<%BANNER%>

Kinetic and Catalytic Characterization of Lanthanum Ferrites and Manganites for Gas Sensor and Fuel Cell Applications

Permanent Link: http://ufdc.ufl.edu/UFE0042073/00001

Material Information

Title: Kinetic and Catalytic Characterization of Lanthanum Ferrites and Manganites for Gas Sensor and Fuel Cell Applications
Physical Description: 1 online resource (192 p.)
Language: english
Creator: Armstrong, Eric
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: characterization, cobaltitie, ferrite, lanthanum, manganite, perovskite, reduction, sensor, surface
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Lanthanum ferrites and manganites have been shown to be excellent catalysts for use as sensing electrodes for potentiometric gas sensors to detect CO, CO2, and NOx gases and for use as cathode materials for solid oxide fuel cells (SOFC) to reduce oxygen. However, these materials have complex surfaces that are not well understood. For further optimization, more understanding of the surface behavior is needed. NOx adsorption behavior on LaFeO3 (LFO) and LaMnO3+delta (LMO) was characterized using temperature controlled methods and mass spectrometry. Temperature program desorption revealed decomposition of complex surface species formation when NO or NO2 was adsorbed on LFO and LMO. LFO exhibited higher adsorption capacity for NOx species than LMO and was shown to be more active for NOx surface conversion. Both effects were attributed to the different B-site cations, with iron in LFO in the 3+ valence state, and manganese in LMO in the 3+ and 4+ valence states. Results from diffuse reflectance infrared spectroscopy were used to identify specific nitrite and nitrate species that are formed on the surfaces of LFO and LMO at room temperature. Temperature programmed reaction revealed a complex NO2 decomposition mechanism to NO and O2 for LFO and LMO in which the formation of nitrite and nitrate species serve as intermediates below ~600degreeC. NOx sensing mechanisms were considered and predicted based on the types and quantities of surface species formed. A novel approach using isotope exchange was developed called isothermal isotope exchange (IIE) to study the oxygen reduction behavior on lanthanum ferrites and manganites. Two common kinetic parameters have been used to characterize the ability of a material to exchange oxygen: the tracer diffusion coefficient (D*) and the surface exchange coefficient (k*). D* is a measure of the bulk diffusivity while k* is a measure of surface exchange. Both aspects are involved in the oxygen reduction reaction (ORR), but in particular k* can be difficult to measure. Techniques used to measure k* currently require dense, thick samples that restrict testing to the diffusion controlled regime. IIE has the capability of testing powder materials which allows for accurate measurement of k* in the surface exchange controlled regime. (La0.6Sr0.4)(Co0.2Fe0.8)O3-delta (LSCF) was shown to have low activation behavior of k* between 500-800degreeC indicating the surface is catalytically unchanged within the temperature range. On the other hand k* for (La0.8Sr0.2)MnO3 (LSM) was observed to increase with decreasing temperature. This behavior is consistent with a precursor-mediated mechanism in which there is no energy barrier for chemisorbed dissociative adsorption. Composite cathodes were tested and results showed that when a pure electronic conductor is combined with a pure ionic conductor, k* increases to higher values than either of the pure materials alone. k* for different A- and B-site stoichiometries of lanthanum ferrites, cobaltites, and manganites were also measured. The mixed ionic and electronic conductors were shown to exhibit higher k* values than the pure electronic conductors indicating the importance of having both electrons and oxygen vacancies present for ORR These results provide guidance for improving k* of SOFC cathodes and suggestions for optimization were made.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Eric Armstrong.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Wachsman, Eric D.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042073:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042073/00001

Material Information

Title: Kinetic and Catalytic Characterization of Lanthanum Ferrites and Manganites for Gas Sensor and Fuel Cell Applications
Physical Description: 1 online resource (192 p.)
Language: english
Creator: Armstrong, Eric
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: characterization, cobaltitie, ferrite, lanthanum, manganite, perovskite, reduction, sensor, surface
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Lanthanum ferrites and manganites have been shown to be excellent catalysts for use as sensing electrodes for potentiometric gas sensors to detect CO, CO2, and NOx gases and for use as cathode materials for solid oxide fuel cells (SOFC) to reduce oxygen. However, these materials have complex surfaces that are not well understood. For further optimization, more understanding of the surface behavior is needed. NOx adsorption behavior on LaFeO3 (LFO) and LaMnO3+delta (LMO) was characterized using temperature controlled methods and mass spectrometry. Temperature program desorption revealed decomposition of complex surface species formation when NO or NO2 was adsorbed on LFO and LMO. LFO exhibited higher adsorption capacity for NOx species than LMO and was shown to be more active for NOx surface conversion. Both effects were attributed to the different B-site cations, with iron in LFO in the 3+ valence state, and manganese in LMO in the 3+ and 4+ valence states. Results from diffuse reflectance infrared spectroscopy were used to identify specific nitrite and nitrate species that are formed on the surfaces of LFO and LMO at room temperature. Temperature programmed reaction revealed a complex NO2 decomposition mechanism to NO and O2 for LFO and LMO in which the formation of nitrite and nitrate species serve as intermediates below ~600degreeC. NOx sensing mechanisms were considered and predicted based on the types and quantities of surface species formed. A novel approach using isotope exchange was developed called isothermal isotope exchange (IIE) to study the oxygen reduction behavior on lanthanum ferrites and manganites. Two common kinetic parameters have been used to characterize the ability of a material to exchange oxygen: the tracer diffusion coefficient (D*) and the surface exchange coefficient (k*). D* is a measure of the bulk diffusivity while k* is a measure of surface exchange. Both aspects are involved in the oxygen reduction reaction (ORR), but in particular k* can be difficult to measure. Techniques used to measure k* currently require dense, thick samples that restrict testing to the diffusion controlled regime. IIE has the capability of testing powder materials which allows for accurate measurement of k* in the surface exchange controlled regime. (La0.6Sr0.4)(Co0.2Fe0.8)O3-delta (LSCF) was shown to have low activation behavior of k* between 500-800degreeC indicating the surface is catalytically unchanged within the temperature range. On the other hand k* for (La0.8Sr0.2)MnO3 (LSM) was observed to increase with decreasing temperature. This behavior is consistent with a precursor-mediated mechanism in which there is no energy barrier for chemisorbed dissociative adsorption. Composite cathodes were tested and results showed that when a pure electronic conductor is combined with a pure ionic conductor, k* increases to higher values than either of the pure materials alone. k* for different A- and B-site stoichiometries of lanthanum ferrites, cobaltites, and manganites were also measured. The mixed ionic and electronic conductors were shown to exhibit higher k* values than the pure electronic conductors indicating the importance of having both electrons and oxygen vacancies present for ORR These results provide guidance for improving k* of SOFC cathodes and suggestions for optimization were made.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Eric Armstrong.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Wachsman, Eric D.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042073:00001


This item has the following downloads:


Full Text

PAGE 1

1 KINETIC AND CATALYTIC CHARACTERIZATION OF LANTHANUM FERRITES AND MANGAN I TES FOR GAS SENSOR AND FUEL CELL APPLICATIONS By ERIC N. ARMSTRONG A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PA RTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Eric N. Armstrong

PAGE 3

3 To my wife, Stacey a nd t o m y g randfather Norman Silveira

PAGE 4

4 ACKNOWLEDGEMENTS I have been blessed with many rel ationships that contributed significantly to my success as a graduate student. First and for emost it all begins and ends with my savior Jesus Christ who has directed my path and given me the abilities that I have to accomplish this work. Second, the lov e of my life and wife, Stacey has been a tremendous support, enduring all the highs and lows of research with me. I love her very much and w ithout her help and support, it would have been tremendously difficult to complete this process. I also want to t hank my parents David and Norma, for their support of my decision to pursue graduate education and for their help getting me started. They have been an encouragement, despite having to hear about all my adversities from 3,000 miles away. My in laws, Mar k and Gayle Comer, have also been very supportive of my studies. In particular, Gayle has recognized important milestones with fun gifts and encouraging words. I want to thank my faculty advisor, Professor Eric D. Wachsman, for taking a chance on me before we even met and patiently enduring my gradual learning process I have learned much from him about research and have genuinely appreciated all of his guidance throughout the Ph.D. process. I would also like to acknowledge my doctoral committee Professor Juan Nin o, Professor David Norton, Professor Scott Perry, and Professor Jason Weaver for serving as my committee members I am thankful for their comments and feedback at my proposal defense and oral qualifying exams as well as their expert opinions and guidance toward completion of this dissertation. It has been a pleasure working with all of my group mates. F rederick Martin Van Assche III taught me much about graduate research, and particularly mass spectrometry during my first couple of years in the group. Briggs White provided

PAGE 5

5 excellent scientific mentoring in my formative years as a graduate student as well We had many discussions that were very productive and helpful To the other four students who joined the group along with me (Bryan Blackburn, Danijel Gostovic, Cynthia Kan, and Dongjo Oh), it was truly a pleasure to share milestones together and sharpen each others intellect along the journey. Bryan Blackburn was particularly helpful with automating equipment and selflessly giving of his time. Other group members who I worked closely with included Jin S oo Ahn, Sean Bishop, M a tthew Cammarata, Dohwon Jung, Kang Taek Lee, ByungWook Lee, Jianlin Li, Eric Macam (coffee buddy) Tak k eun Oh, and Jeremiah Smith. I also want to thank Dr. Heesung Yoon, who alwa ys has a solution and gave me excellent guidance with regard to experimental equipment design and implementation throughout my graduate career. Dr. Keith Duncan is not only a great friend and brother in Christ but he patiently listened to my scientific c onjectures and offered excellent feedback and guidance including advice on how to quantify my results with model solutions He always motivated me with words of encouragement when I most needed them and he played an integral role in my completion of this process I have been blessed to have had an abundance of personal support during the last five years. Regular meetings or conversations with good friends Steve Dorman, B.J. Fregly, Dustin Heinen, Allyn Kyes, Mark Pepple, and Mark Sheplak were extremely encouraging and rejuvenating. It is always wise to share ones burdens with others and this group shared mine more than they realize. In addition to personal support, B.J. Fregly also generously gave of his time and knowledge by helping me with Matlab

PAGE 6

6 implementation and optimization for which I am very grateful. Finally as I let out a long sigh of relief as the last 5 years of work comes to completion, thank you everyone!

PAGE 7

7 TABLE OF CONTENTS Page ACKNOWLEDGEMENTS ................................................................................................... 4 LIST OF TABLES .............................................................................................................. 10 LIST OF FIGURES ............................................................................................................ 11 LIST OF ABBREVIATIONS .............................................................................................. 14 ABSTRACT ........................................................................................................................ 15 CHAPTER 1 INTRODUCTION ........................................................................................................ 18 2 BACKGROUND .......................................................................................................... 21 2.1 Heterogeneous Catalysis ..................................................................................... 21 2.2 Potentiometric Gas Sensors ................................................................................ 21 2.2.1 Nernstian Behavior ..................................................................................... 22 2.2.2 Non -Nernstian Behavior ............................................................................. 23 2.2.3 Mixed Potential Theory ............................................................................... 24 2.2.4 Differential Electrode Equilibria .................................................................. 24 2.3 Solid Oxide Fuel Cells .......................................................................................... 25 2.3.1 Function ....................................................................................................... 26 2.3.2 Efficiency ..................................................................................................... 26 2.3.3 Mechanism of Oxygen Reduction Reaction ............................................... 28 2.4 Perovskite Material s ............................................................................................. 29 2.4.1 Engineering of Lanthanum Based Perovskites .......................................... 29 2.4.2 Lanthanum Manganites .............................................................................. 31 2.4.3 Lanthanum Ferrites ..................................................................................... 32 2.4.4 Lanthanum Cobaltites ................................................................................. 32 2.5 Experimental Techniques ..................................................................................... 34 2.5.1 Mass Spectrometry ..................................................................................... 34 2.5.1.1 Temperature programmed desorption (TPD) ................................... 34 2. 5.1.2 Temperature programmed reaction (TPR) ....................................... 35 2.5.1.3 Isotope exchange .............................................................................. 36 2.5.5 Diffuse reflectance infrared spectroscopy (DRIF T) ................................... 37 3 NOx ADSORPTION BEHAVIOR OF LaFeO3 AND LaMnO3+ AND ITS INFLUENCE ON POTENTIOMETRIC SENSOR RESPONSE ................................. 40 3.1 Introduction ........................................................................................................... 40 3.2 Experimental ......................................................................................................... 41 3.2.1 TPD Procedure ........................................................................................... 42

PAGE 8

8 3.2.2 TPR Procedure ........................................................................................... 43 3.2.3 DRIFT Procedure ........................................................................................ 43 3.3 Results .................................................................................................................. 44 3.3.1 LFO TPD and TP R ..................................................................................... 44 3.3.2 LMO TPD and TPR ..................................................................................... 46 3.3.3 LFO and LMO DRIFT ................................................................................. 47 3.4 Dis cussion ............................................................................................................. 49 3.4.1 TPD ............................................................................................................. 49 3.4.2 LFO TPD ..................................................................................................... 49 3.4.3 LMO TPD .................................................................................................... 51 3.4.4 Identification of NOx Surface Species on LFO ........................................... 51 3.4.5 Identification of NOx Surface Species on LMO .......................................... 53 3.4.6 TPR ............................................................................................................. 54 3.4.7 NOx Sensor Mechanism ............................................................................. 55 3.5 Conclusions .......................................................................................................... 60 4 NOx ADSORPTION ON LaFeO3 SUPPORTED STRUCTURES: DISTINGUISHING SURFACE CONTRIBUTIONS OF COMPOSITE MATERIALS ................................................................................................................ 72 4.1 Introduction ........................................................................................................... 72 4.2 Experimental ......................................................................................................... 72 4.3 Results and Discussion ........................................................................................ 73 4.3.1 TPD ............................................................................................................. 73 4.3.2 Characterization Metric for Two Surface Catalyst ..................................... 74 4.3.3 Characterization Metric for Three Surface Catalyst .................................. 75 4.4 Conclusions .......................................................................................................... 76 5 DETERMINATION OF SURFACE EXCHANGE COEFFICIENTS USING INSITU ISOTHERMAL ISOTOPE EXCHANGE ............................................................ 84 5.1 Introduction ........................................................................................................... 84 5.2 Theoretical Background ....................................................................................... 85 5.3 Experimental ......................................................................................................... 88 5.4 Results .................................................................................................................. 90 5.5 Discussion ............................................................................................................. 92 5.5.1 Accumulation Profiles ................................................................................. 92 5.5.2 Model ........................................................................................................... 93 5.5.3 Diffusion Coefficients .................................................................................. 93 5.5.4 Surface Exchange Coefficients .................................................................. 94 5.5.5 Characteristic Thicknesses ........................................................................ 97 5.6 Conclusions .......................................................................................................... 98 6 DETERMINATION OF SURFACE EXCHANGE COEFFICIENTS OF COMPOSITE CATHODES USING IN -SITU ISOTHERMAL ISOTOPE EXCHANGE .............................................................................................................. 109

PAGE 9

9 6.1 Introduction ......................................................................................................... 109 6.2 Theoretical Backg round ..................................................................................... 111 6.3 Experimental ....................................................................................................... 112 6.4 Results ................................................................................................................ 114 6.5 Discussion ........................................................................................................... 115 6.5.1 Diffusion Coefficients ................................................................................ 1 15 6.5.2 Surface Exchange Coefficients ................................................................ 117 6.5.3 Characteristic Thicknesses ...................................................................... 118 6.5.4 Intermediate Temperature Cathode Analysis .......................................... 119 6.6 Conclusion .......................................................................................................... 121 7 EFFECT OF A AND B-SITE CATIONS ON THE SURFACE EXCHANGE COEFFICIENT FOR ABO3 PEROVSKITE MATERIALS ........................................ 130 7.1 Introduction ......................................................................................................... 130 7.2 Experimental ....................................................................................................... 132 7.3 Results ................................................................................................................ 133 7.4 Discussion ........................................................................................................... 134 7.5 Conclusions ........................................................................................................ 136 8 ELECTROCATALYTIC ISOTHERMAL ISOTOPE EXCHANGE OF SUPPORTED CATALYSTS ..................................................................................... 147 8.1 Introduction ......................................................................................................... 147 8.2 Experimental ....................................................................................................... 147 8.3 Results and Discussion ...................................................................................... 148 8.4 Conclusion .......................................................................................................... 150 9 CONCLUSIONS ........................................................................................................ 156 APPENDIX A MASS SPECTROMETER CA LIBRATION ............................................................... 160 B ERROR ANALYSIS OF SURFACE EXCHANGE COEFFICIENTS ....................... 161 C ISOTHERMAL ISOTOPE EXCHANGE PROFILES ................................................ 165 LIST OF REFERENCES ................................................................................................. 185 BIOGRAPHICAL SKETCH .............................................................................................. 191

PAGE 10

10 LIST OF TABLES Table page 3 -1 Identified NOx surface species from DRIFT spectra. ............................................ 61 3 -2 (a) NO and (b) NO+O2 deconvoluted desorption peaks from LFO ...................... 61 3 -3 ( a) NO2 and (b) NO2+O2 deconvoluted desorption peaks from LFO ................... 62 3 -4 (a) NO and (b) NO+O2 deconvoluted desorption peaks from LMO ..................... 62 3 -5 (a) NO2 and (b) NO2+O2 deconvoluted desorption peaks from LMO .................. 63 4 -1 LFO composite and component sample measurements. ..................................... 77 4 -2 LFO infiltrated composite and component sample measurements. ..................... 77 4 -3 Values used to calculate % surface exposure of the LFO/YSZ composite. ......... 77 4 -4 Values used to calculate % surface exposure of the LFO infiltrated YSZ. .......... 77 7 -1 Specific surface areas, samples masses and particle sizes of perovskites ...... 138

PAGE 11

11 LIST OF FIGURES Figure page 2 -1 Diagram depicting the function of a solid oxide fuel cell. ...................................... 38 2 -2 Solid oxide fuel cell voltage losses compared to theoretical voltage. .................. 38 2 -3 Possible pathways for oxygen reduction. .............................................................. 39 3 -1 LFO (a) NO TPD and (b) NO+O2 TPD. ................................................................. 64 3 -2 Deconvoluted NO signal from the NO TPD over LFO. ......................................... 65 3 -3 LFO (a) NO2 TPD and (b) NO2+O2 TPD. .............................................................. 66 3 -4 LFO NO2 TPR with thermodynamic equilibrium calculations. .............................. 67 3 -5 LMO (a) NO TPD and (b) NO+O2 TPD. ................................................................ 68 3 -6 LMO (a) NO2 TPD and (b) NO2+O2 TPD. .............................................................. 69 3 -7 LMO NO2 TPR with thermodynamic equilibrium calculations. .............................. 70 3 -8 DRIFT spectra for NO, NO+O2, NO2, and NO2+O2 adsorption on LFO. .............. 70 3 -9 DRIFT spectra for NO, NO+O2, NO2, and NO2+O2 adsorption on LMO. ............. 71 4 -1 NO+O2 TPD for the painted LFO sample. ............................................................. 78 4 -2 NO+O2 TPD for the painted LFO/YSZ composite sample. ................................... 78 4 -3 NO+O2 TPD for the painted YSZ composite sample. ........................................... 79 4 -4 NO2+O2 TPD for the painted LFO sample. ............................................................ 79 4 -5 NO2+O2 TPD for the painted LFO/YSZ composite sample. ................................. 80 4 -6 NO2+O2 TPD for the painted YSZ sample. ............................................................ 80 4 -7 NO+O2 TPD for the L FO infiltrated YSZ scaffold composite sample. .................. 81 4 -8 NO2+O2 TPD for the LFO infiltrated YSZ scaffold composite sample. ................. 81 4 -9 NO +O2 TPD for the LFO powder sample. ............................................................. 82 4 -10 NO2+O2 TPD for the LFO powder sample. ............................................................ 82 4 -11 NO+O2 TPD for the LFO infiltrated LSM/Y SZ composite sample. ....................... 83

PAGE 12

12 4 -12 NO2 +O2 TPD for the LFO infiltrated LSM/YSZ composite sample. ................... 83 5 -1 Blank IIE profile. ................................................................................................... 100 5 -2 IIE of (a) LSCF, (b) LSM, (c) GDC, and (d) YSZ at 800C. ................................ 100 5 -3 Accumulation profile processing method demonstrated for LSCF. .................... 102 5 -4 Fraction of 18O exchanged with lattice 16O for LSCF, LSM, GDC, and YSZ .. 103 5 -5 Measured diffusion coefficients of LSCF and LSM ............................................. 105 5 -6 Measured diffusion coefficients of GDC and YSZ .............................................. 105 5 -7 Measured surface exchange coefficients of LSCF and LSM ............................. 106 5 -8 Energy diagram for a precursor mediated dissociative adsorption reaction. ..... 106 5 -9 Measured surface exchange coefficients of G DC and YSZ ............................... 107 5 -10 Calculated characteristic lengths of LSCF and LSM ......................................... 107 5 -11 Calculated characteristic lengths of LSCF and LS M ......................................... 108 6 -1 IIE of (a) LSCF/GDC, (b) LSM/YSZ, and (c) LSM/GDC .................................... 122 6 -2 Conversion fraction for (a) LSCF/GDC, (b) LSM/YSZ, and (c) LSM/GDC ......... 123 6 -3 Measured diffusion coefficients of LSCF/GDC, LSCF, and GDC ..................... 125 6 -4 Measured diffusion coefficients of LSM/YSZ, LSM, and YSZ ........................... 125 6 -5 Measured diffusion coefficients of LSM/GDC, LSM, and GDC ......................... 126 6 -6 Measured surface exchange coefficients of LSCF/GDC, LSCF, and GDC ...... 126 6 -7 Measured surface exchange coefficients of LSM/YSZ, LSM, and YSZ ............ 127 6 -8 Measured surface ex change coefficients of LSM/GDC, LSM, and GDC. .......... 127 6 -9 Calculated characteristic lengths of LSCF/GDC, LSCF, and GDC ................... 128 6 -10 Cal culated characteristic lengths of LSM/YSZ, LSM, and YSZ ......................... 128 6 -11 Calculated characteristic lengths of LSM/GDC, LSM, and GDC ....................... 129 7 -1 IIE of (a) LMO, (b) LFO, (c) LCO, (d) LSM, (e) LSF, (f) LSC, and (g) LSCF .... 139 7 -2 Conversion fraction of (a) LMO, (b) LFO, (c) LCO, (d) LSM, (e) LSF, (f) LSC 142

PAGE 13

13 7 -3 Measured surface exchange coefficients of LMO, LFO, LCO, LSM, LSF ........ 146 8 -1 Electrocatalytic experimental setup. .................................................................... 151 8 -2 Experimental conditions for electrocatalytic isothermal isotope experiments .... 151 8 -3 Electrocatalytic IIE of LSM on GDC .................................................................... 152 8 -4 dm/dt of oxygen ions through the cell calculated from Faradays law. ............... 154 8 -5 dm/dt of oxygen ions measured by mass spectrometry. .................................... 155 C -1 LSCF IIE profiles ................................................................................................. 165 C -2 LSM IIE profiles .................................................................................................... 166 C -3 GDC IIE profiles .................................................................................................. 168 C -4 YSZ IIE profiles ................................................................................................... 170 C -5 LSCF/GDC IIE profiles ........................................................................................ 171 C -6 LSM/YSZ II E profiles ........................................................................................... 173 C -7 LSM/GDC IIE profiles .......................................................................................... 175 C -8 LMO IIE profiles .................................................................................................... 177 C -9 LFO IIE profiles ................................................................................................... 178 C -10 LCO IIE profiles ................................................................................................... 180 C -11 LSF IIE profiles .................................................................................................... 18 1 C -12 LSC IIE profiles ................................................................................................... 183

PAGE 14

14 LIST OF ABBREVIATIONS D* Tracer diffusion coefficient DRIFT Diffuse reflectance infrared spectroscopy GDC Gadolinium doped ceria IIE Isothermal isotope exchange k* Tracer surface exchange coefficient LCO Lanthanum cobaltite LFO Lanthanum ferrite LMO Lanthanum manganite LSC Strontium substituted lanthanum cobaltite LSCF Strontium and cobalt substituted lanthanum ferrite LSF Strontium substituted lanthanum ferrite LSM Stronti um substituted lanthanum manganite PPM Parts per million SOFC Solid oxide fuel cell TPD Temperature programmed desorption TPR Temperature programmed reaction YSZ Yttria stabilized zirconia

PAGE 15

15 Abstract of Dissertation Presented to the Graduate School of t he University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy KINETIC AND CATALYTIC CHARACTERIZATION OF LANTHANUM FERRITES AND MANGANITES FOR GAS SENSOR AND FUEL CELL APPLICATIONS By Eric N. Arm strong August 2010 Chair: Eric D. Wachsman Major: Materials Science and Engineering Lanthanum ferrite s and manganites have been shown to be excellent catalysts for use as sensing electrodes for potentiometric gas sensors to detect CO, CO2, and NOx gases and for use as cathode materials for solid oxide fuel cells (SOFC) to reduce oxygen. However, these materials have complex surfaces that are not well understood. For further optimization, more understanding of the surface behavior i s needed. NOx adsorpt ion behavior on LaFeO3 (LFO) and LaMnO3+ (LMO) was characterized using temperature controlled methods and mass spectrometry. Temperature program desorption revealed decomposition of complex surface species formation when NO or NO2 was adsorbed on LFO and LMO. LFO exhibited higher adsorption c apacity for NOx species than LMO and was shown to be more active for NOx surface conversion. Both effects were attributed to the different B -site cations with iron in LFO in the 3+ valence state, and manganese in LMO in the 3+ and 4+ valence states Res ults from diffuse reflectance infrared spectroscopy were used to identify specific nitrite and nitrate species that are formed on the surfaces of LFO and LMO at room temperature. Temperature programmed reaction revealed a complex NO2 decomposition mechani sm to NO and O2 for LFO and LMO in which the formation of

PAGE 16

16 nitrite and nitrate species serve as intermediates below ~600 C. NOx sensing mechanisms were considered and predicted based on the types and quantities of surface species formed. A novel approach using isotope exchange was developed called isothermal isotope exchange (IIE) to study the oxygen reduction behavior on lanthanum ferrites and manganites. Two common kinetic parameters have been used to characterize the ability of a material to exchange ox ygen: the tracer diffusion coefficient (D ) and the surface exchange coefficient (k ). D is a measure of the bulk diffusivity while k is a measure of surface exchange. Both aspects are involved in the oxygen reduction reaction (ORR), but in particular k can be difficult to measure. Techniques used to measure k currently require dense, thick samples that restrict testing to the diffusion controlled regime. IIE has the capability of testing powder materials which allows for accurate measurement of k in the surface exchange controlled regime. (La0.6Sr0.4)(Co0.2Fe0.8)O3 (LSCF) was shown to have low activation behavior of k* between 500800C indicating the surface is catalytically unchanged within the temperature range. On the other hand k* for (La0. 8Sr0.2)MnO3 (LSM) was observed to increase with decreasing temperature. This behavior is consistent with a precursor mediated mechanism in which there is no energy barrier for chemisorbed dissociative adsorption. Composite cathodes were tested and result s showed that when a pure electronic conductor is combined with a pure ionic conductor, k* increases to higher values than either of the pure materials alone. k for d ifferent A and B -site stoichiometries of lanthanum ferrites, cobaltites, and manganites were also measured. The mixed ionic and electronic conductors were shown to exhibit higher k* values than

PAGE 17

17 the pure electronic conductors indicating the importance of having both electrons and oxygen vacancies present for ORR These result s provide guidance for improving k* of SOFC cathodes and suggestions for optimization were made.

PAGE 18

18 CHAPTER 1 INTRODUCTION Modernization of industrial processes and transportation methods over the last two centuries has created a social and political culture confronted with concerns of environmental health and safety as well as natural resource limitations. An increasing amount of attention is being focused on engineering clean, renewable, and sustainable energy sources to limit harmful emissions created by processes involvi ng the burning and combustion of hydrocarbons derived from fossil fuels. Many technologies such has hydrogen fuel cells, solar cells, hydropower, wind, geothermal, and battery systems have been proposed, and in many cases, are being implemented already. In addition, to accurately measure the low concentration byproducts of existing and new processes, more sensitive and selective gas sensing technologies are b eing developed. In 1971, the United States Congress passed a bill that became the Clean Air Act [1] initiating regulations on stationary and mobile emissions. Since then, it has been amended several times with th e most significant revisions occurring in 1990. Among the harmful emissions to be limited are carbon monoxide, carbon dioxide, nitrogen oxides (NOx), sulfur oxides (SOx), particulate matter, and volatile organic compounds (VOCs). These gas emissions cont ribute to environmental pollution such as acid rain, smog, and ground ozone, and health issues such as respiratory infection, aggravation of existing health conditions, and failure to absorb oxygen resulting in fatality. With the production of more automobiles, planes, ships and other modes of transportation that utilize combustible fuels, there is a pressing need to reduce their harmful emissions. According to the Environmental Protection Agency, regional air quality is being monitored in multiple locations for pollutant levels [2]. This system is useful for

PAGE 19

19 estimating overall levels, but concentrated levels of harmful gases such as NOx are e mitted locally wherever gasoline or diesel is burned. Individuals interacting closely with these sources on a daily basis will be more affected than those who are not in the proximity. Currently there are no devices that can accurately measure the low co ncentration of NOx emissions from automobiles so standards are somewhat subjective. Potentiometric gas sensors, which operate by the generation of a potential response when gas molecules adsorb and catalytically decompose on the surface of catalysts, hav e shown promise to detect such gases. However, characterization regarding sensing mechanism and materials selection are outstanding issues that need to be resolved. Aside from emissions of combustible hydrocarbon fuels, there is also the problem of limite d re sources. Fossil fuels are not unlimited, though it is not known exactly the totality of supply available, and it is estimated that under current consumption rates, supply could be depleted within the next half century if current power generation is no t supplemented by other fuels or renewable energies [3] One promising technology that could be used to help alleviate the burden is s olid o xide f uel c ells (SOFC ). SOFCs convert chemical energy into electrical energy with the potential for up to 70% efficiency compared to 30% efficiency from internal combustion engines limited by the C arnot cycle. SOFCs are most efficient when operating with hydrogen a s the fuel, but have the benefit of still being more efficient than internal combustion engines when operating with hydrocarbon fuels and they evolve significantly lower concentrations of harmful emissions However, there are still obstacles to overcome. SOFCs operate at high temperatures (8001000 C ), which makes materials costs prohibitive. Therefore, there

PAGE 20

20 is a driving force to make these materials less expensive and easier to design, which can be achieved by lower ing the operating temperature. Howev er, w hen the operating temperature is lowered, the kinetics of the electrochemical reactions that take place on the surface of the electrodes, which are thermally activated, decrease causing polarization (efficiency) losses. The electrochemical reaction o f particular concern takes place on the cathode where oxygen is reduced. It is well known that oxygen is reduced faster with certain catalysts, but what is not well understood is the mechanism of the surface reaction. New characterization techniques are needed to aide in the fundamental understanding of the oxygen reduction reaction. Both gas sensor and SOFC technology can be achieved with the same type of materials used for electrolytes and electrodes. In particular, lanthanum ferrites and manganites wi th the perovskite structure have been identified as excellent catalysts for the detection of NOx in gas sensors and the reduction of oxygen in SOFCs. However, further characterization is needed to understand the surface activity and kinetic behavior of th ese materials.

PAGE 21

21 CHAPTER 2 BACKGROUND 2.1 Heterogeneous Catalysis Potentiometric gas sensors and solid oxide fuel cells (SOFC) function because of the use of materials that facilitate or catalyze surface reactions. Catalysis is the acceleration of a reacti on by chemical compounds without the net alteration of the chemical compounds (i.e. the chemical compounds assist in the reaction). Homogenous catalysis occurs when the catalyst is soluble in the reaction medium. When the catalyst is a separate phase from the reaction medium, the term used to describe the reaction is heterogeneous catalysis. In the case of gas sensors, heterogeneous catalysis takes place on the electrodes where gaseous species such as NO or NO2 adsorb, dissociate, and combine with other surface species such as adsorbed and dissociated O2 to form other surface species. Examples of formed surface species include nitrite (NO2 -) and nitrate (NO3 -) species. These reactions taking place simultaneously on two different catalytic electrodes c reates a potential difference, which is used as a sensor response. In the case of an SOFC, the cathode needs to be catalytic toward the reduction of oxygen while the anode needs to be catalytic toward the oxidation of hydrogen. For both gas sensors and SO FC s these reactions take place between a gas phase and a solid phase, thus they are heterogeneous reactions. 2.2 Potentiometric Gas Sensors Potentiometric gas sensors are made up of two electrodes sandwiching a solid electrolyte. In response to electroch emical reactions taking place at each electrode, the electrical potenti al changes and that potential difference is the output signal used to

PAGE 22

22 determine gas concentrations the electrodes are exposed to. There are several types of potentiometric sensors that can be or are used for different applications. The most common is the -sensor used as an oxygen sensor in automotive combustion engines. T he response for this type of sensor is based solely on thermodynamics and can be described by Nernsts law. Howev er, w hen electrodes catalytic toward the oxidation or reduction of multiple reactions are used, the response can no longer be attributed purely to thermodynamics and kinetics must be considered. In this case, non-N ernstian behavior is observed. 2.2.1 Nern stian Behavior A standard -sensor has a dense, oxygen ion conducting yttria stabilized zirconia (YSZ) electrolyte with two porous platinum electrodes. One electrode is exposed to air and is used as a reference, while the other is exposed to the gas stream to be measured. At high enough temperatures, the YSZ becomes conductive and oxygen ions are in equilibrium at the interface of each electrode/electrolyte pair according to Eq. 2 -1. 2 2O 2 e 4 O (2 -1 ) The electrolyte must be in electrochemical equilibrium. Since an i onic gradient exists if the pO2 at the sensing electrode is different than that of the air reference, a compensating electrical gradient also exists which creates and electromotive force (EMF) response according to Nerns t s law (Eq. 2-2 ). ) pO pO ln( F 4 RT E) ref ( 2 ) sense ( 2 (2 2 )

PAGE 23

23 E is the potential response, R is the universal gas constant, T is temperature, F is Faradays constant, pO2(sense) is the oxygen partial pressure of the stream to be measured, and pO2(ref) is the partial pressure of the air reference. This type of sensor has been used successfully in automobiles to monitor and control the air to fuel ratio of combustion engines since the 1970s. It is able to withstand operating temperatures up to 900C and has a lifetime of 100,000 miles [4] However, this type of sensor is only useful for detecting oxygen and canno t be used to detect small concentrations of combusted effluents such as CO, CO2, CH4, and NOx. 2.2.2 NonN ernstian Behavior To measure low concentrations of combusted effluents, metal oxide semiconductor materials have been used as electrodes instead of pl atinum. These metal oxides contain transition metal cations which have been shown to be good catalysts towards the oxidation or reduction of combusted effluents as outlined in several reviews on solid-state gas sensors [5 7] Since a typical combustion stream contains multiple gases that may react on an electrode, depending upon whether electrons are donated or received and the space -charge layer created by adsorbates the electrical potential of the electrode changes thereby changing the potential response between the sensing electrode and the reference electrode. No longer is the EMF response from thermodynamic equilibrium but the kinetics of the reactions must be accounted for to explain the electrochemical potential between the two electrodes Since the Nernst equation cannot describe this behavior, it is considered non-N ernstian. One advantage of using metal oxides as catalysts to measure low concentrations of combustible effluents is that both electrodes can be exposed to the same gas stream. If one electrode is a better catalyst for oxidation or reduction than the other, a potential

PAGE 24

24 response will be generated as the electr ochemical potential of one electrode will change must faster than the other. 2.2.3 Mi xed Potential Theory One of the common interpretations to describe the potential response between two electrodes associated with multiple electrochemical reactions occurring simultaneously is the Mixed Potential Theory [8, 9] This theory states that the potential difference between two different electrodes is attributed solely to the reduction and oxidation couple taking place at each electrode. For example, in an NO and O2 containing gas, the cathodic and anodic reactions can be considered to proceed according to Eqs. 2 -3 and 2 -4 respectively. 2 2O 2 e 4 O (2 -3) e 4 NO 2 O 2 NO 22 2 (2 -4) If the electrodes are in two different environments, then the cathodic reaction takes place on the reference electrode and the anodic reaction takes place on the sensing electrode. If both electrodes are in the same environment then both anodic and cathodic reactions will take place on both electrodes. However, this does not account for adsorption/desorption behavior, which on a semiconductor results in a change in the Fermi level at the surface. 2.2.4 Differential E lectrode Equilibria Differential Electrode Equilibria incorporates both the thermodynamic and kinetic contributions of catalytic semiconducting metal oxides [10] At high temperatures, where kinetics are fast, Mixed Potential Theory may be able to fully explain the potential difference between electrodes, however at lower temperatures the electrochemical

PAGE 25

25 reaction is only one contribution to the potential difference. One of the best examples is the theoretical thermodynamic equilibrium of NO at high temperatures. NO should decompose to N2 and O2, however this has been shown not to occur ov er several metal oxide catalysts [11, 12] Since NO never achieves thermodynamic equilibrium, it is clear kinetic contributions are affecting the potential response that cannot be explained by Mixed Potential Theor y. Differential Electrode Equilibria suggests non equilibrium behaviors exist on the surfaces of the electrodes that change the local gas concentration. In addition, the surface species formed either donate or accept electrons, changing the Fermi level of the electrode and further altering the potential response. The best example of this is the differences in sign of the response signal for oxidizing (NO2) and reducing (CO) species on p-type (LaFeO3) and ntype (WO3) electrodes [13] On the p-type material, NO2 adsorption results in a dec rease in potential response relative to a platinum reference electrode, but on the n-type material an increase in potential is observed. In the case of the reducing gas the opposite response is seen. A dsorption causes an increase in potential response on the p-type material and a decrease in potential on the n -type material. If the electrochemical reactions taking place at the electrodes were the only contributions to the potential response, both p -type and ntype semiconducting materials would give potential responses with the same sign. Therefore Differential Electrode Equilibria has been shown to be a more comprehensive explanation for sensor response than Mixed Potential Theory. 2.3 Solid Oxide Fuel Cells SOFCs are devices that convert chemical energ y into electrical energy, similar to a battery. However instead of stored chemical energy, the chemical energy is constantly

PAGE 26

26 supplied to the device in the form of a gas. SOFCs operate at high temperatures in excess of 8 00 C One of the benefits of SOFCs is the ir ability to reform hydrocarbon fuels, but they function most efficiently with pure hydrogen. When hydrogen is used as the fuel, the byproduct of the overall reaction is water, which makes SOFCs attractive devices for their environmental friendlin ess. 2.3.1 Function An SOFC is composed of two electrodes and an electrolyte. A dense oxygen ion conducting electrolyte separates two porous electrodes. The anode is composed of materials catalytic toward the oxidation of hydrogen according to Eq. 2 -5. e 2 H 2 H2 (2 5) The cathode is composed of materials catalytic toward the reduction of oxygen according to Eq. 2 -6. 2 2O 2 e 4 O (2 6) The oxygen ions formed as a product of reduction then conduct through the electrolyte to the anode where it combines with oxidized hydrogen ions to produce water while electrons produced according to Eq. 2 5 flow through a load connected to the fuel cell (Eq. 2 7) The overall reaction takes place according to Eq. 2 8 O H O H 22 2 (2 -7) O H 2 O H 22 2 2 (2 -8 ) A diagram of an operating fuel cell is shown in Fig. 2 -1. 2.3.2 Efficiency An SOFC generates a voltage which is the potential energy generated by the electrochemical potential difference between the two electrodes. This is commonly

PAGE 27

27 referred to as the open circuit voltage or potential (OCV). For a reversible fuel cell, the OCV is given by Eq. 2 9. zF g Ef (2 -9) Where E is the electromotive force or OCV, fg is the molar specific Gib bs free energy of the overall reaction, z is the number of electrons transferred for each molecule of fuel and F is Faradays constant (96,485 C). In reality, the fuel cell is not reversible and there are several irreversible energy losses. There are thr ee types of irreversible losses associated with the voltage generated by a fuel cell when applied to a load. The losses are seen in Fig. 2 -2. The first portion of loss is due to the activation polarization of the electrodes. These losses are lessened wi th higher catalytically active materials that exhibit rapid adsorption, dissociation, charge -transfer, and incorporation of oxygen into the lattice. The second portion of losses are due to the ohmic resistances of the anode, cathode, electrolyte, and inter connect materials. Unlike activation polarization losses, these losses obey Ohms law, which is why they are called ohmic losses. Using materials with high electronic and ionic conductivities reduces this type of loss as well as using a thin electrolyte layer since the ohmic losses associated with the electrolyte are due to ionic conductivity which is much lower than the electronic conductivity of the rest of the materials. Finally, there can be losses from mass -transport limitations of reactants to the e lectrodes. If products are not swept away or reaction kinetics are faster than the flow of reactant fuel, losses will occur and are commonly referred to as concentration polarization losses.

PAGE 28

28 2.3.3 Mechanism of Oxygen Reduction Reaction The current industr y trend for SOFC fabrication is to lower the operating temperatures to decrease start -up times, decrease expenses associated with using high-temperature materials, and increase seal flexibility. By decreasing the temperature, the activation polarization l osses associated with the electrochemical reduction of oxygen on the cathode increase, and become the highest contribution. To develop ways of improving the kinetics, the mechanisms by which oxygen reduction occur must be understood. Over the last 1520 years, many researchers have sought to explain the complicated ORR with varying degrees of success and no agreed upon standard. A thorough review of the state of ORR understanding was written by Adler in 2004 [14] One overarching conclusion is that there is not one rate limiting step that can be applied to all electrodes and may not be applied to any. Since O2 is thermodynamically stable even at high temperatures, the chemical steps involved, including adsorption, dissociation, and transport often times mask or overlap time-scales associated with charge transfer or interfacial electroche mical kinetics. In particular, the microstructure and polarization history of the electrode play a significant role in emphasizing certain steps For example, lower sintering temperatures and times provide an increase in porosity and surface area for rea ctions to occur, but decrease the interfacial contact area between the electrode and electrolyte where incorporation and transport occur. On the other hand, higher sintering temperatures and times result in increased interfacial contact, with less porosit y and surface area for the reactions to occur. Usually there is a compromise between the two. Additionally composite electrodes that are designed to increase interfacial contact and surface area of the catalyst further

PAGE 29

29 convolute the chemical and electro chemical steps of the ORR. Fig. 2 3 shows a depiction of all possible steps involved in the ORR on a platinum/ YSZ [15] Combining Adlers conclusions with the diagram of po ssible elementary steps, it is no surprise that deconvolution is a non-trivial matter. Furthermore, Adler also concludes that there is a need for in -situ measurements of surface reactions in ambient pressure as a function of temperature, pO2, and polariza tion, which at this time has eluded researchers in this field. 2.4 Perovskite Materials 2.4.1 Engineering of Lanthanum Based Perovskites The ABO3 perovskite materials, where A and B are metal cations, have proven to be the best materials for cathodes up to this point because of their tunability. Properties such as electronic and ionic conductivity, phase stability, and thermal expansion coefficients (TEC) can be adjusted with the choice of A and B -site metal cations. The most stable configuration at SOFC operating temperatures is when the A -site cation has a similar ionic radius to oxygen (1.4 ), the B -site cation is small enough to fit in the BO6 octahedra position between the O anions, and both A and B -site cations are in the 3+ valence state. Goldschm idt developed the toleranc e factor as a metric for configuration stability according to Eq. 2 10 [16] ) r r ( 2 r r tO B O A (2 -10) In this equation, t is the tolerance factor, rA and rB are the radii of the A and B -site cations respectively, and rO is the radius of the oxygen anion. When t=1, the cubic phase, and the most closed packed configuration, of the perovskite is preserved. The further from t=1, the more distorted the lattice becomes. For these reasons, lanthanum

PAGE 30

30 is chosen as the A -site cation because it has an ionic radius closest to oxygen Transition metals are chosen as B -site cations to adjust electronic properties, but they must also fit in the octahedra sites without creat ing a strain in the lattice. The above reasoning was the framework that generated well known cathode materials (La0.8Sr0.2)MnO3 (LSM ) and (La0.6Sr0.4)(Co0.2Fe0.8)O3 (LSCF ). The substitution of Sr2+ on the La3+ site created a charge imbalance that was compensated for by the oxidation of Mn3+ Mn4+. The result was an increase in the electronic conductivity, which for the Mn3+/Mn4+ couple resulted in p-type semiconduc ting behavior. In LSM, like many perovskites with electrical conductivity, conduction occurs by small polaron hopping along the B -O -B chains made up by the Mn3+/Mn4+ couple according to Eq. 2 -11 1 n 1 nM h M (2 -11) The closer the packing a nd less the cell is distorted, the easier it is for electronic species to move. It was thought that since the ORR requires oxygen vacancies and electrons, creating an ionically and electronically conductive material would be ideal for fuel cell operation, especially at lower temperatures. LSCF was engineered for this purpose. Sr2+ substitution on the La3+ sites still resulted in a charge imbalance, but unlike manganese cobalt and iron are thermodynamically more stable in the 2+ and 3+ valence states. This results in charge compensation through the generation of oxygen vacancies and oxygen non-stoichiometry results [17] Because of this, ionic conductivity as well as p -type semiconducting electronic conductivity exists. However,

PAGE 31

31 t otal conductivity and thus electronic conductivity is still lower than LSM at high temperatures [18] 2.4.2 Lanthanum Manganites LaMnO3+ ( LMO ) is unique among the perovskites in that it is oxygen superstoichiometric. Of the three B -site cations reviewed here, it favors oxidation the most. It has exce ss oxygen because manganese favors the 3+ and 4+ valence states, so additional oxygen ions balance the charge on the lattice. Because of this, LMO exhibits intrinsic p-type conductivity. No evidence has been found for the existence of interstitial oxygen atoms, so to accommodate the additional oxygen, it has been shown by neutron diffraction and h ighr esolution t ransmission s pectroscopy that cation vacancies exist on the A and B -sites [19, 20] However, there is disagreement as to the ratios of lanthanum and manganese vacancies. The general molecular formula, which assumes equal vacancies between A and B -site cations, was written by Van Roosmalen et. al. [20] as follows where is a cation vacancy 3 3 4 3 6 3 3 6 3 3 3 3 2 3O Mn Mn La 3 3 O 2 LaMnO The tendency for Mn3+ to oxidize makes it difficult to create oxygen vacancies in LMO. Normally for perovskites this is done by substituting a cation of lower valence, such as Sr2+, on the A -site, but with LM O, manganese compensates for the divalent substitution by oxidizing from the 3+ to the 4+ valence state thereby creating more Mn3+/Mn4+ couples, which increases the p-type conductivity from 150 S/cm [21] to 200 S/cm [22] at 900C (La0.8Sr0.2)MnO3+ (LSM ) has high electronic conductivity but very minimal ionic conductivity because of manganese oxidation. However, LSM does have a TEC (11.7x106 K1) [22] close to (Y2O3)0.08(ZrO2)0.92 ( YSZ ) (10.5E 6 K1) [23] and is

PAGE 32

32 electrocatalytic toward the reduction of oxygen, which is why it has become a standard cathode material. 2.4.3 Lanthanum Ferrites Mizusaki et. al. conducted a thorough study on the electronic conductivity and defect structure of LaFeO3 (LFO ) [24] LFO is oxygen deficient and was shown to have p -type semiconducting behavior. Iron can be stable in valence states 2+, 3+, and 4+. Mizusaki showed LFO has a low concentration of lanthanum vacancies, which is balanced by a low concentration of oxygen vacancies and Fe3+/Fe4+ oxidation couples. Holes conduct via small polaron hopping from B -site to B -site b y changing valence state. Conductivity is very low for LFO. Mizusaki measured an electrical conductivity of 0.6 S/cm at 1000C By substituting Sr2+ on the La3+ site, electrical and ionic conductivity is substantially increased. Bongio et. al. reports electrical conductivity as high as 350 S/cm at 550C [25] for La0.5Sr0.5FeO3 (LSF) with a transition from semiconducti ve to metallic conductivity at about the same temperature, while Yaremchenko et. al. reports ionic conductivities between 0.100.35 S/cm at 750950C for La0.3Sr0.7FeO3 [26] The increase in ionic conductivity is thought to be of particular aide in reducing polarization losses associated with ORR at lower temperatures. The compatible electrolyte, because of similar TECs (~12E 6 K1) [27, 28] is (Ce0.9Gd0.1)O2 (GDC), which is a fast ionic conductor at intermediate temperature operation. 2.4.4 Lanthanum Cobaltites (La1 xSrx)CoO3 (LCO ), unlike LMO, is oxygen deficient in the temperature range 25800C Seebeck coefficient measurements by Mizusaki et. al. [29] show that LCO is a ptype semiconductor up to ~800C where conduction changes to metallic. Since

PAGE 33

33 LCO is both a p-type semiconductor and oxygen deficient, the cobalt ion, which like iron, is stable in the 2+, 3+, and 4+ valence states has some proportion of these. Petrov et. al. reported the tetravalent concentration of cobalt (Co4+) to be 0.005 [30] Therefore there is some Co2+ as well to account for the oxygen deficiency. When a divalent ion such as Sr2+ is substituted on the A -site, p-type conductivity increases due to an increase in Co3+/Co4+ couples, and the semiconducto r to metallic transition temperature decreases. Ionic conductivity also increases because of the creation of oxygen vacancies, which also accommodate the charge imbalance from Sr2+ substitution. The chemical formula for LSC can then be written as follows [30] y 3 4 y 2 x 3 y 2 x 1 x x 1O Co Co Sr La Though LSC shows high electrocatalytic activity toward the reduction of oxygen and substantial electronic conductivities, the main problem with using it as an SOFC cathode is that it has a high TEC (18.523.7x106 K1) [27, 31, 32] One way to increase the electrical conductivity without sacrificing ionic conductivity and keeping the TEC close to GDC is to combine the properties of LSC and LSF by substituting some of the iron for cobalt cations. When this is done, electrical conductivity has been shown to be between 100-340 S/cm in the temperature range 500-900 C for La0.6Sr0.4Co0.2Fe0.8O3 [17, 31, 33, 34] Ionic conductivity was shown to be similar to LSC by Stevenson et. al. with measurements of 0.23 S/cm at 900C [35] Tai et. al. confirmed TEC values of 15.3x106 K1 between 100600C were within acceptable limits of GDC [17]

PAGE 34

34 2.5 Experimental Techniques 2.5.1 Mass Spectrometry Mass spectr ometry is a technique that can be used to analyze the composition of a gaseous sample. Gaseous molecules interact with an orthogonally directed highenergy electron beam so that an electron is ejected from the molecule, resulting in a positively charged ion. In the case of a quadrupole mass spectrometer there are four round electrodes oriented in a square with opposite charge polarities on every other one such that an ideal hyperbolic electric field exists. A combination of DC and RF voltages are used to selectively transmit ions through the quadrupole for detection. Ions that are not selectively transmitted follow an unstable path and are neutralized at the quadrupole electrodes. The concentration of ions is measured using a Faraday cage where each ion t hat is detected is counted. If gas concentrations are very low, a multiplier is used to increase the mass spectrometer signal. In this case, a calibration is conducted to convert the signal to real values. A quadrupole mass spectrometer can be used in a variety of experiments to characterize the catalytic and kinetic behavior of materials. 2.5.1.1 Temperature p rogrammed d esorption (TPD) Temperature programmed desorption (TPD) is a powerful technique used to measure the adsorption/desorption behavior as w ell as the irreversible electrochemical reactions at the surface of catalysts A typical experiment involves the pretreatment of a sample in an oxygen atmosphere to ensure the surface is restored to its equilibrium state and is free of any contaminants. Following pre-treatment, the sample is cooled to the temperature at which adsorption takes place. A gas is adsorbed for 30 minutes or

PAGE 35

35 more and cooled to room temperature under the same flow. The long adsorption times are designed to saturate the surface of the material with the adsorbate. By doing so, the adsorption capacity can be determined by integrating the area under the desorption peaks. Once a t room temperature, an inert gas typically helium is set to flow over the sample until a steady state si gnal is reached. With a controlled ramp rate, the sample is heated to 800C while gas effluents are monitored. Adsorbed species have different bond strengths on the surface and will desorb at different temperatures. T he number and size of desorption peak s observed give evidence of different activation energies for desorption and different surface species altogether as some adsorbates may rearrange to form other species. The specific species is unable to be identified from TPD alone and is often used in c onjunction with infrared spectroscopy. 2.5.1.2 Temperature p rogrammed r eaction (TPR) Temperature programmed reaction ( TPR ) is similar to TPD except the sample is under continual flow of reacting gas while it is heated. Following the pretreatment described above, a reacting gas balanced by helium is allowed to come to equilibrium with the sample at room temperature. Once a steady -state signal has been achieved, the sample is linearly heated under flow of the reacting gas. The effluent gas composition is a gain monitored as a function of temperature and time. The benefit of TPR, as noted by Falconer and Schwarz [36] in their review, is that rate of rea ctions are directly measured. As the surface environment changes, the activities and number of sites of catalysts change also. Using T PR, one can monitor the changes unlike stead -state experiments. Though temperature programmed methods provide transient kinetic information, the interpretation of such data is nontrivial and

PAGE 36

36 every system may exhibit different convoluted results. Usually data is combined with results from other experiments such as TPD to draw reasonable conclusions. 2.5.1.3 Isotope e xchange One way of studying the oxygen exchange behavior of metal oxide catalysts is by using tracer oxygen (18O2) One of the most common way s this is done is isotope exchange depth profiling (IEDP) with secondary ion mass spectrometry (SIMS) [37, 38] In this type of experiment, a dense bar is heated to a temperature high enough for rapid oxygen exchan ge (6001000 C ), equilibrated in a high concentration of 16O2, and then the atmosphere is exchanged with 18O2 and the sample is annealed. 18O2 adsorbs on the surface, dissociates and exchanges with lattice 16O. After annealing, the sample is quenched, locking the 18O in the lattice. Finally, the sample is depth profiled layer by layer to determine the diffusion profile of 18O atoms over thickness of the sample. The diffusion profile can then be fit to Cranks [39] solution to the diffusion equation to extract the diffusion coefficient (D), a bulk property, and the surface exchange coefficient (k), a surface property. These kinetic rate coefficients quantify the rate of oxygen exchange. Alternatively, another use of isotope exchange can be done in-situ with powder materials. In an isothermal isotope exchange (IIE) experiment [40] the sample is heated to a temperature to be tested and equilibrated in continuous flow of a 16O2 atmosphere. In a separate line, an equal pO2 of 18O2 is flown. The lines are switched and the rise and fall of m/z signals 32 (16O2), 34 (16O18O), and 36 (18O2) are monitored with a quadrupole mass spectrometer. By this method, the actual exchange of oxygen can be observed with the removal of latt ice oxygen from the fall of the 32 signal, the incorporation of lattice oxygen from the rise of 36 and the surface activity from the rise

PAGE 37

37 and fall of the scrambled product seen from the 34 signal Like SIMS, accumulation profiles verses time can be proces sed and modeled with Cranks [39] solution to the diffusion equation to extract kinetic parameters D and k. 2.5.5 Diffuse r eflectance i nfrared s pectroscopy (DRIFT) When a sample c ontains an adsorbate on its surface, the species each have a vibrational signature related to its bonding. Diffuse reflectance infrared spectroscopy (DRIFT) can be used to identify different vibrational modes of surface species. A sample is typically mul led with potassium bromide (KBr), which serves as a reflective medium to accentuate the amount of signal absorbed by the sample. IR peaks are a result of signal absorbed by the sample at a particular frequency. Peak frequencies are then assigned and surf ace species are identified by the ir vibrational signature.

PAGE 38

38 Figure 2 1 Diagram depicting the function of a solid oxide fuel cell. Fig ure 2 2 Solid oxide fuel cell voltage losses compared to theoretical voltage.

PAGE 39

39 Figure 2 3 Possible pathways for oxygen reduction illustrated with a platinum electrode and YSZ electrolyte [15]

PAGE 40

40 C HAPTER 3 NOx ADSORPTION BEHAVIOR OF LaFeO3 AND LaMnO3+ AND ITS INFLUENCE ON POTENTIOMETRIC SENSOR RESPONSE 3.1 Introduction An incr ease in pollution and health concern over the harmful effects of NOx emissions from automobiles, electricity generation, and other sources have created a demand for improved gas sensors for their detection and catalysts for their reduction [2] Necessary gas sensor requirements include selectivity and sensitivity to NOx detection, stable response for long periods of time at operating temperatur es between 400 and 1000C and low cost. Potentiometric solid-state sensors using metal oxide s (Cr2O3 [41, 42] La2CuO4 [43 -46] LaFeO3 [13, 4754] LaMnO3 [55, 56] and WO3, [12, 57] ), yttria stabilized zirconia (YSZ), and noble metals (Au and Pt) as the sensing electrode, electrolyte, and counter electr ode respectively, have shown promise to meet these stringent criteria. The NOx sensing mechanism of these potentiometric sensors has been described by Mixed Potential Theory which attributes the potential response to the electrochemical reactions taking pl ace on the electrodes [55, 56, 58, 59] However, in the case of metal oxide semiconducting electrodes, such as LaFeO3 and LaMnO3, adsorption of NOx species alters the Fermi level in the oxide and contributes to the potential response. Therefore a more comprehensive mechanism called Differential Electrode Equilibria that considers adsorption behavior in addition to the response from electrochemical reactions was developed [10] Temperature programmed desorption (TPD) and reaction (TPR) are two techniques used to investigate the gas adsorption and catalytic behavior of materials. TPD can be used to investigate relative binding energies of different adsorbed species,

PAGE 41

41 surface complex formation, and total adsorption. TPR can be used to investigate the kinetic limitations of gas phase reactions on a given material in comparison with thermodynamic equilibrium calculations. Both techniques have been used extensively to study the catalytic properties of numerous materials. Diffuse reflectance infrared spectroscopy (DRIFT) is a technique used to measure the b ond vibrational frequencies of adsorbed surface species. Each species has a vibrational signature that can be used for identification, and when used in conjunction, TPD, TPR, and DRIFT can provide information about surface reactions that is useful for mec hanistic evaluation. Huang et al. [60] and Van Assche et. al. [45] used similar methods to conduct extensive mechanistic analysis of NO adsorption and decomposition on lanthanum oxide and lanthanum copper oxide respectively. LaFeO3 (LFO) and LaMnO (LMO) were select ed for TPD, TPR, and DRIFT analysis to compare materials with similar parent crystal structure (perovskite), but possessing different B -site cations. Iron has a valence charge of 3+ in LFO [61] and manganese has valence charges of 3+ and 4+ in LMO [62, 63] The combination of analysis techniques allowed for the mechanistic un derstanding of surface reactions and their impact on potentiometric sensor response. 3.2 Experimental LFO and LMO powders were prepared by aqueous combustion synthesis using ethylene glycol fuel [64] Lanthanum nitrate and iron nitrate solutions were mixed with ethylene glycol in a stoichiometric oxidizer to fuel ratio for LFO synthesis. Manganese nitrate was substituted for iron nitrate for the synthesis of LMO. Once solutions were adequately mixe d, water was evaporated at 70 C until a viscous gel remained. Gels were placed on a hot plate and heated at 10 C /min until auto ignition ensued, resulting

PAGE 42

42 in the production of homogeneous powders. Powders were calcined at 400C followed by grinding and f urther calcination at both 600 and 800C Finally, powders were ball milled in ethanol for 24 hours, dried, and sieved. X -ray powder diffraction (XRD) using a CuKradiation source was used to confirm phase purity of both materials. The final specific su rface areas were 12.8 and 4.6 m2/g for LFO and LMO respectively, measured by BET Final stoichiometries were La(Fe3+ 0.990.01Fe4+ 0.010.01)O3.010.01 and La(Mn3+ 0.630.01Mn4+ 0.370.01)O3.190.01 [64] using a redox titration technique similar to Jaenicke et al. [65] and Porta et al. [66] 3.2.1 TPD Procedure Adsorbing gas concentrations were 2000 ppm NO (NO), 2000 ppm NO in the presence of 20% O2 (NO+O2), 2000 ppm NO2 (NO2), and 2000 ppm NO2 in the presence of 20% O2 (NO2+O2). All gases w ere research grade and balanced by helium Powder samples were normalized by surface area (~0.6 m2) and supported on a quartz frit in the middle of a continuous flow quartz reactor. Upstream of the sample the reactor had a 4 mm inner diameter while down stream of the sample it had a 2 mm inner diameter to reduce gas residence time. A thermocouple in a quartz sheath was placed at the surface of the sample. The reactor was placed in the center of a custom made tube furnace capable of linear heating rates up to 30 C /min. Prior to each run the sample was pretreated at 700C in a 4% oxygen balanced by helium mixture flowing at 25 cm3/min. to ensure consistent lattice oxygen content and to remove any surface contaminants. After the pretreatment, samples wer e cooled to 300C at 30 C /min. under the same gas flow as the pretreatment. Once the sample reached 300C the adsorbing gas was flowed over the sample for 30 min., followed by cooling to 50C at 5C /min.

PAGE 43

43 under the same gas flow. Below 50C the gas was switched to helium and allowed to equilibrate for 16 hours. Once equilibrated, the sample was heated at 30C /min. to 725C in helium and the desorbed gas effluent was measured using an Extrel quadrupole mass spectrometer (MS). All gases were flown at the same rate of 25 cm3/min. 3.2.2 TPR Procedure Samples were cooled to below 50 C at 30C /min. under the same gas flow as the pretreatment. The sample was kept under helium flow while reactant gases were mixed for 30 min. in a separate line. An automatic s witch was initiated and the reacting gases were flowed over the sample. Once a steady state MS signal was reached, the sample was heated at 30C /min. to 725 C while the gas effluent was monitored by the mass spectrometer. In the case of LFO, the samples continued to adsorb NO2 for some time at room temperature. To prevent excessive adsorption, heating was initiated 12 min. after the switch. The reactant gas used was 1000 ppm NO2 balanced by helium 3.2.3 DRIFT Procedure A 95% potassium bromide (KBr) and 5% metal oxide mixture was mulled for 30 seconds. KBr was used as a medium for reflected IR signals since it absorbs little signal The mulled samples were pretreated in the same manner as in TPD/TPR experiments. Once the sample was cooled to below 50 C it was transferred to a sealed container and stored in a dry box until tested. DRIFT spectra were recorded using a Nicolet Magna 760 IR under dry nitrogen at room temperature. Dried KBr was used as a background to further accentuate the adsorbate spec ies on the metal oxide Spectra were processed by subtracting H2O and CO2 characteristic peaks as both

PAGE 44

44 molecules could easily adsorb from atmosphere during transfer. OMNIC software was used to assign peak values for species identification. 3.3 Results 3. 3.1 LFO TPD and TPR TPD spectra of LFO NO and NO+O2 adsorptions are shown in Fig. 3 1. From the NO TPD, multiple corresponding NO and NO2 non-Gaussian desorption peaks were observed, with the addition of O2 peaks at higher temperature s indicating decompo sition of complex surface species. Since the formation of complex surface species depends on the concentration of two reactants (NO and O2 or NO2 and O2) it is likely that the kinetic rat e of desorption is secondorder, though the order of the rate of des orption was not determined from these results Assum ing the reaction rate is second order, the desorption peaks would be symmetric. A simplified approach to separate the individual species using a manual Gaussian peak fitting program in Excel was applied. Though Gaussian peaks are symmetric, secondorder desorption does not stipulate Gaussian shaped peaks exist However, this approach can be used to gain a general idea of the types of species that are decomposing and desorbing. As an example, the decon volution of the NO desorption peaks from the NO TPD is shown in Fig. 3 2. Since the powders being tested are polycrystalline materials, a variety of surface site terminations could exist each with different energetics for bonding. Because of this, the p eak widths may vary. This is seen in the 300 C peak in the NO deco nvolution. The larger width of the peak could mean it has very different energetics for binding of surface species than the other sites or it could be an aberration of the simplified Gaus sian fitting approach. The process was repeated for all desorption

PAGE 45

45 curves to determine peak temperatures and total species desorbed. All c urve fits had greater than 99% goodness of fits Each desorption peak corresponds to a different surface species, each with a unique activation energy for desorption. The presence of NO2 desorption after NO adsorption is a result of surface reactions taking place between NO and solid-state surface oxygen since no oxygen was added to the system in the gas phase. When oxygen is added to the system, there is an increase in total species desorbed, particularly in the case of NO2. The low temperature peaks that were seen in the NO TPD are no longer observed. It is clear that adsorbed oxygen plays a significant role in t he formation of surface species that decompose into and desorb NO2. Peaks and desorption amounts for NO and NO+O2 TPDs are shown in Table 3 2. In the table, it is seen that NO, NO2, and O2 peaks correspond to each other. This means that it is likely sim ultaneous desorption of decompos ition products is occurring Similar to the NO TPDs, mixed desorption products of NO and NO2 are seen when NO2 is adsorbed both alone and in the presence of oxygen (Fig. 3 3). Similar peak temperatures are observed for both TPDs (Table 3 -3). Increases in total desorption of both NO and NO2 were observed when NO2 was adsorbed in the presence of oxygen. In comparison to NO desorption, a larger increase in NO2 was seen. O2 peaks correspond to NOx desorption peaks, indicating decomposition of complex surface species (i.e. nitrite and nitrate species can decompose into NO and O2 and nitrate species can also decompose into NO2 and O2). Unlike the NO TPD, there are no desorption peaks below 200C

PAGE 46

46 TPR results are compared with t hermodynamic calculations in Fig. 3 4. Thermodynamic calculations for NO2 decomposition to NO and O2 according to Eq. 3 -1 were calculated using the Gibbs free energies of formation to determine the equilibrium constant as a function of temperature. TPR res ults show that LFO adsorbs NO2 up to ~200C and then begins to desorb NO2, as evidenced by the large peak at ~290C At 300C LFO begins to decompose NO2 to NO, but the reaction is not complete within the temperature range as evidenced by the nonzero NO2 concentration levels. Using the equilibrium constant and initial concentrations of NO2 and O2, the concentration of NO over the temperature range was calculated. From the stoichiometric relationship between NO and NO2, the concentration of NO2 was subsequently calculated. The calculated curves are seen in Fig. 3 4. 2 2O 2 1 NO NO (3 -1) 3.3.2 LMO TPD and TPR Fig. 3 5 shows NO and NO+O2 LMO TPDs. From the NO TPD, multiple NO and NO2 desorption peaks are observed, with O2 desorption peaks only c orresponding to peaks above ~200 C In the presence of oxygen, NOx desorption increases, as well as the ratio of NO2/NO desorption. As with LFO, NO+O2 significantly affects the surface reactions taking place. In both NO and NO+O2 TPDs, the loss of oxyge n is observed, beginning at ~480C The evolution of oxygen at this temperature is referred to as oxygen and is attributed to lattice oxygen release compensated by manganese reduction from 4+ to 3+ valence state [62] Peaks and desorption amounts for NO TPDs are shown in Table 3 4.

PAGE 47

47 Total NOx desorption from NO2 and NO2+O2 adsorption is similar for LMO (Fig. 3 6), w ith the NO2 TPD showing slightly more desorption. Oxygen does not change the NO2 adsorption behavior much for LMO. The only differences are seen in some of the O2 to NO desorption ratios, which indicates different quantities of nitrite and nitrate specie s desorbed for each experiment (Table 3 -5). Surprisingly the NO to NO2 desorption ratio is greater for the NO2 TPD (Fig. 3 6) than the NO TPD (Fig. 3 -5), confirming that some surface intermediate species form. The NO2 TPR over LMO (Fig. 3 7) shows two smal l desorption peaks at ~80C and ~300C close to the 105C and 308 C peaks seen in the NO2 TPD. NO2 decomposition begins at ~300C and is not complete within the temperature range of the experiment. Thermodynamic equilibrium curves for NO2 and NO are seen in Fig. 3 7 as well. The LMO TPR more closely matches thermodynamic expectations than what was observed with LFO. 3.3.3 LFO and LMO DRIFT LFO DRIFT results are seen in Fig. 3 8. When NO and NO2 were adsorbed on LFO with and without O2, it was noted th at DRIFT peaks were very similar. Total adsorption increased in order of NO, NO+O2, NO2, and NO2+O2, consistent with TPDs. Peak frequencies were identified as seen in the figure. LMO DRIFT spectra for NO, NO+O2, NO2 and NO2+O2 are seen in Fig. 3 -9. Like LFO, peaks were very similar. LMO showed a greater number of peaks for NO adsorption in the presence of oxygen than without. In addition, NO2 gas adsorptions show greater total quantities of surface vibrational frequencies than NO. Fewer peaks were iden tified from LMO Drifts than LFO and in much lower quantities. Identified peaks are labeled in Fig. 3 9.

PAGE 48

48 From TPD it is possible to gain a general understanding of the number of species desorbed and identify the general species type (ionic, covalent, nitri te, or nitrate) from N:O atom ratios by integrating the areas under the peaks identified from the simplified Gaussian model but it is not possible to identify the specific form of the species. DRIFT results were used to identify possible species decompos ing and desorbing from the surfaces of LFO and LMO. However, experiments were conducted at room temperature, so the species identified are representative of the initial species upon heating, but it is possible that the species rearrange to form different species prior to decomposition and desorption. Interpretation of NOx complex formation using DRIFT can be very subjective due to the sensitive nature of NO and multi valent cation interactions. NO contains an 2px antibonding molecular or bital, and depending on the condition of that orbital (whether an electron is donated or accepted), the vibrational frequencies can change significantly. Nonetheless, there has been enough data reported for NO and NO2 adsorption and surface formation on different metal oxide cations that reasonable interpretations can be made, especially when combined with other techniques such as TPD/TPR. A NOx correlation chart from Hadjiivanov et al.s review was used for species identification in this study [67] Identified surface species at room temperature are seen in Ta ble 3 -1. Species with a ratio of N:O of 1.00 are ionic NO+ and NO-, and covalent NO. Species with a ratio of N:O of 0.50 are ionic NO2 +, covalent NO2, and nitrite species with nitro or monodentate nitrito configurations. Finally, the only species with an N:O ratio of 0.33 is transitional nitrate NO3 -.

PAGE 49

49 3.4 Discussion 3.4.1 TPD To determine the type of species desorbed from TPDs, nitrogen and oxygen atoms were counted. When NO or NO2 desorb without corresponding O2 peaks, the ratio of N moles to O moles i s 1 and 0.5 respectively. In the case where O2 peaks correspond to NO and NO2 desorption peaks, nitrite or nitrate species decomposed upon desorption, and the same procedure was used to identify which species desorbed. A nitrite decomposition results in a ratio of N:O of 0.5 and for a nitrate decomposition, the ratio is 0.33. In the cases where the ratio is not equal to these values, it is possible that multiple species decomposed and desorbed or part of the decomposition was retained by the surface Th ese ratios can be found in Tables 3 -2 3 -5 and possible corresponding species can be found in Table 3 1. It is possible that the ratios can be driven down by oxygen desorption from LFO that takes place at 150370C and 390480C and on LMO at 300450C xygen) and above 450C where lattice oxygen desorbs [67, 68] These temperature ranges were determined by O2 TPDs. O2 desorption from O2 TPDs was less than 20% of O2 desorption in the case of NO and NO2 TPDs. I t is unlikely that as much O2 adsorbed on the surface during NO and NO2 TPDs as would have in O2 TPDs, so it is assumed that most of the oxygen desorption is from decomposed species. 3.4.2 LFO TPD From the NO TPD of LFO, the first three temperatures (140, 195, and 300C ) have small amounts of desorbed NO and NO2 relative to the two high temperature peaks, and no corresponding O2 desorption, therefore they are considered to be from ionic/covalent surface species. From the 140C peak, a 0.7 N:O ratio is seen and is

PAGE 50

50 thus a combination of NO and NO2 species. The 195 C and 300C peaks also exhibit mixed ratios of 0.68 and 0.84 with the latter being comprised of more NO than the former. At 420 and 480C O2 desorption peaks correspond with NO and NO2 desorption peaks indicating decomposition of complex adsorbed species. From the 420 C peak, a ratio of 0.39 suggests primarily nitrate species desorb with some accompanying nitrite. From the 480 C peak, the opposite is true with a ratio of 0.44 suggesting primarily nitrite species are desorbed. Therefore when NO is adsorbed on LFO, nitrite species are more stable than nitrate. When oxygen was added, the ionic/covalent species disappear and the first desorption peak at 312C has a ratio of 0.49. This peak is thus d ecomposition of a nitrite species. The three peaks that follow at 338, 412, and 479C all have N:O ratios of 0.33, 0.24, and 0.32, consistent with nitrate species. Since it is impossible to have a species decompose into a ratio lower than 0.33, the 0.24 ratio is an example where the ratio could have been driven down by oxygen desorption. Given the low ratio, it is still conclusive that nitrate species decomposed and desorbed. It is seen that the addition of oxygen encourages the formation of nitrate spe cies in the case of NO adsorption on LFO. The first desorption peak in the NO2 TPD is at 305 C where a ratio consistent with nitrite species exists. As temperature increases, there is a transition to nitrate species that desorb for the pure NO2 adsorption case at 330 C followed by another nitrite peak at 424C and yet another nitrate peak at 474 C In addition, ionic/covalent species

PAGE 51

51 desorb at 365C since no corresponding O2 desorption is seen. The ratio of 0.61 suggests more NO2 than NO desorption. Wh en oxygen is added, unlike the NO TPDs, the species do not change. All peaks exhibit the same type of ratios. For NO+O2, NO2, and NO2+O2, nitrate species are most stable. 3.4.3 LMO TPD LMO TPDs all exhibit low temperature peaks with no corresponding O2 d esorption peaks indicating ionic/covalent species desorption. In addition, hardly any nitrite species are desorbed with nitrate species having the highest stability on the surfaces. When oxygen was added to the NO TPD, an additional nitrat e peak was seen at 426C making it more stable. The addition of oxygen has little effect on the species formed in the case of NO2 adsorption. All peaks are the same except the 235 and 270C peaks from the NO2 TPD combine to form one peak at 250 C when oxygen was added. In all experiments, nitrat e species were the most stable on LMO. 3.4.4 Identification of NOx Surface Species on LFO DRIFT was used to correlate the desorption peaks from TPD with specific species however, since DRIFT results were from room temperature m easurements, it is possible surface species rearranged prior to decomposition and desorption. From the NO TPD, it is seen that the largest amount of species desorbed occurred at 420C At this desorption temperature, the N:O ratio is 0.39. This ratio means the decomposing and desorbing species is a mixture of nitrate and nitrite species with nitrate being the majority. The only nitrate species detected from DRIFT was free like transitional nitrate at a frequency of 1391 cm1. Two nitrite species were d etected, but were unable to be

PAGE 52

52 distinguished due to overlap of peak frequencies Therefore, the remaining nitrite portion is either nitro, monodentate nitrito, or a mixture of the two. The next largest peak from the NO TPD is at 480 C and has a N:O ratio of 0.44. This ratio still indicates a mixture of nitrate and nitrite, but in this case, the nitrite species is the majority. Since no other nitrate or nitrite species were detected, it is assumed that the same type of species are desorbing as in the cas e of the previous peak, just in different quantities. The peaks at lower temperatures were composed of mixtures of NO (NO-, NO+, NO) and NO2 (NO2 + and NO2) species. In the NO+O2 case, there were two more peaks below 420C in which a nitrite or nitrate dec omposed. In this case, free like transitional nitrate species were identified desorbing at 338, 412, and 479C in order of most to least species desorbed. The peak at 312C is purely a nitrite decomposition comprised of either nitro, monodentate nitrito, or a combination of those species. All ionic/covalent species formed more stable nitrite and nitrate species with added oxygen, as none were detected by DRIFT. NO2 adsorption on LFO resulted in larger quantities of nitrite and nitrate formation than NO adsorption. No ionic species were formed at low temperature, but mixed ionic desorption was seen at 365C With a N:O ratio of 0.61, desorption is primarily NO2 by nature and is likely to be ionic (NO2 +) since the peak at 2400 cm1 is much greater for NO2 than NO adsorption. With the addition of oxygen, nitrite and nitrate formation once again increases. The same species are formed, but are formed more readily when more oxygen is present either with the adsorption of NO2 instead of NO, or if O2 is present in the gas stream.

PAGE 53

53 3.4.5 Identification of NOx Surface Species on LMO NO adsorption on LMO resulted in more straight forward N:O ratios for ionic/covalent species desorbing at low temperatures. The N:O ratio for the 140C peak shows a mixed desorption, b ut the peaks at 250 and 323C show pure NO2 and NO desorption respectively. However, the DRIFT spectra showed no measurable amounts of ionic/covalent species at room temperature, so they are unable to be identified. The peaks at 323 and 374 C also are str aight forward. The 323 C peak desorbs as a nitrite species that is identified as nitro species since that was the only nitrite detected from DRIFT. The 374 C peak desorbs as a nitrate with some O2 desorption. The only identified nitrate species is the s ame as LFO: free-like transitional NO3 -. When oxygen was added, more ionic species desorbed, and in this case DRIFT spectra revealed ionic species on the surface of LMO at room temperature. Most of the species were in the form of NO2 and the only NO2 spec ies identified were of the form NO2 +. The three highest temperature peaks have ratios exhibiting nitrate decomposition and desorption with the highest at 426C desorbing some nitro species in addition Unlike LFO, adsorption of NO2 resulted in an increase in ionic species formation and desorption at lower temperatures. The NO2 desorptions at 308 and 175C are all or almost all NO2 and can be identified from peak heights with the 308 C peak being covalent NO2 and the 185C peak being ionic NO2 +. The peak at 270 C is also easily identified since it is all NO. The largest NO desorption is covalent NO. At high temperatures, nitrate desorption increases relative to NO adsorption and no nitrite desorption is seen. The addition of oxygen to NO2 increased nitrate formation and caused the formation of some nitrite species at higher temperatures. In the case of the nitrite

PAGE 54

54 formation, it is possible from the DRIFT spectra that these species could be monodentate nitrito instead of nitro, but they cannot be distinguished due to overlap of peak frequencies. Covalent NO2 and ionic NO2 + still form, but slightly less than with only NO2 adsorption. Overall, LMO is less active for nitrite and nitrate formation and NOx species adsorption than LFO. Where NO2 and O2 streams facilitated the conversion of ionic and covalent species to more stable higher temperature species on LFO, there was not a significant impact on LMO. 3.4.6 TPR Though the TPD desorption ratios and DRIFT spectra match reasonably well for desorbing nitrite and nitrate species, it is likely that surface species are rearranging and forming intermediates before decomposing and desorbing at varying temperatures as the sample is heated. Evidence for such behavior can be seen in the TPRs in Figs. 3 4 and 3 -7 wher e NO2 is converted to NO as the sample is heated in a constant gas stream according to Eq. 3 -1. The reverse reaction of Eq. 3 -1 is thermodynamically favorable at low temperature and the forward reaction is favorable at high temperature. In the case of LFO and LMO, conversion begins at ~300C ~100C higher than thermodynamic calculations suggest. It was previously shown from TPD and DRIFT that nitrite and nitrate species form on the surface of LMO and LFO when NO or NO2 is adsorbed from 300C to room tem perature. Formed nitrite and nitrate species will decompose below the decomposition temperature of NO2 with nitrite species having less stability than nitrate [69] In fact this is what is seen in the TPR where both LFO and LMO start to decompose nitrite species at ~100 C Further decomposition of nitrate species occur at

PAGE 55

55 different temperatures for the two materials. According to Yuvaraj et. al. [70] the s table anhydrous nitrates for iron and manganese Fe(NO3)3 and Mn(NO3)2, have decomposition temperatures of ~160C on LFO and ~200C on LMO. Therefore nitrate species are more stable on LMO than LFO. This is seen from TPR results since LMO begins decompo sing nitrate species at ~200 C and LFO below ~200 C overlapping with nitrite decomposition. Furthermore, LFO was observed to exceed the NO2 to NO conversion predicted by thermodynamics, while LMO lags conversion. This is due to nitrite/nitrate formation and stability differences on LFO and LMO rather than the metal oxides being more or less active toward decomposition of NO2. TPR results also show LFO to have higher adsorption capacity than LMO, consistent with TPD results seen from the amounts of decompo sed surface species following low temperature adsorption. LFO has large quantities of overlapping nitrite and nitrate desorption, but LMO only desorbs small quantities of nitrite and nitrate species with onset separated by ~100C 3.4.7 NOx Sensor Mechani sm According to Mixed Potential Theory [8, 9], the potential difference between a sensing electrode and a reference electrode in a potentiometric sensor is attributed to the redox reactions taking place on both elec trodes according to Eqs. 3-2 and 33. 2 2O NO e 2 NO (3 -2) e 4 O O 22 2 (3 -3) When the sensing electrode is exposed to the reacting gas (NO or NO2) and the reference electrode is exposed to air, both reactions (either forward reacti ons of Eqs. 3-

PAGE 56

56 2 and 33 for NO2 exposure or reverse reactions for NO exposure) occur at the sensing electrode. When NO2, an oxidizing gas, is exposed to the sensing electrode, oxygen is produced at the electrode and increases the potential relative to the air reference. Likewise, when NO, a reducing gas, is exposed to the sensing electrode, oxygen is consumed and the potential at the sensing electrode decreases relative to the reference electrode. Therefore a positive response will be observed for NO2 ex posure and a negative response for NO exposure. This was demonstrated by Miura et. al. [8] for a CdMn2O4 spinel metal oxide sensing elect rode with a platinum air reference. Additionally, Brosha et. al. [56] showed this same behavior when LMO was exposed to CO, a reducing gas like NO, relative to a pl atinum air reference, though their lead configuration was opposite which generated a positive response. However, the mechanism is very complicated and involves many contributions. For example, it was shown that for the configuration where the sensing (LaF eO3 or WO3) and platinum reference electrodes were placed on opposite sides of the electrolyte and both electrodes were exposed to the same gas environment, LaFeO3, a p -type semiconductor, and WO3, an n -type semiconductor, both exhibited positive responses when exposed to NO2 and negative responses when exposed to NO [52, 71] Since both electrodes were exposed to the same gases, both reactions (Eq. 3-2 and 3 -3 for NO2 exposure or NO exposure) occurred on both elect rodes, but the rates of reactions were different. In another configuration where LaFeO3, the sensing electrode, and platinum, the reference electrode, were placed on the same side of the electrolyte as close fingers and both electrodes were exposed to th e same gas environment, LFO exhibited

PAGE 57

57 opposite behavior to that of the configuration where electrodes were on opposite sides of the electrolyte giving a negative potential response to NO2 and a positive potential response when exposed to CO [5052] In the same configuration, WO3 produced opposite potential responses to LaFeO3 exhibiting positive response when exposed to NO2 and a negative response to CO. It was shown that in the close finger configuration, the dif ferent semiconductor types, behaved differently in the same gas environment. This behavior is not consistent with, and cannot be explained by Mixed Potential Theory redox reactions alone, but must include other contributions to sensor response, such as ge ometrical configuration [72] electrode microstructure [44] heterogeneous catalysis [12, 41 43, 57] adsorption of charged surface species [12, 43, 45, 46] and shifts in Fer mi energy [15] In evaluating the sensing mechanism for LFO and LMO, possible paths for formation of nitrite and nitrate species must be considered. Eqs. 34 3 -9 comprise possible formation pathways [45, 46] When NO is adsorbed without any oxygen present, nitrite and nitrate species are formed with lattice oxygen according to Eqs. 3 -4 and 3 -6. When oxygen is present, additional paths may exist according to Eqs. 35 and 3 -7 e V O N O NOO ) ad ( 2 x O ) g ( (3 -4) h O N O NO) ad ( 2 ad ) g ( (3 -5) e 2 V O N O O NO ) ad ( 3 x O ) ad ( 2 (3 -6) ) ad ( 3 ad ) ad ( 2O N O O N (3 -7)

PAGE 58

58 Likewise for NO2 gas adsorption, Eq. 3-8 is the likely formation pathway of nitrate species without gas phase oxygen present and Eq. 39 is another possibility when gas phase oxygen is present. e V O N O NOO ) ad ( 3 x O ) g ( 2 (3 -8) h O N O NO) ad ( 3 ad ) g ( 2 (3 -9) When NO was adsorbed on LFO and LMO in TPD and DRIFT experiments, transitional nitrate species predominately formed, probabl y according to Eq. 3 4 and 3 6, using lattice oxygen to form complex species since adsorbed oxygen was unavailable. When this happens, electrons are injected into the electrode as the charged nitrates are formed. This results in a change in the Fermi ene rgy at the electrode where the nitrates are formed. Since nitrate formation increased when NO was adsorbed in the presence of oxygen, it is likely that formation of complex species proceed according to Eqs. 35 and 3 -7, in which adsorbed oxygen from the gas phase is used to form nitrate species, as well as Eqs. 3-4 and 36, in which lattice oxygen alone is used to form nitrate species. This would cause the injection of holes in the case of nitrate formation with adsorbed oxygen in concert with injection of electrons in the case of nitrate formation with lattice oxygen resulting in a change in the Fermi level. In the NO2 case, nitrite species were predominately formed and the adsorption levels and types of species did not change with or without oxygen. Eqs. 3 4 3 -9 cannot explain this type of species formation from NO2 adsorption unless NO2 decomposes to NO and O2 prior to formation of nitrite species. This is likely to happen at higher temperatures (~600C or higher), but different reactions are likely to occur at lower

PAGE 59

59 temperatures. Another scenario is adsorbed NO2 directly draws electrons from the lattice according to Eq. 310. h O N NO) ad ( 2 ) g ( 2 (3 -10) This would cause the injection of holes in the sensing electrode and a different response than th e NO case where electrons are injected. This behavior can partially explain the opposite responses seen from NO and NO2 gas exposure on LaFeO3. It is also possible that more complicated surface reactions are occurring simultaneously that are unable to be identified from these results. It is not possible to explain the overall changes of sign of sensor response seen in the different geometrical configurations, but it is conclusive that the formation of these charged species, which when formed change the F ermi energy of the electrode, are contributing to the changes in sensor response. With the previous discussion in mind, it is obvious the overall response mechanism varies depending on the operating temperature. At temperatures below ~600C the sensor re sponse occurs by the formation and decomposition of complex nitrite and nitrate surface species, but above ~600 C the response is due to NO2 decomposition according to Eq. 3-1. The lower the temperature, the more species exist on the surface, which cause s the increase in response observed by Di Bartolomeo [13, 51, 52] However lower operating temperatures also causes more stable surface species, which causes the decrease in response time, also observed by Di Bartolomeo [13, 51, 52] It is evident from these results that Mixed Potential Theory cannot adequately explain this behavior, but rather this behavior can be explained by Differential Electrode Equilibria.

PAGE 60

60 When compari ng the expected sensor responses of NO and NO2 for LFO and LMO, since LFO is more active for the formation of surface species than LMO and nitrate species are less stable on the surface of LFO compared to LMO, LFO is expected to be more sensitive (exhibit higher response signal) and exhibit faster response time for NO and NO2 gas sensing than LMO. 3.5 Conclusions From TPD, TPR, and DRIFT results, LFO and LMO show complex nitrite and nitrate surface formation when NO or NO2 are adsorbed. TPDs were conducted to identify the general number of, and general type (ionic, covalent, nitrite, or nitrate) of desorbing species using a simplified Gaussian model DRIFT results were used to identify NOx adsorbates at room temperature, and were correlated to TPD peaks. TPR was used to probe catalytic and kinetic activities for the reduction of NO2. LFO formed surface nitrite and nitrate species in greater quantities and surface species were less stable than those on LMO. LFO was also shown to be more active for NOx sur face conversion. NOx sensor mechanisms on LFO and LMO were proposed to be similar in response sign but differing in sensitivity and response time. Finally, it is clear that any NOx sensor mechanism for these types of semiconducting oxides must include t he formation of complex surface species, such as nitrites and nitrates as intermediates and Mixed Potential Theory does not.

PAGE 61

61 Table 3 1 Identified NOx surface species from DRIFT spectra of NO, NO+O2, NO2, and NO2+O2 adsorbed on LFO and LMO (italicized). Table 3 2 (a) NO and (b) NO+O2 deconvoluted desorption peaks and amounts of gas desorbed from LFO. (a) NO NO 2 O 2 Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) N:O Ratio 140 0.39 135 0.22 0.79 195 0.15 190 0.20 0.68 300 0.23 310 0.12 0.84 420 1.40 390 0.42 415 1.16 0.39 480 0.62 460 0.17 478 0.42 0.44 (b) NO NO 2 O 2 Peak (C) De sorbed 2 ) Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) N:O Ratio 312 0.52 291 1.31 308 0.31 0.49 338 2.78 345 0.94 0.33 412 1.21 410 1.94 0.24 479 1.17 470 1.28 0.32

PAGE 62

62 Table 3 3 (a) NO2 and (b) NO2+O2 deconvoluted desorption peaks and amounts of gas desorbed from LFO. (a) NO NO 2 O 2 Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) N:O Ratio 305 0.31 309 1.76 316 0.22 0.48 330 0.09 335 2.31 340 0.58 0.38 365 0.37 379 1.39 0. 61 424 2.16 415 1.13 0.49 474 0.68 460 0.65 0.34 (b) NO NO 2 O 2 Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) N:O Ratio 305 0.31 317 2.78 321 0.31 0.47 330 0.11 335 2.01 342 0.62 0.36 365 0.51 355 1. 70 0.59 425 2.27 417 1.19 0.49 480 0.69 462 0.68 0.34 Table 3 4 (a) NO and (b) NO+O2 deconvoluted desorption peaks and amounts of gas desorbed from LMO. (a) NO NO 2 O 2 Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) Peak (C) Des orbed 2 ) N:O Ratio 140 0.32 140 0.23 0.77 250 0.08 1.00 323 0.69 0.50 374 0.53 366 0.83 0.25 (b) NO NO 2 O 2 Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) N:O Ratio 102 0.01 1.00 115 0.08 120 0.05 0.86 170 0.08 170 0.80 0.54 250 0.79 0.50 316 0.94 315 0.39 0.30 370 0.65 345 0.05 372 1.17 0.22 426 0.86 458 0.66 0.40

PAGE 63

63 Table 3 5 (a) NO2 and (b) NO2+O2 deconvoluted desorption peaks and amounts of gas desorbed from LMO. (a) NO NO 2 O 2 Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) N:O Ratio 105 0.07 110 0.13 0.67 175 0.35 185 1.16 0.59 235 0.12 0.50 270 0.59 0.51 308 1.53 0.50 372 1.28 375 1.71 0.27 440 0.62 460 0.77 0.29 (b) NO NO 2 O 2 Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) Peak (C) Desorbed 2 ) N:O Ratio 110 0.08 110 0.20 0.65 180 0.05 195 0.93 0.53 255 0.13 0.50 290 0.09 310 1.39 0.53 373 1.46 370 1.33 0.35 442 0.85 445 0.62 0.41

PAGE 64

64 Fig ure 3 1 LFO (a) NO TPD and (b) NO+O2 TPD.

PAGE 65

65 Fig ure 3 2 Deconvoluted NO signal from the NO TPD over LFO.

PAGE 66

66 Fig ure 3 3 LFO (a) NO2 TPD and (b) NO2+O2 TPD.

PAGE 67

67 Fig ure 3 4 LFO NO2 TPR with thermodynamic equilibrium calculations.

PAGE 68

68 Fig ure 3 5 LMO (a) NO TPD and (b) NO+O2 TPD.

PAGE 69

69 Fig ure 3 6 LMO (a) NO2 TPD and (b) NO2+O2 TPD.

PAGE 70

70 Fig ure 3 7 LMO NO2 TPR with thermodynamic equilibrium calculations. Fig ure 3 8 DRIFT spectra for NO, NO+O2, NO2, and NO2+O2 adsorption on LFO.

PAGE 71

71 Fig ure 3 9 DRIFT spectra for NO, NO+O2, NO2, and NO2+O2 adsorption on LMO.

PAGE 72

72 CHAPTER 4 NOx ADSORPTION ON LaFeO3 SUPPORTED STRUCTURES: DISTINGUISHING SURFACE CONTRIBUTIONS OF COMPOSITE MATERIALS 4.1 Introduction Based on the results of NOx adsorption behavior on LaFeO3 (LFO) and LaMnO(LMO) in chapter 3, LFO was chosen to study the NOx adsorption behavior of composite structures, particularly the ability to distinguish between surface contributions of the different compon ents of the composites Composite sensing electrodes were for med by combing LFO with a compatible electrolyte (Y2O3)0.08(ZrO2)0.92 (YSZ) to increase the surface area of the catalyst and maintain structural integrity. As part of the characterization, a new metric using mass spectrometry was designed to quantify the catalyst surface exposure of two and three surface composites. 4.2 Experimental LFO was processed by the methods described in section 3.1. The powder was dispersed in a solvent and painted onto a thin (~500 m) YSZ substrate. Samples were fired and broken up into piec es to fit in the quartz reactor described in section 3. 2. 1. In the same way, LFO and YSZ powders were mixed in a 50:50 composition and painted onto a YSZ substrate. Finally, YSZ powder was dis persed and painted on a YSZ substrate. The specific surface areas of the samples were measured by BET and are shown in Table 4 1. All samples were normalized by surface area to ~0.6 m2 for direct comparison. In addition to a 50:50 composite, a separate sample with infiltrated LFO was made. In this case, the powder, processed according to section 3.1, was characterized as the LFO parent component. A paint ed YSZ scaffold on a YSZ substrate was infiltrated with LFO powder dispersed in a solvent and then the whole structure was

PAGE 73

73 fired. Table 4-2 shows BET results and masses of samples with normalized surface areas (~0.6 m2). Finally, a third sample was made with a 50/50 LFO/YSZ painted electrode with additional LFO infiltrated. Temperature programmed des orption ( TPD ) procedures were conducted as described in section 3.2.1 for NO+O2 and NO2+O2 gas compositions. 4.3 Results and Discussion 4.3.1 TPD Figs. 4 1 -4 3 show the NO, NO2 and O2 desorption after NO+O2 adsorption on the painted LFO on YSZ, painted 50: 50 LFO/YSZ composite painted on YSZ, and painted YSZ on YSZ samples Total NOx desorption increases with LFO content from 1.7 moles/m2 from YSZ to 4.2 moles/m2 from LFO/YSZ, to 10.7 moles/m2 from LFO. Additionally species that decompose and desorb hav e higher binding energies with increasing YSZ content seen by the increase in temperature s for NO and NO2 desorption peaks. As was discussed in section 3.3.1, O2 desorption accompanying NO and NO2 desorption indicates decomposition of NOx surface species. Figs. 4 4 -4 6 show the desorption peaks after NO2+O2 adsorption on the same samples. Once again, total NOx desorption increases with LFO content, but with more desorption than from NO+O2. Total NO and NO2 desorbed for the three samples were 1.9 moles/m2 for YSZ, 5.1 moles/m2 for LFO/YSZ, and 16.3 moles/m2 for LFO. Similar to NO adsorption, higher binding energies of surface species decomposing and desorbing are observed with increasing YSZ content. When LFO is infiltrated into a YSZ scaffold, an decrease in desorption is observed from both NO+O2 and NO2+O2 TPD s (Fig s. 4 7 and 48 ). The LFO infiltrated sample desorbed 3.7 moles/m2 of NO and NO2 compared to 4.2 moles/m2 for the 50:50

PAGE 74

74 composite from the NO+O2 TPD and 4.8 moles/m2 compared to 5.1 moles/m2 from the NO2+O2 TPD. This indicates that the infiltrated sample has less surface exposure of LFO than the 50:50 composite sample Figs. 4-9 and 410 show the LFO powder NO+O2 and NO2+O2 TPDs respectively. Both are seen to desorb much higher quantities of NO and NO2 as expected since the powder has only LFO surfaces for adsorption. When the LFO/YSZ composite was infiltrated with LFO, the catalyst surface coverage increased. This is evident by increased desorption of NO and NO2 from both NO+ O2 and NO2+O2 TPDs as compared to the 50:50 and the infiltrated composites Total desorption was 5.3 moles/m2 from the NO+O2 TPD (Fig. 4 11) and 7.6 moles/m2 from the NO2+O2 TPD (Fig. 4 12). 4.3.2 Characterization Metric for Two Surface Catalyst To separate the contributions of the LFO and YSZ in the 50:50 composite and the infiltrated composite, a char acterization metric was developed. For the 50:50 composite sample, the painted LFO sample was used as the standard for maximum LFO surface coverage while the YSZ painted sample was used as the minimum. Therefore, the % surface exposure (%SE) can be calculated from the measured desorption of total NO and NO2 species according to Eq. 4 1. SE % 100 x Minimum Maximum Minimum Measured (4 -1) From this calculation, it was shown that the %SE for LFO is 28% from NO+O2 TPDs and 22% from NO2 TPDs. Values can be found in Table 4 3. This shows a variance in surface exposure of LFO calculated from the two different experiments, but nonetheless the values are close enough to conclude that calculated values are reasonable.

PAGE 75

75 The same procedure was applied to the infiltrated sample to calc ulate surface exposures of 20 and 13 % for NO+O2 and NO2+O2 respectively. Results are shown in table 4 4. For reasons unknown, when the metric is applied for the NO2 case, the surface exposure is shown to be less for both samples. 4.3.3 Characterization M etric for Three Surface Catalyst A more complicated metric was developed to determine the surface exposure for the LFO/YSZ composite that was infiltrated with LFO. In this case, the areas of the samples as well as the desorbed NO and NO2 per unit area are required. The following relationships (Eqs. 42 -4 4) were solved simultaneously to determine the surface area exposures for infiltrated LFO, composite LFO, and YSZ 3 A 2 A 1 A Atotal (4 -2) 3 x 3 A 2 x 2 A 1 x 1 A Dx (4 -3) 3 x 3 A 2 x 2 A 1 x 1 A Dy (4 -4) Where Atotal is the toal surface area, A1 is the s urface area of infiltrated LFO A2 is the surface area of LFO in the 50:50 composite structure, A3 is the surface area of YSZ, Dx is the total desorbed NO and NO2 from the NO+O2 TPD for the measured s ample Dy is the total desorbed NO and NO2 from the NO2+O2 TPD for the measured sample, x is the total desorbed NO and NO2 per unit area from the NO+O2 TPD of the standard samples and y is the total desorbed NO and NO2 per unit area from the NO2+O2 TPD of the standard samples From the equations it was determined that the surface exposure of the infiltrated LFO was 0. 7 %, that of the composite LFO was 38.8% and that of YSZ was 60. 5 %. The infiltrated surface exposure was lower than expected, but the overal l surface exposure of LFO increased compared to the other two composites. This

PAGE 76

76 demonstrates that mass spectrometry can be used to determine surface exposures of two and three surface composite catalysts. 4.4 Conclusions Different composite sensor structur es were tested for NOx catalytic behavior. Surface species increased with LFO surface exposure for all samples but species were less stable as they desorbed at lower temperatures The infiltrated sample had less surface exposure than the 50:50 composi te because of lower quantities of NO and NO2 desorption in both TPDs. Using mass spectrometry, two and three surface characterization metrics were developed to determine surface exposure of the different phases.

PAGE 77

77 Table 4 1 LFO composite and component sample measurements. Catalyst Specific Surface Area (m 2 /g) MS Sample Mass (mg) MS Sample Surface Area Painted LFO 0.45 1320 0.6 Painted 50/50 LFO/YSZ Composite 0.99 590 0.6 Painted YSZ 1.59 370 0.6 Table 4 2 LFO infiltrated composite and component sam ple measurements. Catalyst Specific Surface Area (m 2 /g) MS Sample Mass (mg) MS Sample Surface Area LFO Powder 5.49 108 0.6 Painted 50/50 LFO/YSZ Composite 1.71 350 0.6 Painted YSZ 1.59 370 0.6 Table 4 3 Values used to calculate % surface exposure of the 50:50 LFO/YSZ composite. TPD Maximum (Painted LFO moles/m 2 ) Minimum (Painted YSZ moles/m 2 ) Measured (Painted LFO/YSZ moles/m 2 ) % Surface Exposure NO+O 2 10.7 4.2 1.7 28 NO 2 +O 2 16.3 5.1 1.9 22 Table 4 4 Values used to calculate % surface exposure of the LFO infiltrated YSZ scaffold composite TPD Maximum (Painted LFO moles/m 2 ) Minimum (Painted YSZ moles/m 2 ) Measured (Painted LFO/YSZ moles/m 2 ) % Surface Exposure NO+O 2 11.9 3.7 1.7 20 NO 2 +O 2 24.9 4.8 1.9 13

PAGE 78

78 Figure 4 1 NO+O2 TPD for the painted LFO sample. Figure 4 2 NO+O2 TPD for the painted LFO/YSZ composite sample.

PAGE 79

79 Figure 4 3 NO+O2 TPD for the painted YSZ composite sample. Figure 4 4 NO2+O2 TPD for the painted LFO sample.

PAGE 80

80 Figure 4 5 NO2+O2 TPD for the painted LFO /YSZ composite sample. Figure 4 6 NO2+O2 TPD for the painted YSZ sample.

PAGE 81

81 Figure 4 7 NO+O2 TPD for the LFO infiltrated YSZ scaffold composite sample. Figure 4 8 NO2+O2 TPD for the LFO infiltrated YSZ scaffold composite sample.

PAGE 82

82 Figure 4 9 NO+O2 TP D for the LFO powder sample. Figure 4 10 NO2+O2 TPD for the LFO powder sample.

PAGE 83

83 Figure 4 11 NO+O2 TPD for the LFO infiltrated LSM/YSZ composite sample. Figure 4 12 NO2 +O2 TPD for the LFO infiltrated LSM/YSZ composite sample.

PAGE 84

84 CHAPTER 5 DETERMI NATION OF SURFACE EXCHANGE COEFFICIENTS USING IN-SITU ISOTHERMAL ISOTOPE EXCHANGE 5 .1 Introduction There is a driving force to reduce operating temperatures in s olid o xide f uel c ells (SOFC) from high temperatures (800-1000 C ) to intermediate and low temper atures (400-800 C ) for the purpose of reducing degradation rates, improving thermal compatibility of materials, and reducing materials costs [73, 74] Lowering the operating temperature results in challenges of maintaining high power densities since the electrochemical processes rely on thermally activated catalytic processes. One of the greatest challenges is improving the performance of the cathode where oxygen reduction occurs and is the pro cess which is considered to contribute most to polarization losses at low temperatures. To improve the catalytic behavior of cathode materials, a better understanding of the oxygen reduction reaction (ORR) is needed. Much effort has been made to understan d the mechanism of the ORR that takes place on the cathode of a SOFC, but understanding is still lacking. The two kinetic parameters commonly used to characterize oxygen exchange in cathode materials are the diffusion coefficient, D (cm2/s), which is a bulk property, and the surface exchange coefficient, k (cm/s), which is a surface property. Much progress has been made determining meaningful D values, however k is difficult to measure from the experimental techniques currently being used. The two techni ques most used to determine these parameters are isotopic exchange depth profiling with secondary ion mass spectrometry (SIMS) [37, 38, 7580] and electrical conductivity relaxation (ECR) [8184] However, the D and k values differ in meaning for each experiment [82, 83, 85]

PAGE 85

85 In a SIMS experiment, dense samples are annealed in 18O2, followed by quenching, and depth profiling of diffused 18O ions. The depth profile is modeled using Cranks [39] mathematical solution to the diffusion equation given the specific experimental conditions and D and k are extracted. In this case, D is a tracer diffusion coefficient, D*, and k is a tracer surface exchange coefficient, k*. In an ECR experiment, a dense sample is subjected to an oxygen partial pressure change and concurrent changes in the electrical conductivity of the material are measured as the sample equilibrates. The resulting profile is modeled using Cranks solution to the diffusion equation, but D and k are chemical diffusion (Dchem) and surface exchange coefficients (kchem). These values are related to the tra cer values by a thermodynamic factor [82, 83, 85] SIMS is limited to ex -situ analysis since samples must be quenched and removed. ECR is limited to non equilibrium exchange during changes in pO2. Both experiments require dense samples that are difficult to fabricate thin, which is the reason extracted k values may not be accurate. To overcome these limitations, a novel approach using isotope exchange called in -situ isothermal isotope exchange (IIE) was used to ch aracterize common cathode materials (La0.6Sr0.4)(Co0.2Fe0.8)O3 (LSCF) and (La0.8Sr0.2)MnO3 (LSM) and common electrolyte materials (Ce0.9Gd0.1)O2 (GDC) and (Y2O3)0.08(Zr O2)0.92 (YSZ). Tracer D* and k* values were extracted and compared to values from SIMS experiments, since the D and k values have the same mean ing. 5 .2 Theoretical Background Though from different experimental approaches, SIMS and ECR samples are processed, and experimental profiles are modeled, in the same manner. For these experiments samples must be dense to prevent rapid diffusion to different parts of the

PAGE 86

86 sample, which would distort the diffusion profile. If the sample is thick, bulk diffusivity dominates and limits the oxygen flux through the sample. If the sample is very thin, surface exchange dominates over the bulk. Thus a characteris tic thickness (Lc) has been defined [86] which provides a reference for whether bulk diffusion or surface exchange limits the ORR [83, 84] The value is the ratio of D to k as seen in Eq. 5 -1. k D Lc (5 -1) For SIMS and ECR experiments, samples are made to have thicknesses as close to Lc a s possible to extract a measurable k. Making a sample with thickness L>>Lc allows for an accurate measure of D. Likewise, a sample with thickness L<
PAGE 87

87 In the case of ECR experiments, Cranks solution for diffusion in a plan e sheet under surface evaporation conditions (Eq. 5 -3) is used to model the accumulation of 18O from the conductivity profile [39] 1 n 2 2 n 2 n 2 2 n 2 t) L L ( ) l / Dt exp( L 2 1 M M (5 -3) Mt is the amount of oxygen entering or leaving the sheet at time t, M is the total amount accumulated when diffusion occurs over an infinite amount of time, l is the thickness of n is a root of Eq. 5 4, where L is a dimensionless variable defined in Eq. 5 -5. L tan n n (5 -4) cL l D k l L (5 -5) When l>>Lc cancel, which eliminates ks influence, leaving D to control the behavior of the solution. For IIE, the same approach can be taken to model diffusion in powder par ticles assumed to be spheres. Using Cranks solution to the diffusion equation for diffusion in a sphere with surface evaporation (Eq. 5 6) [39] it is possible to extract D* and k* values from accumulation isotope exchange profiles according to the following relationships. 1 n 2 n 2 n 2 2 n 2 t)) L L ( L ( ) a / Dt exp( L 6 1 M M (5 -6) 0 1 L cot n n (5 -7) cL a D k a L (5 -8)

PAGE 88

88 All variables have the same meaning as previously defined. The only additional variable is a, the radius of the sphere, instead of l in the case of a plane sheet. Since diffusion occurs in powder particles, a is expected to be on the order of, or lower than Lc, meaning the oxygen exchange rate is surface exchange dominate d and meaningful k* values can be extracted. 5 .3 Experimental LSCF (Praxair) and LSM (Fuel Cell Materials) cathode powders were used in this study. YSZ (Tosoh) and GDC (Anan) were the electrolyte powders chosen for this study. Cathode samples were normal ized by surface area to ~0.1 m2. Specific surface areas measured by BET for LSCF and LSM were 6.7 and 5.7 m2/g respectively, providing for sample sizes of ~15 and ~18 mg. Due to higher specific surface areas of YSZ and GDC (12.59 and 20.80 m2/g measured by BET), normalizing to ~0.1 m2 would mean sample sizes were much too small for testing. Therefore, electrolyte samples were normalized to ~0.3 m2 providing for GDC and YSZ sample sizes of ~14 and ~24 mg respectively. However, since D* and k* describe intrinsic kinetic behavior, values are directly comparable for all samples regardless of mass. Average particle sizes of LSCF, LSM, GDC, and YSZ were measured by laser light scattering using a Coulter LS13320 to be 282, 158, 96, and 86 nm respectively. Deta ils and schematics of the experimental setup were given previously [40] Each powder was placed in a continuous flow quartz reactor tube on a quartz frit in the middle of the tube. The portion upstream of the sample had a 4 mm inner diameter, while that downstream of the sample had a 2 mm inner d iameter to reduce gas residence time. Reynolds number calculations for flow through the packed bed yielded values between 0.1 0.3, depending on the sample, which means gas flow through the

PAGE 89

89 reactor is laminar. Furthermore, the flow rate is fixed at 20 scc m by MKS mass flow controllers, which means the time taken for the gas to traverse the entire sample (~24 mm in height) is ~0.1 0.15 s, which is sufficiently fast compared to the kinetic reaction rates to assume surface reactions on particles do not inter fere with one another. The ideal pO2 chosen for this technique was 0.05 atm since it was high enough that it was close to realistic operating conditions, but not too high as pO2=0.20 atm saturated the portion of the mass spectrometer signal attributed to the kinetic behavior of the material. A flow rate of 20 sccm of 0.05 atm O2 was flown over the sample as the sample was heated to the desired temperature. Once the target temperature was reached, the sample was allowed to equilibrate with the oxygen gas phase, while in a separate line an equal pO2 was established containing 99.8% 18O2 (Cambridge Isotope). Once equilibrated, the lines were changed using a twoway automatic switch and the rise and fall of m/z ratios of 32 (16O2), 34 (16O18O), and 36 (18O2) using an Extrel quadrupole mass spectrometer were monitored. 1000 ppm argon tracer was also included in the isotope gas stream to show the switch was complete. The temperature range tested was from 500850C Fig. 5 1 shows a blank IIE profile, while F ig. 5 2 shows raw IIE profiles of LSCF, LSM, GDC, and YSZ at 800C Since accumulation of 18O in the sample wa s the desired profile to model the signals were processed in the following way. 2 18 18O in O inN 2 N (5 -9) O O out O out O out18 16 2 18 18r r 2 N (5 -10 ) O out O in O net18 18 18N N r (5 -11)

PAGE 90

90 t 0 t O net tdt r M18 (5 -12 ) Where N is the flow rate into and out of the reactor, rout is the rate of evolution of the different species after interacting with the sample, rnet is the rate of incorporation of 18O int o the sample, and Mt is the amount of 18O accumulated at time t. Since a 34 signal is detected, this indicates a scrambled product of 16O18O is being formed on the surface and released. In this case, an 18O atom that had been taken from the gas stream i s given back and therefore must be subtracted from the total accumulation. This process is demonstrated in Fig. 5 3 for LSCF at 800C Once the accum ulation profiles were processed, using a custom made program for linear regression optimization in Matlab and the relationships in Eq.s 5 6 -5 8, the data was modeled to extract D* and k* values. 5 .4 Results IIE profiles for LSCF, LSM, GDC, and YSZ (Fig. 5 -2) show the incorporation behavior of the different materials. A brief drop in pressure is seen by the initial drop in the 32 and total O2 signals upon engagement of the switch. However, the argon tracer rapidly reaches its 1000 ppm threshold without being affected by the pressure drop, which demonstrates that the 18O2 gas remains unaffected as well, and t he initial 34 and 36 signals are from gas/sample interactions. Furthermore, the 34 and 36 signals are used to process an accumulation profile, which means the drop in both 32 and total O2 signals does not interfere with the kinetic behavior of the materials. LSCF (Fig. 5 2a) rapidly incorporates 18O into its lattice evidenced by the increase and subsequent decrease to ~0 ppm of the scrambled product 16O18O. The fact that a scrambled product exists is evidence of surface activity and since the 18O2 signal reaches the baseline concentration while the 16O18O reaches ~0 ppm, this means that LSCF has

PAGE 91

91 fully exchanged its lattice oxygen. LSM (Fig. 5 2b) shows surface activity as well, but in contrast to LSCF the scrambled product 16O18O species continues to evolve at high ppm levels (~1000 ppm) despite the longer anneal time. Furthermore, 18O2 does not reach the baseline concentration within the time tested indicating that the lattice oxygen in LSM is not completely exchanged. The electrolyte materials have hi gh ionic conductivities and will rapidly exchange as seen in Fig. 5 -2. GDC (Fig. 5 -2c) fully exchanges rapidly like LSCF. YSZ (Fig. 5 -2d) almost completely exchanges at 850C, but takes longer with a more prolonged 16O18O signal, similar to LSM. From th ese plots, GDC shows more rapid oxygen exchange than YSZ. Accumulation profiles with model fits to Eqs. 5 6 -5 8 for LSCF, LSM, GDC, and YSZ are seen in Fig. 5 -4 The accumulation profiles for LSCF (Fig. 5 -4a) are almost identical over the temperature rang e tested (500 -800C ) and have goodness of fits greater than 99%. Complete conversion of 18O with lattice 16O was observed at all temperatures. On the other hand, LSM (Fig. 5 4b) in a much more limited temperature range (775 850C ), showed significant diff erences in its ability to exchange the 18O with lattice 16O. Furthermore LSM never reaches complete conversion despite the higher temperatures and longer anneal times. Even though LSM was annealed in a lower pO2, this does not play a factor in the lack o f exchange. This was confirmed by the fact that saturation of the kinetic behavior occurred at higher pO2s, so it is not a matter of gas concentration limitations. Goodness of fits greater than 99% were achieved, but only at temperatures greater than 775 C Likewise, when comparing the fluorite electrolyte materials, a similar comparison to the perovskites is made. GDC accumulation profiles (Fig. 5 4c) are almost identical.

PAGE 92

92 All have greater than 99% goodness of fit and show full exchange within the temp eratures tested (600850C ). YSZ accumulation profiles (Fig. 5 4d), in a more limited temperature range (700-850 C ), show differences in 18O conversion fraction for the different temperatures. All curves have a goodness of fit of greater than 97%. Close to complete conversion is only observed for the sample at 850C Each sample going down in temperature exhibits less conversion and slower incorporation as evidenced by the change in slope of the initial part of the curve. Overall, even at 850C the in corporation is slower than GDC as the slope of GDC is much steeper. 5 .5 Discussion 5 .5.1 Accumulation Profiles When analyzing surface exchange behavior, it is important to consider two species: oxygen vacancies and electrons. The ORR takes places according to Eq. 5 1 3 x O O 2O 2 e 4 V 2 O (5 -1 3 ) Therefore, in order for the complete ORR (adsorption, dissociation, charge transfer, incorporation) to take place, oxygen vacancies and electrons must be present. In an operating fuel cell, electrons are pro vided from oxidation of a fuel at the anode by an external circuit. However, the surface reaction at the cathode is accelerated when electrons are present in the material as well. This concept is observed in the accumulation profiles of the different typ es of conductors: MIEC conductors LSCF and GDC, electronic conductor LSM, and ionic conductor YSZ. From the accumulation profiles, it is seen that those for LSCF and GDC are independent of temperature. They both rapidly and completely exchange oxygen. As

PAGE 93

93 MIECs, both these materials have electrons and oxygen vacancies present, which facilitate the exchange. The other two materials, LSM and YSZ, are considered pure electronic and pure ionic conductors respectively. Both of these materials do not exchange rapidly nor completely. In the case of LSM, the material lacks oxygen vacancies needed for exchange, while in the case of YSZ, the material lacks electrons to facilitate exchange. This confirms that both species are needed to facilitate rapid exchange k inetics. 5 .5.2 Model It was found that the model was insensitive to D* for all samples except LSM, but very sensitive to k* for all samples, confirming that experimental conditions wer e such that the sample length (particle size ) was below Lc, and oxygen exchange was dominated by the surface reaction. In the case of LSM, the model was sensitive to D* and lower values were determined than previously reported. For the other three samples, a wide range of D* values spanning over several orders of magnitude d ifference provided equally good fits while k* changed less than 5%. Since these experiments were not sensitive to D*, values reported from SIMS studies which are sensitive to D*, were used as initial guesses for optimization. Therefore, extracted D* val ues are consistent with past results. 5 .5.3 Diffusion Coefficients The D* values for the perovskite and fluorite materials extracted from the model and literature [75, 79, 87] are shown in Figs. 5 5 and 5 6. Since the literature values were used as initial guesses in the model, it is no surprise that they agree, having similar activation energies. LSCF has higher ionic conductivity than LSM, which is why D* is much higher as well. Since the model was sensitive to D* for LSM, these values

PAGE 94

94 are unique which is why they are slightly different than those reported by De Souza and Kilner [75] GDC and YSZ in comparison to LSCF and LSM have higher D* values and lower activation energies than LSCF and LSM. This is due to the fact that diffusivity of oxygen in the bulk of the fluorite structure is much easier than in the perovskite structure. Between the fluorites, YSZ has higher D* values at temperatures above ~650 C but below ~650C GDC maintains higher D* values due to its lower activation energy (75 kJ/mol compared to 11 3 kJ/mol). Therefore, GDC is a more suitable electrolyte for lower and intermediate temperature operation. 5 .5.4 Surface Exchange Coefficients k* values were extracted and are shown verses temperature in Figs. 5 -7 and 5 -9 When in a surface exchanged controlled regime for LSCF, the activation energy (3 kJ/mol) is much lower than the previously reported value by Benson (180 kJ/mol) [87] If it is assumed that LSCF behaves like its parent material, LaFeO3 (LFO), Chan Woo et. al. [88] showed from ab initio calcul ations that within the pO2 and temperature ranges of this study, the surface of LFO maintains a constant structure. If the surface remains unchanged over the temperature range, it is reasonable that the surface activity would also remain the same. This explains the low activation energy behavior and has positive implications for low and intermediate temperature operation. In the case of LSM, what was previously seen from De Souza et. al [75] was an activation energy of 101 kJ/mol. However, from IIE, a positive slope with an apparent negative activation energy is obser ved. This is not the first time such behavior has been observed. Using ECR, Ganeshananthan et. al. [84] tested both dense and porous La0.6Sr0.4CoO3 samples. They found that the porous sample had an apparent negative

PAGE 95

95 activation energy while the dense sample exhibited thermally activated behavior. One possible explanation that was given was an existing relationship between k* and surface coverage. Meaning, as temperature is decreased, more coverage was expected with oxygen ions being formed on the surface and not incorporating into the bulk. A different interpretation that better describes the kinetic behavior with temperature is a precursor medi ated mechanism that involves a reversible first step of molecular oxygen adsorption followed by a dissociation step according to Eq. 5-14 [89] ) ads ( O 2 ) ads ( O ) g ( O2 2 (5 -14) By this mechanism, the kinetic rate of molecular oxygen in the gas phase (O2(g)) is equal to the flux of molecules at the surface ( ) times the probability for trapping of the oxygen molecule into an adsorbed state ( ). Once trapped in the adsorbed state, the molecule can desorb back into the gas phase or dissociate on the surface. The rate s of desorption (rd) and dissociation (rr) are considered elementary first order reactions governed by the rate coefficients for desorption (kd) and dissociation (kr) shown in Eqs. 5 -15 and 516, where [O2(ads)] is the concentration of adsorbed species on the surface. ) ads ( O k r2 d d (5 -1 5) ) ads ( O k r2 r r (5 -16) Applying the steady state approximation, which is valid in the limit of zero coverage of O2(ads) (z ero coverage is a reasonable assumption given the high temperatures of the experiments ), the overall rate of the reaction is described according to Eq. 5-17 where S0 is the probability of molecular dissociative adsorption. ) k k k ( S rd r r 0 (5 -17)

PAGE 96

96 Therefore, the reaction rate is the probability of gas phase oxygen adsorbing on the surface multiplied by the branching probability for dissociation of the adsorbed molecule The branching probability is the effective k for the dissociative adsorption reaction, which is k* from the experimental data. If the activation energy for dissociation of the adsorbed molecule (Er) is less than the activation energy for desorption (Ed), the general rate equation (517) shows the effective k will decrease with increasing temperature, which is consistent with k* observed for LSM. When kd/kr>>1, a simple Arrhenius relationship exists according to Eq. 5-18. RT ) E E ( exp k k kd r d r d r (5 -18) In this equation, r and d are preexponential factors, R is the universal gas constant, and T is the temperature. From this expression, it is clear that if ErEd where an energy barrier for dissociation exists (EA). This explains the increase in k* with increasing temperature. A diagram showing the energy barriers for both cases is seen in Fig. 58 The idea that LSM could have a more active surface for oxygen reduction than LSCF at low temperature is unexpected, but not unreasonable. Since experiments used to extract D* and k* were in diffusion controlled regimes, and LSM has slower diffusion behavior than LSCF, the surface behavior was masked by bulk diffusion. It is important to note t hat even if LSM shows a higher rate of surface exchange at low temperature that does not necessarily make it a better SOFC cathode for low and

PAGE 97

97 intermediate temperature operation. The overall ORR involves incorporation and bulk transport as well. When com paring LSCF and LSM, LSCF has much higher D* values, which in conjunction with its k* values, make it more suitable for intermediate temperature use. However, if the surface exchange behavior of LSM could be incorporated into the electrode structure, a superior composite could be fabricated. When in the surface exchanged controlled regime, GDC and YSZ both have lower k activation energies compared to literature (Fig. 5 9 ). Whereas previous results from Manning et. al. [79] showed large discrepancies in activation energies be tween the two fluorite materials, IIE shows closer activation energies of 3 and 38 kJ/mol for GDC and YSZ respectively. Since they are of the same type of material, this is expected. The lower activation energy of GDC and higher k* values compared to tha t of YSZ can be explained by the higher electron concentration in GDC [90 92] Once again comparing the different types of conductors, the MIECs LSCF and GDC both have low activation energy for surface exchange and higher k* values than the other two pure conductors. Now it is seen that the combination of electrons and oxygen vacancies is important for the surface exchange aspect of the ORR. In contrast, the two pure conductors have opposite behaviors with temperat ure. While k* of YSZ, the ionic conductor, increases with temperature, that of the electronic conductor (LSM) decreases with temperature. Based on the kinetic model described previously, Er>Ed for YSZ and an energy barrier exists for dissociation, while Ed>Er for LSM and there is no energy barrier for dissociative adsorption 5 .5.5 Characteristic Thicknesses Lc values were calculated from measured D* and k* values and compared to those calculated from literature values (Figs. 5 10 and 5 -1 1 ). LSCF has slightly higher Lc

PAGE 98

98 values than those computed from Bensons D* and k* values [87] and LSM has lower values than computed from De Souzas reported D* and k* values [75] SIMS samples typically have thicknesses of ~0.5 -3 mm. As seen in Fig. 5 9, SIMS samples would only be above or around Lc for LSCF at high temperatures. LSCF and LSM samples tested with IIE are ~150280 nm well below the thicknesses of SIMS samples. Comparing the IIE sample thicknesses to Lc, LSCF samples have thicknesses below Lc for all tem peratures tested. IIE samples may have thicknesses above Lc for LSM, but it would still be more sensitive to surface exchange behavior than SIMS. From the Lc values, in order to have a sample below Lc at 850C for LSM, the thickness would have to be on t he order of or less than 5 nm. Because of this, isotope exchange by SIMS or IIE may not be the best method to determine k values. However, it is possible that LSM could be more active for surface exchange than LSCF at low to intermediate temperatures, but was unable to be confirmed below 775 C for the reasons mentioned earlier. Lc values for GDC and YSZ are similar (Fig. 5 1 1 ). YSZ has a larger Lc than GDC between 700850C due in large part to the higher D* of YSZ in the same temperature range. SIMS sa mples are below Lc for YSZ and below Lc for GDC at temperatures 600C and higher. Even so, IIE thicknesses are much lower, falling well below Lc, ensuring the technique is more sensitive to k*. Therefore IIE is a more suitable technique to extract k* and the values reported here can be considered closer to accurate values. 5 .6 Conclusions IIE was shown to be more sensitive than SIMS or ECR to measure the kinetic surface behavior of catalytic materials, particularly fast ion conductors. k* values of cathod e materials LSCF and LSM and electrolyte materials GDC and YSZ were

PAGE 99

99 extracted due to the ability to test in-situ porous powder samples. D*, k* and Lc values from IIE were used to compare cathode and electrolyte materials. Results show that the activation energy for surface exchange on LSCF is very low and k* varies little over the temperatures tested, while LSM exhibits an apparent negative activation energy though it was only able to be tested above 775 C due to slow oxygen incorporation behavior. Like wise, k* values of GDC vary little over temperature and are greater than YSZ, which has a higher activation energy. This variance is attributed to the higher electron concentration in GDC compared to YSZ. The MIEC materials LSCF and GDC exhibit similar a ccumulation behavior and surface exchange behavior because of the presence of both electrons and oxygen vacancies. In contrast, the pure electronic (LSM) and ionic (YSZ) materials had similar accumulation behavior, but opposite surface exchange behavior. A precursor mediated mechanism was proposed to explain the opposite temperature dependence of LSM Furthermore IIE was shown to be sensitive to k* with sample thicknesses below Lc for LSCF, GDC, and YSZ, and close to LSM.

PAGE 100

100 Fig ure 5 1 Blank IIE profile showing 32 (16O2), 34 (16O18O), 36 (18O2), 40 (Ar) and total O2 MS signals. Figure 5 2. IIE of (a) LSCF, (b) LSM, (c) GDC, and (d) YSZ at 800C. LSCF, YSZ and GDC were exchanged in 0.05 atm O2 and LSM was exchanged in 0.01 atm O2.

PAGE 101

101 Figure 52. Continued.

PAGE 102

102 Fig ure 5 2 Continued. Fig ure 5 3 Accumulation profile processing method demonstrated for LSCF at 800 C in 0.05 atm O2.

PAGE 103

103 Figure 54. Fraction of 18O exchanged with lattice 16O (open symbols) with model fits (lines) for (a) LSCF, (b) LSM, (c) GDC, and (d) YSZ.

PAGE 104

104 Fig ure 5 4 Continued.

PAGE 105

105 Fig ure 5 5 Measured diffusion coefficients (solid symbols) of LSCF and LSM compared to literature values (open symbols) [75, 87] Fig ure 5 6 Measured diffusion coefficients (solid symbols) of GDC and YSZ compared to literature values (open symbols) [79]

PAGE 106

106 Fig ure 5 7 Measured surf ace exchange coefficients (solid symbols) of LSCF and LSM compared to literature values (open symbols) [75, 87] Figure 58. Energy diagram for a precursor mediated dissociative adsorption reaction.

PAGE 107

107 Fig ure 5 9 Measured surface exchange coefficients (solid symbols) of GDC and YSZ compared to literature values (open symbols) [79] Fig ure 5 10 Calculated characteristic lengths (solid symbols) of LSCF and LSM compared to literature values (open symbols) [75, 87] with typical SIMS and IIE sample thicknesses.

PAGE 108

108 Fig ure 5 1 1 Calculated characteristic lengths (solid symbols) of GDC and YSZ compared to literature values (open symbols) [79] with typical SIMS and IIE sample th icknesses.

PAGE 109

109 CHAPTER 6 DETERMINATION OF SURFACE EXCHANGE COEFFICIENTS OF COMPOSITE CATHODES USING IN-SITU ISOTHERMAL ISOTOPE EXCHANGE 6 .1 Introduction Solid Oxide Fuel Cells (SOFC) are solid state ionic devices that efficiently convert chemical energy into electrical energy. Catalytic and ionic electrochemical process are activated, which require high operating temperatures, between 800-1000 C with common cathode and electrolyte materials (La0.8Sr0.2)MnO3 (LSM) and (Y2O3)0.08(ZrO2)0.92 (YSZ) for efficient operation. When cells operate at these temperatures, there are issues of material compatibility (thermal expansion mismatches) and instability, requiring expensive and complicated interconnect and sealing materials. Therefore, there is a need to lower operating temperatures to overcome these issue s. Mixed ionic and electronic conductors (MIEC), such as (La0.6Sr0.4)(Co0.2Fe0.8)O3 (LSCF), have been used instead of LSM as the cathode to operate at lower temperatures. However, LSCF is not compatible with YS Z as an electrolyte due to the formation of an insulating La2Zr2O7 phase [93] In addition, to take advantage of LSCFs MIEC properties, a fast ion conductor (102 S/cm) is desirable to operate at lower temperatures so that once oxygen reduction and incorporation is complete, oxygen ions will quickly diffuse to the anode [94] The compatible elec trolyte used with LSCF is (Ce0.9Gd0.1)O2 (GDC), which has higher conductivity than YSZ at low to intermediate temperatures. The LSCF/GDC cathode/electrolyte combination allows for lower operating temperatures, but not low enough for economically feasibl e operation since ionic conductivity in LSCF decreases as temperature decreases [95] One solution is to combine cathode and electrolyte materials to form composites of interspersed ionic and

PAGE 110

110 electronic conductive phases in a way that will improve the oxygen reduction reaction (ORR), where the complete ORR involves adsorption and dissociation of oxygen molecules, charge transfer and incorporation of oxygen ions into the bulk lattice. Many electrochemical impedance spectroscopy (EIS) studies have been conducted on LSCF/GDC [18, 96, 97] LSM/YSZ [98, 99] and LSM/GDC [18, 22, 100, 101] composite cathodes to compare ORR rates. All of these composites exhibited lower polarization resistances than the pure cathode materials, demonstrating that these composite combinations provide improvement for the ORR. Though improvements have been made, the surface reaction is still not well understood and further understanding would aide in optimization. Therefor e these three composites were selected for this study to characterize their surface activity contribution to the ORR. The kinetic parameter that is used to quantify the surface reactivity is commonly referred to as the surface exchange coefficient (k), whi le the kinetic parameter used to characterize bulk diffusion is the diffusion coefficient (D). It is common to measure these quantities from diffusion profiles using isotope exchange depth profiling with secondary ion mass spectrometry (SIMS) [12 -19] or electrical conductivity relaxation (ECR) [8184] In a SIMS experiment, a dense sample is equilibrated in a p16O2 atmosphere. The atmosphere is then replaced by the same p18O2 composed of isotopical ly labeled oxygen. The diffusion profile is then fit to Cranks solution to the diffusion equation for a semi -infinite medium [39] The diffusion equation contains D and k, whic h can be extracted. These coefficients are considered to be tracer values which are labeled D* and k*. ECR is similar except the pO2 of the sample atmosphere is changed and the resulting conductivity profile is monitored. This profile is fit to

PAGE 111

111 Cranks solution to the diffusion equation for a plane sheet [39] and D and k can be extracted. In this case, the kinetic coefficients are chemical coefficients labeled Dchem and kchem and are related to D* and k* by a thermodynamic factor [82, 83, 85] Both of these experiments require dense samples, usually ~0.53 mm thick, which makes it difficult to isolate the surface behavior. Each materi al has a characteristic thickness (Lc) [86] in which samples with thicknesses greater than Lc are diffusion (bulk) controlled and samples with thicknesses less than Lc are surface exchange controlled, where Lc is the ratio of D/k. In most c ases, samples used in SIMS and ECR experiments have thicknesses that are equal to or greater than these values and the results are accurate for D, but not k. More recently, a novel approach utilizing isotope exchange called isothermal isotope exchange (IIE ) [102] was developed to isolate the surface behavior. IIE utilizes powder materials, which allow the use of materials with relative thicknesses (particle sizes) on the order of nanometers. Therefore it is possible to characterize materials with thicknesses equal to or less than Lc, ensuring surface exchange behavior is isolated. 6.2 Theoretical Background A two step mechanism consisting of dissociative adsorption followed by incorporation into the solid was used to mod el isotope exchange behavior of LSCF and LSM [40, 103] It was determined that in the temperature range from 600-800 C LSCF was limited by dissociative adsorption. In the same temperature range, LSM was shown to be limited by incorporation due to slow bulk diffusion. The two step mechanism demonstrates the differences in the pure electronic conductor (LSM) and

PAGE 112

112 the MIEC (LSCF) for the overall ORR, but does not specifically compare the differences in surface activi ties of these materials. It was shown in chapter 5 that unlike LSCF, k* for LSM increa sed with decreasing temperature, exhibit ing an apparent negative activation energy consistent with a precursor mediated mechanism The results were unexpected and inconc lusive since the sample thickness was shown to be above Lc for LSM. Nonetheless, if this behavior was valid it would be significant for composite cathode optimization. One of the ways to confirm this behavior is to test composites that have LSM and a fast ion conductor (YSZ or GDC) as the two phases. This would increase Lc and ensure surface exchange controlled behavior for experiments. If LSM does in fact have increased surface exchange behavior with decreasing temperature, the composite would show a compromise between the LSM and the YSZ or GDC phases as the electrolyte materials exhibit activated behavior. 6 3 Experimental LSCF/GDC, LSM/YSZ, and LSM/GDC composites were provided by Fuel Cell Materials. 50 weight % (La0.6Sr0.4)(Co0.2Fe0.8)O3 and 50 weight % (Ce0.9Gd0.1)O0.95 were calcined and ball milled to form powder particles with a specific surface area of 5.7 m2/g. The same procedure was also done for (La0.8Sr0.2)MnO3 and (Y2O3)0.08(ZrO2)0.92 and (La0.8Sr0.2)MnO3 and (Ce0.9Gd0.1)O0.95 that resulted in specific surface areas of 6.1 and 5.0 m2/g respectively. All specific surface areas were measured by BET. Samples were normalized by surface area to ~0.1 m2, resulting in samples masses of ~15, ~16, and ~20 mg for LSCF/GDC, LSM/YSZ, and LSM/G DC respectively. Particle sizes were measured to be 300, 280, and 325 nm, respectively, by laser light scattering using a Coulter LS13320.

PAGE 113

113 Details of the experimental setup have been given previously [40] Powder samples were loaded into a quartz continuous flow reactor with a 4 mm inner diameter upstream of the sample and a 2 mm inner diameter downstream of the sample to reduce gas residence time. In the center, the sample was placed on a porous quartz frit to allow for rapid gas permeation. The quartz reactor was placed into a custom made tube furnace that allowed for controlled heating up to 850 C Samples were heated to the temperature to be tested and allowed to equilibrate in 20 sccm of an oxygen balanced by helium mixture consisting of a p16O2 of 0.05 atm. While the sample equilibrated, an equal flow rate and p18O2 of isotopically labeled oxygen (99.8% purity by Cambridge Isotopes) was mixed in a separate line. Once equilibrated, an automatic switch was triggered and the rise and fall of m/z signals for 32 (16O2), 34 (16O18O), and 36 (18O2) were monitored using an Extrel quadrupole mass spectrometer. A 1000 ppm argon tracer was included in the isotopically labeled oxygen mixture to show the switch was complete. Diffusion profiles were processed according to previously reported procedures [102] and were fit to Cranks solution to the diffusion equation for a sphere in surface evaporation conditions [39] according to Eqs. 6 1 -6 3 1 n 2 n 2 n 2 2 n 2 t)) L L ( L ( ) a / Dt exp( L 6 1 M M (6 -1) 0 1 L cot n n (6 -2) cL a D k a L (6 -3) Mt is the total amount of 18O accumulated in the sample, M is the total amount of accumulation as time goes to

PAGE 114

114 n is the root to Eq. 2, and t is time. From these equations, using a custom made Matlab program for linear regression analysis, k* and D* were extracted. 6 4 Results IIE profiles for all samples at 800 C in 0.05 atm O2 are shown in Fig. 6 -1. A brief drop in pressure is seen by the initial drop in the 32 and total O2 signals upon en gagement of the switch. However, the argon tracer rapidly reaches its 1000 ppm threshold without being affected by the pressure drop, which demonstrates that the 18O2 gas remains unaffected as well, and the initial 34 and 36 signals are from gas/sample in teractions. Furthermore, the 34 and 36 signals are used to process accumulation profiles, which means the drop in both 32 and total O2 signals does not interfere with the kinetic behavior of the materials. LSCF/GDC (Fig. 6 -1a) shows complete conversion w ithin 15 minutes of exchange as the 18O2 signal reaches the baseline concentration while the scrambled product 16O18O falls to ~0 ppm. LSM/YSZ (Fig. 6 -1b) and LSM/GDC (Fig. 6 -1c) have exchange profiles very similar to LSCF/GDC. The main difference is nei ther completely exchanges within the 15 minutes of testing. Curves showing the accumulation of 18O into the lattice with time were processed from the raw IIE data according to the method described in section 5.3. The accumulation curves with model fits fo r LSCF/GDC, LSM/YSZ, and LSM/GDC composites are seen in Fig. 62. All modeled curves fit the data with 97% or greater goodness of fit. The accumulation profiles for LSCF/GDC composites are identical with the exception of 500 C Accumulation profiles of runs above 500 C all show complete conversion. This is consistent with the single phase LSCF and GDC profiles previously shown to be independent of temperature within the same range [102] Due to slow

PAGE 115

115 incorporation kin etics of LSM, both LSM/YSZ and LSM/GDC composites could only be fit to the model in the temperature range between 700850C Both samples have similar profiles at 700 and 750C but show increases in accumulation at 800 and 850C At 850 C the sample alm ost completely exchanges which is similar to the YSZ samples at 850C [102] Accumulation profiles were processed and the model was fit in the same way for all samples and temperatures tested to extract D* and k*. 6 .5 D iscussion Composite samples in this study are considered to behave macroscopically with one D* and one k* value for each at a given temperature and pO2. Since the samples are nanopowders, it is expected that all would be below Lc due to their small partic le sizes. The model confirms this behavior since it is sensitive to selecting k*, which changes less than 5% with equally good fits spanning several orders of magnitude of D* values. The model fit all samples well, but samples containing LSM were unable to be tested below 700C due to slow incorporation effects. LSM behavior was described in chapter 5 6 5 .1 Diffusion Coefficients Since the powders are small, the samples are not sensitive to D*. Therefore values from experiments that are sensitive to D* such as SIMS experiments, were used as initial guesses for the model. Fig. 6 -3 shows D* values of LSCF and GDC from results in chapter 5 and D* values for the LSCF/GDC composite compared with Esquirol et. al.s [76] exper imental values for LSCF/GDC (70 wt% LSCF and 30 wt% GDC in the form of a dense bar) from SIMS experiments are shown to be consistent. Likewise, LSCF and GDC are consistent with SIMS results from Benson [87] and Manning et. al. [79] respectively. GDC has higher D* values than LSCF and the composites D*s and

PAGE 116

116 activation energy are a compromise between the two. The LSCF/GDC composite was shown to have higher overall conductivity than pure LSCF from oxygen permeation experiments [18] The high ionic conductivity of GDC interspersed with LSCF results in increases in D* values and a decrease in activ ation energy (EA) relative to LSCF. Within the temperatures tested, only one literature value for D* was found for LSM/YSZ (40wt% LSM and 60 wt% YSZ in the form of a dense bar tested with SIMS) from Ji et. al. [77] Therefore, the other initial guesses had to be interpolated. It is assumed that the relative difference in the literature data point between the D* values of its component materials will remain the same for the other data points, as the LSCF/GDC D* values did. The other D* values were determined using this method. Once again, the composite D* values for LSM/YSZ lie between the pure materials of YSZ [79] and LSM [75] seen in Fig. 6-4. EIS has been used to show the polarization resistance of LSM/YSZ is less than LSM [99] Part of the reason for that is the increase in oxy gen conductivity as demonstrated by the increase in D*. There were no reported D* values at the time of writing for LSM/GDC, so all values were interpolated in the same manner as the LSM/YSZ. These values however did not have a data point to reference, so following the trends of LSCF/GDC and LSM/YSZ it was assumed that the D* values lay between the component materials LSM and GDC with similar magnitudes to LSM/YSZ. The results are shown in Fig. 6 -5. In fact this behavior is also confirmed in literature f rom both oxygen permeation experiments and EIS experiments. Kharton et. al. [18] showed LSM/GDC (50 wt% (La0.7Sr0.3)MnO3 and 50 wt% GDC in the form of a dense disk) has a higher ionic conductivity than LSM, which was expected since ionic conductivity in LSM is negligible. Murray et. al. [100]

PAGE 117

117 showed that LSM/GDC (50 wt% LSM and 50 wt% GDC in the form of a porous electrode) not only had lower polarization resistance than LSM, but that it also had half the polarization resistance of LSM/YSZ. This behavior cannot be interpreted from the D* data alone, but should be considered along with the surface behavior. 6 5 .2 Surface Exchange Coefficients The k* results for LSCF/GDC, seen in Fig. 66, are lower values than previously reported. These values are also close to GDC k* values showing a slight improvement when combining LSCF and GDC at high temperatures. Following the trend of activation, the biggest advantage comes at high temperatures above 700C where ionic conduction is fast fo r LSCF. Activation energies for these components are very low, indicating the surface behavior does not change much over the temperature range from 500-850 C It was reported in chapter 5 that k* for LSM might increase with decreasing temperature according to a precursor mediated mechanism However, there was uncertainty in these findings since the LSM particle size was above Lc. However, combining LSM and YSZ causes an increase in Lc for the composite material, which means composite samples can be used to confirm and validate the LSM behavior. It is seen in Fig. 67 that k* for LSM/YSZ also has an apparent negative activation energy since k* increases with decreasing temperature. k* for the LSM/YSZ composite, like D*, is a compromise between activation behaviors of both components LSM and YSZ as seen in the slopes of the linear fits. However, unlike D*, k* of the composite is shown to be higher in magnitude than either of the pure components. The combination of the high ionically conductive YSZ and h igh electronically conductive LSM phases create a more conducive environment for oxygen surface exchange than either material

PAGE 118

118 separately, and furthermore the composite maintains the apparent negative activation energy of LSM. L SM/GDC also exhibits an appar ent negative activation energy (Fig. 6 -8), combining the activation behaviors of LSM and GDC. In this case, LSM/GDC exceeds surface exchange of GDC as temperature decreases. Since GDC is also a high ionically conductive material, like YSZ, the combinatio n with the high electronically conductive LSM also creates favorable surface exchange behavior. Both composite behaviors make sense because of the need for oxygen vacancies, primarily provided by the electrolyte material, and surface O2 dissociation sites primarily provided by LSM, for the ORR to occur. 6 5 .3 Characteristic Thicknesses The major benefit of IIE is the ability to use powder samples, which means the effective thickness of the sample is the particle size. SIMS and ECR experiments require dense samples, which are difficult to fabricate thin. As a result, sample thicknesses are typically between ~0.5-3 mm. In contrast, particle sizes of powders used in these experiments are between 280-325 nm, three orders of magnitude lower in thickness. D* values computed from initial guesses using SIMS and interpolated values and extracted k* values were used to determine Lc, where Lc=D*/k*, for the composite samples along with previously reported pure cathode and electrolyte samples. Fig. 69 shows larger Lc values for LSCF/GDC than from values calculated from Esquirol et. al. [76] In Fig. 6 -9, the thicknesses of SIMS and IIE samples are compared to Lc. Where SIMS may be sensitive to k* at high temperature, when sample th icknesses are below Lc, IIE is sensitive to k* for all temperatures tested. Therefore more accurate

PAGE 119

119 determination of k* is ensured. Likewise, the same difference can be seen for LSM/YSZ in Fig. 6 -10 and LSM/GDC in Fig. 6 -11, however, in the case of these composites, SIMS sample sizes would be above Lc in every case except 850 C for LSM/YSZ. Additionally it is shown that the IIE sample thicknesses are below Lc for LSM/YSZ. This is significant to confirm the previously observed LSM behavior, despite LSM h aving thicknesses above Lc. Since the composite still exhibits an apparent negative activation energy this validates the LSM behavior previously observed. The difference in sample thicknesses between SIMS and IIE samples and the fact that IIE sample thi cknesses are below Lc for all composite samples proves the reliability of k* values from IIE. LSCF/GDC had the highest Lc values, which is expected since it is composed of two MIEC materials. Among the LSM composites, LSM/YSZ and LSM/GDC have very similar Lc values. This is due to the two materials have similar D* and k* values. 6 5 .4 Intermediate Temperature Cathode Analysis When comparing k* for the three composites, LSCF/GDC has the highest values above ~825C and LSM/YSZ has the highest values below ~ 825C Just considering the LSM composites, LSM/YSZ has the higher k* values at all tested, but trends indicate a shift to LSM/GDC at lower temperatures. Previous results from Murray et. al. [100] reported LSM/GDC exhibited lower polarization resistances than LSM/YSZ (50 vol% LSM and 50 vol% YSZ) in the temperature range 600-750 C This indicates that in the temper atures where D* for LSM/YSZ is greater than LSM/GDC, the higher k* of LSM/GDC is more significant in reducing polarization losses associated with the ORR. Murray et. al. [97] in another study showed LSCF/GDC (50 wt% LSCF and 50 wt% GDC) to have even lower polarization resistance compared to LSM/GDC in the temperature range 5 00750C In this case, Even though both LSM/GDC and LSM/YSZ

PAGE 120

120 have higher k* values than LSCF/GDC below 825C the difference is much less than the difference in D* values, in which those of LSCF/GDC are much greater. Therefore it is clear that D* and k* are both important in reducing polarization losses and one is not necessarily more important than the other, but rather a balance of the two is required. Not only are high D* and k* materials required, but the way in which the materials are combined is i mportant as well. To illustrate this concept, Jiang et. al. [101] reported on GDC impregnated LSM composite cathodes and found that impregnated composites had even lower polarization resistances than Murray et. al.s LSM/GDC composite. In this case, it was s hown that by adding nanosized (100-200 nm) interdispersed high D* GDC particles into a porous LSM scaffold, performance was improved from the 50:50 weight% LSM/GDC composite. The GDC particles were determined not to be in a continuous network, but still f illed the pores of the LSM scaffold. The decrease in polarization resistance was attributed to the loading of GDC particles, known for their oxygen storage, conduction, and surface exchange abilities, at the TPB sites which facilitated the ORR. One result from this work that may not be explained from EIS polarization resistance results is the apparent negative activation energy of k* for the LSM materials. This implies that there is no energy barrier for dissociative adsorption of oxygen molecules on the surface If this is true, a structure similar to Jiang et. al.s except opposite (i.e. nanosized impregnated LSM into a GDC scaffold), could combine the high k* LSM behavior with the high D* GDC behavior to provide further improvement at low and intermedi ate temperatures If LSM particles are small enough, the surface properties will be isolated such that the lack of ionic conductivity in LSM would be

PAGE 121

121 irrelevant as the high ionic conductivity of GDC would contribute wholly to the incorporation and transport of oxygen. 6 6 Conclusion IIE has been demonstrated as a useful technique to determine accurate k* values, which is difficult to do with SIMS and ECR experiments. LSCF/GDC, LSM/YSZ, and LSM/GDC were tested and compared to EIS polarization results from literature. It is clear that a balance between high D* and k* behavior is required to improve the ORR on the cathode. The much higher D* values of LSCF/GDC compared to LSM/YSZ and LSM/GDC overshadow the lower k* values it exhibits below 825 C This means that simply combing the highest D* and k* materials will not always produce the highest performance, but rather intelligent microstuctural design of the materials must be employed. The apparent negative activation energy of k* for LSM compared to the a ctivated behavior for LSCF at lo w and intermediate temperatures could be taken advantage of if particles were small enough such that ionic conductivity in that phase would be irrelevant, but the surface reactivity was solely utilized. The small particles wo uld also increase the triple phase boundary sites where the cathode, electrolyte, and oxygen molecules meet, and as a result, the high D* of GDC could be effectively utilized for incorporation and conduction. It is expected that combining these materials in this manner would improve the ORR at low and intermediate temperatures and reduce polarization resistances associated with the cathode.

PAGE 122

122 Figure 61. IIE of (a) LSCF/GDC, (b) LSM/YSZ, and (c) LSM/GDC, at 800C in 0.05 atm O2.

PAGE 123

123 Figure 61. Continued Figure 62. Fraction of 18O exchanged with lattice 16O (open symbols) with model fits (lines) for (a) LSCF /GDC, (b) LSM/YSZ, and (c) LSM/GDC.

PAGE 124

124 Fig ure 6 2 Continued.

PAGE 125

125 Figure 6 3 Measured diffusion coefficients (solid symbols) of LSCF/GDC, LSCF, and GDC compared to literature values (open symbols) [76, 79, 87] Figure 6 4 Measured diffusion coefficients (solid Symbols) of LSM/YSZ, LSM, and YSZ compared to literature values (open symbols) [75, 77, 79]

PAGE 126

126 Figure 6 5 Measured diffusion coefficients (solid symbols) of LSM/GDC, LSM, and GDC compared to literature values for LSM and GDC (open symbols) [75, 77, 79] Figure 6 6 Measured surface exchange coefficients (solid symbols) of LSCF/GDC, LSCF and GDC compared to literature values for LSCF/GDC (open symbols) [76]

PAGE 127

127 Figure 6 7 Measured surface exchange coeffic ients (solid symbols) of LSM/YSZ, LSM and YSZ compared to literature values for LSM/YSZ (open symbol) [77] Figure 6 8 Measured surface exchange coefficients of LSM/GDC, LSM and GDC.

PAGE 128

128 Figure 6 9 Calculated characteristic lengths (solid symbols) of LSCF/GDC, LSCF, and GDC compared to literature values (open symbols) [76] with typical SIMS and IIE sample thicknesses. Fig ure 6 10 Calculated characteristic lengths (solid symbols) of LSM/YSZ, LSM, and YSZ compared to literature values (open symbol) [77] with typical SIMS and IIE sample thicknesses.

PAGE 129

129 Figure 6 11 Calculated characteristic lengths of LSM/GDC, LSM, and GDC with typical SIMS and IIE sample thicknesses.

PAGE 130

130 CHAPTER 7 EFFECT OF A AND B-SITE CATIONS ON THE SURFACE EXCHANGE COEFFICIENT FOR ABO3 PEROVSKITE MATERIALS 7 .1 Introduction The ABO3 perovskite famil y of materials with A=La, Sr and B=Co, Fe, Mn, have been show to give the best combination of electronic and ionic conductivity, stability, and compatibility as cathode materials for Solid Oxide Fuel Cells (SOFC). An SOFC converts chemical energy into ele ctrical energy by the separation of redox reactions at the anode and cathode, which is made possible by oxygen ion conducting electrolytes yttria stabilized zirconia (YSZ) or gadolinia doped ceria (GDC). (La0.8Sr0.2)MnO3 (LSM), which is considered a pure electronic conductor, has been proven to be a compatible cathode material with YSZ, a pure ionic conductor, but the combination of these two materials requires operating temperatures in excess of 800C for a rapid oxygen reduction reaction (ORR). (La0.6Sr0.4)(Co0.2Fe0.8)O3 (LSCF) has been shown to be a compatible cathode with GDC. Since LSCF is a mixed ionic and electronic conductor (MIEC), and GDC exhibits fast ionic conduction below 800C this combination has shown promise for intermediate temperature operation. Even with the combination of LSCF and GDC for intermediate temperature operation, there still exists significant polarization losses associated with the ORR at the cathode. The complete ORR consists of oxygen adsorption, dissociation, charge transfer, and inc orporation. To understand the cause of the loss, ideally the mechanism for ORR should be understood, but such complicated materials systems have proven this to be a challenging task. Another way to understand ORR behavior is to determine the kinetic para meters associated with the overall reaction. In particular, the diffusion coefficient (D) and surface exchange coefficient (k) are parameters that quantify the rate

PAGE 131

131 of bulk and surface reactions respectively. These parameters are commonly extracted from isotope diffusion profiles using isotope exchange depth profiling with secondary ion mass spectrometry (SIMS) [37, 38, 78] or from conductivity profiles from electrical conductivity relaxation (ECR) [8184] experiments. D values are well characterized from these experiments, but k values are less understood. Since these experiments require dense samples that are difficult to fabricate thin, inherent errors are associated wit h calculating an accurate k from diffusion profiles. The characteristic thickness (Lc) [86] of a sample, which is the ratio of D/k, is used to determine whether the sample is in a diffusion limited (bulk) regime, or a surface exchange controlled regime. Samples with thicknesses greater than Lc are diffusion limited, while samples with thicknesses less than Lc are surface exchange limited. Since SIMS and ECR require dense samples, typically thicknesses are on the order of or greater than Lc. Therefore, the surface behavior is masked by the diffusion behavior. Recently, a novel approach using isotope exchange was developed called isothermal isotope exchange (IIE) [102] to extract accurate k values. This technique relies on the exchange of isotopically labeled oxygen for lattice oxygen in a similar way to SIMS experiments, but can be conducted on powder materials. Using powder materials allows for the relative thickness to be the particle size of the material, which means the thickness of the sample is below Lc. IIE is used to extract k* (tracer surface exchange coefficients) from perovskite materials with varying A and B -site cations to better understand their roles in oxygen surface exchange behavior, w hich will aide in the optimization of intermediate

PAGE 132

132 temperature cathode materials. Samples tested include LSM, LSCF, LaMnO (LMO), LaFeO3 (LFO), LaCoO3 (LCO), (La0.6Sr0.4)FeO3 (LSF), and (La0.8Sr0.2)0.99CoO3 (LSC). 7 .2 Experimental LFO and LMO were prepared by aqueous combustion synthesis, which was previously described [102] LSM (P ra xair) LCO, LSF, LSC and LSCF (Fuel Cell Materials) were obtained from manufacturers as commercial powders. The specific surface areas of all materials were measured by BET. The particle sizes were measured by laser light scattering using a Coulter LS 13320. All samples were normalized by surface area to ~0.1 m2 and specific surface areas, sample masses, and particle sizes can be found in Table 1 Details and schematics of the experimental setup were given previously [40] Powder samples were placed on a quartz frit in the center of a quartz continuous flow reactor with a 4 mm inner diameter upstream of the sample and a 2 mm inner diameter downstream of the sample to reduce gas residence time. The quartz reactor was placed in the center of a custom made tube furnace capable of testing samples at temperatures up to 850C Samples were tested between 500850C with lower temperatures only being used for high ionic conducting materials. Samples were heated to the testing temperature and allowed to equilibrate in 20 sccm of a p16O2. In a separ ated line, an equal flow rate of p18O2 was mixed. Once equilibrated, an automatic switch was triggered to rapidly change the oxygen atmospheres. A quadrupole mass spectrometer was used to measure downstream effluents of m/z ratios of 32 (16O2), 34 (16O18O), and 36 (18O2). 1000 ppm argon was also included in the isotope gas stream to show the switch was complete.

PAGE 133

133 Diffusion profiles were processed using the method described in section 5.3. The solution to the diffusion equation for a sphere with surface evaporation derived by Crank [39] (Eq. 7 -1 -7 3) was fit to the data using linear regression optimization by a custom made Matlab program. 1 n 2 n 2 n 2 2 n 2 t)) L L ( L ( ) a / Dt exp( L 6 1 M M (7 -1) 0 1 L cot n n (7 -2) cL a D k a L (7 -3) Mt is the total amount of 18O accumulated in the sample, M is the total amount of accumulation as time goes to radius n is the root to Eq. 7 -2, and t is time. Since all samples were in a surface exc hange limited regime, the model was sensitive to k*, b ut not to D* as the model gave variations of D* over several orders of magnitude with equally good fits, while k* varied less than 5%. Therefore, k* can be considered to be accurate and was extracted h ere for analysis. 7 .3 Results IIE profiles for all samples at 800 C are seen in Fig. 7 -1. Due to slow incorporation, LMO, LFO, and LSM were all tested in a pO2 of 0.01 atm instead of 0.05 atm like the other samples. By testing these samples in a lower pO2, their kinetic signal is not saturated by the baseline concentration of gas discussed in chapter 5. LMO (Fig. 7 -1a), LFO (Fig. 7 -1b), and LSM (Fig. 7 -1d) all show incomplete exchange of the lattice within 40 -50 minutes of exchange. The rise and fall of m/z 34 (16O18O) shows the surface is active to adsorb and dissociate 18O2 into 18O where it combines with a lattice

PAGE 134

134 16O and desorbs. For all three samples the 34 signal continues to evolve surface species throughout the entire run. In contrast, the high electrically conductive material LCO (Fig. 7 -1c) and the MIEC materials, LSF (Fig. 7 -1e), LSC (Fig. 7 1f), and LSCF (Fig. 7 1g) also show a rise and fall of m/z 34 (16O18O), but eventually falls to ~0 ppm as the m/z 36 (18O2) signal reaches the baseline c oncentration within 12 -25 minutes of exchange. This indicates that there is no lattice 16O left to exchange and the lattice is saturated with 18O. Fig. 7 2 shows the 18O conversion fraction of the lattice along with model fits to the accumulation curves f rom the IIE experiments at 800C Once again, the differences between the pure electronic conductors and the MIECs can be clearly seen. LMO (Fig. 7 -2a), LFO (Fig. 7 -2b), and LSM (Fig. 7 -2d) all show incomplete conversion and evidence of an unwillingness to completely convert if the experiment was run for an extended period of time. Additionally, the rate of conversion and amount converted is a function of temperature for LFO and LSM, while LMO has already reached its plateau for all tem peratures tested. I n contrast the MIECs all show complete conversion and are furthermore independent of temperature. 7.4 Discussion From isotope exchange results of LSCF and LSM, a two step mechanism consisting of dissociative adsorption and incorporation of oxygen, was applied to demonstrate LSCF is dissociative adsorption limited and LSM is incorporation limited in the temperature range 600-800 C [40, 103] This behavior can be observed by the temperature dependence, or lack of c omplete conversion for oxygen exchange, of the pure electronic conductors LMO, LFO, and LSM and temperature independent exchange of the MIEC materials LSF, LSC, and LSCF in Fig. 72. The one exception to

PAGE 135

135 this behavior is LCO, which is a pure electronic conductor, but has very high conductivity (up to 1000 S/cm) [29, 30] While the two step kinetic model can separate dissociative adsorption and incorporation to determine which step is rate limiting, a comparison of the differences in surface exchange alone among the materials is not possible. However, using IIE to extract the k* values for comparison, it is possible to combine the results of the two step mechanism and the differences in k* values to have a more comp rehensive understanding of the ORR mechanism for these perovskite materials. Fig. 7 3 shows the k* values for the different A and B -site perovskites. When comparing the differences in magnitude of k*, the pure electronic conductors (LMO, LFO and LSM) are seen to have lower surface exchange abilities, while the MIEC materials ( LSF, LSC, and LSCF) have higher surface exchange abilities. Again, the one exception is LCO, which has similar magnitude k* values as the MIEC materials because of its high electr onic conductivity. Since ionic conductivity makes incorporation fast, from the two step model it was not obvious that surface exchange would be higher in these materials than the pure electronic conductors which were shown to be limited by incorporation due to lack of ionic conductivity. Here it is shown that in fact, the MIEC materials do exhibit faster surface exchange than the pure electronic conductors within the temperatures tested. Oxygen vacancies are shown to be important for surface exchange and possibly more important than electron concentration. This is evident by the higher k* for LSF, a material with high ionic conductivity and relatively low electronic conductivity, and the lower k* for LCO, a material with high electronic conductivity and no ionic conductivity Additionally, LMO, a pure electronic conductor with low conductivity has even lower k*s than LCO. Nonetheless, both oxygen vacancies and

PAGE 136

136 electrons are important as seen by the material with the highest concentration of both, LSC, has the highest k* values. However, the differences in magnitude do not tell the complete story for surface exchange, but rather differences in behavior with temperature are evident also. When interpreting the slope behavior, possibilities for obvious trends including electronic conductivity, ionic conductivity, valence state of the B -site transition metal cation, and adsorption/desorption behavior were ruled out due to lack of correlation. One way of comparing the behavior is to first observe the slopes of LMO and LCO, which exhibit similar activated behavior. Upon strontium substitution on the lanthanum sites, the behaviors similarly change to an apparent negative activation energy, which can be explained by a precursor mediated mechanism described in s ection 5.5.4 Since manganese is primarily in the 3+ and 4+ valence state and cobalt is primarily in the 2+ and 3+ state, the same behavior for the lanthanum manganites and cobaltites is surprising. Now consider the lanthanum ferrites. They behave oppos ite to that of the lanthanum manganites and cobaltites. Since iron is primarily in the 2+ and 3+ valence states, a correlation with LCO was considered more likely Therefore, less obvious surface behavior exists for oxyg en reduction on these materials an d more understanding is needed. 7.5 Conclusions IIE was conducted on varying A and B -site perovskite materials in the ABO3 configuration. Pure electronic conductors LMO, LFO, and LSM, exhibited temperature dependent exchange and an unwillingness for comp lete exchange. MIEC materials LSF, LSC, and LSCF exhibited temperature independent complete exchange. Even though LCO is not an ionic conductor, it does have high electronic conductivity and was

PAGE 137

137 shown to behave as the MIEC materials did for oxygen exchange. Since IIE experiments are demonstrated to be in a surface exchange regime, reliable k* values were extracted. The pure electronic conductors were shown to have the lowest k* values while the MIEC materials had the highest. Different temperature dep endencies were observed for the materials. The lanthanum manganites and cobaltites were observed to behave the same with opposite slopes upon strontium substitution on the lanthanum site, while lanthanum ferrites behaved opposite to the lanthanum manganit es and cobaltites No obvious trend based on electronic conductivity, ionic conductivity, valence state, or adsorption/desorption behavior was observed to explain the slope differences and more understanding is needed. The materials exhibiting positive s lopes (LFO, LSM, and LSC) have apparent negative activation energies consistent with a precursor mediated mechanism.

PAGE 138

138 Table 7 1 Specific surface areas, samples masses and particle sizes of tested samples. LMO LFO LCO LSM LSF LSC LSCF Specific Surface A rea (m 2 /g) 4.6 12.8 6.6 5.7 10.4 11.5 6.7 Sample Mass (mg) 22 8 15 18 10 9 15 Particle Size (nm) 310 177 3 07 158 259 296 282

PAGE 139

139 Figure 71. IIE of (a) LMO, (b) LFO, (c) LCO, (d) LSM, (e) LSF, (f) LSC, and (g) LSCF at 800C. LCO, LSF LSC, and LSCF were exchanged in 0.05 atm O2 and LMO, LFO, and LSM was exchanged in 0.01 atm O2.

PAGE 140

140 Figure 71. Continued.

PAGE 141

141 Figure 71. Continued.

PAGE 142

142 Figure 7 1 Continued. Figure 72. Fraction of 18O exchanged with lattice 16O (open symbols) with model fits (lines) for (a) LMO, (b) LFO, (c) LCO, (d) LSM, (e) LSF, (f) LSC, and (g) LSCF.

PAGE 143

143 Figure 72. Continued.

PAGE 144

144 Figure 72. Continued.

PAGE 145

145 Fig ure 7 2 Continued.

PAGE 146

146 Figure 7 3 Measured surface exchange coefficients of LMO, LFO, LCO, LSM, LSF, LSC, and LSCF.

PAGE 147

147 CHAPTER 8 ELECTROCATALYTIC ISOTHERMAL ISOTOPE EXCHANGE OF SUPPORTED CATALYSTS 8.1 Introduction Isothermal Isotope Exchange (IIE), described in chapters 5 7 is a powerful tool to measure surface exchange rates of catalysts for oxygen reduction. Though charge transfer is inherent in the exchange of 18O for lattice ions, it cannot be directly measured by IIE since no current is drawn. When a solid oxide fuel cell (SOFC) is operating the device is under a load current is conducted through the cell. S OFC materials can behave differently under a load, therefore it is also important to measure the catalytic activity of cathode materials in -situ An electrocatalytic experimental setup was built for this purpose. 8.2 Experimental To use as a supported sub strate, (Ce0.9Gd0.1)O2 (GDC) powder from Anan was pressed into a1.5 in. diameter pellet. The green body was sintered at 1550 C for 4 h until dense. Following sintering, the pellet was polished on both sides and such that the final thickness was ~ 0.5 m m. (La0.8Sr0.2)O3 (LSM) powder from Praxair was mixed with a commercially available vehicle to make an ink for screen -printing. LSM electrodes were screen -printed on both sides of the GDC substrate so that the sample was symmetric. The sample was sinter ed at 8 50 C for 1 h. Final electrode thicknesses were ~20 m. Platinum leads were adhered to the sample using brushed on platinum ink. The sample was loaded into a custom made electrocatalytic setup that is capable of in -situ current/voltage application with simultaneous current/voltage measurement using a Princ eton PARSTAT 2247 and gas effluent analysis using an Extrel quadrupole

PAGE 148

148 mass spectrometer (MS). A diagram of the experimental setup is shown in Fig. 81. A tape -casted glass seal was used to fuse the sample to the reactor to ensure a gas tight measuremen t chamber The sample was heated to 800 C and allowed to equilibrate in a pO2 of ~0.05 atm. A flow rate of 25 sccm of 16O2 balanced by helium was used to establish equilibrium. Once equilibrated, constant voltages between 0.11.0 V were applied to the s ample such that the lead attached to the electrode being monitored by the MS was positive and the electrode in air was attached to the negative lead. In this case, a constant flux of 16O2 was pumped from the air side to the MS side. In a separate line, an equal pO2 composed of 18O2 balanced by helium with a 1000 ppm argon tracer was mixed. Once the MS signals were stable, the gas lines were switched and the rise and fall of m/z signals 32 (16O2), 34 (16O18O), 36 (18O2), and 40 (Ar) were recorded. A diag ram of the experimental conditions is seen in Fig. 82. 8.3 Results and Discussion Isothermal isotope exchange plots (IIE) with applied voltages are seen in Fig. 83. From the existence of a scrambled product (16O18O), it is evident that even though 16O i ons are being pumped from the air side to the MS side, 18O2 gas molecules are adsorbing on the surface, dissociating and combing with 16O from the lattice. This type of experiment isolates the surface reaction because the applied voltage results in a curr ent that opposes the incorporation of oxygen ions on the MS side. However, the IIE plots look almost identical for applied voltages of 0.1, 0.3, 0.5, 0.75, and 1.0 V. This indicates that the surface reaction is not changing with respect to the different voltages. I t is assumed that the scrambled product is a measure of the number of 16O ions being pumped from the air side to the MS side (each 16O combines with an 18O) which is reasonable since upon switching the gas environments, the 32 (16O2) signal drops

PAGE 149

149 rapidly and continues to drop as the experiment progresses indicating no new 16O2 is being generated. T hen dm/dt of the oxygen ions can be calculated and compared to dm/dt calculated from Faradays law (Eq. 8-1) dt dm nF i (8 -1) In Eq. 8 1, i is the measured current, n is the number of electrons associated with each ion (2), and F is Faradays constant (96,485 C/mol). Figs. 8-4 and 85 show the dm/dt of 16O ions calculated using the current and the MS data respectively. dm/dt determined from t he generation of the scrambled product is higher than the flow of oxygen ions through the cell calculated from the current and shows little variation with voltage. However, the MS signal does mimic the drop in current as the experiment progresses seen clearly in the drop in dm/dt from the current plot for 0.5, 0.75 and 1.0 V. Since the scrambled product is larger, it is likely that the surrounding GDC substrate exposed to the gas is exchanging with the isotopically labeled oxygen also, since it was shown in chapters 5 and 6 that it is highly active for surface exchange. This would account for both the higher signals and the fact that the different voltages cannot be separated. In order to isolate the generation of scrambled product from the cat hode, the GDC would need to be shielded or the generation of scrambled product from the GDC would need to be subtracted. One way to accomplish this is to make the electrode larger, leaving little surface area of GDC exposed. Nonetheless, this result demonstrates the usefulness of this type of experimental setup for testing electrocatalytic behavior of cathodes under a load.

PAGE 150

150 8.4 Conclusion An electrocatalytic setup was designed to measure surface exchange reactions of fuel cell cathode materials in-situ. It has been demonstrated that it is possible to combine the IIE technique used to study powders with applied current/voltage to simulate working conditions. More development is needed to extract the kinetic parameters as well as completely isolate the cat hode interaction with isotopically labeled oxygen.

PAGE 151

151 Figure 8 1 Electrocatalytic experimental setup capable for in -situ testing with applied current/voltage and measurement of gas effluent. Figure 8 2 Experimental conditions for isothermal isotope experiments with applied voltages.

PAGE 152

152 Figure 83. Electrocatalytic IIE of LSM on GDC in 0.05 atm O2 with applied voltages of (a) 0.1, (b) 0.3, (c) 0.5, (d) 0.75, and (e) 1.0 V.

PAGE 153

153 Figure 83. Continued.

PAGE 154

154 Fig ure 8 3 Continued. Figure 8 4 dm/dt of o xygen ions through the cell calculated from Farad ays l aw.

PAGE 155

155 Fig ure 8 5 dm/dt of oxygen ions measured by mass spectrometry.

PAGE 156

156 CHAPTER 9 CONCLUSIONS Understanding the surface behavior of catalysts used for potentiometric sensors and solid oxide fuel cells (SOFC) is necessary for informed optimization. To this end, the NOx adsorption/desorption behavior and oxygen exchange behavior were characterized for perovskites in the lanthanum ferrite and manganite families. Additionally, the oxygen reduction reaction (ORR) on lanthanum ferrites, manganites, and cobaltites was characterized. From this work, a better understanding of NOx sensing behavior and the ORR has been achieved To aide in understanding of the potentiometric response of LaFeO3 (LFO) and LaMnO3+ (LMO) to oxidizing (NO2) and reducing (NO) gases, NOx adsorption behavior was characterized. Tempera ture programmed desorption revealed NOx adsorption to be much great er on LFO than LMO This is attributed to the different valence states of the iron an d manganese B -site cations. In LFO, i ron is primarily in the 3+ valence state which results in a stoichiometric lattice On the other hand, in LMO, manganese is in the 3+ and 4+ valence states. Charge compensation for the manganese 4+ valence state is achieved by the addition of excess oxygen ions, which are site accommodated by A and B -site vacancies. Though LFO exhibits more adsorption, species are more stable on LMO since they decompose and desorb at higher temperatures. Both materials had multipl e NO, NO2, and O2 desorption peaks following adsorption of either NO or NO2 giving evidence of complex surface species formation with different binding energies for each species Using diffuse reflectance infrared spectroscopy surface species were identif ied at room temperature in nitrite and nitrate formations giving further evidence that adsorbed

PAGE 157

157 gas was rearranging on the surface to form different species. This is significant for sensor response as nitrite and nitrate species are charged. When charged species are formed, the Fermi level at the surface changes which results in a potential change of the sensing elec trode relative to a reference. From temperature programmed reaction LFO was observed to exceed the NO2 to NO conver sion predicted by thermody namic calculations while LMO lagged conversion. This is due to nitrite/nitrate formation and stability differences on LFO and LMO rather than to the metal oxides being more or less active toward decomposition of NO2. This behavior cannot be explained by Mixed Potential Theory, but is better described by Differential Electrode Equilibria. Next, the oxygen exchange behavior was studied on a var iety of pure and composite lanthanum ferrite and manganite materials used as cathodes for SOFCs. A novel approach using isotope exchange, called isothermal isotope exchange (IIE) was developed to extract the surface exchange coefficient (k*), a kinetic rate parameter. This has been done before using other techniques, but they required dense and thus thick samples w hich limits the kinetic behavior of the material to a diffusion controlled regime. When in a diffusion controlled regime, ORR is controlled by bulk diffusion and surface activity is masked. Therefore, any k* values extracted are unreliable IIE is an ox ygen tracer technique that was shown to be sensitive to k*, primarily due to the ability to test powders instead of dense samples. Using this technique, common cathode materials (La0.6Sr0.4)(Co0.2Fe0.8)O3 (LSCF) and (La0.8Sr0.2)O(LSM) and common electrolyte materials (Ce0.9Gd0.1)O3 (GDC) and (Y2O3)0.8(ZrO2)0.92 (YSZ) were characterized While the mixed ionic and electronic conductors (MIEC) LSCF and GDC

PAGE 158

158 had low activation energies showing little variation of catalytic activity over the temperature range 500800C the pure conductors LSM (electronic) and YSZ (ionic) had opposite temperature dependencies. LSM had an apparent negative activation energy for dissociative adsorption and YSZ was activa ted. It was concluded that the behavior of LSM can be explained by a precursor mediated mechanism as described in section 5.5.4. To further explore the behaviors of the pure materials, composite powders of LSCF/GDC, LSM/YSZ, and LSM/GDC were tested as wel l. LSM composites maintained the apparent negative activation energy seen from pure LSM which validated the initial results, and showed overall improvement of k* while combining the MIEC materials showed little improvement. It is concluded that the inc rease in k* with decreasing temperature could be taken advantage of at low and intermediate temperatures if the particles were small enough such that the surface reactivity was solely utilized. In this configuration, the ionic conductors which have high diffusion coefficients (D*) would be responsible for incorporation and transport. Pure perovskite materials with different A and B -site cations were tested to determine the effect of the different cations on k*. Pure electronic conductors LaMnO (LMO), LaFeO3 (LFO), LaCoO3 (LCO) and LSM exhibited lower k* values than MIEC materials (La0.6Sr0.4)FeO3 (LSF), (La0.8Sr0.2)0.99CoO3 (LSC), and LSCF demonstrating the importance of both electrons and oxygen vacancies for surface exchange. However, the behav ior of k* with temperature varied unexpectedly showing no correlation with electronic conductivity, ionic conductivity, B -site transition metal cation valence state, or oxygen adsorption/desorption behavior. Therefore, more

PAGE 159

159 understanding is needed to expl ain the slope variation among these materials However, it is clear that the behaviors of LFO, LSM, and LSC can be described by the precursor mediated mechanism. Comparing the NOx adsorption/desorption behavior of LFO and LMO with their oxygen exchange be havior, it is clear that two different mechanisms exist for these materials. Where LFO was more active for NOx adsorption, LMO is more active for oxygen surface exchange at high temperatures. Furthermore, LFO and LMO exhibited opposite temperature depend encies for surface exchange. Finally, an electrocatalytic setup was developed to test the surface exchange behavior of supported catalysts with applied current/voltage and simultaneous measurement of gas phase effluent s With this setup IIE can be used w ith applied currents, simulating a working fuel cell. This allows for the ability to measure surface properties in -situ for more comprehensive mechanistic analysis. Initial results from an LSM/GDC/LSM symmetric cell were shown as a proof of concept.

PAGE 160

160 APPENDIX A MASS SPECTROMETER CA LIBRATION To obtain accurate concentration measurements from the mass spectrometer (MS), calibration is necessary. Following each experiment, each component gas was calibrated by flowing three different known concentrations of each gas at a constant flow rate of 20 sccm and recording the intensity signal from the MS. From these three points, a linear relationship between gas concentration and MS intensities was established for each gas. The linear equation was then used to convert the recorded experimental intensities into concentrations of parts per million (ppm). In the case of NO2, the molecule has a NO cracking fraction that was subtracted from the NO signal. To do this, the NO intensities were recorded relative to known concentrations of NO2 in ppm. This way, the MS signal of NO generated from the cracking fraction of NO2 could be subtracted from the NO MS signal. Once the concentration in ppm was known, since the volumetric flow rate was also known, the flow rate in m ole/s was calculated from the ideal gas law using a pressure of 1 atm and a temperature of 25 C. This calibration procedure was used for all temperature programmed desorption and reaction experiments.

PAGE 161

161 APPENDIX B ERROR ANALYSIS OF SURFACE EXCHANGE COEFFI CIENTS When considering possible error associated with the surface exchange coefficient calculations, the first source to consider is the raw isothermal isotope exchange (IIE) profile. The raw IIE profile is converted to an accumulation curve according to the procedure outlined in the experimental section in chapter 5. Therefore, the error in the measurements of concentration and time measured by the mass spectrometer must be considered since the accumulation curve is processed from the IIE profile. Sinc e the accumulation curve is then made into a fraction of conversion curve, the error is defined as follows. M M ) t t C C C C ( M M t 36 36 34 34 t (B-1) t/M 34 36 are the errors in the concentration mea measurement of time, C34 and C36 are the instantaneous measurements of the 34 and 36 signal at an instantaneous time t, and Mt/M is the instantaneous fraction of conversion. The limits of the mas s spectrometer for concentration and time measurements are listed below. 34 = 1 ppm 36 = 1 ppm t/M can be determined. For this analysis, values for C34, C36, and t were chosen at Mt/ M=0.5 for all samples except L MO, LFO, and LSM, which were chosen at Mt/ M=0.3 due to lower conversion fractions.

PAGE 162

162 the extraction of the surface exchange coefficient from the model according to Eqs. B 2 B-4. 1 n 2 n 2 n 2 2 n 2 t)) L L ( L ( ) a / Dt exp( L 6 1 M M (B-2) 0 1 L cot n n (B-3) cL a D k a L (B-4) To do this, Eq. B 2 must be differentiated with respect to L and t. The solutions are shown in Eqs. B 5 and B 6 respectively. 1 n 2 2 n 2 n 2 2 n 2 n 2 n 2 n 2 n 2 t))) 1 L ( L ( ( )) a / Dt exp( ) L 2 ( ))) 1 L ( L ( ( 2 ( L 6 dL M M d (B-5) 1 n 2 n 2 2 n 2 2 n 2 n 2 t)) 1 L ( L ( a )) a / Dt exp( D L 6 dt M M d (B-6) t/M determined from the following relationship. dL M M d t dt M Md M M L t t t (B-7) -4) according to Eq. ( B-8).

PAGE 163

163 k ) L L ( k (B-8) The error in the measurement of a, the radius of the particle, should also be considered. Due to the size of particles, measurements were limited to using the laser light scattering technique for analysis. When using laser light scattering, the model applied to interpret the data involves the complex index of refraction at light wavelengths of 450, 600, and 900 nm. Since the perovskite and fluorite materials used in this study are complex, these values were unable t o be obtained. Rather, a complex index found from a laser light scattering technical application for (La0.6Sr0.4)(Co0.2Fe0.8)O3 (LSCF) was used for all perovskite samples, while a complex index of refraction found for (Ce0.9Gd0.1)O2 (GDC) was used for the fluorite samples. It was found that when this model was applied to each sample for two to four runs each sample, that a single peak was observed for the distribution of particle sizes when plotting the results in number%, and the mean particle sizes and standard deviations were consistent among all runs for each sample. However, a large standard deviation was observed due to agglomeration of a limited number of particles. Because of this, if the standard deviation was used as the error, the errors w ould be too large to plot. However, if the measured results were extrapolated out to 100 measurements, the distribution would become normal and the standard error would be much lower than the standard deviation based on t test calculations according to Eq B 9. n SD t SE (B-9) In this equation, SE is the standard error, t is the t -test value based on the degrees of freedom and % confidence in the results, SD is the standard deviation, and n

PAGE 164

164 is the number of samples. Therefore, error associ ated with the particle size measurements were not included in this analysis due to estimation of complex indices that could have resulted in larger standard deviations, and too small of a sample size to determine an accurate standard error. Finally, using this previously described error analysis, only the (La0.8Sr0.2)MnO (LSM) sample at 775 and 800C had errors large enough to be viewed on the scales plotted.

PAGE 165

165 APPENDIX C ISOTHERMAL ISOTOPE EXCHANGE PROFILES Figure C -1. LSCF IIE profiles exchanged in 0.05 atm O2 at (a) 500C, (b) 600C and (c) 700C.

PAGE 166

166 Fig ure C -1 Continued. Figure C -2. LSM IIE profiles exchanged in 0.01 atm O2 at (a) 775C, (b) 825C and (c) 850C.

PAGE 167

167 Fig ure C -2 Continued.

PAGE 168

168 Figure C -3. GDC IIE profiles exchanged in 0.05 atm O2 at (a) 500C, (b) 600C, (c) 700C, and (d) 850C.

PAGE 169

169 Figure C -3. Continued.

PAGE 170

170 Figure C -4. YSZ IIE profiles exchanged in 0.05 atm O2 at (a) 700C, (b) 750C, and (c) 850C.

PAGE 171

171 Fig ure C -4 Continued. Figure C -5. LSCF/ GDC IIE profiles exchanged in 0.05 atm O2 at (a) 500C, (b) 600C, (c) 700C and (d) 850C.

PAGE 172

172 Figure C -5. Continued.

PAGE 173

173 Fig ure C -5 Continued. Figure C -6. LSM/YSZ IIE profiles exchanged in 0.05 atm O2 at (a) 700C, (b) 750C, and (c) 850C.

PAGE 174

174 Fig ure C -6 Continued.

PAGE 175

175 Figure C -7. LSM/GDC IIE profiles exchanged in 0.05 atm O2 at (a) 650C (b) 700C, (c) 750C, and (d) 850C.

PAGE 176

176 Fig ure C -7 Continued.

PAGE 177

177 Figure C -8. LMO IIE profiles exchanged in 0.01 atm O2 at (a) 700C, (b) 750C and (c) 850C.

PAGE 178

178 Figure C -8. Continued. Figure C -9. LFO IIE profiles exchanged in 0.01 atm O2 at (a) 700C, (b) 750C and (c) 850C.

PAGE 179

179 Fig ure C -9 Continued.

PAGE 180

180 Figure C -10. LCO IIE profiles exchanged in 0.05 atm O2 at (a) 700C, (b) 750C, and (c) 850C.

PAGE 181

181 Figure C -10. Continued. Figure C -11. LSF IIE profiles exchanged in 0.05 atm O2 at (a) 700C (b) 750C, and (c) 850C.

PAGE 182

182 Fig ure C -11 Continued.

PAGE 183

183 Figure C -12. LSC IIE profiles exchanged in 0.05 atm O2 at (a) 650C, (b) 700C, (c) 750C, and (d) 850C.

PAGE 184

184 Fig ure C -12 Continued.

PAGE 185

185 LIST OF REFERENCES [1] U nited States E nvironmental Protection Agency, Clean Air Act, http://www.epa.gov/air/caa/ (2010). [2] U nited States Environmental Protection Agency Nitrogen Dioxide, http://www.epa.gov/air/nitrogenoxides/ (2010). [3] United States Department of Energy Strategic Signifi cance of America's Oil Shale Resource, Volume I Assesment of Strategic Issues, (2004). [4] W. Gopel, G. Reinhardt, M. Rosch, Solid State Ionics 136 (2000) 519. [5] A.M. Azad, S.A. Akbar, S.G. Mhaisalkar, L.D. Birkefeld, K.S. Goto, Journal of the Electroc hemical Society 139 (1992) (12) 3690. [6] F. Menil, V. Coillard, C. Lucat, Sensors and Actuators B -Chemical 67 (2000) (12) 1. [7] S. Zhuiykov, N. Miura, Sensors and Actuators B -Chemical 121 (2007) (2) 639. [8] N. Miura, H. Kurosawa, M. Hasei, G.Y. Lu, N. Yamazoe, Solid State Ionics 868 (1996) 1069. [9] N. Miura, T. Raisen, G. Lu, N. Yamazoe, Sensors and Actuators B -Chemical 47 (1998) (13) 84. [10] E.D. Waschman, P. Jayaweera, Selective Detection of NOx By Differential Electrode Equilibria, In: E.D. Wach sman, W. Weppner, E. Traversa, M. Liu, P. Vanysek, N. Yamazoe, Editors, Solid-State Ionics II: Ceramic Sensors, The Electrochemical Soceity (2001), p.298304. [11] E.D. Wachsman, P. Jayaweera, G. Krishnan, A. Sanjurjo, Solid State Ionics 136 (2000) 775. [ 12] J. Yoo, F.M. Van Assche, E.D. Wachsman, Journal of the Electrochemical Society 153 (2006) (6) H115. [13] E. Di Bartolomeo, M.L. Grilli, J.W. Yoon, E. Traversa, Journal of the American Ceramic Society 87 (2004) (10) 1883. [14] S.B. Adler, Chemical Reviews 104 (2004) (10) 4791. [15] J. Nowotny, T. Bak, M.K. Nowotny, C.C. Sorrell, Advances in Applied Ceramics 104 (2005) (4) 154. [16] V. Goldschmidt, T. Barth, G. Lunde, W. Zachariasen, VII: Die Gesetze der Krystallochemie, Geochemische Verteilungsgesetze de r Elemente, Oslo (1926).

PAGE 186

186 [17] L.W. Tai, M.M. Nasrallah, H.U. Anderson, D.M. Sparlin, S.R. Sehlin, Solid State Ionics 76 (1995) (3 4) 273. [18] V.V. Kharton, A.V. Kovalevsky, A.P. Viskup, A.L. Shaula, F.M. Figueiredo, E.N. Naumovich, F.M.B. Marques, Solid S tate Ionics 160 (2003) (3-4) 247. [19] B.C. Tofield, W.R. Scott, Journal of Solid State Chemistry 10 (1974) (3) 183. [20] J.A.M. Vanroosmalen, E.H.P. Cordfunke, R.B. Helmholdt, H.W. Zandbergen, Journal of Solid State Chemistry 110 (1994) (1) 100. [21] J.H. Kuo, H.U. Anderson, D.M. Sparlin, Journal of Solid State Chemistry 87 (1990) (1) 55. [22] V.V. Kharton, A.V. Kovalevsky, A.P. Viskup, F.M. Figueiredo, A.A. Yaremchenko, E.N. Naumovich, F.M.B. Marques, Journal of the Electrochemical Society 147 (2000) (7) 2814. [23] S.B. Sohn, S.Y. Choi, G.H. Kim, H.S. Song, G.D. Kim, Journal of Non -Crystalline Solids 297 (2002) (23) 103. [24] J. Mizusaki, T. Sasamoto, W.R. Cannon, H.K. Bowen, Journal of the American Ceramic Society 65 (1982) (8) 363. [25] E.V. Bongio, H. Black, F.C. Raszewski, D. Edwards, C.J. McConville, V.R.W. Amarakoon, Journal of Electroceramics 14 (2005) (3) 193. [26] A.A. Yaremchenko, M.V. Patrakeev, V.V. Kharton, F.M.B. Marques, I.A. Leonidov, V.L. Kozhevnikov, Solid State Sciences 6 (2004) (4) 357. [27] L.W. Tai, M.M. Nasrallah, H.U. Anderson, D.M. Sparlin, S.R. Sehlin, Solid State Ionics 76 (1995) (3 4) 259. [28] M. Mogensen, N.M. Sammes, G.A. Tompsett, Solid State Ionics 129 (2000) (14) 63. [29] J. Mizusaki, J. Tabuchi, T. Matsuura, S. Yamauchi, K. Fueki, Journal of the Electrochemical Society 136 (1989) (7) 2082. [30] A.N. Petrov, O.F. Kononchuk, A.V. Andreev, V.A. Cherepanov, P. Kofstad, Solid State Ionics 80 (1995) (34) 189. [31] A. Petric, P. Huang, F. Tietz, Solid State Ionics 135 (2000) (1 -4) 719. [32] Y. Ohno, S. Nagata, H. Sato, Solid State Ionics 9-10 (1983) (DEC) 1001. [33] G.C. Kostogloudis, C. Ftikos, Solid State Ionics 126 (1999) (12) 143.

PAGE 187

187 [34] Y. Teraoka, H.M. Zhang, K. Okamoto, N. Yamazoe, Materials Research Bulletin 23 (1988) (1) 51. [35] J.W. Stevenson, T.R. Armstrong, R.D. Carneim, L.R. Pederson, W.J. Weber, Journal of the Electrochemical Society 143 (1996) (9) 2722. [36] J.L. Falconer, J.A. Schwarz, Catalysis Reviews -Science and Engineering 25 (1983) (2) 141. [37] R.J. Chater, S Carter, J.A. Kilner, B.C.H. Steele, Solid State Ionics 53 (1992) 859. [38] J.A. Kilner, B.C.H. Steele, L. Ilkov, Solid State Ionics 12 (1984) (MAR) 89. [39] J. Crank, The Mathematics of Diffusion, Oxford, Clarendon Press (1975) p. 36, 60, 96. [40] C.C. Kan, H.H. Kan, F.M. Van Assche, E.N. Armstrong, E.D. Wachsman, Journal of the Electrochemical Society 155 (2008) (10) B985. [41] J. Yoo, H. Yoon, E.D. Wachsman, Journal of the Electrochemical Society 153 (2006) (11) H217. [42] J. Yoo, E.D. Wachsman, Senso rs and Actuators B -Chemical 123 (2007) (2) 915. [43] J. Yoo, S. Chatterjee, F.M. Van Assche, E.D. Wachsman, Journal of the Electrochemical Society 154 (2007) (7) J190. [44] B. White, E. Traversa, E. Wachsman, Journal of the Electrochemical Society 155 (2008) (1) J11. [45] F.M. van Assche, J.C. Nino, E.D. Wachsman, Journal of the Electrochemical Society 155 (2008) (7) J198. [46] F.M. Van Assche, E.D. Wachsman, Solid State Ionics 179 (2008) (39) 2225. [47] J.W. Yoon, M.L. Grilli, E. Di Bartolomeo, R. Polini, E. Traversa, Sensors and Actuators B-Chemical 76 (2001) (13) 483. [48] M.L. Grilli, E. Di Bartolomeo, E. Traversa, Journal of the Electrochemical Society 148 (2001) (9) H98. [49] N.N. Toan, S. Saukko, V. Lantto, Physica B -Condensed Matter 327 (2003) (24) 279. [50] E. Di Bartolomeo, N. Kaabbuathong, A. D'Epifanio, M.L. Grilli, E. Traversa, H. Aono, Y. Sadaoka, Journal of the European Ceramic Society 24 (2004) (6) 1187.

PAGE 188

188 [51] E. Di Bartolomeo, N. Kaabbuathong, M.L. Grilli, E. Traversa, Solid State Ionics 171 (2004) (34) 173. [52] E. Di Bartolomeo, M.L. Grilli, E. Traversa, Journal of the Electrochemical Society 151 (2004) (5) H133. [53] V. Lantto, S. Saukko, N.N. Toan, L.F. Reyes, C.G. Granqvist, Journal of Electroceramics 13 (2004) (13) 721. [54] R.D. Zhan g, H. Alamdari, S. Kallaguine, Journal of Catalysis 242 (2006) (2) 241. [55] E.L. Brosha, R. Mukundan, D.R. Brown, F.H. Garzon, J.H. Visser, Solid State Ionics 148 (2002) (1 -2) 61. [56] E.L. Brosha, R. Mukundan, D.R. Brown, F.H. Garzon, Sensors and Actuators BChemical 87 (2002) (1) 47. [57] J. Yoo, S. Chatterjee, E.D. Wachsman, Sensors and Actuators B -Chemical 122 (2007) (2) 644. [58] F.H. Garzon, R. Mukundan, E.L. Brosha, Solid State Ionics 136 (2000) 633. [59] N. Miura, G.Y. Lu, N. Yamazoe, Solid State Ionics 136 (2000) 533. [60] S.J. Huang, A.B. Walters, M.A. Vannice, Journal of Catalysis 192 (2000) (1) 29. [61] S.E. Dann, D.B. Currie, M.T. Weller, M.F. Thomas, A.D. Alrawwas, Journal of Solid State Chemistry 109 (1994) (1) 134. [62] R.J. Bell, G.J. Mil lar, J. Drennan, Solid State Ionics 131 (2000) (3-4) 211. [63] N. Sakait, H. Fjellvag, Acta Chemica Scandinavica 50 (1996) 580. [64] T. Striker, J.A. Ruud, J. Am. Ceram. Soc. In Press (2010). [65] S. Jaenicke, G.K. Chuah, Berichte Der Bunsen-Gesellschaft -Physical Chemistry Chemical Physics 96 (1992) (1) 1. [66] P. Porta, S. Cimino, S. De Rossi, M. Faticanti, G. Minelli, I. Pettiti, Materials Chemistry and Physics 71 (2001) (2) 165. [67] K.I. Hadjiivanov, Catalysis Reviews -Science and Engineering 42 (2000) (1 2) 71. [68] Y. Teraoka, M. Yoshimatsu, N. Yamazoe, T. Seiyama, Chemistry Letters (1984) (6) 893. [69] K.H. Stern, The Journal Of Physical Chemistry Reference Data 1 (1972) (3) 747.

PAGE 189

189 [70] S. Yuvaraj, F.Y. Lin, T.H. Chang, C.T. Yeh, Journal of Physical Chemistry B 107 (2003) (4) 1044. [71] E. Di Bartolomeo, E. Traversa, M. Baroncini, V. Kotzeva, R.V. Kumar, Journal of the European Ceramic Society 20 (2000) (16) 2691. [72] E.R. Macam, B.M. White, B.M. Blackburn, E. Traversa, E.D. Wachsman, In Progress (2010) [73] M.C. Williams, Fuel Cells 7 (2007) (1) 78. [74] M.C. Williams, J.P. Strakey, W.A. Surdoval, L.C. Wilson, Solid State Ionics 177 (2006) (19-25) 2039. [75] R.A. De Souza, J.A. Kilner, Solid State Ionics 106 (1998) (34) 175. [76] A. Esquirol, J. Kilne r, N. Brandon, Solid State Ionics 175 (2004) (1-4) 63. [77] Y. Ji, J.A. Kilner, M.F. Carolan, Solid State Ionics 176 (2005) (9-10) 937. [78] J.A. Kilner, R.A. DeSouza, I.C. Fullarton, Solid State Ionics 868 (1996) 703. [79] P.S. Manning, J.D. Sirman, J.A Kilner, Solid State Ionics 93 (1996) (1-2) 125. [80] P.S. Manning, J.D. Sirman, R.A. DeSouza, J.A. Kilner, Solid State Ionics 100 (1997) (12) 1. [81] J.A. Lane, S.J. Benson, D. Waller, J.A. Kilner, Solid State Ionics 121 (1999) (14) 201. [82] S. Kim, Y .L. Yang, A.J. Jacobson, B. Abeles, Solid State Ionics 121 (1999) (14) 31. [83] S. Wang, P.A.W. van der Heide, C. Chavez, A.J. Jacobson, S.B. Adler, Solid State Ionics 156 (2003) (1-2) 201. [84] R. Ganeshananthan, A.V. Virkar, Journal of the Electrochemic al Society 152 (2005) (8) A1620. [85] J. Maier, Solid State Ionics 112 (1998) (34) 197. [86] P.J. Gellings, H.J.M. Bouwmeester, The CRC handbook of solid state electrochemistry, CRC Press, Boca Raton, Fla. (1997) p. 504 [87] S.J. Benson, PhD thesis Impe rial College, London, U.K.,1999.. [88] C. W. Lee, R.K. Behera, S. Okamoto, R. Devanathan, E.D. Wachsman, S.R. Phillpot, S.B. Sinnott, Journal of American Ceramic Society (Submitted March, 2010).

PAGE 190

190 [89] J.F. Weaver, S.P. Devarajan, C. Hakanoglu, J. Phys. Chem C 113 (2009) (22) 9773. [90] K.L. Duncan, Y.L. Wang, S.R. Bishop, F. Ebrahimi, E.D. Wachsman, Journal of Applied Physics 101 (2007) (4). [91] K.L. Duncan, Y.L. Wang, S.R. Bishop, F. Ebrahimi, E.D. Wachsman, Journal of the American Ceramic Society 89 (200 6) (10) 3162. [92] S.R. Bishop, K.L. Duncan, E.D. Wachsman, Acta Materialia 57 (2009) (12) 3596. [93] T. Kawada, N. Sakai, H. Yokokawa, M. Dokiya, I. Anzai, Solid State Ionics 50 (1992) (34) 189. [94] B.C.H. Steele, Journal of Materials Science 36 (2001) (5) 1053. [95] A. Esquirol, N.P. Brandon, J.A. Kilner, M. Mogensen, Journal of the Electrochemical Society 151 (2004) (11) A1847. [96] Y.J. Leng, S.H. Chan, Q.L. Liu, International Journal of Hydrogen Energy 33 (2008) (14) 3808. [97] E.P. Murray, M.J. Seve r, S.A. Barnett, Solid State Ionics 148 (2002) (1-2) 27. [98] A.C. Co, S.J. Xia, V.I. Birss, Journal of the Electrochemical Society 152 (2005) (3) A570. [99] E.P. Murray, T. Tsai, S.A. Barnett, Solid State Ionics 110 (1998) (34) 235. [100] E.P. Murray, S. A. Barnett, Solid State Ionics 143 (2001) (3-4) 265. [101] S.P. Jiang, W. Wang, Solid State Ionics 176 (2005) (1516) 1351. [102] E. Armstrong, K. Duncan, D. Oh, E. Wachsman, In Progress. (2010). [103] C.C. Kan, E.D. Wachsman, Solid State Ionics 181 (2010) (5 7) 338.

PAGE 191

191 BIOGRAPHICAL SKETCH Eric was born to parents David and Norma Armstrong in 1981 and was raised in Turlock, California, a small town in the heart of the central valley. He grew up excelling in school, but enjoying math ematics in particular. I n high school, he continued to excel in academics while he also lettered in tennis and cross country During his sophomore year he came to recognize Gods love for him, which changed the way he would live for the rest of his life. He made up his mind ar ound his junior year of high school that he wanted to attend the University of California, Los Angeles (UCLA), the same university his father attended. He was accepted as a m aterials s cience and e ngineering major and began attending UCLA in the fall of 19 99. At UCLA, Eric was involved in many activities including competitive Ultimate [Frisbee] residential life as a resident assistant for two years, and the association for careers in technology which hosted career fairs for technical majors twice a year. Early on, he struggled making the connection between what he was learning and the application of it in the real world, but quickly became interested in semiconductor materials seeing many ways in which he could apply his knowledge in industry He gradu ated with his Bachelor of Science in m aterials s cience and e ngineering with an electronic materials specialty in March of 2004. After graduating, while looking for a job, Eric moved to Scotts Valley, California where he worked as a staff counselor, host a nd even did some construction work for a general contract or at Mission Springs Christian Camp and Conference Center. It was during this time that Eric decided graduate education would be an excellent way to develop more skills that he could apply in the real world. Within a month of visiting the University of Florida and committing to attend there to study m aterials s cience and

PAGE 192

192 e ngineering in the fall of 2005, Eric met his beautiful wife Stacey while they were attending the same Bible study. Their relationship rapidly blossomed and after five months of dating, Eric moved 3,000 miles away to begin graduate school, while Stacey remained in California. In November they were engaged, and in June of 2006, they were married. From that point on, they made their home together in Gainesville, Florida while Eric comp leted his studies, working on characterizing the catalytic and kinetic behavior of metal oxide s for applications as potentiometric gas sensors and solid oxide fuel cells. He graduated with a Master of Science d egree in m aterials s cience and e ngineering fr om the University of Florida in December of 2006 It was not always clear throughout Erics studies what the path would be, but looking back it is clear that Gods hand played an integral role in guiding it and he remains thankful and humbled by the way hi s God loves and cares for him in all circumstances.