly high. For ceramic materials used for electrolyte application, the presence of unwanted impurities in the powder is a serious problem. To avoid this, chemically inert and highly pure yttria stabilized zirconia media with 3 and 10 mm di ameter spheres were used. The ball milled ceramic slurries were then dried in the oven at 120oC for 16 h. The agglomerated powders were ground using mort ar and pestle to fine particles which were subsequently separated out usi ng sieve with the aperture opening of 212 m. After obtaining the homogeneous mixture of raw oxides, po wders were then treated to high temperature for solid state reaction. T he calcination temperature and time were optimized for different doped ceri a systems. The dissolution of the dopant cations inside the host latti ce depends upon its solubili ty limit at a particular temperature. The width of t he solubility increases with the decrease in the ionic radius difference between the host and dopant cations. The maximum solubility exist when the difference between the ionic radii become minimum.65 For example, for Lu0.10Ce0.90O2the calcination temperature and time were optimized to be 1350oC for 10 h.
50 30405060 1000oC(222) (311) (220) (200)1400oC1350oC1300oC 1200oC 1100oCIntensity (arb. units)2(Degrees)Lu2O3 Mixed(111) Figure 3-2. XRD patterns of Lu0.10Ce0.90O2calcined at different temperatures for 10 h. To verify the complete dissolution of dopants in ceria, phase analysis was performed using X-ray diffracti on (XRD). Curved position-sens itive (CPS) diffractometer (INEL, France) was used to obtain the XRD pattern of each com position using Cu K radiation. Figure 3-2 shows the XRD profiles of Lu0.10Ce0.90O2calcined at different temperatures for 10 h. It can be seen around 1350oC, there are no extra peaks of Lu2O3 present in the XRD pattern. As Lu3+ exhibits similar ionic radius as that of Ce4+, thus it requires lower therma l energy to dissolve in host ceria lattice when compare to other dopant cations which show higher ionic radii (rLu VIII = 0.977 rCe VIII = 0.97 ).48 In order to widen the solubility li mit for other trivalent dopant s cations, calcination was performed at 1450oC for 10 h. Figure 3-3 shows the XRD pa tterns of different doped ceria systems taken at room tem perature in air. It can be s een that all the compositions
51 are phase pure with cubic fluorit e structure. In the inset, (200) plane peak of all the compositions is shown. The peak position provides inform ation about the lattice change (expansion or contraction) of the host ceri a which will be discussed in Chapter 4 and Chapter 5. 3040506070 32.733.033.3 Intensity 2 (Degree)400 222 311 220 200Sm0.10Ce0.90O2-Gd0.10Ce0.90O2-Lu0.10Ce0.90O2-Y0.10Ce0.90O2-Intensity (Arbitary Units)2 (Degree)CeO2Er0.10Ce0.90O2-Room Temperature111200 Figure 3-3. XRD patterns of different doped ceria systems calcined at 1450oC for 10 h. After the calcination, aggl omerated powders were again ball-milled for 24 h in deionized water, dispersant, forming a 60 vol % slurry. The ball-milled ceramic slurries were then dried in the oven at 120oC for 16 h. The agglome rated powders were ground using mortar and pestle to fine size par ticles, which were subsequently separated out using sieve with the aperture opening of 212 m. The particle size distribution of these powders was measured using Beckman Coulter LS13320. Figure 3-4 shows the
52 number particle size distri bution of co-doped ceria ceramic powder synthesized using the experimental procedure descri bed earlier. In order to ac hieve high density ceramic, it is desired that the particle size shou ld be less than 1 m with particle size distribution to be narrow and monodisperse.59 It can be seen that the most of the particles are less than 2 m in size with the mean size of 0.84 m. Further, the particle size distribution is also mono-disperse which is desired for good sintering in reasonable time. 0.1 1 10 0 2 4 6 8 10 Diffrential Number (%)Particle Size (m)Sm0.075Nd0.075Ce0.85O2-Using Solid Oxide Route Figure 3-4. Particle size distribution of Sm0.075Nd0.075Ce0.85O2synthesized using solid oxide route. 126.96.36.199 Co-precipitation route The co-precipitation technique was us ed to synthesize phase pure powder of Sm0.075Nd0.075Ce0.85O2.58 By doing so, the effect of processing on the electrical property of co-doped ceria wa s studied. One of the main objectives of using wet chemical route is to obtain fine particle size powder which as a result enhance the
53 sintering kinetic of the ceramic powder. Thus, high density pellets can be obtained using lower sintering temper ature and time (especially w hen compared to conventional solid state route). Highly pure cerium nitrate (Ce(NO3)3.6H2O, Alfa Aesar, 99.99%), samarium nitrate (Sm(NO3)3.6H2O, Aldrich, 99.999%), and neodymium nitrate (Nd(NO3)3.6H2O Alfa Aesar, 99.9%) were used as a starting raw materials. They were weighed in the stoichiometric proportions and dissolved in de-ionized water to produce aqueous solution. Excess ammonia soluti on (Acros Organics, 28-30 vol% of NH3 solution in water) was added to the stirred so lution to increase the pH to 12. The addition of ammonia solution results in the formation of yellowish brown color precipitate. The precip itate was filter ed and then subsequently dried at 80oC for 12 h. The agglomerated powder was then ground to fi ne particles using mortar and pestle. The powder was then calcined at 900oC for 10 h in air. The CPS diffractometer was used to obtain the XRD pattern of Sm0.075Nd0.075Ce0.85O2using Cu K radiation. A monochromator crystal was us ed to separate out Cu K 1 from the incident X-ray beam. Peak positions in the XRD pattern were det ermined by fitting each individual peak with a symmetric Pearson VII profiles to model Cu K 1 using commercially available software (i.e., Solver add-in within Micros oft Excel spreadsheet package). Figure 3-5 shows the XRD profile of the ca lcined powder of Sm0.075Nd0.075Ce0.85O2taken at room temperature. Powder looks phase pure with c ubic fluorite structure. The best estimate of the lattice constant (ao) was calculated using the least-squares extrapolation method.66 The lattice parameter of Sm0.075Nd0.075Ce0.85O2synthesized using coprecipitation technique was 5. 4299 0.0013 The estima ted lattice parameter value is close to the value (5.4314 ) obtained from the empirical relationship given by Kim.22
54 304050607080 331 420 400 222 311 220 200Intensity (Arbitary Units)2 (Degrees)111XRD pattern of Sm0.075Nd0.075Ce0.85O2synthesized using Co-precipitation method Figure 3-5. XRD pattern of Sm0.075Nd0.075Ce0.85O2synthesized using co-precipitation technique. After the calcination, aggl omerated powder was ground us ing mortar and pestle to fine size particles which were then separat ed using sieve with the aperture opening of 212 m. The particle size distribution of the grounded powder wa s then measured and is shown in Figure 3-6. It can be s een that the particle size of the powder synthesized using co-pr ecipitation is lower than that in solid state method. The most of the particles are less than 0.2 m in size with the mean size of 0.095 m. Similar to the conventional solid state route, the particle size distribution obtained using this technique is somewhat narrow and mono-disperse which is desirable for good sinterability.
55 0.1 1 0 2 4 6 8 10 12 Diffrential Number (%)Particle Size (m)Sm0.075Nd0.075Ce0.85O2-Using Co-precipitation Route Figure 3-6. Particle size distribution of Sm0.075Nd0.075Ce0.85O2synthesized using coprecipitation technique. 3.2.2 Forming After synthesizing the phase pure powder s from conventional solid oxide route and/or co-precipitation techniques, the nex t step was to fabricate a green ceramic body. Approximately 2 wt% of polyvinyl alcohol (P VA) binder was added into the powder to assist in forming. Powder was then mixed with the binder using a mortar and pestle. This was performed until the powder stopped st icking to the wall of the mortar. The agglomerates were separated fr om the powder using a sieve with the aperture opening of 212 m. Powder consisting of fine size particles was t hen uniaxially pressed into disk-shaped pellets (8 mm in diameter and 4 mm thick) under a pressure of 180 MPa. This was followed by the isostatic pressing at 200 MPa for 3 minutes.
56 3.2.3 Solid State Sintering Sintering is a process in whic h the ceramic green body is heat treated to a temperature (below the melting point) such t hat its particles adhere to each other to give a desired microstructure. Thus, the porous ceramic body when heated to the sintering temperature, it densify to the t heoretical density of the material. This densification rate involves the transport of material which in turn depends upon key process parameters such as particle size, temperature, time a nd pressure. In this work, all the ceramic samples were sintered in air. Thus, the only paramet ers which were varied were temperature, time and particle size of t he powder. Based on this, the effect of processing conditions on the electrical property of doped ceria is studied. Although the high sintering te mperature and time promotes solid state sintering, it has a detrimental effect on t he ionic conductivity of doped ce ria materials. Fabrication of highly dense ceramics at lower sintering temperatures and shorter times remains one of the challenges for the applicat ion of this material. Micr owave sintering technique was used to fabricate dense cera mics of doped ceria at low temperature and time. The effect of microstructure on the ionic c onductivity will be discussed in Chapter 6. 188.8.131.52 Conventional sintering The green ceramic pellets obtained fr om the phas e pure pow ders synthesized using solid oxide route we re sintered in air at 1550oC for 10 h. The pellets pressed from the Sm0.075Nd0.075Ce0.85O2powder synthesized using co-p recipitation technique were sintered at 1350oC for 10 h. The ramp rate in these runs while heating and cooling was 200oC/hour. Binder burn out was performed at 400oC for 1 h. Densit ies of the ceramic samples sintered at 1550oC for 10 h were measured in water using Archimedess principle and were estimated to be 98% of theoretical density or above. However, the
57 density of the sintered samples fabric ated from powder sy nthesized using coprecipitation was around 93% of theoretical density. 184.108.40.206 Microw ave sintering The green ceramic pellets of Sm0.075Nd0.075Ce0.85O2powder synthesized using co-precipitation technique were microwave sintered in a 2.45 GHz microwave furnace (Thermwave Mod III). Microwave furnace wa s operated at 1.3 kilowatts power output. The microwave sintering was performed at 1450oC for 1 h. Figure 3-7. Sample placed in the t hermal pod for microwave sintering. Figure 3-7 shows the green ceramic pellet pos itioned in a zirconia sample holder inside the refractory insulation box or t hermal pod. Silicon carbide thermocepts were used to boost the temperature rise of both microwave absorbing samples and non-absorbing or low absorbing materials. They can micr owave heat materials through temperature zones were the material does not readily absorb microw ave energy at 2.4 GHZ. These thermocepts are arranged around the inside wall of the thermal pod. The
58 Archimedes density of the pellets fabric ated from the microw ave sintering was estimated to be around 94% of theoretical density. 3.2.4 Sample Preparati on for Characterization After fabricating the high dens ity ceramic pe llets, they were processed for different characterizations. In this section, sample fabrication for the diffe rent characterization techniques utilized will be discussed. 220.127.116.11 Ionic conductivity The high t emperature ionic conductivi ty measurements were performed on dense sintered ceramic samples of doped ceria materials. The sintered pellets were polished to obtain planar surfaces. Pt paste (CL115349, Heraeus) was brushed onto both sides of the disk-shaped pellets to serve as the el ectrode. The pellets were then co-fired at 900oC for 1 h. Pt wires (99.9% pure) with diameter 0.127 mm were attached to the cell using Ag adhesive paste to perform ionic conductivity measurements. 18.104.22.168 X-ray diffraction High temperature X-ray diffraction was performed on the doped ceria samples to understand the relationships between the crystal structure and ionic conductivity at higher temperatures. After the ionic c onductivity measurements, samples were mechanically polished with SiC abrasiv e papers (of different grit sizes), and subsequently cleaned using sonicator. Pell ets were then ground to fine particles using mortar and pestle. X-ray di ffraction was performed on the sintered fine particles of doped ceria. 22.214.171.124 Scanning electron microscop y The microstructural characterization was performed using scanning electron microscope (SEM). The si ntered ceramic samples were mechanically polished to a
59 mirror finish using alumina dispersed polymer grinding discs of different grit size. After polishing, ceramic samples were cleaned us ing sonication for 30 min. This was followed by thermal etching process. The clean polished surface of ceramic samples was rapidly taken to the temperature 100oC below the respective si ntering temperature. The heating ramp rate was 600oC/hour. Samples were kept at the respective temperatures for 1 h. After thermal etchin g, they were again cleaned using sonication for 30 minutes. They were then coated with carbon for SEM. 126.96.36.199 Transmission electron microscopy The grain boundary study was performed th rough conventional TEM, using its various modes of analysis. In conventional TE M, these were mainly bright field (BF) and dark field (DF) imaging, selected ar ea diffraction patterns (SADP), and energy dispersive X-ray analysis (EDX). These were performed on a JEOL JEM-2010 at 200 kV having an ultra thin Be window for EDX, attached with a Link Analyzer. Sintered disks, 3 mm in di ameter of doped ceria were mechanically polished to a thickness of ~30 m using Gatans precision diamond discs of different grades. Using superglue, the ceramic sample was t hen mounted on the molybdenum TEM sample holder (with the shape of a ring). The sa mple placed on Mo sample holder was then thinned in the precision ion po lishing system (PIPS) with ar gon ions at an accelerating voltage of 5 kV, which were incident at both surfaces at an angle of 4 to yield a reasonably thin specimen. After 4 to 10 h of ion milling (depen ding upon the sample thickness), a hole was seen in the centre of the ceramic sample. The samples were then coated with carbon, and were placed under the microscope for microstructural study.
60 3.2.5 Impedance Spectrosco py Experimental Setup Ionic conductivity measurements were done using two point probe ac impedance spectroscopy technique. Figure 3-8 shows the electroded ceramic sample, quartz reactor, and impedance spectroscopy setup. Figure 3-8. AC impedance spec troscopy experimental setup. Processed ceramic sample was individual ly heated in a quartz reactor which was placed inside a small tube furnace. The quartz tube consisted of an inlet and outlet for gas flow, and gold wires running through al umina rods coated with platinum for shielding. The gold wires were connected to t he Pt wires, which were attached to the Pt electrodes of the ceramic sample using Ag paste. At different temperatures, the complex impedance (Z) of the sample was measured using Solartron 1260 over the
61 frequency range of 0.10 Hz to 32 MHz. A 50 mV AC voltage was applied and the induced current was measured to produce the impedance spectra. Measurements were performed using 2-point connecti on to the Solartron. The Solartron 1260 is interfaced with the computer through Zplot software. Data acqu isition was done through Z-plot software. All the measurements were take n in air, in the te mperature range of 250oC to 700oC.
62 CHAPTER 4 IONIC CONDUCTIVITY AND DOPING ST RATEGI ES FOR IONIC CONDUCTIVITY ENHANCEMENT 4.1 Ionic Conductivity Lowering the operating tem perature range to intermedi ate temperatures (500 C) can significantly reduce the cost and extend the SOFC a pplication domain to portable power market sectors.67 High ionic conductivity of ox ide electrolytes is critical for the development of such SOFCs. Although yttria stabilized zirconia (YSZ) is considered to be the most reliable electrolyt e in terms of structur al and thermodynamic stability, its lower ionic conductivity in intermediate temperatures restricts SOFC usage to high temperatures (1000C).7,67 In recent years, doped ceria electrolytes have opened up the possibility for su ch intermediate temperatures SOFCs due to their higher ionic conductivity and good thermodynamic stability.9 Among doped ceria electrolytes, Gd0.10Ce0.90O2is widely accepted to exhibits the highest ionic conductivity.13 In search of advanced ceramic materials that show higher ionic conductivity than Gd0.10Ce0.90O2, it is important to investigate the effect of dopants and doping level on the ionic conductivity of the doped ceria materials. This involves investi gation of local defect structures or complex defect associates (as mentioned in Chapter 2) for different doped ceria systems. In addition, thermodynamic properties su ch as activation energy for oxygen diffusion and entropy of the doped ceria system ar e also influenced by the dopant type and their concentration.68,69 Thus, it is essential to understand these relationships, and based on them rationally des ign higher ionic conductive electrolyte materials.
63 As mentioned in Chapt er 2, ionic conduction in these materials is a thermally activated process. Ionic conductivity () dependence on temperatur e can be described by Arrhenius relationship, kT H Tm oexp (4-1) where o, Hm, k and T are pre-exponential coefficient, migration enthalpy for oxygen diffusion, Boltzmanns constant and absolute temperature, respectively. Equation (4-2) is the detailed form of the abov ementioned Arrhenius relationship, kT H k S aNV k q Tm m ooO Vexp exp ][2 2 (4-2) where No is the number of oxygen sites per unit volume, ] [ OV is the fraction of free mobile oxygen vacancy present in the oxygen sublattice, qv is the charge oxygen vacancy, a is the ion jump distance, o is an appropriate lattice vibration, and Sm and Hm is the entropy and enthalpy change, respectively, during oxygen ion diffusion. However, the above relationship assumes that there are negligible interactions between dopant cations and oxygen vacanc ies sites. Due to both electrostatic and elastic interactions between these two types of point defects, there is some additional energy barrier involved (in addition to Hm) for oxygen diffusion. Thus there is a possibility of the formation local defect structures (also discussed in Chapter 2) such as / / A OADVD and OAVD/ (i.e., trimers and dimers, respectively). The additional thermal energy (which can be termed as asso ciation energy) is required to break free the oxygen vacancies from thes e complex defect associates.52 Taking / / A OADVD
64 and OAVD/ defect structures into account, equat ion (4-2) can be wr itten as equation (4-3) and equation (4-4), respectively. kT HH k SS aAN kT qDimers m Dimers m oo V iexp exp2 2 (4-3) kT H H k S S aNcA kT qTrimers m Trimers m oodop V i3 exp 3 exp 4 123/13/1 2 3/1 (4-4) where HDimers and SDimers, and HTrimers and STrimers are the associat ion enthalpy and entropy of dimers and trimers, respective ly. A complete derivation of the above relationships can be found in Appendix B. As these defect structures lower the free mobile oxygen vacancy concentration, one of the main objectives while designing the electrolyte material is to minimize this association enthalpy (i.e., HDimers and HTrimers). The concentration of these defect stru ctures depends on the dopant content and the physical properties of acceptor dopants such as ionic radius, charge valence, polarizability, etc.22,70 Hence, the selection of dopants fo r host ceria based on activation energy is critical for the devel opment of novel electrolytes. In addition, it can be seen from equations (4-3) and (4-4), that the ionic conductivity also depends upon the entropy of the system. As mentioned in Appendix A, the total entropy of a colle ction of atoms is a sum of vibrational entropy and configurational entropy.24 The uncertainty in the exact value of energy in which the atoms vibrates in a given energy level constitutes vibrational entropy. On increasing the temperature, the probability of atoms excited to the higher energy level increases, which in turn, increases the vibrational entropy of the system. The c hange in vibrational entropy as a function of tem perature is given in Appendix A. Simi lar to free mobile
65 oxygen vacancy concentration (in the pr e-exponential term of the Arrhenius relationship), entropy term also depends on the temperature. Further, vibrational entropy is also the function of the characteristic frequency in which the atoms vibrate. Dopants which exhibit stronger bond strength with oxygen ions will show lower thermal entropy, while dopant which shows weaker bond strengt h will exhibit higher vibrational entropy.24 The total entropy also contains the confi gurational term which is the measure of the number of ways in which the atom c an be arranged in a given number of lattice sites. Thus, using two or more dopants, t he number of configurat ions can be increased which in turn increase the total entropy in the system. Previous works have shown that using two or more dopants can signific antly improve the ionic conductivity.16,18,19 This was attributed to the increase in configur ational entropy which consequently enhances o.69,71 In the present work, this approach was taken and the co-dopants were used to suppress the oxygen vacancy ordering with an aim to enhance the ionic conductivity. 4.2 Co-doping Strategies In the following sections, two different co-doping strategies will be discussed followed by results and discussion. The main objectives in both the strategies are to minimize the association energy and maximize the pre-exponentia l coefficient (or entropy) which as a result increases t he ionic conductivity of the material. 4.2.1 Critical Dopa nt Ionic Radius The critical dopant ionic radi us was first proposed by Kim22, which was later revisited by many other researchers72-74. In this section, the effect of co-doping based on critical dopant ionic radius on the ionic conductivity in ceria will be discussed.
66 188.8.131.52 Theory As discussed in previous section, ac tivation energy for oxygen diffusion at intermediate temperatures cons ists of migration enthalpy ( Hmig) of the oxygen ion, and the association enthalpy ( Hassoc) of the local defect structures From previous research it has been found that in the dilute regime, Hmig is independent of dopant concentration.75 However, with the increase in oxygen vacancy concentration, the probability of the format ion of local defect structure increases. Thus, to enhance the ionic conductivity, the minimization of the association of defects is required. As expressed before, this association energy is a function of the coulombic attraction between oxygen vacancies and dopant cations si tes and the elastic strain field around the defect complex.22 According to Kilner et al.21, maximum ionic conductivity in doped oxide fluorites is achieved when there is mini mum elastic strain present in the lattice. Kim22 studied the effect of ionic radius and valence of the dopant cation on the lattice parameter of oxide fluorites and a critical ionic radius, rc, was defined as the ionic radius of the dopant that neither causes expansion nor contraction of the host fluorite lattice. Specifically, for a trivalent dopant cation in host ceria, rc was calculated to be 1.038 Figure 4-1 shows the grain (bulk) ionic conduc tivity as a function of trivalent dopant cation. It is important to note that the grain ionic conducti vity data presented in this figure are taken from literature. Gd0.10Ce0.90O2exhibits the highest grain ionic conductivity among singly doped ceria electrolytes This is consistent with the critical ionic radius concept as the ionic radius of Gd3+ (r3+ Gd,VIII = 0.1053 nm) is relatively close to the ideal rc value.48
67 0.0960.1000.1040.1080.112 -1.0 -0.5 0.0 0.5 rc3+Nd3+Sm3+Gd3+Y3+Log(grainT[S.cm-1.K])Radius of trivalent dopant cation/nmLu3+ LuxNdyCe0.90O2GdxYyCe0.90O2Figure 4-1. Grain ionic (GI) conductivity comparison of D0.10Ce0.90O2at 400oC. GI conductivity data for Gd0.10Ce0.90O2, Sm0.10Ce0.90O2, Y0.10Ce0.90O2and Nd0.10Ce0.90O2are taken after Steele13, Zhan et al.76, Zhang et al.77, and Li et al.78, respectively, while Gd0.056Y0.044Ce0.90O2, LuxNdyCe0.90O2and Lu0.10Ce0.90 O2are from this work. 184.108.40.206 Materials selection As Kim22 suggested, ideal trivalent dopant fo r host ceria should have the ionic radius of 1.038 In the lanthanide series of the periodic table, Tb3+ cation with VIII coordination shows ioni c radius of 1.040 .48 Hence, based on cr itical dopant ionic radius scheme, Tb3+ is the ideal dopant for host ceria. Unfortunately, Tb element exists in different valence states. This w ill promote electronic conduction through the electrolyte and lower the ionic transference number.54 In order to test Kims hypothesis, co-doping pairs were selected ba sed on their ionic radius. Lu3+ and Nd3+, and Gd3+ and
68 Y3+ were selected as trivalent co-dopant pairs. They were added in a proportion such that the weighed average dopant ionic radius of co-dopants matches rc. For example if xLu and xNd are the proportion of Lu3+ and Nd3+ co-dopants, then they must satisfy the following equation: cNd Lu NdNd LuLurxxrxrx )( (4-5) where rLu and rNd are the ionic radius of trivalent dopant Lu3+ and Nd3+ with VIII coordination, respectively. By doing so, it was expected that the positive elastic strain due to larger dopant cation (i.e., Nd3+ or Gd3+) can be compensated by the negative elastic strain due to smaller dopant cation (i.e., Lu3+ or Y3+). This in turn prevents any distortion in fluorite lattice which is usua lly present in singly doped ceria systems. Lu0.054Nd0.046Ce0.90O2and Gd0.056Y0.044Ce0.90O2ceramic samples were synthesized using solid oxide route as described in Chapter 3. In order to study the elastic strain behavior as a function of dopant concent ration, different composition of LuxNdyCe1x yO2(where x +y = 0.05, 0.10, 0.15 and 0.20) were also synthesized. In addition, to understand the effect of electrostatic interactions alone, LuxCe1xO2(where x = 0.05, 0.10, and 0.15) was also investigated as the ionic radius of Lu3+ is approximately the same as Ce4+ (r3+ Lu,VIII= 0.977 and r4+ Ce,VIII= 0.97 ).48 Moreover, singly doped ceria D0.10Ce0.90O2(where D3+ = Y3+, Gd3+, Sm3+ and Nd3+) samples were also processed under similar experimental procedure. 220.127.116.11 Results and discussion In this section, relationship between the elastic strain (measured at room temperature) and the ionic conductivity measured at 400oC for co-doped ceria systems
69 will be discussed. Further, the comparison between the ionic conductivity of co-doped ceria with that of Gd0.10Ce0.90O2will be presented. X-ray diffraction analysis: To ensure the complete disso lution of dopant in CeO2, phase analysis was performed using X-Ray diffraction (XRD) technique. Figure 4-2 and Figure 4-3 show the XRD profiles of the calcined powders of LuxNdyCe1 -x-yO2(where x +y = 0.05, 0.10, and 0.15 and 0.20), Gd0.056Y0.044Ce0.90O2, CeO2 and D0.10Ce0.90O2(where D3+ = Lu3+, Y3+, Gd3+, Sm3+and Nd3+). It can be observed that all the compositions studied are singl e phase with a cubic fluori te structure like pure CeO2. 30405060 x+y = 0 GdxYyCe0.90O2-CeO2LuxNdyCe0.80O2-LuxNdyCe0.85O2-LuxNdyCe0.95O2-LuxNdyCe0.90O2-(111) (200) (220) (311)x+y = 0.10x+y = 0.05x+y = 0.10x+y = 0.15(222)x+y = 0.20Intensity (arb. units)2 (Degrees) Figure 4-2. XRD profiles of LuxNdyCe1x-yO2and Gd0.056Y0.044Ce0.90O2taken at room temperature.
70 3040506070 Nd0.10Ce0.90O2-Sm0.10Ce0.90O2-Gd0.10Ce0.90O2-Y0.10Ce0.90O2-CeO2Lu0.10Ce0.90O2-(111) (400) (200) (220) (311) (222)Intensity (arb. units)2 (Degrees) Figure 4-3. XRD profiles of D0.10Ce0.90O2taken at room temperature. The effect of total dopant concentration ( x+y) on the lattice constant ao of the cubic fluorite structure of LuxNdyCe1 -x-yO2was studied. The ao of the calcined powders of LuxNdyCe1 -x-yO2, and LuxCe1xO2with different total do pant concentrations was calculated using maximum likelihood estimati on method with tungsten as an internal standard.79 Figure 4-4 shows the variation of the elastic strain present in the cubic fluorite lattice of LuxNdyCe1 -x-yO2, NdxCe1xO2and LuxCe1xO2as a function of total dopant concentration. The lattice constant data for the different compositions of NdxCe1xO2was taken after Stephens et al.80 Elastic strain is calculated using the following equation: Elastic strain a aao (4-6)
71 where a is the lattice constant of a pure ceria. Positive and negative elastic strain is respectively observed upon the separate addition of Nd3+ and Lu3+. By contrast, there is almost no elastic strain present in the fluorite lattice of LuxNdyCe1 -x-yO2-, even at high dopant concentrations. This validates the hy pothesis that, when combined, the positive elastic strain generated by t he addition of the larger Nd3+ dopant ion is compensated by the negative elastic strain caused by the addition of smaller Lu3+ dopant ion. Further, the lattice parameter of Gd0.056Y0.044Ce0.90O2ceramic was calculated using extrapolation method.66 The estimated lattice parameter of Gd0.056Y0.044Ce0.90O2( a = 5.4133 0.0018 ) was close to that of standard value of ceria ( a = 5.41134 ) at room temperature which is similar to what we obtained in LuxNdyCe1 -x-yO2system.47 0.000.050.100.150.20 -0.001 0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 LuxNdyCe1 -x-yO2NdxCe1 -xO2LuxCe1 -xO2Elastic StrainMol. Dopant Concentartion ( x+y) Figure 4-4. Elastic strain as a function of dopant concentration. La ttice parameter data for NdxCe1xO2were taken after Stephens et al.7,80
72 Ionic conductivity: Ionic conductivity measurem ent was performed using ac impedance spectroscopy from 250 700oC, in air. The details of this technique can be found in Chapter 2. Figure 4-5 shows the grain ionic conductivit y of Gd0.10Ce0.90O2, Lu0.054Nd0.046Ce0.90O2, and Lu0.10Ce0.90O2as a function of temp erature. Further, the grain ionic conductivity data of Gd0.10Ce0.90O2taken from literature were also plotted. It can be seen that the grain ionic conductivity data follo ws Arrhenius behavior. The typical correlation coefficients for a linear least squares fit lie between 0.9995 and 0.9999. Comparison of the grain ionic conductivity for co-doped ceria with that of singly doped ceria is shown in Figure 4-1. As mentioned earlie r, the grain conductivity values of different doped c eria systems are taken from literature. It can be seen that Lu0.054Nd0.046Ce0.90O2exhibits higher grain ionic conductivity than either of Lu0.10Ce0.90O2and Nd0.10Ce0.90O2, which is consistent with the elastic strain behavior observed in Figure 4-4. However, in the intermedi ate temperature range, the grain ionic conductivit y of Lu0.054Nd0.046Ce0.90O2is lower than that of Gd0.10Ce0.90O2. Thus, even though Lu0.054Nd0.046Ce0.90O2 shows higher ionic conducti vity than singly doped ceria (Lu3+ or Nd3+), its ionic conductivity is less than that of Gd0.10Ce0.90O2. It may be due to the large ionic radius difference between the Lu3+ and Nd3+ dopant cations ( rLu 3+ =0.977 and rNd 3+ = 1.109 ).48 On testing the Gd0.056Y0.044Ce0.90O2system ( rGd 3+ =1.053 and rY 3+ = 1.019 )48, it is found that the ionic conductivity value of Gd0.056Y0.044Ce0.90O2lies in between that of Gd0.10Ce0.90O2and Y0.10Ce0.90O2(see Figure 4-4). Thus, even though both co-doped c eria systems show minimal elastic strain, their ionic conductivity is less than that of Gd0.10Ce0.90O2. This shows that minimizing elastic strains is good but not enough. Here, it is important to note that in the structure-
73 property relationship proposed by Kim22, both elastic strain and ionic conductivity are measured in different conditions. Ionic conductivities for all the doped ceria electrolytes were measured at high temperatures (~ 400oC), while lattice parameters were calculated at room tem perature. This will be revisited in the Chapter 5 where structureproperty relationships were determined at hi gher temperatures. T hat will provide the correct rc values of the dopant that causes neither expansion nor contraction in the host ceria lattice at high temperatures. 1 01 21 41 6 -1 0 1 2 700600500 400 1000/T (K-1) Lu 0.054 Nd 0.046 Ce 0.90 O 2Lu 0.10 Ce 0.90 O 2Gd 0.10 Ce 0.90 O 2Gd 0.10 Ce 0.90 O 2(A fter Zhang et al.) Gd 0.10 Ce 0.90 O 2(After Steele)Log(grainT[S.cm-1.K])T (oC) Figure 4-5. Arrhenius plot for the bulk ionic conductivity of Gd0.10Ce0.90O2, Lu0.10Ce0.90O2, and Lu0.054Nd0.046Ce0.90O2systems in air. Comparison between the present wo rk and literature13,77 for the grain ionic conductivity of Gd0.10Ce0.90O2.
74 18.104.22.168 Summary and conclusion Based on critical dopant ionic radius, co -dopant strategy was used to investigate the effect of elastic strain in the lattice on the grain ionic conductivity of the doped ceria. The Lu3+ and Nd3+, and Y3+ and Gd3+ were selected as co-dopant pairs for the host CeO2 and they were added in a proportion such that the weighed av erage dopant ionic radius matches rc for all the compositions. Elastic strain present in LuxNdyCe1 -x-yO2 and Gd0.056Y0.044Ce0.90O2systems was calculated to be negligible when compared to singly doped ceria systems. It was observed that both Lu0.054Nd0.046Ce0.90O2 and Gd0.056Y0.044Ce0.90O2ceramics show lower grain i onic conductivity than that of Gd0.10Ce0.90O2. The results clearly suggest t hat the co-dopant st rategy based on rc does not lead to ionic conducti vity higher than that of Gd0.10Ce0.90O2. However, rc proposed by Kim is derived through the empiri cal relationship, which was obtained for room temperature. The clear view of this co-doping strategy will only be possible, if rc is obtained at higher temperatures where the ionic conductivity of these materials is measured. This topic will be revisited in the Chapter 5. 4.2.2 Density Functiona l Theory Calculation Many of the properties of materials ar e dependent upon interatomic interactions which include the nuclei-nuclei, nuc lei-electron, and electron-electron.81 Density function theory (DFT) incorporates these interactions by solving Schrdingers equations typically by using a pproximations like Kohn-Sham82 to obtain electronelectron exchange and correlation functions. DFT provides high mate rials fidelity and is successfully used to describe different types of material systems (metals, semiconductors, oxides, etc.). A review of the DFT approach to solve materials science related problems can be found elsewhere.81 In this section, co-doping strategy (for
75 ceria) based on the DFT predict ion of the interaction ener gy between oxygen vacancy and dopant cation will be described. The ionic conductivity of co-doped ceria with different dopant concentrati on will be compared with Gd0.10Ce0.90O2. Further, the activation energy and the preexponential coefficient of co-doped sample will be presented in detail. 22.214.171.124 Theory In the recent DFT work by Andersson et al.83 has shown that the total interaction energy values for the oxygen va cancy sitting in nearest neighbor (NN) site and the next to nearest neighbor (NNN) site for Pm3+ dopant are almost same. The defect properties were modeled within 3 2 2 super cells with 4.2 % of dopant content. Figure 4-6 shows the total inter action as a function of dopant ionic radii for different trivalent dopants. This figure is modified after Andersson et al.83 The total interaction energy which is the sum of electrostatic and elas tic interaction energies, between trivalent dopant cation site and the oxygen vacancy residing in NN and NNN sites, is calculated for different trivalent dopant cations using DFT calculations. Th e negative sign in the interaction energy imply attractive in teractions. In its original form, Figure 4-6 is plotted between the total inter action energy and the at omic number of the dopant cation. The atomic number was used to de scribe the doped ceria system as this is the only physical parameter (except for the crystal structure) that is used as an input in DFT calculations. The specification of the atom ic number in DFT ca lculations takes into account ionic radii, and interactions between electrons. Ho wever, from the expe rimental point of view, it is reasonable to plot the interacti on energy as a function of ionic radii of the dopant, as this physical property is more re levant while understanding the effect of different dopants on the crystal structure.22
76 126.96.36.199.121.101.081.061.041.021.00 -0.6 -0.4 -0.2 0.0 La Total Interaction Energy (eV)Dopant Ionic Radius ()Pr Ce Nd Pm Sm GdTbDyHoEr Total Interaction Energy (NNN sites) Total Interaction Energy (NN sites) rEffective Sm/Nd Figure 4-6. Total interaction energy between dopant cation and oxygen vacancy sitting in nearest neighbor (NN) and next to nearest neighbor (NNN) site (of dopant cations) for different trivalent cations.48,83 The main idea of the present co-doping stra tegy is to maximize the concentration of equi-interaction energy oxygen vacancy sites inside the host ceria lattice. Figure 4-7 shows another view of cubic fluorite struct ure of doped ceria, where oxygen ions form the lattice and cerium ions (and dopant ions ) sit inside the latti ce. Here, oxygen vacancy sits in NN site and NNN site of dopant cation. The probability of oxygen vacancy to sit in any particu lar site depends upon its inte raction energy with dopant cation site. The oxygen vacancies will try to diffuse to the sites where the interaction energy is higher. However, once they ar e trapped in these higher interaction energy sites, they will require additional thermal energy to jump out of there. If all the sites are
77 of equal interaction energy, then there will be no deep traps for oxygen vacancies. Thus, the probability of oxygen vacancies jump out of the traps increases which will promote diffusion through the material and mini mize the concentration of local defect structures. A B Figure 4-7. Cubic fluorite structure of doped ceria, wher e oxygen vacancy reside in nearest neighbor (NN) site A) and next to nearest nei ghbor (NNN) site B) of dopant cation. It can be seen from Figure 4-6 that for Pm3+, and for ions to the right of Pm3+ in the lanthanide series, oxygen vacancy interact s strongly with the dopant cation when it occupies NN position of the dop ant cation. Moreover, cations with atomic number less than the Pm3+ in the lanthanide series usually hold oxygen vacancies in the NNN position of the dopant ca tion. Thus, around Pm3+ there will be no site preference for oxygen vacancies, resulting in an increase in the number of equi-interaction energy sites which facilitate the oxygen vacancy diffusion. It was predicted then that Pm3+ doped ceria should exhibits higher ionic co nductivity than any other singly doped ceria material.
78 188.8.131.52 Materials selection Using DFT results, it was shown that Pm3+ is the ideal trivalent dopant cation for ceria to exhibit highest ionic conductivity. Unfortunately, Pm is radioactive and cannot be used for electrolyte applications. The ideal dopant should have an effective atomic number around Pm3+ (61) which shows ionic radius of 1.093 .48 Therefore, as predicted by Andersson et al., a co-dopant scheme using Sm3+ and Nd3+ can provide an experimental scenario for testi ng this hypothesis. In terms of ionic radius, 1:1 Sm:Nd co-doping can be seen as having a weighed averag e dopant ionic radius (Effective NdSmr/) of 1.094 which is close to Pm3+ ionic radius of 1.093 (shown in Figure 4-6).48 Thus, based on the Pm3+ atomic number, in the present work the effect of co-dopant pair Sm3+ and Nd3+ on the ionic conductivity of ceria based electrolyte is investigated. Dopants were added in equal proportion to achiev e the effective atomic number of Pm3+, i.e. 61. By doing so, similar total interactions betw een the oxygen vacancies sitting in NN and NNN sites and the dopant cation are expect ed, with the consequent enhancement of the ionic conductivity. In order to study the effect of dopant concentration on the crystal structure and ionic conductivity, polycrystalline samples of Smx /2Ndx /2Ce1xO2(where x = 0.01, 0.03, 0.05, 0.08, 0.10, 0.12, 0.15, 0.18, and 0.20) were synt hesized by conventional solid state reaction method (as descr ibed in Chapter 2). As Gd0.10Ce0.90O2exhibits the highest ionic conductivity among doped ceria syst ems, therefore for comparison, phase pure sample of Gd0.10Ce0.90O2was also processed using the identical experimental procedure. Densities of all the sintered samples were measured in water using Archimedess principle and were estimated to be 98% of theoret ical density or above.53
79 184.108.40.206 Results and discussion In this section, the microstructure, lattice parameter, and the ionic conductivity of Smx /2Ndx /2Ce1xO2will be covered. Also, th e comparison between the ionic conductivity of Smx /2Ndx /2Ce1xO2and Gd0.10Ce0.90O2will be presented. Further, the effect of dopant concentration on the crysta l lattice and the ionic conductivity at intermediate temperatures will be discussed. Microstructure analysis: The microstructure of the sintered pellets of Smx /2Ndx /2Ce1xO2was studied using SEM. A typical micrograph of the surface of the Sm0.05Nd0.05Ce0.90O2sample is shown in Figure 4-8. There is almost no residual porosity observed in Figure 4-8 which is consistent wi th the m easured density of the sintered pellet. The mean lineal intercept technique was used to determine the mean grain sizes. The average grain size in each sample was found to be in the range of 4 5 m. Figure 4-8. Scanning electron micrograph of the surface of Sm0.05Nd0.05Ce0.90O2sintered pellet.
80 X-ray diffraction analysis: To verify the complete diss olution of dopants in ceria, phase analysis was performed using X-ray di ffraction (XRD). Curved position-sensitive diffractometer was used to obtain the XRD pattern of each composition using Cu K radiation. Figure 4-9 shows the XRD pr ofiles measured at room temperature for all the compositions of Smx /2Ndx /2Ce1xO2. Tungsten was used as an internal standard in the powder sample. It can be observed that all the compositions studied are single phase with a cubic fluorite structure like pure CeO2. 3040506070 (400)* *(222) (311) (220)x = 0.20 x = 0.18 x = 0.15 x = 0.12 x = 0.10 x = 0.08 x = 0.05 x = 0.03 x = 0.01Intensity (Arb. Units)2 (Degrees)* W Standard (200) (111)x = 0Smx /2Ndx /2Ce1xO2Figure 4-9. XRD patterns of Smx /2Ndx /2Ce1xO2measured at room temperature. Tungsten was used as an internal standard. Figure 4-10 shows the (111) and (200) pea ks of the materials and (110) peak of the tungsten standard for differ ent dopant concentration. It can be observed that with the increase in the total dopant content ( x ), (111) and (200) peaks of the material shift to
81 the lower 2 angles while (110) peak positions of the tungsten standard remain stable in all the compositions. This clearly indicates that the doped ceria lattice expands with the increase in dopant content. 30 334041 (110) x = 0.20 x = 0.18 x = 0.15 x = 0.12 x = 0.10 x = 0.08 x = 0.05 x = 0.03 x = 0.01 Intensity (Arb. Units)2 (Degrees) W Standard (200) (111)x = 0Smx /2Ndx /2Ce1xO2Figure 4-10. Shift in (111) and (200) XRD peaks position of Smx /2Ndx /2Ce1xO2with increase in dopant concentration ( x ). Tungsten was used as an internal standard. Precise lattice constant ( ao) for all the compositions of Smx /2Ndx /2Ce1xO2was calculated using the least-s quares extrapolation method.66 Figure 4-11 shows the lattice parameter of Smx /2Ndx /2Ce1xO2solid solution as a function of total dopant concentration ( x ). For comparison, the la ttice parameter data of NdxCe1xO2as a function of dopant content is also shown taken after Stephens et al.80 It can be seen that for Smx /2Ndx /2Ce1xO2system, the lattice parameter increases linearly with the
82 increase in Sm3+ and Nd3+ concentration following Vegards law.84 Using a leastsquares fitting algorithm, a linear relation ship was obtained between the lattice parameters of Smx /2Ndx /2Ce1xO2and dopant content ( x ). This can be represented as, ao( x ) = 0.5412 + 0.0154 x (4-7) 0246810121416182022 5.405 5.410 5.415 5.420 5.425 5.430 5.435 5.440 5.445 5.450 Nd x Ce 1x O 2after Stephens et al. Sm x /2 Nd x /2 Ce 1x O 2Lattice Constant ( ) x Dopant Concentration (%) Figure 4-11. Lattice parameter as a fu nction of dopant concentration (mol%) for Smx /2Ndx /2Ce1xO2and NdxCe1x O2taken after Stephens et al.67,80 It is important to note that a gradual non-linear increase in t he lattice parameter with dopant concentration was observed in ceria-neodymium system. Similar quadratic expansion was also reported in YbxCe1 -xO2, DyxCe1 -xO2, GdxCe1 -xO2, and YxCe1 -xO2systems.85 The lattice parameter behavio r for these systems was fitted using the quadratic function of the dopant mole fraction. T he first-order term in the quadratic relationship was interpreted as the ionic radii mismatch between the dopant ( rd) and the host cations ( rh), which consequently results in the expansion (for rd > rh) or
83 contraction (for rd < rh) of the host lattice with the incr ease in dopant concentration. The coefficient of the second-degree term was interpreted as a measure of the degree of ordering present in the material. The decreas e in the lattice param eter at higher dopant concentration can be attributed to the effect of the attractive interactions between dopant cations and oxygen vacancies which tend to contract the unit cell. As seen from equation (4-7), the coefficient of the second-degree term in the lattice parameter dopant co ntent relationship in Smx /2Ndx /2Ce1xO2system is essentially negligible. This points to the theory that the oxygen vacancy ordering in Smx /2Ndx /2Ce1xO2system is significantly lower compared to other ceria based systems. Ionic conductivity: Ionic conductivity measurem ent was performed on the sintered pellets of Smx /2Ndx /2Ce1xO2using the ac impedance spectroscopy technique in air, from 250 oC. It is essential to note that it is the grain ionic c onductivity that is the intrinsic property of the material; whereas, total conductivity includes microstructural, impurity and other contributions to the samples resistance.86 Thus, in the present work, comparison between the grain ionic conductivity ra ther than the total ionic conductivity for different materials is reported. Figure 4-12 shows the typi cal impedance s pectrum of an electroded polycrystalline ceramic sample of Sm0.075Nd0.075Ce0.85O2. Three arcs, corresponding to the polarizations of the gr ain, grain boundar y, and the electrode, were shown in the impedance spectrum. The high frequency semi-circle in the spectrum which is associated with intra-grain polar ization is also shown in the inset.
84 0.0 5.0x1041.0x1051.5x105 0.0 5.0x1041.0x1051.5x105 0.0 4.0x1038.0x103 0.0 4.0x1038.0x103 Grain-Z// [ohms]Z/ [ohms]Electrode -Z// [ohms]Z/ [ohms]0.1 Hz 174 Hz 173 KHz Grain Boundary Grain Figure 4-12. Typical impedance s pectrum of a polycrystalline Sm0.075Nd0.075Ce0.85O2ceramic measured in air at 300oC. Inset shows the in tra-grain polarization contribution. The grain ionic resistance was calculated by fitting the observed impedance spectrum with the analog equival ent circuit having three para llel resistor-constant phase element (R-CPE) pairs in series. Further details about impedance spectroscopy are given in Chapter 2. The r eported grain ionic conductivity for each composition is the average of the grain ionic conductivity values measured for three different test samples of the same composition with varying geometrical aspect ratio. The typical error assiated with the values of the reported grain ionic conductivity is < 2%. Furthermore, no density correction has been m ade on the reported grain io nic conductivity data as the density of all the sintered pellets is 98% of the theoretic al density or above.
85 1.01.11.21.220.127.116.11.71.81.92.02.1 -2 -1 0 1 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 700600500400300 200 5 mol% 3 mol% 1 mol%Log[T(S.cm-1.K)] 1000/T (K-1) Dilute RegimeSmx /2Ndx /2Ce1xO2T (oC)8 mol% 12 mol%10 mol%15 mol%Log[T(S.cm-1.K)] 18 mol%475oC Figure 4-13. Arrhenius plots for t he grain ionic conductivity of Smx /2Ndx /2Ce1xO2. Dopant concentration ( x ) is reported in mol%. Figure 4-13 shows the Arrhenius plot of the grain ionic conductivit y for Smx /2Ndx /2Ce1xO2. In the dilute regime, the grain ionic conductivity increases with the increase in the dopant concentra tion. This can be explaine d as there are limited defect interactions, and the ionic conductivity mainly depends upon the oxygen vacancy concentration. However above 5 mol%, the grain ionic conductivity decreases with increase in total dopant concentration ( x ) in the lower intermediate temperatures (<
86 450oC). This can be attributed to the incr ease in the defect interactions, which consequently decreases the number of lo w energy migration paths for oxygen vacancies, and lower the mobile oxygen vacancy concentration. 0.640.680.720.760.80 4.4 4.8 5.2 5.6 6.0 ( x = 0.18)( x = 0.15)( x = 0.12) ( x = 0.10) ( x = 0.08) ( x = 0.05) ( x = 0.03) Log oActivation Energy (eV)( x = 0.01) Figure 4-14. Log o as a function of activa tion energy measured below 475oC for Smx /2Ndx /2Ce1xO2. Line connecting data points for each composition is shown for visual aid. Further, it can be seen from the Figure 4-13 that the Arr henius plots for the grain ionic conductivity of all the compositions (> 5 mol %) converges to a common transition point, i.e. around 475oC. Around this temperature, conductivity becomes independent of dopant concentration. Such behavior was also previously reported in a wide range of materials and is commonly known as Meyer-Neldel rule.87 Hohnke reported a similar trend for the ionic conductivity of GdxCe1 -xO2.88 Recently, Stephens et al. observed
87 Meyer-Neldel behavior for t he ionic conductivity in NdxCe1 -xO2at higher Nd3+ concentration.80 Based on the Meyer-Neldel rule, t he magnitudes of t he pre-exponential factor ( o), and activation energy ( H ) can be linked by a following relationship, log o = H + (4-8) where and are constants. Figure 4-14 is a plot between the log o and H for Smx /2Ndx /2Ce1xO2. For the higher dopant concentration (> 5 mole %), log o shows a linear relationship with H which is consistent with the above mentioned relationship. Using least-squares linear r egression, the values for and were found to be 6.6839 and 0.6443, respectively. According to Almond et al., To, which is a common transition point for the Arrhenius plots for the conductivity data of all the composition, corresponds to an orderdisorder transition in the mobile ion subl attices in a ionically conductive material.89 Thus, above To, mobile ions no longer remained ordered on their lattice sites. According to Nowick et al., To can be seen as a transition temperature between the stage where most of the carriers are bound at various traps and the stage where all the carriers are free.87 This suggests that above To, almost all the ox ygen vacancies are free to migrate for any dopant concentration. It is difficult to confirm this hypothesis as the Arrhenius conduc tivity plots of Smx/2Ndx/2Ce1-xO2exhibit a gradual, rather than a sharp, change in slope with temperature. In addition, the extent of formation of local defect structures also increas es with the increase in dopant concentration; therefore, higher oxygen vacancy concentration systems wi ll require higher thermal energy (or temperature) for the oxygen vacancies to surpass the dopant interaction barrier.
88 Furthermore, for the dopant concentration in t he dilute regime, Arrhenius plots never exhibit any cr ossover point. 0246810121416182022 1.0x10-22.0x10-23.0x10-24.0x10-2 650oC 600oC GdxCe1xO2after Zhang et al. Gd0.10Ce0.90O2-(This work) Smx /2Ndx /2Ce1xO2500oC 450oCGrain Ionic Conductivity (S.cm-1)x, Dopant Concentration (%)550oC Figure 4-15. Grain ionic conductivity as a function of x dopant concentration (mol%) in Smx /2Ndx /2Ce1xO2and GdxCe1xO2. The grain ionic conductivity of Gd0.10Ce0.90O2(this work) at different te mperatures is also plotted.53,67,77 Figure 4-15 shows the grain ionic conductivity of Smx /2Ndx /2Ce1xO2at different temperatures. For compar ison, measurements of Gd0.10Ce0.90O2synthesized under identical experimental conditions in our lab ar e presented, as well as recent data on the same GdxCe1-xO2composition from Zhang et al.77 Our Gd0.10Ce0.90O2data is comparable with grain ionic conductivity values reported by Zhang et al. At lower temperatures, the grain ionic conductivity of Smx/2Ndx/2Ce1-xO2looks comparable to that of GdxCe1xO2. However, with the in crease in temperature Smx /2Ndx /2Ce1xO2
89 clearly surpasses GdxCe1xO2. Among all the compositions of Smx /2Ndx /2Ce1xO2, Sm0.09Nd0.09Ce0.82O2exhibits the highest grain ionic conductivity above 600oC (3.47 10-2 Scm-1 at 650oC). The reported ionic conductivity is 35% higher than that of GDC. Below 600oC, Sm0.075Nd0.075Ce0.85O2is the highest ionically conductive material in doped ceria material. 1.01.11.21.31.41.5 0.0 0.5 1.0 1.5 700650600550500450400 1000/T(K-1) Sm 0.075 Nd 0.075 Ce 0.85 O 2Sm 0.05 Nd 0.05 Ce 0.90 O 2Gd 0.10 Ce 0.90 O 2Log[grainT(S.cm-1.K)]T(oC) Figure 4-16. Arrhenius plot for the grain ionic conductivity of Sm0.075Nd0.075Ce0.85O2and Sm0.05Nd0.05Ce0.90O2. The grain ionic conductivity of Gd0.10Ce0.90O2(our work) at different tem peratures is also plotted.67
90 At 550oC, the grain ionic conductivity of Sm0.075Nd0.075Ce0.85O2and Sm0.05Nd0.05Ce0.90O2was found to be 14.0 10-3 Scm-1 and 12.2 10-3 Scm-1, respectively. The measured grain ionic conductivity data of Sm0.075Nd0.075Ce0.85O2and Sm0.05Nd0.05Ce0.90O2was 30% and 14% higher than that of Gd0.10Ce0.90O2(~ 10.7 10-3 Scm-1), respectively, which is the widely acc epted material to exhibit highest ionic conductivity among doped ceria. This is a proof that, on the basis of conductivity, Smx /2Ndx /2Ce1xO2is the most promising ceria-based el ectrolyte for SOFCs operating in the intermediate temperature range. As mentioned earlier, the higher grain ionic conductivity of Smx /2Ndx /2Ce1xO2electrolytes can be attributed to the incr ease in the number of equi-interaction energy sites which lower the activation energy and facilitate oxygen ion diffusion in the lattice. In addition to the activation energy, usi ng co-dopants also enhances the configuration entropy of the system which, in turn, incr eases the pre-exponential factor in the Arrhenius relationship. Figure 4-16 shows the Arrhenius plot for the grain ionic conductivity of Sm0.05Nd0.05Ce0.90O2, Sm0.075Nd0.075Ce0.85O2and Gd0.10Ce0.90O2in intermediate temperatures. The results clearly show that Sm0.05Nd0.05Ce0.90O2exhibits higher grain ionic conductivity than that of Gd0.10Ce0.90O2at almost all the temperatures. However, in the high intermediate temperature, Sm0.075Nd0.075Ce0.85O2is the material that exhibits t he highest grain ionic conductivity.67 Table 4-1 compares the H and o of Sm0.05Nd0.05Ce0.90O2, Sm0.075Nd0.075Ce0.85O2and Gd0.10Ce0.90O2. The error in the values of H and o was calculated by measuring the grain ionic conductivity of the test sample s with different geometrical aspect ratios. Given the error associated with the exper imentation, it can be seen from the Table 4-1
91 that Sm0.05Nd0.05Ce0.90O2exhibits higher than that of Gd0.10Ce0.90O2. The result can be attributed to the higher o and lower H of Sm0.05Nd0.05Ce0.90O2than that of Gd0.10Ce0.90O2. This validates the hypothe sis that co-dopants based on Pm3+ atomic number (with ionic radii in the close proximity to Pm3+) results in the enhancement of grain ionic conductivity. Although, Sm0.075Nd0.075Ce0.85O2exhibits the highest ionic conductivity among doped ceria, it shows higher H than that of Gd0.10Ce0.90O2and Sm0.05Nd0.05Ce0.90O2. However, it possesses higher o which shows dominant effect on ionic conductivity compared to H This indicates that fo r a high ionic conductive material, both a low H and a maximum o are required. Table 4-1. Comparison of the H and log o for Gd0.10Ce0.90O2, Sm0.05Nd0.05Ce0.90O2. and Sm0.075Nd0.075Ce0.85O2measured below 550oC. Further, of these oxide materials at 550oC is also reported.67 Materials (Scm-1.K)] H (eV) Log10 [ o (Scm-1.K)] Gd0.10Ce0.90O210.7 0.4 10-30.6900 0.00305.184 0.020 Sm0.05Nd0.05Ce0.90O212.2 0.3 10-30.6867 0.00035.215 0.002 Sm0.075Nd0.075Ce0.85O214.0 0.2 10-30.7634 0.00385.5565 0.020 Figure 4-17 shows the grain ionic conductivity trend of Smx/2Ndx/2Ce1-xO2as a function of dopant concentration at different temperatures. Initially, the grain ionic conductivity increases with the increase in d opant concentration. However, it reaches a maximum and then substantially drops beyond certain dopant concentration at lower temperatures. Degradation of the grain ionic conductivity at higher dopant concentration is related to the formation of local defect structures, which lowers the mobile oxygen vacancy concentration. Moreover, it is interesting to observe that as the temperature increases, the maxima of the grain ionic conductivity in the Figure 4-17 shifts toward the higher dopant concentration. A sim ila r trend was also observed in GdxCe1-xO2and YxCe1-xO2by Zhang et al.77 and Tian et al.62, respectively. This
92 behavior indicates the dissociation of complex defect structures (present mainly at higher dopant concentration) at higher temperatures to give mobile oxygen vacancies. Since the ionic conductivity depends main ly on the mobile oxygen vacancy concentration, therefore, with the increase in temperatur e ionic conductivity maxima shift toward higher dopant concentrations. 02468101214161820 10-510-410-310-210-1 250oC 700oC 650oC 600oC Grain Ionic Conductivity (S.cm-1)x, Dopant Concentration (%) 300oC 400oC 350oC 450oC 500oC 550oC Figure 4-17. Grain i onic conductivity of Smx /2Ndx /2Ce1xO2as a function of x dopant concentration (mol%). Lines connecting dat a points for different temperatures is shown for visual aid. Since the Arrhenius plot for the grain ioni c conductivity exhibits a gradual decrease in the slope at higher intermedi ate temperatures, it is difficu lt to distinguish the regimes where the association-dissociat ion of oxygen vacancies is taking place. As mentioned earlier, the cross-over point of Arrhenius plots for the grain ionic conductivity of
93 Smx /2Ndx /2Ce1xO2(for x > 0.05) was interpreted as the association dissociation temperature of oxygen vacancies, hence, for t he purpose of this study it is assumed that the transition occurs at 475oC. Using a least-squares algorithm, high and low temperature region of Arrhenius plot is linearly fitted. The fit of the data is generally quite good, with correlation coefficients for a linear least-squares fit between 0.9997 and 0.9999. Activation energy values for the ionic conduction in these materials are calculated from the high and low temperature gradients of the fitted Arrhenius plots. 048121620 0.60 0.65 0.70 0.75 1.02 1.04 migrationH Measured below 475oC Measured above 475oCn associatio migrationH H Smx /2Ndx /2Ce1xO2-Activation Enerrgy (eV)x Dopant Concentration (%) Figure 4-18. Activation energy for oxygen ion diffusion in Smx /2Ndx /2Ce1xO2measured below 475oC and above 475oC, as a function of x Line connecting data points for different temperatures is shown for visual aid. Figure 4-18 shows the plots of activation e nergy for the grain ionic conduc tivity of Smx /2Ndx /2Ce1xO2, measured below 475oC and above 475oC, as a function of dopant
94 concentration ( x ). The activation energy m easured in the range of 475oC 250oC, can be seen as a sum of two contribut ions, the association enthalpy ( Hassociation) and migration enthalpy ( Hmigration). Initially, the addition of dopant cations in ceria decreases the activation energy for oxygen vacancies diffusion. However, around 5 mol% dopant concentrations, activation energy pl ot exhibits a minimum. A similar trend of activation energy as a function of dop ant concentration was earlier reported by Faber et al.14 for various rare earth oxi des doped ceria systems. Acco rding to Faber et al., this decrease is due to the presence of attracti ve interactions between dopant cations and oxygen vacancies, with the in crease in dopant concentration leading to an increase in the number of percolation paths. However, above 5 mol% dopant concentration, with increasing dopant concentration, there is an increasing probability that the dopant cations sit in close proximity. This resu lts in the formation of deep traps that accommodate all existing mobile oxygen vacancies.14 Consequently, the oxygen vacancy diffusion paths become more resistiv e and therefore the activation energy for ionic conduction increases. As most of the oxygen vacancies above 475oC are assumed to be mobile, the reported activation energy measured above 475oC (in Figure 4-18) is equal to the migration enthalpy of the oxygen ion diffusion. It can be seen that the Hmigration also experiences a minima at 5 mol% dopant c oncentration. Similar results have been observed for activation energy measur ed at higher temperatures for NdxCe1xO2system, by Stephens et al.80 However, based on nuclear magnetic resonance (NMR) studies, Fuda et al.90 concluded that the migration enthalpy is independent of dopant concentration. Their results are not in agr eement with the migrat ion enthalpy data.
95 According to Ralph et al.91, compositional independence of migration enthalpy only holds good in the dilute regime. Assuming the dilute regime in Smx /2Ndx /2Ce1xO2system extends to 5 mol%, it can be seen that the migration enthalpy decreases with the increase in dopant concent ration. This behavior of Hmigration can be explained in terms of continuous network of oxygen vacancies. As the dopant concentration increase in the dilute regime, number of low energy migration pat hs increase which effectively decrease the migration enthalpy. 18.104.22.168 Summary and conclusion In search of advanced materials that show improved ionic co nductivity, both preexponential factor and the acti vation energy (in the Arrhenius relationship) for oxygen diffusion need to be optimized. Co-dopants we re used to suppress the oxygen vacancy ordering and increase the preexponential factor in the system. The relative co-dopant concentration was based on the effective atomic number of Pm3+. This novel approach was based on the computationa l work which suggests using co-dopant with an average effective atomic number of Pm (61). By doing so, the in crease in the number of equiinteraction energy sites of oxygen vacancy wa s expected, which as a result, facilitates oxygen diffusion. This in turn increased the ionic conductivity of the material. Therefore, a co-doping scheme using Sm3+ and Nd3+ provided the experimental scenario to test this hypothesis. Lattice parameter as a function composition obeys Vegards law which indicates, short range oxygen ordering in Smx /2Ndx /2Ce1xO2system is not as pronounced as in other singly doped ce ria system. Among all the compositions of Smx /2Ndx /2Ce1xO2, Sm0.09Nd0.09Ce0.82O2exhibits the highest grain ionic conductivity above 600oC (3.47 10-2 Scm-1 at 650oC). The grain ionic conductivity of Sm0.09Nd0.09Ce0.82O2was found to be 35% higher than that of
96 Gd0.10Ce0.90O2at 650oC. Below 600oC, Sm0.075Nd0.075Ce0.85O2is the highest ionically conductive material in doped ceria material. At 550oC, the grain ionic conductivity of Sm0.075Nd0.075Ce0.85O2was found to be 30% higher than that of Gd0.10Ce0.90O2. Thus, it was shown that co-doping based on Pm atomic number results in the enhancement of ionic conductivity. Further, at higher dopant concentrations, Smx /2Ndx /2Ce1xO2system follows Meyer-Neldel rule. Taking the comm on transition point of Arrhenius plots as association-dissociation tem perature of oxygen vacancie s, migration enthalpy in Smx /2Ndx /2Ce1xO2system was calculated. It was s hown that the migration enthalpy do have dependence on dopant concentration in dilute regime. Furthermore, shift in the grain ionic conductivity maxima toward high er dopant concentration with the increase in temperature, was observed in Smx /2Ndx /2Ce1xO2system. 4.3 Processing Effects on Ionic Conductivity In light of the comparison here proposed between singly doped and co-doped electrolytes, it is important to note that, fo r example, a literature survey of the reported grain ionic conductivity of Gd0.10Ce0.90O2reveals a wide range of values as presented in Figure 4-1 and Figure 4-19. Inspection of the ex perimental procedure followed in each work indic ates a strong correlation bet ween the powder synthesis technique and the conductivity observed. Typically, co-precipitation techniques lead to higher conductivity values and solid state processing methods result in lower ones.63 Figure 4-19 compares the gr ain ionic conductivity at 600oC of singly doped ceria reported here with that from literature. Since in our work solid state processing was employed, as expected, the grain io nic conductivity here reported for Gd0.10Ce0.90O2is lower than that of Steele.13 On the other hand, its conducti vity is higher that of Reddy et al.61 Similar is the case with Sm0.10Ce0.90O2. However, our Nd0.10Ce0.90O2exhibits
97 much higher grain ionic conductivi ty than reported by Stephens et al.80 and Li et al.78 It can be inferred that besides co mposition, processing related variables such as grain size distribution, dopant redistribution within the grain, grain impurities, and extended grain defects also play a vital role in the enhancement of the grain ionic conductivity of the electrolytes. 0.1050.1080.111 0.6 0.9 1.2 1.5 (2) (1) (3) (4) From the Literature Our WorkNd3+Sm3+Gd3+ Log(grainT [S.cm-1.K])Ionic radius of trivalent dopant cation (nm)(5) Figure 4-19. Grain ionic c onductivity comparison at 600oC between our work and literature. Grain i onic conductivity data (1), (2), (3 ), (4), and (5) is taken from Reddy et al.61, Steele13, Zhan et al.76, Stephens et al.80, and Li et al.78, respectively. Clearly, a consistent set of data for the ionic conductivity of all the doped ceria systems is required in which the processi ng variable to synthesize ceramic samples remain constant. This will provide a clear idea about the ideal dopant for ceria which
98 shows the highest ionic conductivity, when all the materials are synthesized under similar processing conditions. After identifying the highest ionic con ductive material, the ionic conductivity can be im proved using lower sintering temperat ure and higher purity processing conditions. The va riation of processing condi tions to improve the ionic conductivity of doped ceria will be discussed in the Chapter 6.
99 CHAPTER 5 STRUCTURE-PROPERTY RELATIONSHI PS IN DOPED CERIA MATERIAL S AT HIGH TEMPERTATURES 5.1 Introduction High ionically conductive elec trolyte materials are critic al for the development of SOFCs operating in intermediate temperatur e range. Acceptor doped ceria materials are potential candidates for this application. For a given dopant valence (dior trivalent) and concentration, oxygen ion conductivity in ceria system depends primarily upon the dopant size. According to Kilner et al.21, maximum ionic conductivity is achieved in the system which shows neither contracti on nor expansion upon the addition of dopant cation. For example, most of the ionic radii of commonly us ed dopant cations lie closer to the ionic radius of Ce4+ than Zr4+. Thus, it is expected that for the same dopant cation, activation energy for oxygen vacancy diffusion involved in doped CeO2 systems must be lower than that in doped ZrO2. From Chapter 2, it is known that activation energy is comprised of migrat ion and association enthalpies. The migration enthalpy in both the systems is almost com parable; therefore, it is the association enthalpy (of local defect structures) that dict ates the oxygen ionic conduct ivity in the material. Table 5-1 compares the migrat ion and as sociation enthalpies in ZrO2 and CeO2 doped with Ca2+.21 It can be seen that the asso ciation enthalpy in doped CeO2 system is lower than that in doped ZrO2, which clearly indicates that the lower ionic radius difference between the dopant ( rd) and host ( rh) cations, results in higher ionic conductivity. Although the difference in the hos t and dopant cations ionic radius ( rh-rd) is important, the creation of ox ygen vacancies in the host lattice (to maintain charge neutrality) by the addition of acceptor dopants, also influences the lattice geometry. The presence of oxygen vacancies in stead of oxygen anions results in the contraction of the
100 lattice. Given (rh-rd) is negligible, the increase in the dopant concentration results in the contraction of the host lattice.92 For example, as presented in Chapter 4, in LuxCe1xO2system, although the ionic radius difference between Lu3+ and Ce4+ is very small (r3+ Lu,VIII= 0.977 and r4+ Ce,VIII= 0.97 ), the increase in the dopant concentration contracts the host ceria lattice.7,48 Table 5-1. Activation energies for ZrO2 and CeO2 systems doped with Ca2+.21 Activation Energy Ca-doped ZrO2 Ca-doped CeO2 Migration Enthalpy (eV) 0.68 0.61 Association Enthalpy (eV) 1.00 0.64 Aleksandrov et al.93 estimated the lattice strain in Y2O3-ZrO2 solid solution as a function of dopant concentration, which was later modified by Ingel et al.94, by adding the dopant size effect on the lattice. Howeve r, the proposed model does not fit to other fluorite AO2 type oxide (where A is a cation) systems, as it does not precisely take into account the effect of oxygen vacancies. In order to compensate the oxygen ion vacancies, some arbitrary values of i onic radii were used. Glushkova et al.92 used both ionic radius difference (between host an d dopant cations), and t he oxygen vacancies creation factors into account to describe the change in the lattice c onstant of lanthanide doped fluorite oxides. The contraction of t he lattice due to the formation of oxygen vacancies was described in terms of an effe ctive oxygen-ionic radius, which becomes smaller with the increase in dopant content. The effective oxygen ionic radius (effective Or ) was defined as, 6.3 1138.1 m reffective O (5-1)
101 where 1.38 is the ionic radius of O2(four-fold co-ordination), m is the molar concentration of lanthanide oxide (LnO1.5) and 3.6 is the correction factor for the association of oxygen vacancies. The units of effective Or is The model shows good agreement with the experimental data for LnO1.5-ZrO2 and LnO1.5-HfO2 systems. However, lattice parameter calculated fr om the model shows deviation from the measured values for LnO1.5-CeO2 and LnO1.5-ThO2 systems. The deviation is even larger, if alkaline dopant ca tions are used instead of LnO1.5. Kim22 studied the elastic strain formed in the host fluorite oxide lattice by the substitution of acceptor dopant cations for host cations. Using multiple regression analysis, empirical relationship for the la ttice parameter change in the doped ceria system, as a function of the difference in the ionic ra dius between host and dopant cation ( rh-rd), dopant valence and dopant concentration was established for room temperature. Equation (5-2) shows the la ttice change for the ceria solid solutions, k kk k Cemz r a 0015.022.0413. 5 (5-2) where aCe (in ) is the lattice parameter of the ceria solid solution at room temperature, rk (in ) is the ionic radius difference ( rk-rh) between kth dopant ( rk) and host ( rh) cations, zk is the valence difference ( zk-zh) between kth dopant ( zk) and host ( zh) cations, and mk is the mole percent of the kth dopant cation. It was assumed that the solid solution of doped ceria obeys Vegards law i.e., a linear relationship exists between lattice parameter and concentration of solute.84 Using equation (5-2), the critical ionic radius ( rc) of the dopant was proposed. Dopant with ionic radius equal to rc, will neither cause expansi on nor contraction in the host fluorite oxide lattice. For the trivalent dopant cation and host ceria, rc was
102 determined to be 1.038 .22 Based on Kilner et al.s prediction, dopant with rc is the ideal dopant for ceria to show highest ionic c onductivity. Thus, the ionic conductivity in doped ceria materials depends upon the difference ( cdrr) between the ionic radius of the dopant cation and rc. The lower the cdrr value for a particular dopant, the higher the ionic conductivity for that system is expected to be. Using a similar argument, Kim suggested that Gd3+ exhibits the highest ionic conductivity as the ionic radius of Gd3+ (r3+ Gd,VIII= 1.053 ) lies close to rc.48 Hong et al.74 proposed a model to estimate latti ce parameter of trivalent doped ceria at room temperature using a different approach. Oxygen vacancy was considered to be one of the species, and a radius (OVr ) was assigned to it which was assumed to be constant with respect to dopant concentra tion and dopant ionic radius. Thus, OVr has a unique value for a given solid solution. Using this approach, lattice parameter ( a ) as a function of dopant concentration ( x ) and dopant ionic radius ( rd) was proposed and is given by, OCe V O Cedrr xrr rr aO 9971.0 3 4 25.025.0 9971.0 3 4 (5-3) where rCe and ro are the ionic radius of Ce4+ and O2-, respectively. Equation (5-3) obeys Vegards law. Using this model the critical ionic radius ( rc) for the trivalent dopant cation was proposed for ceria, the definition of whic h is similar to Kims definition of rc. The rc is defined for which the Vegards slope = 0. For the trivalent dopant cations, the rc value is estimated to be 1. 024 by plotting measured values of Vegards slopes for a number of dopant cations against t he dopant cation ionic radius.
103 In the literature, it has been reported that Sm3+ and Gd3+ are the ideal dopants for ceria to show higher conductivity among other dopant cations.9,71 The ionic radius of Sm3+ (r3+ Sm,VIII= 1.079 ) is higher than that of Gd3+ and shows cdrr value of 0.041 .48 The rc value is taken from the Kims22 proposed model. Also, it is known that Dy3+ shows lower ionic conductivity than that of Gd3+ and Sm3+.9 The cdrr value is lowest for Dy3+ (0.011 ), when compared with Gd3+ (0.015 ), and Sm3+ (0.041 ). These results contradict Kilner et al.s hypothesis. It is important to note that the rc derived by Kim, was obtained from the elastic strain (or lattice parameter) measured at room temperature. The ionic conductivity of doped ceria electrolyte materials is usually measured between 400oC to 700oC. It may be possible that the rc value increases with the increasin g temperature. Hence, it is important to test the Kilner et al.s hypothesis by obtaining rc value at higher temperatures.21 Also important is the fact that the ionic conductivity of these electrolyte materials depend upon processing variables.7,63 Keeping the processing conditions constant, it is important to synthesize ce ria based materials with different trivalent dopants. This will provide a c onsistent set of ionic conducti vity data that can be used to obtain elastic strain conductivity rela tionships at higher temperatures. As discussed by Inaba et al.9 in the review of ceria based electrolytes, trivalent acceptor cations as dopants exhibit higher i onic conductivity than divalent cations in ceria, therefore, in the present work, trivalent cations were used as dopants for host ceria. Polycrystalline ceramic materials of D0.10Ce0.90O2(where D3+ = Lu3+, Yb3+, Er3+, Y3+, Dy3+, Gd3+, Sm3+, Nd3+) were synthesized under similar experimental conditions described in Chapter 3. In addition to this, different compositions of Smx /2Ndx /2Ce1xO2
104 ( x = 0.01, 0.03, 0.05, 0.08, 0.010, 0.12, 0.15, 0.20) were also processed using same experimental procedures. Ionic conductivity measurements were performed using two point probe ac impedance spectroscopy, from 250oC to 700oC in air. This was followed by high temperature X-ray diffraction (XRD ) of the crushed sint ered samples of doped ceria materials from room temperature to 600oC. Curved position sensitive diffractometer CPS120 (with modified Debye Scherrer geomet ry) was used to perform high temperature XRD. A monochromator crystal was used to separate out Cu K 1 from the incident X-ray beam. The ground powder s were placed on top of an alumina tube, inside a water-cooled furnace. Furnace tem perature was controlled using a commercial software with 2oC tolerance. Samples were first heated to 600oC, and then measurements were performed at different temperatures (in air), while the furnace temperature is lowered. A ramp rate of 8oC/minute was maintained for the furnace to heat up to 600oC and then cool down to room temperatur e. Before the start of the data acquisition at any temperatur e, around 10 minutes of wait ing time was given for the material to come into equilibrium with the temperature. Data acquisition at each temperature was performed for 1 h. Peak pos itions in the XRD pattern were determined by fitting each individual peak with a symmetric Pearson VII profile to model Cu K1 using a commercially available software (i.e ., Solver add-in within Microsoft Excel spreadsheet package). The bes t estimate of the lattice constant was calculated using the extrapolation method.66 This method corrects for the most common error in lattice parameter determination, that of vertical sample displacement.95 Further information about this method is given in Appendix D.
105 76.8 77.0 77.2 Gd0.10Ce0.90O2while cooling Gd0.10Ce0.90O2while heatingIntensity (Arbitary Units)2 (Degree)(331) peak at 400 oC Figure 5-1. The (331) peak in XRD profiles of Gd0.10Ce0.90O2measured at 400oC in different ramping conditions. Dots are the experimental data, and the line is the symmetric Pearson VII function peak fit. In order to confirm that the XRD patterns obtained while the furnace is cooling down and heating up are same, meas urements were taken for Gd0.10Ce0.90O2ceramic in both situations. Figure 5-1 shows the fitted (331) peak of Gd0.10Ce0.90O2at 400oC in air. The fitted peak profile parameters fo r both the conditions are given below (in Table 5-2). Table 5-2. Fitted (331) XRD peak profile parameters for Gd0.10Ce0.90O2at 400oC, measured while the furnace is heating up and cooling down. 2 (Degrees) Full width of half max. (Degrees) Heating 76.939 0.077 Cooling 76.935 0.081
106 It can be seen that in both profiles, peak position values are very close to each other. This clearly indicates that there is no evidence of significant thermal hysteresis present in the material. 5.2 Crystal Structure of Trivalent Acceptor Doped Ce ria at High Temperature 5.2.1 High Temperatur e X-Ray Diffraction High temperature XRD patterns were ta ken for different doped ceria systems at temperatures ranging from 25oC to 600oC, in air. Figure 5-2 and Figure 5-3 show the XRD profiles of Sm0.10Ce0.90O2and Lu0.10Ce0.90O2, respectively, measured at different temperatures. All the patterns look similar to that of pure ceria shown in Chapter 2, which clearly indicates that, both Sm0.10Ce0.90O2and Lu0.10Ce0.90O2ceramics, remain phase pure cubic at intermediate temperature range in air. 3033505560 56.857.057.2 Intensity (Arbitary Units)2 (Degrees) 100 oC 200 oC 300 oC 400 oC 500 oC(222) (311) (220) (200)Intensity (Arbitary Units)2 (Degrees)(111)Sm0.10Ce0.90O2(311) Peak Figure 5-2. XRD prof iles of measured Sm0.10Ce0.90O2at different temperature in air.
107 3033505560 56.056.456.8 25 oC 400 oC 650 oCIntensity (Arbitary Units)2 (Degrees)(222) (311) (220) (200)Intensity (Arbitary Units)2 (Degrees)(111)Lu0.10Ce0.90O2-500 oC (311) Peak Figure 5-3. XRD prof iles of measured Lu0.10Ce0.90O2at different temperature in air. In the insets of Figure 5-2 and Figure 5-3, the (311) peak of the XRD pat terns measured at different temperatures for Sm0.10Ce0.90O2and Lu0.10Ce0.90O2, respectively, are also shown. With the in crease in temperature, the (311) peak shifts toward lower 2 angle which indicates the thermal expansion of the lattice of doped ceria system. The XRD profile s measured for other dope d ceria systems at higher temperatures are presented in Appendix C. All the doped ceria materials are phase pure with no extra peaks of other phase present in XRD profiles.
108 5.2.2 Lattice Strain 22.214.171.124 Thermal strain The solid oxide electrolyte is an ess ential component of SOFC device which operates in the tem perature range between 400oC to 800oC. Together with electrolyte, an SOFC consists of an anode, a cathode, a nd interconnector. Thus, the device can be regarded as a typical example of composite mate rials. For such composite materials to avoid thermal stresses at high temperatures, it is required that a ll the components of the materials possess nearly the same ther mal expansion coefficient over a wide temperature range.96,97 For example, it is recogn ized that the highest accepted mismatch in thermal expans ion coefficient between cathode and electrolyte materials should not exceed 30%.98 Higher difference in thermal expansion results in the mechanical fracture of the components duri ng the heating and cooling cycles of the device. Hence, thermal properties of SOFC component s are essential for this application. Since the present work is devot ed on the study of solid oxide electrolytes, therefore in this section, t hermal expansion coefficient of doped ceria electrolytes will be discussed. The linear thermal ex pansion coefficient ( ) of a material can be defined as the fractional change in length with the change in temperature at constant pressure, or P oT l l 1 where lo is the original length.24 It is well known that the thermal expansion is related to the shape of the potential energy inter-atomic distance between atoms in the crystal lattice. The larger the asymmetry in the shape of the po tential energy curve, the higher the thermal expansion coefficient of the material. Further, asymmetry of the energy well depends upon the bond strength bet ween the two atoms. Weak bond
109 strength materials will show higher asym metry, and consequent ly higher thermal expansion coefficient. For example, thermal expansion coeffici ent of solid Ar is in the range of 10-3 /oC, while for most of ceramics a nd metals, it is in range of 10-5 /oC.24 100200300400500600 5.40 5.41 5.42 5.43 5.44 5.45 5.46 5.47 300400500600700800900 epeatue( ) Nd0.10Ce0.90O2Sm0.05Nd0.05Ce0.90O2Sm0.10Ce0.90O2Gd0.10Ce0.90O2CeO2 Dy0.10Ce0.90O2Er0.10Ce0.90O2Yb0.10Ce0.90O2Lu0.10Ce0.90O2-Lattice Parameter ()Temperature (oC) Figure 5-4. Therma l expansion of D0.10Ce0.90O2ceramic systems (where D3+ = Lu3+, Yb3+, Er3+, Dy3+, Gd3+, Sm3+, Sm3+/Nd3+, Nd3+) in air. Polynomial equation was used to fit the thermal expansion of doped ceria systems. Since the thermal expansion coefficient nor mally increases with the increase in temperature, it is usually wri tten in terms of temperature as, 2 1T c Tbbo (5-4)
110 where bo, b1, and c are constants independ ent of temperature.96 Figure 5-4 shows the lattice expansion of the of D0.10Ce0.90O2(where D3+ = Lu3+, Yb3+, Er3+, Y3+, Dy3+, Gd3+, Sm3+, Sm3+/Nd3+, Nd3+) ceramic systems with the increase in temperature. The curves shown in the Figure 5-4 are plotted as a function of temperature (with Kelvin as a unit) and are fitted using a least square algor ithm to the following polynomial, d T c Tb Tbaaoo22 1 (5-5) where d is a constant, a and ao are the lattice parameter of the material at any given temperature and at room temperature, respectively. Table 5-3. Thermal expansion coefficient () constants for different doped ceria systems. Thermal Expansion Coefficient () Constants Doped Ceria Electrolyte Materials (with different trivalent dopants) bo (1/K) 10-10 B1 (1/K2) 10-9 c (K) 10-1 Yb0.10Ce0.90O21.48 8.70 7.58 Dy0.10Ce0.90O21.48 8.76 8.61 Gd0.10Ce0.90O21.48 9.69 6.24 Sm0.10Ce0.90O21.48 10.21 5.55 Sm0.05 Nd0.05Ce0.90O21.48 10.57 4.46 Nd0.10Ce0.90O21.47 11.21 3.63 CeO2 1.48 8.85 7.01 Table 5-3 shows the ther mal expansion coeffici ent constants (i.e., bo, b1, and c) of doped ceria systems, obtain ed through fitting. For Lu0.10Ce0.90O2, and Er0.10Ce0.90O2ceramic systems, the curve cannot be fitt ed as there was insufficient number of experimentally measured data points. In Figure 5-4, the line co nnecting data points for these materials is plotted for visual aid.
111 0.991.021.051.081.11 8.5 9.0 9.5 10.0 Yb3+Dy3+Sm3+/Nd3+Nd3+Sm3+Dopant Ionic Radius ()Gd3+ [1/K] (0-6)600 oC Figure 5-5. Thermal expansion coefficient (at 600oC) as a function of dopant ionic radius for D0.10Ce0.90O2systems. Line connecting da ta is for visual aid. According to Hayashi et al.99, the thermal expansion coefficient is dependent on the total binding energy of the material. This binding energy can be influenced by changing the bond distances be tween anions and cations, and also by creating oxygen vacancies inside the material. It becomes sma ller with the increase in the ionic distance between anions and cations. Further, with the formation of oxygen vacancies the binding energy in the crystal decreases.99 Keeping the oxygen vacancy concentration (or dopant content i.e., 10 mol%) fixed, the inco rporation of dopant cations with different ionic radii results in the change in Ce-O and D-O distances (Dopant cation = D3+). This is critical as it determines whether the total binding energy in the system increase or decrease. As Hayashi et al.97,99 suggested, the lower the total binding energy in the material, the higher the ther mal expansion coefficient. Figure 5-5 shows the effect on (at 600oC) with the change in dopant ionic radius. It can be seen that with the increase
112 in the ionic radius of the dopant, the thermal expansion coefficient for doped ceria systems at this temperature increases. Th is suggests that the total binding energy involved in the system decreases with the in crease in ionic radii of the dopant. Thus, Nd3+ dopant cation shows lower binding energy than Yb3+ cations. 0100200300400500600 5.40 5.41 5.42 5.43 5.44 5.45 5.46 5.47 5.48 5.49 300400500600700800900 Sm0.04Nd0.04Ce0.92O2Sm0.025Nd0.025Ce0.95O2Sm0.015Nd0.015Ce0.97O2Sm0.005Nd0.005Ce0.99O2Sm0.10Nd0.10Ce0.80O2Sm0.075Nd0.075Ce0.85O2Sm0.060Nd0.060Ce0.82O2Sm0.05Nd0.05Ce0.90O2CeO2Temperature (K)Lattice Parameter ()Temperature (oC) Figure 5-6. Therma l expansion of Smx /2Ndx /2Ce1xO2ceramic systems (where x = 0.01. 0.03, 0.05, 0.08, 0.10, 0.12, 0.15, and 0.20) in air. Polynomial equation (given in text) was used to fit the t hermal expansion of doped ceria systems. The thermal lattice expansion for the different compositions of Smx /2Ndx /2Ce1xO2was also measured using high temperature XRD. Figure 5-6 shows the thermal expansion curve for Smx /2Ndx /2Ce1xO2system. As expected, all the curves exhibit
113 non-linear behavior. They were fitted using equation (5-5). Table 5-4 shows the thermal expansion coefficient constants (i.e., bo, b1, and c) of doped ceria systems, used in equation (5-4). Table 5-4. Thermal expansion coefficient () constants for Smx /2Nd x /2Ce1xO2. Thermal Expansion Coefficient () Constants Smx /2Nd x /2Ce1xO2bo (1/K) 10-10 b1 (1/K2) 10-9 c (K) 10-1 Sm0.005 Nd0.005Ce0.99O21.478 9.275 5.778 Sm0.015 Nd0.015Ce0.97O21.478 9.731 5.469 Sm0.040 Nd0.040Ce0.92O21.475 10.019 5.700 Sm0.05 Nd0.05Ce0.90O21.480 10.210 5.550 Sm0.060 Nd0.060Ce0.88O21.474 10.312 5.451 Sm0.075 Nd0.075Ce0.85O21.473 10.597 5.520 Sm0.010 Nd0.010Ce0.80O21.471 11.314 4.085 Figure 5-7 shows the effect on (at 600oC) with the change in dopant content in ceria. It can be seen that with the incr ease in the dopant conc entration, the thermal expansion coefficient for doped ceria systems (at this temperature) increases. Zhou et al.100 and Hayashi et al.97 observed similar results for YxCe1xO2and GdxCe1xO2, respectively. According to Hayashi et al., this trend can be explained by the formation of oxygen vacancy, which reduc es the binding energy of t he system. For example, in GdxCe1xO2system, the Ce-O distance decreases with the increase of Gd content, while Gd-O bond length remains almost unchanged with increase in x.99 However, the increase in the binding energy due to the decr ease in Ce-O distances is lower than the decrease in binding energy durin g the creation of oxygen vacanc ies. This results in the lowering of binding energy which explains t he higher thermal expans ion coefficient with the increase in dopant concentration.
114 05101520 9.0 9.5 10.0 10.5 Mole % Do p ant Concentration [1/K] (-6)600 oCSmx /2Ndx /2Ce1xO2Figure 5-7. Thermal ex pansion coefficient () at 600oC for Smx /2Nd x /2Ce1xO2as a function of dopant concentration. Line connecting data is for visual aid. 126.96.36.199 Chemical strain The formation of point defects has an influence on the lattice dimension of the host crystal structure.101 These point defects have thei r own specific volume generally different from that of their host lattice, and are non-uniformly distributed in the host crystal structure.102 This results in the mechanical strain in the lattice which is also called chemical strain. For example, addition of trivalent dopant cation (D3+) results in the formation of / CeD and OV point defects in the host ceria lattice. Both these defects independently affect the lattice dimension of host ceria crystal structure. For / CeDpoint defect, the ionic radius mismatch bet ween the host and dopant cations (rh-rd) governs the nature of chemical elastic strain in the lattice. In the case rh < rd, the addition of / CeD point defect, results in the expansi on of host lattice. However, if rh > rd, the lattice will
115 contract.21 To maintain charge neutrality in the system, OV point defect is also created (in addition to / CeD). Hong et al.74 estimated the effective oxygen vacancy radius for ceria to be 1.164 at room temp erature which is significantly smaller than the radius of the oxide ion of 1.38 .48,74 Thus, it can be said inco rporation of oxygen vacancies result in the contraction of the host lattice For example, at r oom temperature, the lattice constant of Lu0.10Ce0.90O2(5.3987 ) is lower than t hat of pure ceria (5.41134 ), even though rh-rd value is almost negligible (r3+ Lu,VIII = 0.977 and r4+ Ce,VIII = 0.97 ).7,47,48 The effect of both the point defects on t he host lattice will decide the nature of elastic strain present in the host lattice. The extent (or magnitude) of the elastic strain is a function of the concentration of these point defects. In the present section, the chemical strain (or lattice strain) by incor porating trivalent dopant cations (of different physical properties) inside the ceria latti ce, from room temperature to 600oC will be discussed. Also, the effect of dopant concentration (x) on the lattice parameter of Smx /2Ndx /2Ce1xO2ceramic at higher temper ature is investigated. As presented in the previous section, according to Kims22 proposed empirical relationship for doped ceria, at room temper ature, lattice paramet er change in doped ceria system is linearly depende nt on the ionic radius mism atch between the host and dopant cations. Similar type of linear relati onship was also given by Hong et al.74 in their work on rare earth element doped ceria electrolytes. However, whether this linear dependency also exists at higher temperatur es, where these materials are normally operated is still unknown.
116 In order to investigate the effect of dopant ionic radius on the lattice parameter at higher temperatures, different doped ceria systems (D0.10Ce0.90O2) were synthesized under the same experimental conditions, as described in Chapt er 3. The dopant content (and thus expected oxygen vacancy concentration) in these systems is kept constant to study the separat e effect of ionic radius mi smatch between host and dopant cations. The high temperature lattice param eters were estimated using XRD described earlier in this chapter. 0.981.001.021.041.061.081.101.12 5.40 5.42 5.44 5.46 Nd Sm/Nd Sm Gd Dy Y Er Yb 500 oC 400 oC 300 oC 200 oC 100 oC Lattice Parameter ()Trivalent Dopant Ionic Radius () 25 oCLu Figure 5-8. Lattice expansion of D0.10Ce0.90O2systems as a function of dopant ionic radius. The linear least-square al gorithm was used to fit the data. Figure 5-8 shows the isothermal curves for the lattice expansion of doped ceria as a functi on of dopant ionic radius. It c an be seen even at higher temperatures, lattice
117 parameter shows a linear depen dency on the dopant ionic radi us. All the curves were fitted using a linear least-square fit. Table 5-5. Relationship between lattice parameter (a) and dopant ionic radius (rd) for D0.10Ce0.90O2at different temperatures. Temperature (oC) Relationship between la ttice parameter (a) and dopant ionic radius (rd). The units of a and rd are in 25 1946.5r2088.0ad 100 2033.5r2049.0ad 200 2049.52096.0 dr a 300 2090.5r2122.0ad 400 2096.5 2189.0 dr a 500 2059.5r2302.0ad Table 5-5 shows the lattice parameter and dopant ionic radius relationships obtained by linearly fitting each isothermal curve in Figure 5-8. The empirical relationship proposed by Kim22 (taking 10 mole % dopant concentration and trivalent dopant cations) for room temperature is given below. 185.5 2200.0 dr a (5-6) where a is the lattice parameter of the D0.10Ce0.90O2and rd is the ionic radius of the trivalent dopant cation. The units of a and rd are in It is important to note that Kims lattice constant of pure ceria is 0.00166 higher than the standar d value of 5.41134 .47 Kim verified the relationship against considerable amount of data, and the typical uncertainty in the parameters was observ ed to be 0.003 Although in the present work, the relationship between lattice paramet er and ionic radius of dopant follows a
118 linear behavior, the obtai ned linear fit for D0.10Ce0.90O2(at room temperature) does not match with the relation ship proposed by Kim. The effect of dopant concentration on the lattice parameter of doped ceria systems is already discussed in Chapter 4. According to Mogensen et al.73, doped ceria solid solutions should follow Vegards law. Howe ver, experimentally it has been observed for various singly doped ceria systems that latt ice parameter exhibits quadratic behavior with dopant concentration.80,85 The second degree term in the quadratic expansion or contraction (depending upon the ionic radius mismatch between dopant and host cations) has been interpreted as due to intera ctions between the point defects. In the dilute regime, the host lattice parameter c hange is not affected by these interactions as the point defects (in this case / CeD and OV) are usually far away from each other. However, with the increase in dopant content, the probability of OV sitting next to / CeD increases. This results in the formation of stable local defect structures (such as / /DVDO) which tend to contract the host ce ria lattice, and, thus, precludes Vegards law behavior. However, in Smx /2Ndx /2Ce1xO2system, it has been shown that at room temperature, the lattice expands linearly as a function of dopant concentration.67 The second degree in t he lattice expansion of Smx /2Ndx /2Ce1xO2system is negligible which indica tes that the local defect st ructures in this system are not as pronounced as in other systems. In the present work, lattice parameter relationships as a function of dopant content in Smx /2Ndx /2Ce1xO2system at higher temperatures were determined. Differ ent polycrystalline sintered samples of Smx /2Ndx /2Ce1xO2(where x = 0.01, 0.03, 0.05, 0.08, 0. 10, 0.12, 0.15, and 0.20) were synthesized using the conventional solid oxide route (described in C hapter 3). Sintered
119 pellets were crushed into fine particles usin g mortar and pestle. The XRD profiles were collected at high temperatur es as described earlier. 05101520 5.41 5.42 5.43 5.44 5.45 5.46 5.47 5.48 Smx /2Ndx /2Ce1xO2600 oC 500 oC 400 oC 300 oC 200 oC 25 oC Lattice Parameter ()x Dopant Concentration (mol %) 100 oC Figure 5-9. Lattice expansion in Smx /2Ndx /2Ce1xO2system as a func tion of dopant concentration at different temperatures. Figure 5-9 shows the la ttice expansion of Smx /2Ndx /2Ce1xO2as a function of dopant concentration at different temperat ures. It can be se en that in all the temperatures, Smx /2Ndx /2Ce1xO2system follows linear behavio r i.e., Vegards law. All the curves are fitted using a linear least-squares fit. Table 5-6 shows the fitted linear rel ationship for Smx /2Ndx /2Ce1xO2system at different temperatures.
120 Table 5-6. Relationship between lattice parameter (a) and dopant content (x) for Smx /2Ndx /2Ce1xO2system at higher temperatures. Temperature (oC) Relationship between la ttice parameter (a) and dopant content (x). The unit of a is 25 )21(40973.5)2(00138.0 x a 100 )24(41398.5)2(00140.0 x a 200 )22(42050.5)2(00140.0 x a 300 )27(42694.5)3(00145.0 x a 400 )22(43444.5)2(00150.0 x a 500 )18(44205.5)2(00156.0 x a 600 )21(44992.5)2(00170.0 x a 300400500600700800 1.4x10-31.4x10-31.5x10-31.5x10-31.5x10-31.6x10-3 Vegard's SlopeTemperature (K) Figure 5-10. Vegards slope as a function of temperature for Smx /2Ndx /2Ce1xO2system. The experimental data is fitted using an exponential curve. The slope of the fitted line is termed as Vegards slope. Figure 5-10 shows the variation of Vegards slope (SV) with temperature. It can be seen that the SV in the
121 Smx /2Ndx /2Ce1xO2system increases with the increase in temperature. The variation of the estimated SV datasets with temperature was then fitted using different curve equations. The best fit was observed using exponential curve, with R2 value of 0.9947. The relationship between SV and the absolute temperature (T) is given below, T SV 00479.0exp101453.500136.06 (5-7) According to Kim22, SV is the algebraic sum of elastic and electrostatic interactions among the point defects in t he doped fluorite oxide system. The solubility limit of a solute depends on the elastic energy in t he lattice which is introduced when a dopant cation (with different ionic r adius and charge valence than hos t cation) is incorporated into the host lattice. The higher the elastic strain energy, the sm aller the extent of solubility. This elastic strain energy ( E) in a substitutional solid solution is defined as, 21 3 2 o oV V GVE (5-8) where G is a the shear modulus, and Vo and V are the molar volume of a pure metal and its solid solution. As doped ceria exhibi ts the cubic fluorite structure, the molar volume in equation (5-8) can be wri tten in terms of lattice parameter. 2 31 3 2 o o oa aa GVE (5-9) where ao is the lattice constant of the pure ceria and and a is the change in lattice constant after the incorporation of dopant. Since a << ao, therefore, E can be written as, 26 aGVEo (5-10) Now, a can be described in terms of Vegards slope (SV). Therefore,
122 2 26cSGVEVo (5-11) where c is the concentration of solute. As m entioned earlier, the ex tent of solubility depends inversely upon the el astic strain energy ( E ). Using equation (5-11), it can be said that the solubility limit of t he dopant increases with the decrease in SV. 21VS s (5-12) It is well known that the extent of solubility widens with the increase in temperature. Thus, following equation (5-12) it is expected that with the increase in temperature, SV should decrease. However, Vegards slope in Smx /2Ndx /2Ce1xO2system increases with the increase in temperature (as shown in Figure 5-10). This indic ates there is an additional term invo lved in the Vegards slope, which with the increase in temperature bec omes predominant compared to elastic and electrostatic interaction energy. Also, it is important to not e that with the increase in temperature, the extent of solubility depends not only on the elastic energy but also on the entropy term (-T S ). Thus, it is the Gibbs free energy ( G ) which governs the solubility limit rather than the elastic energy ( E ). 5.3 Ionic Conductivity of Trivalent Acceptor Doped Ceria at High Temperatures Due to higher ionic conductivity and good thermodynamic stability, doped ceria materials open up the possibility of SOFCs operating at intermedi ate temperatures. Ionic conductivity in these materials takes place through the oxygen vacancy diffusion mechanism. The concentration of oxygen va cancies which are the charge carriers in these oxide materials is dependent on the am ount of acceptor dopant content present in the host oxide. For the same dopant conc entration, ionic conductivity differs for different type of dopants. Fo r example, the trivalent dopant cation system shows higher
123 ionic conductivity compared to the divalent dopant cation system.9 Also, Gd3+ is widely accepted to show highest ionic conductivity.13 This difference in ionic conductivity occurs due to the interactions which exist among various types of point defects. The highest ionic conductivity is shown in the do pant cation system where these interactions are minimal. This condition leads to oxyg en vacancy with higher mobility without any type of local ordering. Numerous studi es have been done to understand the effect of different dopants on the oxygen vacancy diffusion in doped ceria materials.11,14,56,75,103 However, as mentioned in Chapter 4, a wide range of ionic conductivity data have been reported in the literature.13,63,77 Examination of the exper imental procedure followed in each work indicates a strong correlation between the processing conditions and the conductivity observed. Typically, powder synthesized using co-precipitation technique exhibits higher ionic conductivity than t hat processed through solid state route.63 In order to identify the best dopant cations for ceria, and to establish the relationships between dopant physical properties and ionic conductivity, it is essential to develop a consistent set of ionic conductivity dat a for all the doped ceria systems synthesized using the same experimental conditions. This will provide a clear idea about the ideal dopant for ceria, which shows the highest ioni c conductivity, when all the materials are synthesized under similar processing conditions. In the present work, D0.10Ce0.90O2systems (where D3+ = Lu3+, Yb3+, Er3+, Y3+, Dy3+, Gd3+, Sm3+ and Nd3+) were synthesized using solid oxide route as described in Chapter 3. Dopant concentration in these systems is kept fixed, to investigate the effect of ionic radius mismatch between host and dopant cations on the ionic conductivity. In addition, to study the effect of dopant concent ration on the ionic c onductivity, different
124 compositions of Smx /2Ndx /2Ce1xO2were also synthesized und er similar experimental conditions. The ionic conductivity meas urements were taken on the Pt electroded sintered samples using two-point probe elec trochemical impedance spectroscopy from 250oC to 700oC in air. Complex impedance measurements were taken over the frequency range from 10 MHz to 0.1 Hz. It is e ssential to note that in the present work, the grain ionic conductivity wh ich is the fundamental materi al property, rather than the overall ionic conductivity (which includ es microstructural, impurities and other contributions) of doped ceria syst ems is reported. Further the reported grain ionic conductivity for each composition is the average grain ionic conductivity values measured for three different samples of the same composition with varying geometrical aspect ratio. The typical error associated wi th the values of t he reported grain ionic conductivity is < 2%. Figure 5-11 shows the grain ioni c conductivity comparison of D0.10Ce0.90O2materials synthesized using similar exper imental procedures, measured at different temperatures in air. The grai n ionic conductivit y values of Sm0.05Nd0.05Ce0.90O2are also shown. The weighed average ionic radius of the dopant cations (i.e., Sm3+ and Nd3+) was taken as an effective ionic radius of Sm3+/Nd3+ (i.e., 1.094 ) while plotting this graph. From the Figure 5-11, it can be seen that with the increase in dopant ionic radius, the ionic conductivity increases. Even though the dopant ionic radius of Lu3+ is very close to that of Ce4+, the grain ionic of Lu0.10Ce0.90O2exhibits the lowest ionic conductivity in all the temperatures. This may be due to the electrostatic interactions between dopant cations and oxygen vacancies wh ich restrain oxygen vacancy diffusion inside the crystal lattice. At higher i onic radius of the dopant cation, the ionic
125 conductivity data of D0.10Ce0.90O2seems to reach saturation (with increasing dopant ionic radius). However, between 500oC and 600oC, Nd0.10Ce0.90O2shows the highest ionic conductivity among other trivalent acceptor doped ceria systems with similar dopant concentration. It is also important to note that the ionic conductivity of co-doped ceria (i.e., Sm0.05Nd0.05Ce0.90O2) lies between that of Sm0.10Ce0.90O2and Nd0.10Ce0.90O2for all the temperatures. 0.981.001.021.041.061.081.10 10-310-2 400oC 450oC 500oC 550oC Sm/NdNd Sm Gd Dy Y Er Yb Lugrain(S.cm-1)Radius of trivalent dopant cation () 600oC Figure 5-11. Grain ionic conductivity compar ison of different doped ceria materials (with 10 mol% dopant content). All the materials were synthesized using solid oxide route.
126 A recent study by Kamiya et al. on Nd-doped ceria system has shown that Nd0.23Ce0.77O2exhibits higher oxygen diffusion coefficient than Y0.20Ce0.80O2.104 Also, the obtained oxygen diffusion coefficient value for Nd0.23Ce0.77O2is close to that of Gd0.31Ce0.69O2and Y0.40Ce0.60O2. It is well known that the ionic conductivity is directly proportional to the oxygen diffusion coeffi cient, which depends u pon the concentration of mobile charge carriers. Based on the oxygen diffusion coefficient of Nd0.23Ce0.77O2and Gd0.31Ce0.69O2, it can be said that there is less association of point defects in Nddoped ceria system.104 Further, the activation energy for oxygen ion diffu sion in these cubic fluorite structure oxides is inversely proportional to the lattice spacing in the unit cell. As doping with Nd3+ leads to a higher lattice parameter compared to Gd3+ and Y3+ in doped ceria, the activation energy in Nd-doped CeO2 is lower than that of Y-doped CeO2 and Gd-doped CeO2.9 These results indica te that doping with Nd3+ is much more effective compared to Gd3+ and Y3+. The ionic conductivity as a func tion of dopant concentration in Smx/2Ndx/2Ce1-xO2system is already discussed in Chapter 4. Figure 5-12 shows the 3-dimens ional graph between the grain ionic conductivity, dopant c oncentration and temperature. For all the temperatures, the ionic conductivity initia lly increases with the increase in dopant content. It reaches maximum at certain dopant concentration, and then degrades. The decrease in the ionic conductivity at higher dopant concentration can be attributed to the increase in the concentration of local defect structures. These stable defect structures hinder the flow of oxygen vacancies, and thus lower the mobile oxygen vacancy concentration.
127 250 325 400 475 550 625 700 2 4 6 8 10 12 14 16 18 0 0.01 0.02 0.03 0.04 0.05 0.06 T e m p e r a t u r e (oC ) Dopant Concentration (mol %) Grain Ionic Conductivity (S.cm-1) 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05 0.055 Figure 5-12. Grain ionic conductivity as a function of dopant concentration and temperature. Figure 5-13 shows the 3-dimensional gr aph between the normalized grain ionic conductivity, dopant concentration, and temperat ure. The ionic conductivity of all the compositions at a given temperature is normalized by the maximum ionic conductivity obtained at that temper ature. The graph shows that with the increase in temperature, the ionic conductivity maximum shifts to ward higher dopant concentration. This behavior indicates the dissociation of complex defect structures (present mainly at higher dopant concentration) at higher temperatures to give mobile oxygen vacancies. Since the ionic conductivity depends main ly on the mobile oxygen vacancy concentration, with increase in temperature, ionic conducti vity maximum shift toward higher dopant concentrations.
128 250 325 400 475 550 625 700 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 0 0.2 0.4 0.6 0.8 1Te m pe r a t ur e ( oC ) Dopant Concentration (mol %) Normalized Grain Ionic Conductivity 0.1 0. 2 0. 3 0. 4 0. 5 0. 6 0. 7 0. 8 0. 9 1 Figure 5-13. Normalized grain ionic conductivity as a function of dopant concentration and temperature. 5.3.1 Activation Energy As discussed in Chapter 4, the activa tion energy for oxygen vacancy diffusion inside the crystal lattice, is the sum of migration enthalpy ( HMigration) of oxygen vacancy and association enthalpy ( HAssociation) of local defect structures. The HMigration is defined as the amount of energy required for the oxygen vacancy to jump from one oxygen ion site to another oxygen ion site, assuming there are no interactions offered from the point defects residing next to oxy gen vacancy site. At lower temperatures, point defects (oxygen vacancie s and dopant cations) usually interact with each other to form stable local defect structures. Thus, an additional amount of energy is required to dissociate oxygen vacancies from these local defect structures at these temperatures. This additional energy is termed as association enthalpy or HAssociation. In this
129 temperature range, activati on energy is the sum of HMigration and HAssociation. According to Steele13, above 400oC, most of the oxygen vacancies overcome the dopant cations interaction barrier, and are free to diffuse inside the crystal lattice. Thus, activation energy for oxygen vacancy di ffusion above this temperature can be considered to be equal to HMigration. However, experimentally there is no such set temperature at which this transition takes place. Typically, the Arrhenius conductivity plots of these oxide material s exhibit a gradual, rather t han a sharp, change in slope with temperature. Therefore, using Arrhenius conductivity plo t, it is difficult to separate HAssociation from the total activation energy. In the present work, activation energy is determined from the Arrhenius plot of the ionic conducti vity data measured from 250oC to 700oC for D0.10Ce0.90O2systems. It was assumed that the change in slope is negligible in the Arrhenius plots. Figure 5-14 shows the activation energy f or oxygen vacancy di ffusion in different doped ceria systems with 10 mo l% dopant concentration. For comparison, activation energy in co-doped ce ria systems of Gd0.056Y0.044Ce0.90O2and Sm0.05Nd0.05Ce0.90O2is also shown. The weighed average ionic radius of the dopant cations (i.e., Gd3+ and Y3+, and Sm3+ and Nd3+ ) was taken as an effective ionic radius of Gd3+/Y3+ (i.e., 1.038 ) and Sm3+/Nd3+ (i.e., 1.094 ), respectively, while plotting this graph. The error bar associated with the activation energy value in Figure 5-14 was determined by measuring activation energy values in th ree different test samples of similar composition. Initially, with the increase in dopant ionic radius, activation energy decreases. It reaches minimum at Sm0.10Ce0.90O2(0.662 0.002 eV), and then increases. Nd0.10Ce0.90O2(0.672 0.004 eV) exhibits higher activation energy than
130 Sm0.10Ce0.90O2system but lower activation energy than Sm0.05Nd0.05Ce0.90O2(0.676 0.004 eV) and Gd0.10Ce0.90O2(0.678 0.003 eV) systems. Interestingly, for co-doped ceria i.e., Gd0.056Y0.044Ce0.90O2(0.719 0.003 eV), activation energy is in between that of Y0.10Ce0.90O2(0.764 0.002 eV) and Gd0.10Ce0.90O2. As mentioned earlier, in order to design higher ionically conductive materials, activation energy for oxygen vacancy diffusion needs to be minimized. From Figure 5-14, it is clear that t he trivalent dopant with ionic radius within the range of 1.05 to 1.11 exhibits lower activation energy, and consequently, higher ionic conductivity. 0.960.991.021.051.081.11 0.65 0.70 0.75 0.80 0.85 0.90 Yb3+Sm3+/Nd3+Nd3+Y3+Sm3+Gd3+Gd3+/Y3+Dy3+Er3+Activation Energy (eV)Dopant Ionic Radius ()Lu3+ Figure 5-14. Activation energy for oxygen vacancy diffusion det ermined from the Arrhenius plot of the ionic c onductivity data m easured from 250oC to 700oC in doped ceria systems with 10 mol% dopant.
131 0.960.991.021.051.081.11 0.65 0.70 0.75 0.80 0.85 0.90 0.95 Yb Sm/Nd Nd Y Sm Gd Dy Er Activation Energy (eV)Dopant Ionic Radius ()Lu Measured above 475 oC HMigration+HAssociation Measured below 475 oC HMigration Figure 5-15. Migration and Association enthalpies for oxygen vacancy diffusion determined from the Arrhenius plot of doped ceria systems. While describing the activation energy for oxygen diffusion in Smx /2Ndx /2Ce1xO2system in Chapter 4, the MeyerNeldel rule was discussed. According to this rule, the Arrhenius plots for the grain ionic conductivity of all the compositions (> 5 mol % dopant concentration) converges to a comm on transition temperature point ( To). Around this temperature, conductivity becomes independent of dopant concentration. According to Nowick et al., To can be seen as a transition te mperature between the stage where most of the carriers are bound at various traps and the stage where all the carriers are free.87 This suggests that above To, almost all the oxygen vac ancies are free to migrate for any dopant concentration. Since the Arr henius plot for the grain ionic conductivity exhibits a gradual decrease in the slope at higher intermediate temperatures, it is
132 difficult to confirm this hypothesis. However, in order to distinguish between migration and association enthalpies, it was assumed that the associat ion dissociation transition occurs at To. For Smx /2Ndx /2Ce1xO2system, To is found to be around 475oC. In the present study, it was further assumed that the transition temperature for all the doped ceria systems is around 475oC. Using a least-squares algorithm, the high and low temperature region of Arrhenius plot was linearly fitted. T he fit of the data was generally quite good, with correlation coefficients for linear least-squares fit between 0.9997 and 0.9999. Activation energy values for the ionic conduction in t hese materials were calculated from the high and low temperature gradients of the fitted Arrhenius plots. Figure 5-15 shows the activation energy for the grain ionic conductivity of D0.10Ce0.90O2systems, measured below 475oC and above 475oC, as a function of dopant cation ionic radius. The activation energy measured in the range of 475 oC 250oC is a sum of Hassociation and Hmigration, while the activation energy measured above 475oC is equal to Hmigration. It can be seen that both the curves show similar behavior. The migration entha lpy decreases with the incr ease in dopant ionic radius and reaches a minimum at Sm0.10Ce0.90O2(0.645 0.009 eV), and then increases. According to Nowick et al., Hmigration should be independent of dopant cations physical properties and most importantly dopant concentration. Using this argument, Nowick et al. measured the ionic conductivity of highly pure CeO2, and separately estimated the migration enthalpy to be 0.67 eV.105 Wang et al.103 studied the dielectric and anelastic loss originating in the OCeVY/defect pair. The activation energy for the relaxation process was estimated to be 0.64 eV. Similarly, Kilner et al. a nd Hohnke reported the migration enthalpy to be 0.6 eV and 0.61 eV, respectively, for doped ceria systems.88,106
133 The literature results are not in agreement with the migrat ion enthalpy data presented in this work. The data presented in Figure 5-15 indicates that there might be some additional energy term involv ed in the total activation energy. The activation energy for oxygen diffusion as a function of dopant concentration in Smx /2Ndx /2Ce1xO2system is already discussed in Chapter 4. The migration enthalpy in Smx /2Ndx /2Ce1xO2system exhibits dependency on d opant content which is again not in agreement with the abov e mentioned literature. 5.3.2 Pre-Exponential Coefficient The pre-exponential term in t he Arrhenius relationship for the ionic conductivity of these oxide materials is described in C hapter 2. The pre-exponential factor ( o) is a complex function of different va riables, and can be written as, k S aNVqm ooOVexp ][2 2 0 (5-13) where qv is the charge of an oxygen vacancy, ] [ OV is the fraction of mobile oxygen vacancies, No is the number of oxygen ion sites per unit volume, a is the ion jump distance, o is an appropriate lattice vibration, and Sm is the entropy change during oxygen ion diffusion. In order to develop higher ionically c onductive materials, it is essential to select dopant cations which show a high pre-exponential coefficient. In the present work, the pre-exponentia l coefficient in the Arrhenius ionic conductivity plot in doped ceria systems is compared. Figure 5-16 shows the preexponential factor of D0.10Ce0.90O2systems. It is important to not e that the report ed pre-exponential coefficient is determined from the intercept of the Arrhenius plot, assuming there is no change in the slope at higher te mperatures, due to the release of oxygen vacancies
134 from local defect structures. It can be seen that the preexponential coefficient shows a similar trend when compared with activati on energy in doped ceria systems shown in Figure 5-14. It decreases with the increase in dopant ionic radius. Sm0.10Ce0.90O2exhibits the lowest while Lu0.10Ce0.90O2shows the highest pre-exponential coefficient value. 0.960.991.021.051.081.11 5.0 5.2 5.4 5.6 5.8 Gd3+/Y3+Sm3+/Nd3+Yb3+Log(Pre-exponential Coefficient)Nd3+Sm3+Gd3+Dy3+Y3+Er3+Dopant Ionic Radius ()Lu3+ Figure 5-16. Pre-exponential c oefficient in the Arrhenius ionic conductivity plot of D0.10Ce0.90O2systems. The higher pre-exponent ial coefficient of Lu0.10Ce0.90O2may be attributed to the higher lattice vibration (o ) in this system. As shown in Figure 5-8, among all the doped ceria systems, Lu0.10Ce0.90O2shows the lowest lattice par ameter. It is expected that ions in this system, are com paratively closer to each other, and exhibit stronger bond strength. The two ions (with unlike charges) inside the crystal lattice can be seen as a
135 harmonic oscillator where two blocks (of similar mass) are attached with the spring. The vibration frequency (o ) of the spring is given by, m ko 2 1 (5-14) where m is the mass of the block and k is the spring constant. The k depends upon the stiffness of the spring, which in turn determi nes the force required to stretch the spring. Similarly, ions with stronger bond strength will show higher stiffness, and consequently, higher vibration frequency. After reaching the minimum at Sm3+, the pre-exponential coefficient again increases with the increase in t he ionic radius of the dopant. Nd0.10Ce0.90O2shows a higher pre-exponential coe fficient than that of Sm0.10Ce0.90O2. This may be due to the increase in the latti ce spacing of Sm0.10Ce0.90O2, which results in the increase in the jump distance ( a ) of the oxygen vacancy. From the present work, it is known that the activation energy of Nd0.10Ce0.90O2is higher than that of Sm0.10Ce0.90O2. Although, the ionic conductivity depends upon both the ac tivation energy and the pre-exponential coefficient, it is the latter that is contribut ing more towards ionic conductivity in this dopant ionic radius range. Since, the pre-exponential coefficient value of Nd0.10Ce0.90O2is higher than that of Sm0.10Ce0.90O2, Nd0.10Ce0.90O2exhibits higher ionic conductivity. As discussed in Chapter 4, co-doping in ceria can l ead to the increase in configurational entropy of the system which consequently enhances the preexponential coefficient value. For Sm0.05Nd0.05Ce0.90O2system, the pre-exponential coefficient was higher than that of Nd0.10Ce0.90O2and Sm0.10Ce0.90O2. However,
136 Gd0.056Y0.044Ce0.90O2exhibits a pre-exponential coe fficient value between that of Gd0.10Ce0.90O2and Y0.10Ce0.90O2ceramics. Experimentally, it is difficult to separate out the contribution from different variables on the pre-exponential coefficient value. Th us, a computational a pproach, like density functional theory, may be extremely helpful to investigate the effect of each variable independently. This work will allow better understanding of the pre-expone ntial factor for different type of acceptor dopants shown in Figure 5-16. This w ill allo w selecting next generation co-dopants, wh ich not only show higher pre-exponential coefficient values, but also lower activation ener gies for oxygen vacancy diffusion. 5.4 Revisiting Critical Dopant Ionic Radius In the introduction of this chapter, it wa s mentioned that fluorite structured oxides exhibit maximum ionic conductivity when doped with an trivalent cation that causes very little expansion or contracti on of the lattice. To support this argument, it has been said that doped ZrO2 exhibits lower ionic conductivity than that of doped CeO2, because most of the trivalent dopant ca tions show ionic radius that closely lie to that of Ce4+ than Zr4+. Thus, the ionic radius mismatch bet ween host and dopant cati ons will be smaller in doped ceria systems co mpared to doped ZrO2. This leads to more elastic strain in doped ZrO2 system, and consequently, lower ioni c conductivity. However, among doped ceria systems, Lu3+ cation shows the lowest ionic conductivity even though the ionic radius of Lu3+ is very close to that of Ce4+. The addition of a cceptor dopant cations creates oxygen vacancies (to achieve electr ical neutrality), which also leads to the contraction of the lattice. In the literature, diffe rent approaches were taken to estimate the effect of dopant cations in side the fluorite oxide lattice Using multiple regression analysis, Kim proposed the empirical relationship between the ionic radius of the dopant
137 cation dissolved in ceria (and ot her fluorite structure oxides). A critical ionic radius ( rc) for the dopant cation is defin ed which on substituting fo r host cation does not change the lattice parameter. Based on lattice strain ionic conductivity relationship, cation with ionic radius rc, is the ideal dopant for ceria to ex hibit the highest ionic conductivity. The rc value obtained from Kims22 proposed model for the trivalent dopant cations in ceria was estimated to be 1.038 In the same line, Hong et al.74 derived rc value to be 1.024 using a model in which radius was assigned to oxygen vacancy. In principle, the rc value can be derived from the plot of latti ce parameter versus dopant ionic radius for a number of different oxide solutions of the same concentration. Eguchi et al. published such a plot usi ng eight lanthanide oxides with 20 mol% dopant concentration.10 The rc value obtained from the graph is 1.010 Mogensen et al. reported different rc values derived from the past lit erature, ranging from 1.010 to 1.038 .73 It is widely accepted that Gd3+ dopant cation exhibits the highest ionic conductivity among doped ceria materials.13 According to Kim, Gd3+ shows the highest ionic conductivity because the ionic radius of Gd3+ (1.053 ) lies very close to the rc value (1.038 ).48 In Chapter 4, co-doping strategy was discussed in which co-dopants were added such that the weighed average dopan t ionic radius matches to the rc value. The ionic conductivity obtained from the codoped ceria was lower t han that of Gd-doped ceria with the same total dopant concentration. The rc value reported in all the past work is obtained from the plot of dopant cation ionic radius a nd lattice expansion measured at room temperatur e. The ionic conductivity (which is measured at high temperatures) results indicate t hat there might be an increase in rc value toward Nd3+
138 ionic radius value, at higher temperatures. To test this hypothesis, in the present work, lattice expansion as a functi on of dopant cation ioni c radius is measured at higher temperatures. The rc value is obtained, wh ere the lattice paramet er mismatch between doped ceria and pure ceria becomes zero. Th is work will not only provide the rc value at higher temperatures, but also verify Kilner et al.s hypothesis, that maximum ionic conductivity is achieved in the system when the dopant cation substitute for host cation without any change in lattice. Table 5-7. Critical dopant ionic radius for trivalent acceptor dopant cation for host ceria, for different temperatures. Temperature (oC) Critical dopant ionic radius ( rc) value () 25oC 1.0311 100oC 1.0293 200oC 1.0291 300oC 1.0278 400oC 1.0274 500oC 1.0252 The relationship between t he lattice parameter of D0.10Ce0.90O2as a function of dopant cation ionic radius, at higher temperatures is shown ea rlier in this chapter in Table 5-5. The lattice parameter of doped ceria depends line arly upon dopant cation ionic radius, even at higher temperatures The lattic e parameter of pure CeO2 as a function of temperature was also determined using the same experimental procedure (given in Chapter 3). The lattice parameter of pure ceria was es timated to be 5.4099 at room temperature. The Joint Commi ttee on Powder Diffraction Standards card 340394 for ceria shows the lattice parameter value of 5.41134 .47 The difference in the
139 estimated and standard value (0.00144 ) was adjusted in all the lattice parameter values that were determined using this procedure. Table 5-7 shows the rc value determined at different temperatur es. At room temperature, the rc value was estimated to be 1.0311 which is in between the values reported by Kim et al. (1.038 ) and Hong et al. (1.024 ). With t he increase in temperature, the rc value decreases. 0.981.001.021.041.061.081.101.12 -0.015 -0.010 -0.005 0.000 0.005 0.010 0.015 0.020 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 LuYb Er Dy Gd Sm Sm/Nd NdGrain Ionic Conductivity (S cm-1) Change in Lattice Parameter, a-aCeria ()Dopant Cation Ionic Radius () Lattice Parameter Mismatch between D0.10Ce0.90O2and CeO2 500oC 1.0252 Grain Ionic Conductivity Y Figure 5-17. Lattice param eter mismatch between D0.10Ce0.90O2and pure ceria, as a function of dopant cation ionic radius at 500oC. The grain ionic conductivity of D0.10Ce0.90O2at 500oC is also shown. Figure 5-17 shows the lattice parameter mismatch between D0.10Ce0.90O2( a ) and pure ceria, as a function of dopa nt cation ionic radius, at 500oC. It can be seen that for dopant cations with ionic radius less than 1.025 the mismatch is negative, while dopant cations which show higher ionic radius than 1.025 t he mismatch is positive.
140 At 1.025 there is no mismatch bet ween doped ceria system and pure ceria. Figure 517 also compares th e grain ionic conducti vity of different doped ceria systems as a function of dopant cation ionic radius, at 500oC. Assuming Kilner et al.21s hypothesis is correct, Dy3+ should exhibits the highest ionic conduc tivity, as its ionic radius (1.027 ) is very close to the rc value. However, it can be seen that the grain ionic conductivity continue to increase even for the dopant cations which show ionic radius higher than the rc value. The Nd0.10Ce0.90O2cation exhibits the highest ionic conductivity among all the other doped ceria materials, with latti ce parameter difference with pure ceria ~0.0196 at 500oC. This clearly indicates that the Kilner et al.21s structure-ionic conductivity relationship is not app licable in doped ceria materials. In Chapter 4, DFT work performed by Andersson et al.83 on the total interaction energy for different trivalent dopant cations was shown. Since the oxygen vacancy site preference in Pm3+ to reside in nearest neighbor site or next to neighbor site of dopant cation is almost the same, ther efore it was pr edicted that Pm3+ is the ideal dopant cation for ceria. However, DFT calculations were performed at 0 K. As Nd3+ lies next to Pm3+ in the lanthanide series, theref ore there is a possibility that the cross-over of the total interaction energies shown in Figure 4-6 shifts toward Nd3+ at higher temperatures. Thus, for Nd-doped ceria system, the number of equi-interaction energy sites is comparatively higher compared to other doped ceria materials. Future work is required to further investigate this possibility in detail. 5.5 Summary and Conclusions To design higher ionically conductive dop ed ceria materials, it is imperative to relate the crystal structure with ionic conductivity at higher temperatures. In this work, crystal structure of diffe rent doped ceria systems were characterized at higher
141 temperature using XRD. All t he compositions exhibit cubic fluorite structure similar to that of pure ceria. Both t hermal and chemical elastic stra ins in the different systems of doped ceria were determined by estimati ng the lattice parameter at higher temperatures. The thermal expansion coeffi cient increases with the increase in dopant ionic radius and dopant concentration. The lattice parameter shows linear behavior with dopant ionic radius, from room temperature to 500oC. Also, the Smx /2Ndx /2Ce1xO2system obeys Vegards law, even at high temperatures. As ionic conductivity depends upon the pr ocessing conditions, a wide range of ionic conductivity values is reported for any given doped ceria system. Thus, it was difficult to compare the ionic conductivity data from the literature. In order to understand the ionic conductivity trend among different trivalent acceptor doped ceria materials, processing variables in the synthesis of these materials were kept constant. A consistent set of ionic conductivity data (at higher temperatures ) was prepared for the different doped ceria materials (D0.10Ce0.90O2) synthesized under similar experimental conditions. Dopant concentration in thes e compositions is fixed to understand the separate effect of dopant ionic radius. Among trivalent acceptor dopant cations, Nd3+ cation exhibits the highest ionic conductivity in ceria. To study the effect of dopant content, different compositions of Smx /2Ndx /2Ce1xO2ceramics were also synthesized. Initially, with the increase in the dopant concentration, the io nic conductivity increases. It reaches maximum at certain dopant conc entration, and then degr ades. Among all the tested doped ceria systems, Sm0.075Nd0.075Ce0.85O2exhibits the highest ionic conductivity, at 500oC in air.
142 According to Kilner et al., for a fl uorite structure oxid es maximum ionic conductivity is achieved by alloying it with dopant cations which show minimum elastic strain in the host lattice. Following this hypothesis, Kim introduced the critical dopant ionic radius ( rc) concept for doped ceria materials. The rc was defined as the ionic radius of the dopant cation that causes no elastic strain in the host lattice. Thus, dopant with ionic radius with rc value should exhibit the highest i onic conductivity in ceria. On comparing the ionic conductivity of all the doped ceria materials synthesized under similar experimental conditions, it was observed that Nd3+ exhibits the highest ionic conductivity. However, ionic radius of Nd3+ is far away from the rc value. It is important to note that in the structure ( rc) property (ionic conductivi ty) relationship proposed by Kim22, both the parameters were measured in di fferent conditions (temperatures). In order to verify the rc concept, rc values were determined at te mperatures where the ionic conductivity is measured. The graph between the elastic strain and the dopant cation ionic radius was plotted at higher temperatures, to derive rc values. The rc value decreases with the increase in temperature. At 500oC, rc value was estimated to be 1.0252 for trivalent dopant cations in host ceria. The rc value obtained at 500oC lies close to the ionic radius of Dy3+. However, it is Nd3+ that exhibits the highest ionic conductivity. This clearly indicates that the Kiln er et al.s hypothesis is not applicable for doped ceria materials.
143 CHAPTER 6 PROCESSING EFFECTS ON THE IONIC CONDUCTIVITY 6.1 Introduction In Chapter 4, a comparison between the ionic conductivity data for singly doped ceria systems taken from literature was shown. A wide range of grain ionic conductivity values have been reported for a mate rial with similar composition.7 A careful survey of the experimental procedures from literature, it indicates that processing variables show an effect on the ionic conductivity of the mate rial. Typically, powder synthesized using wet chemical route exhibit higher ionic conduc tivity compared to conventional solid state route.58 The soft chemical route is capable of producing relatively higher purity, homogeneous, and ultrafine powder at low temperatures.58,63,64 For the polycrystalline ceramic electrolyte materials, the presence of even slight concentrat ion of impurities is deleterious to both the grain and the grain boundary conductivities.107,108 Typically in all the ceramic processing techniques, SiO2 is a ubiquitous background impurity.109 It is difficult to avoid it presence even when us ing a relatively pure sample. Zhang et al.107 studied the influence of SiO2 content on the grain and the grain boundary ionic conductivity in GdxCe1xO2system. It was shown that with the increase in SiO2 content, the grain ionic conductivity consiste ntly decreases. Similar results have been reported by Martin et al.110 in SiO2 doped ZrO2 system. With the increase in SiO2 content from 1 wt % to 10 wt %, grain conductivity of 8 mo l% yttrium stabilized zirconia at 400oC deceases by approximately 25 %. According to Zhang et al.107, this decrease in the conductivity can be explained by the dissolution of SiO2 in GdxCe1xO2crystallites. At higher tem peratures, small amount of SiO2 dissolve inside the doped
144 ceria lattice, either at substitutional (shown in equation 6-1) or interstitial sites (shown in equation 6-2). O Ce CeOOSi SiO 222 (6-1) O i CeO OOSi VSiO2 222 (6-2) The Si4+ cation entering the lattice as a substitution for Ce4+ does not create any charge carriers inside the lattice, however due to the difference in the ionic radius between the Si4+ (r4+ Si,VIII= 0.4 ) and Ce4+ (r4+ Ce,VIII= 0.97 ) results in the elastic strain which not only hinders the mobility of oxygen vacancies but also increases the association activation energy of local defect structures (discussed in Chapter 2). According to the Hume-Rothery rules, for the substitutional dissolution, the ionic radius mismatch between the host and dopant cations of less than 15 % is required. The ionic radius mismatch between the Si4+ and Ce4+ case is quite large for Si4+ to substitute for Ce4+. Thus, there is a higher probability that Si4+ cations reside on the interstitial sites. It can be seen from equation (6-2), interstitial point defects results in the consumption of oxygen vacancies which are the charge carrier in these oxide materials. This type of dissolution mechanism also results in the degradation of ionic conductivity. In polycrystalline ceramic samples grain boundaries also have a significant role to the overall ionic conduct ivity. The deleterious effect of impurities on the grain boundary conductivity has been recognized in doped ceria ceramic for over 10 years.62,111 The impurities incorporated during the powder processi ng tend to segregat e near the grain boundaries during sintering and grain growth. Figure 6-1 shows high resolution images of doped ceria systems from o ur work. Figure 6-1 A) shows c ontinuous layer of an segregated phase present in the grain boundary of Gd0.10Ce0.90O2. The layer is
145 around 2 nm in thickness. Figure 6-1 B) shows the gl assy amorphous phas e formed on the three-grain juncti on in the sintered Sm0.05Nd0.05Ce0.90O2ceramic sample. Both the samples were synthesized using solid state reac tion. The partial or complete blocking of charge carriers by these segregated im purities leads to higher grain boundaries resistivity. A B Figure 6-1. A) Continuous layer of an siliceous phase in the grain boundary of Gd0.10Ce0.90O2B) Glassy phase formed at the three-grai n junction in Sm0.05Nd0.05Ce0.90O2. Using transmission electron mi croscopy, Gerhardt et al.112 studied the grain boundaries in doped ceria systems. The presence of Si in the form of a continuous glassy phase is responsible for the large decr ease in the effective io nic conductivity in these materials. Even a few hundred ppm of a siliceous impurity can increase the grain-boundary resistivity of Gd0.10Ce0.90O2by up to 100 times.112,113 According to Steele13, for a highly pure Gd0.10Ce0.90O2sample, the grain boundary impedance cannot be detected above 500oC. However, for an impure Gd0.10Ce0.90O2sample, the
146 total ionic conductivity is mainly dominated by grain boundary conductivity, upto 1000oC. This clearly suggests the need for hi ghly pure materials for the electrolyte application in SOFCs. In addition to the impurities, the processi ng of the final microstructure from the starting powder also affects t he ionic conductivity. Recent work by Eposito et al.114, have shown that the sharp grai ns and the unrelaxed grain boundary interfaces lead to higher grain boundary conductivi ty in doped ceria systems. Such microstructure possesses grain boundar ies that are consid ered to be highly def ective and segregationfree. In order to obtain such grain bounda ry structure, fast firing was performed to promote lattice diffusion of the material (from the grain boundary source) which results in the densification (rather t han the grain growth and coars ening) in the polycrystalline green ceramic. Also, it is well known that the accept or dopant cations in ceria materials have tendency to segregate near the grain boundaries.115 This results in the formation of Schottky barrier for oxygen vacancy diffu sion which significantly degrades the ionic conductivity in polycrystalline ionic materials. Dopant segregation can be attributed to the ionic radius mismatch between the dopant and host cations. In addition, the positive potential of the grain boundary core act as a driving forc e in segregating dopant cations near the grain boundary.116 Also, at high sintering tem peratures, acceptor cations are sufficiently mobile to segregate to the gr ain boundary core, and accumulate in the space-charge layer as a result of elastic st rain and electrostatic interactions with the positively charged grain boundary core. Thus, it is essential to lower the sintering temperature and time of the green ceramic material, such that the sintering process
147 does not provide enough thermal drive for the acceptor dopant cation to segregate near the grain boundary. O ne of the approach, as pointed out by Herring117, to lower the sintering temperature is to reduce the particle size of the starting powder. The densification can be enhanced by using nano-size powder wh ich exhibits higher surface energy to drive mass diffusion transport.118 Other approach to synthesize dense ceria samples at lower temperatures is by usi ng transition metal oxides sintering additives which promote densification.119,120 However, following this methodology may have a detrimental effects on the electric al properties of the material.121,122 In recent years, microwave heating have ga ined lot of interest especially in the fabrication of ceramic materials with controlled microstructure.123 In comparison to the conventional direct heating, microwave heating of ceramics o ffers lot of advantages. It not only promotes the material densification in a consider ably reduced processing time and temperature, but also prov ides uniform heat treatment (v ia volumetric heating) and rapid heating rate which improve the materi al properties and microstructure. It has been reported for alumina that the microstructure evolut ion in the microwave and conventional sintering are not the same, and that the mi crowave sintering produces finergrained alumina.124 Recently, Rambabu et al.125 reported 92 % rela tive theoretical density of the ceramic samples of CeO2 synthesized using microwave sintering at 1200oC for 60 min. In the present work, micr owave sintering was used as a fast firing technique to obtain dense polycrystalline cera mics with sharp grains and unrelaxed grain boundary interfaces. In Chapter 4, a novel co -doped ceria system was desi gned that exhibits higher ionic conductivity than Gd-doped ceria. After identifying the co-dopants for the host
148 ceria, optimization of the dopant concentration was performed. This leads to the development of Sm0.075Nd0.075Ce0.85O2material that exhibits 30 % higher ionic conductivity than Gd0.10Ce0.90O2, at 550oC in air. In the pr evious work, co-doped samples were synthesized using solid state reac tion. As mentioned earlier, this ceramic processing technique may cont ain impurities which degrade t he ionic conductivity of the materials. To captur e the full potential of Sm0.075Nd0.075Ce0.85O2material, powder was synthesized using co-precipitation technique. This will not only im prove the grain ionic conductivity but also promotes transport of charge carriers across the grain boundaries. Further, microwave sintering was used as a fast firing te chnique to obtain unrelaxed grain boundary interfaces that can further enhance the grai n boundary conductivity of the material. 6.2 Experimental Procedures The experimental details for the processing of phase pure powder of Sm0.075Nd0.075Ce0.85O2using co-precipitation technique and the microwave sintering of the green ceramic sample are given in Chapter 3. Figure 6-2 compares the temperature time profile in the conventional sintering, and the mi crowave sintering. The conventional sintering of the Sm0.075Nd0.075Ce0.85O2sample was performed at 1550oC for 10 h while the microwav e sintering was done at 1450oC for 1 h. The area under the curve shown in Figure 6-2 is proportional to t he thermal energy given to the ceramic sample during eac h sintering process. The particle size distribution in Sm0.075Nd0.075Ce0.85O2powder obtained using coprecipitation technique was m ono-disperse with most of t he particles were below 0.2 m in size. The mean particle size was around 0.095 m. Similarly, powder synthesized using conventional solid oxide route exhibits mono-disperse particle size distribution.
149 However, the average particle size obtained through this process was 0.84 m. Microwave sintering was performed on the green ceramic sample using the powder synthesized with co-precipitation technique; while the conventional solid state sintering was performed on the sample with the powder synthesized usin g the solid oxide route. The density of the ceramic sample obtaine d using microwave sintering was around 94 % of theoretical density, while in the case of conventional sintering it was above 98 % of theoretical density. For c onvenience, in the text abbrev iations are used for both the samples (shown in Table 6-1). 30060090012001500 0 200 400 600 800 1000 1200 1400 1600 Microwave Sintering Conventional SinteringTemperature (oC)Temperature ProfileTime (Minutes) Figure 6-2. Comparison of t he temperature time profile in conventional solid state sintering and micr owave sintering.
150 Table 6-1. Abbreviation of samples synthesized using different techniques. Sample Abbreviation Conventional Sintering + Solid Oxide Route SamCoSo Microwave Sintering + Co-precipitation Route SamMiCo The microstructural analysis was per formed using both scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Electron dispersive Xrays spectroscopy (EDXS) was used for chemical compositional analysis in the material. Details for the sample preparatio n for both the characterizations are described in Chapter 3. The ionic conductivity m easurements were performed using two-point probe impedance spectroscopy (di scussed in Chapter 2) from 250oC to 700oC, in air. 6.3 Results and Discussion 6.3.1 Microstructure Figure 6-3 shows the SEM images of SamCoSo and SamMiCo. Due to the large particle size of the starting powder, and the higher sintering tem perature and time, the grain size in SamCoSo is much larger compared to SamCoMi. The grain size in the sample synthesized through solid state reacti on method is in t he range of 4 7 m. The arrows in the microstructure image of the SamCoSo sample show the curv ed nature of the grain boundaries. According to Eposito et al.114, this type of grain boun daries are relaxed and in equilibrium state. T he fast firing in the SamMiCo sample, results in the finer grained microstructure. The grain size using micr owave sintering is in the range of 0.5 1 m. In order to study th e grain boundary in SamMiCo sample, TEM was performed on SamMiCo sample.
151 A BFigure 6-3. SEM images of A) Conventional sintering + So lid oxide route B) Microwave sintering + Co-precipitation technique. A BFigure 6-4. TEM images of the microwave sintered Sm0.075Nd0.075Ce0.85O2sample. Figure 6-4 shows the micr ostructure of the SamMiCo sample. It can be seen that the grains are finer with s harp grain boundaries. Clearly uniform and rapid heating of the ceramic sample does not provide enough time for grain growth. Thus, there is no grain boundary interface relaxation in the SamMiCo sample. Figure 6-5 compares the 5 m 5 m
152 typical grain boundary observed in both the samples. Typicall y, the width of the grain boundary in the pure polycrystalline ceramic sample is in the range 1 2 nm.126 As discussed in the introduction, in order to reduce the grain boundary surface energy most of the siliceous impurities segregate near the grain boundar ies. Therefore, depending upon the segregation of the dopant and impurities, the gr ain boundary thickness can increase to above 2 nm.115 In the sample synthesized using co-precipitation and microwave sintering, it can be seen that t he width of the grain boundary is quite small (<1 nm). This clearly suggests that t he segregation near t he grain boundaries is minimal in the sample, and the impurities present in the mate rial are very low. In contrast to SamMiCo sample, the grain bou ndary width in SamCoSo is quite large (~2 nm). This may be due to the segregation of acce ptor dopant cations or due to the presence of impurities near t he grain boundary. A BFigure 6-5. Grain boundary in the polycr ystalline ceramic synthesized using A) microwave sintering and B) conv entional solid state sintering.
153 EDXS was performed for chemical composit ional analysis in the grain boundary of SamCoSo sample. Figure 6-6 shows the EDX spectrum at the grain boundary for the sample synthesized using sol id oxide route and solid state si ntering. Th e extent of dopant segregation profile across the grain boundary cannot be determined using line scan EDXS, as it is difficu lt to resolve the peaks of Ce, Sm and Nd in the EDX spectrum. However, in t he EDXS result shown in Figure 6-6, Si peak was detected. This suggests the presence of Si near the grain boun daries. Thi s is consistent with literature, which states that solid oxide processi ng route to process powder usually contains higher im purity content. 012345678910 Ca Nd Nd Nd Nd Nd Nd Sm Sm Sm Sm Sm Sm Sm SmMoMo Ce Ce Ce Ce Ce Ce Ce Ce OIntensity (Arbitary Units)Energy (eV) Sm0.075Nd0.075Ce0.85O2-CMo SiSynthesized using Solid Oxide Route and Solid State Sintering Figure 6-6. EDX spectrum of SamCoSo sample.
154 6.3.2 Ionic Conductivity Figure 6-7 shows the normalized im pedance spectra of an electroded polycrystalline ceramic samples of Sm0.075Nd0.075Ce0.85O2synthesized using different processing routes, taken at 350oC in air. 0 1x1042x1043x1044x1045x1046x104 0 -1x104-2x104-3x104-4x104-5x104-6x104 0 1x1032x103 0 -1x103-2x103 Electrode Grain BoundaryZ / / (Ohms-cm)Z// (Ohms-cm)Z/ (Ohms-cm) Solid Oxide Route Synthesis + Solid State Sintering Co-precipitation Synthesis + Microwave Sintering Z/ (Ohms-cm)At 350oC, Sm0.075Nd0.075Ce0.85O2in air Grain Figure 6-7. Impedance spectrum of SamMiCo and SamCoSo samples at 350oC, in air. On comparing different Sm0.075Nd0.075Ce0.85O2ceramic samples at 350oC, it can be seen that the intra-grain pol arization semicircle in SamMiCo sample looks smaller when compared with SamCoSo. This may be attributed to t he siliceous impurities present in the SamCoSo sample which dissolve inside the crystal lattice (using interstitial mechanism), and consume the oxygen vacancie s which are the charge carriers in these oxide materials. Similarly, the impedance of the grain boundary polarization semi-circle for the SamCoMi sample is smaller than that of SamCoSo, which is consistent with the
155 microstructural features obser ved in the previous secti on. The sharp and un-relaxed grain boundary interfaces formed in the fast firing results in the lower grain boundary impedance in SamMiCo sample. Further, the rapid and homogeneous heating during the microwave sintering does not provide suffici ent time for the acceptor dopant cations to segregate near the grain bo undaries. Also, due to the higher purity powder (synthesized using co-precipitation technique), extrinsic siliceous impurities are almost absent in the SamMiCo sample. As reported in C hapter 4, at 550oC, the grain ionic conductivity of Sm0.075Nd0.075Ce0.85O2synthesized using conventional solid oxide route is 14 10-3 Scm-1. The conductivity value is 30% higher than that of Gd0.10Ce0.90O2synthesized using the same experimental conditions. Figure 6-8 shows the A rrhenius plot for the grain and t he total ionic conductivity in SamMiCo and SamCoSo samples. In addition, the Arrhenius plot for the grain and t he total ionic conductivity of Gd0.10Ce0.90O2synthesized using solid oxide route is al so shown. The sample synthesized using microwave sintering and co-precipitation technique, shows higher grain ionic conductivity than that of SamCoSo. At 550oC, the grain ioni c conductivity of Sm0.075Nd0.075Ce0.85O2of SamMiCo sample is 15 10-3 S.cm-1, which is around 45 % higher than that of Gd0.10Ce0.90O2(~10.7 10-3 S.cm-1). Table 6-2 shows the grain io nic conductivity value of Sm0.075Nd0.075Ce0.85O2at 600oC, synthesized using different processing techniques. For comparison, the grain ionic conductivity value of Gd0.10Ce0.90O2taken after Steele13 is also shown. It is important to note that the grain ionic conductivity reported by Steele is the highest ever reported for Gd0.10Ce0.90O2. It can be seen from the Table 6-2 that
156 Sm0.075Nd0.075Ce0.85O2synthesized using microwave sintering and co-precipitation technique exhibits higher grain i onic conductivity t han that of Gd0.10Ce0.90O2reported by Steele. This result shows the potential of Sm0.075Nd0.075Ce0.85O2as an electrolyte material for solid oxide fuel cells. 188.8.131.52.4 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 650600550500450400 Sm0.075Nd0.075Ce0.85O2(Coppt. Techn. + Microwave Sint.) Sm0.075Nd0.075Ce0.85O2(Solid State Reaction) Gd0.10Ce0.90O2(Solid State Reaction)T (oC)Total Ionic Conductivity Sm0.075Nd0.075Ce0.85O2(Coppt. Techn. + Microwave Sint.) Sm0.075Nd0.075Ce0.85O2(Solid State Reaction) Gd0.10Ce0.90O2(Solid State Reaction)Log[T(S.cm-1.K)] 1000/T (1/K)Grain Ionic Conductivity Figure 6-8. Arrhenius plot for the ionic conductivity of Sm0.075Nd0.075Ce0.85O2and Gd0.10Ce0.90O2materials.
157 Table 6-2. Comparison of the grain ionic conductivity of Sm0.075Nd0.075Ce0.85O2and Gd0.10Ce0.90O2at 600oC. SamCoMi SamCoSo Gd0.10Ce0.90O2(taken after Steele) Grain Ionic Conductivity (S.cm-1) 2.74 10-2 2.37 10-2 2.50 10-2 The total ionic conductivity of both Gd0.10Ce0.90O2and Sm0.075Nd0.075Ce0.85O2samples is lower than their corresponding grain ionic conductivity. This is basically due to the lower grain boundary conductivity contri bution towards the total ionic conductivity. Typically, the grain boundary ionic conductivity of acceptor doped ceria is two orders of magnitude lower than the corresponding grain ionic conductivity values, depending upon the dopant concentration and temperature.115 Thus, the lower grain boundary conductivity results in lower total ionic conductivity. Further, the total ionic conductivity of SamCoMi sample is higher than that of SamCoSo. This is consistent with the microstructural results discussed earlier. Since SamCoMi sample exhibits higher purity compared with SamCoSo sample, therefore, the grain boundary im purity phase blocking is comparatively lower in SamCoMi sample. In addition, the fa st firing in the microwave sintering does not provide su fficient time for dopant cations or impurities to segregate the near the grain boundar ies in the material. Further work need to be performed to separate out the individual e ffect of the powder synthesis, and the microwave sintering on the total ionic conductivity. Further, from Figure 6-8, it can be seen that the Sm0.075Nd0.075Ce0.85O2synthesized using solid stat sintering route exhibits higher total ionic conductivity than Gd0.10Ce0.90O2. This can be attributed to the comparatively higher grain and grain boundary resistivity of Gd0.10Ce0.90O2. It is already shown in previous chapter that
158 Sm0.075Nd0.075Ce0.85O2exhibits higher grain ionic conductivity than that of Gd0.10Ce0.90O2. However, the main contribution to the total resistivity of Gd0.10Ce0.90O2is from the grain boundary re sistivity. The lower gr ain boundary conductivity of Gd0.10Ce0.90O2may be associated with the Gd3+ segregation near the positively charged grain boundaries. The segregation of dopant blocks the oxygen vacancy diffusion across the grain boundaries. Future work is required to investigate the chemical compositional analysis near the grai n boundaries and to study the extent of dopant segregation in thes e two materials. 6.4 Summary and Conclusions In recent years, ionic conductivity of doped ceria electrolyte materials has been extensively studied. In lit erature a wide range of ionic conductivity data for Gd-doped ceria (which is commonly used for electrolyt e application) are given. After inspecting the experimental procedures followed in the work, it was observed that powder synthesized using wet chemical route usua lly shows higher ionic conductivity when compared with powder synthesized using solid st ate reaction. Further from liter ature it is known that the higher si ntering time and te mperature leads to lower grain boundary conductivity due to the segregation of accept or dopant cations (or impurities) near the grain boundaries. In order to prevent any segregation, fast firing technique to sinter ceramics is usually preferred over conventional sintering. This will promote lattice diffusion of material during sintering process, which results in s harp grains with unrelaxed grain boundaries. In the work presen ted in the previous chapter, co-doping in ceria leads to the development of Sm0.075Nd0.075Ce0.85O2electrolyte, which exhibits 30% higher grain ionic conductivity than Gd0.10Ce0.90O2. However, Sm0.075Nd0.075Ce0.85O2was synthesized using solid st ate reaction method. In the
159 present chapter, Sm0.075Nd0.075Ce0.85O2powder was synthesized using co-precipitation technique. This results in ul trafine particle size powder with lower impurity content. Sintering of the compacted green ceramic sample was carried out using microwave heating, at 1450oC for 1 h. Microstructure of t he microwave sintered ceramic sample shows the sharp grains with clean grain boundaries. Ioni c conductivity was measured using two-point probe ac impedance spectrosc opy technique. Both the grain and the total conductivity was higher in the sample synthesized using co-precipitation technique and microwave sintered, when compared with the sample synthesized using conventional solid state r eaction method. At 550oC, the obtained grain ionic conductivity was around 45 % higher than that of Gd0.10Ce0.90O2. On comparing with Steeles Gd0.10Ce0.90O2(which is the best ever repo rted grain ionic conductivity data for Gd0.10Ce0.90O2), the grain ionic conductivity of Sm0.075Nd0.075Ce0.85O2sample synthesized using co-precipitation techni que and microwave sinter ing, was 10 % higher at 600oC in air. The blocking effect of grai n boundaries is higher in the solid state reaction sample compared to the sample syn thesized using co-precipitation technique and microwave sintering. This can be attributed to the higher dopant cation and impurities segregation near the grain boundaries in the sample synthesized using conventional solid state reaction. These re sults indicate the co mbined effect of the powder synthesis route and the sintering technique on the ionic conductivity in doped ceria materials. Higher purity ceramic pr ocessing route, and the low temperature and time sintering technique are essential to ob tain better conductivity in these materials. However, further work needs to be done to s eparate out the individ ual effect of powder synthesis technique and microwave sintering on the grain boundary ionic conductivity.
160 CHAPTER 7 BUTTON CELL PERFORMANCE TESTING 7.1 Introduction Lowering the operating temper ature of solid oxide fuel cells (SOFCs) requires development of novel materials that ex hibit improved propert ies at intermediate temperatures. In the present work, focus is on designing higher ionically conductive materials for the electrolyte application in SOFCs. Using co-doping strategy, it has been shown in Chapter 4, that Sm0.075Nd0.075Ce0.85O2exhibits 30 % higher ionic conductivity than the commercially employed Gd0.10Ce0.90O2(GDC) electrolyte at 550oC, in air. In the present chapter, the potential of Sm0.075Nd0.075Ce0.85O2as an alternative electrolyte to G DC will be further investigated by testing the performance of SOFC device with flat-plate design. It is the most common cell design which offers simple cell geometry and multip le fabrication options. Furt her information about the cell design is discussed elsewhere.6 The performance of the SOFC device at intermediate temperatures will be discussed in this chapter. The fabrication of the anode-support ed prototype SOFC devices was assisted by Jin-Soo Ahn, graduate student in Materials Science & Engineering Department at University of Florida. 7.2 Experimental Procedure An anode-supported prot otype SOFC is f abricated, with Sm0.075Nd0.075Ce0.85O2electrolyte deposited on NiO-G DC composite anode using the conventional colloidal process. The La0.6Sr0.4Co0.2Fe0.8O3GDC composite is used as a cathode. Figure 7-1 shows the schematic of an anode-support ed flat-plate design of SOFC.
161 Figure 7-1. Schematic of anode-support ed prototype solid oxide fuel cell. The NiO GDC anode supports were synthes ized by tape casting a mixture of NiO (Alfa Aesar, 99% purity, CAS 1313) and GDC (Rhodia, LOT H-050708) powders. The powder mixture contai ns 65 weight % of NiO. Both the ra w oxide powders were accordingly weighed (to obtain 150 g batch of powder mixture), and are mixed to the organic mixture of toluene (23.5 cm3) and ethanol (21.5 cm3), with 1.5 g of solsperse (Air Products and Chemic als) which act as a dispersant in the slurry. The role of dispersant is to keep the dispersed cerami c particles in suspension in the dispersing medium. Further, a mixture of di-n butyl Phthalate (Alfa Aesar) (6.13 cm3 in volume), and polyethylene glycol (Fisher scientific) (1 .2 g in weight) was added as a plasticizer, while polyvinyl butyral (Acros organic) (6.1 cm3 in volume) was added as a binder to this slurry. Mixing was performed using ball milli ng for 24 h. After ball milling, the resulting slurry was transferred to a vacuum chamber, for de-airing. This process prevents any formation of pinhole defects or cracking (during tape ca sting), due to the air bubbles trapped inside the slurry. During the de-airing pr ocess, the slurry was constantly stirred with magnetic stirrer to avoi d solidification of t he slurry surface. The slurry was then tape-casted using tape-cast (Procast from DHI, Inc.). The NiO GDC tape was subsequently dried for 2 h at 100 C. Circular green tapes wi th 32 mm diameter were
162 then punched out from the tape. Figure 7-2 shows the green tape and the circular punched tape of NiO-GDC. The circular anode tapes were partially sintered at 900 C for 2 h. A BFigure 7-2. Green tape of NiO-GDC. A) and the circular punched tape B). The co-precipitation technique was us ed to synthesize phase pure powder of Sm0.075Nd0.075Ce0.85O2. One of the main objectives of using wet chemical route is to obtain fine particle size powder, which as a re sult enhances the sint ering kinetic of the ceramic powder. Thus, highly dense ceramic electrolyte can be synthesized using lower sintering temperature and time. The details about the powder synthesis using coprecipitation techniqu e are provided in Chapt er 3. The particle size distribution in Sm0.075Nd0.075Ce0.85O2powder obtained using co-preci pitation technique was monodisperse with most of t he particles were below 0.2 m in size. The mean particle size was around 0.095 m. For the deposition of electrolyte on anode support, the Sm0.075Nd0.075Ce0.85O2powder synthesized using co-precipitation technique was ball milled for 24 h, in ethanol medium with solsperse as a dispersing agent. After 24 h of ball milling, appropriate amount of polyvinyl butyral and di -n butyl phthalate were added
163 into the slurry. The ceramic slurry was ag ain ball-milled for another 24 h. Before the deposition, ceramic slurry was sonicated for 10 minutes. The ceramic slurry of Sm0.075Nd0.075Ce0.85O2was then deposited twice onto the anode (NiO-GDC) surface using pipette. The electrolyte deposit ed anode samples were subsequently, heattreated at 120 C for 5 h. The bi-layered structure of the electrolyt e and anode was then co-sintered at 1550 C for 4 h using a 3 C/minute ramp rate in air. Cathode ink was pre pared by mixing La0.6Sr0.4Co0.2Fe0.8O3(Praxair Specialty Ceramics, 99.9% purity) and th e GDC (Rhodia) powders in a 1:1 weight ratio, using mortar and pestle. The alpha-terpiniol a nd ethanol were added as a solvent, and polyvinyl butyral and di-n butyl phthalate were used as a binder and plasticizer, respectively. After mixing and grinding the cathode ink in mortar and pestle for 1 h, the ink was brush-painted evenly onto the Sm0.075Nd0.075Ce0.85O2electrolyte (deposited on NiO/GDC tape). The first layer of cathode ink was dried in an oven at 120C for 1 h. The second layer of the same cathode ink was then evenly brush painted on top of the first layer. After applying the cathode on top of the electrolyte, samples were then fired at 1100 C for 1 h. Pt paste (CL115349, Her aeus), as a current collector, was brushpainted onto both electrodes alo ng with Pt mesh, and Au co nnecting wires. Samples with current collectors and connecting wires were then finally heat-treated at 900C for 1 h. Figure 7-3 shows the flow chart for the complete experimental procedure for the fabrication of the prototype of solid oxide f uel cell sample. After the heat treatment, button cells were mounted on the ZrO2 reactor for testing th e power density in the intermediate temperature range.
164 Figure 7-3. Flow chart for the fabrication of anodesupported prototype SOFC. NiO / GDC (Anode) Solsperse Di-n Butyl Phthalate + Polyethylene Glycol Polyvinyl Butyral Acetone / Toluene Ball Milling, 24 h De-Airing of the Slurry Tape Casting Drying 100oC, 2 h Punching 32 mm Tapes Sintering at 900oC, 2 h Synthesis of Sm0.075Nd0.075Ce0.85O2Electrolyte using Co-precipitation Technique Ethanol Solsperse Ball Milling, 24 h Di-n Butyl Phthalate + Polyethylene Glycol Ball Milling, 24 h Electrolyte Slurry De p osition on Anode Drying 120oC, 5 h Sintered 1550oC, 4 h LSCF / GDC -Terpinol / Ethanol Polyvinyl Butyral Di-n Butyl Phthalate Mortar & Pestle, 1 h Cathode Ink Deposition on Electrolyte Sintering at 1550oC 1 h
165 Figure 7-4. Experimental setu p for the I-V characteristi cs measurement of the test SOFC sample. Figure 7-4 shows the experimental setup for the current-voltage characteristics measurements of the SOFC sample at dif ferent temperatur es. The fuel cell samples were sealed (anode side) to a zirconia tube in a custom-made testing apparatus using two-part ceramabond sealant (a mixture of 517-pow der and 517-liquid from Aremco). The setup was then placed into a furnace, cured, and taken up to testing temperature. Dry air and H2/H2O gas mixtures were used as the ox idant and fuel gases, respectively. Flow rates were maintained using mass flow controllers. The 90 cm3/minute of dry air and wet hydrogen (3 volume % H2O) were supplied to the cathode and anode side, respectively. The cell open circuit potential (OCP) was monitored using a Solartron
166 1287 potentiostat until a stable value was reached, and the cu rrent-voltage (I-V) measurements were taken with the same instrument. The impedance measurement was carried out at open circuit conditions using two-point probes ac impedance spectroscopy technique. A Par-stat 2273 (Princeton Applied Research) frequency response analyzer was used for impedance measurement. Impedance spectra were measured from 10 kHz to 0.01 Hz, at differe nt temperatures. From the high frequency intercept of the impedance spectrum with the real axis, the ohmic ASR value was calculated (after normalizing the resistance according to cathode area). Electrode ASR value was determined from the differ ence between the lo w and high frequency intercepts (after normalizing the resistance according to cathode area). Field emission scanning electron microscope (FE-SEM) was used to visualize the microstructure of the fuel cell sample. In or der to view the cross-se ction of all the three layers (anode, electrolyte, and cathode) of the cell, the test sample was fractured through the electrodes. T he fractured sample was em bedded in epoxy-resin, and was mechanically polished. 7.3 Results and Discussion 7.3.1 Particle Size Figure 7-5 shows the number particle si ze distribution of the Sm0.075Nd0.075Ce0.85O2and the commercially obtained La0.6Sr0.4Co0.2Fe0.8O3(LSCF) powders. The particle size distribution of the phase pure Sm0.075Nd0.075Ce0.85O2powder synthesized using co-precipitation technique was measured using Beckman Coulter LS13320. The particle size distribut ion of LSCF powder is provided by the company. In order to achieve high density ce ramic electrolyte, it is desired that the particle size should be less than 1 m with particle size dist ribution to be narrow and
167 monodisperse. It can be seen that the Sm0.075Nd0.075Ce0.85O2particles exhibit size less than 1 m, with the mean size of 0.095 m. Further, the particle size distribution is also monodisperse which is desired for good sintering in reasonable time. 0.11101001000 0 2 4 6 8 10 12 14 16 18 Sm0.075Nd0.075Ce0.85O2La0.6Sr0.4Fe0.8Co0.2O3Diffrential Number (%)Particle Size (m) Figure 7-5. Particle size distribution of Sm0.075Nd0.075Ce0.85O2synthesized using coprecipitation technique and La0.6Sr0.4Co0.2Fe0.8O3(obtained from Praxair Specialty Ceramics). The GDC powder obtained from Rhodia compri sed of very fine particles of size less than 100 nm.127 These particles form aggregates due to their high surface area. On the other hand, the NiO parti cles are mostly micron-sized.127 The performance of the SOFC also depends upon the electrode reactions which prim arily take place at triple phase boundaries (TPB) of elec trolyte, electrode and air. Increasing the TPB length, increases the electrode reaction sites wh ich consequently enhances the performance of SOFCs. The electrodes consist of smaller particles result in larger TPB lengths. Thus, the lower particle size is typically requir ed to increase the surface reactivity, and reduce the activation polarization. However, lowering the particle size enhances the densification of the electrode. The dry air and oxygen gas can easily pass through the
168 dense cathode; however, it is difficult for a dense anode to provide fuel rapidly to the reaction sites, and to remove water molecules e fficiently. This results in the increase in the concentration polarization. To avoid th is issue, the micron-sized NiO powder was used in the fabrication of anode tape. The LSCF powder used for the cathode exhibit the number particle size distri bution which is bi-modal, with most of the particles around 1 m in size (shown in Figure 7-5). 7.3.2 Microstructural Analysis Figure 7-6 shows the mi crostructure view of the crosssection of the fuel cell, and the surface of Sm0.075Nd0.075Ce0.85O2electrolyte. It can be seen both in the crosssectional and surface micrographs that the electrolyte is densely sintered except for a few isolated residual pores. From the SEM image, the thickness of the Sm0.075Nd0.075Ce0.85O2electrolyte is estimated to be ~ 5 m. This suggests that the Sm0.075Nd0.075Ce0.85O2electrolyte can be easily depos ited with conventional ceramic routes such as colloidal deposition. As expected, the nickel (reduced form of NiO) particles in the anode are exc eptionally larger compared to LSCF particles in the cathode. This is mainly due to the large part icle size of the starting powder i.e., NiO. Although, the hydrogen oxidation in the anode side is ki netically more favorable than oxygen reduction in the cathode si de, the significantly large par ticle size of the Ni near the electrolyte and anode inte rface can cause high anodic polarization, at low temperatures.
169 Figure 7-6. FE-SEM images of A) cross-section view of electrodes and electrolyte, and B) surface of t he electrolyte. 7.3.3 Impedance Analysis Figure 7-7 shows the impedance spectrum of the test SO FC sample at 650, 600, 550 and 500C. Using impedance spectrosc opy, it is possible to separate out the electrode and the ohmic contribut ions to the total area specific resistance (ASR) value at each temperature. 0.00.20.40.60.81.01.21.41.6 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 0.00.10.20.3 0.0 0.1 0.2 0.3 -Z// imaginary (Ohms)Z/ real (Ohms) Impedance Spectrum at 650 oC Impedance Spectrum at 600 oC Impedance Spectrum at 550 oC Impedance Spectrum at 500 oC -Z// imaginary (Ohms)Z/ real (Ohms) Figure 7-7. Impedance spectrum of the SOFC cell measured at different temperatures. Electrolyte B Cathode Anode Electrolyte A
170 650600550500 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Total Electrode OhmicArea Specifc Resistance (Ohms-cm2)Temperature (oC) Figure 7-8. Area specific resist ance at different temperatures. Figure 7-8 shows the electrode and the ohmic ASR values s eparated from the total ASR at different temperat ures. The rapid increase in the total ASR value with the decrease in temperature is mainly due to the significant increase in the electrode polarization. On the other hand, ohmic ASR value stays relatively low even at lower temperatures. The total ASR values were also calculated from the gradient of the linearly fitted IV curves. Table 7-1 shows the com parison between the total ASR values obtained from the impedance measurements and I-V characteristics. Impedance measurements were done under open circuit condition, therefore wh ile fitting the I-V cu rve, the region near zero current was taken into account. Both ASR values from the IV characteristics and
171 the impedance measurement are comparable in all temperat ures except 500C. The ohmic contribution towards the total ASR at each temperature is tabulated in Table 7-1. Comparison betwe en the total ASR obtained from I-V characteristic and impedance measurement. The ohmic contribution towards the total ASR is also shown. Temperature (oC) ASRI-V (Ohms-cm2) ASRImpedance (Ohms-cm2) % Ohmic 650 0.083 0.087 30.78 % 600 0.244 0.235 21.74 % 550 0.507 0.536 17.54 % 500 1.189 1.487 12.03 % As shown in Table 7-1, at 650C the ohmic cont ributi on toward the total ASR is 30%. However, with the decreasing tem perature ohmic ASR % contribution becomes smaller. At 500C, the ohmic contribution to the total ASR value is only 12.03%. This can be explained by two aspects. First, t he high activation energy of the oxygen reduction reaction results in higher resist ance from the cathode at low temperatures. Secondly, without the anode functional laye r, large particles of NiO at the anodeelectrolyte interface lowers the hydrogen ox idation reaction kinetics which results in additional resistance. At this point, due to the limitation of two point impedance measurement, we cannot decide which electrode ASR is dominant. 7.3.4 Power Density Figure 7-9 shows the I-V characteristi cs of the prototy pe of SOFC with Sm0.075Nd0.075Ce0.85O2electrolyte at various temper atures, ranging from 500C to 650C. The flow rate of 90 cm3/minute of dry air and wet hy drogen was maintained in cathode and anode side, respectively. The open circuit potential (OCP) values obtained
172 from this test cell were 0. 72, 0.79, 0.77 and 0. 84 V at 650C, 600 C, 550C and 500C, respectively. The obtained OCP values were close to the typical OCP values achieved in the cell with GDC as an electrolyt e, using similar experimental setup.127 The open circuit potential value of the tested cell is also dependent upon the properties of sealant used in the device. The gas sealant should be highly dense with minimal porosity, and also match the thermal ex pansion coefficient of Sm0.075Nd0.075Ce0.85O2electrolyte. Future work need to be performed to identify materials to be used as sealant for SOFCs based on Sm0.075Nd0.075Ce0.85O2electrolyte. 012345 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Power Density (W/cm2) Open Circuit Potential (V)Current (A/cm2)1.38 W/cm2 At 650 oC At 600 oC At 550 oC At 500 oC Figure 7-9. The I-V characteristics of the prototype SO FC sample with Sm0.075Nd0.075Ce0.85O2electrolyte at various tem peratures ranging from 500 to 650C in 90 cm3/minute of both dry air and wet hydrogen. Figure 7-9 also shows the power density as a function of cu rrent. The maxi mum power densities achieved in the test cell were 1.38, 0. 71, 0.34 and 0.17 W/cm2 at 650, 600, 550, and 500C, respectively. Considering the lower OCP values in the test cell, the obtained power dens ities are exceptionall y high. This can be contributed to the high
173 ionic conductivity of Sm0.075Nd0.075Ce0.85O2electrolyte. Also, it is important to note that the I-V characteristics and the maximum power density of the SOFC is a function of numerous processing and material variable s. For this reason, the potential of Sm0.075Nd0.075Ce0.85O2electrolyte cannot be directly co mpared with that of GDC using I-V characteristics. Comparison between the ionic conductivity of Sm0.075Nd0.075Ce0.85O2and GDC electrolyte was investigated in detail in Chapter 4.53,67 However, the performance testing results clearly suggest that the Sm0.075Nd0.075Ce0.85O2electrolyte material can su ccessfully generate high power density in SOFCs operating in the in termediate temperature range. Further, it is well known that the ionic phase in the electrode composite also have an effect on the SOFC efficiency.128,129 Higher ionic conductive material in the electrodes significantly enhances t he power density of the SOFC. As Sm0.075Nd0.075Ce0.85O2exhibit higher ionic conductivity than GDC, it is expected this will further improve the performance of SO FCs. The optimizati on of the processing parameters involved while replacing GDC with Sm0.075Nd0.075Ce0.85O2material in the electrode composites is left as a future work. 7.4 Summary and Conclusion The performance testing of the novel co-doped ceria ma terial as an electrolyte was performed by fabricating anode supported prot otype of SOFC. The electrolyte was deposited on top of the porous NiO-GDC anode tape using conventional colloidal technique. The microstructure result s show that the deposited layer of Sm0.075Nd0.075Ce0.85O2ceramic is highly dense with very few residual pores present. The LSCF-GDC composite was used as a cathode, which was brushed on top of the dense sintered electrolyte layer. The current -voltage characteristics were measured at
174 different temperatures, using 90 cm3/minute of dry air and we t hydrogen in cathode and anode sides, respectively. The excepti onally high power density of 1.38 W/cm2 was obtained at 650oC. The power density of t he tested cell shows that the Sm0.075Nd0.075Ce0.85O2ceramic can successfully generate higher performance in SOFCs working in intermediate temperatur e range. The ASR values obtained from impedance measurements matches very clos ely with that measured using I-V characteristics. On lowering the temperature, the ohmic contribution towards the total ASR decreases. This implies that the elec trode contribution to the total ASR value is significantly higher at lower temperatures. By replacing t he ionic phase in the electrode composite from GDC to Sm0.075Nd0.075Ce0.85O2can further improv e the performance of SOFCs operating in the interm ediate temperature range.
175 CHAPTER 8 SUMMARY AND FUTURE WORK 8.1 Summary Solid oxide fuel cell (SOFC) is a promising technology for the future electrical power generation, as its shows high efficiency, low emissions and fuel flexibility. In order to promote the development of SOFCs for a variety of energy needs, it is essential to lower the operating temperature to the intermediate tem perature range (400 700oC). This will not only lower the fabricat ion and maintenance cos t, but also improves the reliability of SOFCs. However, with current SOFC materials lowering the operating temperature increases the ohmic losses, and degrades the performance and efficiency of the system. Thus, it is required to develop materials for all the components of SOFC that exhibit improved properties in the intermediate temperatur es. This work focused on the development of higher ionic conductive materials for elec trolyte applications in the intermediate temperature range. In recent years, doped ceria materials have received significant attention for electrolyte application, due to their high ionic conductivity and good thermodynamic stability in intermediate tem peratures. Until this work, among doped ceria materials, Gd0.10Ce0.90O2exhibited the highest ionic conductivi ty at intermediate temperatures. The main goal of this work was to further enhance the ionic conductivity in doped ceria systems. Two different doping strategies were used on ceria such that the interactions between point defects formed by the substi tution of acceptor dopant cations for host Ce4+ cations, become minimal. Also, it is essential to maximize the pre-exponential coefficient (in the Arrhenius relationship) which is the function of configurat ional entropy. Co-dopants were used to suppress the ox ygen vacancy ordering, and increase the
176 configurational entropy of the system. The co-dopants pairs were selected based on the individual dopant ionic radius, to test two di fferent doping strategies i.e., critical ionic radius concept, and the results from density functional theory (DFT). According to the critical dopant ionic radius concept, it was suggested that the ideal dopants for ceria to exhibit higher i onic conductivity, should possess ionic radius closer to the critical dopant ionic radius (rc) value. The rc is the ionic radius of the dopant that on addition to the hos t fluorite structure of ceria causes no lattice elastic strain. In order to test this hypothesis, co-dopant pairs were selected and added in host ceria such that the weighed average ionic radius of co-dopants matches rc value. On comparing the ionic conductivity of the novel ceramic materials with that of Gd0.10Ce0.90O2synthesized under similar experimenta l conditions, it was observed that the electrolyte based on t he critical dopant ionic radius concept (i.e., LuxNdyCe1x yO2and Gd0.056Y0.044Ce0.90O2) exhibit lower grain ionic conductivity than that of Gd0.10Ce0.90O2material. Another novel doping strat egy based on recent DFT results was tested which suggests using co-dopant with an average effective atomic number of Pm (Z = 61). By doing so, the increase in the number of equi-interaction ener gy sites of oxygen vacancy was expected, which as a result, facilitates oxygen diffusion. This in turn increased the ionic conductivity of the material. Therefore, a co-dop ing scheme using Sm3+ and Nd3+ provided the experimental scenario to test th is hypothesis. It was found that the lattice parameter as a function com position obeys Vegards law which indicates short range oxygen ordering in Smx /2Ndx /2Ce1xO2system is not as pronoun ced as in other singly doped ceria system. The grain ionic conductivity of Sm0.05Nd 0.05Ce0.90O2was found to
177 be higher than that of Gd0.10Ce0.90O2. Among all the compositions (x) of Smx /2Ndx /2Ce1xO2, Sm0.09Nd0.09Ce0.82O2exhibits the highest grain ionic conductivity above 600oC (3.47 10-2 Scm-1 at 650oC). The grain ionic conductivity of Sm0.09Nd0.09Ce0.82O2was found to be 35% higher than that of Gd0.10Ce0.90O2at 650oC. Below 600oC, Sm0.075Nd0.075Ce0.85O2is the highest ionically conductive material in doped ceria material. At 550oC, the grain ionic conductivity of Sm0.075Nd0.075Ce0.85O2was found to be 30% higher than that of Gd0.10Ce0.90O2. Thus, it was shown that the codoping based on Pm atomic number results in enhanced ionic conductivity. In addition, at higher dopant concentrations, Smx /2Ndx /2Ce1xO2system follows Meyer-Neldel rule. Taking the comm on transition point of Arrhenius plots as association-dissociation tem perature of oxygen vacancie s, migration enthalpy in Smx /2Ndx /2Ce1xO2system was calculated. It was shown that the mi gration enthalpy does depend on dopant concen tration in the dilute regime Furthermore, shift in the grain ionic conductivity maxima toward high er dopant concentration with the increase in temperature, was observed in Smx /2Ndx /2Ce1xO2system. Ionic conductivity also depends upon the pr ocessing conditions. There are a wide range of ionic conductivity values for Gd0.10Ce0.90O2reported in literature. In order to understand the ionic conductivi ty trend among different triv alent acceptor doped ceria materials, processing variable s in the synthesis of these materials were kept constant. Thus, in the present work, a consistent ionic conductivity data set was prepared for different doped ceria materials (D0.10Ce0.90O2) synthesized using similar experimental conditions. Dopant concentration in these compositions is fixed to understand the effect of dopant ionic radius. Among triv alent acceptor dopant cations, Nd3+ exhibits the
178 highest ionic conductivity in the host ceria. To study the effect of dopant content, different compositions of Smx /2Ndx /2Ce1xO2ceramics were also synthesized. Initially, with the increase in the dopant concentration, the ionic conductivity increases. It reaches maximum at certain dopant concentrat ion, and then signifi cantly decreases. Among all the tested dope d ceria systems, Sm0.075Nd0.075Ce0.85O2exhibited the highest ionic conductivity, at 500oC in air. Among all the trivalent acceptor dopants, Nd3+ exhibited the highest ionic conductivity in ceria. Howe ver, the ionic radius of Nd3+ is far away from the rc value. It is important to note th at in the structure (rc) property (ionic conductivity) relationship proposed by Kim22, both the parameters were measured in different conditions (temperatures). In order to verify the rc concept, rc values were determined at temperatures where the ionic conductivity is measured. The graph between the elastic strain and the dopant cation ioni c radius was plotted at hi gher temperatures, to derive rc values. The rc value decreased with the incr ease in temperature. At 500oC, the rc value was estimated to be 1.0252 for trivalent dopant cations in host ceria. The rc value obtained at 500oC lies closest to Dy3+ ionic radius. However, it is Nd3+ that exhibits the highest ionic conductivity. This clearly indica ted that the rc concept is not applicable for doped ceria materials. In order to investigate the effect of processing on the ionic conductivity, different processing routes were used to synthesize material. From lit erature, it is reported that the powder synthesized using wet chemic al route usually shows higher ionic conductivity than conventional solid oxide route. Further, it is also known that higher sintering times and temperatures lead to lower grain boundary conduc tivity due to the
179 segregation of acceptor dopant cations (and impurities) near the grain boundaries. In order to prevent any segregation, fast firing technique to sinter ceramics is usually preferred over conventi onal sintering. As Sm0.075Nd0.075Ce0.85O2exhibited the highest ionic conductivity among the test ed doped ceria materials, therefore in the present work, Sm0.075Nd0.075Ce0.85O2powder was synthesized using co-precipitation technique. Sintering of the compacted green ceramic sample was done using microwave heating, at 1450oC for 1 h. Microstructu re of the microwave sint ered ceramic sample shows sharp grains with clean grai n boundaries. Both the grain and the total ionic conductivity were higher in the sample synthesized usin g co-precipitation tech nique and microwave sintering, when compared with the sample synthesized usin g conventional solid state reaction method. At 550oC, the obtained grain ioni c conductivity was around 45 % higher than that of Gd0.10Ce0.90O2. Comparison with Steele13s Gd0.10Ce0.90O2(which is the best ever reported grai n ionic conductivity data for Gd0.10Ce0.90O2), the grain ionic conductivity of Sm0.075Nd0.075Ce0.85O2sample synthesized using co-precipitation technique and microwave sinteri ng, was 10 % higher at 600oC in air. The blocking effect of the grain bou ndaries is higher in the solid stat e reaction sample compared with the sample synthesized using co-precipitation technique and microwave sintering. This can be attributed to the segr egation of dopant cations and impurities near the grain boundaries in the sample synthes ized using conventional solid state reaction. These results indicated the combined effect of the pow der synthesis rout e and the sintering technique on the ionic conductivity in doped ceria materials. After developing the higher ionically conducti ve material, perform ance testing of an anode supported prototy pe of SOFC based on Sm0.075Nd0.075Ce0.85O2electrolyte was
180 performed. The electrolyte was depos ited on top of t he porous NiOGd0.10Ce0.90O2anode tape using conventional colloidal techniqu e. The microstructu re showed that the deposited layer of Sm0.075Nd0.075Ce0.85O2ceramic is highly dense with very few residual pores present. The La0.6Sr0.4Co0.2Fe0.8O3Gd0.10Ce0.90O2composite was used as a cathode, which was brushed on top of t he dense sintered electrolyte layer. The current voltage characteristics were m easured at different temperatures, using 90 cm3/minute of dry air and wet hydrogen in ca thode and anode sides, respectively. The exceptionally high po wer density of 1.38 W/cm2 was obtained at 650oC. The power density of the tested ce ll shows that the Sm0.075Nd0.075Ce0.85O2ceramic can successfully generate higher performance in SOFCs working in intermediate temperature range. 8.2 Future Work 8.2.1 Doping Strategy As mentioned in Chapter 1, the main sci entific approach to ac hieve the goal of developing higher ionic conductive materials is to identify the doping strategy. It was shown in Chapter 4 that Pm is the ideal dopant to exhibit t he highest ionic conductivity. Since Pm is a radioactive element, and cannot be used for this application, therefore co-dopant strategy based on Sm3+ and Nd3+ was used on ceria. This leads to the development of highly ionically conductive Sm0.075Nd0.075Ce0.85O2material. Future work is needed to further test this strategy using Gd3+ and Nd3+ as co-dopant pair. Using multiple trivalent dopants suppresses the oxygen vacancy ordering, and increases the configurati onal entropy of the system.67 This, in turn, enhances the ionic conductivity of the material. In addition, fr om Chapter 5, it is clear that the dopant cations exhibiting the ionic radius in the range of 1.05 to 1.11 show low activation
181 energy for oxygen diffusion. Si nce dopant ionic radius of Gd3+, Sm3+ and Nd3+ cations lie in this range, it would be interest ing to study the ionic conductivity of Gdx /3Smx /3Ndx /3Ce1xO2system in detail. Also, in Chapter 5, it is shown that Nd3+ dopant cation exhibits the highest ionic conductivity among all the other trivalen t dopant cations. However, the ionic conductivity result does not follow the hos t and dopant cations ionic radius mismatch concept. Therefore, work is required to study in detail the crystal structur e and the grain ionic conductivity in NdxCe1xO2system. As grain boundary is the integral part of the polycrystalline ceramic material, the study of the grain boundary ionic conductivity for different compositions in NdxCe1xO2system need to be performed. Finally, work is required for the optimizati on of dopant content in NdxCe1xO2system to obtain higher ionic conductivity. 8.2.2 Ionic Conductivity Among all the tested doped ceria compositions, Sm0.075Nd0.075Ce0.85O2exhibits the highest grain ionic conductivity in the intermediate temperature range, in air. However, under real oper ation, the solid oxide electrolyt e is also exposed to reducing environment at the anode side. Thus, it is important to measure the ionic conductivity of Sm0.075Nd0.075Ce0.85O2as a function of partial pressure of oxygen. Further, in reducing conditions (~10-20 atmosphere) and at hi gh temperatures (~1000oC), CeO2 is prone to change its valence state from Ce4+ to Ce3+, which promotes electronic conduction in the material. This not only degrades the associ ated cell efficiency, but also expands the material (due to the formation of additional oxygen vacancies) wh ich has a deleterious effect on the mechanical properties.73 Thus, for Sm0.075Nd0.075Ce0.85O2electrolyte, it is
182 required to measure the transference number of Sm0.075Nd0.075Ce0.85O2in reducing atmospheres. In Chapter 5, a trend for t he pre-exponential factor (in Arrhenius relationship) was shown as a function of the tr ivalent dopant ionic radius. Si nce pre-exponential factor depends upon numerous variables (s hown in Appendix B), it is difficult to de-convolute the contribution from different variables to ward the pre-exponential factor. Thus, a computational approach, like density functional theory, ma y be extremely helpful to investigate the effect of each variable independently. This work will allow better understanding of the pre-exponential factor for different type of acceptor dopants. Using this knowledge, novel doping strategy can be designed (to develop higher ionically conductive materials) for ceria and related fluorite based oxides. 8.2.3 Processing In Chapter 6, it has been shown that the ionic conductivity depends on processing conditions. Samples synthesized using co-precipitation technique and microwave sintering exhibit higher grain and total i onic conductivity, than the samples prepared through solid state reaction met hod. As described in chapter 6, microwave sintering is a fast firing technique in which the microstructure with fine grain size is obtained. Thus, it is important to study the effe ct of the microwave sintering over the microstructure of doped ceria material. Further, the amount of impurities segregat ion (near the grain boundary core) is also dependent upon the orientation of the grains forming the grain boundary. The high angle grain boundaries can accommodate more impurities when compared with low angle grain b oundaries. Future work is r equired to study in detail the grain boundary configurat ion in the microwave sintered samples. Furthermore, ionic conductivity measurements need to be performed on the conventional sintered ceramic
183 samples formed using the powder synthesized from co-precipita tion technique. This will separate out the effect of the powder synt hesis route on the ionic conductivity of the material. 8.2.4 Performance Testing In Chapter 7, it wa s shown that the Sm0.075Nd0.075Ce0.85O2electrolyte material can successfully generate high power densit y in SOFCs operating in the intermediate temperature r ange. At 650oC, exceptionally high po wer density of 1.38 W/cm2 was obtained using 90 cm3/minute of dry air and wet hy drogen in cathode and anode sides. However, with the decrease in temperature, the power density of the tested cell degrades significantly. This is basically due to the rapid increase in the electrode area specific resistance (ASR) with the decrease in temperature. This electrode ASR can be lowered by replacing the ionic phase in the electrode composite. It is well known that the ionic phase in the electr ode composite also have an effect on the SOFC efficiency. Higher ionic conductive material in the electrodes significantly enhances the power density of the SOFC. As Sm0.075Nd0.075Ce0.85O2exhibit higher ionic conductivity than GDC, it is expected this will further improve the performance of SOFCs.128,129 The optimization of the proce ssing parameters involved while replacing GDC with Sm0.075Nd0.075Ce0.85O2material in the electrode composit es is left as a future work. In addition, without the anode functional layer, large particles of NiO at the anode-electrolyte interface lowe rs the hydrogen oxidation reac tion kinetics which results in additional resistance.127 Thus, using anode functional layer in the interface of anode and electrolyte can substantially improve the performance of SOFCs operating in the intermediate temperature range. Further, the open circuit potential value of the tested cell is also dependent upon the properties of sealant used in the device. The gas
184 sealant should be highly dense with mini mal porosity, and also match the thermal expansion coefficient of Sm0.075Nd0.075Ce0.85O2electrolyte. Future work need to be done to identify materials to be used as sealant for SOFCs based on Sm0.075Nd0.075Ce0.85O2electrolyte.
185 APPENDIX A DEFECT CHEMISTRY AND THERMODYNAMIC IN POINT DEFECTS Diffusion i n oxide solids can only take pl ace due to the presence of imperfections (point defects) in the lattice. For exampl e, Schottky defect in MgO results in the formation of a pair of Mg+ and O2vacancies in the bulk crystal. The OVdefects play the role of charge carriers and enable O2 gas to diffuse through the solid material. // Mg OVVNull (A-1) Even though introduction of defects involv e breaking of bonds, there is a finite concentration of defects always present in t he crystal structure at all the temperature above 0 K. This is attribut ed to the entropy increase due to the presence of these defects. The total entropy of a collecti on of atoms can be written as a sum of configurational entropy ( Sconfig) and vibrational entropy (or thermal entropy) ( Svibra). vibra configSSS (A-2) Thus, if x mole fraction of certain type of point defect, then the entropy change ( S) due to the formation of these defects is, /ln 1ln1lnxkxxxxkS (A-3) where k, and /are Bolzmanns constant, the vi brational frequency of atoms far away from oxygen vacancy, and the vibrational frequency of at oms in close proximity to oxygen vacancy, respectively. If the energy needed to form these defect is Hfor Joules per mole then the increase in enthalpy ( H) is, forHxH ............... .. .............(A-4)
186 The total change in Gibbs free energy (G) is STHG (A-5) From (A-3) and (A-4), /ln 1ln1lnxkTxxxxkTHxGfor (A-6) The concentration of defect at which the minimum in G occurs (i.e., x G ), is the equilibrium concentration (xeq) of defect for a parti cular temperature, RT STH xvib for eqexp (A-7) Thus, the final expression does not contain the configurat ion entropy term but only depends upon the change in Gibbs energy by the formation of a single defect.
187 APPENDIX B IONIC CONDUCTIVITY AND DEFECT COMPLEXES B.1 Oxygen Ion Conductivity Oxygen ion conductivity (i) in a solid oxide electrolyte obeys Arrhenius relationship, kT Hm oiexp (B-1) where o, Hm, k, and T pre-exponential coefficient, migration enthalpy, and Boltzmann constant. Oxygen vacancy conductivity ca n be expressed in terms of oxygen vacancy mobility as, VVoOVqNV ][ (B-2) where No is the number of oxygen sites per unit volume (cm-3), ] [ OV is the fraction of mobile oxygen vacancy present in the anion sublattice, and qv and v are the charge and mobility of oxygen vacancies, respective ly. According to the Nernst Einstein relationship, oxygen vacancy mobility can be described with the corresponding diffusivity (Dv). kT DqVV V (B-3) The oxygen vacancy diffusivity can be written as, kT H k S aDm m o Vexp exp2 (B-4) where a is the ion jump distance, o is an appropriate lattice vibration, and Sm is the entropy change during oxygen ion diffusion. Oxygen ion diffusivity can be describe in terms of oxygen vacancy diffusivity as,
188 VOiDVD][ (B-5) From (B-4) and (B-5), kT H k S aVDm m o Oiexp exp][2 (B-6) The oxygen ion mobility (i) can be written as, kT H k S aV kT qm m o O V iexp exp][2 (B-7) Using (B-7) and (B-2), oxygen ion conductivity can be written as kT H k S aNCV kT qm m ooOO V iexp exp ][2 2 (B-8) where Co is the mole fraction of oxygen ions present in the material, kT H k S aNVV kT qm m ooO O V iexp exp ])([2 2 (B-9) For smaller values of ] [OV equation (B-9) can be approximated as, kT H k S aNV kT qm m ooO V iexp exp ][2 2 (B-10) B.2 Defect Complexes In equation (B-10), it is assumed that there are no interact ions between dopant cations and oxygen vacancies sites. However, the presence of elastic and electrostatic interactions between these tw o point defects result in t he formation of local defect structures such as / / A OADVD and OAVD/. At lower dopant concentration, it can be assume that only dimers (i.e., OAVD/) are present in the material. However, with the increase in dopant c oncentration, the probability t hat oxygen vacancy has more than one dopant cation in the neighboring position rapidly increases. Thus, at higher
189 dopant concentration, the presence of trimers (i.e., / /A OADVD ) is more likely. In the following section, equation (B-10) is modified taken into consideration both dimers and trimers. B.2.1 Dimers In this section, equation (B10) is modified assuming that only dimers are present. OCe OCeVDVD/ / (B-11) Applying the law of mass-action, ])[( ][/ /OCe O Ce DimersVD VD K (B-12) where Kdimers is the equilibrium constant. Fr om charge neutrality relationship, ])[(/ /Ce O OCeDVVD (B-13) For full association of defects, ])[(/ O OCeV VD (B-14) dop Ce OCecDVD])[(/ / (B-15) where cdop is the concentration of dopant cati ons. Thus, equation  becomes, ][O DimersVK (B-16) Now, equilibrium constant KDimers is a function of temperature (T) and can be written as, kT G AKDimers Dimersexp (B-17) kT STH AKDimers Dimers Dimersexp (B-18)
190 where A, GDimers, HDimers, and SDimers are the pre-exponential constant for the reaction, Gibbs free energy, enthalpy, and entr opy of association, respectively. Using (B-16) and (B-17), kT STH AVDimers Dimers Oexp] [ (B-19) Inputting the equation (B -19) into (B-10), kT H k S aN kT STH A kT qm m oo Dimers Dimers V iexp exp exp2 2 (B-20) kT HH k SS aAN kT qDimers m Dimers m oo V iexp exp2 2 (B-21) Equation (B-21) is the modified equation (B10) assuming only dimers as a local defect structure present. B.2.2 Trimers In this section, it is assumed that only trimers local defect st ructures are present while modifying equation (B-10). OCe Ce OCeVD DVD/ / /2) ( (B-22) Applying the law of mass-action, ])2[( ][/ 2/OCe O Ce TrimersVD VD K (B-23) where KTrimers is the equilibrium constant. Fr om charge neutrality relationship, ][2/Ce ODV (B-24) Using equation (B -23) and (B-24),
191 ])2[( ][4/ 3OCe O TrimersVD V K (B-25) For full association of defects, ])2[(/ O OCeV VD (B-26) dop OCecVD])2[(/ (B-27) where cdop is the total dopant concentration expre ssed as a site fraction of the cation site. Similar to dimers case, equilibrium constant KTrimers is a function of temperature (T) and can be written as, kT G AKTrimers Trimersexp (B-28) kT STH AKTrimers Trimers Trimersexp (B-29] where A, GTrimers, HTrimers, and STrimers are the pre-exponent ial constant for the reaction, Gibbs free energy, enthalpy, and entr opy of association, respectively. Using equation (B-25), (B-27), and (B-29), kT STH cA VTrimers Trimers dop O3 exp 4 1 ][3/13/1 3/1 (B-30) Inputting the equation (B -30) into (B-10), kT H H k S S aNcA kT qTrimers m Trimers m oodop V i3 exp 3 exp 4 123/13/1 2 3/1 (B-31) Equation (B-31) is the modi fied equation (B-10) assuming only trimers as a local defect structure present. The conductivity should follow above trends at higher dopant concentration, when the probability of the presence of trimers is higher.
192 APPENDIX C HIGH TEMPERATURE XRD PATTERNS OF CERIA COMPOUNDS In this section, XRD pr ofiles of doped ceria systems at high temperatures are shown. 304050607080 600 oC 500 oC 400 oC 300 oC 200 oC 100 oC 40 oCIntensity (Arbitary Units)2 (Degrees)CeO2 Figure C-1. XRD profile s of pure ceria collected at high temperatures. 304050607080 29.229.4 Intensity (Arbitary Units)2 (Degrees) 600 oC 500 oC 400 oC 300 oC 200 oC 100 oC 40 oCIntensity (Arbitary Units)2 (Degrees)Yb0.10Ce0.90O2-(111) Figure C-2. XRD profiles of Yb0.10Ce0.90O2collected at high temperatures.
193 304050607080 47.247.447.6 Intensity (Arbitary Units)2 (Degrees) 600 oC 500 oC 400 oC 300 oC 200 oC 100 oC 22 oCIntensity (Arbitary Units)2 (Degrees)Y0.10Ce0.90O2-(220) Figure C-3. XRD profiles of Y0.10Ce0.90O2collected at high temperatures. 304050607080 46.847.147.4 Intensity (Arbitary Units)2 (Degrees) 600 oC 500 oC 400 oC 300 oC 200 oC 100 oC 40 oCIntensity (Arbitary Units)2 (Degrees)Dy0.10Ce0.90O2-(220) Figure C-4. XRD profiles of Dy0.10Ce0.90O2collected at high temperatures.
194 304050607080 33.233.4 Intensity (Arbitary Units)2 (Degrees) 600 oC 500 oC 400 oC 300 oC 200 oC 100 oC 25 oCIntensity (Arbitary Units)2 (Degrees)Gd0.10Ce0.90O2-(200) Figure C-5. XRD profiles of Gd0.10Ce0.90O2collected at high temperatures. 304050607080 28.829.0 Intensity (Arbitary Units)2 (Degrees) 500 oC 400 oC 300 oC 200 oC 100 oC 25 oCIntensity (Arbitary Units)2 (Degrees)Sm0.10Ce0.90O2-(111) Figure C-6. XRD profiles of Sm0.10Ce0.90O2collected at high temperatures.
195 APPENDIX D EXTRAPOLATION METHOD TO CALCULATE LATTICE PARAMETER The curved position sensitive ( CPS) diffr actometer (INEL, France) was used to collect X-ray diffraction profiles of doped ceria materials using Cu K radiation. Monochromator crystal was us ed to separate out Cu K1 from the incident X-ray beam. Peak positions in the XRD pattern were det ermined by fitting each individual peak with a symmetric Pearson VII profiles to model Cu K1 using a commercially available software (i.e., Solver add-in within Microsoft Exce l spreadsheet package). Extrapolation method was used to estimate the lattice parameter of doped ceria material s. This method can be used to correct for vertical sa mple displacements within cert ain limits. Thus, it is not essential to use internal standard to obtain the experimental offset in 2 position. Information regarding this technique can be found elsewhere.66 For a given XRD profile, lattice parameter (a) was determined for each indivi dual peak using the following relationship, 2222lkh sin a (D-1) The measured lattice parameters were th en plotted as a function of cos2 The scattered data was then linearly fitted using leastsquare algorithm. The fitted line is in the form of the following relationship, 2coskaao (D-2) where ao is the true estimation of the lattice parameter, and a is the apparent lattice parameter calculated from the angular position of a particular peak and k is the constant depending upon the sample vertical displacemen t. The implication of equation (D-2) is that the best estimate of lattice parameter, ao, can be obtained by plotting a with respect
196 to cos2, and extrapolating it to cos2 = 0. Figure D-1 shows apparent lattice parameter and cos2 plots for pure ceria material taken at different temperatures. It can be seen that with the increase in tem perature the intercept is also increasing which indicates the thermal lattice expansion of ceria. 0.30.40.50.184.108.40.206.0 5.30 5.32 5.34 5.36 5.38 5.40 5.42 40 oC 100 oC 200 oC 300 oC 400 oC 500 oC 600 oC Measured Lattice Parameter ( a) ()cos2Pure Ceria Figure D-1. Plot of calculated latti ce parameter, a with respect to cos2 for pure ceria at different temperatures.
197 LIST OF REFERENCES 1. Energy Information Admi nistration, 5th May 2008 ( http://www.eia.doe.gov) 2. Energy Information Administration (O fficial Energy Statistics from the U.S. Government), 5th May 2008 ( http://www.eia.doe.gov/environment.html) 3. National Renewable Energy Laboratory (NREL), 20th June 2008 (http://www.nrel.gov) 4. Fuel Cell Vehicles, 15th May 2008 ( http://www.fueleconomy.gov/feg /fuelcell.shtml) 5. N.Q. Minh, "Cer amic Fuel-Cells," J. Am. Ceram. Soc., 76  563-88 (1993). 6. N.Q. Minh, and T. Takahashi, Science and Technology of Ceramic Fuel Cells, Elsevier Science B. V., Amsterdam, 1995. 7. S. Omar, E.D. Wachsman, and J.C. Nino, "A co-doping approach towards enhanced ionic conductivity in fl uorite-based electrolytes," Solid State Ionics, 177 [35-36] 3199-203 (2006). 8. NETL: Solid State Energy Conv ersion Alliance (SECA), 15th May 2008 ( www.seca.doe.gov) 9. H. Inaba, and H. Tagaw a, "Ceria-Based Solid Electrolytes Review," Solid State Ionics, 83 [1-2] 1-16 (1996). 10. K. Eguchi, T. Setoguchi T. Inoue, and H. Arai, "Ele ctrical-Properties of CeriaBased Oxides and Their Application to Solid Oxide Fuel-Cells," Solid State Ionics, 52 [1-3] 165-72 (1992). 11. R.T. Dirstine, R.N. Bl umenthal, and T.F. Kuech, "Ioni c-Conductivity of Calcia, Yttria, and Rare Earth Doped Cerium Dioxide," J. Electrochem. Soc., 126  264-9 (1979). 12. V. Butler, C.R.A. Catl ow, B.E.F. Fender, and J.H. Ha rding, "Dopant Ion Radius and Ionic-Conductivity in Cerium Dioxide," Solid State Ionics, 8  109-13 (1983). 13. B.C.H. Steele, "Appraisal of Ce1 -xGdyO2x /2 Electrolytes for IT-SOFC Operation at 500o C," Solid State Ionics, 129 [1-4] 95-110 (2000). 14. J. Faber, C. Geoffroy, A. Roux, A. Sylvestre, and P. Abelard, "A Systematic Investigation of the DC-El ectrical Conductivity of Rare-Earth Doped Ceria," Appl. Phys. A: Mater. Sci. Process., 49  225 (1989).
198 15. X.Q. Sha, Z. Lu, X.Q. Huang, J.P. Miao, Z.G. Liu, X.S. Xin, Y.H. Zhang, and W.H. Su, "Influence of the Sintering Temperature on Elec trical Property of the Ce0.8Sm0.1Y0.1O1.9 electrolyte," J. Alloys Compd., 433 [1-2] 274-8 (2007). 16. J. Van Herle, D. S eneviratne, and A.J. McEvoy, "Lanthanide Co-Doping of Solid Electrolytes: AC Conductivity Behaviour," J. Eur. Ceram. Soc., 19 [6-7] 837-41 (1999). 17. N. Kim, B.H. Kim, and D. Lee, "Effe ct of Co-Dopant Addi tion on Properties of Gadolinia-Doped Ceria Electrolyte," J. Power Sources, 90  139-43 (2000). 18. F.Y. Wang, S.Y. Chen, W. Qin, S.X. Yu, and S.F. Cheng, "Study on Gd and Mg Co-Doped Ceria Electrolyte for Intermediat e Temperature Solid Oxide Fuel Cells," Catal. Today, 97 [2-3] 189-94 (2004). 19. X.Q. Sha, Z. Lu, X.Q. Huang, J.P. Miao, Z.H. Ding, X.S. Xin, and W.H. Su, "Study on La and Y Co-Doped Ceria-Bas ed Electrolyte Materials," J. Alloys Compd., 428 [1-2] 59-64 (2007). 20. X.Q. Sha, Z. Lu, X.Q. Huang, J.P. Miao, L. Jia, X.S. Xin, and W.H. Su, "Preparation and Properties of Rare Earth Co-Doped Ce0.8Sm0.2xYxO1.9 Electrolyte Materials for SOFC," J. Alloys Compd., 424 [1-2] 315-21 (2006). 21. J.A. Kilner, and R.J. Brook, "A St udy of Oxygen Ion Conductivity in Doped Nonstoichiometric Oxides," Solid State Ionics, 6  237-52 (1982). 22. D.J. Kim, "Lattice-Par ameters, Ionic Conductiviti es, and Solubility Limits in Fluorite-Structure MO2 Oxide (M = Hf4+, Zr4+, Ce4+, Th4+, U4+) Solid-Solutions," J. Am. Ceram. Soc., 72  1415-21 (1989). 23. P.J. Gellings, and H.J.M. Bouwmeester, The CRC Handbook of Solid State Electrochemistry, CRC Press Inc, Boca Raton, 1996. 24. M.W. Barsoum, Fundamentals of Cera mics, McGraw Hill, Bristol, 1997. 25. A.J. Moulson, and J.M. Herbert, Electroceramics: Materials, Properties, and Applications, Wiley, Chichester, 2003. 26. A.M. Adams, F.T. Bac on, and R.G.T. Watson, Fuel Cells, Academic Press, New York, 1963. 27. B.C.H. Steele, "Material Science and Engineering: The Enabling Technology for the Commercialisation of Fuel Cell Systems," J. Mater. Sci., 36  1053-68 (2001). 28. C. Berger, Handbook of Fuel Cell Technology, Prentice-Hall Inc., New Jersey, 1968. 29. A.J. Appleby, Fuel Cell Handbook, Van Nonstrand Reinhold, New York, 1989.
199 30. L.J.N. Blomen, and M.N. Mugerwa, Fuel Cell S ystems, Plenum, New York, 1993. 31. A.J. Appleby, "Issues in Fuel Cell Commercialization," J. Power Sources, 58  153-76 (1996). 32. H.H. Mbius, On the History of Solid El ectrolyte Fuel Cells," J. Solid State Electrochem., 1 [July] 2-16 (1997). 33. E. Baur, and H. Prei s, "On Fuel Chains with Fixed Cables," Z. Elektro. Angew. Phys. Chem., 43 727-32 (1937). 34. L. Blum, H.P. Buchkremer, S. Gross, A. Gubner, L.G. J. de Haart, H. Nabielek, W.J. Quadakkers, U. Reis gen, M.J. Smit h, R. Steinberger-Wilckens, R.W. Steinbrech, F. Tietz, and I.C. Vinke, "S olid Oxide Fuel Cell Development at Forschungszentrum Juelich," Fuel Cells, 7  204-10 (2007). 35. M. Sahibzada, B.C.H. Steele, K. Hellgar dt, D. Barth, A. Effendi, D. Mantzavinos, and I.S. Metcalfe, "Intermediat e Temperature Solid Oxide F uel Cells Operated with Methanol Fuels," Chem. Eng. Sci., 55  3077-83 (2000). 36. S.C. Singhal and S. Gopalan, Processing and Characterization of Electrochemical Materials and Devices. The American Ceramic Society, Ceramic Transactions: Westerville, 1996; Vol. 109. 37. C. Wagner, "The Mechanism of Elec tric Supply Lines in Neon Lamps," Naturwissenschaften, 31 265-8 (1943). 38. T.H. Etsell, and S.N. Flengas, "Electrical Properties of Solid Oxide Electrolytes," Chem. Rev., 70  339-76 (1970). 39. H. Yokokawa, N. Sakai, T. Horita, K. Yamaji, and M.E. Brit o, "Electrolytes for Solid-Oxide Fuel Cells," Mater. Res. Soc. Bull., 30  591-5 (2005). 40. N.X. Jiang, and E.D. Wachsman, "Structural Stabi lity and Conductivity of PhaseStabilized Cubic Bismuth Oxides," J. Am. Ceram. Soc., 82  3057-64 (1999). 41. V.V. Kharton, F. M.B. Marques, and A. Atkinson, "Transport Properties of Solid Oxide Electrolyte Ceramics: A Brief Review," Solid State Ionics, 174 [1-4] 135-49 (2004). 42. X.D. Wu, Q. Liang, D. Weng, J. Fan, and R. R an, "Synthesis of CeO2-MnOx Mixed Oxides and Catalytic Performanc e Under Oxygen-Rich Condition," Catal. Today, 126 [3-4] 430-5 (2007). 43. S.M. Mathew, S.B. Umbarkar, and M.K. Dongare, "NOx Storage Behavior of BaO in Different Structural Envi ronment in NSR Catalysts," Catal. Commun., 8  117882 (2007).
200 44. Z.I. Onsan, "Catalytic Proc esses for Clean Hydrogen Production from Hydrocarbons," Turk. J. Chem., 31  531-50 (2007). 45. M.I. Soumelidis, R.K. Stobart, and R.A. Jackson, "A Chemically Informed, ControlOriented Model of a ThreeWay Catalytic Converter," Proceedings of the I Mech E Part D Journal of Automobile Engineering, 221 [D9] 1169-82 (2007). 46. S. Yabe, and T. Sato, "Cerium Oxide for Sunscreen Cosmetics," J. Solid State Chem., 171 [1-2] 7-11 (2003). 47. JCPDS International Committee for Diffraction Da ta Card 34 0394. In. 48. R.D. Shannon, "Revised Effective Ionic-Radii and Systematic Studies of Interatomic Distances in Halides and Chalcogenides," Acta Crystallogr., Sect. A, 32 [Sep1] 751-67 (1976). 49. B.C.H. Steele, Ceramic Oxygen Ion Conductors and Their Technological Applications. The Institute of Materials: London, 1996. 50. T. Kudo, and K. Fueki, Solid St ate Ionics, Kodansha/ VCH, Tokyo, 1990. 51. T. Takahashi, Physics of Electr olytes, Academic Pr ess, London, 1972. 52. D.M. Smyth, The Defect Chemistry of Me tal Oxides, Oxford Univ ersity Press, Inc., New York, 2000. 53. S. Omar, E.D. Wachsman, and J. C. Nino, "Higher Conductivity Sm3+ and Nd3+ CoDoped Ceria-Based Electrolyte Materials," Solid State Ionics, 178 1890-7 (2008). 54. G.B. Balazs, and R.S. Glass, "ACImpedance Studies of Rare-Earth-Oxide Doped Ceria," Solid State Ionics, 76 [1-2] 155-62 (1995). 55. J.R. Macdonald, and E. Barsoukov, Impedance Spectroscopy: Theory, Experiment, and Applications, John Wiley & Sons, New York, 1987. 56. H. Yahiro, K. Eguchi, and H. Arai, "Ele ctrical-Properties and Reducibilities of Ceria Rare Earth Oxide Systems and Their Appl ication to Solid Oxide Fuel-Cell," Solid State Ionics, 36 [1-2] 71-5 (1989). 57. H. Yahiro, T. Ohuchi, K. Eguchi and H. Arai, "Electrical-Properties and Microstructure in the System Ceria Alkaline-Earth Oxide," J. Mater. Sci., 23  1036-41 (1988). 58. I. Mahata, G. Das, R. K. Mishra, and B.P. Sharma "Combustion Synthesis of Gadolinia Doped Ceria Powder," J. Alloys Compd., 391 [1-2] 129-35 (2005). 59. N.M. Rahaman, Ceramic Processing, CRC Taylor & Francis, Boca Raton, 2007.
201 60. A. Tsoga, A. Naoumidis, and D. Stover, "Total Electrical Conductivity and Defect Structure of ZrO2-CeO2-Y2O3-Gd2O3 Solid Solutions," Solid State Ionics, 135 [1-4] 403-9 (2000). 61. K.R. Reddy, and K. Karan, "Sinterability, Mechani cal, Microstructural, and Electrical Properties of Gadolinium-D oped Ceria Electrolyte for Low-Temperature Solid," J. Electroceram., 15  45-56 (2005). 62. C.Y. Tian, and S.W. C han, "Ionic Conductivities, Si ntering Temperatures and Microstructures of Bulk Ceramic CeO2 Doped with Y2O3," Solid State Ionics, 134 [1-2] 89-102 (2000). 63. R.S. Torrens, N.M. Sammes, and G.A. Tompsett, "Characterisation of (CeO2)0.8(GdO1.5)0.2 Synthesised Using Various Techniques," Solid State Ionics, 111 [1-2] 9-15 (1998). 64. S.W. Zha, C.R. Xia, and G.Y. Meng, "E ffect of Gd (Sm) D oping on Properties of Ceria Electrolyte for Solid Oxide Fuel Cells," J. Power Sources, 115  44-8 (2003). 65. C. Keller, U. Berndt, H. Engerer, and L. Leitner, "Phase Equilibria in Thorium Oxide Lanthanide Oxide Systems," J. Solid State Chem., 4  453-65 (1972). 66. A. Pramanick, S. Omar, J.C. Nino, and J.L. Jones, "Lattice Parameter Determination using a Curved Position S ensitive Detector and Application to SmxNdxCe1-2 xO2Ceramics," unpublished (2007). 67. S. Omar, E.D. Wachsman, and J.C. Nino, "Higher Ionic Conductive Ceria Based Electrolytes for SOFCs," Appl. Phys. Lett., 91 144106 (2007). 68. L. Minervini, M.O. Zacate, and R.W. Grimes, "Defect Cluster Formation in M2O3Doped CeO2," Solid State Ionics, 116 [3-4] 339-49 (1999). 69. H. Yamamura, E. Katoh, M. Ichika wa, K. Kakinuma, T. Mori, and H. Haneda, "Multiple Doping Effect on the Electrical Conductivity in the (Ce1-x-yLaxMy)O2-delta (M = Ca, Sr) System," Electrochemistry, 68  455-9 (2000). 70. E.D. Wachsman, S. Boyapati, and N. Jiang, "Effect of Dopant Polarizability on Oxygen Sublattice Order in PhaseStabilized Cubic Bismuth Oxides," Ionics, 7 1-6 (2001). 71. F.Y. Wang, B.Z. Wan, and S.F. Cheng, "Study on Gd3+ and Sm3+ Co-Doped CeriaBased Electrolytes," J. Solid State Electrochem., 9  168-73 (2005). 72. P. Li, I.W. Chen, J.E. Pennerhahn, and T.Y. Tien, "X -Ray Absorption Studies of Ceria with Trivalent Dopants," J. Am. Ceram. Soc., 74  958-67 (1991).
202 73. M. Mogensen, N.M. Sa mmes, and G.A. Tompsett, "Physical, Chemical and Electrochemical Properties of Pure and Doped Ceria," Solid State Ionics, 129 [1-4] 63-94 (2000). 74. S.J. Hong, and A.V. Virkar, "Lattice-Parameters and Densities of Rare-Earth-Oxide Doped Ceria Electrolytes," J. Am. Ceram. Soc., 78  433-9 (1995). 75. K. Fuda, K. Kishio, S. Yamachi, and K. Fueki, "Study on Vacancy Motion in Y2O3Doped CeO2 by O-17 NMR Technique," J. Phys. Chem. Solids, 46  1141-6 (1985). 76. Z.L. Zhan, T.L. Wen, H.Y. Tu, and Z.Y. Lu, "AC Impedance Investigation of Samarium-Doped Ceria," J. Electrochem. Soc., 148  A427-A32 (2001). 77. T.S. Zhang, J. Ma, L.B. Kong, S.H. Chan, and J.A. Kilner, "Aging Behavior and Ionic Conductivity of Ceria-Base d Ceramics: A Comparative Study," Solid State Ionics, 170 [3-4] 209-17 (2004). 78. L.P. Li, X.M. Lin, G. S. Li, and H. Inomata, "S olid Solubility and Transport Properties of Ce1xNdxO2Nanocrystalline Solid Solutions by a Sol-Gel Route," J. Mater. Res., 16  3207-13 (2001). 79. Y. Tsur, and C.A. Randall, "Analysis of X-Ray Powder Diffraction Data Using the Maximum Likelihood Estimation Method," J. Am. Ceram. Soc., 83  2062-6 (2000). 80. I.E.L. Stephens, and J.A. Kilner "Ionic Conductivity of Ce1xNdxO2x /2," Solid State Ionics, 177 [7-8] 669-76 (2006). 81. A.E. Mattsson, P.A. Schultz, M.P. Desjarlais, T.R. Mattsson, and K. Leung, "Designing meaningful density f unctional theory calculations in materials science a primer," Modelling and Simulation in Materials Science and Engineering, 13  R1-R31 (2005). 82. W. Kohn, and L.J. Sham, "Self-C onsistent Equations Including Exchange and Correlation Effects," Physical Review, 140 [4A] 1133-& (1965). 83. D.A. Andersson, S.I. Simak, N.V. Skorodumova, I.A. Abrikosov, and B. Johansson, "Optimization of Ioni c Conductivity in Doped Ceria," Proc. Natl. Acad. Sci. U. S. A., 103 3518 (2006). 84. L. Vegard, and H. Dale, "Tes ts on Mixed Crystals and Alloys," Z. Kristallogr., 67  148-62 (1928). 85. T.C. Parks, and D.J.M. Be van, "Mixed Oxides of Type MO2 (Fluorite)-MO1.5 VI : Phase Relationships in Systems CeO2-SmO1.5 and CeO2-NdO1.5," Rev. Chim. Miner., 10 [1-2] 15-28 (1973).
203 86. J.E. Bauerle, "Study of Solid Electr olyte Polarization by a Complex Admittance Method," J. Phys. Chem. Solids, 30  2657-70 (1969). 87. A.S. Nowick, W.K. Le e, and H. Jain, "Survey and Interpretation of PreExponentials of Conductivity," Solid State Ionics, 28 89-94 (1988). 88. D.K. Hohnke, "Ionic-C onduction in Doped Oxides with the Fluorite Structure," Solid State Ionics, 5 [Oct] 531-4 (1981). 89. D.P. Almond, and A.R. West, "The Activation Entropy for Transport in Ionic Conductors," Solid State Ionics, 23 [1-2] 27-35 (1987). 90. K. Fuda, K. Kishio, S. Yamauchi, K. Fueki, and Y. Onoda, "O-17 NMR-Study of Y2O3-Doped CeO2," J. Phys. Chem. Solids, 45 [11-1] 1253-7 (1984). 91. J.M. Ralph, and J.A. Kilner, Enhancement of the ionic conductivity in ceria gadolinia solid solutions. The European Solid Oxide Fuel Cell Forum: Oslo, 1996. 92. V.B. Glushkova, L. V. Sazonova, and F. Hanic, "Latti ce Parameters of Cubic Solid Solutions in the System uR2O3-(1-u)MO2," Ceram. Int., 4  176-8 (1978). 93. V.I. Aleksandrov, G.E. Valyano, B.V. Lukin, V.V. Osiko, A.E. Rautbort, V.M. Tatarintsev, and V.N. Filatova, "Structu re of Single-Crystals of Stabilized Zirconium Dioxide," Inorg. Mater., 12  235-8 (1976). 94. R.P. Ingel, and D. Lewis, "Lattice-Parameters and Density for Y2O3-Stabilized ZrO2," J. Am. Ceram. Soc., 69  325-32 (1986). 95. B.D. Cullity, and S.R. Stock, Elements of X-ray Diffraction, Prentice Hall, New Jersey, 3rd1978. 96. T. Hisashige, Y. Ya mamura, and T. Tsuji, "Ther mal Expansion and Debye temperature of Rare Earth-Doped Ceria," J. Alloys Compd., 408 1153-6 (2006). 97. H. Hayashi, M. Kanoh, C.J. Quan, H. I naba, S.R. Wang, M. Dokiya, and H. Tagawa, "Thermal Expansion of Gd-Doped Ceria and Reduced Ceria," Solid State Ionics, 132 [3-4] 227-33 (2000). 98. W. Zajac, K. Swiercze k, and J. Molenda, "Thermochemical Compatibility Between Selected (La,Sr)(Co,Fe,Ni)O3 Cathodes and Rare Earth Doped Ceria Electrolytes," J. Power Sources, 173  675-80 (2007). 99. H. Hayashi, T. Saitou, N. Maruya ma, H. Inaba, K. Kawamura, and M. Mori, "Thermal Expansion Coefficient of Yttria Stabilized Zirconia for Various Yttria Contents," Solid State Ionics, 176 [5-6] 613-9 (2005).
204 100. X.L. Zhou, M.X. Zhu, F.J. Deng, G.Y. Meng, and X.Q. Liu, "Electrical Properties, Sintering and Thermal Expansion Behavior of Composite Ceramic Interconnecting Materials, La0.7Ca0.3CrO3d / Y0.2Ce0.8O1.9 for SOFCs," Acta Mater., 55  2113-8 (2007). 101. N. Swaminathan, and J. Qu, "Interactions Between N on-Stoichiometric Stresses and Defect Transport in a Tubular Electrolyte," Fuel Cells, 7  453-62 (2007). 102. M. Greenberg, E. Wachtel, I. Lubomirsky, J. Fleig, and J. Maier, "Elasticity of Solids with a Large Concentra tion of Point Defects," Adv. Funct. Mater., 16  4852 (2006). 103. D.Y. Wang, D.S. Park, J. Griffith, and A.S. Nowi ck, "Oxygen-Ion Conductivity and Defect Interactions in Yttria-Doped Ceria," Solid State Ionics, 2  95-105 (1981). 104. M. Kamiya, E. Shimada, Y. Ikuma, M. Komatsu, H. Haneda, S. Sameshima, and Y. Hirata, "Oxygen Self-Diffusion in Cerium Oxide Doped with Nd," J. Mater. Res., 16  179-84 (2001). 105. A.S. Nowick, D. Y. Wang, D.S. Park and J. Griffith, Oxygen-Ion Conductivity and Defect Structure of CeO2 Doped with CaO and Y2O3. North Holland: Amsterdam, 1979. 106. J.A. Kilner, and B.C.H. Steele, Nonstoichiometric Oxides. Academic Press: New York, 1981. 107. T.S. Zhang, J. Ma, S.H. Chan, P. Hing, and J.A. Kilner "Intermediate-Temperature Ionic Conductivity of Ceri a-Based Solid Solutions as a Function of Gadolinia and Silica Contents," Solid State Sci., 6  565-72 (2004). 108. Y.H. Cho, P.S. C ho, G. Auchterlonie, D. K. Kim, J.H. Lee, D. Y. Kim, H.M. Park, and J. Drennan, "Enhancement of Grain-Boundary Conduction in Gadolinia-Doped Ceria by the Scavenging of Highl y Resistive Siliceous Phase," Acta Mater., 55  4807-15 (2007). 109. T.S. Zhang, J. Ma, L.B. Kong, P. Hing, Y. J. Leng, S.H. Chan, and J.A. Kilner, "Sinterability and Ionic Conductivity of Co-Precipitated Ce0.8Gd0.2O2-delta Powders Treated via a High-Energy Ball-Milling Process," J. Power Sources, 124  26-33 (2003). 110. M.C. Martin, and M.L. Me cartney, "Grain Boundary Ioni c Conductivity of Yttrium Stabilized Zirconia as a Function of Silica Content and Grain Size," Solid State Ionics, 161 [1-2] 67-79 (2003). 111. T.S. Zhang, J. Ma, Y.Z. Chen, L.H. Luo, L. B. Kong, and S.H. Chan, "Different conduction behaviors of grain boundaries in SiO2-containing 8YSZ and CGO20 electrolytes," Solid State Ionics, 177 [13-14] 1227-35 (2006).
205 112. R. Gerhardt, and A.S. Nowick, "Grain-Boundary Effe ct in Ceria Doped with Trivalent Cations .1. Electrical Measurements," J. Am. Ceram. Soc., 69  641-6 (1986). 113. R. Gerhardt, A.S. Nowi ck, M.E. Mochel, and I. Dumler "Grain-Boundary Effect in Ceria Doped with Trivalent Cations .2. Microstructure and Microanalysis," J. Am. Ceram. Soc., 69  647-51 (1986). 114. V. Esposito, and E. Traver sa, "Design of Electroceramics for Solid Oxide Fuel Cell Applications: Playing with Ceria," J. Am. Ceram. Soc., 91  1037-51 (2008). 115. X. Guo, and R. Waser, "Electrical Properties of the Grain Boundaries of Oxygen Ion Conductors: Acceptor-Doped Zirconia and Ceria," Prog. Mater. Sci., 51  151-210 (2006). 116. X. Guo, and Y. Ding, "Grain boundary space charge effect in zirconia Experimental evidence," J. Electrochem. Soc., 151  J1-J7 (2004). 117. C. Herring, "Effect of Change of Scale on Sintering Phenomena," J. Appl. Phys., 21  301-3 (1950). 118. Y.C. Zhou, and M. N. Rahaman, "Effect of Redo x Reaction on the Sintering Behavior of Cerium Oxide," Acta Mater., 45  3635-9 (1997). 119. J.D. Nicholas, and L.C. De Jonghe, "Prediction and evaluat ion of sintering aids for Cerium Gadolinium Oxide," Solid State Ionics, 178 [19-20] 1187-94 (2007). 120. G.S. Lewis, A. Atkinson, B.C.H. Steele, and J. Drennan, "Effect of Co addition on the lattice parameter, electrical conductivity and sintering of gadolinia-doped ceria," Solid State Ionics, 152 567-73 (2002). 121. E. Jud, Z. Zhang, W. Sigle, and L.J. Gauckler, "Microstructu re of Cobalt Oxide Doped Sintered Ceria Solid Solutions," J. Electroceram., 16  191-7 (2006). 122. V. Gil, J. Tart aj, C. Moure, and P. Duran, "Effect of Bi2O3 Addition on the Sintering and Microstructural Development of Gadolinia-Doped Ceria Ceramics," J. Eur. Ceram. Soc., 27 [2-3] 801-5 (2007). 123. K.H. Brosnan, G.L. Messing, and D.K. Agrawal, "Microwave Sintering of Alumina at 2.45 GHz," J. Am. Ceram. Soc., 86  1307-12 (2003). 124. M.A. Janney, and H.D. Kimrey, Microwave Sintering of Alumina at 28 GHz. American Ceramic Society: Westerville, 1988. 125. B. Rambabu, S. Ghosh, and H. Jena, "Novel Wet-Chemical Synthesis and Characterization of Nanocrystalline CeO2 and Ce0.8Gd0.2O1.9 as Solid Electrolyte for Intermediate Temperature Solid Oxi de Fuel Cell (IT-SOFC) Applications," J. Mater. Sci., 41  7530-6 (2006).
206 126. D. Perez-Coll, D. Marrero-Lopez, P. Nunez, S. Pinol, and J.R. Frade, "Grain Boundary Conductivity of Ce0.8Ln0.2O2d Ceramics (Ln = Y, La, Gd, Sm) with and without Co-Doping," Electrochim. Acta, 51  6463-9 (2006). 127. J.S. Ahn, H.S. Yoon, and E.D. Wachsman, "Perfo rmance of IT-SOFC with Ce0.9Gd0.1O1.95 Functional Layer at the Interface of Ce0.9Gd0.1O1.95 Electrolyte and Ni-Ce0.9Gd0.1O1.95 Anode," Proc. Electrochem. Soc., Submitted (2008). 128. T. Kenjo, S. Osawa, and K. Fujikawa "High-Temperature Air Cathodes Containing Ion Conductive Oxides," J. Electrochem. Soc., 138  349-55 (1991). 129. E.P. Murray, and S.A. Barnett, "(La,Sr)MnO3-(Ce,Gd)O2x Composite Cathodes for Solid Oxide Fuel Cells," Solid State Ionics, 143 [3-4] 265-73 (2001).
207 BIOGRAPHICAL SKETCH Shobit Omar was born in 19th of October 1980, in Jhansi, Uttar Pradesh, India. His father is an electrical engineer, and ran a small scale industry in Jhansi. His mother is a homemaker, while his elder brother is an electrical eng ineer and works in a software company in Pune, In dia. Following the footsteps of his father and brother, Shobit Omar always dream ed of becoming an engineer. After finishing the 12th grade in 1999, and spending one y ear of completely secluded li fe, he was able to get his name in the merit list for studying in one of the IITs. However, he decided to take up ceramic engineering at the In stitute of Technology, Banaras Hindu University (IT-BHU), Varanasi. Spending four year s in the holy city of Va ranasi completely changed his perception about life. Living in the cultural and religious centre of India transformed him into a spiritual person. After graduation, he decided to go for advanced studies in materials science in the US, rather than going for a job directly after graduation. In summer 2004, he obtained a bachelors degr ee from IT-BHU, and got admission in Materials Science and Engineering Department at the University of Florida (UF), Gainesville, Florida. At UF, Shobit Omar joined Dr. Juan C. Ninos research gr oup, where he spent a lonely life in the research laboratory for the first semester. He missed his family, friends, Indian food, and everything related to India. Slowly things changed, and he got some really good friends in Gainesville. As time passed, he got accustomed to US culture. After four years of his research, he is now near completion of his PhD. Throughout this period, he was exposed to numerous new things, which helped him to grow professionally and individually. Studying in Univer sity of Florida provided him the excellent environment and opportunity to build a strong career in the materials science
208 field. Shobit Omar has accepted a post-docto ral position in Ris National Laboratory in Roskilde, Denmark upon completion of his Ph.D. degree.