Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Proton conductive ceramic materials for an intermediate temperature fuel cell Jankovic, Jasna 2011

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2011_fall_jankovic_jasna.pdf [ 7.89MB ]
Metadata
JSON: 24-1.0059159.json
JSON-LD: 24-1.0059159-ld.json
RDF/XML (Pretty): 24-1.0059159-rdf.xml
RDF/JSON: 24-1.0059159-rdf.json
Turtle: 24-1.0059159-turtle.txt
N-Triples: 24-1.0059159-rdf-ntriples.txt
Original Record: 24-1.0059159-source.json
Full Text
24-1.0059159-fulltext.txt
Citation
24-1.0059159.ris

Full Text

PROTON CONDUCTIVE CERAMIC MATERIALS FOR AN INTERMEDIATE TEMPERATURE PROTON EXCHANGE FUEL CELL  by  Jasna Jankovic  B.A.Sc., The University of Belgrade, 1996 M.A.Sc., The University of British Columbia, 2005   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Chemical and Biological Engineering) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   June 2011 © Jasna Jankovic, 2011  ii ABSTRACT Development of intermediate temperature (200-500oC) fuel cells could possibly overcome many disadvantages of both the high temperature (600-1000°C) solid oxide fuel cells (SOFC) and the low temperature (70-100°C) proton exchange membrane fuel cells (PEMFC) in terms of materials durability, cost, application, and overall system structure. A change in materials, especially the proton conductive electrolyte, is required to achieve this. However, to date, no solid proton conductors have been developed that work satisfactorily in this temperature range. The goal of this thesis was to develop a ceramic proton-conducting material to be used as a dense electrolyte, as well as within the anode structure of an intermediate temperature fuel cell. Investigated ceramic materials were based on oxygen deficient ceramic oxides – undoped and Ce- and La-doped Ba2In2O5, which were expected to show proton conductivity within the intermediate temperature range due to water and/or proton incorporation into their defect structure. Five different compositions of brownmillerite materials, Ba2In2-x- yCexLayO5+x/2 (x=0.25 and 0.5; y=0.25 and 0.5) were synthesized via the solid-state reaction and the glycine-nitrate process, characterized and electrochemically investigated in order to find a suitable proton-conductive electrolyte. The materials were characterized using X-ray powder diffraction (XRD), thermogravimetric analysis (TGA), differential scanning calorimetry (DSC), particle size analysis (PSA), scanning electron microscopy (SEM), transmission electron microscopy (TEM), etc. The electrical conductivities of the ceramics were determined using ac impedance spectroscopy. Among the tested materials, undoped Ba2In2O5 produced by the glycine-nitrate process was selected as the material with the highest total conductivity (between 0.02 S/cm and 0.7 S/cm) and stability in hydrogen-  iii containing atmospheres and at temperatures between 300oC and 480oC. High proton transport numbers (e.g., 0.84 at 300oC) and relatively high open circuit voltage values of the air, Pt║Ba2In2O5║Pt, 50%H2/50%N2 cell (e.g., 0.81 V at 300oC) confirmed the predominant proton conductivity of this material. Although highly proton conductive in a hydrogen- containing atmosphere, Ba2In2O5 showed poor performance as an electrolyte in an intermediate temperature fuel cell due to the incorporation of oxygen on the cathode side with associated blocking of the proton conduction. Application of sintered porous Ba2In2O5 in a cermet with a metal catalyst in the anode structure was shown to be beneficial.               iv PREFACE This thesis work was performed by Jasna Jankovic, including the literature review, research proposal preparation, experimental design, performing of the experiments and data analysis, under supervision of Dr. David Wilkinson from the University of British Columbia and Dr. Rob Hui, from the National Research Council ─ Institute for Fuel Cell Innovation. The research resulted in three manuscripts, prepared by Jasna Jankovic and revised and edited by Dr. Wilkinson and Dr. Hui. Jasna Jankovic, Dr. David Wilkinson and Dr. Rob Hui are listed as the authors of the following manuscripts: • The results of the study discussed in Chapter 4 and 5 have been submitted as follows: J. Jankovic, D.P. Wilkinson, R. Hui, (2011) Preparation and Characterization of Ce- and La-doped Ba2In2O5 as Candidates for Intermediate Temperature (200-500oC) Solid Proton Conductor (submitted). • A version of the results in Chapter 6 has been submitted as follows:   J. Jankovic, D.P. Wilkinson, R. Hui, (2011) Impedance Comparison Study of Undoped and Ce-doped Ba2In2O5 (submitted). • The material presented in Chapter 7 has been published as follows: J. Jankovic, D.P. Wilkinson, R. Hui, (2011) Proton Conductivity and Stability of Ba2In2O5 in Hydrogen Containing Atmospheres, Journal of the Electrochemical Society, 158 (1) B61-B68.    v Research results from this thesis work have been presented by Jasna Jankovic at the following conferences: • Jankovic, J., Wilkinson, D.P., Hui R. (2009) High Conductivity of Nanocrystalline Ba2In2O5 in Hydrogen Containing Atmospheres at Intermediate Temperatures. Paper presented at the 17th International Conference on Solid State Ionics, Toronto. • Jankovic, J., Wilkinson, D.P., Hui R. (2008) Electrical Conductivity Investigation of Micro- and Nano- Crystalline Ce- and La-doped Ba2In2O5 in Air and Hydrogen Atmosphere. Paper presented at the Solid State Protonic Conductors-14 conference, Japan. • Jankovic, J., Wilkinson, D.P., Hui R. (2007) Synthesis and Characterization of Proton- Conducting Ceramic Materials for a High Temperature PEMFC. Paper presented at the Hydrogen and Fuel Cell Conference, Vancouver. • Jankovic, J., Wilkinson, D.P., Hui R. (2007) Proton-Conducting Ceramic Materials for a High Temperature PEMFC. Paper presented at the 58th Meeting of the International Society of Electrochemistry, Banff, Canada. One patent application has been submitted: • Wilkinson, D.P., Jankovic, J., Maric, R., Hui, R., Ghosh, D., (2006) New Approaches to the Gas Diffusion Electrode and Electro-catalyst Layer for High Temperature Proton Conducting Fuel Cells.       vi TABLE OF CONTENTS  ABSTRACT ........................................................................................................ii PREFACE .......................................................................................................... iv TABLE OF CONTENTS......................................................................................vi LIST OF TABLES .............................................................................................xii LIST OF FIGURES ........................................................................................... xiv ACKNOWLEDGEMENTS ...............................................................................xxiv 1. Introduction .................................................................................................1 1.1 Motivation for the thesis project ........................................................................1 1.2 Overall approach of the thesis............................................................................2 1.3 Thesis layout .......................................................................................................4 2. Background, Literature Review and Thesis Objectives.............................6 2.1 Fuel cells in general.............................................................................................6 2.1.1 Introduction and history...............................................................................6 2.1.2 Design of fuel cells......................................................................................7 2.1.3 Reactions and thermodynamics .................................................................10 2.1.4 Operational fuel cell voltages ....................................................................17 2.2 Intermediate temperature fuel cells .................................................................19 2.3 Proton conducting ceramic oxides....................................................................22 2.3.1 Electrical conductivity in ceramic oxides...................................................24 2.3.2 Crystal structure of acceptor-doped perovskites and brownmillerite Ba2In2O5 ..........................................................................................................................27  vii 2.3.3 Oxygen ion, proton and electron conductivity of acceptor-doped perovskites and brownmillerite Ba2In2O5 ..............................................................................29 2.3.4 Effect of transition order-disorder temperature on conductivity and possible improvement of conductivity..............................................................................43 2.3.5 Stability of Ba2In2O5 .................................................................................44 2.4 Research opportunity for proton conductivity in Ba2In2O5 and related materials.................................................................................................................45 2.5 Thesis objectives ...............................................................................................47 3. Experimental Approach ............................................................................ 48 3.1 Synopsis.............................................................................................................48 3.2 Powder preparation ..........................................................................................49 3.3 Materials characterization ...............................................................................53 3.3.1 X-ray powder diffraction ...........................................................................53 3.3.2 Particle size analysis..................................................................................53 3.3.3 Thermogravimetric analysis and differential scanning calorimetry.............54 3.3.4 Temperature-profile X-ray diffraction .......................................................54 3.3.5 Stability in humid atmospheres..................................................................55 3.4 Preparation of samples for conductivity and e.m.f. measurements ................56 3.5 Ac impedance spectroscopy and conductivity measurements.........................60 3.6 Stability in hydrogen-containing atmospheres ................................................63 3.7 Proton transport number determination .........................................................63 3.8 Open circuit voltage, polarization curve and potentiostatic measurements ...66 3.9 Preparation of testing samples for anode study...............................................68  viii 4. Material Characterization of Undoped and Ce- and La-doped Ba2In2O5 ......................................................................................................................... 72 4.1 Synopsis.............................................................................................................72 4.2 Crystal phase of the materials ..........................................................................72 4.3 Powder particle characterization .....................................................................75 4.4 Temperature-profile XRD................................................................................78 4.5 Thermogravimetric analysis and DSC.............................................................81 4.6 Stability in humid atmospheres and water ......................................................88 4.7 Summary ...........................................................................................................92 5. Electrical Conductivity of the Materials................................................... 94 5.1 Synopsis.............................................................................................................94 5.2 Ac impedance measurements for conductivity determination........................95 5.2.1 Analysis of electrode contribution .............................................................97 5.2.2 Ac impedance results and conductivity determination................................99 5.3 Electrical conductivity results ........................................................................102 5.3.1 Comparison to the literature ....................................................................102 5.3.2 Conductivity of the solid-state samples in air and a hydrogen-containing atmosphere.......................................................................................................103 5.3.3 Conductivity of the GNP samples in air and a hydrogen-containing atmosphere.......................................................................................................106 5.3.4 Effect of microstructure...........................................................................107 5.4 Summary .........................................................................................................108  ix 6. Electrochemical Impedance Spectroscopy of Ba2In2O5 – Effect of Porosity, Grain Size, Dopant, Atmosphere and Temperature .................. 111 6.1 Synopsis...........................................................................................................111 6.2 Samples for the study......................................................................................112 6.3 Effect of porosity on conductivity ..................................................................114 6.4 Ac impedance of BIO-GNP and BIO-SS in air..............................................116 6.4.1 Effect of grain size ..................................................................................116 6.4.2 Effect of water loss..................................................................................118 6.4.3 Equivalent circuit fit ................................................................................118 6.5 Ac impedance of BIC-GNP and BIC-SS in air ..............................................120 6.5.1 Effect of dopant.......................................................................................120 6.5.2 Effect of water loss..................................................................................123 6.5.3 Change of phase ......................................................................................123 6.5.4 Equivalent circuit fit ................................................................................124 6.6 Ac impedance of BIO-GNP and BIO-SS in 50% H2/50% N2 ........................125 6.6.1 Effect of grain size ..................................................................................125 6.6.2 Effect of water loss..................................................................................129 6.6.3 Decomposition at 500oC..........................................................................132 6.6.4 Equivalent circuit fit ................................................................................133 6.7 Ac impedance of BIC-GNP and BIC-SS in 50% H2 /50% N2........................134 6.7.1 Effect of dopant.......................................................................................134 6.7.2 Effect of water loss..................................................................................137 6.7.3 Change of phase ......................................................................................137  x 6.7.4 Equivalent circuit fit ................................................................................137 6.8 Conductivity comparison................................................................................138 6.9 Summary .........................................................................................................141 7. Proton Conductivity of Ba2In2O5 and Stability in Hydrogen-containing Atmospheres................................................................................................. 144 7.1 Synopsis...........................................................................................................144 7.2 Conductivity of Ba2In2O5 in air, nitrogen and hydrogen ..............................145 7.3 Proton conductivity of Ba2In2O5 ....................................................................149 7.3.1 Proton transport number by e.m.f. method ...............................................151 7.4 Stability of Ba2In2O5 in hydrogen containing atmospheres...........................153 7.5 Summary .........................................................................................................157 8. Evaluation of Ba2In2O5 for Use as an Electrolyte Material or Within the Anode in an Intermediate Temperature Fuel Cell ..................................... 159 8.1 Synopsis...........................................................................................................159 8.2 Evaluation of Ba2In2O5 as the electrolyte.......................................................159 8.2.1 OCV measurements.................................................................................159 8.2.2 Polarization curves ..................................................................................160 8.2.3 Potentiostatic measurements....................................................................164 8.3 Evaluation of Ba2In2O5 within the anode structure.......................................167 8.4 Summary .........................................................................................................172 9. Conclusions............................................................................................... 174 9.1 Materials preparation and characterization..................................................175 9.2 Electrical conductivity of the investigated materials .....................................176  xi 9.3 Proton conductivity and stability of Ba2In2O5 in hydrogen-containing atmospheres .........................................................................................................177 9.4 Evaluation of Ba2In2O5 for use as an electrolyte material or within the anode in an intermediate temperature fuel cell .............................................................179 9.5 Research significance......................................................................................180 9.6 Future work and recommendations ...............................................................181 References .................................................................................................... 184 APPENDIX A – Ba2In2O5 Properties.......................................................... 193 APPENDIX B – Background on Selected Measuring Techniques ............ 204 B.1 Ac impedance spectroscopy ...........................................................................204 B.2 Proton transport number determination by the electromotive force measurement ........................................................................................................218 APPENDIX C – Electrochemical Testing Procedures ............................... 223 C.1 Ac impedance spectrometry ..........................................................................223 C.2 Proton transport number determination by the EMF measurement ..........228 C.3 OCV, polarization curve and potentiostatic measurement ..........................230        xii LIST OF TABLES Table 3.1 – List of prepared samples with compositions and acronyms used........................49 Table 3.2 – Summary of the sintering temperatures and pellets properties. ..........................59 Table 4.1 – Temperature-profile XRD measurement results showing the change in crystal structure with temperature. ..................................................................................................81 Table 4.2 – Stability study of the investigated materials in liquid water and humidified air.  √- stable; X-not stable. Note: Percents are given in mole %. ....................................................90 Table 6.1 – Ba2In2O5 (BIO) sample measured in air – parameters determined by fitting the experimental ac scans to the equivalent circuit shown in Fig. 6.6.......................................119 Table 6.2 – Ce-doped Ba2In2O5 (BIC) samples measured in air – parameters determined by fitting the experimental ac scans from Fig. 6.7 to the equivalent circuit shown in Fig. 6.6. Note: Values at 100oC marked with * represent total resistance (Rb+Rgb), as bulk and grain boundary semicircles could not be resolved. ......................................................................125 Table 6.3 – Ba2In2O5 (BIO) samples measured in 50% H2/50% N2 - parameters determined by fitting the experimental ac scan shown in Fig. 6.9 to the equivalent circuit in Fig. 6.6. Note: Values marked with * and ** represent total resistance (Rb+Rgb), as bulk and grain boundary semicircles could not be resolved, where ** marked the cases where full semicircles were not present and therefore values for λ could not be determined. ..............134 Table 6.4 – Ce-doped Ba2In2O5 (BIC) samples measured in 50% H2/50% N2 - parameters determined by fitting the experimental ac scan shown in Fig. 6.8 to the equivalent circuit in Fig. 6.6. Note: Values marked with * represent total resistance (Rb+Rgb), as bulk and grain boundary semicircles could not be resolved. ......................................................................138 Table 7.1 – Measured e.m.f. across the cell (80%H2/15%N2/5%H2O), Pt║Ba2In2O5║Pt, (48%H2/49%N2/3%H2O) with standard deviation and calculated proton transport numbers (tH+) and proton conductivity at different temperatures. .....................................................152  xiii Table 8.1 – Electrolyte and cathode resistances measured by ac impedance spectroscopy for the air, Pt║Ba2In2O5║Pt, 50%H2/50%N2 cell before the potentiodynamic measurements. Anode resistance could not be resolved from the scans. .....................................................162 Table 8.2 – Results of the potentiostatic measurements (0.5 V applied) and resistances obtained by ac impedance measurements for the Pt║Ba2In2O5║Pt cell. .............................166 Table 8.3 – Electrodes used in a symmetrical cell for anode investigation..........................169 Table B.1 – Impedance expressions for some simple electrical circuit elements.................211               xiv LIST OF FIGURES Figure 1.1 – Schematic of a ceramic-based proton conducting fuel cell..................................3 Figure 2.1 – Schematic of a fuel cell and most common types of fuel cells. ...........................8 Figure 2.2 – Schematic of a fuel cell stack [6]. ......................................................................9 Figure 2.3 – Carnot efficiency and the fuel cell thermodynamic efficiency [1].....................16 Figure 2.4 – Comparison between polarization curve for a low-temperature proton exchange fuel cell and a high temperature solid oxide fuel cell [1]. .....................................................19 Figure 2.5 – Literature data for proton conductivity of selected solid proton conductors. .....21 Figure 2.6 – Literature data on proton conductivity of selected proton-conductive ceramics. Conductivity of a typical electrolyte material for SOFCs, YZS, is shown for comparison. ...23 Figure 2.7 – Presentation of the change from a cubic perovskite to an orthorhombic brownmillerite structure by doping with cations one less in valence. Lattice parameters for the orthorhombic brownmillerite structure are: a=6.09911Å, b=16.73653 Å, c=5.96221 Å and the relation to the perovskite cubic structure is given by perovbrown a2a ≈ , perovbrownb a4≈  and perovbrownc a2≈ , [33, 34, 75]. ...............................................................29 Figure 2.8 – Defect diagram for Ba2In2O5 under different partial pressures of oxygen [adapted from 38, 65 with permission from the Solid State Ionics and the author]. ..............34 Figure 2.9 – Schematic of proton transfer by Grotthuss mechanism in acceptor-doped perovskites [69]...................................................................................................................36 Figure 2.10 – Defect diagram for Ba2In2O5 under different partial pressures of water vapour. [Adapted from 80, with permission from the author]. ..........................................................39  xv Figure 3.1 – Stages of the glycine-nitrate process in this work: Schematic of the steps (a) Precursor solution boiling and thickening; (b) Porous ash formed; (c) Ash transferred into a crucible; (d) Completed reaction after heating in a furnace. .................................................51 Figure 3.2 – Schematic of the solid-state process steps used in this work.............................53 Figure 3.3 – SEM pictures of a Ce-doped Ba2In2O5 sample prepared by the GNP method and sintered at (a) 1300oC for 6h, (b) 1350oC for 6h and (c) 1400oC for 6 hours. Scale 30 µm. ..58 Figure 3.4 – Ac impedance analyzer system and pellet for conductivity testing (top); AMEL 7902 test setup.....................................................................................................................62 Figure 3.5 – Experimental setup for the e.m.f. testing and a schematic of an e.m.f. concentration cell for ion transport number determination. ..................................................65 Figure 3.6 – Setup for OCV and polarization curve testing: (1) Outer rod with sealed testing cell; (2) Inner spring loaded rod for fuel supply; (3) Three-hole rod with two Pt wires; (4) Pt wires-one to measure current, other for potential determination; (5) Pt mesh for contact; (6) Sample cell for testing; (7) Cathode side of the cell; (8) Fuel out; (9) Cathode side spring loaded rod for O2 supply; (10) Three-hole rod with two Pt wires; (11) O2 out; (12) Pt wires- one to measure current, the other for potential determination; (13) Thermocouple. ..............68 Fig. 4.1 – X-ray powder diffraction patterns for five different compositions made by both the solid-state reaction and glycine-nitrate process: (a) orthorhombic brownmillerite structure; (b) cubic perovskite structure. The temperatures needed to achieve a particular phase are given for both the GNP and the solid-state (SS) method of preparation................................74 Fig. 4.2 – Grain size as a function of calcining temperature for powders produced by the GNP process........................................................................................................................75 Fig. 4.3 – Particle size distribution for Ba2In2O5 (BIO) powder: (a) as-synthesized by the GNP method; (b) prepared by the solid-state method (thick line) and after calcination of the GNP powder at 1300oC for 6h (thin line). ............................................................................77  xvi Fig. 4.4 – TEM image of as-synthesized Ce-doped Ba2In2O5 (BIC) powder prepared by the GNP process: a selected grain with crystal fringe (left); a selected particle (right). ..............77 Fig. 4.5 – SEM images showing morphology of the BIC powders: hard agglomerates produced by the SS method (left); easily breakable clusters produced by the GNP method (right). Scale bar 100 µm.....................................................................................................78 Fig. 4.6 – Temperature-profile XRD in air (a) Ba2In2O5 (BIO) and (b) Ce-doped Ba2In2O5 (BIC) sample, both with a starting brownmillerite structure. Note: B-brownmillerite structure; C-cubic perovskite structure.................................................................................80 Figure 4.7 – TGA (thick line)/DSC (thin line) results for (a) BIO made by the GNP process, (b) BIO made by the SS process. Experiments were performed in air, in the temperature range of 25o to 1550oC with a heating rate of 5oC/min. Transition order-disorder temperature Td, and points where loss of H2O and CO2 occur are marked on the graphs..........................82 Figure 4.8 – TGA (thick line)/DSC (thin line) results (a) BIC, (b) BICL025, (c) BICL05 and (d) BIL, all made by the GNP process. All experiments were performed in air, in the temperature range of 25o to 1550oC with a heating rate of 5oC/min. Points where loss of H2O and CO2, and the order-disorder transition occur are marked on the graphs. Td* are estimated based on the temperature-profile XRD, TGA and DSC data. ...............................................84 Fig. 4.9 – Examples of X-ray diffraction patterns of samples decomposed after 24 h treatment in moist air: (a) as-prepared BIC sample for comparison; (b) decomposed BIO sample at 90% RH and 50oC; (c) decomposed BIC sample at 90% RH and 25oC; (d) decomposed BICL025 sample at 90% RH and 50oC; (e) BIC sample decomposed to Ba3In2(OH)12 in water vapour in a covered beaker. Thick arrows BaCO3, thin arrows In(OH)3...........................................91 Figure 5.1 – (Top) Schematic of a polycrystalline material with grain bulk, grain boundary and an interface with the electrode; (Middle) Equivalent circuit diagram for a polycrystalline material: Rb, Rgb, Re represent the bulk, grain boundary and electrode resistances, respectivelly; Cb, Cgb, Cdl represent the bulk, grain boundary and electrode double layer capacitances, respectivelly; (Bottom) Complex plane plot showing the ideal ac impedance  xvii scan response for a polycrystalline ceramic. Detailed explanation is given in Appendix B.1. ............................................................................................................................................96 Figure 5.2 – Ac impedance scans measured in air showing the difference between the sample/electrode interface contribution when two different electrodes are applied (Au and Pt); (a) BIO sample made by the GNP process; (b) BIC sample made by the GNP process; Frequencies are given for the points where the material contribution semicircle transitions to the electrode semicircle. ......................................................................................................98 Figure 5.3 – Ac impedance scans measured in 50% H2/50% N2 for samples with two different electrodes (a) BIO-GNP sample showing difference in the low frequency semicircle due to the electrode contribution; (b) BIC sample showing no difference in the low frequency semicircle. Frequencies are given for the points where high frequency semicircle crosses the axis......................................................................................................................................99 Figure 5.4 – Examples of the ac impedance scans for BIO, BIC, BICL025 and BIL under different environments at 300oC. Arrows point to the values on the real axis taken as total resistance of the material (Rt). ...........................................................................................101 Figure 5.5 – Arrhenius plot of Ba2In2O5 conductivity in air: ○ in this work; ■ in Niwa et al. [44]; ∆ in Zhang et al. [33] (measured in wet air)...............................................................102 Figure 5.6 – Electrical conductivity in (a) air and (b) 50% H2/ 50% N2 for samples prepared by the solid-state method: As-prepared samples: ● BIO, ♦ BIC, ▲BIL; “Dry” samples: ○ BIO, ◊ BIC, □ BICL025, ∆ BIL. Average measurement error from the three repeated measurements for each case was 5% (error bars shown for the case of BIC in (b)).............105 Figure 5.7 – Electrical conductivity in (a) air and (b) H2/N2 for GNP samples: As-prepared samples: ● BIO, ♦ BIC, ▲BIL; “Dry” samples: ○ BIO, ◊ BIC, □ BICL025, ∆ BIL. Average measurement error from the three repeated measurements for each case was 6% (error bars shown for the case of BIC in (b)).......................................................................................107 Figure 6.1 – SEM pictures of the polished and etched cross-sections of the sintered samples obtained by two different methods (GNP pressed and sintered to 1350oC for 6 hours, and the  xviii solid-state method pressed and sintered to 1400oC for 6 hours): low (a) and high (b) magnification SEM picture of a GNP sample. The high resolution picture reveals grain sizes lower than ~ 50 nm; low (c) and high (d) magnification SEM picture of a SS sample. The high resolution picture for the SS sample reveals grain sizes between 60 and 80 nm..........113 Figure 6.2 – Change in ac impedance scans with porosity for BIO-GNP (a) and BIC-GNP (b) samples measured in air at 400oC (samples were dried before measurement to release water). ..........................................................................................................................................114 Figure 6.3 – Change in the total electrical conductivity due to different porosities of the BIO- GNP (a) and BIC-GNP (b) samples in air and BIO-GNP (c) and BIC-GNP (d) samples in 50% H2/50% N2.................................................................................................................115 Figure 6.4 – Ac impedance scans for the BIO-GNP sample (grain size ~40 nm) and BIO-SS (grain size ~60 nm) measured in air. Characteristic frequencies are given in Hz. Arrows show the locations where the total material resistances (Rb+Rgb) were taken. Lines on the 300oC graph show the equivalent circuit fit to the experimental data……………………………..117 Figure 6.5 – Change in the ac impedance spectra for BIO-SS samples at 300oC over time, due to loss of water. Scans stabilize after about 3 hours............................................................118 Figure 6.6 – General equivalent circuit used for representing the contribution of bulk, grain boundary and electrode in the total ac impedance response. In a more specific case (when α=1) instead of a CPE, a capacitor C can be used. .............................................................119 Figure 6.7 – Ac impedance scans for the BIC-GNP samples (grain size ~ 40 nm) and BIC-SS samples (grain size ~60 nm) measured in air. Characteristic frequencies are given in Hz. Arrows show the locations where the bulk (Rb) and the total material resistances (Rb+Rgb) were taken. Lines on the 300oC graph show the equivalent circuit fit to the experimental data………………………………………………………………………………………….122 Figure 6.8 – Ac impedance scans measured in air at around 400oC showing the change in impedance for BIC-GNP samples with time due to the phase change. Scans stabilize after 2 hours. ................................................................................................................................124  xix Figure 6.9 – Ac impedance scans for the BIO-GNP (grain size ~40 nm) and BIO-SS samples (grain size ~60 nm) measured in 50% H2/50% N2. Characteristic frequencies are given in Hz. Arrows show the locations where the bulk (Rb) and the total material resistance (Rb+Rgb) were taken. Lines on the 200oC graph show the equivalent circuit fit to the experimental data………………………………………………………………………………………….127 Figure 6.10 – Ac impedance of Ba2In2O5 in 50% H2/50% N2 at 300oC - correction for the inductance of the external wiring; □ measured data; ○ data corrected for the inductance....128 Figure 6.11 – Change in ac impedance scan for a BIO-SS (top) sample measured in 50% H2/50% N2 atmosphere at 300oC. The change is due to the loss of water and direct incorporation of hydrogen that provides proton conductivity. ............................................130 Figure 6.12 – Change in ac impedance scans for BIO-GNP samples in 50% H2/50% N2 with loss of water and incorporation of hydrogen between 260oC and 300oC (top) and at 300oC over time. ..........................................................................................................................132 Figure 6.13 – Change of ac impedance scans of Ba2In2O5 in 50% H2/50% N2 at 500oC with time due to decomposition to BaCO3 and elemental indium...............................................133 Figure 6.14 – Ac impedance scans for the BIC-GNP (grain size ~40 nm) and BIC-SS samples (grain size ~60 nm) measured in 50% H2/50% N2. Characteristic frequencies are given in Hz. Arrows show the locations where the bulk (Rb) and the total material resistance (Rb+Rgb) were taken. Lines on the 200oC graph show the equivalent circuit fit to the experimental data…………………………………………………………………………...136 Figure 6.15 – Total and bulk conductivities determined from the ac impedance scans; (a) BIO and BIC samples measured in air; (b) BIO and BIC samples measured in 50% H2/50% N2. Note: values for BIC bulk conductivities at 100oC are projected. .......................................139 Figure 7.1 – Arrhenius plot of Ba2In2O5 conductivity in air and N2: Case I (hollow symbols): fresh sample exposed to ambient air for three days before AC conductivity measurement in air (○) or  N2 (△); Case II (solid symbols): fresh sample treated to 500oC for 3h before conductivity testing in air (●) or N2 (▲). ...........................................................................146  xx Figure 7.2 – Arrhenius plot of Ba2In2O5 conductivity in 50% H2/50% N2: Case I (♦): fresh sample exposed to air for three days before conductivity measurement; Case II (◊): fresh sample heated to 500oC for 3h before testing. Arrows show the conductivities on heating and on cooling. ........................................................................................................................148 Figure 7.3 – Comparison of Ba2In2O5 conductivity in different atmospheres: ● in air; △ in N2; ◊ in 50%H2/50%N2; ♦ in 48%H2/49%N2/3% steam; □ sample decomposed. Note:  In all cases samples were heated to 500oC before testing to remove water. .................................150 Figure 7.4 – Measured e.m.f. across the Ba2In2O5 sample as a function of time before and after increase in hydrogen activity in compartment I at 300oC (data corrected for unwanted voltage contributions). .......................................................................................................152 Figure 7.5 − Proton conductivity of selected solid proton-conductors reported in the literature [33, 40, 45-47, 61-63] compared to the proton conductivity of Ba2In2O5 determined in this work Note: Labels on this graph correspond to the labels in Fig. 2.6 (Chapter 2). ……...…..153 Figure 7.6 – Ac conductivity of Ba2In2O5 in 50%H2/50%N2 at 300oC, 350oC, 400oC, 450oC and 500oC for 24 hours......................................................................................................154 Figure 7.7 – XRD scans of Ba2In2O5 before and after ac conductivity testing in 50%H2/50%N2 at 400oC for 24 hours showing no change in structure. ..............................154 Figure 7.8 – XRD scans of Ba2In2O5 after ac conductivity testing in 50%H2/50%N2 at 500oC for 24 h: ○Ba2In2O5, ▲ BaCO3, ● elemental In. ................................................................155 Figure 7.9 – Raman spectra of Ba2In2O5 before testing (bottom scan); after testing in 50%H2/50%N2 at 400oC for 24h (middle scan) and after decomposition at 500oC (top scan); ○Ba2In2O5, ▲ BaCO3, ♦ elemental In................................................................................156 Figure 8.1 – Polarization curves (top) and associated impedance scans (bottom) measured for the air, Pt║Ba2In2O5║Pt, 50%H2/50%N2 cell at different temperatures..............................163 Figure 8.2 – Potentiostatic measurements (0.5 V applied) for the Pt║Ba2In2O5║Pt cell at 400oC while 50%H2/ 50%N2 mixture was constantly flowing on the anode and air flow was  xxi started (solid arrows) and stopped (dashed arrows) at the cathode. On top: associated impedance spectra measured during the tests. ....................................................................165 Figure 8.3 – Schematic of the anode cermet structure, with proton conducting Ba2In2O5 ceramic support and Pt, Ni or Fe as catalyst.......................................................................168 Figure 8.4 – Fe+BIO and Ni+BIO based anodes after heating to 1300oC. Ni is in the form of NiO and Fe is in the form Fe2O3 at this stage (not reduced yet)..........................................169 Figure 8.5 – Ac impedance scans obtained at 350oC in 50%H2/50%N2 for the symmetrical cells prepared with different electrodes..............................................................................170 Figure A.1 – Brownmillerite structure derived from a perovskite structure. Schematic on the right shows the oxygen vacancies (squares) ordered in 101 direction [39, 90 with permission from Journal of Solid State Chemistry]. .............................................................................193 Figure A.2 – XRD pattern of Ba2In2O5 (Adapted from Hashimoto et al. [74] with permission from the Solid State Ionics). ..............................................................................................194 Figure A.3 – Phase diagram of the BaO-InO3 system (obtained from Phase Diagram Viewer, reference Kalinina et al. [110]). .........................................................................................195 Figure A.4 – TG-MS measurement for as prepared Ba2In2O5 showing loss of H2O and CO2 with heating, reported by Hashimoto et al. [74 with permission from the Solid State Ionics]. ..........................................................................................................................................196 Figure A.5 – X-ray diffraction pattern of the orthorhombic brownmillerite (dry) Ba2In2O5 and tetragonal brownmillerite (humid) Ba2In2O5 [72 with permission from the Solid State Ionics]. ..........................................................................................................................................197 Figure A.6 – Dilatometric measurement for Ba2In2O5 showing anomaly in expansion and shrinking due to the order-disorder phase transformation [111 with permission from the Solid State Ionics]. .....................................................................................................................198  xxii Figure A.7 – In-situ X-ray diffraction data showing gradual transformation of Ba2In2O5 from brownmillerite orthorhombic to tetragonal to cubic structure [112 with permission from the Solid State Ionics]. ............................................................................................................199 Figure A.8 – Arrhenius plot of conductivity for Ba2In2O5 (a) under PO2=10e-6 atm and (b) with PO2 change reported by Goodenough et al. [70 with permission from the Solid State Ionics]. ..............................................................................................................................200 Figure A.9 – (a) Logarithm of total electrical conductivity of Ba2In2O5 over a range of PO2 and at different temperatures; (b) ionic transport number tO determined by e.m.f. measurement, from work by Zhang et al. [38 with permission from the Solid State Ionics]. ..........................................................................................................................................201 Figure A.10 – Experimental data (points) and simulation results (lines) reported by Zhang et al. [38] for Ba2In2O5 defining different types of conductivities (ions, holes or electrons) under different PO2 and temperatures (with permission from the Solid State Ionics).....................202 Figure A.11 – Proton conductivity vs. temperature determined for Ba2In2O5 by Zhang et al. [33 with permission from the Solid State Ionics]. ..............................................................203 Figure B.1 – Presentation of an ac sinusoidal potential signal (E) with a resulting current response (I). ......................................................................................................................206 Figure B.2 – Example of a Nyquist plot showing the impedance and phase angle. .............209 Figure B.3 – Example of a Bode plot showing the relationship between log impedance, phase angle and applied frequency. .............................................................................................209 Figure B.4 – Some simple elements of equivalent circuits and their corresponding Nyquist plots [65 with permission from the author]. .......................................................................212 Figure B.5 – (Top) Schematic of a polycrystalline material with grain bulk, grain boundary and an interface with the electrode; (Middle) Equivalent circuit diagram for a polycrystalline material: Rb, Rgb, Re represent the bulk, grain boundary and electrode resistances, respectivelly; Cb, Cgb, Cdl represent the bulk, grain boundary and electrode double layer  xxiii capacitances, respectivelly; (Bottom) Complex plane plot showing the ideal ac impedance scan response for a polycrystalline ceramic. ......................................................................214 Figure B.6 – General equivalent circuit used for representing the contribution of bulk, grain boundary and electrode, using CPE elements [101]. ..........................................................217 Figure B.7 – Schematic of an e.m.f. concentration cell for ion transport number determination. ...................................................................................................................219 Figure C.1 – Advanced Measurements Integrity (AMI) v.3 interface for temperature and gas flow profile. An example of a profile for an ac impedance test in air is shown on the screen. ..........................................................................................................................................225 Figure C.2 – Thales 4.15 interface for setting up the ac impedance measurements. ............227 Figure C.3 – Example of the potentiodynamic measurements using Solartron MultiSTAT 1480A. ..............................................................................................................................232            xxiv ACKNOWLEDGEMENTS If it was not for this thesis, I would have missed a great experience in my life and the many people that I have met on my way to this point. I am greatly thankful for that. Most of all, I would like to express my sincere gratitude to my supervisors, Dr. David Wilkinson and Dr. Rob Hui. I am thankful to Dr. Wilkinson for giving me this opportunity and believing in me, for his support, patience and guidance, and his dedication to the quality of this work. I would like to thank Dr. Hui for his expertise and suggestions, time and guidance through the solid – state aspect of my thesis. People from the National Research Council Canada–Institute for Fuel Cell Innovation (NRC-IFCI) also deserve a special thanks. For all these years I have been part of the institute, enjoying all the benefits of the well-equipped labs, technical expertise and friendly atmosphere. Many thanks to Dr. Radenka Maric who made this possible and the NRC-IFCI management team for their support. Special thanks go to the people from the High Temperature Fuel Cell Group, especially Dr. Cyrille Decès-Petit, Mark Robertson, Jason Fahlman, Sing Yick, Justin Roller, Dr. Roberto Neagu and Dr. Xinge Zhang, for their help and friendship. I would also like to thank my thesis committee members Dr. Olivera Kesler and Dr. Elod Gyenge for their guidance in the first stages of my research. Many thanks go to my UBC colleagues for their friendship, and especially to Greg Afonso and Mahshid Karimi for their help with the project. Financial support from the Natural Sciences and Engineering Research Council of Canada (NSERC) for this project and my CGS-D scholarship is gratefully acknowledged. I would also like to thank the NRC-IFCI for access to the labs and the financial support that  xxv have been provided for this project. The University of British Columbia must also be thanked for the financial help and the great work and living environment that has been provided for me and my family. Finally, my husband Bosko and my children deserve my endless appreciation and love, for being with me and encouraging me all these years. We have done this together!                xxvi   This thesis is dedicated to my family: To my mom and dad – who would be proud to see me reaching this high; To my aunt – who gave me confidence in life and planted a seed of love for chemistry; To my brother – who taught me everything until I was ready to learn myself;  And most importantly: To my husband Bosko – who went through this journey with me side by side, supporting me patiently and with love all these years; To my kids Iva, Vuk and Dar – who gave me balance in life and brought me endless joy…     1 1. Introduction 1.1 Motivation for the thesis project Global climate change resulting from increasing emissions of greenhouse gases, energy security and sustainability affected by declining reserves of fossil fuels, and the increasing energy demand due to the worldwide population and economic growth are key drivers for the transition from the current fossil fuel economy to a more sustainable energy economy. The future energy production/utilization is expected to rely on a diverse range of low carbon and renewable energy technologies – wind, solar, biomass, hydro, nuclear and fuel cells. Fuel cells − devices that produce electricity via electro-chemical reactions rather than combustion − convert fuel chemical energy into electricity two to three times more efficiently than thermal power plants or internal combustion engines do, and produce less toxic emissions and greenhouse gasses.  This technology has enormous promise for efficient and environmentally friendly power generation. Intense and continuous research in fuel cell technology on one side, and development of new improved ways to produce, process and store hydrogen on the other side, is needed in order to develop efficient, safe and low-cost systems that would enable fuel cell/hydrogen technology to be fully implemented and commercialized. In recent years, research has begun to focus on the development of intermediate temperature fuel cells operating between 200oC and 500oC, a most desirable operating temperature range for many practical applications. Fuel cells operating in this temperature range have the potential to overcome many disadvantages and combine the advantages of both high temperature fuel cells (solid oxide fuel cells – SOFC – operating at 600-1000oC) and low temperature fuel cells (proton exchange membrane fuel cells – PEMFC – operating  2 at 70-100oC) such as fast electrode kinetics, short start-up time, fuel flexibility, use of inexpensive non-platinum catalysts, tolerance toward fuel impurities (e.g., CO) and less degradation problems. Major approaches in the development of intermediate temperature fuel cells would include searching for new electrolyte materials with a sufficiently high ionic conductivity in the intermediate temperature range; developing new and improved techniques to make these new electrolyte materials; investigating new electrode microstructures which would improve electrolyte/electrode interfacial kinetics and transport; and development of highly active low cost catalyst materials. Research in this relatively unexplored area has the potential to address many of the current technology gaps and issues in the R&D and commercialization of fuel cells, and could have significant importance for future energy production and utilisation. 1.2 Overall approach of the thesis The overall approach is to develop a water-independent, intermediate temperature fuel cell, based on a proton conducting ceramic electrolyte and non-noble, structurally optimized electrodes, operating in the temperature range of 200-500oC. The schematic of this fuel cell is shown in Figure 1.1. The features of such a fuel cell would be as follows: • Operating temperature between 200oC and 500oC; • Ceramic proton conducting electrolyte membrane based on nanocrystalline oxygen deficient ceramic oxide materials; • Non-noble and/or low platinum group metal (PGM) content catalyst materials for the cathode and anode; • No need for water-management;  3 • Specific electrode microstructure and composition which would enable high performance; • CO and S-tolerant and potential for internal fuel reforming. In order to develop such a fuel cell, a change in electrolyte and electrode materials from currently existing materials applied in either SOFCs or PEMFCs is needed. Our goal in this thesis was to develop a suitable material to be used as an electrolyte and in the anode of such an intermediate temperature fuel cell.                Figure 1.1 – Schematic of a ceramic-based proton conducting fuel cell. H+ H+ H+ CERAMIC ELECTROLYTE LOAD e- H2 O2 H2 → 2H+ + 2e- 1/2O2 + 2H+ + 2e− → H2O A C CERAMIC-BASED ANODE CERAMIC-BASED CATHODE  4 1.3 Thesis layout The background information and work completed during this thesis project have been divided into 9 chapters. The introductory chapter (Chapter 1) states the motivation for the thesis project as well as the overall approach of this reserach. Chapter 2 contains general background and a literature review of fuel cells, with a special focus on intermediate temperature fuel cells and proton-conducting ceramics, mechanisms of ion incorporation and conduction as well as structure, and some properties of Ba2In2O5 as a base material in this project. The main objectives of the thesis are listed at the end of this chapter. Chapter 3 describes the experimental methods and procedures used in the project. Project results and discussion are presented in Chapters 4 to 8 as follows: • Chapter 4 deals with preparation and material characterisation of five selected ceramic materials, prepared by two different preparation methods. Results show properties of the materials studied by X-ray powder diffraction, particle size analysis, thermogravimetric analysis, transmission electron microscopy, etc. • Chapter 5 summarises the results of the conductivity measurements performed for the prepared ceramics in two different atmospheres – air and 50%H2/ 50% N2. Ba2In2O5 has been selected among all investigated compositions as the candidate with the highest conductivity in hydrogen-containing atmospheres, suitable for further testing. • Chapter 6 gives the details of an ac impedance study for Ba2In2O5 as the selected, high- conductive material and points to the correlation between electrical and structural properties of the material.  5 • Chapter 7 describes the details of the electrochemical testing of the selected material, and reveals results for its proton transport numbers at different temperatures. • Chapter 8 concludes the electrochemical testing of Ba2In2O5 as a proton-conductive electrolyte in a full cell set-up. In addition, a short study of Ba2In2O5 used within the anode microstructure is described. In the final chapter (Chapter 9), the results and conclusions gained through the project are summarised and recommendations for future work are given.              6 2. Background, Literature Review and Thesis Objectives  2.1 Fuel cells in general 2.1.1 Introduction and history A fuel cell is a device that electrochemically converts the chemical energy of a fuel (most commonly hydrogen, but also other fuels such as methane, methanol, ethanol, gasoline, etc.) and an oxidant (air or oxygen) into electricity [1, 2]. Fuel cells chemically combine the molecules of a fuel and oxidizer without burning and therefore they do not produce any of the undesirable products normally associated with the oxidation of fossil fuels in conventional energy conversion systems, such as SO2, oxides of nitrogen, or particulate matter. Fuel cells offer many advantages over conventional power generation systems, such as high efficiency, simplicity, quiet operation, scalable construction, flexible application, etc. Due to these advantages, this technology has a great potential for alternative energy production and use in a range of micro, portable, mobile and stationary applications. The first fuel cell was demonstrated by Sir William Robert Grove in 1839 [3, 4]. However, it was not until the period between the 1950’s and 1960’s that the concept of the modern fuel cell was developed. During this period the first 5 kW stationary fuel cell was developed and patented by the British engineer Francis Thomas Bacon [5]. In the 1960s, the Pratt and Whitney Aircraft Division of United Technologies Corporation licensed Bacon's U.S. patents for use in the U.S. space program to supply electricity and drinking water to NASA’s Apollo mission. Since that time, fuel cell research and development continues to bring fuel cells closer to commercialization.  7 2.1.2 Design of fuel cells A fuel cell generally consists of a fuel electrode (anode) and an oxidant electrode (cathode) separated by an ion-conducting membrane. Oxygen passes over the cathode, and hydrogen (or another fuel) over the anode, generating electricity, water and heat. Fuel cells are classified by their electrolyte material or operating temperature. Figure 2.1 shows the schematic of a fuel cell with the most common types of fuel cells and their associated features. In practical applications, individual fuel cells are usually combined into a fuel cell "stack" to increase the power generation. A schematic of a PEM fuel cell stack with three repeated units is shown in Figure 2.2.         8            Figure 2.1 – Schematic of a fuel cell and most common types of fuel cells. 8  Fuel cell type Fuel in Product out Anode Electrolyte material Cathode Operating temperature Oxidant in Product out Alkaline fuel cell PEM fuel cell Direct methanol fuel cell Phosphoric acid fuel cell Solid oxide fuel cell Molten carbonate fuel cell Power produced Cell efficiency System efficiency CH3OH CO2 H2 H2O H2 H2 H2, CO H2O, CO2 H2, CO H2O, CO2 O2 O2 H2O O2 H2O O2 O2 CO2 O2 H2OH+ H+ H+ OH- O2- CO32- Polymer membrane Polymer membrane Aq. alkaline solution Molten phosphoric acid Ceramic oxide Molten alkaline carbonate 70-100oC 150-250oC 500-1000oC 500-700oC 100-250oC 70-100oC 10-100 kW 100-500 kW <10 MW <100 MW 100 MW 100 mW-1 kW 60-70% 62% 20-30% 10-20% 50-70% 30-50% 55% 40% 60-65% 55-60% 55% 47% 600-1000o C  9    Figure 2.2 – Schematic of a fuel cell stack [6]. Solid oxide fuel cells (SOFCs) and proton exchange membrane fuel cells (PEMFCs) are two of the main fuel cell systems that are being considered for commercialization. Both have advantages and disadvantages in terms of operating temperature, materials used, cost and applications. An SOFC consists of two porous ceramic electrodes separated by a dense, oxygen ion conducting ceramic electrolyte. Operating temperatures of 600-1000oC are needed to achieve sufficient ionic conductivity in the electrolyte. Such high temperatures enable rapid electrochemical reactions on the electrodes without the use of noble metal catalysts, as well as direct reforming of hydrocarbon fuels and the use of carbon monoxide as a fuel. Dense ceramic electrolytes prevent fuel crossover from the anode to the cathode side. High overall (fuel to electricity and heat) efficiencies up to 80% can be achieved due to the  10 high-grade heat that can be used in cogeneration systems [7]. The disadvantages of SOFCs, such as shorter service life due to the materials degradation, long heat-up process, need for expensive heat-tolerant materials for interconnections and other components, etc., are mainly related to the high operating temperatures of these fuel cells. On the other hand, PEM fuel cells generally use perfluorosulfonic acid polymers such as Nafion as the electrolyte, which require liquid water to maintain their high proton conductivity. The maximum operating temperatures are around 100oC, due to the necessary hydration of the polymer and the stability of the polymer acid groups. In a conventional PEMFC, platinum and platinum alloys are the typical catalyst materials. A short start-up time, broad range of applications, and design and package flexibility are some of the advantages for this type of fuel cell. Disadvantages of the currently used PEMFCs are the poor kinetics of the oxygen reduction reaction on the cathode, severe poisoning of the Pt electrode by CO present in the fuel, costly materials (Pt, membrane, etc.) and large overpotential losses [1, 2]. 2.1.3 Reactions and thermodynamics Reactions that occur at the anode and the cathode of a fuel cell depend on the type of the fuel cell. At the anode of a proton exchange fuel cell, which will be considered in this project, the hydrogen oxidation reaction occurs, releasing electrons and protons (H+): 2H2 → 4H+ + 4e-                                          E°=0 V vs. SHE      (2.1) Protons are then conducted through a proton-conductive electrolyte to the cathode, where the oxygen-reduction reaction (ORR) occurs [8]:  11 O2 + 4H+ + 4e− → 2H2O                           E°=1.23 V vs. SHE   (2.2) The overall reaction is then: O2 + 2H2 → 2H2O                                     E°=1.23 V (2.3) Electrons released at the anode flow to the cathode through an external circuit to generate direct current. If the fuel cell is fuelled only by hydrogen, the only waste product at the cathode is water. Equilibrium and standard cell potential Maximum electrical energy that an electrochemical cell, such as a fuel cell, can supply is given through the change of the Gibbs free energy of the overall reaction in the electrochemical cell, and is directly related to the equilibrium (reversible) cell potential, Ecell,e [1, 9, 10]: ecellrxn nFEG ,−=∆  (2.4) Where n is the number of electrons transferred in the reaction, F= 96,485 Cmol-1 is the Faraday’s constant and Ecell, e (V) is the equilibrium cell potential. ∆Grxn (Jmol-1) is the change in Gibbs free energy of the overall reaction, and can be calculated from the Gibbs free energy of formation of the products and the reactants (values given in the literature [11, 12, 13]): .. reactofGprodofGG f j f j rxn ∑∑ −=∆  (2.5)  12 The Gibbs free energy of formation changes with temperature and state (liquid or gas). In the case of standard conditions (298.15 K, 105 Pa, activity for all the species =1), it is: o ecell o rxn nFEG ,−=∆  (2.6) From this expression, the standard cell potential, Eocell,e (or simply Eo) can be determined. The standard cell potential can also be defined as the difference between the standard half-cell potential of the cathode and anode in a fuel cell and represents the reversible, maximum cell voltage at standard conditions: o a o c o EEE −=  (2.7) where Eºc (V) and Eºa (V) are the cathode and the anode standard half cell potentials respectively. Their values can also be calculated from the Gibbs free energy of the half cell reactions, using the same expressions 2.4 and 2.5 for the half cell reactions or experimentally determined measuring the potential difference of each electrode coupled with a standard reference electrode (e.g. standard hydrogen electrode-SHE). In the above reactions given in 2.1, 2.2 and 2.3, the standard half cell potential for hydrogen oxidation is Eoa=0.00 V (vs. SHE), for oxygen reduction reaction is Eºc=1.23 V (vs. SHE), giving the standard cell potential of Eo=1.23 V [14]. Nernst equation When the cell is operated under non-standard conditions the equilibrium cell (or half cell) potential is calculated again using the Gibbs free energy of a reaction under given conditions and the final form is given by the Nernst equation [15]. For a general electrode reaction:  13 ∑∑ ⇔⋅+ − j jjR j jjO dsenOxs Re,,    (2.8) The Gibbs free energy of the reaction is:           +∆=∆ ∏ ∏ j S jOx j S jd o rxnrxn jO jR a a RTGG , , , .Re ln  (2.9) Combining equation 2.9 with equations 2.4 and 2.6 the equilibrium cell potential is obtained. The equation is called the Nernst equation:           ⋅−= ∏ ∏ j S jOx j S jd o e jO jR a a nF RTEE , , , ,Re ln  (2.10) In the above three equations sO,j or sR,j are the stoichiometric coefficients of the oxidized and reduced species, respectively; Oxj and Redj represent the oxidized and the reduced species, respectively; R = ideal gas constant (8.314 Jmol-1K-1); T = temperature (K); a is the activity of the species (for soluble species in an ideal solution a is molar concentration, ci (mol/l); for non-ideal solutions ai = γi·xi; where γi = activity coefficient; xi = mol fraction; for substances in excess such as solids, H2O, etc. a=1; for ideal gasses agas is the partial pressure of the gas (atm); for non-ideal gases agas = fiPi/Po where Pi = partial pressure; fi = fugacity coefficient; Po = standard state pressure); n is the number of electrons exchanged in the stoichiometrically balanced half reactions (n=nanode=n cathode).  14 Effect of temperature The standard cell potential, Eo, in Equation 2.10 is given at 298 K. Since this value is temperature dependent, the standard cell potential at any other temperature is given by: ∫ ⋅∆+= T oo T dTS nF EE 298 298 1   (2.11) Where EoT (V) is the standard cell potential at temperature T; Eº298 (V) is the standard cell potential at 298 K, ∆S (Jmol-1·K-1) is the change in entropy; T (K) is the operating temperature. For a small change of T (<500oC), ∆S can be often assumed constant, and therefore: ( )298298 −∆+≅ T nF SEE ooT  (2.12) For the reaction 2.3 the change in entropy is negative, and therefore the thermodynamic voltage of the cell decreases with increasing temperature (e.g. Eo(PEMFC at 80°C) ≈1.2V, while Eo(SOFC at 600-1000°C) ≈0.9–1.1V). For the same reaction the change of standard potential with the temperature is KmV dT dE o /8.0−= [16]. Although decreasing the thermodynamic voltage of a fuel cell, higher operational temperature can improve the overall cell performance by increasing the conductivity of the electrolytes and electrodes, improving mass transfer and enhancing the activity of the catalysts. Effect of pressure Beside the reactant and product activity and operation temperature, the equilibrium cell potential, Eo, will change with pressure. The relationship is given by:  15      ∆ −= 1 2, ,, ln 12 P P nF RTn EE gmolPePe   (2.13) Where Ee,P1 (V) is the equilibrium cell potential at pressure P1 (105 Pa for the standard state), Ee,P2 (V) is the equilibrium cell potential at pressure P2 (Pa or atm); ∆nmol,g is the change in number of mols of the gaseous species in the reaction; T (K) is the operating temperature. The equation assumes ideal gas behaviour. Thermodynamic efficiency of a fuel cell Efficiency of a fuel cell is often compared to the Carnot efficiency of a heat engine. Carnot efficiency is given by: %100 1 21 ⋅ − = T TT Cη    (2.14) where T1 (K) is the maximum temperature of the heat engine, and T2 (K) is the temperature of the released heated fluid. The efficiency is lower than 100% because of the wasted heat that is always present. Carnot efficiency limit cannot be applied to fuel cells. One of the ways to define fuel cell efficiency is through the maximum thermodynamic efficiency.  Maximum thermodynamic efficiency of a fuel cell is defined as the ratio of the maximum electrical energy that can be produced from a fuel cell, given by equation 2.4, and the heat (or change in the enthalpy formation) that would be produced to burn the same fuel. %100%100 ,max ⋅∆ −=⋅ ∆ ∆ = rxn celle rxn rxn H nFE H Gη  (2.15)  16 where ∆Hrxn (J·mol-1) is the change in enthalpy or so called, calorific value. This value is different, depending if the lower heating value (LHV) or a higher heating value (HHV) of a fuel is used. Actual efficiency of the fuel cell is obtained if instead of using the ideal, equilibrium cell voltage, the actual operating cell voltage is used and other losses such as fuel cross-over are taken into account. Figure 2.3 shows the comparison of the Carnot heat engine efficiency and the thermodynamic fuel cell efficiency, and how these efficiencies change with the temperature. The graph shows that low temperature fuel cells have higher thermodynamic efficiency. However, as mentioned earlier, higher temperatures often improve the overall performance of the fuel cells, by enhancing the catalyst activity, mass transport and conductivity. Also, voltage losses discussed in the following section are often lower at higher temperatures.            Figure 2.3 – Carnot efficiency and the fuel cell thermodynamic efficiency [1]. 10 20 30 40 50 60 70 80 90 0 200 400 600 800 1000 Operating temperature (oC) Th er m o dy n am ic  ef fic ie n cy  (% ) PEMFC DMFC PAFC MCFC SOFC Fuel cell, liquid product Fuel cell, gaseous product Carnot limit, 50 oC exhaust AFC  17 2.1.4 Operational fuel cell voltages Equation 2.10 in the previous section gives the theoretical, equilibrium value of the cell potential when no current is flowing. This is also called the equilibrium open circuit voltage (OCV). However, the actual OCV of a fuel cell is often lower than the theoretical OCV due to the parasitic reactions, etc. As the current is drawn from the fuel cell the cell voltage decreases proportionally with the increase in current. Figure 2.4 shows the performance (polarization curve) of a typical fuel cell operating at temperatures below 100oC, compared to a polarization curve of a high-temperature SOFC operating around 800oC.  For the low temperature fuel cell the open circuit voltage, EOC, is considerably lower than the theoretical OCV value and the initial cell voltage rapidly decreases, but for the high temperature fuel cell EOC is equal or slightly lower than the theoretical value and the initial cell voltage drop is very small. This potential loss at open circuit results from a mixed potential caused by the crossover of fuel from the anode to the cathode. The crossover is generally higher in the polymer electrolyte fuel cells than in the dense electrolyte of an SOFC. The potential loss at OCV can be also due to any small contribution of electronic conduction in the electrolyte. As higher current is drawn from the fuel cell, additional losses play a role as well. These losses cause the characteristic shape of the polarization curves shown in Figure 2.4. After the initial voltage drop at OCV, the voltage continues to decrease in the kinetic or surface activation polarization region due to the slow kinetics of the reaction at the electrode. High temperature in the SOFC promotes the kinetics, making these losses lower. In the ohmic polarization region, the additional drop in voltage is caused by ohmic losses (polarization). Ohmic polarization is caused by the resistance to the flow of electrons through the electrodes and various interconnections in the cell, as well as the resistance to the flow of ions through  18 the electrolyte. Mass transport or concentration losses (polarization) are caused by the fuel and oxidant mass transport limitations and inability to supply sufficient reactants to the electrode surface needed to support the applied load. The equation that describes the actual fuel cell voltage and existing losses is given by: ohmcdadcsasOC VEE ∆−−−−−= ,,,, ηηηη  (2.16)  where E (V) is the actual cell potential; Eoc (V) is open circuit potential; ηs,a (V) is anodic activation (surface) overpotential; ηs,c (V) is cathodic activation overpotential;  ηd,a (V) is anodic concentration (diffusion) overpotential; ηd,c (V) is cathodic concentration overpotential; ∆Vohm=i·r is the ohmic overpotential (V), i is current density (mAcm-2) and r is area-specific resistance (kΩcm2). Each of the overpotentials is mathematically described by specific equations, which will not be discussed here. They are available elsewhere in the literature [1, 2, 14, 15].            19            Figure 2.4 – Comparison between polarization curve for a low-temperature proton exchange fuel cell and a high temperature solid oxide fuel cell [1]. 2.2 Intermediate temperature fuel cells The operational temperature of a fuel cell has a large impact on fuel cell performance, efficiency, the range of applications, materials and fuels used, etc. As mentioned earlier, SOFCs and PEMFCs have a number of advantages and disadvantages that are directly linked to their operating temperatures. Reducing the SOFC operating temperature could have numerous benefits, including increased service life by reducing aging of the materials under high temperatures, shorter heating-up times, less expensive fabrication materials (e.g., metals) than ceramic ones for many components (interconnections and heat exchangers, etc.) and increased potential for mobile applications [7, 17]. On the other hand, increasing the operating temperature of the PEMFC improves the oxygen reduction kinetics (resulting in 0 0.2 0.4 0.6 0.8 1 1.2 0 200 400 600 800 1000 Current density (mAcm-2) Ce ll v o lta ge  (V ) Theoretical open circuit voltage for a PEMFC Theoretical open circuit voltage for a SOFC Ohmic control Kinetic control Mass transport control EOC, PEMFC EOC, SOFC PEMFC SOFC  20 increased power density), enhanced CO tolerance of the fuel cell enabling use of a wider range of direct fuels, and simplification of water management. A PEM fuel cell operating at higher temperatures would enable replacement of expensive Pt catalysts with cheaper metal- based catalysts or metallic perovskite-type oxides, as well as improve the use of the heat rejected [18]. The development of a high temperature PEMFC could also eliminate the need for reactant humidification, water-cooling, and other water requirements. This could reduce the weight, volume, and complexity of the fuel cell system, and thus increase the gravimetric/volumetric power density and reduce system cost. Development of intermediate temperature (200-500oC) fuel cells could possibly overcome many disadvantages of both the SOFC and the PEMFC. New materials are needed for this application. In recent years, more intense research has taken place to develop new electrolyte and electrode materials for fuel cells operating at intermediate temperatures [19- 26]. One of the major approaches in this area of research involves searching for new electrolyte materials with a sufficiently high ionic conductivity in that temperature range. In the search for a suitable intermediate temperature electrolyte material, a range of different materials have been considered: organic materials such as impregnated phosphoric acid- doped polybenzimidazole (PBI) [27-31], membranes with changed polymer chemistry [24, 31], organic-inorganic composites [32], proton-conductive ceramic metal oxides [33-44], etc. However, currently there are no solid proton conductors with suitable conductivity (≥10-1 S/cm) and stability which work satisfactorily in the temperature range between 200oC and 500oC [45]. Figure 2.5 shows a selection of currently known and investigated solid proton conducting materials and their conductivities, reported by Norby [45] in 1999. The figure has been updated with some more recent literature data. The “gap” on the graph shows the lack  21 of proton conducting materials in the 200-500oC temperature range with suitable proton conductivity. Also, the slightly lower temperature range between 100oC and 200oC does not have a wide selection of the materials. It is clear that further research in the area of intermediate temperature fuel cells is very important for further commercialisation of fuel cells. Among the materials, some proton conducting oxides show promise with respect to stability and proton conductivity in that temperature range.             Figure 2.5 – Literature data for proton conductivity of selected solid proton conductors. 1. 1M HCl [45] 2. Nafion 117, fully hydrated [45] 3. PEO-NH4ClO4 [46] 4. H3PO4 [46] 5. PBI-2.3H3PO4 [47] 6. IISPAP (imidazole- intercalated sulfonated polyaromatic polymer) [45] 7. Y:BaZrO3, in humid air [45, 46] 8. Zr(PO3(CH2)5COOH) [46] 9. Ba2YSnO5.5 in humid air [45] 10. Y:BaCeO3 in humid air (theoretical) [45] 11. Gd:BaPrO3 in humid air (proton conductivity not confirmed) [45] 12. Ba:LaErO3 in humid air [45] 13. Sr:LaPO4 in humid air [45] -8.0 -7.0 -6.0 -5.0 -4.0 -3.0 -2.0 -1.0 0.0 1.0 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 1000/T(K-1) lo g  (S /c m )                              1000 800 600   400  300     200          100                20oC 1 2 4 3 5 6 7 8 9 10 11 12 13 “gap” σσ σσ     22 2.3 Proton conducting ceramic oxides In the 1960s protons were discovered as a minor charge carrier in some ceramic oxides [48]. Active investigation of proton conduction in ceramic oxides started in the 1980s, when Iwahara and his coworkers completed their systematic studies on acceptor-doped perovskite- type oxides [49-53]. In their studies they found low proton conduction in hydrogen- containing atmospheres for oxides such as LaAlO3, LaYO3, and SrZrO3. On the other hand, oxides based on SrCeO3 and BaCeO3 showed high proton conductivities and were even tested as electrolytes in fuel cells. Power densities up to 0.2 Wcm-2 have been measured at 800oC for cells with these electrolytes (about 0.5 mm thick) and porous Pt, Ni or Ag as electrodes [52-57]. However, these materials showed instability in CO2 and at high water activities and were not suitable for fuel cell applications with hydrocarbon fuels and in humidified atmospheres [58]. Further research included searching for ceramic oxides with improved conductivity and stability. Figure 2.6 shows conductivities of a selection of proton- conducting acceptor-doped perovskite-type ceramics at different temperatures. Some of the promising materials were found to be Y-doped BaZrO3 oxides that combined high stability with high bulk proton conductivity (line 5 in Fig. 2.6), as well as single crystal Y-doped SrZrO3 (line 2 in Fig. 2.6) [46, 59]. The disadvantage of these materials was that, due to their high grain boundary resistance, the total conductivity of polycrystalline samples was generally an order of magnitude or more lower than the bulk conductivity [59]. As can be seen, the proton conductivity of many perovskite-type proton conducting ceramics reaches values higher than 10-3 Scm-1 only at temperatures higher than 400oC [60]. Still none of these or related materials have achieved satisfactory proton conduction for practical applications in the intermediate temperature range, between 200oC and 500oC. One of the materials that  23 have showed potential for high proton conduction in this temperature range is brownmillerite-structured Ba2In2O5 and its derivatives [33, 40]. Therefore, these materials are the focus of study in the current thesis project.                 Figure 2.6 – Literature data on proton conductivity of selected proton-conductive ceramics. Conductivity of a typical electrolyte material for SOFCs, YZS, is shown for comparison.  Before focusing on the conductivity of Ba2In2O5 and related materials, some background information on general electrical conductivity in oxides will be presented in the following 1. Nd:BaCeO3 , humid air [46] 2. Y:SrZrO3, single crystal, humid air [46] 3. Y:SrCeO3, humid air [46] 4. Y:BaZrO3, bulk cond., experimental, humid air  [46] 5. Y:BaZrO3, bulk cond. theoretical [46] 6. Ce:Ba2In2O5, 50% H2-50%N2 [40] 7. Ba3(Ca1.26Nb1.74)O3-δ, humid air [61] 8. Gd:BaPrO3, humid air (proton conductivity not confirmed) [45] 9. Y:BaCeO3, theoretical [45] 10. BaCe0.8Y0.2O3-δ, humid air [62] 11. BaCe0.9Nd0.1O3-δ, humid air [62] 12. SrCe0.95Yb0.05O3-δ, humid air [62] 13. SrZr0.95Y0.05O3-δ, humid air [62] 14. Ba2(Ca0.79Nb0.66Ta0.55)O6-δ, wet H2 [63] 15. Ba:LaErO3, humid air [45] 16. CaZr0.9In0.1O3-δ, humid air [62] 17.  Ba2In2O5, humid air [33] 18. Fe:LiNbO3, humid air [46] 19.  YSZ oxygen ion conductivity [62] -8.0 -7.0 -6.0 -5.0 -4.0 -3.0 -2.0 -1.0 0.0 1.0 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 1000/T(K-1) lo g  (S /c m )                                 1000 800  600     400   300      200             100                    20oC 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 “gap” σσ σσ     24 section. Also, as the conductivity of the acceptor-doped perovskites and related brownmillerites is related to their crystal structure, Section 2.3.2 will describe the crystal structure of these materials. 2.3.1 Electrical conductivity in ceramic oxides  Electrical conductivity is the net transport of various charge carriers under the influence of an external electric field [64-66]. Ceramic oxides can exhibit electronic and ionic conductivity. While the electronic conductivity involves transport of electrons and holes through the solid, the ionic conductivity involves transport of ions (e.g., oxygen ions and/or protons).  Total electrical conductivity σ of a conducting oxide is the sum of the partial conductivity associated with each type of charge carrier and is given by: +− +++=+==∑ iipnie j j σσσσσσσσ  (2.17) where σe σi,σn,σp,σi-,σi+ (all in Ω−1cm-1 or Scm-1) represent the conductivity contribution from electronic charge carriers, ionic charge carriers, electrons (e-=n), holes (h•=p), anions, and cations, respectively. The fraction of the total conductivity contributed by each charge carrier is the transport number, tj, and is given by: σ σ j jt =  (2.18) Each partial conductivity σj , is defined by: jjjj ezc µσ )(=  (2.19)  25 where cj is the concentration of charge carrier j (number per unit volume), zje is its charge (expressed as a multiple of the charge of an electron, e=1.602189 x 10-19C), and µj is its charge mobility (drift velocity in a constant electric field, m2s−1V−1). In most cases, the concentration cj is the parameter that is modified by the controlled atmosphere and chemical composition of the oxide, while the charge mobility µj is usually a function of temperature only. The charge mobility of an ion is related to its mechanical mobility, Bj, (the drift velocity of an ion per unit of applied electric force, m2s−1C-1V−1) and its diffusion coefficient through the solid, Dj, by the following relationship: jjj eBz=µ  (2.20) jj kTBD =  (2.21) where k=1.381 x 10-23 J/K is the Boltzmann’s constant, T is temperature (K), Dj is the diffusion coefficient (cm2s-1), and Bj is the mechanical mobility (m2s−1C-1V−1). Combining equations 2.19, 2.20 and 2.21 the Nernst-Einstein equation can be obtained, giving the relationship between charge mobility or conductivity and the charge diffusion coefficient [64-66]: kT Dez jj i )( =µ  (2.22) or kT Dezc jjj j 2)( =σ  (2.23)  26 For ionic conductors, the temperature dependence of the charge mobility and ionic conductivity is expressed by the Arrhenius equations:       − = kT E T F ii i exp 0,µµ  (2.24)       +− = kT EE T M i F ii i )( exp0, σ σ  (2.25) where µi,0 is a proportionality constant that depends on attempt frequency (probability that an ion will attempt to break the bond and leave its current position in a structure), the distance moved by the atom, and the size of the external field. σi,0 is a proportionality constant that contains ci and Zie, and also depends on the attempt frequency and jump distance. EiF and EiM are the energies for charge carrier formation and migration, respectively. Both the ionic and the electronic conductivities of any ionic compound exhibit Arrhenius-type temperature dependencies. In general, the temperature dependence of the electrical conductivity is given by the Arrhenius equation:       − = kT E T aexp0 σ σ  (2.26) where Ea is the energy of activation for formation and transport of the charge carriers. This expression is often presented as:       −= kT E T a0lnln σσ  (2.27)  27 and plotted as lnσT against 1/T, which produces a straight line with a slope of -Ea. The expression 2.26 is sometimes presented empirically as:       − = kT Eaexp0σσ  (2.28) and plotted as lnσ vs. 1/T or logσ vs. 1/T. The plots in Figure 2.5 and 2.6 are examples of the presentation of conductivity dependence on temperature. 2.3.2 Crystal structure of acceptor-doped perovskites and brownmillerite Ba2In2O5 Conductivity of acceptor-doped perovkites and related brownmillerites is related to their crystal structure. An ideal, undoped cubic perovskite oxide has the general formula A2+B4+O32-, where A and B are two cations of different sizes (A being larger than B), and O is an oxygen anion that bonds to both. The structure has the B cation surrounded by an octahedron of anions, forming BO6 octahedra which share corners in all 3 dimensions. The A cations occupy every hole which is created by 8 BO6 octahedra, giving the A cation a 12-fold oxygen coordination, and the B-cation a 6-fold oxygen coordination. The cubic perovskite structure is given in Figure 2.7. An acceptor-doped perovskite is formed when the ABO3 perovskite is doped with lower- valent M3+ ions, forming oxygen vacancies to compensate for the charge difference [67]. In terms of Kröger-Vink notation [68] this process can be presented as [69]: 232 222 BOVMOMOB OB x O x B ++′→++ ••  (2.29)  where BBx is the B4+ metal ion at its original position (effective charge =0), OOx is the oxygen ion at its position (effective charge =0), M2O3 is the metal oxide used for doping, M’B is the M3+ dopant ion at the B site (acceptor) (effective charge = -1), VO•• is the oxygen vacancy  28 (effective charge = +2) and BO2 is the formed oxide after the B ion is released from the structure. The general formula of the acceptor-doped perovskite then becomes A2+B4+1-xM3+xO2-3-δ, where δ refers to the oxygen vacancies. Special cases of the acceptor-doped perovskites are the brownmillerite-structured oxides, such as Ba2In2O5, where B4+ ions in a perovskite oxide are completely substituted by lower-valent metal ions, M3+(In3+ in this case) [67, 70-75]. To maintain charge balance, one-sixth of the oxygen anions are removed and a high concentration of oxygen vacancies is formed. At low temperatures, the vacancies are ordered in parallel rows (tetrahedral oxygen vacancy layers) alternating with mainly vacancy-free octahedral perovskite layers. These vacancies are not normally occupied and thus are treated as interstitial sites in the brownmillerite structure.  The structure of a cubic perovskite structure and transformation into a brownmillerite structure is shown in Figure 2.7.             29           Figure 2.7 – Presentation of the change from a cubic perovskite to an orthorhombic brownmillerite structure by doping with cations one less in valence. Lattice parameters for the orthorhombic brownmillerite structure are: a=6.09911Å, b=16.73653 Å, c=5.96221 Å and the relation to the perovskite cubic structure is given by perovbrown a2a ≈ , perovbrownb a4≈  and perovbrownc a2≈ , [33, 34, 75].  2.3.3 Oxygen ion, proton and electron conductivity of acceptor-doped perovskites and brownmillerite Ba2In2O5 A stoichiometric perovskite material is itself not a good ionic conductor [40, 76]. Conductivity of these materials is due to the presence of oxygen vacancies, which, as discussed in the previous section, can be created through doping the perovskite with lower- valent metal ions. Existence of the oxygen vacancies favours oxygen ion transport and causes oxygen ion conduction in these materials. Moreover, mixed oxygen-ion, proton and Perovskite structure Brownmillerite  structure  30 electronic conduction have been found in these systems with oxygen ion conduction predominant at high temperatures (typically above 500oC) and protonic conduction predominant at low temperatures (typically below 500oC) [33, 41-45, 59]. Which conductivity will predominate depends on the dopant species, temperature and the atmosphere the material is exposed to. The concentration of various charge carriers (e.g., oxygen ions, protons, electrons, holes) and defects (e.g., oxygen vacancies, oxygen interstitials) as well as the conducting behaviour in these oxides is controlled by the equilibrium between oxygen or hydrogen/water vapour partial pressure, oxygen vacancies, oxygen interstitials, protons, electrons and holes, under certain operating conditions [77, 78]. Because of these electrical properties, some acceptor-doped perovskite oxides (e.g., BaCeO3, SrCeO3, BaZrO3 and SrZrO3) and related brownmillerite oxides (Ba2In2O5) have attracted significant attention. Brownmillerite-structured Ba2In2O5, as an extreme case of doping, has a high concentration of oxygen vacancies. The defect chemistry of the brownmillerite structure in Ba2In2O5 favours both oxygen ion conduction as well as proton incorporation [33]. As for the other acceptor-doped perovskites, the electrochemical behaviour and electrical conduction of Ba2In2O5 has been found to be very dependent on temperature and atmosphere. In addition to oxide ion conduction at higher temperatures, Ba2In2O5 has been reported to display both electronic conductivity under variable redox conditions and proton conductivity due to water incorporation in its low temperature form [33, 35, 38, 40, 43, 76]. Oxygen ion conductivity in Ba2In2O5 Oxygen ion conductivity in brownmillerite Ba2In2O2 is possible because of the oxygen vacancies created by doping. These vacancies (that are considered interstitial sites) order in  31 the tetrahedral oxygen vacancy layer and present a large space to incorporate oxygen ions from the octahedral oxygen layer through the thermally activated anion Frenkel defect (this defect forms when an atom or an ion leaves its place in the lattice creating a vacancy, and occupies an interstitial site in the lattice) [34]. This process is expected to happen relatively easily, resulting in a large concentration of anion Frenkel defects. Using Kröger-Vink notation, this process can be presented as: ••+′′⇔+ oi x o x i VOOV  (2.30) where xiV is the vacant interstitial site in the oxygen vacancy layer, x oO is oxygen ion at its site, iO ′′ is the oxygen ion at an interstitial site and •• oV is the oxygen vacancy. In dry air, in addition to reaction 2.30, oxygen from the atmosphere incorporates at oxygen vacancies generated by the Frenkel disorder, producing electron holes ( •h ) and p- type electronic conductivity, according to the following reactions [38, 67, 71]: ••• +⇔+ hOVO xOOg 221 )(2   (2.31) or alternatively: x OOg OeVO ⇔′++ •• 221 )(2  (2.32) where, using Kröger-Vink notation, ••OV  is an oxygen vacancy, x OO  is the oxide ion, •h is an electron hole, and e′ is the electron. In this case, the effective charge carriers consist of oxygen vacancies and electron holes (or electrons), which lead to oxygen ion and electronic conductivity. At low temperatures, oxygen vacancies are trapped in the ordered tetrahedral layers of the brownmillerite structure and do not contribute extensively to the oxygen ion conduction.  32 Conductivity increases at high temperatures, due to the higher mobility of holes and ions, with the process being favoured at temperatures above 500oC. At a certain characteristic temperature for these types of materials, the so-called transition order-disorder temperature (Td) [34, 38, 79], oxygen vacancies start to distribute randomly. With further heating, the vacancies become completely disordered so that the structure reverts to that of a highly defective cubic perovskite. The material exhibits a dramatic increase in electrical conductivity as the ‘trapped’ oxygen vacancies become partially disordered. For Ba2In2O5 the transition order-disorder temperature is Td≈ 925oC [38]. While the dominant ionic defects in these materials below the Td are oxygen vacancies created by the anion Frenkel disorder, at temperatures above the Td they are created by complete anion disorder. The main defects are oxygen vacancies with their concentration defined by  [VO••] = ½[InB’] (2.33)  independent of temperature and pressure PO2. Some reported results on Ba2In2O5 oxygen ion conductivity are given in Appendix A. In addition to oxygen ion and electronic p-type conductivity, the material has thermally activated n-type and p-type conductivity, through the reaction: •+′⇔ he0  (2.34) However, this conductivity is expected to be negligible [38].  33 The concentration of the holes, electrons or oxygen vacancies under different conditions can be determined using expressions for equilibrium constant for the reactions 2.30, 2.31, 2.32, and 2.34: K1= [ iO ′′ ] [ ••OV ]          for reaction 2.30 (2.35) p= K2 [ ••OV ]1/2 4/12OP       for reaction 2.31 (2.36) n= K3 [ ••OV ]-1/2 4/12 − OP     for reaction 2.32 (2.37) K4= n· p                       for reaction 2.34 (2.38)  where K1, K2, K3, K4 are the equilibrium constants for reactions 2.30, 2.31, 2.32, and 2.34, respectively, [ ••OV ] is the concentration of oxygen vacancies, [ iO ′′ ] is the concentration of the oxygen ions at interstitial sites, p is the concentration of holes h•, n is the concentration of electrons e′ , and PO2 is the partial pressure of oxygen in the atmosphere.  Given a specific oxygen partial pressure and resulting equilibrium between charged species and defects in the oxide, the concentration of specific mobile species can be calculated. These calculations with specific conditions will not be shown here, but can be found in the literature [34, 65, 68, 80]. Figure 2.8 shows dependence of the concentration of various charge carriers and defects in Ba2In2O5 and the resulting conductivities under different PO2 and temperatures.        34            Figure 2.8 – Defect diagram for Ba2In2O5 under different partial pressures of oxygen [adapted from 38, 65 with permission from the Solid State Ionics and the author]. Proton conductivity in Ba2In2O5 While oxygen ion conductivity of Ba2In2O5 under different conditions is well investigated [33, 35, 38, 43, 70], proton conductivity has been studied only in humid air or inert gases, but not in hydrogen containing atmospheres. Zhang et al. [33] measured the electrical conductivity of Ba2In2O5 in dry air and in 25°C H2O-saturated air (PH2O ≈ 0.03 atm) as a function of temperature and concluded that the conductivity is higher in the H2O- saturated air. That suggested the presence of protonic conduction in this material. Although low (<1·10-5 S/cm), proton conductivity was found to be predominant at temperatures below 500oC. The transport number for proton conduction decreased with increasing temperature. The proton transport number also decreased with increasing PO2 because of the hole  VO•• -1/6 n 1/6 Oi” [Oi”] = [VO••] Oi” p VO•• 1/4 -1/4 p n n p log P(O2) lo g co n c.  T>Td T<Td n-type mixed c. p-type mixed c. ionic cond. [VO••] = ½[InB’]  35 contribution to the total conductivity (according to equation 2.31). A sharp increase of the protonic conductivity (as well as oxygen ion conductivity) was noticed at the order-disorder transition temperature, Td (925°C). Another study by Schober et al. [73] confirmed proton solubility in Ba2In2O5 in moist air atmospheres at temperatures below 800oC. The only study of Ba2In2O5-related materials in hydrogen-containing atmospheres was done by Hui et al. [40], in which they measured the total electrical conductivity of Ce-doped Ba2In2O5 in a humidified gas mixture of 50% H2 and 50% N2, which achieved an electrical conductivity of 7·10-3 S/cm at 300°C. Figure 2.6 shows reported total electrical conductivities measured for Ba2In2O5 and related materials in humidified or hydrogen-containing atmospheres. Proton incorporation in the Ba2In2O5 structure is related to oxygen vacancies in the structure. When the ceramic material is in contact with water or hydrogen molecules, oxygen vacancies play a role in forming hydroxyls ions ( •OH ) [35, 38, 40, 73] significantly enhancing the proton conductivity of these materials at low temperatures (200-500oC). When water vapour is present in the atmosphere, protons incorporate in the crystal structure, by the following mechanism: ••• +⇔+ i x OOg HOVOH 2)(2                                                   (2.39) Oxygen vacancies, ••OV , react with water to fill lattice positions with oxide ions, x OO , and produce interstitial protons, •iH . However, due to the small size of the protons, the protons do not occupy a true interstitial site (oxygen vacancy) but attach to oxide ions, thus forming an hydroxyl ion •OOH : •• ⇔+ Oi x O OHHO  (2.40)  36 The net reaction demonstrating the interaction of oxygen vacancies with water vapour producing proton charge carriers can be written as: ••• ⇔++ O x OOg OHOVOH 2)(2  (2.41) The process can also be represented by the following reaction according to Lefebvre-Joud et al. [62]:   (2.42)  Once incorporated into the structure, the proton migrates by a Grotthus mechanism. It was shown experimentally and by numerical simulation that this transport involves rotational diffusion of the protonic defect in the O-H groups and proton transfer toward a neighbouring oxide ion. During this process, only the proton shows long-range diffusion, whereas the oxygen atoms reside in their crystallographic positions [46, 59, 69]. The simulation results also show that the proton locally “softens” the lattice to allow the transient formation of hydrogen bonds followed by the proton transfer between adjacent oxygen ions [81-84]. Figure 2.9 shows the proton transport by Grotthus mechanism.      Figure 2.9 – Schematic of proton transfer by Grotthuss mechanism in acceptor-doped perovskites [69]. +H H O M M M O M M M M O M H H O Oxygen Proton Metal (In)  37 If oxygen is also present in the atmosphere, it competes with water molecules for the available oxide ion vacancies, producing electron holes ( •h ) and p-type electronic conductivity, according to equation 2.31. On the other hand, when hydrogen is present in the gas stream, Hui et al. [40] and Norby [80] have suggested the possibility that hydrogen is incorporated directly into the material as protons and electrons ( e′ ) through an interaction with oxide ions, according to the following reaction: eOHOH OxOg ′+⇔+ •)(221                           (2.43) Hui et al. suggested that hydrogen incorporated by this mechanism plays a more important role in proton conduction than that of the moisture. This process is predominant at lower temperatures, up to 500oC, causing higher conductivity in hydrogen containing atmospheres. However, this mechanism has not been experimentally confirmed. Bonanos [35] has proposed an alternative mechanism for hydrogen incorporation through protons in the interstitial sites of the lattice: eHH ig ′+⇔ • )(221                                                  (2.44) where using Kröger-Vink notation e′ denotes a negative free electron, and •iH is a proton in the interstitial site. It is considered that protons do not occupy regular interstitial positions (vacancies) due to their small size, but rather attach to oxygen ions [35, 80]. If these reaction mechanisms were happening, an equal number of protons and electrons would be produced. As Bonanos noticed, in that case, proton conduction would always co-exist with n-type electronic conduction, and have a very low transport number due to the lower mobility of protons than electrons. Since this behaviour is not noticed in practice, the above-mentioned  38 mechanisms are not likely. However, it is possible that the electrons interact with other defects in the oxide so that the proton formation is compensated by formation of other negative defects or by the annihilation of positive defects [80]. Under reducing conditions, where hydrogen is stable in the oxidation state 0 (as H2 in the gas phase) it might be possible to anticipate that neutral hydrogen atoms dissolve in oxides, probably interstitially, as xiH [80, 85]. Since proton incorporation in acceptor-doped oxides in hydrogen-containing atmospheres is not fully understood, more investigation of this process is required. The concentration of protons in the case of reaction 2.41 and its dependence on PH2O can be calculated using the expression for the equilibrium constant for the reaction [80]: [ •OH ] = K5 [ ••OV ]1/2 2/1 2OHP  (2.45) where [ •OH ] is the concentration of •OH ions or protons, K5 is the equilibrium constant for reaction 2.41 and OHP 2  is the partial pressure of water vapour. At low PH2O, the concentration of oxygen vacancies is determined by dopant level and is considered to be constant: [ ••OV ] = 1/2[ BnI ′ ] = constant (2.46) resulting in [ •OH ] being proportional to 2/1 2OH P .  At high PH2O the dissolved protons become the important positively charged point defects and the electroneutrality condition is then approximated by: [ •OH ] = [ BnI ′ ] = constant (2.47)  39 and the concentration of protons does not depend on PH2O. Combining equation 2.45 with equation 2.36, and taking into account charge and mass balance conditions, the concentration of oxygen vacancies and holes can be determined as well, as shown by Norby [80]. Figure 2.10 shows the effect of water vapour partial pressure on the concentration of different charge carriers.                        Figure 2.10 – Defect diagram for Ba2In2O5 under different partial pressures of water vapour. [Adapted from 80, with permission from the author].  To conclude, for Ba2In2O5 and related materials, the electronic conductivity is generally not a major contribution to total conductivity. High oxygen partial pressure enhances oxygen ion conductivity and high water partial pressure enhances the contribution  VO•• p log P(H2O) lo g co n c.  [VO••] = ½[InB’] OH• 1/2 [OH•] = [InB’] -1 p -1/2  l o g co n c.   40 of protonic conductivity. Electronic conductivity is very small and usually can be ignored, protonic conductivity is significant in the medium temperature range (300-600°C) and oxygen ion conductivity becomes dominant at higher temperatures (above 500oC). Proton conductivity determination Since Ba2In2O5 can exhibit oxygen ion, proton and electron conductivity under different conditions, it is important to determine the contribution of each of these conductivities to the total conductivity. One of the possible methods to distinguish the oxygen ion, electronic and protonic conductivity in Ba2In2O5 is to compare the total conductivity of a sample in three different atmospheres consisting of dry air, dry nitrogen and a hydrogen-containing atmosphere. As discussed earlier, in air it is expected that Ba2In2O5 should exhibit both oxygen ion and p-type electronic conductivity (according to equation 2.31) in addition to thermally activated n-type electronic conduction, which is expected to be negligible [38]. Hence, the measured conductivity would be given by: σair = σi + σen + σep                                                        (2.48) where σair represents the total conductivity measured in air, σi is the oxygen ion conductivity, σen is the thermally activated n-type electronic conductivity and σep is the p- type electronic conductivity. In nitrogen, beside oxygen ion conductivity and insignificant thermally activated n-type electronic conductivity, the material can still exhibit p-type conductivity due to the presence of a very low  concentration of oxygen (PO2~10-6 atm) in supplied nitrogen. However, as shown by Zhang et al. [38], the p-type conductivity is very low under these conditions and can be neglected. Additional n-type electronic conductivity is  41 expected in nitrogen under certain conditions due to the reduction of the material. The reduction is happening at oxygen partial pressures lower than 10-4 atm, when oxygen leaves the structure, forming oxygen vacancies and accompanying electron centers, as shown by the reaction: ••+′+⇔ Og X O VeOO 221 )(2             (2.49) However, this process is not expected to happen below 700oC for Ba2In2O5, as confirmed by Zhang et al. [38].  Consequently, the main conductivity that can be measured under a nitrogen atmosphere and up to 500oC (σN2) is the oxygen ion conductivity and the thermally activated n-type electron conductivity, given by: σN2 = σi  + σen          (2.50) If we consider σen negligible under lower temperature conditions (< 700oC), the measured conductivity in nitrogen would directly determine the oxygen ion conductivity. From equations 2.48 and 2.50, the p-type electronic conductivity in Ba2In2O5 in oxidizing atmospheres can be determined from the difference of the measured conductivity in air and nitrogen: σep =   σair - σN2          (2.51) In hydrogen containing atmospheres, assuming that Ba2In2O5 does not undergo a reduction process which would result in some electronic conduction, the measured  42 conductivity would be the sum of the protonic conductivity, oxygen ion conductivity and thermally activated electronic conductivity: σH2 = σH+ + σi + σen    (2.52) If the assumption is made that proton conductivity does not affect other charge carriers, the proton conductivity can be determined from the difference of the measured conductivity in hydrogen and nitrogen: σH+ = σH2 - σN2     (2.53) Another widely used method to determine the proton conductivity contribution, or proton transport number, tH+ (ratio between proton conductivity and total conductivity, tH+=σH+/σt), in oxides is the electromotive force method (e.m.f.) [86-88]. This method consists of measurement of the voltage difference across a fully dense sample disk exposed to a gradient in the partial pressure of hydrogen or water vapour, with a constant partial pressure of oxygen on both sides. The proton transport number can be determined as the ratio of the measured voltage to the theoretical electromotive force for the same conditions, given by:                                                                      (2.54)  where tH+ is proton transport number, R is the universal gas constant (J/mol K), F is the Faraday constant (C/mol), PH2I is the high partial pressure of hydrogen (atm) and PH2II is a I H II H HIII P P F RT tE 2 2ln 2 ⋅−= +−  43 low partial pressure of hydrogen (atm). A more detailed explanation of the e.m.f. measurement is given in Appendix B.2. 2.3.4 Effect of transition order-disorder temperature on conductivity and possible improvement of conductivity Since proton incorporation and conduction is related to oxygen vacancies in the brownmillerite structure, the distribution of the vacancies plays a role in proton conductivity of this material. As explained earlier in the section on oxygen ion transport, the material exhibits a significant increase in electrical conductivity above the transition order-disorder temperature due to disordering of the vacancies. However, the transition order-disorder temperature for Ba2In2O5 is Td≈ 925oC [38], and in order to obtain enhanced conductivity in the intermediate temperature range, it is necessary to reduce the Td and stabilize the disordered phase at lower temperatures. The substitution of barium (II) or indium (III) in Ba2In2O5 with other cations has been reported to lead to the reduction of Td and the improvement of oxide ionic conduction at lower temperatures (below 900oC) [44, 70, 89]. Goodnough et al. [70] showed that substituting bigger ions with a larger valence, like Ce4+, for indium (III) ions resulted in stabilizing of the disordered phase below Td=925oC and led to an improvement in total electrical conductivity below 900oC (measured under a low PO2 of 10-6 atm). Niwa et al. [44] measured total electrical conductivity of Ce-doped Ba2In2O5 (Ba2In2−xCexO5+x/2) in O2 and N2 and also confirmed that a gradual increase of Ce4+ content caused reduction of the Td and affected the conductivity results. They reported that, for a sample with a Ce4+ content of x=0.2 and above, the sharp discontinuity in conductivity related to Td disappeared and the sample showed an improvement in conductivity of about one order of magnitude at lower  44 temperatures (500-700oC). However, the achieved conductivities were still too low for practical applications. In order to additionally improve electrical conductivity at low temperatures, it is necessary to further decrease the order-disorder temperature. Mitamura et al.[37] investigated the change in the order-disorder transition temperature with substitution of cations in the B sites of A-B-O brownmillerite and related perovskite structures. The transition temperatures of BaLa2O4, Ba3Y4O9, Ba2In2O5 and Ba3Ga2O6 were found to be 270oC, 350oC, 880oC and 1230oC, respectively. They concluded that the ionic radius of the cations influenced the transition temperatures, and that the temperatures decreased with increasing ionic radius, e.g., the La-doped material showed the lowest transition temperature in the series. Based on these results it is expected to be beneficial to substitute indium (III) in Ba2In2O5 by cerium (IV) and lanthanum (III) cations, in order to lower the order-disorder transition temperature and hence increase electrical conductivity of these materials at lower temperatures. 2.3.5 Stability of Ba2In2O5 Beside good proton conductivity, stability of Ba2In2O5 and related compounds under proposed  fuel cell operating conditions is crucial if this material is to be used as a proton conductive electrolyte. The material should be stable in the intermediate temperature range between 200oC and 500oC, in hydrogen-containing atmospheres, and under humidification. Stability in CO2-rich atmospheres is desirable if this material is to be used with hydrocarbon fuels. Any decomposition of the material would affect its mechanical and electrical properties. In hydrogen-containing atmospheres, a special concern is reduction of the material. Possible reduction of In3+ to a lower valence state could induce electronic conductivity and  45 affect the performance of this material as a proton conductor. However, it is reported in the literature that Ba2In2O5 is resistant to reduction [38, 90, 91] at temperatures below 700oC. Zhang et al. [38] showed that no n-type electronic conduction occurs in the material below 700oC due to the reduction process. Fisher et al. [90] showed that due to the high reduction energy of 4.30 eV/electron, Ba2In2O5 is more resistant to reduction than oxidation. In their study on redox stability of Ba2In2O5 and related compounds by cyclic voltammetry, Rolle et al. [91] confirmed good stability of Ba2In2O5 at 600oC up to -1 V versus the reference oxygen electrode (ROE) in nitrogen, which corresponds to an oxygen partial pressure of 8·10−24 atm. A small and reversible reduction peak was observed at -1.1 V/ROE, which they believed was due to the reduction of In3+ to elemental indium.  They did not observe any reduction peaks that corresponded to reduction of In3+ to In2+ or In1+.  In addition, they performed high- temperature X-ray diffraction (XRD) at 600oC in 1.5% H2 for 17 h and confirmed no change in the crystal structure of Ba2In2O5. Instability in H2O rich atmospheres is not reported for Ba2In2O5, although it is known that this material incorporates H2O in its structure at temperatures below 300oC [72, 74]. In a CO2-rich atmosphere, Zhang et al. [38] found that Ba2In2O5 decomposes easily at temperatures above 650oC. The material decomposes at PO2 between 10-9 to 10-13 atm in the presence of CO2, forming BaCO3. Its stability at lower temperatures has not been reported, but as in the case with H2O, Ba2In2O5 is reported to contain CO2 incorporated in its structure below 700oC, without affecting its structure [74]. 2.4 Research opportunity for proton conductivity in Ba2In2O5 and related materials In the previous sections the potential of Ba2In2O5 and related materials have been shown, as well as background of principles of their conductivity. As can be seen from the literature  46 review, most research on undoped and doped Ba2In2O5 has investigated properties and the electrical conductivity of these materials in oxygen or inert atmospheres, and at temperatures above 500oC. The investigated materials have been mainly prepared by the conventional solid-state process (discussed in the experimental section). In order to determine potential proton conductive ceramic materials for application in the intermediate temperature range (200-500oC) it is important to investigate the properties of these materials in hydrogen- containing atmospheres at temperatures below 500oC. Very little work has been done in this area in hydrogen-containing atmospheres at lower temperatures. Also, other preparation methods for Ba2In2O5 (and related materials) and the effect of preparation method on conductivity have not been investigated much. Some limited work has been reported by Hui et al. [40] where they prepared Ce-doped Ba2In2O5 by the reactive spray deposition technique (RSDT) and measured total electrical conductivity in a humidified gas mixture of 50% H2 and 50% N2, achieving an electrical conductivity of 7·10-3 Scm-1 at 300°C. If Ba2In2O5 and related materials are to be used as proton conductive electrolytes in electrochemical devices such as fuel cells, it is crucial to determine their electrochemical behavior in hydrogen containing atmospheres and evaluate their performance within the fuel cell. In this work our goal was to prepare and characterize some new compositions of Ce- and La-doped Ba2In2O5, along with some previously investigated Ce-doped and undoped Ba2In2O5 for comparison. A particularly important aspect of this work was to study the electrical conductivity and determine the proton conductivity of these materials in hydrogen- containing atmospheres in the temperature range from 100oC to 500oC, conditions that have not been widely investigated before.   47 2.5 Thesis objectives Within the overall approach of the thesis stated in Chapter 1, and based on the literature search, the objective of this thesis project was to develop a ceramic proton-conducting material to be used as the dense electrolyte in the intermediate temperature fuel cell explained in Section 1.2. Investigated ceramic electrolyte materials were based on oxygen deficient ceramic oxides – undoped and Ce- and La-doped Ba2In2O5. The specific objectives of the project were as follows: 1. Select candidate materials to be evaluated as proton-conductive ceramics, based on a background literature search. 2. Prepare and characterize selected candidate materials, in terms of crystal structure, particle and grain size of the powders, thermal properties, stability, and investigate the effect of the powder preparation method on properties. 3. Compare electrical conductivity of the materials in air and hydrogen-containing atmospheres and select a candidate material for the proton-conducting electrolyte. 4. Evaluate the selected proton-conducting material electrochemically. Target a total electrical conductivity of over 10-2 S/cm at 200oC and a proton transport number close to unity (≥0.8). Correlate the electrochemical behaviour to the material properties. 5. Apply the selected material as an electrolyte in an intermediate temperature fuel cell and evaluate its performance.  A highly conductive, dense, gas-tight film of ceramic electrolyte was expected to be developed. 6. An additional objective (time permitting) was to incorporate the selected material within the anode structure and evaluate.  48 3. Experimental Approach 3.1 Synopsis In order to achieve the objectives of this thesis project, the experimental work was organized into several segments. The first goal was to prepare and characterize a number of oxygen deficient ceramic oxide compositions selected through a literature search as potential proton-conductors in the 100-500oC temperature range. Five different compositions of brownmillerite-structured materials, Ce- and La-doped, and undoped Ba2In2O5 were selected and synthesized. Some compositions were synthesized for the first time, such as Ce-/ La- doped Ba2In2O5, while Ce-doped and undoped Ba2In2O5 were prepared for comparison and further testing, unreported in the literature. The materials were produced by two different methods consisting of the glycine-nitrate combustion process or the solid-state reaction in order to investigate the effects of the preparation method on their properties and, as a final objective, their electrical conductivity. The synthesized powders were characterized by a number of material characterization techniques, described in the following sections. In the next stage, electrical conductivity of the materials was measured by electrochemical impedance spectroscopy (also called ac impedance spectroscopy) in different atmospheres in the temperature range between 100oC and 500oC. The effect of dopants, microstructure, temperature and atmosphere on electrical conductivity was investigated. A particularly important aspect of this work was to study the electrical conductivity of these materials in hydrogen-containing atmospheres and in the temperature range from 100oC to 500oC, conditions that have not been studied much or at all for these types of materials. Based on the results, a candidate material for further, more detailed study on proton conductivity was chosen. A detailed study of the conductivity of this material in hydrogen-containing  49 atmosphere was followed by a proton transport number determination study. Finally, the performance of the selected material as an electrolyte and within the anode structure in a fuel cell was tested by standard electrochemical methods, such as electrochemical impedance spectroscopy (EIS), open circuit voltage (OCV) and polarization curves.  The details of all the stages of the experimental work are described in Sections 3.2 to 3.9. 3.2 Powder preparation Samples of Ba2In2-x-yCexLayO5+x/2 (x=0.25 and 0.5; y=0.25 and 0.5) were synthesized via the glycine-nitrate combustion process (GNP) and solid-state reaction (SS). The level of dopants was chosen to be 0.5 in total (x+y=0.5). Based on the previous reported studies by Hui et al. [40], Niwa et al.[44] and Mitamura et al.[37] it was expected that at this level of doping the disordered structure of the material would be stabilized at lower temperatures, and hence electrical conductivity improved, as discussed in Chapter 2, Section 2.3.4. Only one composition of Ba2In1Ce0.5La0.5O5.25 was prepared with a higher content of dopants to investigate the effect of the higher dopant level on material properties. For comparison purposes, non-doped Ba2In2O5 was also synthesized by both procedures. All compositions and their acronyms are listed in Table 3.1. Table 3.1 – List of prepared samples with compositions and acronyms used.       Composition Acronym Ba2In2O5 BIO Ba2In1.5Ce0.5O5.25 BIC Ba2In1.5Ce0.25La0.25O5.125 BICL025 Ba2In1Ce0.5La0.5O5.25 BICL05 Ba2In1.5La0.5O5 BIL   50 Glycine-nitrate combustion process (GNP): The glycine-nitrate combustion process was used to produce samples of all five compositions. Glycine (NH2CH2COOH) (Alfa Aesar), which serves as a complexing agent and a fuel, was dissolved in distilled water. Ba- acetate, (Ba(OOCCH3)2), Ce-, La- and In-nitrate (Ce(NO3)3 x 6H2O, La(NO3)3 x 6H2O and In(NO3)3 x XH2O, respectively), all from Alfa Aesar were added in their appropriate stoichiometric ratio. An example of the stoichiometric reaction for the case of Ba2In2O5 preparation is given below: 2Ba(OOCCH3)2+2In(NO3)3·5H2O+6.5O2 6NO2+16H2O+Ba2In2O5+8CO2         (3.1) Glycine was added in the amount to achieve a glycine/total-nitrates ratio of 5. Such a high glycine/nitrate ratio was necessary in order to prevent precipitation of Ba-acetate. The solution was heated with stirring in a glass beaker for 2 hours until all salts were fully dissolved. The solution was then transferred into a 5 liter stainless steel pot and heated on a hot plate until the viscous liquid started bubbling and forming foam. Usually, the GNP process is reported to end with a very fast combustion and flame, resulting in a fully reacted fine ash in the combustion plume [92]. In our case, the combustion happened slowly, forming a dark, very porous ash of partially reacted precursors. Additional calcining in a crucible at a high temperature (in the range of 1100-1500oC) for 6 hours was needed to achieve a pure crystalline phase. To determine the temperature at which the desired brownmillerite phase was formed, samples of all five compositions were calcined to 1100oC, 1200oC, 1300oC, 1400oC and 1500oC for 6 h. The phase obtained after calcination at each temperature was determined by X-ray diffraction at room temperature. It was important that, during the calcination, the crucible was open to air to allow for the release of CO2.  CO2 in contact with the samples during calcinations could cause the decomposition of the samples into BaCO3  51 and In2O3. Very fine powders were obtained in this procedure. Schematic representing steps in the GNP method, as well as pictures showing different stages of the GNP process are shown in Figure 3.1.                    Figure 3.1 – Stages of the glycine-nitrate process in this work: Schematic of the steps (a) Precursor solution boiling and thickening; (b) Porous ash formed; (c) Ash transferred into a crucible; (d) Completed reaction after heating in a furnace. Precursors (nitrates/acetates) + glycine Dissolving in water Heating on a heater (~350oC) Combustion – forming porous ash Calcination in a furnace to 1100-1500oC    (a) (b) (c) (d)  52 Solid-state reaction process (SS): Stoichiometric quantities of nitrate precursors were ball-milled in ethanol (94-96%, Alfa Aesar) overnight. Since all nitrate precursors were hydrates, upon mixing with ethanol they released water from their crystal structure. Also, some of the precursors, like In(NO3)3 x 5H2O, were hygroscopic and they absorbed moisture from the air. Therefore, it was not possible to dry the samples at room temperature. Thermogravimetric analysis (TGA) was performed to determine the best drying temperature. It was concluded that the precursor mixture lost water (both absorbed and crystal) around 150oC. All samples were subsequently heated to 150oC overnight to release the water. The dried samples were ground, pressed into pellets to achieve better contact, and calcined to a high temperature (in the range of 1100-1500oC) for 10 hours when the brownmillerite structure is formed. Grinding, pressing and calcining steps had to be repeated two times in order to achieve a pure phase. Coarser powders (~ 1.5 µm particle size) were obtained with this solid-state reaction procedure, compared to the GNP process (150 nm). The steps of the solid-state process applied in this work are shown in Figure 3.2.           53           Figure 3.2 – Schematic of the solid-state process steps used in this work. 3.3 Materials characterization 3.3.1 X-ray powder diffraction As-calcined powder samples prepared by both the solid-state and GNP methods were investigated by X-ray powder diffraction at room temperature on a Bruker AXS D8 X-ray diffractometer with a CuKα source, using a 0.02o step increment and 0.1 sec/step rate. Ba2In2O5, as a well investigated composition in the literature, was used in this work as a reference, and all XRD patterns were visually matched and compared to its pattern. 3.3.2 Particle size analysis Particle size analysis was performed by a Malvern Zetasizer ZEN 3600 particle analyzer capable of measuring particle size in the range between 0.6 nm to 6 µm. Prior to the measurement, powder samples were sonicated in ethanol for 3 minutes to break up any Precursors Mixing in EtOH Drying overnight at 150oC Grinding and pressing into a pellet Calcining to 1100-1500oC for 10h Re-grinding and pressing into a pellet Calcining to 1100-1500oC  54 agglomerates of particles. The dispersant Emphos PS-236 (Akzo Nobel) was used to prevent powders from agglomerating during the measurement. Particle morphology and size was further investigated by a FEI TECHNAI high resolution transmission electron microscope (HR-TEM) and an Hitachi S-3500N scanning electron microscope (SEM). Samples for the TEM were prepared as a 0.01 wt% powder sample in ethanol (no dispersant was used) and sonicated for 3 minutes prior to application on a sample slide. The elemental analysis of the TEM samples was done by energy-dispersive X-ray spectroscopy (EDS). 3.3.3 Thermogravimetric analysis and differential scanning calorimetry The thermal properties of the powder materials were evaluated using a SETARAM Setsys Evolution thermal analysis instrument. The thermogravimetric analysis (TGA) of the as-synthesized powders was performed to measure the weight change as a function of temperature, while differential scanning calorimetry (DSC) was used to determine possible phase changes, as well as the order-disorder temperature. TGA and DSC tests were carried out in air in an alumina crucible up to 1500°C with a heating and cooling rate of 5oC/min. X- ray diffraction measurements were performed before and after the thermogravimetric analysis, to identify any phase change during heating. 3.3.4 Temperature-profile X-ray diffraction Phase changes and order-disorder transition temperatures were also confirmed by temperature-profile X-ray diffraction, when as-prepared samples with a starting brownmillerite structure were heated in a chamber in air for three hours at 400oC, 600oC, 800oC, 1000oC, 1300oC, 1400oC and 1500oC and X-ray patterns were recorded at each  55 temperature. The exception was for measurement at 1500oC, where samples were kept for only 1 hour, due to the safety limitations of the instrument at such a high temperature. 3.3.5 Stability in humid atmospheres  During the experimental work on the five compositions, it was noticed that they show some instability in humid atmospheres. In order to investigate the effect of water vapour on the prepared powder samples in this work, they were exposed to air from a humidifier with relative humidities (RH=Pw/PsatT ·100%, where Pw is the partial pressure of water vapour and PsatT is the saturation pressure of water at temperature T) of 50% (which corresponds to e.g., 15 mol% of water vapour in air or Pw=0.15 atm at 70oC) and 90% (which corresponds to e.g., 28 mol% water vapour in air or Pw=0.28 atm at 70oC) and at different temperatures of 25oC, 50oC, and 70oC for 24 hours. The setup for this experiment consisted of a Setaram Wetsys humidifier and a controlled atmosphere furnace connected to it. The humidifier and the furnace were set to the same temperature. Humidified air flowed through heated lines (to prevent condensation) at a rate of 50 sccm (standard cm3/min) and passed the sample in the furnace. After 24 hours under these conditions the samples were tested by X-ray diffraction to determine if any change in composition had occurred. The effect of humidity at higher temperatures in the range between 100oC and 500oC, which could not be achieved by the Wetsys humidifier, was investigated by flowing 100 sccm air through a bubbler at room temperature and over the tested samples in a closed tube furnace, which was heated to a desired temperature between 100oC and 500oC. The humidification achieved was approximately 3 mol %. This level of humidification corresponds to a RH close to 100% at room temperature, but at higher temperatures, between 100oC and 500oC, the relative humidity drops to  RH=3% and below. The later conditions  56 were investigated as likely operational conditions for an intermediate temperature fuel cell where the investigated materials would be potentially used. The stability of the materials was also checked at temperatures below 100oC for the same humidification level (3 mol%) introduced by this method. In addition to the previous two methods, the effect of liquid water on the samples was tested. The samples were immersed in water in a beaker and heated on a hot plate to 25oC, 50oC and 70oC for 24 hours. For all the above cases, X-ray diffraction was used to determine the composition of the samples before and after exposure to humidification. 3.4 Preparation of samples for conductivity and e.m.f. measurements  Pellets for conductivity measurement were prepared from powders made by the GNP process or solid-state reaction. Powders made by the GNP process were hard to handle due to their fine particle size (~150 nm). Therefore, before pressing into pellets and sintering, the GNP powders had to be calcined to 1300oC for 6 h to pre-sinter the particles and make powder with larger particle size for easier handling. Powders made by the solid-state reaction did not require the additional calcining. 2wt% of polyvinyl butyral (PVB) binder (Richard E. Mistler Inc) dissolved in ethanol (10 wt% solution) was added to the powders to help with pellet pressing. The powders were pressed in a 20 mm die using a pneumatic press under 150 MPa pressure. The pellets were sintered at different temperatures, depending on the composition and powder preparation method. A short study was performed to determine the appropriate sintering temperatures for the samples. Table 3.2 shows a summary of the sintering temperatures and achieved pellet properties for different powder samples. The temperatures needed to be sufficient to sinter the samples to a satisfactory density, but low enough to preserve the desired brownmillerite structure (some compositions changed their  57 structure to a cubic phase at higher temperatures). The BICL05 sample was impossible to make into pellets, as the samples always fractured into pieces. Hence, this material was not tested for conductivity. Fig. 3.3 shows the microstructure of a BIC sample sintered at different conditions. The final phase of the sintered GNP and SS samples was confirmed by the X-ray diffraction. Scherrer equation [93] given below was used to calculate the final grain size of the pellets made by the two different methods based on the peak broadening in the X-ray patterns. The grain size D is given by:  θ λ cos 89.0 2 0 2 WW D − =                                                                                       (3.2)  where λ is the wavelength of the Cu source used in XRD, W is the peak width at half of the diffraction peak height (peak with the highest intensity was used, but also compared with several lower intensity peaks), Wo is the instrumental broadening, and θ is the half value of the 2θ angle of diffraction. The instrumental broadened profile was obtained from the measurement of the standard sample NIST SRM 1976 Alumina (National Institute of Standards and Technology – NIST, USA). The grain size of the GNP and SS sintered samples was confirmed by SEM. Sintered pellets were polished (final polishing was with a 1 µm cloth) and etched using 0.3% HNO3 for 1 min. Cross-section of the pellets was observed using a high resolution SEM (FEI DualBeam Strata 235) in order to confirm the grain size of the samples. The density of the pellets was determined by the Archimedes principle in ethanol (94- 96%, Alfa Aesar, ρ20oC=0.7893 g/cm3) and only pellets with a similar total porosity of about  58 20% were chosen for conductivity measurements, in order to minimize the effect of the porosity factor on the conductivity results. Samples with a lower porosity than 20% could not be prepared for all compositions. The thickness of the sintered pellets was 0.8-1.5 mm. They all had a final brownmillerite structure.       Figure 3.3 – SEM pictures of a Ce-doped Ba2In2O5 sample prepared by the GNP method and sintered at (a) 1300oC for 6 h, (b) 1350oC for 6 h and (c) 1400oC for 6 h. Scale 30 µm.  In order to investigate the effect of sample porosity on the conductivity, BIO-GNP and BIC-GNP samples with higher porosities than 20% were prepared by mixing the powders with pore-former graphite flakes 7-10 µm (Alfa Aesar) before sintering. A short study was performed to determine the amount of the pore-former needed to achieve the desired porosity of the samples. Fully dense samples could not be prepared and the lowest porosity obtained (not using the pore-former) was between 16 and 20%. Samples with porosities ~ 20%, 30%, 40% and 50% were prepared and used for the study. Grain sizes were similar for all prepared samples, about 40 nm, to avoid any effect of the grain size on the conductivity. Table 3.2 shows the properties of the porous samples used for this study. Samples for the e.m.f., open circuit voltage and polarization curve measurements were prepared in the same way as described above for the 20% porous samples, but with a larger a b c 30µm 30µm 30µm  59 diameter of about 25-28 mm (to fit the larger diameter of the testing setup), and a thickness of about 2 mm, to avoid gas leakage through the sample. This thickness was selected as the most appropriate by testing the gas leak of the samples with different thicknesses and selecting the samples that did not show the leak higher than 0.05 sccm. Leak tests were performed for these samples by applying 1 psi pressure of He on each sample in a sealed compartment and measuring the leak rate using a mass flow meter. The composition and phase of each sample before conductivity measurement was confirmed by X-ray diffraction. The microstructures of the sintered samples were examined using an Hitachi S-3500N scanning electron microscope (SEM). Pt paste was applied on both sides of the sintered samples to act as contact electrodes (area ~ 0.5 cm2). A picture of a testing sample is shown in Fig. 3.4. Table 3.2 – Summary of the sintering temperatures and pellets properties.          Sintered sample (compos.-method) Sintering T (6h) (oC) Grain size (nm) Porosity (%) BIO-GNP 1350 42 18 BIO-SS 1400 57 20 BIC-GNP 1350 42 20 BIC-SS 1400 60 21 BICL025-GNP 1300 44 20 BICL025-SS 1300 61 19 BIL-GNP 1350 42 20 BIL-SS 1400 65 20 BIO-GNP - 20 1350 42 18 BIO-GNP - 30 1350 40 32 BIO-GNP - 40 1350 42 41 BIO-GNP - 50 1350 41 51 BIC-GNP - 20 1350 43 21 BIC-GNP - 30 1350 41 29 BIC-GNP - 40 1350 40 42 BIC-GNP - 50 1350 41 53  60 3.5 Ac impedance spectroscopy and conductivity measurements The conductivity measurements for four compositions (BICL05 testing pellet could not be prepared) made by both the GNP and the solid–state process were performed in air and hydrogen-containing atmospheres by ac impedance spectroscopy. Ac impedance spectroscopy principles are discussed in Appendix B.1. The measurements were carried out using an impedance analyzer (IM6 by Zahner Electrinks), in the frequency range from 100 mHz to 8 MHz and with an amplitude of 50 mV, and in the temperature range from 100oC to 500oC. Samples were held at each temperature for 3 hours to achieve equilibrium, and impedance measurements were performed every 30 minutes over these 3 hours. Before the actual measurements several tests were performed to determine which part of the impedance spectra was associated with the materials resistivity, and which part with the electrolyte/electrode charge transfer. In order to determine the part of the impedance spectra associated with the electrode charge transfer, samples with the same characteristics (composition, porosity, grain size, thickness) were prepared and different electrodes were applied. For testing in air, one set of samples was prepared with Pt paste electrodes treated to 800oC for 30 min before use, and the other set of samples with Au electrodes sputtered by a Polaron Sputter Coater SC7640. In the hydrogen-containing atmosphere, Au electrodes did not perform well and in this case two different electrodes were used: in one case Pt paste was applied and treated to 800oC for 30 min, as in the previous study, while in the other case the paste was only dried in the oven to 80oC.  In the latter case, the electrode was still active but its activity was expected to be somewhat affected by the microstructure and additives in the Pt paste mixture that did not fully decompose at such a low temperature. Different electrodes were expected to result in a different size of the semicircles associated to the  61 electrolyte/electrode interface impedance, while semicircles associated to the material contribution should remain unaffected. These tests were done for all compositions prepared by the GNP and the solid-state reaction in air and 50% H2/50% N2 atmosphere in the temperature range from 100oC to 500oC.  After the preliminary tests described above, the electrical conductivities of the samples were assessed in the temperature range from 100oC to 500oC in two different atmospheres: air (supplied from the atmosphere and dried containing only up to 15 ppm H2O) and a hydrogen/nitrogen mixture (50% H2/ 50%N2). Nitrogen and hydrogen used in the experiments (Praxair) contained a maximum of 3 ppm (vol) of water and a maximum of 3 ppm oxygen (in N2) and 1 ppm oxygen (in H2). The ceramic samples were exposed to the same atmosphere on both sides with a gas flow rate of 100 sccm. The temperature and gas flow rate were programmed using an AMEL 7902 test setup. The picture and the schematic of the experimental setup are shown in Fig. 3.4. For each composition at least three measurements (three fresh samples with similar properties, but each from a different synthesis batch) were done with the same testing conditions in order to confirm reproducibility of the results and the standard deviation was determined. Some more specific tests were carried out for undoped Ba2In2O5, as the selected material with the highest conductivity in a hydrogen-containing atmosphere. The conditions of these specific tests are discussed in Chapter 6.      62                       Figure 3.4 – Ac impedance analyzer system and pellet for conductivity testing (top); AMEL 7902 test setup. AC impedance Furnace Sample holder Humidity control Power and atmosphere control unit               Testing sample ~18 mm ~5 mm Pt paste 1) Fused silica pipe 2) Alumina anode support 3) Gas spreader and platinum mesh anode contact 4) Gas spreader and platinum mesh cathode contact (pulled from the bottom of the cell) 5) Sample position 6) Anode inlet gas duct 7) Pair of Pt contact leads on both sides of the sample 8) Water cooled support plate 9) Cathode gas inlet 10) Anode gas inlet 1 2 3 4 5 6 7 8 9 10  63 3.6 Stability in hydrogen-containing atmospheres The stability of Ba2In2O5, as a selected material with the highest conductivity, in a hydrogen-containing atmosphere was investigated in the temperature range between 300oC and 500oC, using the same setup as in the conductivity testing. This temperature range was chosen for stability evaluation because it gave the highest conductivities. Samples were heated to temperatures of 300oC, 350oC, 400oC, 450oC and 500oC in a 50% H2/50% N2 atmosphere for 24 h and ac impedance measurements were taken after 2, 8, 16 and 24 h dwell times at each temperature. X-ray diffraction measurements were performed for each sample after the testing to confirm the stability of the Ba2In2O5 under the applied conditions. In order to confirm that there was no reduction of indium during the treatment in hydrogen, Raman analysis (XploRATM, Horiba Jobin Yvon) at ambient conditions, with the 532 nm laser, was performed on the samples before and after testing in hydrogen. Vibration modes characteristic for different oxidation states of indium were expected. 3.7 Proton transport number determination The proton transport number, as explained in Chapter 2, Section 2.3.1, shows the contribution of the proton conductivity to the total conductivity of a sample and is given as a ratio between proton conductivity and total conductivity, tH+=σH+/σt. Proton transport numbers, tH+, were determined in this work by the e.m.f. measurement method, using a concentration cell PH2I, Pt║Ba2In2O5║Pt, PH2II, shown in Fig. 3.5. The principles of the e.m.f. measurement are explained in Appendix B.2. Ba2In2O5 (BIO) prepared by the GNP method was chosen for testing because it demonstrated the highest total conductivity in hydrogen- containing atmospheres. A sintered, leak-free, sample disc of BIO with Pt electrodes was sealed to the alumina tubes on both sides of the concentration cell to avoid any leakage out of  64 the cell. The e.m.f. of the cell was measured by a high impedance MultiSTAT 1480A- Solartron. Measurements were taken at 100oC, 200oC, 300oC and 400oC. Measurements were not performed at temperatures above 400oC, as BIO samples showed mechanical instability when exposed to water vapour at those temperatures. Water vapour was used on both sides of the concentration cell to prevent any possible surface reduction in hydrogen-containing atmosphere, which would affect the e.m.f. measurement. The partial pressure of H2 in both compartments was controlled by mixing H2 with N2 gas and water vapour at 1 atm. The water vapour concentration in the compartment I was controlled by flowing H2/N2 mixture (at 100 sccm) through saturated vapour produced in a hot vessel at 115oC. The amount of saturated vapour in the vessel was controlled by changing the flow of the liquid water to the vessel. The water vapour concentration in the compartment II (3 vol% at all times) was set by flowing H2/N2 mixture (at 100 sccm) through a bubbler at room temperature. The composition on the PII side (low PH2) was kept constant at 48%vol H2/49%vol N2/3%vol H2O. The composition on the PI side was 48%vol H2/49%vol N2/3%vol H2O at the beginning of each test, to determine the reference voltage when there was no gradient across the sample. This reference voltage included all unwanted voltage contributions (static charge, thermovoltage, etc.) and was deducted from the measured voltages when the gradient was present. After determining the reference voltage, the concentration on the PI was changed to 80%vol H2/15%vol N2/5%vol H2O, and the e.m.f. across the sample measured. The ratio of the activities of water vapour on two sides of the compartment was set to be the same as the ratio of hydrogen, so there was no gradient in oxygen activity. Measurement at each temperature was repeated three times and the standard deviation was determined. The experimental setup is shown in Fig. 3.5.  65                       Figure 3.5 – Experimental setup for the e.m.f. testing and a schematic of an e.m.f. concentration cell for ion transport number determination. Humidifier for compartment I Compartment II Bubbler for compartment II Syringe and pump to supply H2O to anode humidifier BIO sample sealed with Ceramabond to the compartment II alumina tube Pt mesh for contact and Pt leads on both sides Furnace Compartment I       High PIH2 Low PIIH2 Ba2In2O5 dense disk with Pt electrodes I II H2 H+ H2  66 3.8 Open circuit voltage, polarization curve and potentiostatic measurements The schematic of the setup for the OCV, polarization curve and potentiostatic measurement is shown in Fig. 3.6. An electrochemical cell: air, Pt║Ba2In2O5║Pt, 50%vol H2/50%vol N2 was constructed using a leak-free Ba2In2O5 sintered pellet with applied Pt electrodes. Gases were passed at 100 sccm on both sides of the cell and the fuel mixture was humidified by passing it through a bubbler at room temperature. Open circuit voltages (OCV) at 100oC, 200oC, 300oC, 350oC, 400oC, 450oC and 480oC were measured for 3 h at each temperature using a Solartron MultiSTAT 1480A (Solartron Analytical, Farnborough, Hampshire, United Kingdom). More frequent measurements between 300oC and 480oC were performed as this is the temperature range where Ba2In2O showed the highest conductivity, hence it is of higher interest. Measurements above 480oC were not performed due to the chemical instability of the material at those temperatures. The test was repeated three times, every time with a new Ba2In2O5 sample, and average value for OCV at each temperature was taken. Full cell polarization and impedance tests were performed at 300oC, 350oC, 400oC 450oC and 480oC, using a Solartron 1260 Frequency Response Analyzer coupled with the Solartron MultiSTAT 1480A. As mentioned earlier, this range of temperatures was selected as a range where Ba2In2O5 showed the highest conductivity and stability in hydrogen- containing atmospheres, as well as stable values of OCV. Ac impedance was recorded from 100 mHz to 8 MHz and with an amplitude of 50 mV. The polarization curves were recorded twice at each temperature. The voltage was changed from the initial potential of OCV to 0 V and back to a final potential of OCV, at a rate of 4 mV/s.  67 Potentiostatic measurements were performed using a Pt║Ba2In2O5║Pt cell to further investigate the electrochemical behaviour of Ba2In2O5 electrolyte when exposed to hydrogen and oxygen atmospheres. The measurements were performed at 300oC, 350oC, 400oC 450oC and 480oC. Before the measurements at each temperature, 50%H2/ 50%N2 was flowed on the anode side (with N2 on the cathode side) until a high conductivity was reached, as confirmed by the impedance measurement. A constant potential of 0.5 V was applied across the cell and current was measured while the atmosphere was changed from 50%H2/ 50%N2 on the anode side, with N2 (no air) on the cathode side to 50%H2/ 50%N2 on the anode side, with 100 sccm air flow on the cathode side. The air flow on the cathode side was started and stopped three times, and the change in current recorded. Ac impedance measurements were performed at the end of the experiments each temperature, with 50%H2/ 50%N2 on the anode side and air on the cathode side.             68             Figure 3.6 – Setup for OCV and polarization curve testing: (1) Outer rod with sealed testing cell; (2) Inner spring loaded rod for fuel supply; (3) Three-hole rod with two Pt wires; (4) Pt wires-one to measure current, other for potential determination; (5) Pt mesh for contact; (6) Sample cell for testing; (7) Cathode side of the cell; (8) Fuel out; (9) Cathode side spring loaded rod for O2 supply; (10) Three-hole rod with two Pt wires; (11) O2 out; (12) Pt wires- one to measure current, the other for potential determination; (13) Thermocouple.  3.9 Preparation of testing samples for anode study This study investigated application of Ba2In2O5 as a proton conductive support for different metal catalysts within the anode of an intermediate temperature fuel cell. Platinum, nickel and iron were considered as potential metal catalysts. Eight different anodes were prepared and applied to a sintered Ba2In2O5 (BIO) electrolyte to form a symmetrical cell for the testing. The anodes were treated to two different temperatures prior to testing, i.e., 800oC H2/N2 supply line Humidifier Furnace Fuel out Air supply line  69 for 30 min and 1300oC for 6 h. The temperature of 800oC was used to provide partial sintering of the metal catalysts, but still retain the porosity of the anode. BIO support in the anode structure was not sintered in this case. The temperature of 1300oC was used to sinter the BIO material in the electrodes in order to form a connected path for proton transport within the anode. The effect of the BIO support sintering on the anode performance was investigated in this case and compared to the non-sintered case. Eight different anodes were prepared as follows: 1) Sample Pt-800oC (baseline sample): Pt paste for ceramic substrates (Alfa Aesar) was applied to the sintered BIO electrolyte on both sides (electrode area 0.25 cm2) to form a symmetrical cell Pt II BIO II Pt. The sample was heated to 800oC for 30 min. 2) Sample Pt-1300oC: Symmetrical cell was prepared in the same way as for sample 1, but it was heated to 1300oC for 6 h. X-ray diffraction was used to confirm that Pt and BIO remained stable. 3) Sample Pt+BIO–800oC: Electrodes for the BIO+Pt II BIO II BIO+Pt symmetrical cell were prepared by mixing the Pt paste (80% Pt content, as stated in the product material safety data sheets) with the BIO powder. The amounts of the Pt paste and the BIO powder were calculated to achieve a 30 wt% Pt/ 70 wt% BIO final mixture.  This is the metal catalyst/ceramic ratio commonly used in the SOFC anode structure, as explained later in Section 8.3. A small amount of the PVB binder and ethanol were added to the Pt+BIO mixture to prepare a smooth paste for application. The Pt+BIO paste was applied to both sides of the sintered BIO pellet and heated to 800oC for 30 min.  70 4) Sample Pt+BIO–1300oC: The symmetrical cell was prepared in the same way as sample 3, but heated to 1300oC for 6 h. X-ray diffraction was used to confirm that Pt and BIO were stable. 5) Sample Ni+BIO–800oC: Ni catalyst powder was prepared by the glycine-nitrate process using Ni(NO3)2·6H2O (Alfa Aesar, 98%) and glycine. The obtained powder after the process was heated in a furnace to 500oC for 6 h to complete the reaction. The final powder contained approximately 85% Ni and 25% NiO, as confirmed by X- ray diffraction. This powder was mixed with the BIO powder in such a proportion to result in a 30 wt% Ni/ 70 wt% BIO final mixture (after reduction). A small amount of PVB and ethanol were added to the powder mixture to form a smooth paste, which was applied to a sintered BIO pellet on both sides to form a BIO+Ni II BIO II BIO+Ni symmetrical cell and heated to 800oC for 30 min. The composition of the treated anode was confirmed to be stable by X-ray diffraction. The anode was reduced to elemental Ni in 50%H2/50%N2 at 450oC before testing. 6) Sample Ni+BIO–1300oC: This sample was prepared in the same way as sample 5, but heated to 1300oC for 6 h before testing. X-ray diffraction of the heated sample confirmed decomposition of the BIO powder within the anode, most likely catalyzed by the presence of Ni catalyst. Therefore, this sample was not used for further testing. 7) Sample Fe+BIO–800oC: Fe catalyst powder was prepared by the glycine-nitrate process using Fe(NO3)3·9H2O (Alfa Aesar, 98%) and glycine. The obtained powder after the process was heated in a furnace to 500oC for 6 h to complete the reaction. X- ray diffraction of the obtained powder confirmed that it contained mainly Fe2O3. This powder was mixed with the BIO powder to prepare a 30 wt% Fe/ 70 wt% BIO final  71 mixture (after reduction). A smooth paste was formed using this powder and a small amount of PVB and ethanol. The paste was applied to a sintered BIO pellet on both sides to form BIO+Fe II BIO II BIO+Fe symmetrical cell and heated to 800oC for 30 min. Composition of the treated anode was confirmed to be stable by X-ray diffraction. The anode was reduced to elemental Fe in 50%H2/50%N2 at 450oC before testing. 8) Sample Fe+BIO–1300oC: This sample was prepared in the same way as sample 7, but heated to 1300oC for 6 h before testing. X-ray diffraction of the heated sample revealed that a reaction between Fe and BIO powder happened, forming BaFe0.5In0.5O0.25. Hence, this sample was also not used for further testing.  Six successfully prepared anode samples incorporated in symmetrical cells were tested by ac impedance spectroscopy in 50%H2/50%N2 at temperatures between 300oC and 480oC.             72 4. Material Characterization of Undoped and Ce- and La-doped Ba2In2O5  4.1 Synopsis Before investigation of the electrical properties of the selected compositions in this work, basic material properties such as crystal structure and composition, particle and grain characteristics, thermal properties, stability, etc., of the prepared materials were studied. Most of these properties are closely related to the electrical behavior of the materials, and therefore their understanding can provide information on the mechanisms and processes of electrical conduction. This chapter discusses material properties of the five different compositions investigated in this work. The crystal phase of the synthesized materials was confirmed by X-ray diffraction (XRD), while powder particles were characterized by particle size analysis (PSA), high resolution transmission electron microscopy (HR-TEM) and scanning electron microscopy (SEM). Phase changes of the materials, transition order- disorder temperature and weight changes with temperature were studied by high-temperature XRD, thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC). Stability of the materials was tested in humid air and liquid water and the composition after exposure was checked by XRD. 4.2 Crystal phase of the materials Five different compositions listed in Table 3.1 were prepared by the solid-state and glycine-nitrate process and their crystal phase determined by X-ray diffraction. Typical X-ray diffraction patterns of the prepared powder samples for all five compositions are given in Fig. 4.1a and b, and indexed X-ray pattern for Ba2In2O5 is shown in Appendix A. As shown in Fig. 4.1, the samples could be indexed to either an orthorhombic brownmillerite or cubic  73 perovskite structure, depending on the calcining temperature. Comparable scans were obtained for powders produced by the two different methods, the GNP and the solid-state method. However, the calcining temperatures needed to achieve particular phases for the powders produced by the GNP method or by the solid-state method were different, as shown in the figure. When produced by the GNP process, all compositions except BIC, needed a 100oC higher temperature to achieve the brownmillerite structure than the samples produced by the solid-state method. During the GNP process the material was in the form of a loose powder, but during the solid state process powders were pressed in pellets. Close contact of the powder particles in the pellets during the solid-state process enabled better heat and material transport, lowering the temperature needed to complete the reaction. All compositions, produced by both methods, achieved the cubic perovskite structure at 1500oC, except for BIL produced by the solid-state reaction, which formed this structure at 1300oC.             74                Fig. 4.1 – X-ray powder diffraction patterns for five different compositions made by both the solid-state reaction and glycine-nitrate process: (a) orthorhombic brownmillerite structure; (b) cubic perovskite structure. The temperatures needed to achieve a particular phase are given for both the GNP and the solid-state (SS) method of preparation. Note: Indexed X-ray diffraction patterns for the orthorhombic brownmillerite and cubic perovskite structure for Ba2In2O5 are given in Appendix A, Fig. A.2 and A.7.  The Scherrer equation, explained in Chapter 3, Section 3.4, was used to calculate the crystallite (grain) size of the powder samples made by the two different methods, based on the peak broadening in the X-ray patterns. Powders produced by the GNP method, right after combustion and before additional calcining, had a starting grain size of about 10 nm. During the additional calcining of these powders to different temperatures, the grains grew up to about 70 nm at 1500oC. Fig. 4.2 shows the effect of calcining temperature on the grain size of        BIL   1350oC       1250oC    BICL05    1400oC       1300oC  BICL025    1300oC       1200oC        BIC   1100oC       1100oC        BIO  1300oC       1200oC        BIL  1400oC       1300oC    BICL05   1500oC       1500oC  BICL025   1500oC       1500oC        BIC  1500oC       1500oC        BIO  1500oC       1500oC               GNP             SS               GNP             SS  75 the powders produced by the GNP method. The grain size of the powders appears to change more significantly above 1300oC. BIO and BIC achieved somewhat lower final grain size than BICL05, BICL025 and BIL at 1500oC. The starting grain size of the powders made by the solid-state reaction was determined to be around 40 nm. This larger grain size compared to the GNP method is due to the fact that the solid-state reaction involved heating to 1300oC until desired composition and structure was achieved. Further heating to higher temperatures gave similar results for grain size as for the GNP powders.           Fig. 4.2 – Grain size as a function of calcining temperature for powders produced by the GNP process. 4.3 Powder particle characterization Particle size distribution of the prepared powders was determined using a Malvern Zetasizer ZEN 3600 particle analyzer. A typical particle size distribution for as-synthesized 0 10 20 30 40 50 60 70 80 1000 1100 1200 1300 1400 1500 1600 Temperature (oC) G ra in  si z e (nm ) BIO BIC BICL025 BICL05 BIL  76 Ba2In2O5 (BIO) powders produced by the GNP process and the solid-state reaction are shown in Fig. 4.3 a and b. Powder produced by the GNP method had particles in the range between 100 nm and 500 nm, with the maximum number of particles around 165 nm. All other powders produced by the GNP method had similar particle size, with 151 nm being the mean particle size for all powders. BIO powder produced by the solid state reaction showed a bimodal distribution in the range between 1000 nm to 7000 nm, with the maximum number of particles with sizes of about 1485 nm and 5560 nm. All other solid state powders showed similar results, with the mean particle size of 1530 nm for the first maximum and the mean particle size of 5658 nm for the second maximum. Fig. 4.3.b also shows the particle size distribution for the BIO powder prepared by the GNP method and additionally calcined at 1300oC for 6 h to achieve a powder suitable for handling and sintering. Somewhat lower particle sizes were obtained by this method compared to the solid-state method. The TEM image in Fig. 4.4 shows an agglomeration of crystalline particles of a BIC powder produced by the GNP process and a selected 15 nm grain. Fig. 4.5 compares the morphology of an as- prepared GNP powder and a solid-state powder.  The GNP powder appears to form easily breakable clusters of fine particles, while the solid-state powder forms hard agglomerates. Other compositions had similar morphology. Energy-dispersive X-ray spectroscopy (EDS) was used to perform elemental analysis of the TEM samples and confirm if the powders had the desired compositions (as listed in table 3.1). Appropriate weight ratios for the metals contained in the powders were confirmed. Oxygen content was higher than stoichiometric (e.g., for Ba2In2O5 the oxygen content was about 5 times higher). This is most likely due to the ability of these materials to absorb oxygen and have variable oxygen stoichiometry, as discussed later in Section 4.5.  No impurities were detected.  77              Fig. 4.3 – Particle size distribution for Ba2In2O5 (BIO) powder: (a) as-synthesized by the GNP method; (b) prepared by the solid-state method (thick line) and after calcination of the GNP powder at 1300oC for 6h (thin line).        Fig. 4.4 – TEM image of as-synthesized Ce-doped Ba2In2O5 (BIC) powder prepared by the GNP process: a selected grain with crystal fringe (left); a selected particle (right). 0 5 10 15 20 25 30 0 1000 2000 3000 4000 5000 6000 7000 8000 Size (nm) N u m be r (% ) BIO-SS BI0-GNP-1300 0 2 4 6 8 10 12 14 16 18 0 100 200 300 400 500 600 700 800 900 1000 Size (nm) N u m be r (% ) BI0-GNP-as prepared b a 50 nm 5 nm  78  Fig. 4.5 – SEM images showing morphology of the BIC powders: hard agglomerates produced by the SS method (left); easily breakable clusters produced by the GNP method (right). Scale bar 100 µm.  4.4 Temperature-profile XRD It has been reported [34, 38, 79] that, at a certain temperature, Ba2In2O5 –based materials undergo a phase transition from orthorhombic brownmillerite structure with ordered oxygen vacancies, to a disordered cubic perovskite structure. The temperature at which this change happens is the earlier mentioned transition order-disorder temperature, Td, and is also related to the change in conductivity of the materials. In order to investigate this change for all five compositions in the present study, temperature-profile XRD measurements were carried out. Each powder sample with a starting brownmillerite structure confirmed by XRD at room temperature was heated consecutively to 400oC, 600oC, 800oC, 1000oC, 1300oC, 1400oC and 1500oC. After 3 hours of equilibration at each temperature XRD was measured and heating to a higher temperature continued. Fig. 4.6a and b shows the changing X-ray diffraction patterns of the powder structures achieved with temperature for  79 the BIO and BIC samples. Structures achieved with temperature for all five compositions are listed in Table 4.1. Samples prepared by both the GNP and solid-state method showed the same behavior. BIO, with a starting brownmillerite structure transformed to a cubic perovskite structure between 800oC and 1000oC, in agreement with 925oC based on the literature [34, 38, 79]. However, when heated for 3 hours at 1300oC, the structure changed back to a brownmillerite structure. Finally, heating the BIO to 1500oC again caused it to change back to the cubic structure. BIC starts its change to a cubic structure at a lower temperature of around 400oC, while BICL025, BICL05 and BIL almost completely change to a cubic structure at around 400oC. In general, the Ce- and La- doped Ba2In2O5 experienced transition from an ordered brownmillerite structure to a disordered cubic structure at lower temperatures (Td) compared to undoped Ba2In2O5. This finding confirms the hypothesis that introduction of Ce and La into the Ba2In2O5 structure would decrease the order-disorder temperature of the material, as mentioned in Chapter 2, Section 2.3.4. BICL05 and BIL materials transformed back to the brownmillerite structure after 3h at 1300oC, while BICL025 transformed to the brownmillerite structure at 1400oC. All compositions changed back to the cubic structure at 1500oC. Although the phenomenon of the order-disorder transition for Ba2In2O5 and related materials has been reported before [34, 38, 79], the multiple shifts between the two phases have not been reported. In addition to the presented temperature-profile XRD, TGA and DSC results in the following section give us additional information on this effect.     80                  Fig. 4.6 – Temperature-profile XRD in air (a) Ba2In2O5 (BIO) and (b) Ce-doped Ba2In2O5 (BIC) sample, both with a starting brownmillerite structure. Note: B-brownmillerite structure; C-cubic perovskite structure.     20 30 40 50 60 70 80 2 Theta In te n si ty As prepared 600oC 1000oC 1300oC 1400oC 800oC 1500oC 400oC   (a) B B B B C B B C    BIO 20 30 40 50 60 70 80 2 Theta In te n si ty As prepared 600oC 1000oC 1300oC 1400oC 800oC 1500oC 400oC   (b) C C C C B B→C B C→B      BIC  81  Table 4.1 – Temperature-profile XRD in air measurement results showing the change in crystal structure with temperature.  B-Brownmillerite structure; C-Cubic perovskite structure * Estimated based on the temperature-profile XRD, TGA and DSC data  4.5 Thermogravimetric analysis and DSC Fig. 4.7 shows the comparison of the TGA/DSC results for the Ba2In2O5 powders prepared by the GNP process (BIO-GNP) and the solid-state reaction (BIO-SS). Similar results were obtained for samples made by the two processes. A small difference was observed in the TGA result, where GNP sample showed somewhat larger initial weight loss (~1.6%), compared to the solid-state sample (~1.4%). All other powders prepared by the solid-state process had similar TGA and DSC results compared to the powders prepared by the GNP process, even without a significant difference in the initial weight loss. Therefore, results for BIC, BIL, BICL025 and BICL05 made by the solid-state reaction are not shown here. Fig. 4.8 shows the TGA/DSC results for these four compositions made by the GNP process.    T(oC) RT 400 600 800 1000 1300 1400 1500 Td (oC) based on DSC BIO B B B B C B B C 925 BIC B BC C C  C B B C 480* BICL025 B C C C C C B BC 410 BICL05 B C C C C B B C 430 BIL B C C C C B B C 420*  82  Figure 4.7 – TGA (thick line)/DSC (thin line) results for (a) BIO made by the GNP process, (b) BIO made by the SS process. Experiments were performed in air, in the temperature range of 25o to 1550oC with a heating rate of 5oC/min. Transition order-disorder temperature Td, and points where loss of H2O and CO2 occur are marked on the graphs.  For all samples TGA reveals a significant weight loss in the 300-400oC temperature range. A sharp endothermic peak in the DSC curve is associated with this change (shown in the small insert in the corner of each picture). As reported in the literature [72, 74] this change is related to the loss of H2O incorporated into the crystal structure of the materials. Hashimoto et al. [74] have concluded that the incorporation of water in as-prepared Ba2In2O5 is due to the absorption of water from the atmosphere and that this process did not affect the crystal structure of the material. They have also reported that, when Ba2In2O5 was annealed in a humid atmosphere at 250oC, the content of the absorbed water increased compared to the as-prepared sample, and the crystal structure of the material changed to a tetragonal structure, -2.5 -2 -1.5 -1 -0.5 0 0 500 1000 1500T (oC) W ei gh t c ha n ge  (w t% ) -10 0 10 20 H ea t f lo w  ( µµ µµ V) DSC -1 0 1 200 400 600 800 DSC 7.5 8.5 1050 1100 (b) BIO-SS H2O CO2 Td -2.5 -2 -1.5 -1 -0.5 0 0 500 1000 1500T (oC) W ei gh t c ha n ge  (w t% ) -15 -5 5 15 H ea t f lo w  ( µµ µµ V ) DSC -5 -4 -3 -2 -1 0 1 800 1000 1200 DSC -0.5 0 0.5 200 400 600 -1 0 -2.0 -1.8 744oC 1090oC (a) BIO-GNP H2O CO2 Td Td Endo Exo  ( µµ µµ W )  ( µµ µµ W )  83 different from the orthorhombic brownmillerite structure. Schober et al. [72] have called this structure a β−structure, while the structure without water is called an α−structure. They also observed a peak in the DTA scan in the same 300-400oC temperature range. As mentioned earlier, BIO sample prepared by the GNP method experienced a larger weight loss (>1.5%) in this region, compared to the sample made by the solid-state method. This suggests higher accumulation of water in the grain boundary region then in the bulk material. Since other investigated compositions did not show difference between GNP and solid-state samples, it is possible that, in their case, water accumulates more in the bulk material. In addition, BIO samples, both the GNP and the solid-state, show smaller weight loss (~1.5%) in this region, compared to all other compositions (~2%). From Fig. 4.7 and 4.8 it is apparent that another smaller weight loss occurs at around 745oC for BIO, 525oC for BIC, 620oC for BICL025, 580oC for BICL05 and 660oC for BIL. This is probably related to loss of CO2, whose incorporation from air into the crystal structure of Ba2In2O5 is also reported by Hashimoto et al. [74]. This process is also observed on the corresponding DSC curves as an endothermic peak, as shown in the DSC inserts of Fig. 4.7a, b and Fig. 4.8a-d.         84  Figure 4.8 – TGA (thick line)/DSC (thin line) results (a) BIC, (b) BICL025, (c) BICL05 and (d) BIL, all made by the GNP process. All experiments were performed in air, in the temperature range of 25o to 1550oC with a heating rate of 5oC/min. Points where loss of H2O and CO2, and the order-disorder transition occur are marked on the graphs. Td* are estimated based on the temperature-profile XRD, TGA and DSC data.   -2.5 -2 -1.5 -1 -0.5 0 0 500 1000 1500T (oC) W ei gh t c ha n ge  (w t% ) -20 -10 0 10 20 30 H ea t f lo w  ( µµ µµ V) DSC -2 -1 0 1 200 400 600 800 TG -2.18 -2.14 -2.1 -2.06 650 750 850 (b) BICL025-GNP CO2 Td H2O -2.5 -2 -1.5 -1 -0.5 0 0 500 1000 1500T (oC) W ei gh t c ha n ge  (w t% ) -40 -20 0 20 40 60 80 H ea t f lo w  ( µµ µµ V) DSC -2 -1 0 1 2 200 400 600 TG -2.1 -2 -1.9 700 800 900 1000 (a) BIC-GNP H2O Td* Endo Exo CO2 -3 -2.5 -2 -1.5 -1 -0.5 0 0 500 1000 1500T (oC) W ei gh t c ha n ge  (w t% ) -20 -10 0 10 20 30 H ea t f lo w  ( µµ µµ V) DSC -2 -1 0 1 2 300 500 700 TG -2.3 -2.28 -2.26 -2.24 -2.22 -2.2 600 800 1000 (d) BIL-GNP H2O CO2 Td* -2.5 -2 -1.5 -1 -0.5 0 0 500 1000 1500T (oC) W ei gh t c ha n ge  (w t% ) -20 -10 0 10 20 30 H ea t f lo w  ( µµ µµ V) DSC -2 -1 0 1 2 200 400 600 TG -2.2 -2.1 -2 -1.9 -1.8 -1.7 -1.6 500 700 900 (c) BICL05-GNP H2O CO2 Td  ( µµ µµ W )  ( µµ µµ W )  ( µµ µµ W )  ( µµ µµ W )  85 In addition to the H2O and CO2 losses, the TG and DSC results for all of the investigated materials revealed changes due to the order-disorder transition and change of oxygen stoichiometry. As discussed earlier, the temperature-profile XRD results for BIO, BIC, BICL025, BICL05 and BIL, show phase changes between ordered brownmillerite and disordered cubic perovskite structures at the temperatures listed in Table 4.1. Based on these results it was expected that the associated characteristic endothermic peaks on heating and exothermic peaks on cooling for reversible order-disorder transition would be observed in the DSC [94-96]. The endothermic peak was observed for the BIO sample at 925oC on heating and an exothermic peak at 1090oC on cooling (Fig. 4.7a and b, DSC insets). For BIC, BICL025, BICL05 and BIL samples (Fig. 4.8) the peaks related to this order-disorder transition could not be easily observed, although the temperature profile XRD confirms the phase change. The temperatures associated with this change are close to the temperatures at which the materials lose water, and the thermal effects in some cases overlap. However, for most samples except BIC and BIL, small endothermic changes in DSC could still be detected. A possible explanation for the thermal effects observed in the DSC curves is given by Yang et al. [97]. The transition of oxygen vacancies from ordered brownmillerite to disordered cubic structure in perovskite-type oxides is accompanied by a change in oxygen content in the material (oxygen sorption). While the process of the structural transition is endothermic, oxygen sorption is an exothermic process. Yin and Lin [95] have proposed that the sorption and desorption mechanism is based on the following reversible defect reaction:    86 ••• +→+ hOVO xOOg 221 )(2            Oxygen sorption: exothermic; weight gain         (4.1) ••• +→+ Og x O VOhO )(2212            Oxygen desorption: endothermic; weight loss    (4.2)  where ••OV is the positive oxygen vacancy, x OO  is the neutralized lattice oxygen and •h is the mobile electron hole. The thermal effects that are occurring during the change from brownmillerite to the cubic perovskite structure in our materials is therefore a combination of the thermal effects due to the two processes – structural transition and oxygen sorption (or desorption). In the case of BIO, the exothermic effect of the oxygen sorption is lower than the endothermic effect of the phase transition; hence the total heat effect is endothermic. In the case of BIC, BICL025, BICL05 and BIL, the heat spent during the phase transformation is similar to the heat released during oxygen sorption, and consequently the total heat effect is very small. In the corresponding TG scans, BIC, BICL025, BICL05 and BIL show a change of slope due to the oxygen sorption, starting from the order-disorder transition temperatures. The weight increase due to the oxygen sorption for BIC and BICL025 is shown in the TG insets in Fig. 4.8a and b. Based on the results obtained from the temperature- profile XRD in Section 4.4 and TGA and DSC results discussed above, we concluded that all the investigated materials undergo the order-disorder transition, accompanied with the increase in oxygen stoichiometry. The order-disorder transition temperatures were determined from the detected DSC peaks, or, in the case of BIC and BIL, estimated based on combined data from XRD, DSC and TGA. The temperatures are given in Table 4.1. With further heating, DSC curves for all materials show quite a large, but slow, exothermic effect over a wide temperature range, with a minimum at around 1300oC (1400oC for BICL025). This effect is probably related to the slow phase transformation back to the  87 brownmillerite structure which completes around 1300oC for BIO, BIC, BICL05 and BIL, and around 1400oC for BICL025, as confirmed also by XRD (Table 4.1). This change seems to be associated with the weight loss around 1000oC, which could be related to desorption of oxygen from the crystal structure. As reported in the literature [98, 99], some perovskite-type oxides can have a variable oxygen stoichiometry, depending on the oxygen partial pressure of the surrounding atmosphere, the temperature and the composition of the perovskite. If the conditions are changed, part of the oxygen is released from or incorporated into the crystal lattice, i.e., the oxygen stoichiometry changes and oxygen vacancies are formed or occupied, respectively. These processes are slow at temperatures up to 900oC, but more enhanced above this temperature. On further heating to about 1500oC, another structural change back to a cubic structure occurs. A sharp endothermic peak was observed in the DSC curve for BIC, due to the transformation to the cubic structure, as confirmed by XRD. A similar process happens with BIO, BICL05 and BIL, but with much smaller endothermic peaks. BIO, actually, showed two endothermic peaks around 1400oC. One is related to the transformation to the cubic phase, while the other one is most likely related to the melting of BIO, which occurs at 1475oC, as shown in the phase diagram in Appendix A. It is also noticed that BIO made by the solid- state reaction had these two peaks at somewhat higher temperatures, probably due to different rate of heat distribution in a coarser powder. BICL025 does not show the endothermic peak at around 1500oC, but it would likely occur at a higher temperature or with longer dwell at 1500oC. This was not confirmed, due to the limitation of the DSC instrument. XRD, however, confirms that the change to the cubic structure starts for this materials at 1500oC, as shown in Table 4.1. TG results in all cases show a small increase in weight with  88 this change into the cubic structure. The XRD, DSC and TG results for the investigated materials reveal multiple shifts between the brownmillerite and cubic structure. Based on the results it can be concluded that these changes are related to the reversible process of absorption and desorption of oxygen, which then cause changes in the structure – ordering it into a brownmillerite structure or forming a disordered cubic structure. 4.6 Stability in humid atmospheres and water It is reported in the literature that Ba2In2O5 with a brownmillerite structure absorbs water from the atmosphere and incorporates it in its crystal structure [72, 74]. Although incorporation of water into the crystal structure is one of the factors that improve the proton conductivity, it can cause instability of the material. The effect of humidity on the stability of the materials in this study was investigated at different temperatures and humidification levels. Table 4.2 summarizes the results of the study. The materials were stable in liquid water at 25oC, but decomposed at higher temperatures. Only BIC showed stability at 50oC. In most cases the materials were not stable if exposed to the flow of humidified air at RH=90% and temperatures between 25oC and 70oC. The materials mostly decomposed into BaCO3 and In(OH)3. Under a lower humidification of RH=50% only BIO decomposed, while the other compositions showed higher stability. Fig. 4.9 shows some examples of the X-ray diffraction patterns after sample exposure to humidity at different temperatures, demonstrating decomposition of the samples. An XRD pattern for as-prepared BIC material is also shown in Fig. 4.9a for comparison. X-ray diffraction patterns in Fig. 4.9b, c and d show decomposition of the materials into BaCO3 (thick arrows) and In(OH)3 (thin arrows) after exposure to the flow of humidified air. However, there were cases when Ba3In2(OH)12 was formed (Fig. 4.9e). In this case the samples were humidified in a covered glass beaker with evaporating  89 water on a hot plate set to 70oC, with no air flow. Consequently, the samples were exposed to a lower concentration of CO2, resulting in formation of Ba3In2(OH)12 instead of BaCO3 and In(OH)3. When the samples were exposed to a low humidification of 3 mol% steam in air flow at temperatures from 100oC to 500oC, X-ray diffraction of the samples did not show any change in the composition, although their color was lighter than the original samples. No difference in stability was noticed between samples prepared by the solid-state process and by the GNP process.                 90 Table 4.2 – Stability study of the investigated materials in liquid water and humidified air.  √-stable; X-not stable. Note: Percents are given in mole %. RH- relative humidity.        Sample Humidification (24h) 25oC 50oC 70oC 100oC-500oC BIO Immersed in liquid water √ X X - -II- Humidifier RH=90% (28% steam) √ X X - -II- Humidifier RH=50% (15% steam) √ X X - -II- Bubbler (3% steam) √ √ √ √ BIC Immersed in liquid water √ √ X - -II- Humidifier RH=90% (28% steam) X X X - -II- Humidifier RH=50% (15% steam) √ √ √ - -II- Bubbler (3% steam) √ √ √ √ BICL025 Immersed in liquid water √ X X - -II- Humidifier RH=90% (28% steam) X X X - -II- Humidifier RH=50% (15% steam) √ √ √ - -II- Bubbler (3% steam) √ √ √ √ BICL05 Immersed in liquid water √ X X - -II- Humidifier RH=90% (28% steam) X X X - -II- Humidifier RH=50% (15% steam) √ √ √ - -II- Bubbler (3% steam) √ √ √ √ BIL Immersed in liquid water √ X X - -II- Humidifier RH=90% (28% steam) X X X - -II- Humidifier RH=50% (15% steam) √ √ √ - -II- Bubbler (3% steam) √ √ √ √  91                     Fig. 4.9 – Examples of X-ray diffraction patterns of samples decomposed after 24 h treatment in moist air: (a) as-prepared BIC sample for comparison; (b) decomposed BIO sample at 90% RH and 50oC; (c) decomposed BIC sample at 90% RH and 25oC; (d) decomposed BICL025 sample at 90% RH and 50oC; (e) BIC sample decomposed to Ba3In2(OH)12 in water vapour in a covered beaker. Thick arrows BaCO3, thin arrows In(OH)3.    BIC-GNP as prepared    BIO-GNP in RH=90% at T=50oC  BIC-GNP in RH=90% at T=25oC  BICL025-GNP in RH=90% at T=50oC BIC-SS in water vapour-covered 20 30 60 70 80 40 50 2 Theta (a) (b) (c) (d) (e)  92 4.7 Summary Five different compositions of undoped and Ce- and La-doped Ba2In2O5 were successfully synthesized by the solid-state process and the glycine-nitrate process. The orthorhombic brownmillerite structure was achieved for all compositions in the lower temperature range of 1100 to 1400oC, while the cubic perovskite structure was achieved for all materials at 1500oC, except for BIL, which achieved this structure already at 1300oC. The mean particle size diameter of the as-prepared powders was 1.5 µm for the powders produced by the solid-state reaction and 150 nm for the powders produced by the glycine-nitrate process. In the case of powders produced by the glycine-nitrate process, the grain size, determined by the powder X-ray diffraction, increased from the initial ~10 nm to ~70 nm with the sintering temperature of 1500oC. In the case of the powders prepared by the solid- state process produced powders had an initial grain size of ~40 nm, increased to ~80 nm with heating to 1500oC. Temperature profile X-ray diffraction revealed that all powders undergo one or more phase transformations with heating in the range of 400oC to 1500oC, which are mainly related to order-disorder transformation between the ordered brownmillerite structure and a disordered cubic structure. However, the results in our study revealed multiple transitions that have not been reported in the literature. In the DSC tests unique peaks associated with the order-disorder transitions at lower temperatures were present, but hard to distinguish, except for BIO. The transition at higher temperatures was accompanied by a change in enthalpy slope and additional endothermic peaks. The transition process between the two structures may be slow, accompanied with a slow absorption or desorption of oxygen, when forming the cubic structure or the brownmillerite structure, respectively. The oxygen  93 absorption and desorption effects were also observed in the TGA curves, showing a small weight gain during change to a cubic structure and significant weight loss during change to the brownmillerite structure. TGA results also showed that at lower temperatures two processes take place with heating: release of H2O and CO2 from the crystal structure. Comparison of the weight loss due to the water release between different compositions showed that BIO had the smallest weight loss among all the samples. In addition, BIO sample prepared by the glycine-nitrate method showed a larger weight loss due to water release, compared to the sample made by the solid-state method, suggesting a higher content of water in the grain boundary region than in the bulk material. This difference was not noticed for other compositions, suggesting that water accumulates more in the bulk of these materials. Stability studies for all five compositions demonstrated that these compositions are not stable in highly humid atmospheres with RH=90% up to 70oC over 24 h, while higher stability was observed for RH=50%. The materials are stable in liquid water at 25oC, while decomposing at higher temperatures. All materials showed stability for a composition of 3% mol steam in air and a flow up to 500oC for at least 24 hours.          94 5. Electrical Conductivity of the Materials 5.1 Synopsis The main goal for the conductivity tests was to compare the electrical conductivities of the five different compositions investigated in this work (listed in Table 3.1) and potentially determine a candidate material to act as a proton conductive electrolyte for an intermediate temperature fuel cell. Therefore, of particular interest was the investigation of the electrical conductivity in hydrogen-containing atmospheres. It was anticipated that conductivity results in hydrogen-containing atmospheres would provide an insight into the proton conduction capability of the materials. Total electrical conductivity of the materials was measured using ac impedance spectroscopy, as described in Chapter 3, Section 3.5. Comparison of the electrical conductivities for all compositions (except for BICL05, which could not be successfully prepared) was done in both air and 50% hydrogen balanced by nitrogen in the temperature range from 100oC to 500oC. The effect of the microstructure was investigated by comparing the conductivity of the samples made by the solid-state process (powder particle size ~1.5 µm, final grain size of the sintered pellets ~60 nm) and by the GNP process (powder particle size ~165 nm, final grain size of the sintered pellets ~40 nm). Ba2In2O5 (BIO) was used as a baseline for comparison, because a number of its characteristics are well studied and reported in the literature. However, its conductivity in hydrogen-containing atmospheres has not been reported yet. Based on the results, the candidate material with the highest conductivity and stability in hydrogen-containing atmospheres was to be selected and further investigated in more detail.  95 5.2 Ac impedance measurements for conductivity determination As mentioned earlier, the total electrical conductivities of the tested samples were determined by ac impedance spectroscopy. The principles of ac impedance spectroscopy are given in Appendix B.1. Obtained ac impedance results were analyzed based on the “Bauerle model” for the polycrystalline ceramic materials. Based on this model, ac impedance scan of a polycrystalline material appears in a series of three semicircles related to the impedance contribution of the bulk material, grain boundary and the charge transfer between electrode and electrolyte, as shown in Fig. 5.1. The semicircles could be sometimes overlapped or out of frequency range, resulting in only one or two semicircles, or partial semicircles. The equivalent electrical circuit that is often used to represent this model is also shown in the figure. More detail of the “Bauerle model” and interpretation of the results for polycrystalline materials is given in Appendix B.1.              96                        Figure 5.1 – (Top) Schematic of a polycrystalline material with grain bulk, grain boundary and an interface with the electrode; (Middle) Equivalent circuit diagram for a polycrystalline material: Rb, Rgb, Re represent the bulk, grain boundary and electrode resistances, respectivelly; Cb, Cgb, Cdl represent the bulk, grain boundary and electrode double layer capacitances, respectivelly; (Bottom) Complex plane plot showing the ideal ac impedance scan response for a polycrystalline ceramic. Detailed explanation is given in Appendix B.1. Rb Rgb Re Cb Cgb Ce Z' (Ω ) or (Ω cm) Z' ' ( ΩΩ ΩΩ ) o r ( ΩΩ ΩΩ c m ) Rb Rb+Rgb Rb+Rgb+Re Frequency, f (Hz) Bulk contribution Grain boundary contribution Electrode- electrolyte transfer contribution Bulk Grain boundary Electrode- electrolyte interface Polycrystalline material  97 5.2.1 Analysis of electrode contribution Before conductivity measurements by ac impedance spectroscopy, it was important to determine which part of the ac impedance scans corresponds to the total material resistance and which part is the contribution from the sample/electrode interface charge transfer impedance. Knowing that, the total contribution from the material itself can be found and its total electrical conductivity determined, based on the above mentioned “Bauerle” model. In order to distinguish the contribution of the sample/electrode interface impedance from the desired material impedance, samples with different electrode materials (Au, and Pt prepared in two different ways, as explained in Chapter 3, Section 3.5) were tested by ac impedance spectroscopy and their scans compared. Fig. 5.2 shows an example of the ac impedance spectra measured in air at 300oC for BIO and BIC samples made by the GNP process and comparison between the scans obtained when two different electrodes were used. A difference in the low frequency semicircles is evident for both cases, showing much higher impedance for the Au electrode, confirming that this semicircle is due to the electrode contribution. The high frequency semicircles are similar (one semicircle in the case of BIO and two in the case of BIC) showing that they represent the material contribution. Arrows show the locations of where the electrode semicircle starts. The same method was done for the other compositions, prepared both by the solid-state and GNP method, for each temperature of interest. BIL and BICL025 samples had similar scans to the BIC sample in air.     98           Figure 5.2 – Ac impedance scans measured in air showing the difference between the sample/electrode interface contribution when two different electrodes are applied (Au and Pt); (a) BIO sample made by the GNP process; (b) BIC sample made by the GNP process; Frequencies are given for the points where the material contribution semicircle transitions to the electrode semicircle.  An example of the ac impedance scan comparison for the samples with different electrodes in hydrogen-containing atmosphere is given for the BIO-GNP and BIC-GNP samples in Fig. 5.3. Fig. 5.3a shows that, for the BIO sample, the low frequency semicircles are different for the two different electrodes. This result confirms that this semicircle is associated with the electrode contribution. However, in Fig. 5.3b for the BIC sample, it can be seen that both the low and high frequency semicircle are comparable, suggesting that both are related to the material properties. In this case, the impedance semicircle associated with the electrode does not show on the impedance scan. Similar tests were performed for all the other compositions prepared by both the GNP and solid-state method and electrode -5000 -4000 -3000 -2000 -1000 0 0 1000 2000 3000 4000 5000 Z' (kΩ cm) Z' ' (k ΩΩ ΩΩ c m ) Au electrode Pt electrode 300oC -100 -50 0 0 50 100 24Hz 353HzBIC in air -8000 -6000 -4000 -2000 0 0 2000 4000 6000 8000 Z' (kΩcm) Z' ' (k ΩΩ ΩΩ c m ) Au electrode Pt electrode 300oC -500 0 0 500 149Hz 1kHz BIO in air 300oC (a) (b) Rb+Rgb Rb+Rgb Rb+Rgb -250 – 2h 250  99 contributions were determined at each temperature. The BIL and BICL025 samples showed similar behaviour to the BIC sample.          Figure 5.3 – Ac impedance scans measured in 50% H2/50% N2 for samples with two different electrodes (a) BIO-GNP sample showing difference in the low frequency semicircle due to the electrode contribution; (b) BIC sample showing no difference in the low frequency semicircle. Frequencies are given for the points where high frequency semicircle crosses the axis.  5.2.2 Ac impedance results and conductivity determination Following the preliminary study discussed in Section 5.2.1, actual testing of the samples by ac impedance was performed to determine their total conductivity. Four different compositions consisting of BIO, BIC, BICL025, and BIL, prepared by the solid-state and the GNP method were tested in air and hydrogen-containing atmospheres at temperatures between 100oC and 500oC. Total electrical resistances, R, of the materials were determined from the ac impedance scans, from the intersection with the real axis of the semicircle -8000 -6000 -4000 -2000 0 0 2000 4000 6000 8000 Z' (kΩ cm) Z' ' (k ΩΩ ΩΩ c m ) Pt not treated Pt treated 300oCBIC in H2 1000Hz -0.50 0.00 46 47 48 49 Z' (Ωcm) Z' ' ( ΩΩ ΩΩ c m ) Pt not treated Pt treated 300oC in H2 51kHz 531Hz BIO in H (a) (b)  100 associated with the material. Conductivities of the samples were calculated from the total resistances, R, of the samples, using the following equation: AR l ⋅ == ρ σ 1  (5.1) where σ is the conductivity of the material in S/cm, ρ is the electrical resistivity in Ωcm, R is the resistance of the material in Ω, l is the thickness of the sample in cm, and A is the area of the applied electrode in cm2. In our case, the geometrical factor l/A in the equation was already taken into account in the ac impedance scans, where both the real and imaginary axes are given in units of Ωcm.  Therefore, for calculating the total conductivities, a simple expression σ=1/ρ was used, where ρ is the total material resistivity given in Ωcm. The conductivities and their dependence on the temperature are plotted in the Arrhenius form, as described in Section 2.3.1. Fig. 5.4 shows examples of the ac impedance scans obtained for different samples under different environments at 300oC. Arrows on the figures show the points taken as the total material resistance which are used for the conductivity determination. In this stage of the project, ac impedance scans were used only to define and compare the total conductivities of the tested materials. A more detailed analysis of the ac impedance data for selected materials of interest is discussed in Chapter 6.        101                Figure 5.4 – Examples of the ac impedance scans for BIO, BIC, BICL025 and BIL under different environments at 300oC. Arrows point to the values on the real axis taken as total resistance of the material (Rt). -1000 -500 0 0 500 1000 Z' (kΩcm) -100 -50 0 0 50 100 Z' (kΩcm) Z' ' (k ΩΩ ΩΩ c m ) -10 -5 0 0 5 10 Z' (kΩcm) -0.30 -0.20 -0.10 0.00 45 46 47 48 49 50 Z' (Ωcm) Z' ' ( ΩΩ ΩΩ c m ) BIO-GNP BIC-GNP BICL025-GNP BIL-GNP (a) In 50% H2/50% N2 at 300oC Rt Rt Rt Rt -8000 -6000 -4000 -2000 0 0 2000 4000 6000 8000 Z' (kΩcm) -6000 -4000 -2000 0 0 2000 4000 6000 Z' (kΩcm) -60 -40 -20 0 0 20 40 60 Z' (kΩcm)Z''  (k ΩΩ ΩΩ c m ) -3000 -2000 -1000 0 0 1000 2000 3000 Z' (kΩcm) (b) In air at 300oC  BIO-SS  BIC-SS  BICL025-SS  BIL-SS  Rt  Rt  Rt  Rt  -25  - -250  - -3  -  102 5.3 Electrical conductivity results 5.3.1 Comparison to the literature Fig. 5.5 shows Arrhenius plots of measured conductivity in air for a Ba2In2O5 sample in this work made by the solid-state process, compared to some literature data. Comparable conductivities were obtained. The somewhat lower conductivities in this work are most likely due to the different conditions applied. Zhang et al. [33] measured conductivity in H2O saturated air (at 25oC), while our measurements were performed in dried atmospheric air (moisture content 15 ppm). As reported by Zhang et al., moisture in the air provides protonic conductivity in Ba2In2O5, especially at temperatures below 500oC, resulting in higher total conductivity.            Figure 5.5 – Arrhenius plot of Ba2In2O5 conductivity in air: ○ in this work; ■ in Niwa et al. [44]; ∆ in Zhang et al. [33] (measured in wet air). -9 -8 -7 -6 -5 -4 -3 -2 -1 0.80 1.20 1.60 2.00 2.40 2.80 1000/T(K-1) lo g σσ σσ  (S /c m )         900  700       500   400      300            200                     100oC  103 5.3.2 Conductivity of the solid-state samples in air and a hydrogen-containing atmosphere Fig. 5.6 compares conductivities in air (a) and 50% H2/50%N2 (b) for samples made by the solid-state process. Solid symbols in both graphs represent as-prepared samples which were tested. These samples contain absorbed water in the structure, as shown from the TGA results. Hollow symbols show the conductivity results for the “dry” samples, which were heated to 380oC for 3 hours to release the water before the conductivity measurements. Measurements in air. When measured in air, as shown in Fig. 5.6a, as-prepared BIO sample shows a sudden drop in conductivity between 300oC and 400oC. This is due to the release of water in this temperature range, as reported in our TGA results. As discussed earlier, water present in the structure provides protonic conduction and contributes to the total electrical conduction. Once the water leaves the structure, conductivity is mainly due to other charge carriers (ions and holes). When the “dry” BIO sample was tested, the total conductivity below 300oC was lower than for the as-prepared sample, and was linearly increasing with the temperature. Higher conductivity below 300oC for the as-prepared sample reveals that the water present in the sample provides a significant protonic contribution to the total conductivity. On the other hand, as shown for the case of BIC in Fig. 5.6a, as-prepared BIC, BICL025, and BIL samples did not show a significant change in conductivity in air due to the release of water from the structure. They also did not show a significant difference between the as-prepared and the “dry” samples. The contribution to the total conductivity provided by the absorbed water is much smaller for these compositions than for the BIO sample. This is opposite to what was expected, since the TGA results shown in Chapter 4, Section 4.5 indicated that BIC, BICL025, and BIL have a higher content of water present in  104 the structure, compared to BIO. However, the TGA suggested also that, for these materials, water accumulates more in the bulk, while for BIO its accumulation is higher in the grain boundary region. Hence, the stronger effect of water on the conductivity of BIO suggests that the grain boundary plays a more important role in proton conductivity than the bulk of the materials with respect to water. When all “dry” samples were compared in air, the conductivity of the BIC sample was the highest, while BICL025 and BIL samples have the lowest conductivities. For these three compositions there is a decrease in the conductivity around 400oC. Since the effect of water is eliminated by heating the materials before testing, this change in conductivity can be related only to the order-disorder transition, confirmed earlier by the temperature-profile XRD and the DSC tests, discussed in Chapter 4, Sections 4.4 and 4.5. This structure transition and loss of oxygen vacancies due to the associated oxygen sorption is most likely causing the loss of conductivity. It is interesting to note that BIO and BIC show similar activation energy for ionic conduction of about 0.64 eV and 0.61 eV, respectively, while BICL025 and BIL have a lower activation energy of about 0.18 eV, suggesting differences in conduction mechanisms for these two groups of materials. Measurements in 50% H2/50% N2. Fig. 5.6b shows the total electrical conductivities measured in the hydrogen-containing atmosphere for the solid-state samples. Again, for the as-prepared BIO sample, an effect of water was noticed, in this case as a sharp increase in conductivity. This is probably due to water release from the structure and direct incorporation of hydrogen in the crystal structure, as suggested in Chapter 2, Section 2.3.3. Decomposition or reduction of the material were excluded as possible causes for increased conductivity, as no changes in composition or structure were observed by XRD after testing in hydrogen.  105 When conductivity was measured for the “dry” BIO sample, a linear increase in conductivity was recorded, with two orders of magnitude higher conductivities at temperatures below 300oC, compared to the as-prepared sample. This suggests a high contribution of protonic conductivity in BIO in the presence of hydrogen. Similar, but a much less enhanced effect of water was found for as-prepared BIC, BICL025, and BIL samples. When all “dry” samples were tested in hydrogen, their conductivities were higher than conductivities in air for all compositions, except BIC whose conductivity was lower. All composition showed a similar energy of activation of 0.45 eV, revealing the same mechanism of conduction in hydrogen. “Dry” BIC, BICL025 and BIL showed a small change in slope at 400oC with decrease in the conductivity, which is, again, due to the transformation from the brownmillerite to the cubic structure. Also, the BIC, BICL025, and BIL samples retained this cubic structure after the testing in hydrogen, as confirmed by X-ray diffraction.        Figure 5.6 – Electrical conductivity in (a) air and (b) 50% H2/ 50% N2 for samples prepared by the solid-state method: As-prepared samples: ● BIO, ♦ BIC, ▲BIL; “Dry” samples: ○ BIO, ◊ BIC, □ BICL025, ∆ BIL. Average measurement error from the three repeated measurements for each case was 5% (error bars shown for the case of BIC in (b)). -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1.20 1.60 2.00 2.40 2.80 1000/T(K-1) lo g σσ σσ  (S /c m )         500     400        300             200                       100oC b -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1.2 1.6 2 2.4 2.8 1000/T(K-1) lo g σσ σσ  (S /c m )         500    400        300             200                      100oC a  106 5.3.3 Conductivity of the GNP samples in air and a hydrogen-containing atmosphere Fig. 5.7 shows a comparison between conductivity in air (a) and 50% H2/50%N2 (b) for samples made by the GNP process. The effect of water on the conductivity of the as-prepared samples in both atmospheres was similar to the solid-state samples, but much more enhanced for the BIO-GNP sample, especially in hydrogen. When “dry” GNP samples were measured in air, the measured conductivities were very similar to the conductivities for the solid-state samples. The change in conductivity in air due to the order-disorder transition is again present for BIC, BICL025, and BIL. In hydrogen-containing atmospheres, the improvement in conductivity compared to the tests in air is much more enhanced for the GNP samples. In this atmosphere, changes in conductivity due to the order-disorder transition for the BIC, BICL025, and BIL samples were noticed around 400oC, as for the solid-state samples. Surprisingly high conductivities between 0.02 S/cm and 0.7 S/cm were obtained for the Ba2In2O5 sample in the temperature range from 300oC to 500oC.           107          Figure 5.7 – Electrical conductivity in (a) air and (b) H2/N2 for GNP samples: As-prepared samples: ● BIO, ♦ BIC, ▲BIL; “Dry” samples: ○ BIO, ◊ BIC, □ BICL025, ∆ BIL. Average measurement error from the three repeated measurements for each case was 6% (error bars shown for the case of BIC in (b)). Lines on the graphs are only shown as a guide for the eyes. 5.3.4 Effect of microstructure In order to investigate the effect of the grain size/grain boundary on the electrical properties of the materials, the conductivities of the GNP samples were compared to the solid-state samples in air (Fig. 5.6a and 5.7a) and in hydrogen-containing atmosphere (Fig. 5.6b and 5.7b). The GNP samples, with their smaller grain size and, hence, larger grain boundary area are expected to show a higher effect of the grain boundary. When measured in air, the conductivity results were very similar for the GNP and the solid-state samples. This suggests that grain boundary does not play a significant role in conductivity of oxygen ions and holes, at least not in this temperature range. It is possible that stronger effects would be noticed at higher temperatures (> 500oC), when the mobility of oxygen ions is higher. On the other hand, it is apparent that the sample preparation method and associated microstructure -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1.20 1.60 2.00 2.40 2.80 1000/T(K-1) lo g σσ σσ  (S /c m )         500     400        300             200                        100oC b -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1.20 1.60 2.00 2.40 2.80 1000/T(K-1) lo g σσ σσ  (S /c m )          500    400        300             200                        100oC a  108 does have an effect on the conductivity in hydrogen, showing conductivities an order of magnitude higher for the GNP samples. The effect is especially enhanced for the BIO sample. The effect of water and its contribution to the total conductivity is also more enhanced for the BIO sample prepared by the GNP method. In the literature it is anticipated that this conductivity in hydrogen and humidity is mainly due to the proton conductivity, as discussed in Chapter 2. Therefore, our results suggest that the grain boundary plays an important role in the conduction mechanism of protons, possibly by having a higher concentration of oxygen vacancies in this region. Easily accessible oxygen vacancies enable oxygen ion incorporation in air, and proton incorporation in water or hydrogen, as represented by equations 2.39 through 2.43, in Chapter 2. In addition, a higher concentration of oxygen vacancies improves charge transport along the grain boundaries. The process is enhanced in hydrogen atmospheres, as the mobility of protons is superior at the low temperatures, compared to oxygen ion mobility. This effect has not been studied in further detail in this work. From the conductivity measurements in hydrogen-containing atmospheres for all prepared samples, undoped Ba2In2O5 (BIO) prepared by the glycine-nitrate method showed the highest conductivity with sufficient stability. Therefore, this material was selected for further, more detailed studies of its proton conductivity and the effects of different parameters on its electrical behavior, discussed in the following chapters. 5.4 Summary Total electrical conductivities of the BIO, BIC, BICL025 and BIL samples, prepared by both the glycine-nitrate process and solid-state reaction, were determined by ac impedance spectroscopy in air and hydrogen containing atmospheres, and in the temperature range  109 between 100oC and 500oC. The total conductivities of the tested materials were compared in order to determine the candidate with the highest conductivity in air and especially in hydrogen-containing atmospheres. The effect of dopant, microstructure, and atmosphere on the conductivity of these materials was investigated. All compositions showed higher conductivity in a hydrogen atmosphere than in oxygen. Compared to Ce- and La-doped Ba2In2O5, undoped Ba2In2O5 showed several orders of magnitude higher conductivity in hydrogen containing atmosphere at temperatures between 300oC and 500oC, especially the samples produced by the glycine-nitrate process (conductivity between 0.02 S/cm and 0.7 S/cm). Comparison between the “as-prepared” (that contained water) and “dry” samples showed that water incorporated in the materials structure had an effect on conductivity results, especially for the BIO samples made by the glycine-nitrate method. The results obtained in air revealed that the water in the structure provides additional conductivity that contributes to the total conductivity, while the total conductivity was lower when the water was not present. When measured in hydrogen, the presence of water lowered the total conductivity, while its release caused the conductivity to increase, possibly by enabling direct incorporation of hydrogen. This effect was particularly enhanced for the BIO sample made by the glycine-nitrate method. Although the effect of the humidification on conductivity of Ba2In2O5 and related materials was investigated in the literature [33, 40], a study of the effect of the internal water bound in the structure on the conductivity, as presented in this work, was not reported before. The results of the conductivity study showed that microstructure has an effect on conductivity, especially in hydrogen. Samples made by the glycine-nitrate process, with a  110 resulting smaller grain size of about 40 nm, showed higher conductivity than samples produced by the solid-state reaction with a grain size of about 60 nm in both atmospheres. The difference is not significant in air, which suggests that grain boundary does not play a significant role in the conductivity of oxygen ions and holes in this temperature range. However, when measured in hydrogen, the conductivities were an order of magnitude higher for the GNP sample, suggesting an important role of the grain boundary in the conduction mechanism of protons. Based on the results discussed in this chapter, undoped Ba2In2O5 produced by the glycine-nitrate process is the most promising material from the investigated group of materials for further proton conductivity investigation. Although conductivity results in hydrogen-containing atmospheres give us an indication of proton conductivity, it is crucial to determine the contribution of proton conduction to the total conductivity measured. This aspect is discussed in the following chapter.            111 6. Electrochemical Impedance Spectroscopy of Ba2In2O5 – Effect of Porosity, Grain Size, Dopant, Atmosphere and Temperature  6.1 Synopsis  Through the conductivity study discussed in Chapter 5, undoped Ba2In2O5 material was selected as the most promising material for an intermediate-temperature proton conductor. Although this material has been well investigated, it has never been studied in hydrogen- containing atmospheres before. Also, the high conductivities measured in our work in these atmospheres have never been reported before. Thus, it is important to investigate the electrical properties of this material in more detail. Electrical properties of Ba2In2O5 and related materials are shown to depend on the type and level of dopant, on preparation procedure and resulting microstructure, and on the atmosphere and sample porosity [44, 74, 100]. The presence of water in the structure and its loss in the temperature range between 200oC and 300oC, as discussed earlier in Chapter 3, Section 3.5, is another factor that affects their electrical properties. An understanding of the electrochemical processes in these materials and the effects of different parameters on their electrical conductivity could help improve the characteristics and applicability of Ba2In2O5 and its derivatives in electrochemical devices. Using the ac impedance technique and theory, a much better understanding of the electrochemical processes and the behavior of Ba2In2O5 and related materials can be obtained. In this stage of the project electrochemical impedance spectroscopy was used to investigate the electrical properties of Ba2In2O5 and the effect of porosity, dopant (Ce), atmosphere, temperature, and grain size on these properties. A detailed study of this kind for Ba2In2O5 has not been reported before. The effect of dopant was investigated by comparing  112 the results of undoped Ba2In2O5 samples to Ce-doped Ba2In2O5 samples. Effect of grain size for both BIO and BIC samples was studied by comparing the samples made by the glycine- nitrate method, which resulted in smaller grain sizes, and the samples made by the solid-state process, which resulted in larger grain sizes. Ac impedance scans were fitted to an equivalent circuit and R and C (or λ) parameters determined. Porosity effects were determined by testing the samples made from the BIO-GNP and BIC-GNP material, over a range of porosity. All samples were tested in two different atmospheres, air and 50% H2/50% N2, and at different temperatures in the range from 100oC to 500oC. 6.2 Samples for the study Undoped Ba2In2O5 (BIO) and Ce-doped Ba2In2O5 (BIC), made by both the solid-state reaction and GNP method, were tested and compared in this study. Characteristics of the used sintered samples are given in Chapter 3, Table 3.2. Fully dense sintered pellet samples could not be prepared due to the limitation of the sintering temperature (melting temperature for Ba2In2O5 is 1475oC, as shown in Appendix A, Fig. A3). The lowest total porosity obtained without the use of the pore-former was between 16 and 20% (with ~4% open porosity). All samples for the ac impedance spectra were produced with a total porosity of about 20% in order to minimize the effect of the porosity factor on the conductivity results. They all had brownmillerite structure as the final phase. Sintered pellet samples produced from the GNP powders had a final grain size of about 40 nm, while the pellet samples made from the solid-state powders had a grain size around 60 nm, as confirmed by X-ray diffraction and the Scherrer formula. Fig. 6.1 shows high resolution SEM pictures of the polished and etched (using 0.3% HNO3 for 1 min) cross-sections of a BIO sample made by the GNP process (BIO-GNP) and a BIO sample made by the solid-state process (BIO-SS). It  113 was challenging to achieve a good contrast between the grains and grain boundaries and a focused picture due to the limitations of the instrument and small grain sizes of the samples. However, it can be seen in the figure that the BIO-SS sample had larger grains compared to the BIO-GNP sample. The grain sizes are in the range determined earlier by the Scherrer formula. Additional BIO–GNP and BIC-GNP samples with porosities of 30%, 40% and 50% were prepared for the study to determine the effect of the porosity on the conductivity. They all had similar grain sizes.             Figure 6.1 – SEM pictures of the polished and etched cross-sections of the sintered samples obtained by two different methods (GNP pressed and sintered to 1350oC for 6 hours, and the solid-state method pressed and sintered to 1400oC for 6 hours): low (a) and high (b) magnification SEM picture of a GNP sample. The high resolution picture reveals grain sizes lower than ~ 50 nm; low (c) and high (d) magnification SEM picture of a SS sample. The high resolution picture for the SS sample reveals grain sizes between 60 and 80 nm. (a) (b) (c) (d)  114 6.3 Effect of porosity on conductivity BIO-GNP and BIC-GNP samples were used for this study, in both air and 50% H2/50% N2. The samples were dried at 380oC for 3 h before measurement to release the water. Investigation of different porosities between 20% and 50% showed that the conductivities of the samples decreased with an increase in porosity. Fig. 6.2 shows an example of the change in the ac impedance scans with the porosity, measured in air at 400oC for the BIO and BIC samples. BIC sample clearly shows the change in the second semicircle with the change of porosity, while the bulk semicircle is unaffected. This is expected, as the bulk impedance does not depend on the geometry and microstructure of the sample. Second semicircle is, therefore, associated not only to the grain boundary resistance, but also to the microstructure of the sample. For the BIO sample, bulk and grain boundary semicircles could not be resolved, and the effect of porosity is shown as an increase of the shown semicircle. Fig. 6.3 shows the change of the conductivity with the porosity for the BIO and BIC samples measured in air and in hydrogen-containing atmospheres.         Figure 6.2 – Change in ac impedance scans with porosity for BIO-GNP (a) and BIC-GNP (b) samples measured in air at 400oC (samples were dried before measurement to release water). -200 -150 -100 -50 0 0 50 100 150 200 Z' (kΩ cm) Z"  (k ΩΩ ΩΩ c m ) 20% 30% 40% 50% (a) BIO-GNP -100 -80 -60 -40 -20 0 0 20 40 60 80 100 Z' (kΩcm) Z' ' (k ΩΩ ΩΩ c m ) 20% 30% 40% 50% (b) BIC-GNP  115                     Figure 6.3 – Change in the total electrical conductivity due to different porosities of the BIO- GNP (a) and BIC-GNP (b) samples in air and BIO-GNP (c) and BIC-GNP (d) samples in 50% H2/50% N2. (b) BIC-GNP in air -9 -8 -7 -6 -5 -4 -3 -2 -1 0 10 20 30 40 50 60 Porosity, % Lo g σσ σσ , S/ c m 2 100oC 500oC 200oC 300oC 400oC -9 -8 -7 -6 -5 -4 -3 -2 -1 0 10 20 30 40 50 60 Porosity, % Lo g σσ σσ , S/ c m 2 100oC 500oC (a) BIO-GNP in air 200oC 400oC 300oC (d) BIC-GNP in H2/N2 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 10 20 30 40 50 60 Porosity, % Lo g σσ σσ , S/ c m 2 100oC 200oC 300oC 500oC 400oC (c) BIO-GNP in H2/N2 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 10 20 30 40 50 60 Porosity, % Lo g σσ σσ , S/ c m 2 100oC 200oC 300oC 400oC 500oC  116 6.4 Ac impedance of BIO-GNP and BIO-SS in air 6.4.1 Effect of grain size Fig. 6.4 shows ac impedance scans for the BIO samples measured in air in the temperature range from 100oC to 500oC.  Results for the samples made by both the glycine- nitrate process (GNP - representing samples with a ~ 40 nm grain size) and by the solid-state reaction (SS - representing samples with a ~ 60 nm grain size) are shown to determine the effect of grain size. One semicircle appears in the high frequency range for each impedance scan, representing the contribution from the material impedance, while the low frequency tail is part of the electrode contribution, as determined earlier. Bulk and grain boundary semicircles cannot be resolved, as they most likely have similar time constants. When comparing samples with the different grain sizes, it is apparent that the samples with larger grains, and hence smaller grain boundary area (SS samples) show higher impedance. Since bulk impedance is an intrinsic property of the material and does not depend on the microstructure, this higher impedance for the SS samples is due to the grain boundary impedance. Hence, it can be concluded that the grain boundary plays a role in the conductivity of BIO in air (mainly oxygen ion and hole conduction [40, 73]), showing higher impedance when the grain boundary area is smaller. However, the role is not as enhanced as in the test performed in hydrogen, as discussed in Chapter 5, Section 5.3.4. The difference in impedance is less apparent at higher temperatures, most likely due to the higher mobility of the oxygen ion and improved conduction both in the in the grain boundaries and in the bulk of the material.    117    Error!                    -200 -150 -100 -50 0 0 50 100 150 200 Z (kΩcm) 58k 169 52k 198 500 oC 500oC ′ -1000 -500 0 0 500 1000 Z (kΩcm) 8k 34 6k 85 400oC 400oC ′ -3000 -2000 -1000 0 0 1000 2000 3000 Z' (kΩ cm) Z' '(k ΩΩ ΩΩ cm ) 24 10k 4k 169 300oC -10000 -5000 0 0 5000 10000 Z (kΩcm) 200oC 4.5k 84 1k 53 200oC ′ -150000 -100000 -50000 0 0 50000 100000 150000 Z (kΩ cm) GNP SS 100 0.5 160 7.5 100oC BIO-in air 100oC ′ Rb+Rgb Rb+Rgb Figure 6.4 – Ac impedance scans for the BIO-GNP sample (grain size ~40 nm) and BIO-SS (grain size ~60 nm) measured in air. Characteristic frequencies are given in Hz. Arrows show the locations where the total material resistances (Rb+Rgb) were taken. Lines on the 300oC graph show the equivalent circuit fit to the experimental data.  118 6.4.2 Effect of water loss During the measurement in air, conductivity of Ba2In2O5 (BIO) starts decreasing as a function of time for both GNP and SS samples at 300oC and eventually stabilizes after about 3 hours. This effect is shown in Fig. 6.5 for the SS sample. This behavior is due to the loss of water from the structure around 300oC that is discussed in Chapter 4. As discussed there, water provides protonic conductivity, significantly contributing to the total conductivity measured in air. As the water leaves the structure, the conductivity decreases until it reaches a stable value, when most of the water has left the structure.        Figure 6.5 – Change in the ac impedance spectra for BIO-SS samples at 300oC over time, due to loss of water. Scans stabilize after about 3 hours.  6.4.3 Equivalent circuit fit  The impedance scans for BIO-GNP and BIO-SS in air can be modelled using an equivalent circuit with a series of RC elements, which represent bulk, grain boundary and electrode impedance. In a more general case, the scans can be modelled using constant phase elements, Z(CPE)=λ/(jω)α [101], in combination with R elements, considering RC elements -1000 -500 0 0 500 1000 1500 2000 Z'(kΩ cm) Z' '(k ΩΩ ΩΩ cm ) 30min 1h 1.5h 2h 2.5h  3h  119 a special case when λ=C, and α=1, as explained in Appendix B.1. The equivalent circuit used for fitting the impedance plots is shown in Fig. 6.6. Determined values for R, λ (or C in the case of α=1) and α are shown in Table 6.1. for both SS and GNP BIO samples. Since the bulk and grain boundary semicircles could not be resolved, the total value of the material resistance is shown. The values for α are all equal 1, showing that material impedance can be represented as a simple RC element. While the total resistances are decreasing with the temperature, the capacitances are generally increasing.     Figure 6.6 – General equivalent circuit used for representing the contribution of bulk, grain boundary and electrode in the total ac impedance response. In a more specific case (when α=1) instead of a CPE, a capacitor C can be used.  Table 6.1 – Ba2In2O5 (BIO) sample measured in air – parameters determined by fitting the experimental ac scans to the equivalent circuit shown in Fig. 6.6.  Rb Rgb Re CPEb CPEgb CPEe T, oC Rb+gb, ohm·cm λ (or Cb+gb), F/cm α Rb+gb, ohm·cm λ (or Cb+gb), F/cm α 100 7.58E+07 2.10E-11 1.00 4.63E+07 2.15E-11 1.00 200 4.80E+06 3.32E-11 1.00 1.30E+06 2.72E-11 1.00 300 1.50E+06 2.54E-11 1.00 3.64E+05 4.13E-11 1.00 400 7.65E+05 2.60E-11 1.00 5.46E+05 4.80E-11 1.00 500 1.02E+05 3.00E-11 1.00 8.40E+04 3.27E-11 1.00 SS (grain size≈ 60 nm) GNP (grain size≈ 40 nm)  120 6.5 Ac impedance of BIC-GNP and BIC-SS in air 6.5.1 Effect of dopant Fig. 6.7 shows the ac impedance scans as a function of temperature measured in air for the Ce-doped Ba2In2O5, denoted as BIC in this work. The effect of dopant on conductivity of the oxide materials is often related to the segregation of the dopant in the grain boundary region [102-104]. It can be also due to a different symmetry, lattice distances, charge accumulation, concentration of oxygen vacancies, etc. From the ac scans it is apparent that the total resistances of BIC in air are lower than the resistances measured for BIO samples. Bulk and grain boundary semicircles could not be resolved at 100oC, but are clearly shown at 200oC and 300oC, especially for the GNP samples. The low frequency tail is related to the electrode contribution as determined earlier. The bulk impedance at each temperature is the same for the SS and the GNP samples, what was expected, as this impedance does not depend on the microstructure. On the other hand, grain boundary impedance is greater for the GNP samples, compared to the SS samples. This result reveals that BIC shows a significant grain boundary resistance to the charge carriers in air, and the larger grain boundary area (for the GNP samples) causes higher grain boundary impedance. This is opposite to what was noticed for the undoped Ba2In2O5 (BIO) and hence shows that presence of Ce affects the conductivity of grain boundaries in Ba2In2O5, most likely by segregating in this area and lowering the concentration of oxygen vacancies (based on the reaction 2.29 occurring to the left, where M=In3+, and B=Ce4+). This would have to be confirmed with a detailed crystal structure comparison to determine the oxygen vacancy, concentration and local effect of Ce as a dopant. However, the total conductivity is still improved in BIC compared to BIO.  121 The grain boundary impedance for the GNP samples is greater than the bulk impedance at 200oC, but decreases and approaches bulk impedance at 300oC. For the SS samples, the grain boundary impedance is evidently smaller than the bulk impedance, showing an improvement in conductivity for the samples with larger grains and a smaller grain boundary area.  122                  -10 -5 0 0 5 10 Z' (kΩ cm) 2.6k 265 500oC 201k 108k Rb+Rgb -30 -20 -10 0 0 10 20 30 Z' (kΩ cm) 1000k 7.5 1000k 266 127k 61k 400oC Rb+Rgb -60 -40 -20 0 0 20 40 60 Z' (kΩcm) Z' ' (k ΩΩ ΩΩ cm ) 106 4 470 656k 633k 32k 300oC Rb Rb+Rgb Rb+Rgb -2500 -2000 -1500 -1000 -500 0 0 500 1000 1500 2000 2500 Z (kΩcm) 0.4 24 18k 23 13k 200oC ′ Rb Rb+Rgb Rb+Rgb -40000 -30000 -20000 -10000 0 0 10000 20000 30000 40000 Z' (kΩ cm) GNP SS BIC-in air 149 10 0.1 0.1 100oC Rb+Rgb Rb+Rgb Figure 6.7 – Ac impedance scans for the BIC-GNP samples (grain size ~ 40 nm) and BIC-SS samples (grain size ~60 nm) measured in air. Characteristic frequencies are given in Hz. Arrows show the locations where the bulk (Rb) and the total material resistances (Rb+Rgb) were taken. Lines on the 300oC graph show the equivalent circuit fit to the experimental data.   123 6.5.2 Effect of water loss BIC did not show the change in conductivity (or ac scans) due to the water loss, as BIO did, although our previous work showed that BIC released more water than BIO with heating. This suggests that water incorporated in the structure of BIC does not provide significant proton conductivity that contributes to the total conductivity, probably because the water in BIC is mostly bound in the bulk, and proton conduction is mainly occurring in the grain boundary, as will be discussed later. 6.5.3 Change of phase At 400oC and 500oC, the bulk and grain boundary semicircles are harder to resolve, but there is still some distinction. The scans at 400oC and 500oC show that the total impedance of the GNP samples decreases and becomes smaller than the total impedance of the SS samples. Bulk impedance seems to be different for the GNP and SS samples. A gradual change in scans is noticed at 400oC over time, as shown in Fig. 6.8. In this figure, the change from the scan after 30 min to scan after 1 h is expected at the equilibration step when the temperature in the test is changed from a lower value to a higher value (300oC to 400oC). However, the further change to the scan after 1.5 h and 2 h is in the opposite direction, towards greater impedances. This change is most likely related to the phase change from a brownmillerite to a cubic perovskite structure that is reported to occur in BIC samples around 400oC, as discussed in Chapter 4. This change to the cubic structure causes an increase in bulk and grain boundary impedance.     124       Figure 6.8 – Ac impedance scans measured in air at around 400oC showing the change in impedance for BIC-GNP samples with time due to the phase change. Scans stabilize after 2 hours. 6.5.4 Equivalent circuit fit The ac impedance results for the BIC samples measured in air were fitted to the equivalent circuit shown in Fig. 6.6. For this case, bulk and grain boundary semicircles were possible to define and values for Rb, λb (or Cb), Rgb and λgb (or Cgb) are given in Table 6.2. An example of a fit is given in Fig. 6.7.  It can be seen in the table that the bulk capacitances are somewhat smaller than in the case of BIO. Grain boundary capacitances are about two orders of magnitude higher, resulting in a well defined semicircle in the ac impedance scans. Values for α in the CPE elements are generally close to one for the SS samples, while having lower values for the GNP values, especially for the grain boundary semicircle. This behaviour is expected and often happens in polycrystalline materials due to the anisotropy of the material, which is higher in the grain boundary area [103].    -10 -5 0 0 5 10 15 20 Z' (kΩcm) Z' '(k ΩΩ ΩΩ cm ) 30min 1h 2h 1.5h 3  125 Table 6.2 – Ce-doped Ba2In2O5 (BIC) samples measured in air – parameters determined by fitting the experimental ac scans from Fig. 6.7 to the equivalent circuit shown in Fig. 6.6. Note: Values at 100oC marked with * represent total resistance (Rb+Rgb), as bulk and grain boundary semicircles could not be resolved.         6.6 Ac impedance of BIO-GNP and BIO-SS in 50% H2/50% N2 6.6.1 Effect of grain size When undoped Ba2In2O5 was measured in 50% H2/50% N2 atmosphere, one semicircle was present in the ac impedance scan at 100oC, for both the GNP and the SS samples, as shown in Fig. 6.9. Bulk and grain boundary semicircles could not be resolved, and the electrode contribution is out of the frequency range (lower limit is 0.1 Hz). Total resistivity, taken as the intercept of the semicircle with the real axis, is quite close to the resistivity measured in air. SS samples showed higher total impedance, suggesting greater grain boundary impedance, as the bulk impedance should be the same for the same samples. This is most likely due to the decreased grain boundary area. At 200oC, while GNP samples still show one semicircle with resistivity close to that measured in air, the SS samples start to T, oC Rb, ohm·cm λ (or Cb), F/cm αb Rgb, ohm·cm λ (or Cgb), F/cm αgb 100 - - -  * 1.33E+07 1.19E-09 1.00 200 3.52E+05 2.51E-11 0.85 1.08E+05 6.20E-09 1.00 300 1.35E+04 1.70E-11 1.00 4.90E+03 1.01E-09 0.90 400 1.10E+04 1.45E-11 1.00 2.60E+03 4.82E-10 1.00 500 3.70E+03 1.43E-11 1.00 8.00E+02 9.90E-10 1.00 T, oC Rb, ohm·cm λ (or Cb), F/cm αb Rgb, ohm·cm λ (or Cgb), F/cm αgb 100 - - -  * 3.36E+07 3.01E-10 0.85 200 3.52E+05 2.51E-11 0.85 1.69E+06 3.93E-09 0.70 300 1.35E+04 1.70E-11 0.82 3.41E+04 9.90E-09 0.59 400 7.60E+03 2.10E-12 0.90 2.20E+03 1.13E-09 0.75 500 1.02E+03 5.95E-12 1.00 5.00E+02 2.95E-09 0.90 SS  (grain size≈ 60 nm) GNP  (grain size≈ 40 nm)  126 show two distinguishable semicircles due to the bulk and the grain boundary contribution. The grain boundary seems to contribute much more to the total resistance. This suggests a significant role that grain boundary plays in transport of charge carriers (most likely protons), and that resistance is lowered by a larger grain boundary area.  For these SS samples, the grain boundary contribution decreases at 300oC, while bulk resistance becomes predominant, which is related to water loss from the grain boundary, as explained later. Fig. 6.9, inset for the scans at 300oC, shows the ac impedance scan for the BIO-GNP sample. A radical decrease in resistivity at 300oC and a change in the ac impedance scan occur. The total resistance drops from ~ 1000 kΩcm at 200oC to ~ 46 Ωcm at 300oC. A more detailed scan at 300oC is shown in Fig. 6.10.               127                             Figure 6.9 – Ac impedance scans for the BIO-GNP (grain size ~40 nm) and BIO-SS samples (grain size ~60 nm) measured in 50% H2/50% N2. Characteristic frequencies are given in Hz. Arrows show the locations where the bulk (Rb) and the total material resistance (Rb+Rgb) were taken. Lines on the 200oC graph show the equivalent circuit fit to the experimental data. -5 -4 -3 -2 -1 0 0 10 20 30 Z (Ω cm) 1.5 4.5k 112 193k -0.10 -0.05 0.00 0.8 0.9 1.0 Z (Ω cm) Z'  ( ΩΩ ΩΩ cm ) 3.5k 353 -0.10 -0.05 0.00 3.0 3.1 3.2 3.3 Z'  ( ΩΩ ΩΩ cm ) 8.5 39.0 39.5 40.0 40.5 Z (Ωcm) 2k 20 134 63 400oC Rb+Rgb Rb+Rgb ′ -15 -10 -5 0 0 5 10 15 Z (kΩcm) Z'  (k ΩΩ ΩΩ cm ) -0.3 -0.2 -0.1 0.0 45 46 47 48 49 50 Z (Ωcm) Z'  ( ΩΩ ΩΩ cm ) 7.5 19k 342k 1.6k 134 300 oC Rb+Rgb Rb Rb+Rgb ′ -1500 -1000 -500 0 0 500 1000 1500 Z (kΩcm) 112 2.6k 19k 4.5k 47 0.1 200oC Rb+Rgb Rb -100000 -50000 0 0 50000 100000 Z (kΩcm) GNP SS BIO in 50%H2-50%N2 0.1 17 0.1 10 100oC Rb+Rgb Z" ( ΩΩ ΩΩ cm ) Z" (k ΩΩ ΩΩ cm )   ′ ′ 500oC ′ ′ ,  ,  ′ ′  128 Data represented by squares in Fig. 6.10 are the actual measured data taken at 300oC, showing inductive contribution in the ac impedance scans due to the parasitic inductance of the external wiring of the testing setup [104].  It is important to correct the measured data for the contribution of the inductance, as it can affect the result for the sample resistance.  This can be done by calibrating the cell by shorting and subtracting the cell inductance contribution from the actual test data. Data represented by circles in Fig. 6.10 show the corrected data. This correction was applied to all ac impedance scans that demonstrated inductive behaviour. The low frequency semicircle represents the electrode resistance. The total resistance of the material was taken to be value Rb+gb on the graph. This value proportionally changed with the thickness of the samples, which confirmed that Rb+gb corresponds to the total resistance of the material. The shown semicircle is associated to the electrode impedance, while the tail in the high frequency range is the beginning of the material impedance semicircle (not fully shown because it is out of the frequency range).          Figure 6.10 – Ac impedance of Ba2In2O5 in 50% H2/50% N2 at 300oC - correction for the inductance of the external wiring; □ measured data; ○ data corrected for the inductance. -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 45 46 47 48 49 50 Z' (Ωcm) Z' ' ( ΩΩ ΩΩ c m ) BIO-GNP 300oC in H2 Rb+gb  129 Ac impedance scans for both BIO SS and GNP samples at 400oC, shown in Fig. 6.9, are similar to impedance scans for the GNP sample at 300oC, but with lower values of impedance. SS samples with smaller grain boundary area had ten times higher total resistance, confirming that the grain boundary plays an important role in conduction of protons. 6.6.2 Effect of water loss Fig. 6.11 shows a change over time that occurs at 300oC for the SS samples. The impedance of the sample decreases, especially in the grain boundary area, shown as the decrease in the size of the grain boundary arc. This change is again related to the loss of water at around 300oC. However, the change is opposite to what was seen in air. In this case the total impedance is decreasing and conductivity improving over time. As discussed in Chapter 4, this is due to the loss of water from the structure and direct hydrogen incorporation into the structure, providing improved protonic conductivity.  As it can be seen in the figure, while bulk impedance is not affected by this change, the grain boundary area is significantly affected. This suggests that most of the water is incorporated in the grain boundary region and that hydrogen also incorporates mostly in that region.         130       Figure 6.11 – Change in ac impedance scan for a BIO-SS (top) sample measured in 50% H2/50% N2 atmosphere at 300oC. The change is due to the loss of water and direct incorporation of hydrogen that provides proton conductivity.  Effect of water loss was even more enhanced for the BIO-GNP sample. Due to this effect a significant change between the ac scans at 200oC and 300oC for the BIO-GNP samples was present when measured in hydrogen-containing atmosphere. To better understand this change, more studies were performed in this temperature range. The ac impedance scans measured between 200oC and 300oC, as well as over the 3-hour period at 300oC are shown in Fig. 6.12. Starting at 240oC, the Warburg component of the impedance scan appears, revealing a diffusion process that affects the charge transfer. This is again due to the loss of water from the structure. Total impedance of the material decreases with the temperature. The Warburg component is present up to 290oC. The scan at 290oC has a similar shape to the BIO-SS samples at 300oC (Fig. 6.9), showing the bulk contribution and a depressed grain boundary contribution. The main resistance seems to be in the bulk of the material. The first scan at 300oC after 30 min in Fig. 6.12 (top), is shown as one semicircle. The change at 300oC over a 3-hour period is shown in Fig. 6.12 (bottom). Impedance continues to decrease, until it reaches a low value of 46 Ωcm. The shape of the scans changes -20 -10 0 0 10 20 30 40 Z' (kΩ cm) Z' ' (k ΩΩ ΩΩ cm ) 30min 1h 3h 2h BIO-SS  131 to show a semicircle associated to the electrode impedance, and a high frequency tail that represents the beginning of the material impedance semicircle.  This short study confirms that as the water leaves the structure between 240oC and 300oC, resistivity of the sample in the hydrogen-containing atmosphere decreases several orders of magnitude, most likely due to the incorporation of hydrogen in the structure and proton conductivity. Decomposition of the sample is excluded as the reason for this high conductivity, as neither X-ray diffraction nor the Raman spectra of the samples after the testing show any change in composition, as discussed later in Chapter 7, Section 7.4. To our knowledge, the effect of water loss and direct incorporation of hydrogen on the enhanced improvement in conductivity of Ba2In2O5 has not been reported in the literature for this or other related materials.               132                  Figure 6.12 – Change in ac impedance scans for BIO-GNP samples in 50% H2/50% N2 with loss of water and incorporation of hydrogen between 260oC and 300oC (top) and at 300oC over time. 6.6.3 Decomposition at 500oC At 500oC, for both GNP and SS samples, initial low impedance increased with time and the shape of the ac scans changed, as shown in Fig. 6.13. When the measurement was continued for 24 h, the impedances continuously increased. X-ray diffraction analysis after conductivity testing showed decomposition of the sample to BaCO3 and elemental indium. -5000 -4000 -3000 -2000 -1000 0 0 1000 2000 3000 4000 5000 Z' (Ω cm) Z' ' ( ΩΩ ΩΩ cm ) 260oC 280oC 290oC 300oC-first scan 260oC 280o 290o 3 0o  – 30 min At 300oC over time 3h 1.5h 2h -1000 -500 0 0 500 1000 Z' (Ω cm) Z' ' ( ΩΩ ΩΩ c m ) 1h 30 min  133 Gradual increase in the resistance and appearance of additional contributions to the total resistance is due to the decomposition and formation of the new phases.           Figure 6.13 – Change of ac impedance scans of Ba2In2O5 in 50% H2/50% N2 at 500oC with time due to decomposition to BaCO3 and elemental indium. 6.6.4 Equivalent circuit fit Table 6.3 gives the values for resistances and capacitances for BIO samples measured in 50% H2/50% N2 determined from fitting the ac impedance scans in Fig. 6.9 to the equivalent circuit presented in Fig. 6.6. Scans at lower temperatures showed one or two semicircles with resistances decreasing with temperature. Bulk capacitances had somewhat higher values compared to measurements in air, but still in the same 10-11 F/cm range. Low resistances were achieved at 300oC and higher temperatures in the range between 0.9 Ωcm and 47 Ωcm. Capacitances could not be determined as semicircles were only partially shown, with no defined maximum. -200 -100 0 0 100 200 300 400 Z (Ωcm) Z'  ( ΩΩ ΩΩ c m )  24h  10h -5 0 5 0 10 20 30 Z (Ωcm) Z'  ( ΩΩ ΩΩ c m )  1h  5h ′ ′ ′  134 Table 6.3 – Ba2In2O5 (BIO) samples measured in 50% H2/50% N2 - parameters determined by fitting the experimental ac scan shown in Fig. 6.9 to the equivalent circuit in Fig. 6.6. Note: Values marked with * and ** represent total resistance (Rb+Rgb), as bulk and grain boundary semicircles could not be resolved, where ** marked the cases where full semicircles were not present and therefore values for λ could not be determined.  6.7 Ac impedance of BIC-GNP and BIC-SS in 50% H2 /50% N2 6.7.1 Effect of dopant Fig. 6.14 shows the ac impedance scans measured in 50% H2/50% N2 for the Ce-doped Ba2In2O5 samples (BIC) made by the GNP and SS method. Effect of Ce-dopant is investigated. At 100oC, for both the GNP and SS samples, one semicircle can be seen, with impedance one order of magnitude lower than when the same samples were measured in air. They are also lower than the impedances in hydrogen-containing atmosphere for BIO samples, showing that Ce as dopant improves the conductivity of Ba2In2O5 in hydrogen- containing atmospheres at low temperatures. At 200oC, for both the SS and GNP sample, two semicircles can be distinguished. The semicircle associated to the bulk contribution to the T, oC Rb, ohm·cm λ (or Cb), F/cm αb Rgb, ohm·cm λ (or Cgb), F/cm αgb 100 - - -     * 9.74E+07 9.57E-11 0.80 200 1.19E+05 7.04E-11 0.90 1.09E+06 3.11E-09 0.70 300 6.30E+03 7.39E-11 0.88 2.00E+03 4.77E-08 0.75 400 - - -   ** 3.91E+01 - - 500 - - - decomp. - - T, oC Rb, ohm·cm λ (or Cb), F/cm αb Rgb, ohm·cm λ (or Cgb), F/cm αgb 100 - - -     * 6.54E+07 2.43E-10 0.88 200 1.19E+05 7.04E-11 1.00 7.70E+05 9.00E-11 1.00 300 - - -   ** 4.67E+01 - - 400 - - -   ** 3.07E+00 - - 500 - - -   ** 9.00E-01 - - SS (grain size≈ 60 nm) GNP (grain size≈ 40 nm)  135 impedance is the same for both samples, as expected, while semicircles due to the grain boundary contribution are larger for the SS samples. This effect is similar to what BIO-SS has shown, suggesting higher resistance in the grain boundary region for the sample with the smaller grain boundary area. Larger grain boundary area enables easier transport of charge carriers. GNP sample even shows that the bulk resistance is predominant in the case when the grain boundary area is large enough, showing that charge transport (most likely proton) is happening mainly in the grain boundary area. For all other temperatures up to 500oC, similar effects were noticed. This is opposite to what was noticed in the tests in air, where it was concluded that segregation of Ce4+ in the grain boundary area causes lower concentration of oxygen vacancies and, hence, higher resistance in this area. It seems that lower concentration of oxygen vacancies does not significantly affect proton conductivity, suggesting that mechanism of proton transport might not be related only or at all to oxygen vacancies, but also to interstitial sites, as suggested by Norby [80, 85].  136                 Figure 6.14 – Ac impedance scans for the BIC-GNP (grain size ~40 nm) and BIC-SS samples (grain size ~60 nm) measured in 50% H2/50% N2. Characteristic frequencies are given in Hz. Arrows show the locations where the bulk (Rb) and the total material resistance (Rb+Rgb) were taken. Lines on the 200oC graph show the equivalent circuit fit to the experimental data. -10 -5 0 0 5 10 Z (kΩcm) 35k 2300k 2300k 16 88k 500oC Rb Rb+Rgb ′ -30 -20 -10 0 0 10 20 30 Z' (kΩ cm) 48k 1000k 1000k 45 400 oC 17k Rb Rb+Rgb -20 -10 0 10 20 Z (kΩcm) Z'  (k ΩΩ ΩΩ cm ) 150 7k 2500k 2500k 300oC Rb+Rgb Rb ′ -150 -100 -50 0 0 50 100 150 Z (kΩcm) 145k 145k 1k 13 406 200oC Rb+Rgb Rb ′ -4000 -3000 -2000 -1000 0 1000 2000 3000 4000 Z (kΩ cm) GNP SS 8k 0.1 4k 10k 100oC Rb+Rgb ′ BIC in 50%H2-50%N2 ′  137 6.7.2 Effect of water loss At 300oC, while BIO showed a radical decrease in impedance due to water loss and hydrogen incorporation, BIC did not show a significant effect and its impedance continued to linearly decrease.  As mentioned before, BIC was shown to contain an even larger amount of water in its structure than BIO, and it was expected that BIC shows a stronger effect of water loss. The lack of the same effect could be explained by the hypothesis concluded in Chapter 4, Section 4.5 that BIC possibly contains most of the water in the bulk (grain interior), while BIO has it in the grain boundary. If the proton transport is mainly occurring in the grain boundary region at this temperature, as some of our results suggest (Fig. 6.9, impedance for BIO and BIC sample at 300oC), then water collected in the grain boundary of BIO would affect it, while water collected in the bulk of BIC would have less effect on this transport. This different effect of water in BIO and BIC caused BIC impedance at 300oC to become higher than the impedance of BIO. 6.7.3 Change of phase At 400oC the impedance of the BIC samples increased (mostly bulk) due to already mentioned phase change from brownmillerite structure to a cubic structure. Grain boundary contribution became negligible for the GNP sample at 400oC and 500oC. 6.7.4 Equivalent circuit fit Table 6.4 gives values for Rb, λb (or Cb), Rgb and λgb (or Cgb) obtained by fitting the ac impedance scans measured for BIC in 50% H2/50% N2 to the equivalent circuit in Fig. 6.6. Bulk capacitances are somewhat higher than the bulk capacitances for BIC in air, but lower than capacitances for BIO in H2. Grain boundary capacitances are generally higher for BIC in  138 hydrogen-containing atmosphere than in air. Parameter αgb approaches values of 0.5 at 400oC for SS samples, and already at 200oC for GNP sample, showing that, although water loss is not affecting the resistance of the material, its diffusion is shown on the scans as a Warburg element.  Table 6.4 – Ce-doped Ba2In2O5 (BIC) samples measured in 50% H2/50% N2 - parameters determined by fitting the experimental ac scan shown in Fig. 6.8 to the equivalent circuit in Fig. 6.6. Note: Values marked with * represent total resistance (Rb+Rgb), as bulk and grain boundary semicircles could not be resolved.  6.8 Conductivity comparison  The total and bulk conductivities of the investigated samples in air and hydrogen- containing atmosphere were calculated from the impedance data and are shown in Fig. 6.15. Although total conductivities given here are the same as in Chapter 4 for BIO and BIC samples, they are presented again for the purpose of the comparison in this study.   T, oC Rb, ohm·cm λ (or Cb), F/cm αb Rgb, ohm·cm λ (or Cgb), F/cm αgb 100 - - -   * 3.06E+06 5.21E-12 0.95 200 3.90E+04 2.82E-11 0.82 9.50E+04 1.68E-09 0.85 300 3.00E+03 2.12E-11 0.80 1.70E+04 1.34E-09 0.70 400 6.00E+03 2.65E-11 0.93 4.00E+04 8.82E-08 0.55 500 3.10E+03 2.28E-11 0.93 1.00E+04 1.00E-06 0.60 T, oC Rb, ohm·cm λ (or Cb), F/cm αb Rgb, ohm·cm λ (or Cgb), F/cm αgb 100 - - -   * 9.84E+05 2.02E-11 1.00 200 3.90E+04 2.82E-11 0.92 2.50E+04 5.05E-07 0.55 300 3.00E+03 2.12E-11 0.90 1.40E+03 7.58E-07 0.50 400 6.00E+03 2.65E-11 0.93 6.00E+02 1.50E-09 1.00 500 3.10E+03 2.28E-11 0.93 2.00E+01 9.00E-08 1.00 SS  (grain size≈ 60 nm) GNP  (grain size≈ 40 nm)  139           Figure 6.15 – Total and bulk conductivities determined from the ac impedance scans; (a) BIO and BIC samples measured in air; (b) BIO and BIC samples measured in 50% H2/50% N2. Note: values for BIC bulk conductivities at 100oC are projected.  As it can be seen from Fig. 6.15a, BIO-GNP has higher conductivity in air than BIO-SS samples due to its larger grain boundary area, as concluded earlier in the ac impedance scan analysis. Bulk conductivity could not be determined for the BIO sample because it could not be separated from the grain boundary contribution in the impedance scans.  However, BIC- GNP samples show lower conductivity in air than BIC-SS samples (except at 400oC and 500oC, when the change to a cubic structure occurs), most likely due to the segregation of Ce-dopant in the large grain boundary area and resistance that it causes to the charge transport. Bulk conductivity for the BIC samples is close to the conductivity of the SS samples, suggesting that the main resistance for BIC in air comes from the grain boundary. However, even with this resistance, BIC shows higher total conductivity in air than BIO. -8 -7 -6 -5 -4 -3 -2 -1 0 1.2 1.6 2 2.4 2.8 1000/T(K-1) lo g σσ σσ  (S /c m ) Temp BIO-SS-total BIO-GNP-total BIC-SS-total BIC-GNP-total BIC-bulk         500  400      300           200                    100oC -8 -7 -6 -5 -4 -3 -2 -1 0 1.2 1.6 2 2.4 2.8 1000/T(K-1) lo g σσ σσ  (S /c m ) Temp BIO-SS-total BIO-GNP-total BIO-bulk BIC-SS-total BIC-GNP-total BIC-bulk         500  400     300            200                    100oC (a)   Air (b)   50% H2/50% N2  140 Fig. 6.15b reveals that BIO-SS and BIO-GNP have similar conductivity in hydrogen at low temperatures. However, as water leaves the structure and hydrogen incorporates at 300oC and above, BIO-GNP shows an order of magnitude higher conductivity. This confirms that hydrogen incorporation and transport is favoured in the grain boundary area. As for the bulk conductivity for the BIO sample, it could be determined only at 200oC and 300oC, while at other temperatures it could not be separated from the bulk contribution in the ac impedance scans. The value at 200oC is an order of magnitude higher than total BIO conductivity, revealing that most of the resistance to the transport of protons from hydrogen is in the grain boundary, due to the presence of water that most likely blocks the sites where protons would otherwise be incorporated. At 300oC, as the water leaves the structure, total conductivity radically increases, with bulk conductivity staying low. This is especially enhanced for the GNP sample, confirming that the process of water release and hydrogen incorporation mainly happens in the grain boundary area. It is interesting to note that BIO bulk conductivity is quite close to the BIC bulk conductivity. For the BIC samples, BIC-GNP samples showed higher conductivity than BIC-SS samples in hydrogen, revealing that grain boundary plays an important role in charge conduction. Total conductivity of the SS samples is lower than the bulk conductivity, while total conductivity of the GNP sample is similar, showing that smaller grain boundary area causes higher resistance to the charge transport. In general, the effect of microstructure is more enhanced in a hydrogen-containing atmosphere than in air, confirming that the grain boundary area plays a more important role in the charge transport in hydrogen than in air.    141 6.9 Summary Ac impedance spectroscopy was used to investigate the electrical properties of Ba2In2O5 and Ce-doped Ba2In2O5 and the effect microstructure (grain size and porosity), atmosphere, and temperature on its electrical impedance. Sintered sample pellets of brownmillerite- structured undoped Ba2In2O5 (BIO) and Ce-doped Ba2In2O5 (BIC) were prepared for the study. From the study on the correlation between conductivity and porosity, it was concluded that conductivity decreases with increasing sample porosity and a relationship between porosity and conductivity was established for the BIO and BIC samples in air and a hydrogen-containing atmosphere. The lowest porosity that could be obtained was in the range of 16 to 20%. Samples with a similar total porosity of 20% were used for the further studies. The solid-state samples (SS) resulted in a coarser microstructure and larger grain size (~ 60 nm) than the glycine-nitrate samples (GNP) which had a finer microstructure and smaller grain size (~40 nm). Ac impedance scans measured in air between 100oC and 500oC showed only one semicircle for BIO, due to overlapping of the bulk and grain boundary semicircles, while BIC showed two distinguishable semicircles at most temperatures allowing the bulk and grain boundary impedance to be determined. BIO-SS samples with their larger grains and, hence, smaller grain boundary area resulted in a higher total impedance compared to BIO-GNP samples, showing that the grain boundary area plays an important role in charge transfer in air, and that a larger grain boundary area favours the charge transport. During the measurement at 300oC BIO samples lose the water bound in their structure and their conductivity drops over time, since water provides some protonic conductivity.  142 Doping with Ce led to a decrease in total impedance in air and therefore improved conductivity of the material in air. BIC-GNP samples showed that most of the resistance is in the grain boundary area, most likely due to the Ce segregation in this area. The smaller grain boundary area of the SS samples resulted in a lower resistance. While the BIC samples did not show a significant effect of water loss at 300oC (conductivity was constant over time), they showed a change in scans at 400oC due to the phase transformation from the brownmillerite structure to a cubic structure. For measurements in hydrogen-containing atmospheres, the total impedance of the BIO SS and GNP samples at 100oC and 200oC was similar to the impedance measured in air, but at higher temperatures a significant decrease in impedance was noticed. Warburg impedance present in the scans between 240oC and 300oC revealed loss of water and the significant decrease in impedance at 300oC was associated with the direct incorporation of hydrogen in the structure to provide proton conductivity. The change in the grain boundary semicircle in the ac impedance scans for the BIO-SS samples in that temperature range suggested that the process of water release and hydrogen incorporation mainly occurs in the grain boundary area. Resistivities as low as a few ohms-cm to several tens of ohms-cm were achieved at 300oC and above. It was confirmed that this high conductivity was not related to a chemical decomposition. However, at 500oC and above both GNP and SS samples start to decompose to BaCO3 and elemental In, which was also indicated in the ac impedance scans. Again in hydrogen-containing atmosphere, the Ce-dopant improved conductivity at lower temperatures compared to the BIO sample. However, BIC samples did not show the significant decrease in impedance due to the release of water and hydrogen incorporation, as BIO did, even though it was shown that BIC contains a larger amounts of water in the  143 structure. This is because most of the BIC water is bound in the bulk and its release would not significantly affect the proton conductivity that is mainly occurring in the grain boundary area. The important role of the grain boundary area in proton conductivity was confirmed by the effect of the microstructure on the grain boundary semicircle of the impedance scan. The larger grain boundary area for GNP samples enables easier transport of protons, and therefore results in a smaller grain boundary impedance semicircle. Lower concentration of oxygen vacancies in the grain boundary area did not seem to affect proton transport, suggesting that proton transport could be also related to interstitial sites. However, this was not confirmed. Also, in hydrogen, the BIC samples experienced a change to a cubic structure at 400oC, which caused an increase in impedance, especially the bulk impedance. All ac impedance scans obtained for BIO and BIC samples in both air and hydrogen- containing atmospheres were fitted to a general equivalent circuit with R and CPE elements and values for the bulk and grain boundary resistances and capacitances were determined. In addition, Arrhenius graphs were constructed from the measured resistances to show the conductivity (total and bulk) changes with the temperatures for all cases. Analysis of the conductivity results confirmed the conclusions from the ac impedance scans.       144 7. Proton Conductivity of Ba2In2O5 and Stability in Hydrogen-containing Atmospheres  7.1 Synopsis High total conductivity of Ba2In2O5 in hydrogen-containing atmospheres reported in Chapter 5 creates an opportunity for use of Ba2In2O5 as a proton conductive material for a range of intermediate temperature electrochemical devices. However, it is important to determine the contribution of the proton conductivity to the total conductivity of Ba2In2O5, since Ba2In2O5 can exhibit mixed oxygen ion, proton and electron conductivity under different conditions. One of the possible methods to distinguish the oxygen ion, electronic and proton conductivity in Ba2In2O5 is to compare the total conductivity of a sample in three different atmospheres consisting of dry air, dry nitrogen and a hydrogen-containing atmosphere. This approach is based on the discussion given in Chapter 2, Section 2.3.3 and the results obtained in this work are presented here. Some performed tests, such as conductivity measurements in air and 50% H2/50% N2, are similar to the tests discussed in Chapter 5, but they were performed in more detail at this stage. Another widely used method to determine the proton conductivity contribution, or proton transport number tH+, in oxides is the e.m.f. method (electromotive force method), explained in Appendix B.2. The details of the measurement method used in this project are explained in Chapter 3, Section 3.7. The results are discussed here. Beside good proton conductivity, stability of Ba2In2O5 in hydrogen containing atmospheres is crucial if this material is to be used as a proton conductive electrolyte. Any decomposition of the material would affect its mechanical and electrical properties. The  145 results of the stability study for Ba2In2O5 in a hydrogen-containing atmosphere are also presented in this chapter. 7.2 Conductivity of Ba2In2O5 in air, nitrogen and hydrogen In order to distinguish the oxygen ion, electronic and protonic conductivity of Ba2In2O5, conductivity measurements in three different atmospheres were performed consisting of air, nitrogen and a hydrogen-containing atmosphere in the temperature range between 100oC and 500oC. Conductivity in air. The electrical conductivity of Ba2In2O5 in air was measured at 100oC, 200oC, 300oC, 400oC and 500oC. It was noticed that the conductivity of the samples depended on the treatment of the samples before the conductivity tests. Longer exposure of the sample to ambient air seemed to improve the conductivity, while heating the sample before the test seemed to reduce the conductivity. To study this behavior, two cases were investigated for each temperature tested. For Case I a fresh sample was exposed to air at 25oC for three days before AC conductivity measurement and for Case II a fresh sample was heated in air to 500oC for 3h before AC conductivity testing. Fig. 7.1 shows the Arrhenius plots of Ba2In2O5 conductivity in air for the two cases revealing different behavior. For both cases conductivity increases with the temperature. However, for Case I, a sudden drop in conductivity occurs between 300oC and 400oC, followed by a further increase in conductivity at higher temperatures. This behavior, due to the water loss from the structure, is discussed in Chapter 4 and 5.     146             Figure 7.1 – Arrhenius plot of Ba2In2O5 conductivity in air and N2: Case I (hollow symbols): fresh sample exposed to ambient air for three days before AC conductivity measurement in air (○) or  N2 (△); Case II (solid symbols): fresh sample treated to 500oC for 3h before conductivity testing in air (●) or N2 (▲).  The process of water release starts around 260oC and is completed around 400oC, as confirmed in Chapter 5. These results support the premise that water incorporated in the oxide structure at low temperatures provides protonic conductivity, which is also confirmed by Zhang et al. [33] by their conductivity measurements in humid air. As water leaves the structure, the proton conductivity does not contribute anymore to the total conduction causing it to decrease. The main charge carriers then become oxygen ions and electronic holes and conductivity starts to increase again with temperature. For Case II, where the sample is heated to 500oC removing all water from the structure prior to conductivity -9 -8 -7 -6 -5 -4 -3 -2 1.20 1.60 2.00 2.40 2.80 1000/T(K-1) lo g σσ σσ  (S /c m )         500    400        300            200                      100oC  147 measurement, the only conductivity that was measured from the start was that from the oxygen ions and electron holes.  Conductivity in nitrogen. For both Case I and Case II Ba2In2O5 exhibited similar conductivity behavior in a nitrogen atmosphere compared to air, with only slightly lower conductivity, as shown in Fig. 7.1. The difference in conductivity in air and nitrogen was, for example, 2.0·10-10 S/cm at 100oC and 9.1·10-6 at 500oC when measured in Case II. According to equation 2.51 in Chapter 2, this difference shows the contribution of p-type electronic conduction in air, which increases with temperature, but is still very low in the temperature range from 100oC to 500oC. Conductivity in hydrogen-containing atmosphere. Again, two cases were investigated for each temperature. For Case I the fresh sample was exposed to air for three days to absorb water from the atmosphere before ac conductivity measurement, and for Case II the fresh sample was heated to 500oC for 3h before testing. Fig. 7.2 shows the comparison of the measured conductivity for the two cases.           148           Figure 7.2 – Arrhenius plot of Ba2In2O5 conductivity in 50% H2/50% N2: Case I (♦): fresh sample exposed to air for three days before conductivity measurement; Case II (◊): fresh sample heated to 500oC for 3h before testing. Arrows show the conductivities on heating and on cooling.  At lower temperature (e.g., 100oC) a somewhat higher conductivity was measured for Case I than for Case II, similar to that of Ba2In2O5 conductivity in air. This is due to the protonic conductivity provided by water bound in the material. As the temperature increased, a transitional stage was noticed for Case I between 200oC and 300oC, followed by surprisingly high conductivities around and above 300oC. The observed transitional stage is again due to the water release from the Ba2In2O5 structure. With further temperature increase most of the water is released and hydrogen incorporates into the structure, significantly improving the conductivity.  For Case II, most of the bound water in the sample is released to the atmosphere during the heating before testing. Lower conductivity was measured at 100oC and 150oC than -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 1.20 1.60 2.00 2.40 2.80 1000/T(K-1) lo g σσ σσ  (S /c m )         500    400       300          200                    100oC  149 in Case I during the heating stage. The transitional range between 200oC and 300oC was absent compared to Case I. However, a sharp increase in conductivity above 150oC was noticed, most likely due to the enhanced mobility of protons above this temperature. Conductivities were significantly higher for Case II versus Case I in the 200oC to 300oC range, beyond which similar conductivities were achieved. On cooling, when the temperature was gradually reduced, the measured conductivity at 400oC, 300oC and 200oC was similar to that previously measured when increasing temperature. However, measured conductivities at 150oC and 100oC were higher in the cooling stage. When the atmosphere was switched from hydrogen back to nitrogen or oxygen results were repeatable indicating no apparent irreversible processes. 7.3 Proton conductivity of Ba2In2O5 In order to determine if the high conductivity of Ba2In2O5 in hydrogen-containing atmospheres is due to the proton conduction, oxygen ion or electronic conduction, a comparison between total conductivities measured in air, nitrogen and 50%H2-50%N2 was done and is shown in Fig. 7.3. For comparison, the conductivity of Nafion 117 submerged in water at 80oC (0.18 S/cm) was added [2].         150          Figure 7.3 – Comparison of Ba2In2O5 conductivity in different atmospheres: ● in air; △ in N2; ◊ in 50%H2/50%N2; ♦ in 48%H2/49%N2/3% steam; □ sample decomposed. Note:  In all cases samples were heated to 500oC before testing to remove water.  A large difference in the conductivities in hydrogen atmosphere compared to air and nitrogen can be found at all temperatures, with the conductivity in hydrogen atmospheres 3-5 orders of magnitude higher. This improvement in conductivity in hydrogen-containing atmospheres is larger than usually reported in the literature for related materials, which is in the range between 1-3 orders of magnitude (e.g., Ce-doped Ba2In2O5 [40], Pr-doped BaCeO3 [105], Tb-doped BaCeO3 [106].) From these results and equation 2.53 in Chapter 2, it can be concluded that the contribution of the oxygen ion and p-type electronic conduction is negligible in hydrogen atmospheres at temperatures above 200oC. This suggests that the dominant charge carriers are protons. The measured conductivities of 0.018 S/cm to 0.32 S/cm in the 300oC to 400oC range are much higher than any reported conductivity for perovskite-related materials in hydrogen [107]. Although even higher conductivities are achieved above 400oC, the materials decompose at 500oC and above. -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 1.20 1.60 2.00 2.40 2.80 1000/T(K-1) lo g σσ σσ  (S /c m )         500    400      300             200                       100oC Nafion  151 7.3.1 Proton transport number by e.m.f. method Proton transport numbers, tH+, for Ba2In2O5 were determined at 100oC, 200oC, 300oC, and 350oC by the e.m.f. method, using a concentration cell PH2I, Pt║Ba2In2O5║Pt, PH2II. When a gradient in hydrogen activity without humidification was applied across the sample, the measured e.m.f. was found to be negligible. However, when humidification was applied, much higher e.m.f.s were measured. It may be possible that a surface reduction process occurs in a highly reductive atmosphere containing hydrogen, affecting the e.m.f. measurement. With application of humidification, this process is prevented, enabling measurement of the actual e.m.f. across the sample. To make sure that humidification does not significantly affect the conductivity in hydrogen-containing atmosphere, ac conductivity was measured in a 48% H2/49% N2/3% H2O atmosphere at each temperature and compared to a 50% H2/50% N2 atmosphere. As shown in Fig. 7.3, the conductivity was slightly lower in case of humidification. E.m.f. test was then performed with the humidification. Fig. 7.4 shows the change in voltage across the sample as the atmosphere was changed from 48% H2/49% N2/3% H2O in both compartments, to 80% H2/15% N2/5% H2O in compartment I, measured at 300oC. Table 7.1 summarizes the results of tests and the calculated proton transport numbers at different temperatures. The results reveal pure proton conductivity at lower temperatures of 100oC and 200oC with high but decreasing proton transport numbers at higher temperatures. The proton transport numbers at 400oC and above could not be determined, as the sample mechanically decomposed in humid atmospheres at such high temperatures.    152           Figure 7.4 – Measured e.m.f. across the Ba2In2O5 sample as a function of time before and after increase in hydrogen activity in compartment I at 300oC (data corrected for unwanted voltage contributions).  Table 7.1 – Measured e.m.f. across the cell (80%H2/15%N2/5%H2O), Pt║Ba2In2O5║Pt, (48%H2/49%N2/3%H2O) with standard deviation and calculated proton transport numbers (tH+) and proton conductivity at different temperatures.  Proton conductivities measured by the e.m.f. method place Ba2In2O5 within the “gap” for the available proton conductive materials in the intermediate temperature range between 200oC and 500oC, as shown in Fig. 7.5.  T (oC) e.m.f. (mV) Theoretical e.m.f. (mV) (Eq. 2.54) tH+ Proton conductivity (S/cm) 100 8.2±0.4 8.2 0.99 6.2·10-6 200 10.3±0.5 10.4 0.99 8.6·10-4 300 10.6±0.5 12.6 0.84 0.015 350 10.1±0.3 13.7 0.74 0.09 -0.020 -0.010 0.000 0.010 0.020 0 5 10 15 20 Time (min) e. m . f. (V ) PI:  48%H2/49%N2/3%H2O PII: 48%H2/49%N2/3%H2O PI:  80%H2/15%N2/5%H2O PII: 48%H2/49%N2/3%H2O  153         Figure 7.5 − Proton conductivity of selected solid proton-conductors reported in the literature [33, 40, 45-47, 61-63] compared to the proton conductivity of Ba2In2O5 determined in this work. Note: Labels on this graph correspond to the labels in Fig. 2.6 (Chapter 2).  7.4 Stability of Ba2In2O5 in hydrogen containing atmospheres It is important to determine the stability of Ba2In2O5 in hydrogen-containing atmospheres especially in the temperature range where it exhibits high conductivity, i.e., 300- 500oC. To determine if extended exposure to a hydrogen-containing atmosphere has an effect on conductivity and chemical stability of Ba2In2O5, ac conductivity was measured in 50%H2/50%N2 at 300oC, 350oC, 400oC, 450oC and 500oC for 24 hours. X-ray diffraction measurement was performed before and after the testing. The results shown in Fig. 7.6 revealed that stable conductivities were obtained at all temperatures, except at 500oC. In addition, a 24 h test was performed at 480oC, and confirmed stable conductivity. X-ray Ba2In2O5  154 diffraction analysis before and after conductivity testing showed no change in composition at temperatures below 500oC, as shown in Fig. 7.7.          Figure 7.6 – Ac conductivity of Ba2In2O5 in 50%H2/50%N2 at 300oC, 350oC, 400oC, 450oC and 500oC for 24 hours.          Figure 7.7 – XRD scans of Ba2In2O5 before and after ac conductivity testing in 50%H2/50%N2 at 400oC for 24 hours showing no change in structure.  -3 -2 -1 0 0 8 16 24 Time (h) lo g σσ σσ  (S /c m ) 450oC 400oC 350oC 500oC 300oC 20 30 40 50 60 70 80 2 Theta In te n s ity Before testing in H2 After 24h testing in H2  155 At 500oC, initial high conductivity decreased after first 3 hours, and continued to decrease with time, as shown in Fig. 7.6. This decrease in conductivity is due to the decomposition of the sample and contribution of the new phases to the total resistivity of the sample. The gradual appearance of the new phases is confirmed by the X-ray diffraction of the sample after 10 hours and after 24 hours, as shown in Fig. 7.8. It can be seen that Ba2In2O5 is decomposing with time to BaCO3 and elemental indium. Carbon in BaCO3 is most likely originating from the incorporated CO2 in the structure of Ba2In2O5, as reported by Hashimoto et al. [74].           Figure 7.8 – XRD scans of Ba2In2O5 after ac conductivity testing in 50%H2/50%N2 at 500oC for 24 h: ○Ba2In2O5, ▲ BaCO3, ● elemental In.  In addition to X-ray measurements, Raman spectra were obtained for a fresh sample of Ba2In2O5, a sample that was tested in 50% H2/50% N2 for 24 h at 400oC, and a sample that decomposed at 500oC, as shown in Fig. 7.9. A few characteristic peaks for Ba2In2O5 [91, 108] were observed in the range between 100 cm-1 and 610 cm-1, with the most intense one at 20 30 40 50 60 70 80 2 Theta In te n si ty Before testing After 10h After 24h  156 607 cm-1. In addition, a small peak at 2440 cm-1 is shown in Fig. 7.9 for Ba2In2O5. The figure shows no noticeable change between the fresh sample and the sample tested for 24 h in a hydrogen-containing atmosphere. However, the Raman scan of the decomposed sample showed a change in intensity of the peaks at 485 cm-1, 607 cm-1 and 2440 cm-1, as well as an additional peak at 2215 cm-1 that is associated with atomic indium and peaks at 696 cm-1 and 1063 cm-1 associated with BaCO3 [109]. Thus, Raman spectra confirmed no degradation of Ba2In2O5 at temperatures lower than 500oC, but decomposition to elemental indium and BaCO3 at 500oC.            Figure 7.9 – Raman spectra of Ba2In2O5 before testing (bottom scan); after testing in 50%H2/50%N2 at 400oC for 24h (middle scan) and after decomposition at 500oC (top scan); ○Ba2In2O5, ▲ BaCO3, ♦ elemental In.  In these studies, the testing confirmed stability of Ba2In2O5 over 24 hours in 50% H2/50% N2 at temperatures below 500oC, and decomposition at 500oC and above. Although 2100 2300 2500 2700 Raman shift (cm-1) In te n si ty Before testing After 24h After decomposition 100 300 500 700 900 1100 Raman shift (cm-1) In te n si ty Before testing After 24h After decomposition  157 literature reports the stability of Ba2In2O5 in reducing atmospheres (PO2=10-19 atm in Ar atmosphere or 1.5%H2 balanced by N2), up to at least 600oC, as discussed in Chapter 2, Section 2.3.5, it appears that the high concentration of hydrogen in our case is the cause for the decomposition. 7.5 Summary The conductivity and stability of Ba2In2O5 (brownmillerite phase) was tested in hydrogen-containing atmospheres in the temperature range from 100oC to 500oC. The conductivity was compared to the conductivity measured in air and nitrogen atmospheres. Conductivity measured in a hydrogen-containing atmosphere was almost five orders of magnitude higher than the conductivity measured in either air or nitrogen, suggesting that the protonic conductivity in Ba2In2O5 was predominant in the hydrogen-containing atmosphere. Conductivities in the range of 0.018 S/cm to 0.32 S/cm were measured in the 300oC to 400oC range in this atmosphere. Proton transport numbers determined by the e.m.f. method in a concentration cell: (80%H2/15%N2/5%H2O), Pt║Ba2In2O5║Pt, (48%H2/49%N2/3%H2O), were determined to be tH+=1 at 100oC and 200oC, 0.84 at 300oC and 0.74 at 350oC. Proton transport numbers could not be determined at 400oC and above as the samples mechanically decomposed in a humidified atmosphere. However, the samples were stable up to, and including 350oC. Chemical and electrical stability of Ba2In2O5 in hydrogen-containing atmospheres was confirmed over 24 hours in the temperature range from 300oC to 480oC by X-ray diffraction and Raman spectroscopy, as well as by consistent conductivity results over the time. At 500oC the conductivity gradually decreased, due to decomposition of Ba2In2O5 to  158 BaCO3 and elemental indium, which was confirmed by both X-ray diffraction and Raman spectroscopy. The measured high electrical conductivity of Ba2In2O5 in a hydrogen-containing atmosphere in the temperature range of 300oC to 480oC is promising for the development of intermediate temperature proton-conductive materials for a range of electrochemical applications. The following chapter discusses the evaluation of the performance of this material as a proton-conducting electrolyte for an intermediate temperature fuel cell.                  159 8. Evaluation of Ba2In2O5 for Use as an Electrolyte Material or Within the Anode in an Intermediate Temperature Fuel Cell  8.1 Synopsis The ultimate goal of the work in this thesis was to find a highly proton-conductive material among the investigated ceramics to be applied as an electrolyte in an intermediate temperature proton exchange electrolyte fuel cell. Additionally, the objective was to evaluate the applicability of this material within the anode structure of the fuel cell. The selected material, as discussed in the previous chapters, was Ba2In2O5 prepared by the glycine-nitrate method. The first part of this chapter discusses the results of the electrochemical testing of the Ba2In2O5 dense electrolyte. Open circuit voltage (OCV) and polarization curves were measured at different temperatures in the high conductivity range for Ba2In2O5 (300-480oC), while ac impedance spectroscopy and potentiostatic measurements were utilized for additional information. The chapter concludes with a short study on the Ba2In2O5 material used within the anode structure. Ac impedance spectroscopy was used to evaluate the performance of various anode compositions containing Ba2In2O5 and several different catalyst materials in a hydrogen- containing atmosphere. 8.2 Evaluation of Ba2In2O5 as the electrolyte 8.2.1 OCV measurements Open circuit potentials for the air, Pt║Ba2In2O5║Pt, 50%H2/50%N2 cells were measured over a range of temperatures, using the setup shown in Fig. 3.6 in Chapter 3. The test was  160 repeated three times, each time with a new sample, and average values were taken as the OCV values at each temperature. For the measurements over 3 h at each temperature, constant and stable OCVs were achieved: 0.81 V at 300oC (standard deviation for the three measurement ±2.1%), 0.69V at 350oC (st. dev. ±2.5%), 0.59 V at 400oC (st. dev. ±1.5%), 0.51 V at 450oC (st. dev. ±1.5%), and 0.50 V at 480oC (st. dev. ±2.1%), while at lower temperatures of 100oC and 200oC the values were not stable and consistent. The measured OCV values were compared to the calculated values of 1.12 V, 1.11 V, 1.10 V, 1.08 V, and 1.07 V at 300oC, 350oC, 400oC, 450oC and 480oC, respectively. The theoretical Nernst potential values were calculated for the temperatures of interest and partial pressures of oxygen of 0.21 atm and hydrogen 0.5 atm, using equations 2.10 and 2.12 and taking into account that produced water is in the gaseous state. The measured OCV values are lower than theoretical, which is reported as well for related materials, e.g., BaCe0.5Zr0.3Y0.16Zn0.04O3–δ [59, 110]. Measured OCV for this material was 0.8 V at 650oC between 97% H2/ 3% H2O and 97% O2/ 3% H2O, compared to the theoretical 1.08 V. The authors assign the difference to the leakage and/or the presence of some electronic conduction, which was not confirmed. In our work the difference between the measured and theoretical OCV values is most likely due to a contribution of the electronic conductivity, since thick samples (~2 mm) were used to avoid leaking. However, the relatively high OCVs measured indicate that the proton conductivity is predominant. 8.2.2 Polarization curves Potentiodynamic measurements were carried out using the air, Pt║Ba2In2O5║Pt, 50%H2/50%N2 cells. Fig. 8.1 shows the obtained polarization curves at 300oC, 350oC, 400oC, 450oC and 480oC. Surprisingly low performance at each temperature was obtained. Ohmic  161 resistance seems to play the greatest role in reducing the performance. Such a result was not expected since Ba2In2O5 showed high conductivity in this temperature range in hydrogen- containing atmospheres, as discussed in Chapter 5, 6 and 7 (over 0.018 S/cm). However, ac impedance spectra measured right before the potentiodynamic measurements for the air, Pt║Ba2In2O5║Pt, 50%H2/50%N2 cell showed high resistances, as shown in Fig. 8.1. The two semicircles in the impedance scans most likely correspond to the electrolyte resistance (high frequency semicircle) and the cathode resistance (low frequency arc). This was concluded based on the previous impedance measurements of the symmetrical cells in a hydrogen- containing atmosphere, and in air (Chapter 6), where it was determined that the anode resistance was very low (below 1 ohm at these temperatures), and the cathode resistance was in the range obtained in this study. The electrolyte and the cathode resistances obtained from the impedance scans of the air, Pt║Ba2In2O5║Pt, 50%H2/50%N2 cell are shown in Table 8.1. The high ohmic resistance of the electrolyte and the cathode is the main reason for the very low performance of the cells. This high electrolyte resistance was not in agreement with the high conductivities of the Ba2In2O5 electrolyte measured in hydrogen-containing atmospheres in the same temperature range. Introduction of the air on the cathode side seemed to have a strong effect on the conductivity of the electrolyte.        162 Table 8.1 – Electrolyte and cathode resistances measured by ac impedance spectroscopy for the air, Pt║Ba2In2O5║Pt, 50%H2/50%N2 cell before the potentiodynamic measurements. Anode resistance could not be resolved from the scans.                    T Measured area specific resistance -electrolyte- Measured area specific resistance -cathode- (oC) (Ohm cm2) (Ohm cm2) 300 10800 249200 350 4800 47200 400 2700 8400 450 900 2100 480 3200 13200 A: 50%H2/ 50%N2  C: air  163                       Figure 8.1 – Polarization curves (top) and associated impedance scans (bottom) measured for the air, Pt║Ba2In2O5║Pt, 50%H2/50%N2 cell at different temperatures. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 Current density (mA/cm²) Ce ll v o lta ge  (V ) 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 Po w e r de n s ity  (m W /c m 2 ) 350oC 450oC 300oC 400oC 480oC -40000 -30000 -20000 -10000 0 0 10000 20000 30000 40000 Re (ohm cm²) Im  (oh m  c m ²) 300°C 350°C 480°C 400°C 450°C  164 8.2.3 Potentiostatic measurements In order to investigate the reason for the behaviour explained in the previous section, a potentiostatic measurement was performed in the temperature range of interest (300- 480oC). The experimental approach is explained in Chapter 3, Section 3.8 and the test procedure is given in Appendix C.3. Under a constant potential of 0.5 V applied to the Pt║Ba2In2O5║Pt cell, the change in current was measured while the atmosphere on the cathode side was changed from N2 to air, with the constant 50%H2/ 50%N2 atmosphere on the anode side. Current densities changed from about 12 mA/cm2 while N2 was on the cathode side to almost zero as soon as air was introduced. Impedance measurement showed a high resistance in the latter case. Fig. 8.2 shows the potentiostatic measurement at 400oC, with repeated steps of introducing and stopping the air flow on the cathode side, and measured impedance before and after air was introduced. Table 8.2 gives the summary of the potentiostatic measurements and resistances obtained by ac impedance measurements for the temperature range between 300oC and 480oC.        165              Figure 8.2 – Potentiostatic measurements (0.5 V applied) for the Pt║Ba2In2O5║Pt cell at 400oC while 50%H2/ 50%N2 mixture was constantly flowing on the anode and air flow was started (solid arrows) and stopped (dashed arrows) at the cathode. On top: associated impedance spectra measured during the tests. -2 0 2 4 6 8 10 12 14 16 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Time (s ) Cu rr en t d en si ty  (m A/ cm 2 ) Air started Air stopped -6000 -4000 -2000 0 0 2000 4000 6000 Z' (Ohm cm2) Z"  (O hm  cm 2 ) A: 50%H2/50%N2 C: air Relectrolyte -50 -40 -30 -20 -10 0 0 10 20 30 40 50 Z' (Ohm cm2) Z"  (O hm  cm 2 ) A: 50%H2/50%N2 C: N2  Relectrolyte   166 Table 8.2 – Results of the potentiostatic measurements (0.5 V applied) and resistances obtained by ac impedance measurements for the Pt║Ba2In2O5║Pt cell.   The results of the potentiostatic measurements suggest that introduction of air on the cathode side causes a sharp drop in conductivity of the electrolyte. This behaviour is most likely related to the oxygen sorption into the Ba2In2O5 electrolyte on the cathode side. As oxygen fills the vacancies, the transport of the protons is hindered and conductivity drops. Performance at each temperature in Fig. 8.1 is affected by a balance between proton conductivity, level of oxygen sorption at the cathode side and blocking of the proton conductivity, as well as contribution of the electronic conductivity. Low electrolyte conductivity due to oxygen sorption in Ba2In2O5 makes this material not suitable for use as an electrolyte in the intermediate temperature fuel cell. A possible solution to this problem could be the application of a thin, dense layer of a proton-conductive material that would prevent contact of Ba2In2O5 and air. This material could have lower proton conductivity than Ba2In2O5 as it would be applied as a very thin layer, but it would have to be stable in contact with air. We did not pursue this approach/problem further in the thesis, but it is definitely a recommendation for further work. T Measured current density Measured area specific el. resistance (ASR) Measured current density Measured area specific el. resistance (ASR) (oC) (mA/cm2) (Ohm cm2) (mA/cm2) (Ohm cm2) 300 4.2 75 0.0020 18480 350 5.6 58 0.0216 7395 400 11.2 37 0.0150 4195 450 42.5 10 0.0050 2310 480 82.0 5 0.0020 4930 A: 50%H2/ 50%N2  C: N2 A: 50%H2/ 50%N2  C: air  167 8.3 Evaluation of Ba2In2O5 within the anode structure The purpose of this study was to investigate if Ba2In2O5 as a proton conductive ceramic could be used within the anode structure of an intermediate temperature fuel cell, in combination with different metal catalysts. The structure of the anode was supposed to resemble the porous cermet anode structure used in solid oxide fuel cells [111]. In a conventional SOFC the porous cermet is composed of the oxygen ion conductive YSZ support material and Ni catalyst that provides both catalytic activity and electronic conductivity. To provide a percolation path for electrons through the anode, a high enough content of Ni is needed (around 30 %wt). Too high a content can cause agglomeration of Ni particles, reducing its overall catalytic activity and blocking ionic transport. Also, such a high content of Ni would increase the thermal expansion coefficient of the electrode and cause higher mismatch with the electrolyte. In this study, Ba2In2O5 (BIO) was used as a proton conductive support in the anode structure, mixed with different metal catalysts (M=Pt, Ni or Fe). Figure 8.3 shows a schematic of the anode cermet structure used in this study, showing an interpenetrating network of pores and conductors − Pt, Ni or Fe for electrons and proton conducting ceramic for protons. The activity of each anode cermet in hydrogen-containing atmospheres was investigated by ac impedance spectroscopy, using a symmetrical cell BIO+M II BIO II BIO+M. The procedure of the catalyst and symmetrical cell preparation and testing is explained in the Experimental section in Chapter 3. Eight different samples were investigated, as given in Table 8.3. For the first four samples, Pt or mixed metal catalyst+BIO paste were applied to both sides of the sintered BIO electrolyte and heated to 800oC to decompose organic binders and solvents and partially sinter the metal catalyst, still  168 retaining the porosity. For the other four samples, the sintered BIO electrolyte disks with applied electrode pastes were heated to 1300oC for 6 hours to sinter the BIO material in the electrodes. Sintered BIO within the electrode was expected to provide a connected path for the proton transport and improve the electrode performance. The temperature of 1300oC was determined earlier (Chapter 3) as the temperature at which sufficient BIO sintering happens, but the material keeps the porosity needed for the electrode structure. In both cases, Ni- and Fe-based catalysts had to be reduced in 50%H2/50%N2 at 450oC from NiO and Fe2O3 to metallic catalysts. Pt catalyst was in its metallic form during the whole process.           Figure 8.3 – Schematic of the anode cermet structure, with proton conducting Ba2In2O5 ceramic support and Pt, Ni or Fe as catalyst.     e- H+ H2  2H++2e- H+ Pt or Ni or Fe metal catalyst Electrolyte Ba2In2O5 Ba2In2O5 ceramic Cermet anode  169 Table 8.3 – Electrodes used in a symmetrical cell for anode investigation.  Electrode  Treatment prior to testing Treatment prior to testing Pt (baseline) Heated to 800oC for 30min Heated to 1300oC for 6 h Pt+BIO Heated to 800oC for 30min Heated to 1300oC for 6 h Ni+BIO Heated to 800oC for 30min Heated to 1300oC for 6 h Fe+BIO Heated to 800oC for 30min Heated to 1300oC for 6 h  During the preparation stage, samples with Ni and Fe catalysts (that were in the form of Ni+/NiO and Fe2O3 at that stage, before reduction in hydrogen) that were heated to 1300oC changed their structure from Ba2In2O5 to Ba4In6O13 (in the case of Ni) and to BaFe0.5In0.5O2.5 and In2O3 (in the case of Fe), as confirmed by X-ray diffraction and shown in Fig. 8.4 a and b. Therefore, these samples were not used for further testing.  All other samples were stable (including the Pt and Pt+BIO samples at 1300oC) and did not change their composition during the preparation or testing in hydrogen.         Figure 8.4 – Fe+BIO and Ni+BIO based anodes after heating to 1300oC. Ni is in the form of NiO and Fe is in the form Fe2O3 at this stage (not reduced yet).  Ba2In2O5 Ba4In6O13 NiO Ba2In2O5 BaFe0.5In0.5O2.5 In2O3 20               30                 40                50                 60                70  20               30                 40                50                 60                70            80 2 Theta scale In te n sit y In te n sit y  170 Electrodes were tested by ac impedance spectroscopy in 50%H2/50%N2 at temperatures between 300oC and 480oC. At these temperatures BIO shows high proton conductivity (and stability) and is expected to contribute to the proton transfer in the anode. Fig. 8.5 shows the ac impedance scans obtained at 350oC for the symmetrical cells prepared with different electrodes.           Figure 8.5 – Ac impedance scans obtained at 350oC in 50%H2/50%N2 for the symmetrical cells prepared with different electrodes.  The semicircles in the impedance scans correspond to the resistances for the charge (proton) transfer between the electrode and the electrolyte (as determined earlier, Chapter 5). The tail in the high frequency range is a part of the semicircle that corresponds to the material (electrolyte) resistance, as determined in previous studies (Chapter 5 and 6). According to the results, the Pt electrode heated to 800oC has the lowest charge resistance. When the BIO ceramic was mixed with Pt and heated to 800oC, the BIO did not improve the  -0.3 -0.2 -0.1 0 13 14 15 16 17 Z' (Ωcm) Z' ' ( ΩΩ ΩΩ cm ) Pt-800°C Pt+BIO-800°C Pt+BIO-1300°C Pt-1300°C Ni+BIO-800°C Fe+BIO-800°C 1-Pt-800oC 2-Pt+BI -800oC 3-Pt+B -13 0oC 4-Pt-1300o 5-Ni+ I -800oC 6-Fe+BIO-800oC 1 2 3 4 5 6  171 characteristics of the electrode. In this case, a high surface area of contact between BIO and Pt in the cermet enables transfer of protons from the catalyst to the proton conductive material. However, BIO in the electrode is not sintered (connected) and the path for easy proton transport to the electrolyte does not exist. The resistance for the charge transfer between separated BIO particles most likely causes the higher total resistance. BIO+Pt and Pt-only electrodes heated to 1300oC prior to the testing showed lower performance than the same electrodes heated to 800oC. This is most likely due to the agglomeration and sintering of the Pt particles, hence lowering its catalytic activity. BIO+Pt- 1300oC cermet electrode showed a better performance than Pt-1300oC electrode. In this case, the BIO ceramic in the anode was sintered and therefore a connective path for proton transport to the electrolyte was enabled. BIO provides proton conductivity, while Pt provides electron conductivity and catalytic activity. This confirms that application of BIO within the anode structure does improve the performance of the anode. With a goal to investigate the application of non-noble metal catalysts instead of Pt, Ni and Fe catalysts were used in the cermet anode with BIO instead of Pt. As mentioned before, only electrodes heated to 800oC were tested. Their performance is evidently lower than that of the Pt or Pt/BIO. Part of the increased resistance is due to the fact that samples were not sintered and therefore BIO particles were not fully connected, but part is most likely due to the lower catalytic activity of the Ni and Fe catalysts. Improving the method of preparation with a sintering step for the BIO ceramic, but avoiding the previously discussed change in composition would possibly improve the performance. However, the charge transfer resistance for Ni/BIO and Fe/BIO anodes is still in the low range, making these metals possible anode catalysts for the intermediate temperature fuel cells.  172 8.4 Summary Ba2In2O5 ceramic material produced by the glycine-nitrate process was selected as a highly proton conductive material. This material was investigated as an electrolyte as well as within the anode structure for an intermediate temperature proton exchange electrolyte fuel cell. Open circuit voltages of the air, Pt║Ba2In2O5║Pt, 50%H2/50%N2 cells were measured, but found to deviate from the theoretical values most likely due to the contribution of the electronic conductivity. Polarization curves of the air, Pt║Ba2In2O5║Pt, 50%H2/50%N2 cell showed very low performance at all temperatures. Although Ba2In2O5 electrolyte showed high conductivity in hydrogen-containing atmospheres (over 0.018 S/cm), ac impedance of the above cell showed high electrolyte resistance, which resulted in the low performance. Potentiostatic measurements confirmed that the high resistance of the cell is due to oxygen sorption into the Ba2On2O5 electrolyte on the cathode side and blocking of the proton conductivity. Therefore, this material is not suitable for use as an electrolyte, unless a protective proton conducting layer is used to prevent this process. Ba2In2O5 (BIO) was investigated as a proton-conductive support in the anode structure, mixed with different metal catalysts (M=Pt, Ni and Fe). Ac impedance measurements for the symmetric BIO+M II BIO II BIO+M cells in 50%H2/50%N2 revealed the following order for the activity of the anode: Pt-800oC > Pt+BIO-800oC > Pt+BIO-1300oC > Pt-1300oC > Ni+BIO-800oC > Fe+BIO-800oC. The results showed that Pt is still the best choice for the anode catalyst. However, although Ni and Fe showed the highest charge transport resistance, these values are still in the ohms range, and therefore possible for application as non-noble  173 catalysts for the intermediate temperature fuel cells. The results of this investigation revealed that application of Ba2In2O5 within the anode structure is beneficial and improves the performance of the anode, but only if the anode is sintered to provide a good path for proton transport.                     174 9. Conclusions Fuel cell technology is holding great promise for efficient, sustainable and clean power generation. With its wide range of applications including power generation systems, transportation and portable electronics, commercialization of this technology would facilitate easier transition from current fossil fuel economy to a more sustainable and environmentally friendly energy economy. In order for fuel cell technology to become widely accepted and commercialized, further development of fuel cell components and materials is needed to improve the efficiency and lower the cost of fuel cells. An important aspect of fuel cell development is research in the area of materials for fuel cells, including electrolyte and electrode materials, and system components. A special area of focus is on materials development for application in intermediate temperature fuel cells, operating between 200oC and 500oC, as these fuel cells have many advantages over low or high temperature fuel cells. This thesis project had the objective to develop a ceramic proton-conducting material with sufficiently high proton conductivity (over 10-2 S/cm at 200oC; Nafion is 0.18 S/cm at 80oC), to be used as a dense electrolyte and within the anode structure of an intermediate temperature fuel cell. The investigated materials were based on oxygen deficient ceramic oxides, undoped and Ce- and La-doped Ba2In2O5. A number of different materials were prepared, characterized and electrochemically investigated. Among them, undoped Ba2In2O5 was selected as the material with the highest proton conductivity and tested within a fuel cell, both as the electrolyte and within the anode of the fuel cell. The following pages summarize the conclusions from each stage of this work, followed by the significance of the research and recommendations for future work.   175 9.1 Materials preparation and characterization Five different compositions including undoped and Ce- and La-doped Ba2In2O5 were synthesized by the solid-state process and the glycine-nitrate process: Ba2In2O5 (BIO), Ba2In1.5Ce0.5O5.25 (BIC), Ba2In1.5Ce0.25La0.25O5.125 (BICL025), Ba2In1Ce0.5La0.5O5.25 (BICL05) and Ba2In1.5La0.5O5 (BIL). The materials were characterized in terms of their crystal structure, powder particle and grain size, thermal properties, and stability in humid atmospheres. X-ray powder diffraction of the powders produced by both the solid-state process and the glycine-nitrate process confirmed the desired orthorhombic brownmillerite structure for all compositions when the final calcining temperature was between 1100oC to 1400oC. As- prepared solid-state powders had a mean particle size diameter of 1.5 µm and a grain size of ~40 nm, while the powders produced by the glycine-nitrate process where much finer with a mean particle size diameter of 150 nm and a grain size of ~10 nm. The structure changes of the as-produced powders with temperature were studied using temperature profile X-ray diffraction and DSC/TGA method. The tests revealed that all compositions undergo one or more phase transformation between the brownmillerite structure and the cubic structure in the temperature range of 400oC to 1500oC. These changes are due to the order-disorder transformation, when the ordered brownmillerite structure changes to a disordered cubic structure and vice versa, and are accompanied by oxygen absorption and desorption from the structure, as confirmed by TGA. TGA results also showed that all materials experience a weight loss in the 300-400oC temperature range, as well as in the 500-750oC temperature range, which was concluded to be due to the loss of H2O and CO2, respectively, from the structure. The comparison of the weight loss due to H2O release from the structure of the  176 different materials led to a conclusion that Ba2In2O5 accumulates water more in the grain boundary region then in the bulk material. However, other compositions appear to accumulate water more in the bulk of the material. In the stability study, all the investigated materials generally showed instability in highly humid atmospheres up to 70oC, but were stable in liquid water at 25oC, and in 3% mol steam in air up to 500oC for at least 24 hours. 9.2 Electrical conductivity of the investigated materials Total electrical conductivities of all investigated compositions (except for BICL05, which could not be sintered into a testing pellet), prepared by both methods, were measured by ac impedance spectroscopy and compared in air and 50% H2/50% N2 in the temperature range of 100oC to 500oC. The final goal was to select a candidate material with the highest conductivity and stability in hydrogen-containing atmospheres to act as a proton conductive electrolyte for an intermediate temperature fuel cell. The results of the study showed that atmosphere, dopant and microstructure have a significant effect on the conductivity of these materials. All compositions showed higher conductivity in a 50% H2/ 50% N2 hydrogen atmosphere than in air. Ce-doped Ba2In2O5 had the highest conductivity in air, while La-doped Ba2In2O5 exhibited the lowest conductivity. In hydrogen-containing atmosphere, Ba2In2O5 showed the highest conductivity reaching values of over 0.018 S/cm at 300oC and above, while La-doped Ba2In2O5 again showed the lowest conductivity. Samples with a smaller grain size (GNP samples) showed higher conductivity than samples with a larger grain size (SS samples) in both atmospheres. This shows that the grain boundary plays an important role in the conductivity of these materials. In hydrogen, the conductivities were an order of magnitude higher for the GNP samples, suggesting an important role of the grain boundary in the conduction mechanism of protons. The difference  177 was not that significant in air, suggesting that the grain boundary plays a less significant role in the conductivity of oxygen ions and holes than that of protons in this temperature range. In addition, it was concluded from the ac impedance scans that doping of Ba2In2O5 with Ce caused higher resistance to the oxygen ion conductivity in the grain boundary area, most likely due to segregation of Ce in that area.  Doping, however, did not increase the resistance to the proton conduction in the grain boundary area. The electrical conductivity study also revealed that water incorporated in the materials structure had an effect on the conductivity, especially for the BIO samples made by the glycine-nitrate method. In air, water in the structure provides additional (protonic) conductivity that contributes to the total conductivity, while upon its release around 300oC, the conductivity drops. In hydrogen, the presence of water in the structure lowers the total conductivity, probably by obstructing the proton transport, while its release around 300oC causes the conductivity to increase, possibly by enabling direct incorporation of hydrogen. BIC, BICL025 and BIL samples did not show a significant effect with water loss at 300oC, but they showed a decrease in conductivity at around 400oC due to a phase transformation from the brownmillerite structure to a cubic structure. Based on the results of this study, undoped Ba2In2O5 produced by the glycine-nitrate process was selected as the material with the highest proton conductivity (between 0.02 S/cm and 0.7 S/cm) at temperatures between 300oC and 500oC. This material was selected for further proton conductivity studies and application in an intermediate temperature fuel cell. 9.3 Proton conductivity and stability of Ba2In2O5 in hydrogen-containing atmospheres Ba2In2O5 can exhibit mixed oxygen ion, proton and electron conductivity at different temperatures and in different atmospheres. In this work, the contribution of the proton  178 conductivity to the total conductivity was determined in two different ways. The first way consisted of the comparison of the total conductivities of a sample measured in dry air, dry nitrogen and a hydrogen-containing atmosphere. Results showed that conductivity in 50%H2/50%N2 was almost five orders of magnitude higher than the conductivity measured in air or nitrogen, suggesting predominantly protonic conductivity in a hydrogen-containing atmosphere. The other way, the electromotive force method (e.m.f.), was applied using a concentration cell: (80%H2/15%N2/5%H2O), Pt║Ba2In2O5║Pt, (48%H2/49%N2/3%H2O), to determine the proton transport number tH+. The resulting values of the proton transport numbers were tH+=1 at 100oC and 200oC, 0.84 at 300oC and 0.74 at 350oC. Since the samples were not mechanically stable above 350oC in the humid atmospheres that were used in this test, tH+ could not be determined at these temperatures. Corresponding proton conductivities were 6.2·10-6 S/cm at 100oC, 8.6·10-4 S/cm at 200oC, 0.015 S/cm at 300oC and 0.09 S/cm at 350oC. The last two values fit in the “gap” for the available proton conductive materials, shown in Fig. 2.5 and 2.6. A stability study of Ba2In2O5 in hydrogen-containing atmospheres revealed chemical and electrical stability of this material in the temperature range from 300oC to 480oC over 24 hours, as confirmed by X-ray diffraction and Raman spectroscopy. Conductivity was also constant over 24 hours in this temperature range. However, the study showed that Ba2In2O5 decomposes at 500oC to BaCO3 and elemental indium, confirmed by X-ray diffraction and Raman spectroscopy, which caused a gradual decrease in conductivity. High electrical conductivity, relatively high proton transport numbers and sufficient stability of Ba2In2O5 in hydrogen-containing atmospheres in the temperature range of 300oC  179 to 480oC made this material a suitable potential candidate for application as a proton conducting electrolyte or within the electrode of an intermediate temperature fuel cell. 9.4 Evaluation of Ba2In2O5 for use as an electrolyte material or within the anode in an intermediate temperature fuel cell In the previous studies Ba2In2O5 produced by the glycine-nitrate process was selected to be investigated as an electrolyte and for use within the anode structure in an intermediate temperature fuel cell. The material was used as an electrolyte in a cell: air, Pt║Ba2In2O5║Pt, 50%H2/50%N2 and its OCV and polarisation curve recorded in the temperature range of 300oC to 480oC. This is the temperature range where Ba2In2O5 shows the highest conductivity in hydrogen-containing atmospheres. Measured OCVs of the cell were 0.81 V at 300oC, 0.69V at 350oC, 0.59 V at 400oC, 0.51 V at 450oC, and 0.50 V at 480oC. OCVs at 100oC and 200oC were not stable and consistent. The measured values deviate from the theoretical values of 1.12 V at 300oC, 1.11 V at 350oC, 1.10 V at 400oC, 1.08 V at 450oC, and 1.07 V at 480oC due to the contribution of the electronic conductivity. Although OCV values were acceptable, performance of the cell was very low at all temperatures. As confirmed by the ac impedance measurements of the cell at each temperature and the potentiostatic measurements, this low performance is due to the high resistance of the electrolyte, caused by oxygen sorption into the Ba2In2O5 electrolyte on the cathode side and the blocking of the proton conductivity. As a result of this problem, Ba2In2O5 cannot be used as an electrolyte. The performance could be improved if a thin protective layer of a proton conductive material is used to prevent the process of oxygen sorption.  180 An investigation of the use of Ba2In2O5 within the anode structure as a proton- conductive support, in combination with a metal catalyst M=Pt, Ni or  Fe, was performed by ac impedance measurements in 50%H2/50%N2 for the symmetrical cells BIO+M II BIO II BIO+M. The activity of the anodes was determined to be in the following order: Pt-heated to 800oC > Pt+BIO-heated to 800oC > Pt+BIO-sintered to1300oC > Pt-sintered to 1300oC > Ni+BIO-heated to 800oC > Fe+BIO-heated to 800oC. The results revealed that all three metal catalysts are applicable for use in the anode in the 300-480oC temperature range, with Pt being the most active.  Application of Ba2In2O5 improved the activity of the anode, but only in the case when the anode is sintered to provide a good path for proton transport. 9.5 Research significance Development of fuel cells that operate in the intermediate temperature range of 200- 500oC has significant potential benefits with respect to fuel cell performance, materials, thermal management (particularly for automotive applications) and the coupling of fuel processing with the fuel cell. Electrolyte materials that have a satisfactory conductivity and stability in this temperature range are still not available. Although proton conductive ceramics show some promise for application as proton conductive electrolytes, none of the reported materials has a proton conductivity high enough for practical operation in this temperature range. The significance of this thesis project is that it focused on development of the electrolyte and electrode materials in this particular temperature range of interest. This area of research is still at an early stage, and the findings from this research are a contribution to the advancement of this research area. Some of the compositions in this work, such as Ba2In1.5Ce0.25La0.25O5.125 (BICL025), Ba2In1Ce0.5La0.5O5.25 (BICL05) and Ba2In1.5La0.5O5  181 (BIL) have been synthesized and characterized for the first time. Even though they did not prove to be good proton conductors, the comparison of their characteristics with Ba2In2O5 has helped to provide an understanding of some of the characteristics and behaviour of the more highly conductive Ba2In2O5 material. Analysis of the material and electrochemical properties of the investigated materials has helped shed light on the nature of these materials and the possible mechanisms of conductivity. The results have also shown the great influence that microstructure and the method of preparation have on the properties of the materials. Of particular significance for this thesis work is that the Ba2In2O5 ceramic was investigated in hydrogen-containing atmospheres for the first time, and although this material is a well-known and researched composition, its high conductivity in hydrogen has never been reported before. The high total conductivity of this material with a high proton conductivity in the temperature range of 300oC to 480oC makes it of interest for development of intermediate temperature fuel cell electrolytes. Although Ba2In2O5 did not show good performance when applied as an electrolyte in a fuel cell, an explanation of the problem has been provided, and further research could be done to overcome this problem. In this thesis, Ba2In2O5 was also investigated for the first time as a proton-conducting support for metal catalysts in an anode of an intermediate temperature fuel cell and proved beneficial. The applied operating temperature range of 300oC to 480oC enabled use of non- noble catalysts, such as Ni and Fe. Further research in this area is needed, as well, to improve the performance of the anode. 9.6 Future work and recommendations Out of five investigated materials, Ba2In2O5 ceramic showed promise for application as a proton-conductive electrolyte and as a catalyst support in an intermediate temperature fuel  182 cell. There are several areas recommended for further research in order to improve its performance and applicability in this area. • Although Ba2In2O5 was stable over at least 24 h (and in some tests over 3 days) in hydrogen-containing atmospheres at temperatures up to and including 480oC, its long range stability in these atmospheres is questionable. Additional research and testing is recommended to improve its stability, possibly by doping with an element that would improve its stability, but not reduce its conductivity. The same applies to the material stability in humid atmospheres, which has to be improved if this material is to be used in a fuel cell. • The low performance of a cell with Ba2In2O5 as an electrolyte is due to the sorption of oxygen on the cathode side and blocking of the proton conductivity. Future work could examine application of a thin protective layer to prevent this process from happening. The layer should be produced from a proton conductive material that is stable under the applied conditions, and chemically and thermally compatible with Ba2In2O5. Since this material would be applied as a very thin layer, its proton conductivity could be somewhat lower than that of Ba2In2O5 and still not cause high ohmic resistance. • Doping of Ba2In2O5 could be investigated to reduce the oxygen sorption, but still retain high proton conductivity. • Another approach that could be investigated to reduce oxygen sorption is the controlled surface treatment of Ba2In2O5 to 1500oC in order to turn it into a cubic phase, while still keeping the highly conductive brownmillerite phase in the inner part of the electrolyte. The cubic phase, having a higher oxygen stoichiometry, might not be as susceptible to oxygen  183 sorption as the brownmillerite phase. The conductivity of the cubic phase in hydrogen- containing atmospheres should be examined. • Further work in needed to improve the preparation of the thin, gas-tight Ba2In2O5 electrolyte material. Possible techniques could be spray pyrolysis or reactive spray deposition technique. • In the anode study Ba2In2O5 proved to be useful in improving the activity of the anode. This is true in the case when Ba2In2O5 is sintered to provide a conductive path for protons. Improved methods of preparation for anode structure are needed to enable sintering of Ba2In2O5 separately and subsequent application of the metal catalyst. This would help avoid formation of undesired phases when the two are sintered together, as was found in our study. • Due to indium metal scarcity, high price (~800 $ US/kg) and wide application in the semiconductor industry, it is recommended that either other dopants are used to replace indium in Ba2In2O5, or other materials with potential for proton conductivity in the intermediate temperature range are investigated (e.g., other perovskite-related oxides, P2O5– ZrO2–SiO2 –TiO2 based glasses, etc.) These are some of the recommendations for future research related to the particular materials investigated in this work. However, continuous research in the area of intermediate temperature fuel cells and particularly applicable materials is crucial for further fuel cell development and commercialisation.      184 References 1. J. Larminie, A. Dicks, Fuel Cell Systems Explained, 2nd Edition, John Wiley & Sons Inc., England, (2003). 2. N. Sammes, Fuel Cell Technology-Reaching Toward Commercialization, Springer, (2006). 3. W. Grove, Philosophical Magazine and Journal of Science, XIV (1839) 127. 4. W. Grove, Philosophical Magazine and Journal of Science, XXI (1842), 417. 5. J. Leo, M.J. Blomen, M.N. Mugerwa, Fuel Cell Systems, Plenum Press, New York, USA, (1993). 6. Web site http://www.futureenergies.com , accessed Feb. 2011. 7. High Temperature SOFC Cells, State of the art and prospects, http://www.cea.fr, accessed Sept. 2005. 8. K. Kinoshita, Electrochemical Oxygen Technology, John Wiley& Sons, Inc., (1992), pg 20. 9. F. Walsh, A First Course in Electrochemical Engineering, The Electrochemical Consultancy, (1993), pg. 34. 10. C.M.A. Brett, A.M.O. Brett, Electrochemistry, Principles, Methods, and Applications, Oxford University Press, (2002), pg 15. 11. R.H. Perry, D.W. Green, Perry’s Chemical Engineers’ Handbook (7th Edition), McGraw-Hill, (1997). 12. J.A. Dean, Lange’s Handbook of Chemistry (15th Edition), McGraw-Hill, (1999). 13. W.M. Haynes (editor), CRC Handbook of Chemistry and Physics, 91st Edition, (2010).  185 14. G. Hoogers, Fuel Cell Technology Handbook, CRC Press, (2003), pg. 6-3. 15. E. Gyenge, ChBE 477: Fuel Cells & Electrochemical Systems Course Notes, (2004). 16. G. Prentice, Electrochemical Engineering Principles, Prentice Hall (1991). 17. C. Xia, M. Liu, Advanced Materials, 14 (2002) No. 7. 18. Z. Liu, J.S.Wainright, R.F. Savinell, Chemical Engineering Science, 59 (2004) 4833. 19. M.K. Debe, S.J. Hamrock, R.T. Atanasoski, IV.A.3 Advanced MEAs for Enhanced Operating Conditions, DOE Hydrogen Program FY 2004 Progress Report. 20. M.K. Debe, S.J. Hamrock, R.T. Atanasoski,  Development of Thin Film Membrane Assemblies with Novel Nanostructured Electrocatalyst for Next Generation Fuel Cell (2004). 21. R.T. Atanasoski, IV.C.5 Novel Approach to Non-Precious Metal Catalysts, DOE Hydrogen Program FY 2004 Progress. 22. L.J. Bonvillec, H.R. Kunzc, J.M. Fentona, Journal of Power Sources, 141 (2005) 250. 23. P. Jannasch, Current Opinion in Colloid and Interface Science, 8 (2003) 96. 24. G. Alberti, M. Casciola, Solid State Ionics, 145 (2001) 3. 25. B. Wang, Recent development of non-platinum catalysts for oxygen reduction reaction, Journal of Power Resources, 152 (2005) 1. 26. B.R. Limogesa, R.J. Stanisa, J.A. Turnerb, A.M. Herringa, Electrochim. Acta, 50 (2005), 1169.  186 27. S.M. Haile, P.N. Pintauro, J. Mater. Chem., 20 (2010) 6211. 28. S.R. Samms, S. Wasmus, and R.F, Savinell, Journal of Electrochemical Society, 143, 4 (1996) 1363. 29. D. Cheddie, N. Munroe, Journal of Power Sources, 156 (2) (2006) 414. 30. R. Bouchet, E. Siebert, Solid State Ionics, 118 (1999) 287. 31. K. Sundmacher, L.K. Rihko-Struckmann , V. Galvita, Catalysis Today, 104 (2005) 185. 32. L. Siwen, L. Meilin, Electrochimica Acta, 48 (2003) 4271. 33. G.B. Zhang, D.M. Smyth, Solid State Ionics, 82 (1995) 153. 34. C.A.J. Fisher, M.S. Islam, Solid State Ionics, 118 (1999) 355. 35. N. Bonanos, Solid State lonics, 53-56 (1992) 967. 36. J. Niwa, T. Suehiro K. Kishi, S. Ikeda, M. Maeda, J. Mat. Sci., 38 (2003) 3791. 37. T. Mitamura, H. Ogino, H. Kobayashi, J. Am. Ceram. Soc., 76 (8) (1993) 2127. 38. G. B. Zhang, D.M. Smyth, Solid State Ionics, 82 (1995) 161. 39. P. Berastegui, S. Hull, F.J. Garcia-Garcia, S.G. Eriksson, Journal of Solid State Chemistry, 164 (2002) 119. 40. R. Hui, R. Maric, C. Deces-Petit, E. Styles, W. Qua, X. Zhang, J. Roller, S. Yick, D. Ghosh, K. Sakata, M. Kenji, Journal of Power Sources, 161 (2006) 40. 41. H. Iwahara, T. Esaka, H. Uchida, Solid State Ionics, 3/4 (1981) 359. 42. S. Shin, H. H. Huang, M. Ishigame, Solid State Ionics, 40/41 (1990) 910. 43. T. Yajima, H. Iwahara, H. Uchida, Solid State Ionics, 47 (1991) 117.  187 44. J. Niwa, T. Suehiro K. Kishi, S. Ikeda, M. Maeda, Journal of Materials Science, 38 (2003) 3791. 45. T. Norby, Solid State Ionics, 125 (1999) 1. 46. K.D. Kreuer, Chem. Mater., 8 (1996) 610. 47. R. Bouchet, E. Siebert, Solid State Ionics, 118 (1999) 287. 48. S. Stotz, C. Wagner, Ber. Bunsenges.Phys. Chem., 70 (1966) 781. 49. T. Takahashi, H. Iwahara, Rev. Chim. Miner. 17 (1980) 243. 50. H. Iwahara, H. Uchida, S. Tanaka, Solid State Ionics, 9–10 (1983) 1021. 51. H. Iwahara, H. Uchida, K. Ono, K. Ogaki, J. Electrochem. Soc., 135 (1988) 529. 52. H. Iwahara, Solid State Ionics, 28–30 (1988) 573. 53. H. Iwahara, H. Uchida, K. Morimoto, J. Electrochem. Soc., 137 (1990) 462. 54. N. Bonanos, B. Ellis, M.N. Mahmood, Solid State Ionics, 44 (1991) 305. 55. N. Tanigushi, K. Hatoh, J. Niikura, T. Gamo, Solid State Ionics, 53–56 (1992) 998. 56. H. Iwahara, T. Yashima, T. Hibino, H. Ushida, J. Electrochem. Soc., 140 (1993) 1687. 57. N. Bonanos, K.S. Knight, B. Ellis, Solid State Ionics, 79 (1995) 706. 58. C.W. Tanner, A.V. Virkar, J. Electrochem.Soc., 143 (1996) 1386. 59. K.D. Kreuer, Annu. Rev. Mater. Res., 33 (2003) 333. 60. G. Alberti, M. Casciolar, Solid State Ionics, 145 (2001) 3. 61. T. Norby, M. Wideroe, R. Glockner, Y. Larring, Dalton Trans., (2004) 3012. 62. F. Lefebvre-Joud, G. Gauthier, J. Mougin, J. Appl. Electrochem., 39 (2009) 535.  188 63. S.S. Bhella, V. Thangadurai, DOI: 10.1016 j.jpowsour.2008.09.110. Journal of Power Sources (2008). 64. L.E. Smart, E. A. Moore, Solid State Chemistry: An Introduction, 3rd Edition, CRC Press (2005). 65. S. Hui, Evaluation of Yttrium-Doped SrTiO3 as a Solid Oxide Fuel Cell Anode, Ph.D. Thesis, McMaster University (2005). 66. W. Hayes, A.M. Stoneham, Defects and Defect Processes in Nonmetallic Solids, Dover Publications (2004). 67. M. Karlsson, A. Matic, C.S. Knee, I. Ahmed, S.G. Eriksson, L. Börjesson, Chem. Mater., 20 (2008) 3480. 68. R.J.D. Tilley, Principles and Applications of Chemical Defects, Stanley Thornes Publishers Ltd, (1998) p. 103. 69. J. Wu, Defect Chemistry and Proton Conductivity in Ba-based Perovskites, Ph.D. Thesis, California Institute of Technology (2005). 70. J. B. Goodenough, J. E. Ruiz-Diaz, Y. S. Zhen, Solid State Ionics, 44 (1990) 2l. 71. W. Fischer, G. Reck, T. Schober, Solid State Ionics, 116 (1999) 211. 72. T. Schober, J. Friedrich, F. Krug, Solid State Ionics, 99 (1997) 9. 73. T. Schober, J. Friedrich, Solid State Ionics, 113–115 (1998) 369. 74. T. Hashimoto, Y. Inagak, A. Kishi, M. Dokiya, Solid State Ionics, 128 (2000) 227. 75. P. Berastegui, S. Hull, F. J. Garcia-Garcia, S.-G. Eriksson, Journal of Solid State Chemistry, 164 (2002) 119.  189 76. T. Ishihara, Perovskite Oxide for Solid Oxide Fuel Cells, Springer (2009). 77. P. Kofstad, Nonstiochiometry, Diffusion and Electrical Conductivity in Binary Metal Oxides, Robert E. Krieger Publishing, Florida (1983). 78. A.J. Moulson, A.J. and J.M. Herbert, Electroceramics, Chapman and Hall, London (1990). 79. Q. Yin and Y.S. Lin, Solid State Ionics, 178 (2007) 83. 80. T. Norby, KJM5120 and KJM9120 Defects and Reactions Lecture Notes, www.uio.no , accessed Jan. 2011. 81. R. Glockner, M.S. Islam, T. Norby, Solid State Ionics, 122 (1-4) (1999) 145. 82. R.A. Davies, M.S. Islam, J.D. Gale, Solid State Ionics, 126 (1999) 323. 83. M.S. Islam, M. Cherry, Solid State Ionics, 97 (1-4) (1997) 33. 84. R.A. Davies, M.S. Islam, A.V. Chadwick, G.E. Rush, Solid State Ionics, 130 (2000) 115. 85. S. Steinsvik, Y. Larring, T. Norby, Solid State Ionics, 143 (2001) 103. 86. T. Norby, Solid State Ionic, 28-30 (1988) 1586. 87. D.P. Sutija, T.Norby, P.Bjiirnbom, Solid State Ionics, 77 (1995) 167. 88. H.H. Uchida, N. Maeda, H. Iwahara, Journal of Applied Electrochemistry, 12 (1982) 645. 89. K. Kakinuma, H. Yamamura, H. Haneda and T. Atake, Solid State Ionics, 140 (2001) 301. 90. C.A.J. Fisher, M.S. Islam, R.J. Brook, Journal of Solid State Chemistry, 128 (1997) 137. 91. A. Rolle, G. Fafilek, R.N. Vannie, Solid State Ionics, 179 (2008) 113.  190 92. L.A. Chick, L.R. Pederson, G.D. Maupin, J.L. Bates, L.E. Thomas and G.J. Exarhos, Mater. Lett., 10 (1990) 6. 93. B.D. Cullity and S.R. Stock, Elements of X-Ray Diffraction, 3rd Edition, Prentice-Hall Inc., (2001). p. 167-171. 94. T. Hashimoto, Y. Ueda, M. Yoshinaga, K. Komazaki, K. Asaoka and S. Wang, J. Electrochem. Soc., 149 (10) (2002) (10) A1381. 95. Q. Yin and Y.S. Lin, Adsorption, 12 (2006) 329. 96. S. McIntosh, J.F. Vente, W.G. Haije, Solid State Ionics, 177 (2006) 1737. 97. Z.H. Yang and Y.S. Lin, Solid State Ionics, 176 (2005) 89. 98. A. Mai, F. Tietz and D. Stföer, Solid State Ionics, 173 (2004) 35. 99.  R. Maric, C. Decès-Petit, R. Hui, X. Zhang, D. Ghosh, K. Sakata, J. Electrochem. Soc., 153 (8) (2006) A1505. 100. J. Jankovic, D. P. Wilkinson, and R. Hui, J. of Electrochem. Soc., 158 (1) (2011) B61. 101. J.E. Bauerle, J. Phys. Chem. Solids, 30 (1969) 2657. 102. W. Lai and S.M. Haile, J. Am. Ceram. Soc., 88 (11) (2005) 2979. 103. J. R. Macdonald, W. R. Kenan, Impedance Spectroscopy Emphasizing Solid Materials and Systems, John Wiley & Sons Inc. (1987). 104. D. Vladikova, J.A. Kilner, S.J. Skinner, G. Raikova , Z. Stoynov, Electrochimica Acta, 51 (2006) 1611. 105. M. Wang, L. Qui, Y. Sun, Journal od Rare Earths, 27 (2009) 819. 106. L. Qui, G. Ma, Chinese Journal od Chemistry, 24 (2006) 1564.  191 107. H.J. Park, C. Kwak, K.H. Lee, S.M. Lee, E.S. Lee, Journal of the European Ceramic Society, 29 (2009) 2429. 108. J.F.Q. Rey, F.F. Ferreira, E.N.S. Muccillo, Solid State Ionics, 179 (2008) 1029. 109. P. Chinho, J. Woo-Sik, Z. Huang, A. Timothy, J. Journal of Material Chemistry, 12 (2002) 356. 110. S. Tao and J.T.S. Irvine, Adv. Mater. , 18 (2006) 1581. 111. S.C. Singhal, K. Kendall, High-temperature Solid Oxide Fuel Cells: Fundamentals, Design and Applications, Elsevier Science (2004). 112. T.A. Kalinina, L.N. Lykova, L.M. Kovba, M.G. Melinikova, and N.V. Porotnikov, Zh. Neorg. Khim., 28 (2) (1983) 466; Russ. J. Inorg. Chem. (Engl. Transl.), 28 (2) (1983) 259. 113. H. Yamamura, Y. Yamada, T. Mori, T. Atake, Solid State Ionics, 108 (1998) 377. 114. S.A. Speakman, J.W. Richardson, B.J. Mitchell, S.T. Misture, Solid State Ionics 149 (2002) 247. 115. E. Barsoukov, J.R. Macdonald, Impedance Spectroscopy - Theory, Experiment, and Applications, 2nd Edition, John Wiley & Sons, Inc. (2005). 116. E.V. Ramana , S.V. Suryanarayana, T.B. Sankaram, Mater. Res. Bull., 41 (2006) 1077. 117. S. Dutta, R.N.P. Choudhary and P.K. Sinha,  J. Appl. Phys., 36 (2006)141. 118. K. Huang, R.S. Tichy and J.B. Goodenough, J. Am. Ceram. Soc., 81 (10) (1998) 2576. 119. D.K. Pradhan, B.K. Samantaray, R.N.P. Choudhary, A.K. Thakur, J. Mat. Sci., 40 (2005) 5419. 120. T.S. Zhang,  J.Ma, Y.Z. Chen, L.H. Luo, L.B. Kong, S.H. Chan, Solid State Ionics, 177 (2006) 1227.  192 121. M.J. Verkerk, B.J. Middelhuis and A.J. Burggraaf, Solid State Ionics, 6 (1982) 159. 122. N.M. Beekmans and L. Heyne, Electrochimica Acta, 21 (1976) 303. 123. C.S. Kim, J.H. Park, B.K. Moon, H.J. Seo, B.C. Choi, Y.H. Hwang, H.K. Kim, J.N. Kim, Ferroelectrics, 326 (2205) 109. 124. A. Hooper, J. Phys. D: Appl. Phys., 10 (1977) 1487. 125. P.G. Bruce, A.R. West, and D.P. Almond, Solid State Ionics, 7 (1982) 57.                 193 APPENDIX A – Ba2In2O5 Properties Some basic properties of Ba2In2O5 reported in the literature are listed here, along with the source of the information. Crystal structure The space lattice of Ba2In2O5 is body-centered orthorhombic from room temperature to 900oC. The structure at higher temperatures is discussed in the following sections. It has a brownmillerite-type structure, pictured in Fig. A.1. This structure is derived from a cubic perovskite structure, when a lower valent metal ion substitutes B ion in the perovskite oxide, forming oxygen vacancies that orient in the 101 direction. This process is discussed in Section 3.3. As calculated by Fisher et al. [34], the unit cell parameters are a=6.08 Å (6.09 Å experimental value), b=16.40 Å (16.79 Å experimental value), and c=5.94 Å (5.88 Å experimental value).             Figure A.1 – Brownmillerite structure derived from a perovskite structure. Schematic on the right shows the oxygen vacancies (squares) ordered in 101 direction [39, 90 with permission from Journal of Solid State Chemistry]. (a) ABO3 Perovskite (b) A2B2O5 Brownmillerite A=Ba O Oxygen vacancy B=In  194  XRD pattern The indexed X-ray diffraction pattern of Ba2In2O5 is shown in Fig. A.2          Figure A.2 – XRD pattern of Ba2In2O5 (Adapted from Hashimoto et al. [74] with permission from the Solid State Ionics).   195 Phase diagram  Figure A.3 – Phase diagram of the BaO-InO3 system (obtained from Phase Diagram Viewer, reference Kalinina et al. [112]).     196 H2O and CO2 uptake Hashimoto et al. [74] confirmed the release of H2O and CO2 from the structure of Ba2In2O5 using thermogravimetry–mass spectroscopy technique. Weight loss at around 100oC was attributed to the water absorbed on the surface of the sample, while weight loss around 300oC was attributed to the loss of water from the crystal structure. Another weight loss around 700oC was due to the loss of CO2 from the crystal structure. This is shown in Fig. A.4. This study reported that a sample prepared in air contained 0.34 mol of H2O to 1 mol Ba2In2O5 and had an orthorhombic brownmillerite structure. The structure does not change as the water is released. When the sample was annealed at 200oC in humid Ar (PH2O=30 hPa), the content of water increased to 0.98 mol of H2O to 1 mol Ba2In2O5, and the structure changed to a tetragonal structure, shown in Fig. A.5. This was also confirmed by Shober et al. [6] and the process was found to be reversible.         Figure A.4 – TG-MS measurement for as prepared Ba2In2O5 showing loss of H2O and CO2 with heating, reported by Hashimoto et al. [74 with permission from the Solid State Ionics].   197          Figure A.5 – X-ray diffraction pattern of the orthorhombic brownmillerite (dry) Ba2In2O5 and tetragonal brownmillerite (humid) Ba2In2O5 [72 with permission from the Solid State Ionics].  Order disorder transition temperature Td Goodenough et al. [70] have found that Ba2In2O5 goes through a sharp increase in electrical conductivity around 930oC and assigned it to a first order transition (order-disorder transition) of the orthorhombic brownmillerite Ba2In2O5 structure to a cubic perovskite-type structure. Yamamura et al. [113] noticed an anomaly in expansion and shrinking of the Ba2In2O5 sample when they performed dilatometric experiments, as shown in Fig. A.6. They concluded that this anomaly could be related to the volume change due to the order-disorder phase transition.  The difference in the heating and the cooling process was explained by the thermal hysteresis.   198           Figure A.6 – Dilatometric measurement for Ba2In2O5 showing anomaly in expansion and shrinking due to the order-disorder phase transformation [113 with permission from the Solid State Ionics].  Speakman et al. [114] suggested that this change from orthorhombic brownmillerite structure to cubic structure is gradual, including a slow change from orthorhombic to tetragonal structure from 900oC to 925oC, followed by a gradual change from tetragonal to cubic structure between 925oC and 1040oC. The in-situ X-ray diffraction data given in their work are shown in Fig. A.7.         199                     Figure A.7 – In-situ X-ray diffraction data showing gradual transformation of Ba2In2O5 from brownmillerite orthorhombic to tetragonal to cubic structure [114 with permission from the Solid State Ionics].  orthorhombic tetragonal tetragonal cubic  200 Electrical conductivity Ba2In2O5, like other oxygen deficient perovskite-related ceramics, can exhibit mixed oxygen ion, proton and electron conductivity, as discussed in Chapter 3. Goodenough et al. [70] reported oxygen ion conductivity in Ba2In2O5 measured under an oxygen partial pressure of PO2=10e-6 atm in the temperature range 720 < T< 1375 K (Fig. A.8 a). They noticed a sharp increase in the conductivity due to the disordering of the oxygen vacancies and transition to a cubic structure, as discussed in the previous section. They also reported the appearance of p-type electronic conductivity as the oxygen partial pressure increases, as shown in Fig. A.8 b.             Figure A.8 – Arrhenius plot of conductivity for Ba2In2O5 (a) under PO2=10e-6 atm and (b) with PO2 change reported by Goodenough et al. [70 with permission from the Solid State Ionics]. (a) (b)  201 Similar behavior was reported by Zhang et al.  [38] at temperatures below the order- disorder transition temperature Td, but no effect of oxygen partial pressure was noticed above Td (Fig A.9 a). In the presence of CO2 (oxygen partial pressure was controlled by flow of CO/CO2 mixture) Ba2In2O5 decomposed at low PO2, as shown in the figure. Oxygen ion transport number determined by the EMF method is shown in Fig. A. 9 b.         Figure A.9 – (a) Logarithm of total electrical conductivity of Ba2In2O5 over a range of PO2 and at different temperatures; (b) ionic transport number tO determined by e.m.f. measurement, from work by Zhang et al. [38 with permission from the Solid State Ionics].  Zhang et al. concluded that total conductivity at temperatures above 500oC is a combination of oxygen ion, hole and electronic conductivity (and protonic at lower temperatures): σt=σi+σh+σe  (A.1) Their experimental and simulation results on the relationship between different types of conductivities are shown in Fig. A.10.  (a) (b) tO  202                      Figure A.10 – Experimental data (points) and simulation results (lines) reported by Zhang et al. [38] for Ba2In2O5 defining different types of conductivities (ions, holes or electrons) under different PO2 and temperatures (with permission from the Solid State Ionics).   203 In another work Zhang et al. [33] reported proton conductivity of Ba2In2O5 measured in wet air at temperatures below 500oC. The proton conductivity is history dependent and Fig. A.11 shows proton conductivity determined under three different conditions (for the details, please refer to the published work by Zhang et al.). Proton conductivity was determined as the difference between the total conductivity measured in wet air (PH2O≈0.03 atm) and the total conductivity measured in dry air (PH2O≤3ppm).              Figure A.11 – Proton conductivity vs. temperature determined for Ba2In2O5 by Zhang et al. [33 with permission from the Solid State Ionics].     204 APPENDIX B – Background on Selected Measuring Techniques B.1 Ac impedance spectroscopy Ac impedance spectroscopy or electrochemical impedance spectroscopy (EIS) is a non- destructive method used to study the correlation between microstructure and electrical properties of solids in a number of research areas such as corrosion, electrode kinetics, membranes, conducting polymers, semiconductors, batteries and fuel cells, etc. [101-104, 115-118]. Experimentally, the method involves the application of a small sinusoidal potential Eac or current Iac perturbation (ac signal) with different frequencies to an equilibrium system (tested material) and measurement of the corresponding current or voltage response. The response is a result of electrical processes occurring in a material upon application of the ac signal. Since the technique is called spectroscopy, parameters are measured as a function of the frequency (ω) of the applied perturbation, which is typically in a range of several millihertz to several megahertz. From the applied perturbation and its measured response the magnitude of the impedance (Z) and phase shift (θ) is determined.  Concept of complex impedance If a dc signal is applied (a special case of an ac signal where the frequency is zero) to a circuit element, its ability to resist the flow of the dc electrical current is given by the electrical resistance (R), and is defined by Ohm's law: Edc = Idc R  (B.1)  205 where Edc is the dc potential (V), Idc is the dc current (A), and R is resistance (Ω). The use of the above relationship is limited to only an ideal resistor, which is the only circuit element that impedes the flow of electrons in a dc circuit. When an ac signal is applied, where the frequency is non-zero, the equivalent equation is: Eac = Iac Z    (B.2)  In this case Eac and Iac are defined in an ac circuit as the potential (V) and current (A), respectively, and are sinusoidal signals with a non-zero frequency. In the following text they will be simply labeled as E and I. Z (Ω) is defined as the impedance, the ac equivalent of the resistance. In the case of an ac circuit, not only resistors but also capacitors and inductors oppose the flow of charge carriers in the circuit. The total impedance in a circuit is the combined opposition of all its resistors, capacitors, and inductors to the flow of charge carriers (electrons, protons, oxygen ions, etc). If the sinusoidal perturbation voltage is applied to a system, it can be expressed as a function of time in the following form: )sin()( 0 tEtE ω=  (B.3) Where E(t) is the potential at time t, E0 (V) is the amplitude of the voltage signal, and ω (rad/sec) is the radial frequency. The relationship between radial frequency ω, expressed in radians/second, and frequency f, expressed in hertz is: fpiω 2=  (B.4)  206 In a linear system, the response signal, I(t), is shifted in phase and has an amplitude, I0(A): )sin()( 0 φω += tItI  (B.5) where φ (radians) is the phase difference between the voltage and the current. It is zero for purely resistive behaviour. The sinusoidal ac voltage signal E and resulting ac current I with a phase shift and amplitudes are presented in Fig. B.1.       Figure B.1 – Presentation of an ac sinusoidal potential signal (E) with a resulting current response (I). The impedance of the system then can be expressed by the equation: )sin( )sin( )sin( )sin( )( )( 0 0 0 φω ω φω ω + = + == t tZ tI tE tI tEZ  (B.6)  The relation between the electrochemical system properties and response to the voltage or current excitation is generally very complex and requires the solution of a system of E Phase shift I E T  207 differential equations. In order to simplify the mathematical solution of such systems, Euler’s relationship can be used: φφφ sincos)exp( jj +=  (B.7) to express the impedance as a complex function. In this equation j is the imaginary number and is equal to 1− . Using this relationship the ac potential signal and current response can be described as: )exp()( 0 tjEtE ω=  (B.8) )exp()( 0 φω jtjItI −=  (B.9) The impedance is then represented as a complex number: )sin(cos)exp()( )( 00 φφφ jZjZtI tEZ +===  (B.10) The expression for Z is composed of a real and an imaginary part, and can be presented in a complex impedance notation: "' jZZZ +=  (B.11) where: Z’=Re(Z)= IZIcosφ      is the real part coordinate and (B.12) Z"= Im(Z)= IZIsinφ    is the imaginery part coordinate                                    (B.13)   208 The impedance analysis of a material gives us data having both resistive (real part) and reactive (imaginary part) components. If the real part is plotted on the x-axis and the imaginary part on the y-axis of a chart, a complex plane plot, called a "Nyquist plot" can be formed, as shown in Fig. B.2. Each point on the Nyquist plot is the impedance at one frequency. However, the frequencies cannot be plotted in this chart. This is the usual representation of the impedance. The value Z' is the projection of the impedance along the x- axis and Z'' along the y-axis. The cell impedance is then: 2/122 )"'( ZZZ +=  (B.14) and ' " tan Z Z =φ  (B.15) Another popular presentation of the impedance results is the Bode-plot, shown in Fig. B.3. The impedance is plotted with the frequency on the x-axis and both the absolute values of the impedance |Z| and the phase-shift on the y-axis. Unlike the Nyquist plot, the Bode plot does show frequency information.         209        Figure B.2 – Example of a Nyquist plot showing the impedance and phase angle.            Figure B.3 – Example of a Bode plot showing the relationship between log impedance, phase angle and applied frequency. Φ Z’ -Z” Z  210 The display of impedance data in the complex plane plot (Nyquist plot) appears in the form of a succession of semicircles attributed to relaxation phenomena with different time constants due to the contribution to the overall electrical properties by various components in the material. Each semicircle is characteristic of a single "time constant". Electrochemical impedance plots often contain several time constants, showing several semicircles which can be well distinguished, but also sometimes overlapped or often only a portion of one or more of their semicircles is seen. Ac impedance data is commonly analyzed by fitting it to an equivalent electrical circuit model consisting of a combination (in series and/or in parallel) of different circuit elements. Most of the circuit elements in the model are common electrical elements such as resistors, capacitors, and inductors. Some of these elements with their corresponding impedance expressions are shown in Table B.1. To be useful, the elements in the model should be based on the physical electrochemistry of the system. Different electrical circuit models that are often used to explain some ac impedance data and the corresponding Nyquist plots are shown in Fig. B.4. However, although fitting ac impedance data to an equivalent electrical circuit model can be useful for interpreting the ac impedance data, it can also be misleading as the same impedance data can be fitted with more than one circuit model. One has to be careful when using this method and the model should have a logical relation to the actual electrochemical system.       211 Table B.1 – Impedance expressions for some simple electrical circuit elements  Circuit element Graphical representation Impedance equation Resistor  jRZ 0+= 1−=j Capacitor  CjZ ω/0 −= fpiω 2= Inductor  LjZ ω+= 0 fpiω 2= Note: R is ohmic resistance given in Ohms (Ω), C is capacitance given in Farads (F) and L is inductance given in Henries (H)                212 Figure B.4 – Some simple elements of equivalent circuits and their corresponding Nyquist plots [65 with permission from the author].  213 Impedance response of a polycrystalline material Electrochemical properties of solid polycrystalline materials are affected by their microstructure – porosity, grain size and orientation, nature of grain boundaries [101-104, 116, 118, 121].  Grain boundaries often significantly affect the electrical properties of these materials due to the presence of a second phase in the grain boundaries, micro cracks, mismatch of the lattices, space charge, etc., or a combination of these effects. Ac impedance responses of such materials contain elements that are related to these effects and the material’s microstructure. Some microstructural models have been developed to describe the ac electrical properties of heterogeneous materials. Three models have been widely used for describing electrical properties of grains and grain boundaries: the “Bauerle model”, the “brick layer model’’ and the “parallel model” [101, 117, 118]. In this thesis we use the Bauerle model to model the behaviour of the materials investigated. Bauerle [101] was first to correlate microstructure and electrical properties of ceramics, specifically zirconia-yittria (ZrO2)0.9(Y2O)0.1. Bauerle found that if a second, ionically insulating, phase was segregated in the grain boundaries of a dense material, an additional arc in the ac impedance spectra appeared, which was not present in the high purity material.  He suggested that this second phase restricted the contact between the grains of the highly conducting phase and therefore introduced an additional source of impedance. Based on his theory, it is expected that the ac impedance data for these materials would appear in the complex plane plot in the form of a succession of semicircles attributed to relaxation phenomena with different time constants due to the contribution of the grain (bulk), grain boundary and electrode/electrolyte interface, as shown in Fig. B.5. The equivalent electrical  214 circuit model for such an electrochemical system can be represented by a series combination of parallel RC units, as shown in Fig. B.5, and described by the corresponding impedance expressions for each contribution, as shown in equations B.16, B.17 and B.18.                 Figure B.5 – (Top) Schematic of a polycrystalline material with grain bulk, grain boundary and an interface with the electrode; (Middle) Equivalent circuit diagram for a polycrystalline material: Rb, Rgb, Re represent the bulk, grain boundary and electrode resistances, respectivelly; Cb, Cgb, Cdl represent the bulk, grain boundary and electrode double layer capacitances, respectivelly; (Bottom) Complex plane plot showing the ideal ac impedance scan response for a polycrystalline ceramic. Rb Rgb Re Cb Cgb Ce Z' (Ω ) or (Ω cm) Z' ' ( ΩΩ ΩΩ ) o r ( ΩΩ ΩΩ c m ) Rb Rb+Rgb Rb+Rgb+Re Frequency, f (Hz) Bulk contribution Grain boundary contribution Electrode- electrolyte transfer contribution Bulk Grain boundary Electrode- electrolyte interface Polycrystalline material  215 ])(1[])(1[ 2 2 2 bb bb bb b b CR CRj CR R Z ω ω ω + − + =  (B.16)  ])(1[])(1[ 2 2 2 gbgb gbgb gbgb gb gb CR CRj CR R Z ω ω ω + − + =  (B.17)  ])(1[])(1[ 2 2 2 dle dle dle e e CR CRj CR R Z ω ω ω + − + =  (B.18)  In reality, the impedance spectrum of a polycrystalline material may be more complicated than the simple combination of RC elements and semicircles usually appear depressed. The impedance of such a non-ideal behavior is often represented by a Constant Phase Element (CPE) [101]. The CPE behavior is generally attributed to a distribution of a physical property of the system, such as distributed surface reactivity, surface inhomogeneity, roughness, porosity, current and potential distributions associated with electrode geometry, etc., depending on the system. The impedance expression of the CPE impedance is represented as shown in equation B.19, where λ and α are real constants, and ω is the angular frequency:                αω λ ω )()( jjZCPE =    (B.19) CPE elements can be placed instead of the capacitors in Fig. B.5, as shown in Fig. B.6 and can have several interpretations. For an ideal case, λ is equal to C and α = 1, and the element represents the ideal capacitor. For a case when α = 0.5 the CPE impedance is simplified into a Warburg impedance (ZCPE = ZW), representing semi-infinite diffusion. For finite-length pore diffusion, α= 0.25. The CPE represents a resistor for α= 0, a capacitor for  216 α= 1, an inductor for α=−1, R–C combinations for 0≤ α ≤ 1, R–L combinations for −1 ≤ α ≤ 0 and R–C–L combinations for −1≤  α ≤ 1. An analysis of the impedance data provides unique relaxation frequency describing the relaxation processes occurring within a polycrystalline sample. The peak of the semicircle in the complex plane plot enables us to evaluate the relaxation frequency (fmax) of the bulk material using the relation:  121 maxmaxmax =⇒== bbbb CRfCR piωω τ  (B.19)  bbCR f pi2 1 max =⇒        and max2 1 fpiτ =  (B.20)  where Rb, Cb and τ refer to bulk resistance, bulk capacitance and relaxation time, respectively. The relaxation frequency (ωmax) of the material, at a given temperature, is a unique intrinsic property of the material independent of the sample geometric factors. Bauerle’s model is applicable for polycrystalline ceramics containing a second phase segregated in the grain boundary area.  However, although finding similar behavior in calcia- stabilized zirconia, Beekman and Heyne [122] suggested representation of this behavior through a distribution of time constants using a constant-phase element (CPE elements) rather then the single-RC-time constant, as suggested by Bauerle. Several other researchers [104, 123] also concluded that polycrystalline samples (except for the very dense materials) always showed some anomalous frequency dispersion, which affects the bulk impedance behavior even before a second separate arc appears due to the grain boundary effect. Hence,  217 each of the circular arcs in the ac impedance spectra of a polycrystalline material can be described with the form shown in Figure B.6.    Figure B.6 – General equivalent circuit used for representing the contribution of bulk, grain boundary and electrode, using CPE elements [101].  Research has shown that grain boundaries make a contribution to the impedance even if a second phase is not present. Hooper [124] found that bulk conductivity had the same activation energy as the single-crystal materials they were investigating, while the grain boundary conductivity had greater activation energy. On the other side Bruce and West [125] found the same values for both activation energies for their material, which suggested that the same physical processes are involved in both bulk and grain boundary conduction. They suggested that the grain boundary contribution to the impedance is due to a constriction effect, i.e., a smaller area of contact between grains. It is also possible that in the grain boundary area transport properties are affected by structural defects or imperfections, which are expected to be present there in higher concentrations than in the bulk. Therefore, Bauerle model can be applied to any polycrystalline material, not only to materials with segregation of a second phase in the grain boundary. In general, using the Bauerle model the ac impedance data of a polycrystalline sample can be described and analyzed. Application of an equivalent circuits and the associated complex plot fitting to the experimental results enable estimation of the values for bulk, grain boundary and electrode resistances and capacitances, as well as better understanding of the electrical Rb Rgb Re CPEb CPEgb CPEe  218 processes in the polycrystalline material.  However, one has to be careful with the interpretation and correlate the conclusions to the possible physical processes in the material.  B.2 Proton transport number determination by the electromotive force measurement The electromotive force (e.m.f.) method for determination of the ionic transport number in oxides consists of measurement of the voltage difference (OCV) across a fully dense testing sample with two reversible electrodes exposed to two different partial pressures of oxygen [86-88]. The obtained transport number is considered to be the sum of the transport numbers for metal and oxygen ions (both referred to as native ions), which cannot be further separated by this method. Also, foreign ionic species may contribute to the total conductivity in oxides if there is a gradient in their activity. The e.m.f. method can be used to determine the transport number of these ions as well. A general case of the e.m.f. concentration cell is shown in Fig. B.7, where a sample with two electrodes is exposed to a different activity of species X on two sides, I and II.                    219                  Figure B.7 – Schematic of an e.m.f. concentration cell for ion transport number determination.   Element X is ionizing to Xz at electrode I, and the reverse reaction is occurring at electrode II according to equation B.21:  −+⇔ zeXX z  (B.21)  The charge z may be positive or negative, depending on whether the electrons are formed or consumed in the process. A gradient in the activity, Xa , will cause a voltage difference across the sample (e.m.f.). If we assume that two electrodes on both sides of the sample are equal and reversible and at the same temperature T, the voltage measured across the sample (when no external current is drawn) will be:  High PIX Low PIIX Tested material I II X Xz X  220 ∫ − = − II I XXIII adt ze kTE Z ln  (B.22)  where k=1.381 x 10-23 J/K is the Boltzmann’s constant, T is temperature (K), z is the charge, ZXt  is the transport number for the X z , and Xa  is the activity of the element X. If the transport number of ions Xz, ZXt , is greater than 0, reaction B.21 will take place at the electrodes, and current will flow within the sample. Equation B.22 is valid only if the electrodes are non-polarizable and do not impede reaction B.21. Porous Pt electrodes are generally good for this purpose. If a small activity gradient is used, the transport number can be assumed constant within the gradient, and equation B.22 can be integrated into the following form:                                                                        (B.23)  where ZXt  is the transport number of the X z  ions and represents a mean value between conditions I and II,  IXa  and II Xa are the activities of the species X in compartment I and II (usually partial pressures) and EII-I is the measured voltage when the sample is exposed to two different and known activities of X. The transport number can be determined as the ratio of the measured voltage to the theoretical electromotive force for the same conditions. If Faraday’s constant F is used instead of the Boltzmann’s constant, knowing that F eR N Rk A ⋅ == , where R=8.314 J/mol K is the ideal gas constant, F= e· NA= 96,485 C/mol is      − ⋅= − I X II X XIII a a ze kT tE Z ln  221 the Faraday’s constant, NA= 6.022·1023 mol-1 is Avogadro’s constant, and e= 1.602·10-19C is the elementary charge or the magnitude of the charge of an electron, equation B.23 can be written as:                                                                      (B.24)  For the case of oxygen, z =-4 and equation B.24 obtains the form:                                                                            (B.25)  where −2Ot  is the transport number for both oxygen ions and native metal ions, PO2 I  is the high partial pressure of oxygen and PO2II is a low partial pressure of oxygen. . In the case when protons are introduced (as foreign species) through the reaction B.21, equation B.24 becomes:                                                                        (B.26)  where +Ht  is the proton transport number , PH2 I  is the high partial pressure of hydrogen and PH2II is a low partial pressure of hydrogen. If water is present in the compartments, due to the equilibrium reaction H2O(g)↔H2+1/2O2, equation B.26 can be expressed also as:    (B.27)         ⋅−= + − I H II H HIII P P F RT tE 2 2ln 2      − ⋅= − I X II X XIII a a Fz RT tE Z ln         ⋅= − − I O II O OIII P P F RT tE 2 2 2 ln4                 +         −⋅=         − ⋅= ++ − I O II O I OH II OH HI H II H HIII P P P P F RT t P P F RT tE 2 2 2 2 2 2 lnln2 4 ln 2  222  Therefore, the proton transport number, +Ht , in oxides can be determined by measurement of the voltage difference across a fully dense disk sample exposed to a gradient in the partial pressure of hydrogen or water vapour, with a constant partial pressure of oxygen on both sides.                   223 APPENDIX C – Electrochemical Testing Procedures C.1 Ac impedance spectrometry Setup preparation 1. Place the ceramic pellet sample with the applied electrodes in the sample holder in the AMEL 7902 test setup (Fig. 2.4) 2. Ensure good contact between the sample and the Pt mesh on both sides of the sample 3. Make sure that there is no short circuit caused by Pt wires touching 4. Place the fused silica cover over the sample holder, secure it by 4 screws and place the Styrofoam thermal insulator around the bottom of the silica cover 5. Insert the holder into the furnace and lock 6. Connect gas (air or H2/ N2) supply to the cathode and anode gas inlet 7. Connect four electrodes (two electrodes for potential, two for current) to the ac impedance analyzer (IM6 by Zahner Electrinks) 8. Turn on the cooling water in the supporting plate Temperature/gas flow profile set up After placing the sample in the holder and setting up the testing system, log into the Advanced Measurements Integrity (AMI) v.3 software interface to program the procedure for heating/gas profile. The interface is shown in Fig. C.1. By selecting “Configure” and following the instructions, temperature and gas flow profile can be programmed and saved. 1. Select the “Test selection” tab  224 2. From the “Sequence selection” drop down menu select a desired test 3. Enter the sample info in UUT info window 4. In the “Test channels”  select the temperature and gas flow data to be recorded 5. After setting up desired conditions, click “Run sequence” or “Run selected”, depending if all or only selected steps will be run Fig. C.1 shows an example of the test profile for Ac impedance testing in air. Similar method can be used to program the testing in H2/ N2. Ramp to every temperature was at 2oC/min and dwell time at each temperature was set to 3 h. Gas flow (air or 50% H2/50% N2) was set to 100 sccm.                225  Figure C.1 – Advanced Measurements Integrity (AMI) v.3 interface for temperature and gas flow profile. An example of a profile for an ac impedance test in air is shown on the screen.  Ac impedance testing To setup the sequence for the ac impedance measurement, enter the Zahner electrochemical testing interface, Thales 4.15, and choose EIS measurement. The window that appears is shown in Fig. C.2 (top). 1. Select the frequency range in the “ Recording parameters” box (100 mHz to 8 MHz range used in the tests)  226 2. Click on “Control potentiostat“ to set up the ac perturbation amplitude (50 mV used in the tests) 3. Select the graph type in “ Display spectrum” (Nyquist used) 4. Select “Define series measurement“ to program the test sequence. Screen that appears is shown in Fig. C.2 (bottom). 5. Click on “linear scan” and enter: start time of the first ac scan (aligned with the time when temperature of the sample reaches 100oC), delay between each scan (15 min) and total duration of the measurements ( 70 h) 6. Click  “file op’s“ button to select the directory where data will be saved 7. In the box “inputs“ write the file name 8. Select “time” for the variable parameters 9. Click “go” to start measurement            227                      Figure C.2 – Thales 4.15 interface for setting up the ac impedance measurements.   228 C.2 Proton transport number determination by the EMF measurement The principles of the EMF measurement and the schematic of the utilized concentration cell are explained in Appendix B. Setup preparation 1. Place the ceramic sample pellet between the two vertical alumina tubes of the concentration cell 2. Apply Ceramabond to seal the pellet to the alumina tubes on both sides to avoid leak 3. Apply Pt mesh with Pt wires (electrodes) on both sides of the sample and ensure they are connected to the MultiSTAT 1480A-Solartron 4. Heat up the hot vessel for the supply of the vapour to the compartment I to 115oC 5. Fill up a 100 ml syringe with water, place it in an automatic pump and connect to the hot vessel with a hose 6. Set up the desired flow rate of water to be supplied by the pump to the hot vessel in order to ensure the desired flow of saturated vapour to the compartment I 7. Ensure that the water bubbler for the supply of the humidified gas to the compartment II is filled with water Temperature/gas flow profile set up and test procedure Set the temperature/gas flow profile using the same AMI interface as described in appendix C.1. Ramp rates to each temperature were 2oC/min. The test procedure was as follows:  229 1. Ramp up to 94oC and hold for 2 h (no gas flow) 2. Ramp up to 240oC and hold for 4 h (no gas flow) 3. Heat to 400oC and hold for 3 h to release all absorbed water from the sample 4. Cool down to 100oC 5. Set compartment I to 48%vol H2/49%vol N2/3%vol H2O (100 sccm total flow) 6. Set compartment II to 48%vol H2/49%vol N2/3%vol H2O (100 sccm total flow) 7. Measure ac impedance 8. Measure voltage difference between the two compartments (e.m.f.) until the stable value is obtained to determine the reference voltage when there was no gradient across the sample 9. Change compartment I to 80%vol H2/15%vol N2/5%vol H2O and record the change in the e.m.f. 10. Change compartment I back to 48%vol H2/49%vol N2/3%vol H2O and record the change in e.m.f. 11. Measure impedance at the end of the cycle 12. Repeat steps 8 to 11 three times at the same temperature 13. Repeat steps 5 to 12 for each of the temperatures (200oC, 300oC and 400oC) 14. Cool down and turn off the gases  Ceramabond curing  230 C.3 OCV, polarization curve and potentiostatic measurement The schematic of the test setup is given in Chapter 3, Fig. 3.6. Setup preparation 1. Place the gas-tight ceramic sample pellet with electrodes on both sides between the two horizontal alumina tubes of the testing cell 2. Apply Ceramabond to seal the pellet to the alumina tubes on both sides to avoid leak 3. Apply Pt mesh with Pt wires-one for the current, the other for the potential determination on both sides of the sample and ensure they are connected to the Solartron 1260 Frequency Response Analyzer coupled with the Solartron MultiSTAT 1480A 4. Close the furnace Temperature/gas flow profile set up and test procedure – OCV testing Set the temperature/gas flow profile using the same AMI interface as described in appendix C.1. Ramp rates to each temperature were 2oC/min. The test procedure was as follows: 1. Ramp up to 94oC and hold for 2 h (no gas flow) 2. Ramp up to 240oC and hold for 4 h (no gas flow) 3. Heat to 400oC and hold for 3 h to release all absorbed water from the sample 4. Cool down to 100oC 5. Set anode gas supply to 48%vol H2/49%vol N2/3%vol H2O (100 sccm total flow) Ceramabond curing  231 6. Set cathode supply to 100 sccm air 7. Measure OCV until it reaches a stable value and for 3 h after that 8. Heat to 200oC while keeping the anode and cathode gas supply unchanged 9. Measure OCV until it reaches a stable value and for 3 h after that 10. Repeat steps 8 and 9 for 300oC, 400oC and 500oC 11. Cool down to room temperature and turn off the gases Temperature/gas flow profile set up and test procedure – polarization curve Using the same setup as for the OCV measurement, the polarization curve was measured for the BIO sample at 300oC, 350oC, 400oC, 450oC and 480oC. The temperature/gas flow profile was set using the same AMI interface as described in appendix C.1. Ramp rates to each temperature were 2oC/min. The test procedure was as follows: 1. Ramp up to 94oC and hold for 2 h (no gas flow) 2. Ramp up to 240oC and hold for 4 h (no gas flow) 3. Heat to 400oC and hold for 3 h to release all absorbed water from the sample 4. Set anode gas supply to 48%vol H2/49%vol N2/3%vol H2O (100 sccm total flow) (tests with 50%vol H2/50%vol N2, no water were also performed) (no gas flow on the cathode) 5. Measure ac impedance every 15 min until high conductivity is reached 6. Set cathode supply to 100 sccm air 7. Measure OCV until it reaches a stable value Ceramabond curing  232 8. Apply a potentiodynamic measurement and record the polarization curve 9. Measure ac impedance 10. Repeat steps 8 and 9 three times at the same temperature 11. Increase temperature to 350oC and repeat steps 7 – 10 12. Do step 11 for 400oC, 450oC and 480oC 13. Cool down to room temperature and stop gas flow An example of the test procedure is given in Fig. C.3.  Figure C.3 – Example of the potentiodynamic measurements using Solartron MultiSTAT 1480A.  233 Temperature/gas flow profile set up and test procedure – potentiostatic measurement Using the same setup as for the OCV and polarization curve measurement, the potentiostatic measurement was performed at 300oC, 350oC, 400oC, 450oC and 480oC. The temperature/gas flow profile was set using the same AMI interface as previously described. Ramp rates to each temperature were 2oC/min. The test procedure was as follows: 1. Ramp up to 94oC and hold for 2 h (no gas flow) 2. Ramp up to 240oC and hold for 4 h (no gas flow) 3. Heat to 400oC and hold for 3 h to release all absorbed water from the sample 4. Set anode gas supply to 48%vol H2/49%vol N2/3%vol H2O (100 sccm total flow) (tests with 50%vol H2/50%vol N2, no water were also performed) (N2 on the cathode) 5. Measure ac impedance (A: 100 sccm H2/N2, C: N2) every 15 min until high conductivity is reached 6. Reduce the temperature to 300oC 7. Measure ac impedance (A: 100 sccm H2/N2, C: N2) to make sure sample is still highly conductive 8. Perform potentiostatic measurement  - apply voltage V=0.5V across the sample and measure the current conducted through the sample over 500 sec 9. Set 100 sccm air on cathode and record the change in current (at the same applied voltage) over next 500 sec Ceramabond curing  234 10. Measure ac impedance (A: 100 sccm H2/N2, C: 100 sccm air) 11. Continue recording the potentiostatic measurement 12. Stop the flow of air on the cathode and keep recording the current for the next 500 sec 13. Perform ac impedance again (A: 100 sccm H2/ N2, C: no air flow) 14. Repeat steps 7 – 13 for three times 15. Change the temperature to 350oC, 400oC, 450oC and 480oC  and repeat steps 7 – 13 at each temperature 16. Cool down to room temperature and stop gas flow           

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0059159/manifest

Comment

Related Items