<%BANNER%>

Comprehensive Development of High Performance Solid Oxide Fuel Cells for Intermediate and Low Temperature Applications

Permanent Link: http://ufdc.ufl.edu/UFE0042114/00001

Material Information

Title: Comprehensive Development of High Performance Solid Oxide Fuel Cells for Intermediate and Low Temperature Applications
Physical Description: 1 online resource (182 p.)
Language: english
Creator: Lee, Kang
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: anode, cathode, electrolyte, energy, performance, power, sofc
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: In order to develop high performance solid oxide fuel cells (SOFCs) operating at low to intermediate temperatures, the three main SOFC components--the anode, electrolyte, and cathode--were comprehensively studied. In order to lower anodic polarization losses in anode-supported SOFCs, a novel composite anode functional layer (AFL) having bimodal (nano/micro) structure was developed. Application of this AFL involved a simple process where a precursor solution was coated onto a conventional submicron sized colloidally-deposited Ni-GDC AFL. Cells prepared in this manner yielded maximum power densities (MPD) of 1.29, 1.16, 0.7 and 0.38 W/cm2 at 650, 600, 550 and 500 oC, respectively. Electrochemical impedance results showed a striking decrease in both ohmic and non-ohmic area specific resistances (ASRs) for these cells compared to those with either no AFL, or a conventional AFL. In addition, the effect composition of the conventional submironsized AFL on performance was examined. The highest MPD (1.15 W/cm2 at 650 degreeC) was achieved at a composition of 60wt% NiO. This composition had the best performance over the intermediate temperature range (450 to 650 degreeC). For the potentio-static test, the cell exhibited stable performance over 200 hrs of operation at 1.1 W/cm2. It was also revealed that electrode ASR has an inverse linear relationship with maximum power density at 650oC. To better understand the effect of AFL composition, microstructural features of AFLs having various Ni-GDC compositions were quantified by a 3 dimensional (3D) reconstruction technique using a FIB/SEM dual beam system. Of the compositions tested, the highest triple phase boundary (TPB) density was achieved at 60wt% NiO, which corresponds to a 1:1 volume ratio of Ni to GDC phase. The quantified TPB density showed an inverse proportionality to electrode ASR. Using a wet chemical co-precipitation method, nano-sized ESB particles were successfully synthesized at temperatures as low as ~ 500 oC. Due to the high sinterability of this powder, a dense erbia stabilized bismuth oxide (ESB) layer was successfully formed on a gadolinia doped ceria (GDC) electrolyte by a simple colloidal coating method. A systematic study on the sintering behavior of ESB was conducted to determine the optimum sintering conditions for these materials. I-V measurement a cell using this bilayered electrolyte sytem showed a high power density ( ~ 1.5 W/cm2) at 650 oC due to an enhancement in OCP and a significant reduction in ASR when compared to a GDC single cell. The performance of conventional (La0.80Sr0.20)MnO3-? (LSM) cathodes were dramatically improved at the IT range by combining it with a highly conductive ESB phase. The electrode ASR measured from a symmetric cell consisting of LSM-ESB electrodes on an ESB electrolyte was only 0.08 ?-cm2 at 700oC which is ~60% lower than that of LSM-ESB on GDC electrolytes (0.19 ?-cm2). This exemplifies the synergetic effect the ESB phase has both in the cathode bulk and at the electrolyte/electrode interface. The MPDs of the anode-supported SOFCs with LSM-ESB cathodes on ESB/GDC bilayered electrolytes were ~836 mW/cm2 at 650 oC, which is the highest value reported for SOFCs using LSM-bismuth oxide composite cathodes.
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 Kang Lee.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Wachsman, Eric D.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-02-28

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042114:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042114/00001

Material Information

Title: Comprehensive Development of High Performance Solid Oxide Fuel Cells for Intermediate and Low Temperature Applications
Physical Description: 1 online resource (182 p.)
Language: english
Creator: Lee, Kang
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: anode, cathode, electrolyte, energy, performance, power, sofc
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: In order to develop high performance solid oxide fuel cells (SOFCs) operating at low to intermediate temperatures, the three main SOFC components--the anode, electrolyte, and cathode--were comprehensively studied. In order to lower anodic polarization losses in anode-supported SOFCs, a novel composite anode functional layer (AFL) having bimodal (nano/micro) structure was developed. Application of this AFL involved a simple process where a precursor solution was coated onto a conventional submicron sized colloidally-deposited Ni-GDC AFL. Cells prepared in this manner yielded maximum power densities (MPD) of 1.29, 1.16, 0.7 and 0.38 W/cm2 at 650, 600, 550 and 500 oC, respectively. Electrochemical impedance results showed a striking decrease in both ohmic and non-ohmic area specific resistances (ASRs) for these cells compared to those with either no AFL, or a conventional AFL. In addition, the effect composition of the conventional submironsized AFL on performance was examined. The highest MPD (1.15 W/cm2 at 650 degreeC) was achieved at a composition of 60wt% NiO. This composition had the best performance over the intermediate temperature range (450 to 650 degreeC). For the potentio-static test, the cell exhibited stable performance over 200 hrs of operation at 1.1 W/cm2. It was also revealed that electrode ASR has an inverse linear relationship with maximum power density at 650oC. To better understand the effect of AFL composition, microstructural features of AFLs having various Ni-GDC compositions were quantified by a 3 dimensional (3D) reconstruction technique using a FIB/SEM dual beam system. Of the compositions tested, the highest triple phase boundary (TPB) density was achieved at 60wt% NiO, which corresponds to a 1:1 volume ratio of Ni to GDC phase. The quantified TPB density showed an inverse proportionality to electrode ASR. Using a wet chemical co-precipitation method, nano-sized ESB particles were successfully synthesized at temperatures as low as ~ 500 oC. Due to the high sinterability of this powder, a dense erbia stabilized bismuth oxide (ESB) layer was successfully formed on a gadolinia doped ceria (GDC) electrolyte by a simple colloidal coating method. A systematic study on the sintering behavior of ESB was conducted to determine the optimum sintering conditions for these materials. I-V measurement a cell using this bilayered electrolyte sytem showed a high power density ( ~ 1.5 W/cm2) at 650 oC due to an enhancement in OCP and a significant reduction in ASR when compared to a GDC single cell. The performance of conventional (La0.80Sr0.20)MnO3-? (LSM) cathodes were dramatically improved at the IT range by combining it with a highly conductive ESB phase. The electrode ASR measured from a symmetric cell consisting of LSM-ESB electrodes on an ESB electrolyte was only 0.08 ?-cm2 at 700oC which is ~60% lower than that of LSM-ESB on GDC electrolytes (0.19 ?-cm2). This exemplifies the synergetic effect the ESB phase has both in the cathode bulk and at the electrolyte/electrode interface. The MPDs of the anode-supported SOFCs with LSM-ESB cathodes on ESB/GDC bilayered electrolytes were ~836 mW/cm2 at 650 oC, which is the highest value reported for SOFCs using LSM-bismuth oxide composite cathodes.
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 Kang Lee.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Wachsman, Eric D.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-02-28

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042114:00001


This item has the following downloads:


Full Text





COMPREHENSIVE DEVELOPMENT OF HIGH PERFORMANCE SOLID OXIDE FUEL
CELLS FOR INTERMEDIATE AND LOW TEMPERATURE APPLICATIONS




















By

KANG TAEK LEE


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2010

































2010 Kang Taek Lee
































To my mom and wife









ACKNOWLEDGMENTS

First of all, I glorify God (and Jesus Christ), who is my savior, for this dissertation.

During last four year journey, I could not complete anything without His guidance and

protection. I strongly wish that this dissertation and all my works in the future would be a

confession of my faith in Him. For this dissertation, I wrote six experimental notes. On

the first pages for all these notes, it is written that 'Do not be deceived. God cannot be

mocked. A man reaps what he sows.' (Galatians 6:7). Relying on these words, I always

did try to do my best for all works. Moreover, I should say that this work would not have

been possible were it not for the support of many people.

I would like to thank my advisor, Prof. Eric D. Wachsman, for his support and

guidance. His encouragement helped me to reach a higher level of success and expand

my potential. I also would like to thank Prof. Juan C. Nino, Prof. Simon Phillpot, Prof.

Wolfgang Sigmund, Prof. Mark Orazem and Prof. Valentin Craciun for their advice,

guidance and constructive comments.

I also thank Dr. Heesung Yoon who taught me most of the processes of SOFC

fabrication. I also wish to acknowledge other former and current group members; Dr.

Keith Duncan, Dr. Takkeun Oh, Dr. Sean Bishop, Dr. Dongjo Oh, Byung-Wook Lee, Eric

Armstrong, Dr. Bryan Black Burn, Eric Macam and other members for providing me with

an excellent research environment and helpful comments. I would especially like to

thank Dr. Dohwon Jung, Dr. Matthew Camaratta, Dr. Jin Soo Ahn and Nick Vito for their

sincere friendship and co-working partnership.

I truly want to give my thanks to Kwanjeong Educational Foundation for

unconditional financial support during my entire doctoral research with a very honorable

scholarship.









It was also truly helpful and enjoyable for me to share the time with all my Timothy

group friends at the Korean Baptist Church of Gainesville. All worship services and

activities were great encouragement and support for me.

I would like to thank my family members. Most of all, I want to thank my wife's

parents, Dae Heun Kang and Jung Suk Kim for entrusting their precious and only

daughter to me and welcoming me into their family. I am also thoroughly grateful to my

father, Sam Soo Lee who is a great supporter of mine and who prayed for me through

this process. I also strongly pray to God for restoring his health. At this moment, I

strongly thank and miss my brother, Kang Yong Lee who is my mentor and supporter

with his endless trust and encouragement. I'd like to also mention his wife, Ji Eun Kim

and their precious daughter for their support.

Finally, I dedicate this dissertation to my mom and wife. My mother, Yang Soon

Kim has dedicated herself to me for over 30 years. She always trusted and encouraged

me in any circumstance. My beautiful and lovely wife, Yoo Jin Kang moved to here,

leaving her family and friends to be with me. She was by my side during my struggles

with endless patience and love. I love you so much! Thank you.









TABLE OF CONTENTS

page

ACKNOW LEDGMENTS .............................. ....... ..... .................. 4

LIST O F TA BLES ......... ................ ..................... ...... ............... 9

LIST O F FIG URES........................................... ............... 10

LIST OF ABBREVIATIONS............................................................. 14

A B S T R A C T ......... ...... ........... ................................. ........................... 16

CHAPTER

1 INTRO DUCTIO N ........... ......... ......... ..... .................... ........... 19

2 BACKGROUND ................ ......... ........ ..... ............... 23

2.1 Basic Principle of SOFC Operation........ ......................... ..... ............ 23
2.2 Actual SO FCs O operation .................................................. .... ..... ............... 24
2.2.1 Open-circuit Potential (OCP) and Transference Number (ti) ................... 25
2.2.2 Irreversible Losses ........ ........................................................ ... ...... 27
2.2.2.1 Activation polarization losses ........ ........ ..... .. ............... 27
2.2.2.2 Leakage current polarization losses................ ................... ...... 28
2.2.2.3 Ohm ic polarization .................................... ....... .. ............... 29
2.2.2.4 Concentration polarization ................................. ........ ... ...... 29
2.3. Materials and Design .................... ..... ............. 30
2.3.1 Stabilized Zirconia Electrolytes........................ ................... 30
2.3.2 Aliovalent Cations-Doped Ceria Electrolytes.......... ...... ..................... 31
2.3.3 Stabilized Bismuth Oxide Electrolytes ................ ........................ 32
2.3.4 Bilayered Electrolyte Concept for High Performance IT-SOFCs ............ 34
2.3.4.1 Ceria / Zirconia bilayer electrolyte ......... ................................. 34
2.3.4.2 Ceria / Bismuth oxide bilayer electrolyte.................................... 35

3 INTERGRATING NANO- AND MICRO- STRUCTURED ANODE FUNCTIONAL
LAYERS FOR IMPROVED IT-SOFC PERFORMANCE..................................... 44

3.1 Introduction ........................................ ............... 44
3.2 Experim mental ............... .................................... ............ ......... 47
3.3 R result and D discussion ............. .......................... ............... .............. 49
3.4 Conclusions .......................... ............. ................... 54

4 EFFECT OF NI-GDC AFL COMPOSITION ON PERFORMANCE OF IT-SOFCS 62

4.1 Introduction ........................................ ............... 62
4.2 Experimental .................. ......... ......... 64









4.2.1 C ell Fabrication ............................................................... 64
4.2.3 C haracterization ........................ ........ .......... .. ............................ 65
4.3 Results and Discussion.......................... ...... ........ ......... 66
4.3.1 M icrostructural Analysis............................. ........ ... ............... 66
4.3.2 Effect of AFL Composition on Power Density............................... 67
4.3.2.1 I-V characteristics at 650 C.......................................... ......... 67
4.3.2.2 Temperature dependence......................... .... ....... ..... 70
4.3.2.3 Long term stability ........................... ........ ......... .............. 71
4.3.2.4 Effect of AFL composition on ASR................ ....... .............. 71
4.4 Conclusions .............. ......... .......... ...... .. .......... ....... ........... 72

5 COMPREHENSIVE QUANTIFICATION OF NIO-GDC ANODE FUNCTIONAL
LAYER MICROSTRUCTURE BY THREE-DIMENSIONAL RECONSTRUCTION
USING FIB/SEM .............. ..... ....................... ........ .. ... ............ 84

5.1 Introduction ......................................... 84
5.2 Experimental .................. ..... .................. 85
5.3 Results and Discussion............................ ......... 87
5.4 Conclusions .............. ...... .... ... ...... .......... ............... .......... 93

6 HIGH PERFORMANCE IT-SOFC WITH CERIA/BISMUTH OXIDE BILAYERED
ELECTROLYTES FABRICATED BY A SIMPLE COLLOIDAL ROUTE USING
NANO-SIZED ESB POWDER ........................................... 107

6.1 Introduction ................ ... ......... ................. 107
6.2 Experim mental Procedure ............................................. .......................... 110
6.2.1 ESB Powder Fabrication .......... ....................... 110
6 .2 .2 F ue l C e ll F a bricatio n ............................. ................. ............. 111
6 .2 .3 C characterization ......................................... .............. ........... 112
6.3 R result and D discussion ......................................... .............. .............. 113
6.3.1 Powder Characterization ..................................................... ......... 113
6.3.2 Effect of Sintering Temperature on ESB/GDC Bilayered Electrolyte..... 114
6.3.3 Microstructure of a Full Button Cell with ESB/GDC Bilayered
E lectro lyte ............... ........ .. ... .. ........... .............. .................. 117
6.3.4 Performance of a Button Cell with ESB/GDC Bilayered Electrolyte ...... 118
6.4 Conclusions ........ .......... .............. .................. ... ............ 121

7 HIGH PERFORMANCE LSM-BASED CATHODE BOOSTED BY STABILIZED
BISMUTH OXIDE FOR LOW TO INTERMEDIATE TEMPERATURE SOFCS..... 131

7.1 Introduction ............................................................. 131
7.2 Experimental ........................ ......... ........... 133
7.2.1 Sam ple Fabrication...................... ............... 133
7.2.2 Characterization ........................ ........... .............. 135
7.3 R result and D discussion .......................................... ........................... 136
7.3.1 Impedance Spectroscopy for Symmetric Cells.............. ................ 136
7.3.2 I-V Characterization for Button Cells ...... ........ ..... .................. 140









7.4 Conclusions ................ ......... ................. 144

8 CONCLUSIONS ................ ......... ..................... 158

APPENDIX

A DEPENDENCE OF OCP ON GDC ELECTROLYTE THICKNESS.................... 162

B LONG TERM STABILITY FOR A SOFC WITH NI-GDC AFL .............................. 168

C EXPERIMENTAL SETUP ............. ............. ..... ............... 173

LIST OF REFERENCES .......... ............ ......... ................ ............... 175

BIOGRAPHICAL SKETCH .............. ............ ... ......................... 182









LIST OF TABLES


Table page

2-1 Calculated Po2 and Nernst voltage at open-circuit condition (T=500~7000C)..... 43

2-2 Conductivity Data for Stabilized ZrO2 Doped with Rare-Earth Oxides ............ 43

3-1 Detailed OCP, MPD and ASR values of the fuel cell samples with N+C-AFL,
C-AFL, no AFL at 6000C.................................... .......................... .. ..... 61

4-1 Detailed ASR values of the testing cells with various NiO contents in AFL ........ 83

5-1 3D reconstruction dimension and total volume fractions of Ni, GDC and pore
phase and solid volume fractions of Ni and GDC ............................................ 106

5-2 Summary of quantification of microstructural features of AFL with various
co m po sitio ns ................................................................. 106

6-1 Calcination condition and crystallite size of ESB powders synthesis by co-
precipitation and solid state route............................... .............. 130

6-2 Comparison of specification and electrochemical performance of the studied
ce lls. ......... .. ......... .. .............. ......... ......... ..................... 13 0

7-1 Detailed total, ohmic, and electrode ASR values for Cell-1 and Cell-2 at 650
C ...................... .. .. ......... .. .. ............................................ 1 5 7

A-1 Summary of sample description and OCP result.................... ............ 167









LIST OF FIGURES


Figure page

2-1 Schematic diagram of reactions in SOFCs based on oxygen-ion conductors .... 37

2-2 SOFC current-voltage behavior indicating relative polarization losses............... 38

2-3 Variation of ionic conductivity of stabilized ZrO2 with dopant concentration
(T=8070C) .............. ..... ....... ........... .. ...... .. 39

2-4 Conductivities of selected electrolyte materials ............. .............. 40

2-5 ESB conductivity versus Po2 in purified argon atmosphere.............................41

2-6 Conceptual representation of a bilayer electrolyte showing the effect of
relative thickness on interfacial oxygen partial pressure (Po2 )....................... 41

2-7 Bulk electrolyte ASR at 500 OC as a function of relative (t = LESB/LSDC) and
total thickness for bilayers .............................................................. .... ..... 42

3-1 Schematic illustration of the proposed N+C-AFL structure on anode-
supported SOFC and effect of N+C-AFL on expending TPB length. Yellow
triangles represent TPBs in conventional AFL (C-AFL) and red triangles. ......... 55

3-2 SEM micrographs of the anode surface after deposition and pre-sintering (a,
c, e) and after full sintering followed by simulated testing atmospheric
conditions (b, d, f) for samples with no AFL (a, b), C-AFL (c, d) ................... 56

3-3 Comparison of I-V characteristics for the fuel cell samples with N+C AFL, C-
AFL, and no AFL at 600 oC. (a) I-V plots at the temperature ranging from 650
to 500 oC for N+C-AFL (b), C-AFL (c), and no AFL (d). .............. ............... ... 57

3-4 Electrochemical impedance spectra of the testing samples with N+C AFL, C-
AFL and no AFL at various temperature; 650 oC (a), 600 oC (b), 550 oC (c),
and 500 C (d). ...................................... ......... .. ................... ........ ............ 59

3-5 MPD (a) and ASR plots (b) for the different samples tested between 500 and
650 C .............. ............. ..................................... ............. ...... 60

4-1 Backscattered images showing a cross-sectional view of anode-supported
SOFCs with different NiO content in the anode functional layers; no AFL(a),
40wt% (b), 50wt%(c), 60wt%(d), 65wt%(e), and 80wt%(f) NiO......................... 74

4-2 Magnified microstructures of the anode or AFLs with different NiO content no
AFL(a), 40wt% (b), 50wt%(c), 60wt%(d), 65wt%(e), and 80wt%(f) NiO.
Backscattering mode provides better contrast to distinguish Ni (dark gray)....... 75









4-3 I-V plots of fuel cells with various AFL compositions at 6500C; 40(4), 50(A),
60(*), 65(*) and 80(1)wt% of NiO in AFL, and no AFL(m). The gas
condition was 90sccm of air and 3% of wet hydrogen on the anode ............... 76

4-4 Open circuit potential of the fuel cells with various NiO contents in NiO-GDC
AFL. Solid line (red) shows linear fit of the measured data (square) ................ 77

4-5 MPD (Red square) and total ASRiv estimated from IV curves (blue star) are
plotted with NiO contents in AFL. The open symbols represent no AFL cell...... 78

4-6 Maximum power densities of fuel cells with various AFL compositions at the
temperature range from 450 to 650 C. Open symbols represent MPD of no
AFL cell at each temperature...................................................... 79

4-7 Long term stability test of fuel cell with 60wt% of NiO in the AFL and the no
AFL cell for 200 hrs at 650 C. Potentiostatic tests were conducted with an
applied voltage of 0.379 V for the NiO 60wt% AFL cell and 0.380 V ................. 80

4-8 Impedance spectra with various AFL compositions (a), and total, electrode,
and ohmic ASRs of fuel cells with different NiO content (b) calculated from
impedance spectra (a). Open symbols represent no AFL results .................. 81

4-9 MPD plots with electrode ASR shows a linear relationship Red line is linear
fitting of the measured data (black dots) .. .............. .. ................. ................... 82

5-1 Schematic diagram of FIB/SEM dual beam system with sample (a) and 3D
reconstruction process (b) ............................... ... ... ......................... .. 95

5-2 3D reconstruction of Ni-GDC anode (a), and AFLs with initial composition of
50 (b), 60 (c), 65 (d), and 80 (e)wt% NiO nearby at anode(or AFL)/electrolyte
interface ....................... ...... ... .. ............. ... ............... ........ 96

5-3 Individually reconstructed phases from the 3D reconstruction of AFL with 65
wt% NiO ; GDC (a), Ni (b), Pore (c), and combination of Ni and Pore phases ... 97

5-4 Phase gradient of reconstructed samples with no AFL (a), 50 (b), 60 (c), 65
(d), and 80 (e)wt% NiO in Ni-GDC AFL ......... ... .......... ............... ......... 98

5-5 Volume fraction of Ni, GDC and pore phase in total volume (a), and volume
fraction of Ni and GDC in solid volume of AFLs with various compositions.
Open symbols represent theoretical values............ ................................. 101

5-6 Effective particle diameters of Ni (rectangular), GDC (circle), and pore
(triangle) phase of AFLs with various compositions.................................... 102









5-7 Schematic diagram of TPB length calculation from 3D reconstruction. A
rectangular parallelepiped represents a voxel in a 3D reconstruction and
each one is labeled as one of phases; Ni, GDC, or Pore phase.................. 103

5-8 Plot of quantified surface area and TPB density of AFL with various NiO
contents. Dotted lines are only for guide purpose................. ... ............ 104

5-9 (a) TPB density and electrode ASR with various AFL compositions (Dotted
lines are only for guide purpose.) (b) plot of 1 over electrode ASR with TPB
density. A red line represents linear fit for the plot showing inverse ................ 105

6-1 XRD diffraction pattern of ESB powders synthesized by coprecipitation route
(red line) and solid state route (black line) (a). The magnified XRD diffraction
pattern of the (111) peak is shown at the 20 range from 27 to 290 (b)............. 123

6-2 SEM of ESB powders synthesized by wet-chemical co-precipitation method
(a) and solid-state route (b). ........... ...... ...... ......................... 124

6-3 Evolution of ESB layers on GDC electrolyte at various sintering temperatures
using ss-ESB (a,c,e,g) and cp-ESB (b,d,f,h). It is noted that the magnification
of images for ESB electrolyte using cp-ESB powder is higher. ........................ 125

6-4 Cross-sectional view of GDC electrolyte under ss-ESB layer after sintering at
900 C, which is the magnified image from Fig. 5-3g. In backscattering mode,
ESB (white), GDC (light gray), and NiO (dark gray) phases are well .............. 126

6-5 Cross-sectional SEM image of a full button cell with ESB/GDC bilayered
electrolyte. EDX line scan was conducted along the straight base line
(yellow) and the intensity of each elements are presented as red (Bi) ............ 126

6-6 SEM image of cross-sectional view of a single GDC electrolyte cell (a) and
ESB/GDC bilayered cell (b). Surface views of GDC electrolyte (c) and ESB
electrolyte (d) are shown. ............................................................. .. .... .... 127

6-7 IV-Characteristics (a) and impedance spectra (b) of ESB/GDC bilayer (red
square) and GDC single layer (blue triangle) cell ............. ......... .......... 129

7-1 XRD pattern of LSM, ESB, and LSM+ESB (50:50wt%) powers before and
after annealing at 900 C 50 hrs ....... .... ........................................... ........ 145

7-2 SEM images of LSM-ESB cathode on GDC electrolyte (a) and ESB
electrolyte (b). The insets are backscattered images. ............................... 146

7-3 Impedance spectra of the LSM-ESB cathode on ESB and GDC pellets at the
temperature ranges from 500 to 700 OC ........ .... ............. ................... 147

7-4 Electrode ASRs of LSM-ESB cathode on GDC (blue circles) and ESB (red
squares) electrolytes, and ASR reduction rate (black stars) .......................... 148









7-5 Comparison of the electrode polarization resistance of LSM-bismuth oxide
cathodes at IT ranges. LBSM is short for Lao.74Bio.10Sro.16MnO3- ..................... 149

7-6 Long term stability test of LSM-ESB cathode on ESB electrolyte at 700 C for
100 hours. ................ ...... .......... .............. .. ................. 150

7-7 SEM images of cross-sectional views for cell-1 (a) and cell-2(b), and surface
views for cell-1 (c) and cell-2 (d)... ......................... ............... 151

7-8 I-V characteristics (a) and impedance spectroscopy (b) of cell-1 and cell-2 at
650 C ............................................... .......... 153

7-9 I-V characteristics at various temperature for cell-1 (a) and cell-2 (b) .............. 154

7-10 Maximum power density improvement (black stars) of cell-2 (red squares) at
various temperatures compared to cell-1 (blue circles) ............... .............. 155

7-11 Comparison of maximum power density of SOFCs using LSM-bismuth oxide
composite cathodes at IT ranges................ .... ....................................... 156

A-1 Microstructures of Ni-GDC anode/GDC electrolyte/LSCF cathode SOFCs
with various electrolyte thicknesses............................... ............. ........... 164

A-2 Experimental OCP values from electrochemical test at 500, 600, and 650 C
as a function of GDC electrolyte thickness ....... ..... ............................. 165

A-3 Fit of the OCP model (eq. A-1) to experimental data for OCP as a function of
electrolyte thickness. ...................... ................... .. ............... 166

B-1 Long term stability test of a SOFC with 60wt% NiO in Ni-GDC AFL. The
potentiostatic test was conducted at 650 OC for 600 hours under an applied
voltage of 0.379 V ...... .. .. ........... ......... ....................... ............ ........ 170

B-2 Comparison of I-V plots of the testing sample between before long term test
and after long term test for 600 hours at 6500C(a), 6000C(b), 5500C(c),
5000C(d). ............................................ 171

C-1 Schematic SOFC testing setup a button cell testing setup configuration and
I-V and EIS testing equipment ....................... ...... ............. ............... 173

C-2 Illustration of symmetric cell configuration for EIS test (top) and EIS testing
se tu p (b o tto m ) ............................................................................ 17 4









LIST OF ABBREVIATIONS

AFL Anode functional layer

ASR Area specific resistance

BET Brunauer-Emmett-Teller method

BRO Bismuth ruthenate

BSCF Barium strontium cobalt ferrite

DBP Di-butyl phthalate

EDX Energy dispersive X-ray analysis

EIS Electrochemical impedance spectroscopy

ESB Erbium stabilized bismuth oxide

FIB Focused ion beam

FWHM Full width at half maximum

GDC Gadolina doped ceria

IT Intermediate temperature

LMIS Liquid metal-organic ion source

LSCF Lanthanum strontium cobalt ferrite

LSM Lanthanum strontium manganite

MIEC Mixed ionic-electronic conductor

MPD Maximum power density

OCP Open circuit potential

PLD Pulsed laser deposition

PVB Polyvinyl butyral

ROI Region of interest

ScSZ Scandia stabilized zirconia

SDC Samaria doped ceria









SEM Scanning electron microscope

SOFC Solid oxide fuel cell

TLD Through-lens-detector

XRD X-ray diffraction

YSB Yttrium stabilized bismuth oxide

YSZ Yttrium stabilized zirconia









Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

COMPREHENSIVE DEVELOPMENT OF HIGH PERFORMANCE SOLID OXIDE FUEL
CELLS FOR INTERMEDIATE AND LOW TEMPERATURE APPLICATIONS

By

Kang Taek Lee

August 2010

Chair: Eric. D. Wachsman
Major: Materials Science and Engineering

In order to develop high performance solid oxide fuel cells (SOFCs) operating at

low to intermediate temperatures, the three main SOFC components--the anode,

electrolyte, and cathode--were comprehensively studied.

In order to lower anodic polarization losses in anode-supported SOFCs, a novel

composite anode functional layer (AFL) having bimodal (nano/micro) structure was

developed. Application of this AFL involved a simple process where a precursor solution

was coated onto a conventional submicron sized colloidally-deposited Ni-GDC AFL.

Cells prepared in this manner yielded maximum power densities (MPD) of 1.29, 1.16,

0.7 and 0.38 W/cm2 at 650, 600, 550 and 500 OC, respectively. Electrochemical

impedance results showed a striking decrease in both ohmic and non-ohmic area

specific resistances (ASRs) for these cells compared to those with either no AFL, or a

conventional AFL.

In addition, the effect composition of the conventional submironsized AFL on

performance was examined. The highest MPD (1.15 W/cm2 at 650 oC) was achieved at

a composition of 60wt% NiO. This composition had the best performance over the

intermediate temperature range (450 to 650 oC). For the potentio-static test, the cell









exhibited stable performance over 200 hrs of operation at 1.1 W/cm2. It was also

revealed that electrode ASR has an inverse linear relationship with maximum power

density at 6500C.

To better understand the effect of AFL composition, microstructural features of

AFLs having various Ni-GDC compositions were quantified by a 3 dimensional (3D)

reconstruction technique using a FIB/SEM dual beam system. Of the compositions

tested, the highest triple phase boundary (TPB) density was achieved at 60wt% NiO,

which corresponds to a 1:1 volume ratio of Ni to GDC phase. The quantified TPB

density showed an inverse proportionality to electrode ASR.

Using a wet chemical co-precipitation method, nano-sized ESB particles were

successfully synthesized at temperatures as low as ~ 500 C. Due to the high

sinterability of this powder, a dense erbia stabilized bismuth oxide (ESB) layer was

successfully formed on a gadolinia doped ceria (GDC) electrolyte by a simple colloidal

coating method. A systematic study on the sintering behavior of ESB was conducted to

determine the optimum sintering conditions for these materials. I-V measurement a cell

using this bilayered electrolyte system showed a high power density ( ~ 1.5 W/cm2) at

650 C due to an enhancement in OCP and a significant reduction in ASR when

compared to a GDC single cell.

The performance of conventional (Lao.8oSro.20)MnO3-6 (LSM) cathodes were

dramatically improved at the IT range by combining it with a highly conductive ESB

phase. The electrode ASR measured from a symmetric cell consisting of LSM-ESB

electrodes on an ESB electrolyte was only 0.08 Q-cm2 at 7000C which is ~60% lower

than that of LSM-ESB on GDC electrolytes (0.19 Q-cm2). This exemplifies the









synergetic effect the ESB phase has both in the cathode bulk and at the

electrolyte/electrode interface. The MPDs of the anode-supported SOFCs with LSM-

ESB cathodes on ESB/GDC bilayered electrolytes were -836 mW/cm2 at 650 C, which

is the highest value reported for SOFCs using LSM-bismuth oxide composite cathodes.









CHAPTER 1
INTRODUCTION

A fuel cell is an energy conversion device which directly produces electrical energy

from the chemical energy contained in various fuels by electrochemical reactions. In

1838, the basic principle of the fuel cell was written in one of the scientific magazines of

the time by German chemist, Christian Friedrich Schonbein. One month later, Sir

William Grove reported the first functional fuel cell in 1839. He used a dilute sulfuric acid

solution as an electrolyte at room temperature, which produced water and electricity[1].

However the history of SOFCs began much later, in 1899, with the discovery of the

solid-oxide electrolyte by Nernst and followed with the first SOFC invented by Baur and

Peris in 1937 [2].

Since that time, and especially in the last several decades, tremendous effort and

progress has been made to commercialize SOFCs. For instance, Siemens-

Westinghouse has successfully developed and operated a 100 kW system for over

20,000 h without significant deterioration in performance [3]. Recently the Solid State

Energy Conversion Alliance (SECA), the fuel cell program under United States

Department of energy (DOE), announced their road map, including the development of

a prototype SOFC stack with megawatt capability and fuel-flexiblity by 2015 [4].

One of the biggest challenges to SOFC commercialization is to reduce the

operation temperature while maintaining high power densities. At intermediate

temperatures (IT, 500~700 OC), the system cost can be significantly reduced by allowing

the use of cheap stainless steel for the bipolar plates and the balance-of-plant, as well

as the use of high temperature gaskets rather than rigid glass-based seals, which can

also enhance mechanical stability and life time [5].









Conventional SOFCs with yttria stabilized zirconia (YSZ) electrolytes operate at

high temperatures (over ~10000C) due to its thermally activated ionic conduction, and

thus have unacceptable system cost and slow start-up times [4-7]. Two main strategies

have been studied to reduce ohmic losses in the electrolyte at reduced temperatures.

First is the thin electrolyte approach--the electrolyte resistance is inversely

proportional to the electrolyte thickness [8]. To accommodate thin electrolyte films,

anode supported cells has been developed [5, 9]. In this configuration however, anodic

polarization can limit performance due to the relatively high anode thickness. It is

believed that most fuel oxidation reactions take place at near the anode/electrolyte

interface, indicating that most anodic losses occur in this region [10]. Therefore,

engineering of the interfacial region has received much attention as a way to reduce

losses at the anode [11-14].

The other approach is to use materials with enhanced ionic conductivity. For

example, erbia stabilized bismuth oxide (ESB) and gadolinia doped ceria (GDC) have

one to two orders of magnitude higher ionic conductivity in the IT range than YSZ [15].

However, these two materials have disadvantages, including thermodynamic instability

at the low Po2 conditions experienced at the anode side of fuel cell systems [3, 16, 17].

To overcome these limitations, a bismuth oxide/ceria bilayer electrolyte concept has

been proposed [18]. In order to produce high power densities in the low to intermediate

temperature range, one can combine the bilayer electrolyte concept with a thin film

approach. Using this concept with a thin and dense ESB (~ 4 pm) and GDC (~ 10 pm)

bilayered electrolyte, the author and colleagues recently demonstrated a cell having an

exceptionally high power density of ~ 2 W/cm2 at 6500C [19, 20]. In that study, the









dense ESB layer was deposited by pulsed laser deposition (PLD), which is not a viable

technique for mass production. Therefore, a more simple and cost-effective fabrication

process is necessary.

The focus of this dissertation is the development of SOFCs producing high power

densities in the IT range and prepared with practical and cost-effective fabrication

processes. Each component of the SOFC-- the anode, electrolyte, and cathode--was

investigated to reduce its major polarization losses. In order to control anodic

polarization losses, a novel AFL was developed at the anode/electrolyte interface by

integrating nano- and micron- particle structures. For further improvement of the Ni-

GDC AFL, the effect of composition was carried out. Microstructural features of the

AFLs were quantified using a state-of-the-art 3D reconstruction technique by a FIB/SEM

dual beam system. From this work, the understanding of the relationship between

electrochemical performance and microstructures was enhanced.

In order to improve electrolyte performance, the ESB/GDC bilayered electrolyte

system was investigated. Cost-effective fabrication of dense ESB electrolytes was

achieved by a simple colloidal deposition technique. In order to accomplish this, nano-

sized ESB particles with high sinterability were synthesized by a wet chemical co-

precipitation method. The reproducibility of the high performance exhibited by these

bilayered electrolyte cells was carefully demonstrated.

In addition, an (Lao.8Sro.2)o.9MnO3-6 (LSM)-ESB composite cathode was studied as

an alternative cathode for low to intermediate temperatures. The use of conventional

LSM cathodes has been limited to high temperature SOFCs due to its low ionic

conduction at reduced temperatures [21]. In this work LSM was mixed with the fast ion









conductor, ESB. The performance of the LSM-ESB cathode was investigated in the IT

range. This work demonstrated that, coupled to an ESB electrolyte, the performance of

LSM-ESB was stable and significantly better than that of the same cathode on

conventional GDC or YSZ electrolytes below 650 OC.









CHAPTER 2
BACKGROUND

2.1 Basic Principle of SOFC Operation

In principle, overall SOFC reaction is expressed as a simple reaction formula;

02 + H2 H20 (2-1)
2

In order to complete this reaction in the actual fuel cell operation, the reaction is

divided into two half cell reactions [2, 3, 9]. Fig. 2-1 shows a schematic diagram of a

general SOFC structure and half cell reactions at anode and cathode side [22]. In the

cathode side, oxygen which generally comes from air is reduced to 02- with electrons

provided from outside fuel cell which written as;

1
0, +2e 02 (2-2)
2

For this half reaction, a cathode conducts adsorption of oxygen molecules and

dissociation of adsorbed oxygen. This is followed by formation of oxygen ion by electron

transfer and charge transfer (02- and electron) at the triple phase boundaries (TPBs)

between gas, ionic, and electronic conducting phases. Therefore, a cathode should be a

high catalysis to dissociate molecules and have high ionic and electronic conduction

with good compatibility to the electrolyte as well. Related work with high catalytic

cathode for IT-SOFC has be done at ch. 7 in this dissertation

Transferred Oxygen ions move to anode via electrolyte, for which high ionic

conductivity of the electrolyte is necessary. A driving force of the ion migration is the

Nernst potential due to Po2 difference between cathode and anode. Therefore, the

electrolyte should be a 'good barrier' between the air side and the fuel side to maintain

the low Po2 at the anode, which can be achieved by highly dense electrolyte. Highly









dense electrolyte with a special design for high stability and conductivity was

researched at ch. 6.

At the anode, the fuel is oxidized with the migrated oxygen ions.

H2 +02- H20 + 2e (2-3)

CO +02 CO2 + 2e (2-4)

CH2n+2 + (3n +1)02 nCO2 + (n + 1)H20 + (6n + 2)e (2-5)

Each equation presents the oxidation reaction for different fuels; eq. (2-3) for H2, eq. (2-

4) for CO, and eq. (2-5) for hydrocarbon fuel. In this case, the catalytic properties of

anode is important, which takes place at TPBs. Therefore, concentration and spatial

distribution of TPB can be a key factor to improve anode performance for SOFCs. In

this dissertation, control and mechanism of TPB extension in anode nearby the anode

and electrolyte interface was intensively studied (ch. 3 ~5).

2.2 Actual SOFCs Operation

In actual SOFCs operation condition, the performance of SOFCs is commonly

measured by voltage out as a function of applied current density. Fig. 2-2 shows a

representative voltage-current plot of SOFC [4]. As mentioned above, the Nernst

potential by Po2 difference between anode and cathode produces a driving force to

operate SOFCs. However, this ideal voltage can not be maintained under applied

current due to various irreversible polarization mechanisms. The actual operational cell

voltage (E) as a function of current density can be written as;

E = EOCP 7act 7ohm 17conc (2-6)









where EocP is open circuit potential including leakage current and r1act, ilohm and

r7conc represent the activation, ohmic, and concentration polarization, respectively.

Detailed polarization mechanisms are explained in following sub-sections.

2.2.1 Open-circuit Potential (OCP) and Transference Number (ti)

The voltage is generated across a cell by various gas mixtures with two different

oxygen partial pressures (Po2). The open-circuit potential for the oxygen potential

gradient cells is given by the well-known Nernst equation in a cell [3],

RT Po'
E = In (2-7)
4F Po2"

where Po2" and Po2' are the equilibrium partial pressure of oxygen at the two sides of

the cell, R is the ideal gas constant, F is Faraday's constant and T is the absolute

temperature. The reaction for oxygen is,

O2(g)+4e -> 202 (2-8)

If one knows the oxygen partial pressure of the reference electrode (Po2") and

measures the OCP at a given temperature, the equilibrium value of oxygen pressure at

the working electrode may be determined from eq. (2-7). Generally, low oxygen

pressure can be easily obtained under CO/C02 or H2/H20 mixture [23]. However,

previous results showed that in CO/C02 gas mixtures, the equilibrium was not readily

attained, while H2/H20 gas mixtures showed equilibration for low Po2. In this study,

H2/H20 gas mixtures were used to maintain low oxygen partial pressure on the anode

side. The hydrogen was bubbled with 3% of H20 through a membrane submersed in

water. The condition of a controlled oxygen partial pressure can be obtained via

thermodynamic relations at equilibrium. At high temperature, H2 and H20 gases react

with traces of oxygen as following.









H2(g)+ 2 0-> H20 (2-9)



And the Gibbs free energy is

AG = AGO(T)+RT1n H20 (2-10)
PH2 PO2

where AGO (T) is the standard Gibbs free energy of the reaction, R is ideal gas

constant, and T is temperature. From the thermodynamic data [23],

AG(T) = -242,000(Jol)+ 44.7(mo K)x T(K) (2-11)

At equilibrium, AG = 0

AG(T) =-RT1n HO0 (2-12)
PH Po2


Po2 = PH exp(AGo ))2 (2-13)
PH, (2-13)

Therefore, with the reference oxygen potential, temperature, and H2/H20 partial

pressures, the controlled oxygen partial pressure can be calculated. For example, the

experimental condition can be maintained with reference oxygen partial pressure of

0.21 atm on the cathode side (1 atm air) and 3% of H20 with H2 on the anode side.

Based on these experimental conditions, the theoretical Nernst voltage was calculated

by eq. (2-7) to (2-13) over the temperature range of 500 to 700 C in 50 C increments

as shown in Table 2-1.

The transference number (or transport number), ti, is defined as the fraction of the

total conductivity due to each charged species;









t = (2-14)
Total

The ionic transference number, ti, is equal to 1 for a purely ionic conductor, such as

YSZ. Approaching unity of tj means that there is no significant electron or hole

conduction. Any electronic conductivity causes an internal short circuit in the electrolyte

of a fuel cell. The ratio between the measured and the theoretical OCP or Nernst

voltage is transference number (ti);

t measured (2-15)
OCPtheorectcal

The transference number will be evaluated from the experimentally measured

OCP values with the theoretical values in Table 2-1.

2.2.2 Irreversible Losses

2.2.2.1 Activation polarization losses

At low current density condition, the slow reaction kinetics at the cathode and the

anode can cause the activation polarization. In other words, the excessive energy to

overcome a energy barrier for electrode reaction such as the oxygen reduction and

hydrogen oxidation produces the voltage drop which is increased with the current

density drawn.

The phenomena can be formulated by the Butler-Volmer equation [24];


i = io, exp aanta '- exp a CnF a (2-16)
oe RT RT

S= (anFi ( -aenFri t
i = exp -Ract expy ac-nFct, (2-17)
RT RT









where io,a, io,c are the exchange current densities of the anode and cathode,

respectively, 0a,a and ac,a are the anodic (i>0) and cathodic (i<0) charge transfer

coefficients of the anode, and a,,c and ac,c are the anodic and cathodic charge transfer

coefficients of the cathode. The charge transfer coefficients depend on the

electrocatalytic reaction mechanism and usually ~0.5 is used for SOFCs It is noted that

as shown above, the Butler-Volmer equation should be applied for each electrode

separately. This non linear Butler-Volmer equation can be simplified as;


Fact iRoIn0 (2-18)
a nF \ i

In this case, this equation only considers on forward direction reaction, that is,

reduction at the cathode side and oxidation at the anode side. Surprisingly, this

simplified form was already predicted in empirical equation by Tafel in 1905 [25], which

is written as;


7a =alogln i (2-19)


where a is a called the Tafel slope.

2.2.2.2 Leakage current polarization losses

In principle, voltage measurement of SOFC under open circuit condition (OCP)

should show the theoretical Nernst voltage at the testing temperature. However, the

actual measurement of OCP usually has some deviation from the theoretical voltage.

Even YSZ known as purely ionic conductor still sometimes shows some OCP deviation,

which might be comes from gas leaks across the electrolyte itself or poor seal. For

mixed ionic and electronic conductors (MIECs), such as dope-ceria, the partial

electronic conduction causes OCP drop from theoretical one. For example, the









theoretical OCP at 650 C is ~1.14V but the reported OCPs of the GDC electrolyte cells

are 0.7 ~ 0.8V [7]. As explained above, the ratio of measured OCP to theoretical one is

expressed as transference number, ti.

2.2.2.3 Ohmic polarization

The ohmic polarization is the losses due to total electrical resistances from

electrodes, electrolyte and lead wires. This polarization loss simply follows Ohm's law

(V=IR). Therefore, ohmic polarization which is function of current can be expressed by

Uohm = I(Relectrode + Relectrolyte + Rcontac) (2-20)

where Relectrode, Relectrolyte, and Rcontact are resistances from electrode (both cathode and

anode), electrolyte, and electrode-electrolyte contact, respectively, and I is current

density. For SOFC structure, it has been generally accepted that most of the ohmic

resistance comes from electrolyte due to much slow conduction process of ion migration

rather than that of electrons. In this case, comparison of resistance between electrolytes

is difficult due to its thickness dependence. Moreover, performance of SOFC is

measured as a function of current density (i) not current (I). Therefore, ohmic

polarization of electrochemical devices is generally expressed by;

Uohm = I Rot = i A. REo = i ASRto (2-21)

where ASRtot is total area specific resistance.

2.2.2.4 Concentration polarization

Concentration polarization is generally observed at high current density regime of

I-V curves due to restriction to the transport of the fuel gas molecules to the anodic

reaction site. At high current density, excess water byproduct can block the reaction

sites. Therefore significant deactivation of reaction sites can be occurred. The









concentration polarization can be alleviated by higher gas pressure to drive out excess

water from the reaction sites, reduction of anode thickness to shorten the distance to

electrolyte, or higher porosity formation with same reason.

2.3. Materials and Design

2.3.1 Stabilized Zirconia Electrolytes

For high temperature SOFCs, stabilized zirconia, such as yittria-stabilized zironia

(YSZ) has been most widely used as an electrolyte due to its high stability, reasonable

ionic conductivity at high temperature (> ~900 C), and relatively inexpensive cost [6].

Although pure zirconia (ZrO2) is also chemically stable in both oxidizing and reducing

conditions, it has not been chosen as a solid electrolyte due to its poor ionic

conductivity. In addition to low conduction, pure ZrO2 shows phase transition from

monoclinic to tetragonal and from tetragonal to cubic fluorite at 1170 and 2370 OC,

respectively, accompanying unacceptable volume change (3 ~ 5%) in the fabrication

temperature ranges [26]. However, it has been known that some aliovalent cations,

such as cations of Ca, Y, Mg and Sc, can stabilize the ZrO2 phase as cubic fluorite

structure from room temperature to high temperature [27]. Moreover, this aliovalent

cation doping in ZrO2 produces higher vacancy concentration, leading higher ionic

conduction at the wider Po2 ranges. This aliovalent dopant effect on vacancy

concentration can be explained by Kroger-Vink notation in which the negative charge

produced by substituting a dopant is indicated by prime or a superscript dot if it is

positive. The amount of charge is indicated by the number or prime or dot. For neutrality

after substituting, it is marked as the superscript 'x'. For example, the incorporation

reaction between trivalent dopant and ZrO2 can be written as [3];









ZrO2
M203> 2M'z, +30x + V"


(2-22)


where the M is trivalent dopant and V is vacancy, indicating that two M dopants

produce one oxygen vacancy. Various tri- and divalent dopants has been studied to

make the stabilize Zr02 with high ionic conduction. It was shown that there is certain

dopant concentration to give maximum conductivity. For yttria stabilized zirconia (YSZ),

as Y dopant concentration increases, the conductivity is increased upto 8 mol%, while

over 8 mol% it shows degradation [28]. As shown in Fig. 2-3 Most of the stabilized

zirconia shows similar trend [2]. It is explained that at higher dopant concentration,

defect ordering or vacancy clustering occurs leading the reduction of total number of

active vacancy [29]. Among the dopants, Y is most widely used due to cost and stability,

while highest conductivity has been reported for Sc dopant (Table 2-2) [29]. The ionic

conductivity of YSZ is strongly depends on the concentration and mobility of ions, which

is known to a thermally activated process. Therefore, conductivity of YSZ suffers

significant conductivity reduction at low temperature, which limits it operational

temperature ~1000 OC.

2.3.2 Aliovalent Cations-Doped Ceria Electrolytes

Recently, aliovalent cations-doped ceria (Ce02) has been given much attention as

a potential solid electrolyte because of its higher ionic conductivity over a range of high

to intermediate temperature. The ionic conductivity of ceria is considerably increased by

aliovalent cation doping which increases the oxygen vacancy concentration in ceria [30,

31]. The magnitude of electrical conductivity and the stability under reducing conditions

for ceria-based oxides depend greatly on the kind and quantity of doping elements.

Alkaline earth oxides(e.g. CaO and SrO) and rare earth oxides(e.g. Gd203 and Sm203)









are highly soluble in the ceria sublattice. Among these, Sm and Gd-doped ceria shows

the highest electrical conductivity in ceria based oxides with 10~20% of dopant

concentration [31]. It is considered because of the similar ionic radii of Sm+3 and Gd+3 to

that of Ce+4. Since Steele calmed the ionic conductivity of Gdo.iCeo.901.95(10GDC) is the

highest among various Gd and Sm dopant concentrations at a temperature range of

500~7000C, 10GDC has been paid attention as one of the most suitable candidates for

IT-SOFCs electrolyte [7]. Moreover, as shown in Fig. 2-4 the ionic conductivity of

doped-ceria is approximately one to two order of magnitude greater than that of

stabilized zirconia, which is most widely used electrolyte material up to present [3]. It is

considered because Ce4+ (0.87 A) ion has larger ionic radius that Zr4+ (0.72 A) causing

easier oxygen ion migration through a more open structure. However, one big drawback

of ceria-based electrolyte makes us hesitate to select it as a best electrolyte material for

IT-SOFCs despite of its high ionic conductivity. When ceria-based oxides are reduced

at low oxygen partial pressures (<10-14 atm), Ce4+ transfers into Ce3+ leading significant

n-type electronic conduction with a P(02)-1/4 dependence [7]. This phenomenon reduces

the ionic transference number (ti) and the open circuit potential (OCP), thereby making

ceria less efficient for application as an IT-SOFCs. To increase the electrolyte domain

and to preserve the ionic conductivity of the doped ceria by any means is important.

2.3.3 Stabilized Bismuth Oxide Electrolytes

Various polymorphism in bismuth oxide based materials have been identified with

a, (3, y and 5 phases [32]. Even though the cubic 5 phase at high temperature (> 729

OC) shows an excellent ionic conductivity, attributed to the presence of such a large

concentration of oxygen vacancies, it is unstable and transforms into a monoclinic

phase with below 729 OC resulting in a discontinuous decrease in conductivity [33].









However, when a solid solution of bismuth oxide is formed with erbia(Er203) or several

other rare-earth oxides, bismuth oxide is known to remain stable in the cubic phase.

20% erbia stabilized oxide (ESB) has the excellent conductivity among the stabilized-

bismuth oxides. The greater conductivity of stabilized-bismuth oxide electrolytes has

tremendous potential for lower operating temperature, thus considerably growing the

number of applications for SOFCs [34].Despite the high conductivity of stabilized

bismuth oxide electrolytes, they have not been used in solid electrochemical devices

such as SOFCs due to their thermodynamic instability. Takahashi et al. indicated that

the critical PO2 value below which stabilized bismuth oxide would decompose is the

equilibrium oxygen pressure of a Bi/Bi203 mixture [16]. The decomposition process may

be simplified as:

3
Bi203 = 2Bi(s) + (g) (2-23)
2

The open-circuit potential from galvanic cells with an air cathode and metal/metal

oxide anode were stable in oxygen partial pressures above 10-131 atm at 6000C. This

result showed that there was no contribution of electronic conduction to the total

conductivity above the equilibrium oxygen potential of Bi/Bi203 mixture. Therefore, they

concluded that the minimum oxygen partial pressure at 6000C is 10-13. atm. On the

other hand, Wang and other researches reported the ionic conductivity of Bi203 could

be measured without critical decomposition of Bi/Bi203 under an H2/H20 atmosphere

[35, 36]. Wachsman et al. reported that the measured conductivity of ESB was

independent of Po2 over the range 1 to 10-22 atm under 02/Ar atmosphere as shown in

Fig. 2-5 while ESB was decomposed at low Po2(10-21atm) with H2, which was also

confirmed by XRD [37]. It was considered that the stability of ESB in the Ar/O2









atmosphere is probably due to slow heterogeneous kinetics in the absence of an active

reducing agent, such as H2. From these results, it can considered that bismuth oxide

based electrolytes may be kinetically stable in the absence of contact with an active

reducing agent. Therefore, in order to use bismuth oxide as an IT-SOFC electrolyte,

exposure to the reducing environment of the fuel gases must be prevented due to

decomposition of the bismuth oxide.

2.3.4 Bilayered Electrolyte Concept for High Performance IT-SOFCs

2.3.4.1 Ceria / Zirconia bilayer electrolyte

Bilayered electrolytes have been proposed as an alternative of overcoming the

decomposition by the thermodynamic instability of highly conductive oxides. Yahiro et

al. demonstrated that a thin and dense layer of YSZ on the fuel side of ceria avoided the

effect of reduction of electrolytes by blocking the electronic conduction and

consequently increased the OCP and power density [38]. Alternatively, some

researchers have suggested placing the YSZ layer, which has a low electronic

conductivity, on the air (i.e., oxidizing) side of the SOFC where its function is only to

block electronic flux (thereby increasing the efficiency of the SOFC) [23]. Of the two

approaches the latter has been the most successful. However, in both cases the YSZ

layer could not be made thin enough for the total ionic conductance of the bilayer to be

high enough for efficient power generation at low temperatures. Generally, a YSZ/SDC

or SDC/YSZ bilayer electrolyte has no intrinsic advantage over just a thin YSZ

electrolyte itself, other than providing a non-porous substrate for YSZ deposition, due to

the relatively low conductivity of YSZ [18].









2.3.4.2 Ceria / Bismuth oxide bilayer electrolyte

Wachsman et al. proposed a bilayer electrolyte consisting of a layer of erbia-

stabilizaed bismuth oxide (ESB) on the oxidizing side and a layer of SDC or GDC on the

reducing side[9]. In this arrangement, the ceria layer would protect the bismuth oxide

layer from decomposing by shielding it from very low Po2 and the ESB layer would

serve to block electronic flux. They demonstrated concepts of bilayered electrolytes by

using a system, which consists of SDC (Ceo.8Smo.201.5) on the reducing side and ESB

(Bio.8Er0.201.5) on the oxidizing side [39]. As shown in Fig. 2-6, it was considered that the

gas phase on either side is fixed by the gradient of oxygen partial pressure and the

relative electronic conductivity relies on the local oxygen activity. In this bilayer

structure, the SDC layer prevents the ESB layer from decomposing at very low Po2.

That is, in bilayered bismuth/ceria electrolytes, thermodynamics stability of the bismuth

oxide electrolytes can be elevated and ceria can be act as both an electrolyte and

anode depending on local oxygen partial pressure. As a result, higher OCP can be

obtainable in the bilayered bismuth/ceria electrolytes, since transference number of

bismuth oxide electrolyte is unity [18]. Based on the gradient oxygen partial pressure

and vacancy transport theory, modeling results showed that the interfacial oxygen

partial pressure can be mainly determined by the relative thickness ratio between two

oxide electrolyte layers[24]. These results implies that the relative thickness ratio is the

key parameter to the electrochemical performance of bilayer electrolytes, since the P02

at the ESB/SDC interface can be controlled by the thickness ratio of SDC and ESB

layers. Recently Park et al. reported study result on SDC/ESB bilayer electrolyte [39,

40]. In this paper, they successfully deposited thin ESB layers with pulsed laser

deposition (PLD) technique and a dip coating method on 1.7mm thick SDC pellet and









showed that there is no interfacial phase formation. Such a formation can lower the total

electrolyte conductivity and cause the ESB layer to be ineffective in blocking electron

flow from ceria electrolyte causing higher open-circuit potential. Although the studied

relative thickness ratio(LEsB/LsDc) of the bismuth/ceria bilayer electrolytes was up to 10-2

level, it was expected that the higher relative thickness ratio can obtain higher electron

open circuit potential without loss of high ionic conductivity on electrolyte. Leng et al.

also demonstrated the bilayer concept through YDB (yittria-doped bismuth)/ GDC

electrolyte system [41]. Their results showed not only higher open-circuit potential but

also lowering electrode polarization effect due to the bismuth oxide interlayer. In this

study they used relatively higher thickness ratio of 0.3, which implies the possibility of

thermodynamic stability of thin film bilayer electrolyte even in higher ratio. However,

only one thickness ratio was used and the cathode (Pt) was different from previous

studies (Au). The thickness ratio effect cannot be directly compared with other results.

Recently, modeling of the transport in ESB/SDC bilayer electrolytes has shown that the

thickness of the bismuth oxide layer can be increased relative to the ceria layer, due to

the increase in the electrolytic domain with decreasing operation temperature [42]. As

shown in Fig. 2-7, it has been expected that, at higher the ESB/SDC electrolyte

thickness ratio, the total electrolyte resistance can be dominantly influenced by higher

conductivity bismuth oxide layer, from which we can expect a significantly lower ASR

(area specific resistance) as well as very high OCP.











HZ, CO, H20. C02


ANODE
CO + H20- C02 + H2


INTERFACE -- /
0=+ H2--- 20+ 2e

CATHODE
02 + 4e--- 2=


OVERALL REACTION
H2 + '02-- 2
CO + 02 -4 CO,


Figure 2-1. Schematic diagram of reactions in SOFCs based on oxygen-ion conductors
[22]











Theoretcal EMF or Ideal Voltage


Region of Activation Polarization
\(Reacton Rate Loss) I
STotal




Region of Ohmic Polarzabon
(Resistance Loss)


Loss


Region of
Concentration Polarizaton
(Gas Transpo Loss)


C p 0
r h ev,-'


.-----aon
--- ,- f
--Anode polarizaion Operan Voltage V, Curve
; _-,------- ode-Anode-PO

Current Densty (mAJcm2)

Figure 2-2. SOFC current-voltage behavior indicating relative polarization losses [4]


10-





0.5













b203


4 8 12 16 20


MOL% M203 Of MO


Figure 2-3. Variation of ionic conductivity of stabilized Zr02 with
(T=807oC) [2]


dopant concentration


10-1


10-2


1I4


~I








Temperature [oC]


800 700


600


1.0


500


1.2 1.4 1.6


1000/T [K-']


Figure 2-4. Conductivities of selected electrolyte materials [3]


400


300


E

0




.-j1


1.8












0.0 -


S-0,2-
E
-




S-0.8-

- .In-


Figure 2-5. ESB conductivity


Po2F
Fuel


700C

650C






25 -20 -IS .I 0


Slog (P0)u
versus Po2 in purified argon atmosphere [37]


Po2
Air


Po2 Profile


Interracial Po2


ESB decomrposes ESB is stable
Figure 2-6. Conceptual representation of a bilayer electrolyte showing the effect of
relative thickness on interfacial oxygen partial pressure (Po2 ) [39]















0.1

0
0.08


< T=500'C
0.0 Thicker ESB
Total thickness= 10om Thicker

0.04
0.001 0.01 0.1 1 10 100 1000

Thickness Ratio (LESB/LSDC)


Figure 2-7. Bulk electrolyte ASR at 500 oC as a function of relative (t = LESB/LSDC)
and total thickness for bilayers









Table 2-1. Calculated Po2 and Nernst voltage at open-circuit condition (T=500~7000C)
Temperature AG (T) Po2 Nernst Voltage; Experimental
(C) (J/mol) (atm) OCPtheoretical (V) Conditions
700 -198507 4.641x10-25 1.1415 PH2
650 -200742 1.816 x10-26 1.1473
600 -202977 4.900 x10-28 1.1531 PH2 =0.0309
550 -205212 8.526 x10-30 1.1588 Po2,ref=0.21
500 -207447 8.784 x10-32 1.1646 atm


Table 2-2. Conductivity Data for Stabilized ZrO2 Doped with Rare-Earth Oxides [43]
Dopant Composition Conductivity (10000C) Activation energy


(M203) (mol% M203)


(X 10-2 -1 m-1)


Y203 8 10 96
Nd203 15 1.4 104
Sm203 10 5.8 92
Yb203 10 11 82
Sc203 10 25 62


(kJ/mol)









CHAPTER 3
INTERGRATING NANO- AND MICRO- STRUCTURED ANODE FUNCTIONAL
LAYERS FOR IMPROVED IT-SOFC PERFORMANCE

3.1 Introduction

The commercialization of solid oxide fuel cells (SOFCs) as highly efficient, low

pollution power sources can be realized by lowering operational temperatures [3]. As

Steele claimed, in the lower temperature operation regime (< ~ 700 OC), the system cost

can be significantly reduced with the use of cheap stainless steel for the bipolar plates

and the balance of the plant, combined with the use of high temperature gaskets rather

than rigid glass-based seals. The lower operation temperature can also enhance the

SOFC's mechanical stability and life time [5]. However, critical points to achieve this

goal are to alleviate the significantly increased ohmic and activation polarizations at

reduced temperatures due to their thermally activated nature [9, 44].

An anode-supported designs for low (~ 500 OC) to intermediate (~ 600 OC)

temperature SOFCs have recently received much attention since they can

accommodate very thin electrolytes (< ~ 10 pm), thereby eliminating a large fraction of

the cell's ohmic resistance [5]. Until now, many anode-supported SOFCs with state-of-

the-art thin, highly conductive electrolytes, and highly electrocatalytic cathodes have

been shown to obtain high power density at intermediate temperatures (IT). For

example, Shao and Haile reported a high performance of ~ 1.01 W/cm2 at 600 OC with a

barium doped perovskite cathode (Bao.sSro.s-Coo.aFeo.203-6, BSCF) accompanied with

thin doped ceria-electrolyte [45]. Recent study also demonstrated a nickel and gadolinia

doped ceria (Ceo.gGdo.101.95, GDC) composite anode-supported SOFC with ~10 pm

thick of bismuth/ceria bilayered electrolyte coupled to a bismuth-ruthenate cathode that

achieved a high maximum power density (MPD) of ~ 1.95 W/cm2 at 650 OC [20]. In the









anode-supported design, however, anodic polarization at the interface between

electrolyte and anode can dominate due to its relatively high volume fraction compared

to the electrolyte and cathode.

Generally the SOFC anode provides the conducting phase for charge transfer as

well as reaction sites for the electrochemical oxidation of the fuel [2]. For example with

hydrogen as the fuel, the following reaction occurs

H2 +02 H20 + 2e (3-1)

For low temperature SOFC application, Ni-GDC cermet anode has been widely

used due to high electrocatalytic effect of Ni both for the direct oxidation of hydrogen

and for steam reforming of methane [46]. The GDC in the cermet anode, which high

ionic conductivity at low temperature, extends the reaction zone and compatibility with

GDC electrolyte in addition to preventing Ni sintering. Moreover, it has been reported

that doped ceria showed high resistance to carbon deposition in hydrocarbon fuel [47].

In this composite structure, to achieve higher anode performance, the anode

microstructures should be carefully controlled. Recent research by Suzuki et al. showed

that high MPD of ~ 1.1 and 0.5 W/cm2 at 600 and 550 C with a anode-supported

microtubular fuel cell using ~3 pm thick Sc-doped zirconia (ScSZ) and a GDC interlayer

as electrolyte [48]. In this study, they demonstrated that IT-SOFC performance greatly

depends on anode microstructural factors. For example, porosity and size of Ni particles

in anode influenced concentration polarization and amount of triple phase boundary

(TPB) which is believed as anodic reaction sites between the gas phase, ionic and

electronic conduction phases. Moreover, it is generally accepted that most of the fuel

oxidation reaction, such as eq.(3-1), take place in a limited zone inside ~ 10 pm









thickness of anode adjacent to the anode/electrolyte interface [10]. For this reason,

interfacial anode functional layers (AFLs) have been explored to increase the TPB

length. To date many studies have successfully demonstrated that graded AFL

interlayers with submicron sized NiO particles at the anode/electrolyte interface can

effectively extend active TPB site reducing anodic polarization and give higher

mechanical and chemical stability [11, 49]. Most of these AFLs were fabricated by

conventional colloidal slurry deposition.

In the previous work author and co-workers introduced a new method for

fabricating AFLs by dispersing a GDC precursor at the interface between the anode and

the electrolyte [14]. The resulting nano-sized particles formed a smooth interfacial

region between the anode and electrolyte by filling in pores and crevices and also

extending TPBs. This resulted in an improvement in electrolyte deposition quality as

well as higher electrochemical performance.

Based on this study, the author developed a novel AFL which combines a

conventional particle size (~ < 1pm) AFL applied by colloidal deposition and a nano-

sized Ni-GDC applied by a precursor solution coating. Due to the nature of the

precursor solution, the nano-scale Ni-GDC can penetrate into the AFL and be well

distributed. As shown in Fig. 3-1 we expect two major benefits from this bimodally

integrated AFL concept. First, the TPB length of the AFL can be significantly higher than

that of the conventional AFL because very fine Ni-GDC particles surround the

submicron-size Ni-GDC AFL particles. Secondly, precursor solutions form very fine

particles which fill submicron-sized pores at the interfacial region and thereby increase

the actual 2-dimensional contact area with the electrolyte (2-phase boundaries) so that









it is in closer agreement with the nominal measured contact area, thus reducing the

interfacial ohmic resistance. In addition to higher electrochemical performance, this

novel AFL is very attractive for mass production of IT-SOFCs because all materials are

commercially available (NiO and GDC powder, and metal cation nitrates) and this

simple process is readily applicable to most anode-supported fuel cell designs with

precise control.

In this chapter, the cost-effective and high performance novel AFL for IT-SOFCs is

described. To investigate the effect of the novel AFL on SOFC performance, three cells

all from the same anode tape were compared. These include no AFL, conventional AFL

(referred to as C-AFL), and one sprayed precursor onto the conventional AFL to form a

nano/micro composite AFL (referred to as N+C-AFL). Thin GDC electrolyte and

La0.6Sr0.4C0.2Fo.803-6 (LSCF)-GDC composite cathodes were used for the balance of the

button cells. The microstructural evolution was analyzed and the electrochemical

performance of the SOFC with this novel AFL were measured and characterized.

3.2 Experimental

The NiO-GDC anode support was fabricated by tape-casting using a 65:35 (wt%)

mixture of NiO (micron-scale, Alfa Aesar) and GDC (Rhodia) powders. Based on the

ethanol solvent, an appropriate binder system was prepared with Solsperse, di-n-butyl

phthalate (DBP) and poly-vinyl butyral (PVB), with mixing ratio of 5.9 : 44.1 : 49.9 wt%,

as the dispersant, plasticizer, and binder respectively. For a homogeneous slurry with

proper viscosity and strength before and after tape casting, the binder system was

mixed with the powder mixture and ball-milled for 24 hrs. After a de-airing step to avoid

cracks or defects caused by air bubbles during the tape-casting process, a Procast tape

casting system (DHI, Inc) produced a NiO-GDC anode tape from the slurry. To make a









button type fuel cell, the dried tape was punched out into a circular shape with a 32 mm

diameter and pre-sintered at 900 C for 2 hrs.

Fabrication of the N+C-AFL on anode substrate involved two steps. First, a NiO-

GDC colloidal slurry containing submicron size NiO (JT Baker) and GDC (Rhodia)

(65:35wt%) mixed with the proper binder system was deposited on one side of anode

support by spin coating. The binder system consisted of Solsperse, di-n-butyl phthalate

(DBP) and poly-vinyl buteral (PVB), with mixing ratio of 52.2 : 28.2 : 19.6 wt% as the

dispersant, plasticizer, and binder respectively. In this work, the C-AFL thickness was

maintained by coating the support two times at 1500 rpm for 15 s. Subsequent heat

treatment at 400 C was carried out for the removal of the binder system. Next, the Ni-

GDC nitrate precursor was coated onto the first AFL. A 1 M solution of Ni-GDC

precursor having the same mole ratio of each element as the anode substrate was

synthesized by dissolving Ni(N03)2-6H20, Gd(N03)3-6H20 and Ce(N03)3-6H20 in ethyl

alcohol using ultra-sonication for 30 min. The precursor solution was transferred to a

spray gun (Excell), sprayed onto the C-AFL surface and pre-sintered at 900 C for 1 hr.

Thin and even GDC electrolytes were deposited by spin coating with a GDC

colloidal slurry. For the GDC colloidal slurry, the Rhodia GDC powder was ball milled for

48 hrs with a binder system based on an ethanol solution. As a binder system, for 10 g

of GDC powder, 0.5g of Solsperse dispersantt), 0.3g of PVB (binder) and 0.2g of DBP

plasticizerr) were used with 70cc of ethanol. The spin coating was conducted at 1500

rpm for 15s for each deposition. After drying at room temperature for 10 hrs the

anode/N+C-AFL/electrolyte multilayer structure was sintered at 1450 C for 4 hrs using

a 3 OC /minute ramp rate in air.









A Lao.6Sro.4Coo.2Feo.803-6 (LSCF) GDC composite cathode was prepared and

applied on the GDC electrolyte surface. Cathode inks were synthesized by mixing LSCF

(Praxair) and GDC (Rhodia) at 50:50wt%. For a solvent, Alpha-terpiniol and ethanol

were used. DBP and PVB were used as the plastisizer and binder, respectively. After

mixing and grinding the cathode ink for 1 hour, the ink was brush-painted onto the GDC

electrolyte evenly. The first layer of cathode ink was dried in an oven for 1 hour at 120

C, and a second layer of the same cathode ink was brush painted onto the first layer.

The active cathode area was ~0.4 cm2. The cathode was fired at 1100 oC for 1 hour. Ag

mesh and Pt wire were bonded onto both electrode surfaces using Pt paste for current

collecting and then fired at 900 oC for 1 hour.

The AFL microstructures on Ni-GDC anode supports were observed using

scanning electron microscope (SEM, JEOL 6400 / 6335F). For electrochemical

performance, fuel cell samples were loaded in sealed fuel cell testing apparatuses.

Current-voltage (I-V) characteristics were conducted by a Solartron 1287 using 30 sccm

of dry air on the cathode side and 30 sccm of humidified (3 vol% H20) hydrogen on the

anode side. In addition to the I-V measurement, 2-point electrochemical impedance

analysis was carried out under open circuit condition using a Par-stat 2273 (Princeton

Applied Research) with a frequency range of 100 KHz to 100 mHz.

3.3 Result and Discussion

Fig. 3-2 shows the microstructures of the anode substrate surfaces after

application of the different AFL types and the evolution of these microstructures after

various heat treatments. The AFL surfaces were compared after deposition and

presintering at 900 OC (Fig. 3-2-a, c, e) in order to inspect and compare the initial

morphological state of N+C-AFL deposition on the anode. As seen in Fig. 3-2-a, the









substrate with no AFL has numerous micron-sized pores. Such a microstructure can

cause poor mechanical contact between the anode and electrolyte and lead to micro-

cracking during high temperature operation [11]. In contrast, the microstructure of the

N+C-AFL-covered surface (Fig. 3-2-e) exhibits little if any micron-sized porosity and the

surface particles are fine and well distributed. The surface coated with C-AFL still

exhibits a degree of micron size porosity (Fig. 3-2-c).

Next, we sought to observe the microstructure of the N+C-AFL coating after

testing. However, due to the presence of the electrolyte on the surface, it is difficult to

observe the reduced anode surface after electrochemical testing. Therefore, in order to

simulate the effect of testing, the three anode samples were sintered at 1450 C with no

electrolyte coating and were reduced at 650 C under simulated operational gas flow

conditions for 10 hrs. Fig. 3-2-b shows again that the bare anode forms big pores on the

interfacial surface, compounded by the reduction of large NiO particles into Ni metal.

The conventional C-AFL exhibits a much finer particulate microstructure (Fig. 3-2-d).

The characteristic microstructure of the N+C-AFL is presented in Fig. 3-2-f. Compared

with Fig. 3-2-d, the very fine Ni and GDC particles (marked as dotted circles) appear to

be better distributed between pores of the submicron size Ni-GDC AFL network

structure, indicating likelihood for extended TPB lengths and 2-phase contact area with

the electrolyte. Fig. 3-2-g exhibits a characteristic N+C-AFL structure magnified from Fig.

3-2-f. In this figure it is clearly shown that a nanosized particle is necking with submicron

or micron sized AFL particles, which corresponds well with the schematic diagram of

N+C-AFL in Fig. 3-1-b. Fig. 3-2-h shows the cross-sectional view of actual anode/N+C-









AFL/electrolyte multilayered structure after cell testing, indicating the very fine AFL

structure is well constructed with dense electrolyte and porous anode.

Fig. 3-3 shows the I-V characteristics and power densities for the three different

types of samples at the temperature ranges from 500 to 650 C. In Fig. 3-3-a, I-V plots

of 3 cells were compared at 600 C and the detailed values are tabulated in Table 3-1.

The C-AFL sample exhibited a maximum power density (MPD) of 681 mW/cm2, an

increase of 127% compared to the sample which used no AFL, which measured only

300mW/cm2. In addition, the open circuit potential (OCP) of increased from 0.82 V for

the cell without an AFL to 0.86 V for the cell using the conventional AFL. This supports

the theory that particle size graded anode functional layers may improve the quality of

GDC electrolyte deposition by partial filling the interfacial porosity [14]. Although the

OCP of the Ni-GDC AFL sample was about 0.04 V higher than the sample with no AFL,

as shown I-V plot in Fig. 3-3-a, the major contribution to the improvement in

electrochemical performance comes from its lower area specific resistance (ASR).

For the novel N+C-AFL cell, the MPD reached 1160 mW/cm2-a 287% increase

compared to sample with no AFL and 70% higher than that of sample using a C-AFL,

while its OCP (0.85 V) was comparable to that of the C-AFL sample (0.86 V). Again,

the highly improved performance is caused by the further reduction of the polarization

losses. This result indicate that the precursor solution penetrated into the graded anode

functional layer (Ni-GDC AFL) and formed nano-particles with proper percolation and

distribution of both GDC and Ni particles, thus increasing active TPB length effectively

at the interface between anode and GDC electrolyte.









To further investigate the effect of the N+C-AFL on performance enhancement,

electrochemical impedance testing was conducted under open circuit condition on each

sample at the same temperature range as used for I-V measurement (Fig. 3-4).

Comparison of the nyquist plots of the three samples at 600 C are shown in Fig. 3-4-b.

The total, ohmic, and electrode ASR values at 600 C were extracted and are given in

Table 3-1. As expected from the I-V curves, the total ASR value of the N+C-AFL sample

(0.206 Q-cm2) was much less than that of the C-AFL sample (0.387 Q-cm2) and the

sample with no AFL (0.607 Q-cm2), showing and ASR reduction of 47% and 66%

respectively, which came largely from reduced electrode polarization drops. The

electrode ASR of the N+C-AFL sample was reduced by 52% and 70%, respectively,

compared to C-AFL and AFL-less samples, suggesting again that the use of the N+C-

AFL did have a positive effect on extending TBP lengths. In addition, it should be noted

that the ohmic ASR (0.070 Q-cm2) of the N+C-AFL cell was also reduced by 31% and

53%, respectively, compared with C-AFL and no AFL cells. Based on previously

reported ionic conductivity of GDC, ohmic ASR of ~10 pm thick GDC electrolyte is 0.054

)-cm2, which is close to that of N+C AFL, while the ohmic ASRs of no AFL (0.149 0-

cm2) and C-AFL (0.102 Q-cm2) cell shows much higher values. As reported by Koide et.

al, these additional IR resistance can be caused by higher interfacial contact resistance

between anode and electrolyte [50]. Therefore, this ohmic ASR drop of the novel AFL

cell indicates that the N+C-AFL effectively improved the interfacial wetting and

expended the active contact area of two phase (electrolyte and anode) boundaries at

the GDC electrolyte/Ni-GDC anode interface, causing the lower resistance to the flow of









oxygen ions through the electrolyte to anode, as well as a two-phase boundary area

that more closely matches the nominal active area used in power density calculations.

This ASR analysis implies two additional points with respect to anode-supported

SOFCs. First, the anodic polarization in this SOFC design is a very large portion of the

total electrode polarization, since modifying the anode alone using N+C-AFL reduced

electrode ASR by 70%. Second, the fact that only ~ 5 pm of N+C-AFL lowered the total

ASR by 66% shows that most of this anodic polarization occurs near the

electrolyte/anode interfacial region (~ pm of depth) which can be mitigated by the use of

a proper anode functional layer at the interface.

Fig. 3-5 summarizes MPD obtained from I-V plots shown in Fig. 3-3 and total ASR

values given in Fig. 4 of SOFCs with N+C-AFL, C-AFL, and no AFL at temperatures

ranging from 650 C down to 500 C. The MPD of the N+C-AFL cell reached 1296, 697,

and 380 mW/cm2 at 650, 550, and 500 C (Fig. 3-5-a). Thus, the improvement in

performance becomes even greater at lower temperatures-the power density of the

N+C-AFL cell compared to that of the cell with no AFL increased by 107% (625 to 1296

mW/cm2) at 650 C and by 407% (75 to 380 mW/cm2) at 500 C. The same trend is

observed for the total ASR values as shown in Fig. 3-5-b. The total ASRs of the N+C-

AFL were 0.091, 0.497 and 0.654 Q-cm2 at 650, 550 and 500 C, respectively, showing

total ASR reduction of 67.2 and 80.2% at 650 and 500 C. This implies that at the low

end of the IT range (below 650 C), the ability of the N+C-AFL to reduce the ASR and

improve MPD was confirmed and the improvements were significant. On the other hand,

for all samples, the portion of electrode ASR in total ASR is getting greater as

decreasing temperature (Fig. 3-4). Even N+C-AFL showed electrode ASR fraction









increase in total ASR from 54.8% at 650 OC to 75.8% at 500 OC. We believe this is

because the sluggish oxygen reduction reaction occurred at the conventional perovskite

cathode due to high activation energy at the lower temperature. Coupled with recently

reported highly catalytic cathodes, even higher performance of this SOFC is expected at

low temperature.

3.4 Conclusions

In conclusion, bimodally integrated nano-/micron- composite AFL was developed

by simple spray coating a precursor solution into conventional submicron sized Ni-GDC

in a functional layer. This combined structure produced a novel N+C-AFL.

Microstructural analysis revealed that very fine Ni and GDC particles were

homogeneously distributed into the conventional AFL and formed network structures in

3D, leading to a significant increase in TBP length. A SOFC using this novel AFL

exhibited a MPD of 1.16 W/cm2 at 6000C. Due to its characteristic structure, N+C-AFL

reduced both electrode and ohmic ASR. Compared with the performance of a SOFC

without AFL, the cell using the N+C-AFL showed a 287% increase in power density as

well as a 66% reduction in total ASR. This effect was observed throughout the IT range

tested, indicating the N+C-AFL is an excellent structure for use in high performance IT-

SOFCs.













Ni-GDC precursor <
solution

Submiro-sized
NiO particles

Micro-sized
NiO particles

GDC particles2


Figure 3-1. Schematic illustration of the proposed N+C-AFL structure on anode-
supported SOFC and effect of N+C-AFL on expending TPB length. Yellow
triangles represent TPBs in conventional AFL (C-AFL) and red triangles
represent TPBs by N+C-AFL.




















































Figure 3-2. SEM micrographs of the anode surface after deposition and pre-sintering
(a, c, e) and after full sintering followed by simulated testing atmospheric
conditions (b, d, f) for samples with no AFL (a, b), C-AFL (c, d) and N+C-AFL
(e, f). Highly magnified image shows the characteristic N+C-AFL structure
(g). The cross sectional image presents actual multilayered fuel cell with N+C-
AFL(h).


56


























0.0 ? ----- --------------- 0.0
1.0 .1 1.2
6000C

SN+ C-AFL
1.0
0.8-

-0.8
0.6- C-AFL


00.
4 0.4

00.4
0.20.2
no AFL L

0.0 0.0
0 1 2 3 4
Current Density (A/cm2)

(a)


1.0 a n 1.4
tN+C-AFL (, C L (,- 650Ca
,~ -1.2
0.8. /
6000C
-1.0 "

0.6- 'm El,
El- 0.8

5500C 0.6 (n

S0.0.4
0.2- 5000C
0.2

0.0 0.0
0 1 2 3 4
Current Density (A/cm2)

(b)

Figure 3-3. Comparison of I-V characteristics for the fuel cell samples with N+C AFL, C-
AFL, and no AFL at 600 oC. (a) I-V plots at the temperature ranging from 650
to 500 oC for N+C-AFL (b), C-AFL (c), and no AFL (d).












1.2
0.8- -U
0 21.0
06500C

-0.6-


-00..
0.4- 6000 0.6



0.2











1.2
0.0 5000C I 0.0

0 1 2 3 4



Current Density (A/cm2)

(d)


Figure 3-3. Continued
no AFL
-"1.2
0.8-
-1.0 0

-0.8
c6500C 0.
0.6


0.2 -


C ,,-0.4


01 2 3
Current Density (A/cm2)

(c)



Figure 3-3. Continued













(a) 6500C 0 N+C-AFL A C-AFL no AFL


AA .AA. *-
002< / ^.A AAA--k


--S00


"E



I


E


0.41


0.0
C1


I.U


E .
0.5-


0.0-
0.0


M;.. *A


0.5 1.0 1.5 2.0
Z' (cm2)


2.5 3.0


Figure 3-4. Electrochemical impedance spectra of the testing samples with N+C AFL,
C-AFL and no AFL at various temperature; 650 C (a), 600 C (b), 550 C (c),
and 500 C (d).


0.10-

E
0
0.05-


S0.00-


(b) 6000C



I I-'; I A III


0.05 0.10 0.15 0.20 0.25


0.0 0.1 0.2 0.3 0.4 0.5 0.6

(c) 5500C


0.2 0.4 0.6 0.8 1.0 1.2 1.4


d) 5000o


3.5


I


j


*-
\-~


- -


nQ.00


I


U./


0.30


-- --


-- --


--


0.0


.


I












1.2-

1.0-

0.8-

0.6-

0.4-

0.2-


0.0







4.0

E 3.5
o
3.0

2 2.5
o
c 2.0

L 1.5

(- 1.0

S0.5

0.0


I I I I
500 550 600 650

Temperature (C)


(a)


500 550 600 650
Temperature (C)


(b)

Figure 3-5. MPD (a) and ASR plots (b) for the different samples tested between 500
and 650 oC.


- N+C-AFL
C---AFL
- no AFL









Table 3-1. Detailed OCP, MPD and ASR values of the fuel cell samples with N+C-AFL,
C-AFL, no AFL at 6000C.
Cell Type OCP MPD Total ASR OhmicASR Electrode ASR
unit V mW/cm2 )cm2 )cm2 )cm2
no AFL 0.82 300 0.607 0.149 0.458
C-AFL 0.86 682 0.387 0.102 0.285
N+C-AFL 0.85 1156 0.206 0.070 0.136









CHAPTER 4
EFFECT OF NI-GDC AFL COMPOSITION ON PERFORMANCE OF IT-SOFCS

4.1 Introduction

Wider commercialization of solid oxide fuel cells (SOFCs) can be achieved by

lowering the operation temperature [5]. Critical points to achieve this goal are to

alleviate the significantly increased ohmic and activation polarizations at reduced

temperatures due to their thermally activated nature [5, 7, 44]. Anode-supported

designs for solid oxide fuel cells operating at intermediate temperatures (IT, 500 ~ 650

C) are widely used due to their ability to utilize a very thin electrolyte (below ~10 pm),

eliminating a large fraction of the ohmic polarization loss [51, 52]. Currently, state of the

art thin electrolytes and highly electrocatalytic cathodes are applied on anode-supported

cells, showing high performance at IT ranges. For example, we reported that anode

supported SOFCs using a Nil Ceo.gGdo.101.95 (GDC) composite anode, bismuth

oxide/GDC bilayered electrolyte and bismuth/ruthenate composite cathode achieved an

exceptionally high maximum power density (MPD) of ~ 2 W/cm2 at 650 oC [20]. In the

anode-supported design, however, anodic polarization at the interface can dominate

due to its relatively high fraction of volume compared to the electrolyte and cathode.

Generally, the anode-support is fabricated using submicron-sized NiO particles with

pore-former to increase porosity in the anode [11, 53]. Even though sufficient porosity is

readily achieved, the large sized pores at the electrolyte interface cause a large

interfacial anodic loss and low mechanical strength due to the poor quality of electrolyte

deposition [11]. To overcome these problems, interfacial anode functional layers have

been developed. Up to this date, many researchers have reported that thin anode

functional layers (<10 pm) using fine particles without pore-former effectively reduce









activation polarization by increasing triple phase boundary (TPB) density, strengthening

mechanical properties and lowering ohmic losses by improving the quality of electrolyte

deposition due to significantly reducing interfacial porosity [11, 53-56].

For the last several years, we have concentrated on developing high performance

IT-SOFCs with a tape cast Ni-GDC anode using large micron-sized NiO particles [14].

Although this anode design provided sufficient mechanical properties and porosity

without additional pore-former, it still suffered from low performance. The low

performance is due to the coarse microstructure with large size pores at the electrolyte

interface, causing poor electrolyte deposition. Recently, Ahn et al. introduced a new

method to establish AFLs by dispersing a GDC precursor solution at the

anode/electrolyte interface by simple spray coating. This resulted in ultra fine particles

forming a much smoother interface, leading to extended TPB density [14]. Based on this

study, Lee et al. developed a novel AFL integrating a nanosized Ni-GDC particles into a

submicron-sized AFL by a simple precursor solution coating [57]. Because of the nature

of the precursor solution, very fine Ni-GDC particles were well-distributed and formed a

bimodally structured AFL. The AFL led to a high performance of ~1.3 W/cm2 at 650 C

with a 10 pm thick GDC electrolyte. We have shown how microstructural changes due

to different particle sizes influence the electrochemical performance for SOFCs with

respect to the surface porosity and anode active reaction sites, TPBs. However, the

spatial distribution, amount of porosity and TPBs in two phase composite anodes can

be significantly affected by the amount of each phase [12, 58]. In this study, we

fabricated and investigated IT-SOFCs having AFLs of various compositions using

submicron-sized NiO and GDC particles at the electrolyte interface. Electrochemical









performance studies were conducted by current-volage (I-V) and impedance testing.

The optimal composition of this AFL was investigated, and the relationships between

composition and performance were analyzed.

4.2 Experimental

4.2.1 Cell Fabrication

Flat NiO-GDC anode supports were fabricated by the tape-casting method. In

order to provide a sufficient gas channel without the aid of pore-former, micron-sized

large NiO particles (Alfa Aesar) were mixed with nano sized GDC from Rhodia. In this

study, the composition of the anode support was fixed at 65 to 35wt% of NiO to GDC.

Based on the ethanol solvent, an appropriate binder system was prepared with

Solsperse, di-n-butyl phthalate (DBP) and poly-vinyl butyral (PVB) as the dispersant,

plasticizer, and binder respectively. For a homogeneous slurry with proper viscosity and

strength before and after tape casting, the binder system was mixed with the powder

mixture and ball-milled for 24 hrs. After a de-airing step to avoid cracks or defects

caused by air bubbles during the tape-casting process, a Procast tape casting system

(DHI, Inc) produced a NiO-GDC anode tape from the slurry. To make a button type fuel

cell, the dried tape was punched out into a circular shape with a 32 mm diameter and

pre-sintered at 900 C for 2 hrs.

For a finer and graded AFL structure, submicron-sized NiO (JT-Baker) was mixed

with nano-sized GDC (Rhodia). To investigate the effect of the composition of the AFL,

various NiO-GDC AFL contents, from 40wt% to 80wt% of NiO, were fabricated on the

anode surface by spin coating. Similar to the tape casting slurry, a proper binder system

was added to each AFL powder, leading to a well dispersed colloidal slurry. During the

spin coating process, the thickness of the Ni-GDC AFL layer was controlled by the









number of depositions at the same spin speed and time. In this work, we produced the

same thickness of AFL by applying 3 coats at 1500 rpm for 15s. After deposition, the

AFL layer was pre-sintered at 900 C for 1 hour to remove the binder system.

For the electrolyte, the same GDC powder for the anode and AFL was ball milled

for 48 hrs with a binder system based on an ethanol solution. As a binder system,

Solsperse dispersantt), PVB (binder) and DBP plasticizerr) were used. The thin GDC

electrolyte was coated by the spin coating method with same process as the AFL

deposition. After deposition, samples were dried at room temperature for 10 hrs. After

drying the multilayer anode/AFL/electrolyte structure was sintered at 1450 C for 4 hrs.

A Lao.6Sro.4Coo.2Feo.8O3-6 (LSCF) GDC composite cathode was prepared and

applied on the GDC electrolyte surface. Cathode inks were synthesized by mixing LSCF

(Praxair) and GDC (Rhodia) at 50:50wt%. For a solvent, Alpha-terpiniol and ethanol

were used. DBP and PVB were used as the plastisizer and binder, respectively. After

mixing and grinding the cathode ink for 1 hour, the ink was brush-painted onto the GDC

electrolyte evenly. The first layer of cathode ink was dried in an oven for 1 hour at 120

C, and a second layer of the same cathode ink was brush painted onto the first layer.

The active cathode area was ~0.4 cm2. The cathode was fired at 1100 C for 1 hour. Ag

mesh and Pt wire were bonded onto both electrode surfaces using Pt paste for current

collecting and then fired at 900 C for 1 hour.

4.2.3 Characterization

For electrochemical performance, the prepared cells were loaded on a fuel cell

testing set-up. In order to obtain the gas-tight sealing, the edge of the cell and testing

tube were covered with a mixture of two parts ceramic sealant using ceramabond-517

(Aremco). I-V tests were carried out by a Solartron 1407E with 3% wet hydrogen as a









fuel on the anode side and dry air as an oxidant on the cathode side at various

temperatures. The gas flow rate was controlled by a mass flow controller (MKS 647C).

In addition to current-voltage characteristics, electrochemical impedance spectroscopy

with a two-point probe was measured by a Solartron 1400 using a frequency range of

100 kHz to 100 mHz under the same gas and temperature conditions.

For microstructural analysis, the tested fuel cells were fractured and the cross-

sections of the multilayered structures were observed using a scanning electron

microscope (SEM, JEOL 6400 / 6335F) with the back-scattering mode.

4.3 Results and Discussion

In this study, five kinds of fuel cells with AFLs containing 40, 50, 60, 65, and 80

wt% of NiO were prepared in order to investigate the effect of AFL composition on the

microstructure and electrochemical performance. At the same time, a cell without AFL

was tested as a reference cell.

4.3.1 Microstructural Analysis

Figure 4-1 shows the microstructures from a cross-sectional view of the samples

with different AFL compositions after reducing and testing under 90 sccm of air on the

cathode side and hydrogen on the anode side. In order to directly compare the effect of

the AFL, the thickness of the electrolyte, anode, cathode, and active cathode area were

designed to be the same between the different samples. As seen in Figure 4-1-a~f, the

dense GDC electrolyte for each cell shows an even thickness of 17 to 19 pm, with a few

closed pores. For the other anode and cathode, the thicknesses were measured by

SEM and found to be nearly identical between the samples at 290 pm to 300 pm for

anode and ~30 pm for cathode. For the AFLs, each AFL thickness measured was about

10 pm. In addition, it is clearly shown that interfaces between the AFLs and GDC









electrolyte (Fig. 4-1-b~f) look very flat and continuous while the interface between the

GDC electrolyte and anode support (no AFL sample, Fig. 4-1-a) shows a very irregular

and discontinuous morphology due to large pores caused by reduction of large size NiO

into Ni.

In Fig. 4-2, the magnified microstructures of the anode/AFLs nearby the GDC

electrolyte after reducing and testing are presented. In comparing Fig. 4-2-a to 4-2-b~f,

particulate size differences between the AFLs and anode are clearly shown indicating

that submicron-sized NiO successfully formed a very fine Ni-GDC AFL structure. Using

backscattered mode the Ni (dark gray), GDC (white), and pore (black) phases are well

distinguishable due to contrast difference. As shown in Fig. 4-2-b~f, it is observed that

the amount of both Ni phase and porosity due to reduction of NiO into Ni during the

operation increased with higher NiOwt% in AFL.

4.3.2 Effect of AFL Composition on Power Density

4.3.2.1 I-V characteristics at 650 C

Fig. 4-3 shows I-V characteristics of fuel cells with various compositions of AFL at

650 C. The open circuit potential of the tested cells were 0.806, 0.834, 0.838, 0.814,

0.807, and 0.801 V for the cells with no AFL, 40, 50, 60, 65, and 80wt% NiO AFL. It has

been commonly reported that formation of AFL at the interface between the anode and

thin electrolyte enhances OCP [13, 14, 57]. Generally the theoretical Nernst voltage for

thin electrolytes below 10 pm can be lowered by gas permeation through electrolyte

which is caused by various structural defects such as open pores and microcracks in

addition to inherent internal shorting of a mixed ionic and electronic conductor (MIEC)

such as doped ceria. In that case, AFL can help to enhance the quality of the electrolyte

deposition by reducing the possibility of crack and pore formation at the interface due to









its finer and reduced surface porosity [14, 57]. In addition, as Chen et al. reported, AFL

itself can act as a barrier to gas permeation along with the electrolyte although this can

cause concentration polarization [13].

In this study, however, the OCP does not increase only by the existence of the

AFL, as shown between the AFL cell (0.806V) and 65% wt NiO AFL cell (0.807 V). It is

considered that the electrolyte thickness (~20um) was much larger than the size of bare

anode surface pore (below 5 pm), suggesting a similar gas permeability in all

electrolytes regardless of AFL existence. Instead, the observed OCP seems to be

influenced by AFL composition.

Fig. 4-4 shows the plot of OCP versus NiO composition of the AFL or anode at the

anode/electrolyte interface. Higher OCP is observed with decreasing NiO wt% in the

AFL. SEM observation of the anode and AFL structure before reduction reveals that the

anode without pore-former and AFL are relatively dense with little porosity. Therefore, it

is considered that the total pore volume in the anode or AFL after testing increases

proportionally with NiO content because the only source to produce porosity is the

reduction of NiO under operational conditions. This indicates that an AFL with lower NiO

content forms less porosity after reduction and is a better barrier of gas permeation,

leading to a higher OCP. In addition, the higher Ni content can cause lower OCP

because of greater electronic conductivity. It has been reported that for thin MIEC

electrolyte, OCP greatly depends on electrolyte thickness [59, 60]. Therefore it is noted

that for the ultra thin electrolyte system, the OCP change with AFL composition can be

much greater and have a significant impact on cell performance.









The maximum power densities (MPDs) of cells with no AFL, 40, 50, 60, 65, and 80

wt% NiO AFL were obtained from the power density curves in Fig. 4-3 as 882, 732,

1033, 1147, 1077, and 711 mW/cm2, respectively. In addition to MPDs, total area

specific resistance (ASR) value for each cell also estimated from the IV curves near

OCP region. Resultant MPDs and total ASRs are plotted in Fig. 4-5. Compared to the

no AFL cell, cells with 50, 60, and 65wt% NiO AFL showed significant improvement in

MPD at 650 C. While the OCPs of these cells (Fig. 4-4) shows a different trend and

relatively trivial change (~0.03V), the resultant performance enhancement mostly comes

from decreasing polarization with AFL, as shown in Fig. 4-5. This implies that the finer

AFL structures (Fig. 4-2-c,d,e) effectively increased TPB density at the

anode/electrolyte interface. In contrast, MPDs of cells with 40 and 80wt% NiO AFL

compared to that of cell with no AFL were decreased. In this case, it is considered that

although the AFLs formed very fine particulate structures (Fig. 4-2-b,f), excessively high

content of NiO or GDC can reduce the connectivity of the TPBs (Ni, GDC, and pore)

leading to inactive regions. Therefore, the AFL composition should be considered as an

important factor in terms of SOFC performance.

As mentioned above, the highest MPD was obtained at the AFL with 60wt% of NiO

(1147 mW/cm2) showing a 30% increase compared to no AFL cell (882 mW/cm2), while

higher OCPs were observed at higher NiOwt% AFLs. Assuming full reduction of NiO

into Ni during operation, the Ni to GDC volume ratio of this AFL is easily calculated with

the densities of Ni, NiO and GDC. The results is 48.6vol% Ni to 51.4vol% GDC, that is

almost 1:1 volume ratio. This result is in good agreement with a recent study of NiO-

SDC (samarium-doped ceria) AFL system by Ai et al [12]. It was also predicted by









previous modeling work. Schnieder et al. conducted analytical modeling for estimating

TPB length of composite electrode using the discrete element method and showed that

the TPB length is maximized at 50vol% of the ionic conducting phase in the composite

anode [61]. On the other hand, another recent study by Wilson and Barnett reported the

lowest electrode ASR with the highest TPB length at 50wt% NiO in a NiO-YSZ(yttria-

stabilized zirconia) active layer, which is a 1:2 volume ratio of Ni to YSZ [62]. However,

it is reasonably considered that the ionic conductivity of stabilized zirconia is much lower

than that of doped ceria at the IT range (500 C ~ 800 C). The NiO particle size (-2.5

pm) used in that study was much larger compared to this study's NiO size (< 1 pm).

Therefore, the lowest resistance with the highest TPB length can be achieved at lower

GDC content and higher NiOwt%, leading to higher porosity. For further understanding,

the quantification of microstructural properties will be needed to find relationship with

electrochemical performances.

4.3.2.2 Temperature dependence

In order to verify the validity of the optimal AFL composition of 60wt% NiO at the IT

range, a current-voltage measurement for each cell was conducted at various

temperatures from 450 to 650 C with a 50 C interval. The resultant MPDs with AFL

composition are plotted in Fig. 4-6, including the MPDs of the no AFL cell separately

marked as open symbols. For all temperatures tested, cells with 60wt% NiO in the AFL

shows highest MPDs with 1147, 626, 271, 118, and 45 mW/cm2 at 650, 600, 550, 500,

and 450 C, respectively. This result shows that the effect of AFL optimal composition is

valid through the intermediate to low temperature ranges.









4.3.2.3 Long term stability

To see the effect of optimal composition AFL on high performance, a potentiostatic

test under an applied voltage where cells reached 98% of their MPD was done for each

cell. The applied voltages were 0.379 V for the AFL cell and 0.380 V for the no AFL cell.

Fig. 4-7 shows preliminary results of a 200 hrs long term stability test for the no AFL and

60wt% NiO AFL cells under the 90 sccm of H2/Air condition. While the no AFL cell

shows initial degradation of power density and stabilized behavior, the effect of the

optimal composition AFL was retained for 200 hrs with high power density of ~1.1

W/cm2. For practical application of this AFL, however, further long term testing under

various temperatures and gas conditions should be conducted.

4.3.2.4 Effect of AFL composition on ASR

In order for further investigation, electrochemical impedance tests were carried out

for all samples. Fig. 4-8-a shows the impedance spectra of each AFL composition at

650 C for which an I-V test was conducted. From the high and low frequency complex-

plane intercepts of the impedance spectrum with the real axis, the ohmic, electrode, and

total ASR values were calculated while normalizing the resistance according to cathode

area. The detailed values are tabulated in Table 4-1 and plotted in Fig. 4-8-b. As shown

in the Table, in this study, total ASR values from electrochemical impedance (ASREIS)

are within 4% deviation from the ASR at IV curves (ASRiv) in Fig. 4-5.

As expected from the previous section, total ASRs from AFL composition show a

similar trend to the MPD. The lowest total ASR (0.188 Qcm2) at 650 C was achieved at

60wt% NiO AFL, which was decreased by 25.7% from that of the no AFL cell (0.253

0-cm2). The ohmic ASRs are similar for all samples because the samples tested in this

study have similar thicknesses of electrolytes with similar densities as supported by









microstructures shown in Fig. 4-1. Therefore, most of the ASR drop comes from the

electrode ASR drop, which is considered as anodic polarization reduction since the

cathodes should be identical. From this result, it is expected that different AFL

compositions directly influence a change in TPB density, affecting anodic polarization.

Most likely the TPB density is the highest at 60wt% of NiO in AFL, which is at 1:1

volume ratio of Ni to GDC after reduction.

Based on these results, the relationship between electrode ASR and MPD with

various AFL compositions are examined. As shown in Fig. 4-5, the performance

enhancement of power density reflects the ASR trend. In Fig. 4-9, it is clearly shown

that electrode ASR has an almost linear relationship with MPD. However, there are still

some deviations from the linearity. At this point it is noted that the ASRs were measured

at the open circuit condition, whereas the MPDs occurred at higher current densities

which have complex contributions from various polarizations, such as activation, ohmic

and concentration polarization as shown in I-V curves in Fig. 4-3. Therefore, for a more

precise study, the ASRs measured under applied currents will be done in the future. In

addition, to completely understand electrode ASR, comprehensive microstructural

features should be considered, such as surface area, porosity, tortuosity and TPB

density.

4.4 Conclusions

In this study, the effect of AFL composition on the electrochemical performance

was investigated for IT SOFCs. For this, the various AFLs with composition from 40 to

80wt% NiO were fabricated. The fine and well-distributed AFL structures with different

NiO amounts were confirmed by microstructural analysis. The optimal AFL composition

was achieved at 1:1 volume ratio of Ni to GDC in the AFL, which is 60wt% NiO. The









effect of the optimal AFL composition on MPD was valid at intermediate to low

temperature ranges. In addition, a preliminary long-term stability test showed the

possibility of practical application of this optimal composition. The measured MPD and

ASR show a linear relationship implying that the performance enhancement greatly

depends on the AFL composition, which might be caused by microstructural features

such as TPB density, porosity, tortuosity of pores and surface area. For further

understanding of the AFL effect on electrochemical performances, the quantitative

analysis of AFL microstructures should be done.











SSCF-G DCcathode



: '- GDC electrolyte




--t'.-; Anode functional layer P:
-

( d)


d


. A.


(fI ^ '. ***
b .I
'. "* *" *. .- .a* --

.-.


Figure 4-1. Backscattered images showing a cross-sectional view of anode-supported
SOFCs with different NiO content in the anode functional layers; no AFL(a),
40wt% (b), 50wt%(c), 60wt%(d), 65wt%(e), and 80wt%(f) NiO.


.
.."^
ft


r


'(a) ,































Figure 4-2. Magnified microstructures of the anode or AFLs with different NiO content
no AFL(a), 40wt% (b), 50wt%(c), 60wt%(d), 65wt%(e), and 80wt%(f) NiO.
Backscattering mode provides better contrast to distinguish Ni (dark gray),
GDC (white), and pore (black) phases.


mlp- 9r

SR*3i~








1.2


1.0 -*-60U:40UAFL -1.0
-e*- 65:35 AFL
\1.0
-80:20 AFL

0.8- 0.8 o


S0 0.6- 06


0.4- 0.4


0.2- -0.2 9


0.0 6, 0.0
0 1 2 3 4 5 6

Current Density (A/cm2)


Figure 4-3. I-V plots of fuel cells with various AFL compositions at 6500C; 40(4),
50(A), 60(*), 65(*) and 80(1)wt% of NiO in AFL, and no AFL(.). The gas
condition was 90sccm of air and 3% of wet hydrogen on the anode and
cathode side, respectively.


1.2







0.85


0.84-


0.83-

5 0.82-

0

0.81-


0.80-


0.79 ,-
20 30 40 50 60 70 80 90 100
NiO in AFL (wt%)
Figure 4-4. Open circuit potential of the fuel cells with various NiO contents in NiO-GDC
AFL. Solid line (red) shows linear fit of the measured data (square)







1400 0.30

1200-
,- 0.25
E 1000- 1

E 800- -0.20 ^

600-
-0.15 0
400-

200 ,-. d 0.10
40 50 60 70 80
NiO in AFL (wt%)

Figure 4-5. MPD (Red square) and total ASRiv estimated from IV curves (blue star) are
plotted with NiO contents in AFL. The open symbols represent no AFL cell.







1200

1000- /650C

E 800-
000C
E 600 C
0- 400

200-
200 a_500(

0 --- 4 ------6
40 50 60 70 80

NiO in AFL (wt %)

Figure 4-6. Maximum power densities of fuel cells with various AFL compositions at the
temperature range from 450 to 650 C. Open symbols represent MPD of no
AFL cell at each temperature.










12



1.0
60:40 wt% AFL


0.8



=0-6-,,

noAFL

0 .4



02



0-0
0.0 II 1

0 50 100 150 200

Time (hours)


Figure 4-7. Long term stability test of fuel cell with 60wt% of NiO in the AFL and the no
AFL cell for 200 hrs at 650 C. Potentiostatic tests were conducted with an
applied voltage of 0.379 V for the NiO 60wt% AFL cell and 0.380 V for the no
AFL cell, at which the cells showed 98% of MPD. The gas condition was
90sccm of air and 3% of wet hydrogen on the anode and cathode side,
respectively.









030


N


6500C no AFL
0.25- 40:60 AFL
A 50:50 AFL
0.20 60:40 AFL
65:35 AFL
80:20 AFL
0.15-

0.10-

005 4444444


0.00 1 ,
0.00 0.05 0.10 0.15 0.20 0.25

Z' (Q cm2

(a)


0.35-

S0.30-

S0.25-

-I 0.20-


.0
S0.10-

S0.05-

< 0.00-


0.30
0.30


NiO in AFL (wt%)

(b)


Figure 4-8. Impedance spectra with various AFL compositions (a), and total, electrode,
and ohmic ASRs of fuel cells with different NiO content (b) calculated from
impedance spectra (a). Open symbols represent no AFL results.


" total ASR



Electrode ASR


Ohmic ASR












E 50:50 AFL
. 1000-
E
S900- no AFL

800-
40: AFL
700-
80:2 FL
0.14 0.16 0.18 0.20 0.,

Electrode ASR (Qcm2)
Figure 4-9. MPD plots with electrode ASR shows a linear relationship. Red line is
linear fitting of the measured data (black dots).









Table 4-1. Detailed ASR values of the testing cells with various NiO contents in AFL
Total ASR Ohmic Electrode
AFL NiO Total ASRASR ASR
AFL EIS ASR AS
(wt%) () cm2) E2 ASR ASR
(wt%) ( cm2) (Q cm2) (Q cm2)
40:60 40 0.259 0.260 0.071 0.189
50:50 50 0.227 0.231 0.070 0.161
60:40 60 0.182 0.188 0.055 0.133
65:35 65 0.196 0.203 0.061 0.142
80:20 80 0.286 0.281 0.076 0.205
no AFL 65 0.253 0.253 0.074 0.179









CHAPTER 5
COMPREHENSIVE QUANTIFICATION OF NIO-GDC ANODE FUNCTIONAL LAYER
MICROSTRUCTURE BY THREE-DIMENSIONAL RECONSTRUCTION USING
FIB/SEM

5.1 Introduction

The inevitable demand on lowering solid oxide fuel cell (SOFC) operational

temperatures has been fueled by commercialization of SOFCs [3, 5, 7, 9, 63]. Currently

the anode supported SOFC design has been widely studied due to possible

accommodation of ultra thin electrolyte [8, 20, 55]. For this design, however, the anode

polarization is possibly dominant due to the anode having the largest volume fraction in

the SOFC. To solve this problem, the electrode polarization losses at the anode should

be effectively reduced. It has been well known that the microstructures of an electrode

greatly influence the electrochemical properties in a SOFC. For example, the volume

fraction of the each component in the electrode can modify microstructures in an anode

or cathode[62]. In previous work, it was shown that tailoring different compositions of

the anode functional layer changed the electrochemical performance and power density

of the SOFC [64]. In the work, 60wt% of NiO in NiO-GDC AFL, which is Ni-GDC AFL

after reduction, showed highest maximum power density with 1.15 W/cm2 at 650 oC.

Reducing the electrode ASR is the major factor for increase performance, and the ASR

was found to have an optimal NiO composition (60wt%) of AFL. Moreover, the electrode

ASR and AFL composition change showed an inverse linear relation to maximum power

density. It was expected that at the optimal composition the number of reaction sites,

that is, triple phase boundaries (TPBs), which consisted of an electronic conductor (Ni),

ionic conductor (GDC), and gas diffusion path (pore), is highest. However, the

evaluation of TPB density is not easily achieved with the conventional two dimensional









(2D) SEM analysis. Some studies were conducted for estimating TPB length or density

by stereological methods using analysis of 2D SEM images [62, 65]. However, accuracy

of active TPB length estimation was limited due to phase interconnectivity issues, which

is three-dimensional (3D) property. Recently, 3D reconstruction techniques were

employed to deal with microstructural analysis of SOFCs using FIB/SEM dual beam

system [66-72]. Using this technique, more accurate and realistic quantifications of the

SOFC electrode microstructure has been available. Moreover, quantified microstructural

features were attempted to link directly with electrochemical properties. For example,

TPB density was linked to charge transfer and adsorption properties and the tortuosity

was used to describe concentration polarization[68]. However, most of the studies were

conducted for composite cathodes and Ni-YSZ anode for intermediate to high

temperature (> -700 C) SOFC applications.

In this study, the microstructural features of Ni-GDC AFLs, which is widely used for

intermediate to low temperature (400 ~ 700 C) SOFCs, with different NiO contents

were investigated by 3D reconstruction technique using a FIB/SEM dual beam system.

After reconstruction, the comprehensive quantification for various properties of the

studied samples was conducted and the values were analyzed. Finally, the active TPB

density was calculated and linked with electrochemical performances

5.2 Experimental

In the work four AFL samples were reconstructed and quantified, which include

cells with 50, 60, 65, and 80wt% NiO AFL. As a reference, a sample without an AFL

was also reconstructed. The detailed fabrication process and electrochemical

performance test of the samples were described in previous work [64]. First of all, all

samples were mounted in an epoxy supporter using a Struers EpoVac System. At the









same time, this epoxy infiltrates into the samples and fills open pores inside porous Ni-

GDC anode/AFL structures, providing better contrast for SEM imaging. After the

mounting process, the sample stub was grinded and polished using sandpaper and

diamond paste down to 1 pm roughness to obtain an even surface of the mount

exposing anode-electrolyte interface.

The automated sectioning and imaging were carried out with a FIB/SEM duel

beam system (FEI Strata DB 235). In this system, the electron beam pole (SEM)

against the ion beam pole (FIB) leans at a 520 angle. (Fig. 5-1-a) The image at each

slice was captured by SEM. To get better contrast difference among the phases, a

through-lens-detector (TLD) in backscatter mode was utilized. The FIB was used to

create a trench around the region of interest (ROI). The slicing distance (z-axis

resolution) was 60 nm. To avoid charging effect during SEM process and protect from

ion damage, protective platinum layers were deposited with an in-situ liquid metal-

organic ion source (LMIS). This repeated imaging and slicing processes were

automatically controlled using the Auto Slice and View software system (FEI Company).

After collecting the cross-sectional images for each sample, the alignment,

segment, cropping and labeling for the three dimensional reconstruction were

conducted by Amira software ResolveRTTM (ver 4.0, Mercury Computer System Inc.).

Fig. 5-1-b shows the schematic diagram of 3D reconstruction process. Amira was also

utilized to quantify the various microstructural features from the 3D reconstructions of

the samples, such as each phase (Ni, GDC, and pore) volume and surface area, phase

gradient, and tortuosity of pores.









5.3 Results and Discussion

Fig. 5-2 shows the 3D reconstructions of the cells, which allows for direct

qualitative comparisons between the anode/AFL structures. The z-axis dimension

reconstructed for each sample was approximated at the distance of 8~10 pm from GDC

electrolyte interface. Considering AFL thickness (~10pm) measured in previous study,

the analyzed AFL depth in this study covers almost full cross-section of each AFL [64].

The detailed dimention for reconstructed samples are summariezed in Table 5-

1 .Compared to the no AFL sample (Fig. 5-2-a), cells with AFLs (Fig. 5-2-b,c,d,e) show

much smaller particulate structures. Among the AFLs, the change of amount of Ni

(green in 3D reconstruction) and GDC (red in 3D reconstructions) is clear as NiOwt%

increases. This result is in good agreement with 2D SEM image observation discussed

in the previous section. Moreover, this 3D reconstruction allows the separation of each

phase, or to combine only two phases. As illustrated in Fig. 5-3, each phase of GDC, Ni,

pore, and combination of Ni-pore phases were reconstructed individually. From this

phase separation, the quantification of the phase gradient, volume fraction, and surface

area for each phase is possible. For the spatial distribution of each phase in AFLs, the

phase gradient for each AFL sample was plotted in Fig. 5-4. It is evident in Fig. 5-4 that

from b to e, the level of Ni and pore phase is getting higher, while the amount of GDC

phase is decreased. In addition, the all AFLs (Fig. 5-4-b,c,d,e) have a very narrow

transition zone (~ below 500nm) which is a region for significant increase of the GDC

phase nearby GDC electrolyte, compared to over 2000 nm for bare anode (Fig. 5-4-a).

This result indicates a very fine microstructure in the AFL and its features such as high

surface area and large amount of reaction sites (TPBs) are well retained at the interface

of the AFL/electrolyte. The anode without AFL lose many reaction sites and gas









diffusion paths due to penetration of the GDC electrolyte into large anode pores at the

interfacial zone which is known to be the most important active reaction zone for the

hydrogen oxidation.

For Ni, GDC and pore phases, the phase volume fractions were quantified using

AmiraTM tissue-statistics module. In this module, the number of voxels for each phase

were counted and based on this result the numerical volume fraction for each phase is

easily able to be calculated. This result can be one of the basic criteria for the practical

confirmation of the credibility of the resultant 3D reconstructions. It is because there are

theoretical values of Ni reduced from NiO, GDC, and pore volume fraction based on

initial AFL composition, which can be calculated from the material properties such as Ni,

NiO, and GDC densities when assuming full reduction of NiO into Ni. The theoretical

and measured values of total volume fraction of each phase for different AFL are

tabulated in Table 5-1 and plotted in Fig. 5-5-a. As observed in phase gradient graphs

(Fig. 5-4), the Ni contents and porosity in AFL increases with increasing NiOwt%, while

GDC is inversely proportional to initial NiOwt% in AFL. It is clearly shown that the Ni and

GDC volume fractions extracted from the 3D reconstruction are well matched with

theoretical values. For example, the volume fractions of Ni were 29.1 1.0, 37.8 5.9,

43.2 6.7 and 51.1 6.4vol% for 50, 60, 65, and 80wt% NiO AFL samples,

respectively, which are very close to the theoretical values of 30.4, 36.3, 39.2, and

47.7vol%. However, porosity is little below the theoretical value, which might be an

attributable to sampling resolution. In addition to the total volume fraction, the solid

phase volume fraction between GDC and Ni was also calculated (Fig. 5-5-b). As

expected from the theoretical value, at 60wt% NiO AFL the Ni to GDC volume ratio was









almost 1:1 with measured value of 48.1 6.7 and 51.9 6.7, at which ratio the highest

performance among studied samples were shown in previous section. At this point, it is

noted that these volume fraction results from the 3D reconstructions were based on

around 150 slices of 2D SEM images and the graded phase plots in Fig. 5-4 showed the

continuous fluctuation of each phase along the distance from the electrolyte. Therefore,

to quantify the structural analysis of the composite SOFC anodes, the 3D reconstruction

is more reliable and necessary compared to 2D SEM image analysis method.

The total surface area values of the AFL structures with different NiO contents

were calculated using Amira tissue statistics module. The resultant values are

normalized by the total volume of the region of interest (ROI) (Table 5-2).

The effective particle or pore diameters (d) for each AFL composition were

calculated with a Brunauer-Emmett-Teller (BET) method using a general formula written

as;

d=6V (5-1)
S

where V and S are the volume and surface area of the each phase, respectively.

The Amira tissue statistics program was utilized for the calculation of the phase volume

and surface area of Ni, GDC and pore in each sample. The effective diameters of Ni

(dNi) phase in 50, 60, 65 and 80wt% NiO AFL correspond to 683, 771, 917, and 1120

nm, respectively, showing the expected trend. The complete data set is tabulated in

Table 5-2 and plotted in Fig. 5-6. For all three phases, the linear relationship between

effective diameter and AFL composition is shown, which is the proportionality for Ni and

porosity, and an inverse proportionality for GDC. This result implies that the high NiO

content in the AFL produce a structure with large Ni particles and very small GDC









particles, which can cause poor connectivity of the smaller particle phase (GDC),

increasing the deactivated TPB sites. In contrast, for low NiO AFL, same situation for Ni

phase can occur but worse due to lower porosity, which interferes with the fast gas

diffusion. Therefore, it is expected that at the medium point, around 55~60wt% NiO AFL

can have optimal structure for highest TPB density. In this case, particle size control

should be considered with other factors, such as composition and porosity.

The tortuosity was estimated using the moment of inertia module in the Amira

software. Using this module, tracking of the center of the open pore phase through

sample from the beginning of the AFL to the GDC electrolyte interface is possible, which

allows us to measure the accumulated 3D Euclidian distance through the region. The

accumulated Euclidean distance divided by AFL thickness yielded the tortuosity of the

each sample (Table 5-2). For the 50, 60, 65, and 80wt% NiO AFL, the tortuosity values

were 2.77, 1.91, 1.91, and 1.69, respectively. This result is in good agreement with the

general theory that the higher porosity with larger pore diameter provides less

complicated gas diffusion paths thorough the open pores, which means low tortuosity.

This tortuosity concept can be combined with volume fraction of porosity to estimate

effective diffusion coefficient, which is directly related to concentration polarization [68].

Therefore, for further analysis of this property and cell performance the deconvolution of

the electrochemical impedance under applied current is in progress.

The TPB density for each AFL composition was quantified. The TPB site is where

the three phases such as Ni, GDC, and pore meet at the same place. However, the

TPB works only when these phases are properly connected. For example, the GDC

phase of a TPB site should be connected to the GDC electrolyte to make a path for









oxygen ions from anode(AFL)/ GDC electrolyte interface to the active site. At the same

time, the Ni phase and pores should have percolation to the interconnect (or current

collector) and fuel side (outside anode), respectively. The reasons are to bring hydrogen

gas from the fuel source to TPB sites, to conduct charge transfer during hydrogen

oxidation at the TPB, and to complete the extraction of electrons from the reaction site

outside circuit. Therefore, counting the active TPB and removing the inactive (dead)

TPB from the total TPB measurement is a critical issue[70, 72]. In this study, the TPB

density was calculated based on the 3D reconstruction. For 3D reconstruction, every

voxel was label one of the phases including GDC, Ni, and pore. Generally one edge is

shared by 4 voxels and if an edge is shared by all 3 kinds of phases, then it is counted

as TPB length (Fig. 5-7). In order to estimate the actual working TPB density, the TPB

sites were classified into 3 categories, which are active, inactive, and unknown TPB. For

this, connectivity of each phase at the TPB was traced along xy, yz, and zx plane. If the

Ni, GDC and pores are connected across the AFL from GDC electrolyte to end of the

AFL, it was counted as 'active' TPB. If one of the phases connected to the TPB site was

isolated inside the reconstructed region, which was referred to 'inactive TPB'. Other

cases are sorted as 'unknown' TPBs. The total TPB length (LTPB) was estimated by

summation of the length of the voxel edges counted as the TPB. The TPB density (PTPB)

is calculated by

LTPB
PTPB = (5-2)
Total

where the Vtotai is the total volume of ROI. The unit of TPB density is pm-2 (= Pm/

pm3). For unknown TPBs, it was assumed that the same connectivity of the each phase

exists out of the ROI and the unknown TPBs might have the same portion of the active









and dead TPB in known region. Based on this assumption, the total active TPB density

(PTPB, active, total) was estimated by

P_ TPB,acve I TPB,~unknown (5-3)
PTPB,active,total ~
total L TPBactive +LTPB,dead

where the LTPB,active, LTPB,dead, and LTPB,unknown are active, dead, and unknown TPB

length in a measured volume, respectively. The calculated total active TPB densities

were 8.5 1.4, 15.6 5.0, 14.2 3.4, and 6.3 2.3 pm-2, respectively (Table 5-2). As

expected from total surface area and effective particle size analysis discussed above,

the highest TPB density was achieved in AFL with 60wt% NiO, which is 1:1 volume ratio

of Ni to GDC. Moreover this is in good agreement with the electrochemical impedance

analysis in previous chapter. However, the standard deviation of TPB is greater for

higher values. This result might reflect that the AFL with larger TPB density has higher

randomness and complicity of the structure. This scattering of the TPB density can be

reduced by introducing computational simulation method [72] or morphological

correction factor [71]. Fig. 5-8 shows the TPB result compared to surface area. The

surface area for each sample does not show any trend with AFL composition contrast to

TPB density. It does not seem to be in agreement with results from other studies that

the higher surface area produces higher probability of TPB density as previous shown

previous studies [65, 67]. However, it should be considered that in this study the

effective particle size and porosity was controlled by compositional change of AFL

accompanying volume fraction change between phases, while other studies the phase

composition of the electrodes were fixed. Fig. 5-9-a shows the TPB and electrode ASR

with various AFL compositions. For the electrode ASR, the result was taken from the

previous study [64]. In that study the electrode ASR measured from the AFL with









different NiO contents were presumable assumed that the cathode ASRs were the

same value due to same process of the cathode fabrication for each cell. Therefore, the

electrode ASR trend can show the anodic ASR change. As shown in Fig. 5-9-a, TPB

density shows the inverse trend of electrode (anode) ASR change. Previously, Bieberie

et. al reported that the main electrode process directly depends on the TPB length [73].

In that research, using a Ni pattern anode as a model electrode, they showed that the

electrode conductivity under open circuit condition is correlated to TPB length, which is

linear relationship between inverse resistance and TPB length. Fig. 5-9-b shows the

plotting of one over ASR corresponds to TPB density, also showing the linear

relationship. However, some deviation from the linearity is shown. This can be occurred

by scattering of the TPB length from the 3D reconstruction accuracy or the different

electrochemical mechanism dependence of compositional change of AFL on

electrochemical polarization. Previously, Smith et. al reported that for the LSM-YSZ

cathode the charge transfer and adsorption have a different dependence on TPB length,

which was evaluated by deconvolution of impedance[68]. Therefore, further study about

the effect of TPB density on anode polarization loss will be conducted through the

deconvolution of anode impedance under applied current.

5.4 Conclusions

In this work, microstructural properties of Ni-GDC anode functional layers for IT-

SOFCs were quantified by state-of-the-art 3D reconstruction technique. For 3D

reconstruction, each sample was automatically sectioned and each sectional image was

acquired using FIB/SEM dual beam system. After labeling of each image to give phase

separation, the series of sectioned images were incorporated and reconstructed in 3D

utilizing the Amira software. From this 3D reconstruction, the phase gradient through the









sample depth and volume fraction, effective diameter, surface area for each phases (Ni,

GDC, and pore), and pore tortuosity were quantified. This result showed that the volume

fraction was well matched with theoretical value showing each sample was well

reduced. In addition to volume fraction, the graded phase plot showed that the actual

AFL/ anode structure have some degree of deviation of the each phase volume fraction,

so the accurate quantification microstructural properties could be achieved by bulk

analysis. As one of the most important features in the anode microstructure, the active

TPB density was evaluated with the algorithm of checking the connectivity of voxels for

each phase. The highest TPB length was found at the 1:1 volume ratio of the Ni to GDC

in AFL. Moreover, the TPB densities showed a linear relation to the inverse of electrode

ASR. For the more accurate TPB estimation, the mathematical adjustment of

rectangular voxel will be needed. In addition, the detailed analysis of anode reaction

mechanism by deconvolution of impedance spectrum should be conducted for the direct

relation to electrochemical properties with quantified microstructural features.








FIB



52\
SE.---\


XY
Plane


10Ijm


50ni


Serial sectioning Alignment, Cropping,
-2D images Segmentation


3D reconstruction


Figure 5-1. Schematic diagram of FIB/SEM dual beam system with sample (a) and 3D
reconstruction process (b)


.4~R





















(a) (b)


(c) (d)


Figure 5-2. 3D reconstruction of Ni-GDC anode (a), and AFLs with initial composition of
50 (b), 60 (c), 65 (d), and 80 (e)wt% NiO nearby at anode(or AFL)/electrolyte
interface.





















(a) (b)


(c) (d)


Figure 5-3. Individually reconstructed phases from the 3D reconstruction of AFL with 65
wt% NiO ; GDC (a), Ni (b), Pore (c), and combination of Ni and Pore phases

























Distance from electrolyte (nm)
(a)


Distance from Electrolyte (nm)
(b)
Figure 5-4. Phase gradient of reconstructed samples with no AFL (a), 50 (b), 60 (c), 65
(d), and 80 (e)wt% NiO in Ni-GDC AFL


98








100


Distance from Electrolyte (nm)
(c)


0 2000 4000 6000 8000 1000(


Distance from Electrolyte (nm)
(d)


Figure 5-4. Continued













GDC
S 60-
E
> 40


20-


0 -
0 2000 4000 6000 8000
Distance from Electrolyte (nm)
(e)
Figure 5-4. Continued


100



























I I I I I I
40 50 60 70 80
NiO in AFL (wt %)
(a)


NiO (wt %)
(b)

Figure 5-5. Volume fraction of Ni, GDC and pore phase in total volume (a), and volume
fraction of Ni and GDC in solid volume of AFLs with various compositions.
Open symbols represent theoretical values.


101


Ni
GDC
A Pore





Oi

A A 4 4


0

0








* Ni
* GDC
A Pore







1200
Ni
GDC
A Pore
1000-

E ..-
800-




0 ...... .....

400-



50 60 70 80

NiO in AFL (wt %)

Figure 5-6. Effective particle diameters of Ni (rectangular), GDC (circle), and pore
(triangle) phase of AFLs with various compositions


102



























GDC

Ni

Pore

TPB


Figure 5-7. Schematic diagram of TPB length calculation from 3D reconstruction. A
rectangular parallelepiped represents a voxel in a 3D reconstruction and each
one is labeled as one of phases; Ni, GDC, or Pore phase. A edge which is
shared by all three phases is counted as a TPB length.


103







4.0 -40
I-4
E 3.5- -35

E 3.0 -30

S2.5 .. A .........................A -25

S2.0 -20 (

S 1.5 ............ -15 i

) 1.0- H O "-...... -1o 3
(U)
0.5 -5

0.0 -0
50 60 70 80

NiO (wt %)
Figure 5-8. Plot of quantified surface area and TPB density of AFL with various NiO
contents. Dotted lines are only for guide purpose.


104









40 0 925


-0.20


-0.15 a
CD

-0.10 C>


-0.05


-0.00


NiO in AFL (wt %)
(a)


N








IJ
a)

0

U,
w


0 5 10


15 20 25


TPB Density (pn2)
(b)

Figure 5-9. (a) TPB density and electrode ASR with various AFL compositions (Dotted
lines are only for guide purpose.) (b) plot of 1 over electrode ASR with TPB
density. A red line represents linear fit for the plot showing inverse
relationship between TPB and electrode ASR.


105


E






I-
H-









Table 5-1. 3D reconstruction dimension and total volume fractions of Ni, GDC and pore
phase and solid volume fractions of Ni and GDC
Initial AFL NiO (wt%) 50 60 65 80
Composition GDC (wt%) 50 40 35 20

X (pm) 12.95 8.75 10.45 10.42

3D Y(pm) 10.22 6.06 7.95 7.88
Reconstruction Z(pm) 8.90 10.64 10.76 9.32
Dimension
Total Reconstructed
1177.91 564.21 893.91 765.26
volume (pm3)
Ni in Total Theoretical 30.4 36.3 39.2 47.7
volume(%) Measured(SD) 29.1(-1.0) 37.8(~5.9) 43.2(~6.7) 51.1(-6.4)

GDC in total Theoretical 48.3 38.4 33.4 19.0
volume(%) Measured(SD) 54.6(~3.0) 40.6(~4.8) 35(~5.5) 20.0(~5.9)
Phase
Volume Pore in total Theoretical 21.2 25.3 27.4 33.3
After volume(%) Measured(SD) 16.3(~2.9) 21.6(~3.0) 21.8(~3.2) 28.9(~2.9)
Reduction
Ni in solid Theoretical 38.6 48.6 53.9 71.6
volume(%) Measured(SD) 34.8(~1.0) 48.1(-6.7) 55.1(-7.3) 71.9(~8.0)

GDC in solid Theoretical 61.4 51.4 46.1 28.4
volume(%) Measured(SD) 65.2(~1.0) 51.9(~6.7) 44.9(~7.3) 28.1(-8.0)



Table 5-2. Summary of quantification of microstructural features of AFL with various
compositions
Initial NiO composition in wt% 50 60 65 80
AFL
Effective Ni diameter, d Ni nm 683 771 917 1120
Effective GDC diameter,d GDC nm 890 642 650 316
Effective Pore diameter,d pore nm 381 452 489 560
Surface area per volume, SA/V pm-1 2.4 2.9 2.4 2.6
Tortuosity,T N/A 2.77 1.91 1.91 1.69
TPB density,pTPB(SD) m-2 8.5(~1.4) 15.6(5.0 14.2(3.4 3(2.3)
*Electrode ASR at 650C 2 0.161 0.133 0.142 0.205
*Electrode ASR at 6500C 0-cm2 0.161 0.133 0.142 0.205


106









CHAPTER 6
HIGH PERFORMANCE IT-SOFC WITH CERIA/BISMUTH OXIDE BILAYERED
ELECTROLYTES FABRICATED BY A SIMPLE COLLOIDAL ROUTE USING NANO-
SIZED ESB POWDER

6.1 Introduction

Solid oxide fuel cells (SOFCs) have been widely accepted and studied as a next-

generation energy conversion device. They produces electricity by electrochemically

combining fuel and oxidant across a ceramic ionic conductor, i.e., a solid state

electrolyte [2]. The efficiency of SOFCs is not limited by theoretical Carnot efficiency,

unlike that of combustion-type systems is limited. The fuel-to-electrical efficiency can

reach approximately 45 to 60%. Considering utilization of the by-product heat in co-

generation or bottoming cycles, the projected system efficiency can exceed 80%. In

addition to high efficiency, SOFCs are attractive because of their reduced production of

SOx and NOx, and significantly lower green house gas emissions compared to

combustion engines [3].

High system cost is the one of the largest barriers to the commercialization of

SOFC technology. Ionic conduction in the solid electrolytes is a thermally activated

process, leading conventional SOFCs to operate in a temperature range from 900 OC to

1000 OC. Consequently this high operation temperature requires the use of ceramic

interconnects, high temperature seals, and supper-alloy based balance-of-plant

components, resulting in prohibitive system costs [9]. As Steele claimed, in lower

temperature operation (500~700 OC), the system cost can be significantly reduced by

allowing for the use of cheap stainless steel for the bipolar plates and the balance-of-

plant, combined with the use of high temperature gaskets rather than rigid glass-based

seals, which can also enhance its mechanical stability and lifetime [5].


107









Recent research efforts aim to develop alternative SOFC materials that operate in

the 500 to 800 C range. There are two major issues for lower temperature operation of

SOFCs. The first is that the reduced oxygen ion conduction in ceramic electrolytes

causes a significant increase in ohmic polarization at low and intermediate

temperatures. The second is that electrode polarization is significant due to reduced

cathode activity at the electrode and electrolyte interface. Therefore, development of

electrolytes with high ionic conductivity in the IT range (500 ~ 700 C) has been widely

investigated by many researchers [5, 7].

Among the studied electrolyte materials, doped-ceria and stabilized bismuth oxide

have been reported to have 1~2 orders of magnitude of higher ionic conductivity than

conventional yttria stabilized zirconia electrolytes [74]. However, doped ceria shows

lower open circuit potential (0.7~0.8 V) than the theoretical Nernst voltage due to its

mixed ionic electronic conductivity (MIEC) causing electronic leakage and low power

density [31]. Moreover, although stabilized bismuth oxide has very high oxygen ionic

conductivity, its inherent thermodynamic instability under reducing conditions makes it a

poor choice by itself as an electrolyte for SOFCs [37].

To overcome these problems, Wachsman et al. suggested a bismuth oxide/ceria

bilayer electrolyte consisting of a layer of stabilized bismuth oxide on the oxidizing side

and a layer of doped ceria on the reducing side [18]. In this arrangement, the ceria layer

can improve the thermodynamic stability of the bismuth oxide layer by shielding it from

very low Po2 and the bismuth oxide layer can serve to block electronic flux from doped

ceria in reducing atmospheres, theoretically yielding high OCPs approaching the Nernst

potential at the IT range. For over a decade, several researches have demonstrated


108









that the OCP can be effectively improved using the bismuth oxide/ceria bilayered

electrolyte concept [18, 39-41]. However, high power density at low temperature was

not readily achieved due to high ohmic losses from thick electrolyte-supported cell

designs and the high reactivity of bismuth oxide with conventional perovskite cathodes,

such as Lao.6Sro.4Coo.2Feo.803-6 (LSCF) due to weak metal-oxide bonds [75]. Therefore,

recent research has focused on development of thin erbia stabilized bismuth oxide

(ESB) / gadolinia doped ceria (GDC) bilayered electrolytes and ESB-compatible

cathodes. Previously, Park and Wachsman reported that an ultra thin (~0.2 pm) and

dense ESB layer can be deposited on Sm-doped ceria(SDC) pellets by pulsed laser

deposition (PLD) [39, 40]. In addition, to overcome the reactivity of ESB with

conventional perovskite cathodes, a bismuth ruthenate (Bi2Ru207, BRO7)- ESB

composite cathode was developed [76]. A recent optimization study by Camaratta et al.

reported that BRO7-ESB composite cathodes exhibited very low ASR (0.73 and 0.03 0

cm2 at 500 and 700 C, respectively) [77].

Based on these studies, we demonstrated an impressively high performance of

~1.94 W/cm2 at 650 C using a thin ESB/GDC electrolyte on an anode-supported cell

with highly optimized BRO7-ESB cathodes [19, 20]. To accommodate dense and thin

electrolytes on porous anode-supports, we integrated a recently developed a GDC

anode functional layer (AFL) between the anode and GDC electrolyte by spreading of a

precursor solution [14]. In this study thin (~4 pm) and relatively dense ESB was

successfully deposited by PLD technique on 10 pm thick GDC film. The obtained total

ASR was only 0.079 Q-cm2, showing a ~40% decrease compared to a cell with a single

GDC electrolyte.


109









Up to now, these various studies have soundly proven that the bilayered

electrolyte concept is highly encouraging in practical SOFC applications for low

temperature operation. One of the key factors to fabricate ESB/GDC bilayered

electrolytes is to obtain a dense and thin ESB layer on the sintered GDC electrolyte.

Although PLD can be good for demonstration purposes in lab-scale experiments, a

more cost-effective and practical fabrication process is necessary for future application

of the ESB/GDC bilayered electrolyte.

In this study, we fabricated a thin and dense ESB/GDC bilayered electrolyte on

anode-supported SOFCs by a simple and cost-effective colloidal deposition process. To

obtain a dense ESB layer, nano-sized ESB particles used for the colloidal coating slurry

were synthesized by a wet chemical co-precipitation method. In order to optimize the

sintering conditions of the ESB layer on dense GDC, the evolution of the ESB

microstructure with sintering temperature was investigated. To gauge the performance

of the developed ESB/GDC electrolyte, current-voltage characteristics and

electrochemical impedance tests were carried out.

6.2 Experimental Procedure

6.2.1 ESB Powder Fabrication

To synthesis very fine ESB powder, a coprecipitation route was employed. Pure Bi

nitrate and Er nitrate were used as starting raw materials. They were weighed in

stoichiometric proportions and dissolved in 70% nitric acid to produce a solution. An

excess ammonia solution (Acros Organics, 28- 30% of NH3 solution in water) was

added to the stirred solution to increase the pH value to 12. The addition of the

ammonia solution resulted in the formation of a yellowish-brown color precipitate. The

precipitate was filtered, and then subsequently dried at 80 C for 12 hrs. The


110









agglomerated powder was then ground into fine particles using a mortar and pestle. The

powder was then calcined at 900 C for 10 hrs in air.

For comparison purpose, the ESB powder was synthesis by the conventional

solid-state route. A stoichiometric mixture of Bi203 (99.9995% pure) and Er203 (99.99%

pure), from Alfa Aesar, were mixed and ball-milled with zirconia ball media in a high-

density polyethylene bottle for 24 hrs. After drying, the mixed powders of ESB were

calcined at 800 C for 16 hrs. Agglomerated powders were ground using mortar and

pestle and sieved using a 325 pm mesh.

6.2.2 Fuel Cell Fabrication

The SOFC fabrication involved GDC spin coating on tape-cast anodes followed by

ESB colloidal drop coating. The anode support was prepared by tapecasting 65 wt% of

NiO (Alfa Aesar) and 35 wt% of GDC (Rhodia) with an appropriate amount of solvents

and organic compounds. The anode tapes were presintered at 900 C for 2 hrs. The

GDC AFL was deposited by spraying GDC precursor solution on presintered anode

surface. Subsequent heat treatment at 900 C was carried out for the removal of the

binder system. Detailed preparation and fabrication of GDC AFL on tape-cast anode

were described the previous study [14].

Thin and uniform GDC electrolytes were deposited by spin coating with a GDC

colloidal slurry. The Rhodia GDC powder was ball milled for 24 hrs with solsperse

dispersantt) in ethanol. PVB (binder) and DBP plasticizerr) were added after the first

ball-milling step and the solution was ball-milled for an additional 24 hrs. For spin

coating, the anode substrates were fixed on the vacuum chuck of the spin coater. The

thickness of the GDC electrolyte was controlled by the number of depositions with same

spin speed. After deposition, samples were dried at room temperature for 10 hrs. After


111









drying, the multilayer anode/electrolyte structure was sintered at 1450 C for 4 hrs using

a ramp rate of 3 OC /m in air.

For the ESB/GDC bilayered electrolyte, the ESB layer was colloidally deposited by

drop coating with co-precipitated ESB (cp-ESB) powder. To make a ESB colloidal

slurry, co-precipitated ESB powder was mixed with a binder system which consisted of

Solsperse, DBP and PVB as the dispersant, plasticizer, and binder, respectively. This

mixture was ball-milled in ethanol for 24 hrs, and drop-coated onto the sintered GDC

electrolyte surface. The drop-coating was repeated until a desired thickness was

achieved. To see the effect of sintering temperature on formation of ESB layer on dense

GDC electrolyte, as-deposited ESB/GDC bilayered cell was divided into several pieces

by a diamond saw and each sample was sintered at 700, 800, and 900 C for 4 hrs

using a 400 C 1 h binder burnout step, and a 5 C/min ramp rate. For comparison, the

same experiments were repeated using solid state ESB powder.

Two different composite cathodes were used for this study-- LSCF-GDC (50:50

wt%) on GDC electrolyte and A BRO7-ESB (50:50 wt%) on ESB/GDC bilayered

electrolyte. The cathode development and cathoding procedure can be found earlier

work [19, 77].

6.2.3 Characterization

The phase and size of the crystallites of as-calcined ESB powders were

investigated by means of X-ray diffraction analysis (XRD, Philips APD 3720).

Microstructures of ESB powders and fuel cell structures with ESB/GDC bilayered

electrolytes were observed using scanning electron microscopy (SEM, JEOL 6400 /

6335F). Qualitative elemental analysis of the fuel cell structure was conducted by

energy-dispersive X-ray spectroscopy (EDX).


112









For electrochemical performance measurement, fuel cell samples were loaded in a

fuel cell testing set-up. Current-voltage (I-V) characteristics were conducted by a

Solartron 1287 potentiostat. The gas condition used was 30 sccm of dry air and 90

sccm of wet hydrogen to the cathode and anode side, respectively. To avoid gas

leakage, the edge of cell and testing alumina tube were covered with ceramic sealant

using the mixture of two part ceramabond-517 (liquid and power, Aremco). After the I-V

measurement, two-point probe impedance analysis was carried out under open circuit

condition using a Par-stat 2273 (Princeton Applied Research) over a frequency range of

100 KHz to 100 mHz.

6.3 Result and Discussion

6.3.1 Powder Characterization

Fig. 6-1 shows the XRD result of ESB powders synthesized by co-precipitation

(cp) and solid-state (ss) route. For cp-ESB, the precursor powder was calcined at 500

C for 4 hrs, while ss-ESB powder was synthesized by calcined at 800 C for 16 hrs. As

shown in Fig. 6-1-a, both cp-ESB and ss-ESB show a cubic-fluorite structure of doped-

bismuth oxide without any other phases. This result indicates that the wet chemical co-

precipitation method for synthesis of ESB powder is a greatly effective way to reduce

the calcination temperature and processing time compared to the conventional solid-

state route. Due to less thermal energy input during cubic-fluorite phase formation, we

can expect lowered grain growth and significantly reduced crystallite size of cp-ESB

powder. Generally the crystallite size of the material can be calculated by Scherrer

equation [78];

B(20) = (6-1)
Lcos


113









where K is the shape factor, A is the x-ray wavelength, typically 1.54 A, B is the

line broadening at half the maximum intensity (FWHM) in radians, e is the Bragg angle,

and L is the crystallite length. Fig. 6-1-b shows the magnified x-ray diffraction pattern of

the (111) peak from Fig. 6-1-a at the theta range from 27 to 29 degree. Using this

graph, we calculated the crystallite size of the both ESB powers by eq. (1) and they are

summarized in Table 6-1. The crystallite size of cp-ESB was 1/3 the size of ss-ESB,

indicating a high possibility of smaller particle sizes of cp-ESB powder.

The microstructural analysis by SEM was carried out to investigate the size and

morphology of the ESB powder. Fig. 6-2 shows SEM images of the ESB prepared from

different synthesis methods, that is, conventional solid state and wet-chemical route

using co-precipitation. As shown in Fig. 6-2-a, the particle size of cp ESB powder is

much less than 5 pm and each particle consists of soft agglomeration of nano-sized rod-

shaped particulates with high aspect ratio. In contrast, the ss-ESB powder in Fig. 6-2-b

shows particle size over ~5 pm and each particle appears to be hard-agglomerated.

This result indicates that the co-precipitation process successfully reduced the resultant

ESB particle size and significantly enlarged surface area, allowing for much higher

sinterability.

6.3.2 Effect of Sintering Temperature on ESB/GDC Bilayered Electrolyte

In order to investigate the effect of particle size and sintering temperature on the

formation of an ESB layer on a GDC electrolyte, ESB/GDC bilayered electrolytes were

fabricated using cp-ESB and ss-ESB powder at various sintering temperatures (700,

800, and 900 C).

The SEM images in Fig. 6-3 show cross-sectional views of the bilayered structures

on anode supports fired at various temperatures. Using backscatter imagery, the


114









different phases-ESB (white), GDC (light gray), and NiO (dark gray)-are easily

distinguished due to differences in elemental contrast. It should be noted that the SEM

images for each ESB layer (Fig. 6-3-a, c, e, g for ss-ESB and b, d, f, h for cp-ESB) were

taken at different magnifications. From these microstructures, it is clear that both cp-

and ss-ESB layers increase in density with increasing sintering temperature. However,

the ESB layer which was prepared from ss-ESB powder and fired at 900 C is not fully

dense. Instead, this ss-ESB electrolyte looks less dense when fired at 900 C (Fig. 6-3-

g) than when fired at 800 C (Fig. 6-3-e) due to the formation of large pores. It is known

that the bismuth oxide melting temperature is ~ 825 C [79]. As Jiang and Wachsman

reported, however, doped bismuth oxides have much higher melting temperature of

over 2000 C and differential thermal analysis (DTA) also found no melting endotherm

for the stabilized bismuth oxide up to 1100 C [80]. Therefore, we believe that this

phenomenon is possibly caused by the sublimation of bismuth oxide phase over 8250C.

Moreover, the sintering of ESB layer is processed on a highly densified GDC substrate

which was sintered at a much higher temperature (~1450 C). In this case, the sintering

mechanism of ESB powder is primarily dependent on vertical shrinkage without lateral

shrinkage. Therefore, ss-ESB powder might not establish a dense layer below 800 C

due to its insufficient surface area leading low sinterability. Additionally, ss-ESB powder

will be porous when fired above 900 C due to the sublimation of bismuth oxide phase.

This result indicates that micron-sized ESB powder prepared by conventional solid state

synthesis is not appropriate for colloidal deposition of the ESB layer in ESB/GDC

bilayered electrolytes.


115









On the other hand, as shown in Fig. 6-3-b, d, f, h, when cp-ESB powder was used,

the ESB layer was highly dense when fired at 800 C. SEM powder analysis (Fig. 6-3-

a), of the cp-ESB powder shows that it has a very high surface area due to the nano-

sized nature of the particles. Therefore, it is possible for cp-ESB powder to produce

sufficient surface energy at lower sintering temperatures for the particles to neck and

densify by surface and lattice diffusion. Unlike the layer prepared from ss-ESB powder,

the cp-ESB layer sintered at 900 C (Fig. 6-3-h) looks very dense. However the ESB

thickness was reduced compared to the same layer sintered at 800 C (Fig. 6-3-f).

Since 900 C is above the pure bismuth oxide melting temperature, as discussed

above, this indicates that partial sublimation of bismuth oxide phase or penetration of

bismuth oxide into GDC layer have possibly occurred.

It should be noted that as seen in Fig 6-3-g, h, both ss- and cp- ESB layers

sintered at 900 C show diffusion of ESB phase into ceria layer and segregation of it

into GDC grain boundaries. This is also clearly shown in Fig. 6-4 which is a magnified

image of Fig. 6-3-g. It is considered that the over melting temperature of pure bismuth

oxide (~825 OC) the partial pressure or fugacity of bismuth oxide might be high, which

can cause higher activity of bismuth oxide or bismuth in ESB lattice. Therefore, this high

activity can accelerate motion of bismuth oxide phase, leading diffusion of bismuth

oxide phase along the GDC grain boundaries which generally has a high surface

energy. It has also been reported that bismuth oxide can be soluble in CeO2 forming a

solid solution of Bil-xCexO2-xO2-x/2 with a cubic fluorite structure [81]. Gil et al recently

reported that the solid solubility limit of bismuth oxide in doped ceria is ~ 0.8wt% [82].

Park and Wachsman predicted the possible existence of solid solutioning at the


116









interface between the stabilized bismuth oxide and doped ceria bilayered electrolytes

[40]. In this study, the conductivity of 2 mm thick Sm doped ceria (SDC) was lower than

that of ESB (~ 0.2 pm)/SDC (~ 2 mm) bilayered electrolyte due to the high grain

boundary conductivity of ESB/SDC. This phenomenon was explained as the formation

of a solid solution of ESB into the grain boundary of SDC, which can lower the activation

energy for oxygen ion transfer in SDC grain boundaries by a scavenging effect on

impurity phases. Therefore, we can expect a positive effect on lowering ohmic

polarization in the observed integrated ESB/GDC structure.

On the other hand, it is also shown that the fracture mode of GDC electrolyte with

ESB sintered at 900 C changed from inter- granular to intra-granular (Fig. 6-4),

compared to GDC cross-sections near ESB layers sintered at lower temperature (Fig. 6-

3-a ~ f). This might cause deterioration in the mechanical strength of the GDC

electrolyte due to highly segregated bismuth oxide phase along the GDC grain

boundary. Therefore, this phenomenon should be carefully controlled and further

investigation will be needed.

Consequently, we could obtain dense ESB/GDC bilayered electrolytes by a simple

colloidal deposition using fine ESB powders made by a co-precipitation method. The

optimal sintering temperature for synthesizing dense ESB layers by colloidal deposition

was ~800 C, which limited bismuth oxide sublimation and penetration into the GDC

electrolyte.

6.3.3 Microstructure of a Full Button Cell with ESB/GDC Bilayered Electrolyte

For comparison of electrochemical performance, two kinds of button cells were

prepared a single GDC electrolyte cell and an ESB/GDC bilayered electrolyte cell

based on above study. To obtain similar electrolyte thickness, the colloidal deposition


117









process for each layer was repeated. For microstructural analysis before

electrochemical performance, another button cell with ESB/GDC bilayered electrolyte

was fabricated with the similar process. An ESB layer was deposited using colloidal

slurry containing cp-ESB powder and sintered at 800 C for 4 hours. Fig. 6-5 shows the

cross-sectional SEM image of NiO-GDC anode/ ESB-GDC bilayered electrolyte/ BRO7-

ESB composite cathode. As shown in this figure, a dense ESB/GDC layer was obtained

and every layer is clearly distinguishable with good interfacial contact. To examine the

possible interdiffusion between layers, an energy dispersed x-ray (EDX) line scan

analysis was conducted along the base line (yellow) in Fig. 6-5. Three elements were

traced along the line, which includes Bi (red line), Ce (blue line), and Ni (white line). The

resultant data are overlapped in Fig. 6-5 and no solid state reaction or interdiffusion

between layers was observed.

Fig. 6-6 shows the microstructures of cross section and surface of anode

supported cells with a single GDC electrolyte and an ESB/GDC electrolyte after

electrochemical performance test. Both cross-section and surface view of the GDC

layer (Fig. 6-6-a,c) shows very high density without pores. In contrast, the ESB surface

(Fig. 6-6-d) shows some porosity and lower density compared to the GDC surface,

while the cross-section of ESB (Fig. 6-6-b) looks dense. For further densification, it

needs to be opmisized. As shown in Fig. 6-6, each electrolyte thickness was measured

from the SEM image. The GDC electrolyte thickness was identical, ~9 pm for each

cells. The ESB electrolyte thickness was ~4 pm.

6.3.4 Performance of a Button Cell with ESB/GDC Bilayered Electrolyte

Fig. 6-7-a shows the I-V characteristics of cells with GDC single layer and

ESB/GDC bilayered electrolyte at 650 C. The measurement was conducted under 90


118









sccm of 3% wet hydrogen to anode side and 30 sccm of dry air to cathode side. The

maximum power density (MPD) of the ESB/GDC electrolyte reached ~1.47 W/cm2 with

a 69% increase in MPD compared to a GDC single layer (~ 0.87 W/cm2). The OCPs of

the single GDC electrolyte cell and ESB/GDC bilayered electrolyte cell were 0.75 and

0.80 V, respectively. This OCP improvement with bilayered electrolyte can be evidence

that the ESB layer in ESB/GDC bilayered electrolyte may effectively block the electronic

conduction from the GDC due to reduction of Ce4+ to Ce3+ at low Po2 region. Moreover,

we previously reported similar OCP enhancement with a thin ESB/GDC bilyaered

electrolyte. At that time we fabricated a dense ESB layer of ~ 4 pm by PLD technique

on ~ 10 pm thick GDC electrolyte by spray coating method, resulting OCP increase

from 0.72 to 0.77 V [20]. It is noted that in this study we obtained higher OCP for both

single and bilayered electrolyte cells compared to previous research results. We believe

that this is caused by higher GDC density due to spin coating method instead spraying

coating for GDC electrolyte in the previous study. Therefore, this result demonstrates

that thin and dense ESB/GDC bilayered electrolytes can be achieved by conventional

colloidal process by control of initial powder morphology and sintering condition.

Although the ~ 0.05 V increase in OCP was achieved in a dense bilayered electrolyte,

the major contribution of this outstanding enhancement in MPD came from ASR drop.

The estimated total ASRs from IV curves near OCP region were 0.084 Qcm2 and 0.164

)cm2 for cells with ESB/GDC bilayer and GDC single layer, respectively.

For the further analysis electrochemical impedance test was conducted and the

result is shown in Fig. 6-7-b. In these Nyquist plots the total and ohmic ASR were

extracted from the low and high frequency intercepts of the impedance spectra with the


119









real axis, respectively, and the electrode ASR was calculated from the difference

between these two ASR values. The detailed values are tabulated in Table 6-2. The

total ASR of ESB/GDC bilayered cell (0.088 0-cm2) was decreased by 47.6% compared

to that of cell with GDC single layer (0.168 0-cm2). In addition, in this study the total

ASRs from this AC impedance tests are identical to the ASRs from the I-V curves (DC

impedances) within 5% deviation showing the good validity of the results. The electrode

ASR of the bilayered electrolyte cell was decreased by 62.6% compared to that of GDC

single cell. This ASR drop (0.062 0-cm2) is expected to mostly come from the cathodic

polarization reduction at the interface between high catalytic BRO7-ESB cathode and

ESB electrolyte, considering the both cells utilized the same anode material using

identical procedures. This result has been also well-supported by the former studies [19,

77]. The ohmic ASR was also decreased (26.1%, 0.018 0-cm2) at 650 C. Previously, it

was shown that the ohmic ASR (ASRelectrolyte) of a bilyared electrolyte can be estimated

by a simple modeling for two series resistors written as


ASelectroyte = LGDC- + J (6-2)
C 0ESB -GDC )

where LGDC is thickness of GDC electrolyte, T is thickness ration of ESB to GDC,

and oESB and OGDC are the conductivities of ESB and GDC, respectively [19, 42]. Based

on known ESB and GDC conductivities [83], a similar or slightly higher ohmic ASR of

the ESB (-4 pm)/GDC (-9 pm) bilayered electrolyte than that of ~a single 9pm thick

GDC can be expected by eq. (2) due to much higher conductivity of ESB than GDC at

IT ranges. However, previous studies repeatedly showed that bilayer electrolyte can

effectively reduce the electrolyte ASR even when we used same GDC thickness with a

single GDC layer and an additional ESB layer was deposited [19, 20, 39]. Therefore, we


120









believe that this phenomenon is related to penetration and segregation of bismuth oxide

phase at the GDC grain boundary, even though it was not defective in EDX analysis.

Further discussion about ASR reduction of ESB/GDC bilayered electrolyte was

exposited elsewhere [19, 20].

However, this MPD result was somewhat lower compared to the previous highest

MPD with thin ESB by PLD and GDC bilayered electrolyte cell (~1.95 W/cm2) [20]. As

ohmic ASR results suggested, the density of cp-ESB layer might be still less dense

compared to PLD-ESB. Moreover, in this study, we used both solid state BRO7 and

ESB powder for the cathode, while in the previous study microstructurally optimized

BRO7-ESB cathode was used. Optimization of BRO7-ESB was described in detail by

Camaratta el. al [77]. Therefore, we can expect even higher MPD result by further

improvement of ESB density and cathode performance.

6.4 Conclusions

We fabricated thin and dense ESB/GDC bilayered electrolytes by a simple

colloidal deposition process. For producing highly sinterableESB particles, a wet

chemical co-precipitation route was introduced. Using this nano-scale ESB powder with

high surface area, a dense and very thin (~ 4 pm) ESB layer was established on a GDC

electrolyte by simple colloidal drop coating. The optimal sintering temperature of the

ESB layer was found to be ~800 C due to its poor densification at lower temperatures

and evaporation at higher temperatures. Finally, an anode-supported SOFC with the

thin and dense ESB/GDC bilayered electrolyte coupled to a BRO7-ESB cathode

produced a very high maximum power density of ~ 1.5 W/cm2 at 650 C. This was

possible due to the effect of the bilayered electrolyte on significantly decreasing eletrode

ASR as well as ohmic ASR. This study demonstrated that the high performance of


121









ESB/GDC bilayered electrolyte can be reproducible by the cost-effective and practical

colloidal deposition. Based on this study, fabrication of large-scale planar cells with

bilayered electrolytes for stack cell application is underway.


122
























20 30 40 50 60 70 80
20 angle

(a)


7 28 29
28 angle


(b)

Figure 6-1. XRD diffraction pattern of ESB powders synthesized by coprecipitation
route (red line) and solid state route (black line) (a). The magnified XRD
diffraction pattern of the (111) peak is shown at the 20 range from 27 to 290
(b).


123


Coprecipitation
(5000C 4h)


Solid State
(8000C 16h)





























































Figure 6-2. SEM of ESB powders synthesized by wet-chemical co-precipitation method
(a) and solid-state route (b).


124


hh I
























":..'%" r ,, '


0 2..
r 6- E o G e
N 00




















tha tha of ssES podr
I..
CD CD CD3





II I


I It

CDI






4L





~1 M
1, .1









tha th t ofs-E Bpodr
I.' a)
a r ;lf~*





& s r 1

Figue 63. Eoluion f EB lyerson DG eectolye atvarous intrin
tepratre usn sEB(cg n pEB(bdfh.I sntdta h
magnEifiato ofr imge for ES lcrlt snpEBpweshge
that tha ofssES pwdr


125





























Figure 6-4. Cross-sectional view of GDC electrolyte under ss-ESB layer after sintering
at 900 C, which is the magnified image from Fig. 5-3g. In backscattering
mode, ESB (white), GDC (light gray), and NiO (dark gray) phases are well-
distinguishable, showing ESB penetration into GDC grain boundaries.


Figure 6-5. Cross-sectional SEM image of a full button cell with ESB/GDC bilayered
electrolyte. EDX line scan was conducted along the straight base line (yellow)
and the intensity of each elements are presented as red (Bi), blue (Ce), and
white (Ni) lines


126


M









Nli im


Figure 6-6. SEM image of cross-sectional view of a single GDC electrolyte cell (a) and
ESB/GDC bilayered cell (b). Surface views of GDC electrolyte (c) and ESB
electrolyte (d) are shown.


127




















































Figure 6-6. Continued


128


































0.06-

S0.04-

0.02-
ri


0.00 0.02
0.00 0.02


1.5




0
1.0



()

0.5





0.0


Current Density (A/cm2)
(a)


GDC
ESB/GDC A '--A-A


0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18


Z (-cm2)
(b)



Figure 6-7. IV-Characteristics (a) and impedance spectra (b) of ESB/GDC bilayer (red
square) and GDC single layer (blue triangle) cell


129


I










Table 6-1. Calcination condition and crystallite size of ESB powders synthesis by co-
precipitation and solid state route.
Synthesis Calcine condition FWHM (026) Crystallite size (A)

Co-precipitation 5000C for 4 h 0.48 171
Solid State 8000C for 16h 0.16 512



Table 6-2. Comparison of specification and electrochemical performance of the studied
cells.
GDC ESB Total Total Ohmic Electrode
OCP MPD
Cell Type thickness thickness ASRiv ASREIS ASR ASR Ref
pm pm V W/cm2 Q.cm2 Q.cm2 Q.cm2 Q.cm2
GDC 9 0.75 0.87 0.164 0.168 0.069 0.099 th
study
this
cp-ESB/GDC 9 4 0.80 1.47 0.084 0.088 0.051 0.037 stu
PLD-ESB/GDC 10 4 0.77 1.94 0.075 0.079 0.046 0.033 [20]


130









CHAPTER 7
HIGH PERFORMANCE LSM-BASED CATHODE BOOSTED BY STABILIZED
BISMUTH OXIDE FOR LOW TO INTERMEDIATE TEMPERATURE SOFCS

7.1 Introduction

Lal-xSrxMnO3-6 (LSM) -based cathodes are some of the most widely used

cathodes for high temperature (over 800 OC) solid oxide fuel cells (SOFCs) due to their

high thermal and chemical stability, electrical conductivity, as well as their catalytic

activity for oxygen reduction. In addition, LSM-based cathodes exhibit good

compatibility with conventional electrolyte materials, such as stabilized zirconia and

doped ceria [2, 84-86]. However, despite these advantages, LSM-based cathodes are

not a popular choice for reduced temperature SOFCs. This is due to the fact that LSM

has negligible ionic conductivity and a high activation energy for oxygen reduction due

to slow oxygen incorporation reaction into the solid lattice, leading to significantly

deteriorated electrochemical catalytic effect at low temperatures [7, 63, 87, 88]. To

overcome this problem, dual-phase composite cathode systems have been explored.

Conventional ionic conducting phases such as yttria-stabilized zirconia (YSZ) and

godolia-doped ceria (GDC) have been mixed with LSM leading to significantly enhanced

electrochemical performance [84, 89-94]. For instance, Murray and Barnett reported

that the area specific resistance (ASR) of pure LSM was lowered from 51.03 Q-cm2 to

11.37 O-cm2 for LSM-YSZ and 6.81 O-cm2 for LSM-GDC [90]. The improved

performance is a result of the increased number of three phase boundary (TPB)

reaction sites between gas, electronic and ionic conducting phases, and by the addition

of a pathway for ionic species to be transported thru the cathode. Thus it is expected

that dual phase cathodes which incorporate materials with an ionic conductivity will yield

further improvements in performance.


131









Stabilized bismuth oxide has been reported to have one of the highest known ionic

conductivities-one or two orders of magnitude higher than the conventional material

YSZ at intermediate temperatures (IT) [17, 95]. In addition to its high ionic conductivity,

bismuth-oxide enhances surface oxygen exchange rate as well as charge transfer and

oxygen dissociation, which is believed to be the rate-limiting step in the oxygen

reduction reaction at the cathode [96-100]. Therefore, stabilized bismuth oxide is a good

candidate for the ionic conducting phase in composite cathodes. For example, the

addition of yttria or erbia stabilized bismuth oxide (YSB or ESB) into a metallic cathode

silver (Ag) improved the performance of the cathode both by increasing TPBs and

forming an ionic conduction path to the electrolyte [101-103]. However, the long term

stability of Ag-based cathodes is poor [103]. Recently, we developed a bismuth-

ruthenate Bi2Ru207 (BRO7)-ESB composite cathode [76, 77]. Using this cathode on an

ESB/GDC bilayered electrolyte, we demonstrated exceptionally high power density (~

1.94 W/cm2 at 650 C) [19, 20]. It was suggested in the above study that cathode

performance can be promoted not only by the addition of ESB into the cathode but also

by using an ESB electrolyte at the cathode interface due to effect of ESB on

improvement of oxygen dissociation followed by reduction of cathode polarization

losses.

Recently, in order to overcome limitations of LSM-based cathodes, the utilization

of stabilized bismuth oxides as the ionic conducting phase has received much attention

[104-113]. Jiang et al. reported the performance of LSM-YSB composite cathodes,

showing an interfacial resistance of ~1.0 Q-cm2 at 600 C, which is 11 times lower than

reported for LSM-YSZ cathodes [106]. Moreover, various fabrication techniques were


132









used in order to obtain increased TPB lengths by microstructural evolution [106, 107,

112]. Although these efforts progressively reduced the cathode ASR, the reported

power densities of SOFCs utilizing LSM-bismuth oxide cathodes are still relatively low at

IT (-600 mW/cm2 at 6500C) [107]. To date, most of the studies of LSM-bismuth oxide

composite cathodes have been conducted on zirconia-based electrolytes which exhibit

significant ohmic resistance at intermediate temperatures due to its high activation

energy for ionic conduction. Even though ultra thin YSZ electrolytes successfully reduce

total ohmic polarization losses at reduced temperature [8], significant deterioration of

ionic conduction and oxygen dissociation rate at the cathode/electrolyte interface is

inevitable due to the abrupt change in ionic conduction phase from bismuth oxide in the

cathode bulk to the zirconia electrolyte. Therefore, modification of the interface between

the LSM-bismuth oxide cathode and the electrolyte is needed in order to achieve high

performance at lower temperatures.

In this study, to overcome this, we introduced an (Er203)o.20(Bi203)0.80 (ESB)

electrolyte layer coupled to the (Lao.80oSr.20)MnO3-6 (LSM)-ESB composite cathode. To

gauge the effect of the ESB electrolyte on the cathode performance, the interfacial

resistance of the cathode on ESB and on GDC was measured on symmetric cells. The

actual performance improvement of the LSM-ESB cathode on an ESB electrolyte layer

was evaluated on anode-supported cells by I-V characterization.

7.2 Experimental

7.2.1 Sample Fabrication

The LSM-ESB composite electrode was prepared by simple powder mixing.

Commercial LSM powder with a surface area of 5.6 m2/g was purchased from Fuel Cell

Materials. The ESB powder for the cathode was synthesis by the conventional solid-


133









state method. A stoichiometric mixture of Bi203 (99.9995% pure), and Er203 (99.99%

pure), from Alfa Aesar, were mixed and ball-milled with zirconia ball media in a high-

density polyethylene bottle for 24 hrs. After drying, the mixed powders of ESB were

calcined at 800 C for 16 hrs. Agglomerated powders were ground using a mortar and

pestle and sieved using a 325 pm mesh to get uniform particle sizes. For the LSM-ESB

electrode ink, LSM powder and ESB powder of the same weight ratio (50:50wt%) were

mixed with a binder system, which consists of alpha terpineol (Alfa Aesar), Di-n-butyl

phthalate (DBP) and ethanol as binder, plasticizer and solvent, respectively. To make

symmetric cells, we prepared both ESB and GDC electrolyte pellets by uniaxial pressing

and sintering at 890 C for 16 hrs and 1500 C for 10 hrs, respectively. Once an

appropriate viscosity was reached, the electrode slurry was applied to both sides of the

electrolyte substrates by brush painting. After drying the symmetric cells at 120 C for 1

hrs, a second coat of electrode slurry was applied to the electrolyte substrates. The

doubly-coated cells were then sintered at 800 C for 2 hrs in air. After sintering the

electrode, silver mesh current collectors and platinum lead wires were pressed against

the samples in a quartz reactor using a ceramic screw-and-bolt assembly.

For full button cell fabrication, the anode-supported cell structures with thin

electrolytes were employed. NiO-YSZ (65:35wt%) anode-supports and anode functional

layers (AFLs) consisting of NiO-GDC (65:35wt%) were tape-casted and attached to

each other by uniaxial press on heating substrate and presintered at 900 C. Thin and

uniform GDC electrolytes were prepared by spin coating. In order to investigate the

effect of the ESB electrolyte on LSM-ESB cathode, ESB/GDC bilayered electrolyte was

prepared. To fabricate a thin and dense ESB layer on a GDC electrolyte, nano-sized


134









ESB powder was synthesized by wet chemical coprecipitation method. Using a colloidal

solution containing nano-sized ESB particles, the ESB layer was deposited by spin

coating. A detailed fabrication and discussion of ESB/GDC bilayered electrolytes was

described in the previous work [19]. The multilayer of ESB/GDC/AFL/NiO-GDC anode

structures is co-sintered at 1450 C at 4 hrs. After sintering, the LSM-ESB cathode was

applied in two coats on both ESB and GDC electrolyte surfaces by brush painting and

sintered at 800 C for 2hrs.

7.2.2 Characterization

The phase identification of LSM, ESB, and LSM-ESB composite cathodes were

investigated by means of X-ray diffraction analysis (XRD, Philips APD 3720).

Microstructures of fuel cells with LSM-ESB cathodes were obtaining by scanning

electron microscopy (SEM, JEOL 6400 / 6335F).

The interfacial polarization resistance of the LSM-ESB electrodes on symmetric

cells was conducted through two-point probe electrochemical impedance spectroscopy

(EIS) using a Solartron 1260. The measurement condition of impedance was an AC

voltage amplitude of 50 mV over the frequency range of 0.1 MHz to 0.1 Hz in air. The

frequency response analyzer was used in standalone mode and interfaced to a

computer using Zplot software. Measurements were made from 500 to 700 C with an

interval of 50 OC.

For electrochemical performance of the button cells, samples were loaded in the

testing setup and sealed with cerambond sealant system. Current-voltage (I-V)

characteristics were conducted by a Solartron 1407E under 90 sccm of dry air and 3%

wet hydrogen to the cathode and anode side, respectively.


135









7.3 Result and Discussion

7.3.1 Impedance Spectroscopy for Symmetric Cells

In order to confirm the compatibility of ESB with LSM, a LSM and ESB powder

mixture (50 to 50wt%) was annealed at 9000C for 50 hrs. Fig. 7-1 shows the resultant X-

ray diffraction (XRD) patterns of LSM, ESB powder and mixtures of both powders

before and after heat-treatment. After annealing, the XRD pattern of LSM and ESB

mixture contains only the peaks of cubic fluorite structure from ESB and perovskite

structure from LSM without any other peaks. This result indicates no inter-phase

formation, suggesting that LSM and ESB are chemically stable even at 900 C. This

result of compatibility between LSM and ESB is in good agreement with previously

reported results [106, 108, 113]. In contrast, it has been known that the pervoskite Lai_

xSrxCo1_yFeyO3-6 (LSCF)-based cathode, which is widely used as the conventional

cathode for IT SOFCs, is highly reactive with ESB electrolyte due to weak metal-oxygen

bond of bismuth oxide [19, 77].

Fig. 7-2 shows a cross-sectional view of the symmetric cells with LSM-ESB

cathodes on ESB and GDC electrolyte pellets. In both cases, the composite cathodes

show good adhesion to electrolyte substrates at the cathode/electrolyte interface. In

order to distinguish each phase, backscattered SEM images were also observed, which

are shown as insets in Fig. 7-2. It is clearly shown that relatively large ESB particles are

well-distributed in a very fine LSM particle matrix. However, the majority in LSM-ESB

cathode seems to be LSM particles, even though we mixed 50: 50wt% of LSM to ESB

powders. It is noted that in this study the ESB powder in the composite cathode is large

(1~5 pm) and agglomerated due to solid state synthesis, while commercial LSM powder

is much smaller (< 1 pm) than ESB powder. As many studies reported, microstructural


136









optimization, such as particle size and spatial distribution can modify the connectivity

and increase the surface area of the composite cathodes to produce more active TPB

sites. Therefore, at this point, the performance of LSM-ESB composite cathode might

not be optimal.

Fig. 7-3 is the resultant impedance spectra measured under open circuit condition

from 500 to 700 C with 50 C interval for LSM-ESB/ESB/LSM-ESB and LSM-

ESB/GDC/LSM-ESB symmetric cells. For direct comparison of electrode ASR on the

different electrolytes, all of the impedance spectra have been ohmic resistance-

corrected. That is, the high frequency intercept at the real axis of each spectrum, which

corresponds to bulk electrolyte, electrode-sheet, and lead-contact resistance, has been

subtracted from each data point. The electrode ASRs of the LSM-ESB cathode on GDC

at 500, 550, 600, 650, and 700 C were 10.56, 3.31, 1.11, 0.44, and 0.19 -cm2,

respectively. This result is very comparable with the ASRs of the LSM-YSB cathode on

Sm-doped ceria (SDC) electrolytes reported by Li et al. In their study, the electrode

ASRs were ~ 1.08 and 0.15 Q-cm2 at 600 and 700 C, respectively. This suggests that

the performance of the LSM-ESB cathode in this study is reasonable. It should be

noted, however, that despite the fact that ESB has higher ionic conductivity than YSB

[114], the ASR of LSM-ESB on GDC in the present study is slightly higher than for LSM-

YSB on SDC. This might be due to microstructural features associated with infiltrated

nano-scale YSB particles into LSM backbones, while in this study, LSM-ESB cathode

was fabricated by a simple mechanical mixing method with micron-sized ESB powders.

As mentioned above, further improvements in electrochemical performance of this LSM-

ESB system can be expected by microstructural optimization. On the other hand, the


137









electrode ASRs of the LSM-ESB on ESB substrate were substantially lower--4.18, 1.29,

0.43, 0.19, and 0.08 at 500, 550, 600, 650, and 700 C, respectively.

In Fig. 7-4, the electrode ASR values from the impedance spectra in Fig. 7-3 are

plotted with various temperatures (right Y-axis). In addition, the percent electrode ASR

reduction on ESB relative to that on GDC was calculated at each temperature (left Y-

axis in Fig. 7-4), and a ~ 60% reduction in ASR was retained for all testing

temperatures. This result indicates that the reduced interfacial resistance of LSM-ESB

cathode on the ESB electrolyte remained in effect throughout the IT range. We believe

that this dramatic electrode ASR drop using the ESB electrolyte is caused by the high

oxygen dissociation rate of the bismuth oxide and the continuous fast ionic conduction

pathways from the cathode to the electrolyte without an abrupt change at the LSM-ESB

cathode and ESB electrolyte interface.

As illustrated in Fig. 7-5, the measured cathodic resistances of LSM-ESB on both

ESB and GDC electrolyte are compared to that of LSM-GDC and LSM-bismuth oxide

composite cathodes on various electrolytes in recent literature studies. As expected, all

LSM-bismuth oxide cathodes including the results in this study showed lower electrode

polarization losses than those of LSM-GDC cathodes which were reported by Murray et

al [90]. This result indicates that bismuth oxide phase integrated into LSM based-

cathodes increases the effective TPB length due to its higher ionic conduction.

However, even though doped ceria such as GDC and SDC has higher ionic conductivity

than YSZ (or ScSZ) electrolytes at IT, the reported LSM-bismuth oxide cathodes (and

the LSM-ESB cathode in this study) on GDC or SDC show a similar level of electrode

ASRs compared to cathodes using zirconia-based electrolytes, while LSM-GDC


138









cathodes have shown lower ASR on GDC electrolytes compared to YSZ electrolytes. It

is considered that due to the superior ionic conductivity of ESB or YSB in the composite

cathode, the effect of the GDC electrolyte at the cathode/electrolyte interface against

the YSZ electrolyte can be negligible. Instead, the performance of the LSM-bismuth

oxide cathode on GDC or YSZ seems to depend on bismuth oxide phase fraction and

cathode microstructures as shown in previous studies.

On the other hand, it is clearly shown that the LSM-ESB cathode on an ESB

electrolyte exhibits significantly lower electrode polarization losses than that of any other

LSM-based cathode in all measured temperature ranges from 500 to 700 C. This

outstanding result can be explained that the synergetic effect of the ESB phases both in

cathode bulk and at the electrolyte/cathode interface on enhancing oxygen dissociation

rate and accelerating ionic conduction leading to a dramatic reduction of activation

polarization for oxygen reduction in the cathode. Previously we reported a similar effect

of the ESB electrolyte on reducing electrode ASR for BRO7-ESB composite cathode on

ESB, showing 26% reduction of the ASR relative to that on GDC electrolyte [19].

Therefore, this study again demonstrates that the ESB electrolyte has a highly

significant effect on boosting the SOFC cathode reaction rate at the IT range.

Meanwhile, the activation energy of LSM-ESB cathodes on GDC and ESB was

estimated from ASR plots in Fig. 7-5 using an Arrhenius relationship. The calculated

activation energies for both cells were the same, ~1.24 eV, which is in good agreement

with low end of other reported values for LSM-bismuth oxide cathodes (1.23 ~ 1.5 eV),

which depend on microstructure and composition of the composite cathodes [104, 106,

108, 112]. In addition, it is been reported that the activation energy for the pure LSM on


139









SDC electrolyte is ~ 1.5 eV [106]. This result shows that the addition of ESB phase into

the cathode bulk or at the electrode/electrolyte interface might change the mechanism

of the oxygen reduction at cathode. In addition, this estimated activation energy implies

that the surface oxygen exchange reaction of which activation energy is known to ~ 1.25

eV at low temperature might be a rate-limiting step for the cathode reaction, providing a

good reason for use of ESB with high oxygen surface exchange rate to reduce the

electrode polarization [45].

In order to gauge the long-term stability of the LSM-ESB cathode, the electrode

ASR of the LSM-ESB cathode on an ESB electrolyte pellet was measured at 700 OC.

Fig. 7-6 is the resultant plot of the electrode ASR at 7000C for 100 hrs. The ASR of the

LSM-ESB cathode maintained a constant value for 100 hrs of 0.08 0.001 Q-cm2

indicating no initial degradation in electrode performance. To verify the stability of the

LSM-ESB system for IT SOFCs, however, further long term testing in the lower

temperature ranges and under various applied current conditions should be carried out.

7.3.2 I-V Characterization for Button Cells

To further investigate the effect of LSM-ESB cathodes coupled to an ESB

electrolyte on the actual IT-SOFC performance, current-voltage measurements were

conducted on anode-supported button cells. In order to obtain an ESB electrolyte, an

ESB-GDC bilayered electrolyte cell design was utilized due to the thermodynamic

instability of the ESB electrolyte at low Po2 conditions [16, 18]. Further details regarding

the ESB-GDC bilayered electrolyte are available in the recent studies [19]. In this study

two cells were fabricated, which included LSM-ESB (cathode) / GDC (electrolyte) /Ni-

GDC and LSM-ESB/ESB-GDC/Ni-GDC.


140









Fig.7-7 shows SEM images of the two cells after testing. As shown in the cross-

sectional views (Fig. 7-7-a,b), the structure of two cells are identical except a thin ( 3

pm) ESB electrolyte is observed in ESB/GDC bilayer cell between the LSM-ESB

cathode and GDC electrolyte (Fig. 7-7-b). Fig. 7-7-c,d show the surface views of the

GDC and ESB electrolyte for the GDC single electrolyte cell and GDC/ESB bilayered

electrolyte cell, respectively, and some cracks and pores can be observed. This

indicates that the densities of both electrolytes were not optimal. This issue will be

discussed later in this section.

The current-voltage measurement results of both cells at 6500C are plotted in Fig.

7-8-a. The maximum power densities (MPDs) obtained were 658 and 836 mW/cm2 for

LSM-ESB on GDC and LSM-ESB on ESB/GDC, respectively. Even though the OCP of

the cell with LSM-ESB on ESB/GDC is slightly lower than that of LSM-ESB on GDC, the

MPD of LSM-ESB on ESB/GDC was increased by 27% compared to that of GDC

single layer cell. The increased power density is due to the significant reduction in the

total ASR calculated from the IV curves in Fig. 7-8-a, which were reduced by ~ 54%

from 0.566 )-cm2 for cell-1 to 0.263 )-cm2 for LSM-ESB on ESB/GDC. In this study we

used the same fabrication procedure and confirmed identical microstructures for the two

cells with the exception of the ESB interlayer in ESB/GDC bilayer structure, as shown in

Fig. 7-7-a,b. Thus, we believe that MPD enhancement of the LSM-ESB on ESB/GDC

cell mostly came from the effect of ESB electrolyte on reducing the electrode cathodicc)

ASR at the LSM-ESB cathode/ ESB electrolyte interface.

For further analysis, two points probe impedance tests were conducted. Fig. 7-8-b

illustrates the resultant nyquist plots for each cell under open circuit condition with the


141









same gas condition for I-V test at 650 C. From the high frequency intercepts (RH) and

low frequency intercepts (RL) at the real axis of the complex plane, the ohmic and total

ASR values were estimated, respectively. The electrode ASR including anode and

cathode polarization resistance was calculated using the equation written as;

ASR,,e,lro = R, RH (7-1)

The detailed values are tabulated in Table 7-1. In this study, the total ASR values

obtained using DC measurement from IV (ASRiv) and electrochemical impedance

(ASREIS) showed less than 1% deviation indicating high reliability of the resultant data.

As expected from symmetric cell measurements, the electrode ASR was significantly

reduced for the bilayered electrolyte cell (0.164 Q-cm2), showing ~ 64% reduction

compared to single layer cell (0.450 Q-cm2). This result is quite consistent with the

cathodic ASR change for the symmetric cell measurement in the previous section.

These I-V and impedance results demonstrate that the effect of the ESB electrolyte on

LSM-ESB cathode polarization losses directly influences the SOFCs performance at IT,

showing reasonably high maximum power density of over 850 mW/cm2 at 650 C. At

this point it is noted that the GDC electrolyte densities in the testing cells were not high

and have many small pores and the ESB layer also showed poor density affected by

GDC layer as shown in Fig. 7-7. It is well known that the density and thickness of the

electrolyte strongly influence ohmic resistance and SOFC performance [14]. Therefore,

the MPD of the LSM-ESB cathode can be much higher if denser and thinner GDC and

ESB electrolytes are used.

Fig. 7-9 shows IV-curves of both cells at various temperature ranging from 450 to

650 C. The MPDs of the cell with LSM-ESB on GDC were 25, 72, 182, 389, and 658


142









mW/cm2 at 450, 500, 550, 600, and 650 C, respectively. In case of the bilayered cell,

higher MPDs were exhibited and are reported to be 50, 124, 275, 533, and 836 mW/cm2

at 450, 500, 550, 600, and 650 C, respectively. The improvement in MPD from the cell

with LSM-ESB on GDC to the cell with LSM-ESB on ESB at each temperature was

calculated and plotted with actual MPD values for each cell in Fig. 7-10. It is noted that

in order to give better visualization of the MPDs at low temperature, a log scale was

used for the power density (left axis in Fig. 7-10). In this plot, it is clearly shown that the

enhancement in MPD linearly increases as temperature decreases. For instance, the

MPD of ESB/GDC bilayered cell compared to that of GDC single layer increased by less

than 30% at 650 C but 51 and 100% at 550 and 450 C, respectively. This indicates

that the effect of the ESB electrolyte on LSM-ESB performance is valid and even

greater at low temperature. It is believed that the portion of cathodic polarization losses

is larger at lower temperature due to its thermally activated nature [7]. Therefore, the

beneficial effect of the ESB electrolyte layer on cathode performance should be

emphasized on using LSM-ESB cathodes for low temperature SOFC application.

Fig. 7-11 shows a comparison of MPDs for SOFCs with various LSM-bismuth

oxide cathodes at low to intermediate temperatures. For all temperatures, the LSM-ESB

cathode coupled to ESB electrolyte shows highest MPD. To my knowledge, the MPDs

(LSM-ESB on ESB/GDC electrolyte) in this study are the highest for any SOFCs using

LSM-bismuth oxide composite cathodes reported to date. Furthermore, this LSM-ESB

cathode on an ESB electrolyte can be expected to produce much higher power density

at intermediate and even lower temperatures through microstructural tailoring, including

cathode structure as well as electrolyte density and thickness control.


143









7.4 Conclusions

Conventional LSM cathodes for high temperature SOFCs were prepared for low

temperature SOFC application by pairing it with ESB in a composite cathode. Due to the

inherent high conductivity and fast oxygen exchange rate of ESB, the LSM-ESB

composite cathode showed much lower cathodic polarization losses than any other

LSM-based cathode at the IT range. Moreover, when ESB was used as an electrolyte,

the electrode ASR was reduced further (by ~60%) compared to that of the same LSM-

ESB cathode on a GDC electrolyte. Using the LSM-ESB cathode on a ESB-GDC

bilayered electrolyte, the MPD produced at 650 C was ~ 865 mW/cm2, which is the

highest reported MPD to date for SOFCs using LSM-bismuth oxide cathodes. This

study demonstrated that the performance of LSM-ESB cathodes can be effectively

boosted when using ESB electrolytes. This is a very promising finding for development

of SOFCs which operate at low to intermediate temperatures.


144













at LSM+ESB before annealing




> I 20ESB


SIIPure LSM



20 30 40 50 60 70 80
2 0 (degree)

Figure 7-1. XRD pattern of LSM, ESB, and LSM+ESB (50:50wt%) powers before and
after annealing at 900 OC 50 hrs


145



















































Figure 7-2. SEM images of LSM-ESB cathode on GDC electrolyte (a) and ESB
electrolyte (b). The insets are backscattered images.


146


,-j-~ wanwal-* *^r-**W















1



5



5


1.

E1.



N
0.(
O.4

E 0.
0.





0.2(

E 0.1

0.1,

S0.0(


-40.1
(N
S0.0

0.0(


0o.o

0.0(


5

0
5

0


3
3

4)
2
0


0.00 0.05 0.10 ,0.15
Z (Q-cmn)


0.20


Figure 7-3. Impedance spectra of the LSM-ESB cathode on ESB and
the temperature ranges from 500 to 700 C


GDC pellets at


147


(a) 5000C 0oo0000000000
0
-.. LSM-ESB on GDCO

LSM-ESB on ESBEI

0 2 4 6 8 10 12


(b) 5500C

00000000000 00


0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

S(c) 6000C
3 ,00000000000




0.0 0.2 0.4 0.6 0.8 1.0 1.2
13


(d) 6500C
000000000


0000


(e) 7000C

0000000000000
00000 0


0.25


-


-

-




-
-
-
-


-











10 --


E
0
d

ry


(D

Iii


50
500


550
550


I6
600


I6
650


700
700


' I


Temperature (oC)

Figure 7-4. Electrode ASRs of LSM-ESB cathode on GDC (blue circles) and ESB (red
squares) electrolytes, and ASR reduction rate (black stars).


148


I I


* *


'0...
''
'
'
''


"...:............::::::::.:::::::


I I


I I


-100

-90

-80

-70

-60

-50 (D
-5
-40

-30 (D

-20

-10

-0










.*


'' A *


....< :;,,,i :.*
... /
:: .... "


o


.V7
A'


10-


N E




C/)


0
-1
a)
LUi


.I I
0.95 1.00


.05
1.05


.-**-- LSM-GDC(50wt%) on YSZ [a]
---A-- LSM-GDC (50wt%) on GDC[b]
-<- LSM-20ESB (15wt%) on YSZ[c]
-- LBSM-30ESB(50wt%) on ScSZ[d]
---- LSM-30ESB (50wt%)on ScSZ[e]
--*-- LSM-YSB(50wt%) on YSZ [f]
---- LSM-YSB(50wt%)on SDC [g]
---- LSM-YSB(77wt%) on YSZ[h]
---- LSM-20ESB(50wt%) on GDC, this study
---- LSM-20ESB(50wt%) on ESB, this study


I I I I I1 .
1.10 1.15 1.20 1.25 1.30


1000/T (K')

Figure 7-5. Comparison of the electrode polarization resistance of LSM-bismuth oxide
cathodes at IT ranges. LBSM is short for Lao.74Bio.1oSro.16MnO3-.6
Ref: a,b-[90], c-[111], d-[104], e-[108], f-[109], g-[106], h-[112]


149


x' ~3*
A'~ ~

V K a
AII
K
K

U''" .~














cI
E




CM
0






o
0)
0)
5,
,4U


Time (hours)


Figure 7-6. Long term stability test of
for 100 hours.


LSM-ESB cathode on ESB electrolyte at 700 C


150


0.10


0.08



0.06



0.04



0.02


----------- ------.....-----......... ...---- ..------...
















0 20 40 60 80 100


0.00
















































Figure 7-7. SEM images of cross-sectional views for cell-1(a) and cell-2(b), and surface
views for cell-1 (c) and cell-2 (d)


151













































Figure 7-7. Continued


152


WRr
' ^^ 2s a ^ S










0.8

-0.7 -
0.8-
-0.6

S0.6- -0.5 0
S-0.4 0.
c-.
0.4 -

0.2- 0.2 0
-0.1
02- Cell-1 (LSM-ESB on GDC) -01 3

0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Current Density (A/cm2)
(a)
0.20
5 65 C Cell-1
0.15- 650 o0 0 0 /
10- Cell-2 b
K 0.05-
0.00 ,
0.0 0.1 0.2 0.3 0.4 0.5 0.6

Z' (Q-cm2)
(b)
Figure 7-8. I-V characteristics (a) and impedance spectroscopy (b) of cell-1 and cell-2
at 650 C


153























0.2


0.0
(






1.0-


0.8-


Current Density (A/cm2)
(a)


-u./ -"
0
-0.6

-0.5

-0.4

-0.3

-0.2

-0.1 ,

-0.0
0




0.9

0.8

0.7 O
0
0.6

0.5

0.4 ,

0.3

0.2

0.1

0.0


Current Density (A/cm2)
(b)

Figure 7-9. I-V characteristics at various temperature for cell-1 (a) and cell-2 (b)


154








1000-
:.. -100
-.

*E 80-u


E -60
100 ,

S. -40


o' -20 2 o


10 I 0
450 500 550 600 650

Temperature (C)
Figure 7-10. Maximum power density improvement (black stars) of cell-2 (red squares)
at various temperatures compared to cell-1 (blue circles)


155







1100
LSM-20ESB(50wt%) on ESB, this study
1000- LSM-20ESB(50wt%) on GDC, this study
90 A LSM-YSB(50wt%) on SDC [a]
900 < LSM-YSB(75wt%) on YSZ [b]
800- v LSM-30ESB(50wt%) on YSZ [c]
E O LBSM-30ESB(50wt%) on ScSZ [d]
o 700- > LSM-YSB(50wt%) on YSZ [e]
>
600 -
S500-
Q 400-
2 300- *
200-
100 -

450 500 550 600 650 700

Temperature (OC)

Figure 7-11. Comparison of maximum power density of SOFCs using LSM-bismuth
oxide composite cathodes at IT ranges.
Ref: a-[107], b-[106], c-[112], d-[108], e-[111], f-[109]


156









Table 7-1. Detailed total, ohmic, and electrode ASR values for Cell-1 and Cell-2 at 650
oC


Cathode Electrolyte


Total
ASRiv


Total
ASREIS


Ohmic
ASR


(unit: O-cm2)
Electrode
ASR


Cell-1 LSM-ESB GDC 0.566 0.569 0.119 0.450
Cell-2 LSM-ESB ESB-GDC 0.263 0.261 0.097 0.164


157









CHAPTER 8
CONCLUSIONS

In this dissertation, several works have been conducted to reduce polarization

losses for the anode, electrolyte and cathode at reduced temperatures. In all, the goal

was to develop high performance solid oxide fuel cells running at low to intermediate

temperatures.

A novel AFL was developed to dramatically improve SOFC performance (Chapter

3). The bimodally integrated nano-/micron- composite AFL was fabricated by simply

spray coating a precursor solution into a conventional submicron Ni-GDC functional

layer. The composition used for the anode substrate, the colloidal AFL (C-AFL), and

the precursor- integrated AFL (N+C-AFL) was Ni- GDC. The electrolyte composition

was GDC, a system that has received much attention for its potential for use in IT-

SOFCs. A systematic study comparing cells using no AFL, a C-AFL, and the newly-

developed N+C-AFL was conducted. It showed that the N+C-AFL sample exhibited a

maximum power density of 1160 mW/cm2 at 600 oC-a 287% increase compared to the

sample with no AFL and a 70% increase compared to the sample using a conventional

AFL. Both ohmic and non-ohmic losses were lowered, suggesting that the 2D interfacial

region between the anode and the electrolyte was enhanced as well as TBP lengths.

Additionally, it was shown that the fractional improvement in power increases with

decreasing temperature, a critical point for reduced operating temperatures. These

findings are very encouraging not only based on dramatic improvement in performance,

but also in the simplicity of the technique itself. It is further believed that this N+C-AFL

technique is versatile enough to be applied to other SOFC systems for similar gains in

performance, and will be a milestone in the field of reduced temperature SOFCs.


158









In addition to the effect of particle size in AFL, the effect of AFL composition on the

electrochemical performance was investigated using submicronsized Ni-GDC AFL

(Chapter 4). For this, AFLs with composition ranging from 40 to 80wt% NiO were

fabricated. Microstructural analysis confirmed these functional layers to have phases

which are fine and well-distributed. The optimal AFL composition was achieved at 1:1

volume ratio of Ni to GDC, which corresponds to 60 wt% NiO. This composition

exhibited the highest MPD over the intermediate to low temperature range. In addition, a

preliminary long-term stability test showed the possibility of using this system in

practical SOFC applications. The measured MPD and ASR show an inverse linear

relationship implying that the performance enhancement greatly depends on AFL

composition.

To better understand the relationship between microstructure and electrochemical

performance, the effect composition on the microstructures of Ni-GDC AFLs was

investigated (Chapter 5). AFLs with various Ni-GDC compositions (50 80 wt% NiO

before reduction) were quantified by a three dimensional (3D) reconstruction technique

using a FIB/SEM dual beam system. Each AFL sample was automatically sectioned into

150 slices with 60 nm intervals. The Amira software package allowed for alignment,

segmentation, and reconstruction of the 2D images to a 3D image. From these

reconstructions, the volume fraction, effective particle size, phase gradient, and surface

area for Ni, GDC, and pore phases as well as pore tortuosity for each sample were

quantified. The estimated phase volume fraction was well matched to the theoretically

calculated value. The optimal effective particle size and phase fraction were seen at 60

wt% NiO. The active TPB densities were calculated based on the connectivity of voxels


159









labeled with each phase. The highest TPB density achieved was ~ 15.6 pm-2 and

corresponded to 60wt% NiO The TPB density showed an inverse proportionality to

electrode ASR. This result implies that the quantified microstructural values, which are

controllable, can be directly applied to predict the electrochemical performance of

SOFCs

In Chapter 6, a discussion was given on the fabrication of a thin and dense bilayer

electrolyte consisting of erbium stabilized bismuth oxide (ESB) and gadolinium-doped

ceria (GDC) applied on a tape-cast anode-supported SOFC using a practical and cost-

effective colloidal deposition process. Using a wet chemical co-precipitation method,

nano-sized ESB particles were successfully synthesized at temperatures as low as ~

500 C, which is much lower than those needed for powders prepared by the

conventional solid state route (~ 800 C). Due to the high sinterability of this powder, a

dense erbia stabilized bismuth oxide (ESB) layer was successfully formed on a

gadolinia doped ceria (GDC) electrolyte by a simple colloidal coating method.. A

systematic study on the sintering behavior of ESB revealed that at higher sintering

temperatures, bismuth oxide can sublime or penetrate into the GDC sublayer. SEM and

EDX analysis of a full button cell with an ESB/GDC bilayer sintered at 800 C showed

no visible interfacial diffusion between each layer. I-V measurement of the cell showed

high power density ( ~ 1.5 W/cm2) at 650 C due to an enhancement in OCP and a

significant reduction in ASR when compared to a GDC single cell. This result

demonstrates that this ESB/GDC bilayer electrolyte is practical for high performance

SOFCs at low operational temperature.


160









In Chapter 7, an alternative high performance composite cathode for low to

intermediate temperature SOFCs was described. Using a highly conductive ESB phase,

the performance of conventional (Lao.80oSr.20)MnO3-6 (LSM) cathodes was dramatically

improved. The ESB phase was utilized not only as the ion-conducting phase in the

LSM-ESB composite cathode, but also as an electrolyte coupled to LSM-ESB cathode.

The electrode ASR measured from a symmetric cell consisting LSM-ESB electrodes on

an ESB electrolyte was only 0.08 Q-cm2 at 7000C, which is -60% lower than that of

LSM-ESB on GDC electrolytes (0.19 Q-cm2). This exemplifies the synergetic effect the

ESB phase has both in the cathode bulk and at the electrolyte/electrode interface. The

ESB phase is presumed to increase the effective TPB length as well as enhance the

oxygen surface exchange reaction. This effect was shown to occur at all temperatures

tested, from 500 to 700 C. The MPDs of the anode-supported SOFCs with LSM-ESB

cathodes on ESB/GDC bilayered electrolytes were 50, 124, 275, 533, and 836 mW/cm2

at 450, 500, 550, 600, and 650 C, respectively. These are to date the highest reported

MPDs for SOFCs using LSM-bismuth oxide composite cathodes and demonstrate that

the LSM-ESB composite cathode boosted by an ESB electrolyte is very promising for

low-to-intermediate SOFC applications.


161









APPENDIX A
DEPENDENCE OF OCP ON GDC ELECTROLYTE THICKNESS

In this dissertation, the GDC electrolyte has been used for all works. Contrast to

the conventional YSZ electrolyte with pure ionic conductivity, SOFCs with doped ceria

electrolyte have shown lower OCP from theoretical Nernst voltage due to electronic

leakage current and oxygen permeation of GDC [60]. It has been also reported that

these leakage current and oxygen permeation properties are function of the electrolyte

thickness [59, 60]. In this appendix, the dependence of the OCP on GDC electrolyte

thickness on Ni-GDC anode-supported SOFCs was briefly studied.

In order to gauge the effect of GDC electrolyte thickness on OCP of SOFCs in the

IT range, eight anode-supported SOFCs with different electrolyte thicknesses were

fabricated. The NiO-GDC anode supports were made by a tape-casting process. For

the better electrolyte deposition, submicronsized Ni-GDC AFL (50wt% NiO) was

colloidally deposited. Next, GDC electrolyte was coated on the AFL surface by a spin

coating method. Thickness of the electrolyte was controlled by the number of repeated

coating process in the fixed spin speed and time condition. Detailed fabrication process

of SOFCs with thin GDC electrolyte on the anode-support was described on Chapter 4.

Fig. A-1 shows the SEM images of the cross-sectional microstructures of the

tested samples. From these micrographs, it was observed that different GDC electrolyte

thicknesses of the prepared samples were obtained using the control of spin coating

process. The measured electrolyte thicknesses were varied from 6.2 to 32.9 pm. The

detailed values of electrolyte thickness measured were tabulated in Table A-1.

The OCP was measured under 90 sccm of hydrogen with 3% water as a fuel at

the anode side and 30 sccm of dry air as an oxidant at the cathode side in the


162









temperature range from 500 to 650 C. The resultant OCP values at 500, 600, and 650

C for each sample with electrolyte thickness were plotted in Fig. A-2, and summarized

in Table A-1. At all temperatures, the OCP was decreased as electrolyte thickness

decreased. These experimental results demonstrate the theory that the electronic

leakage current and oxygen permeation in GDC electrolyte is greater at thinner

electrolyte thickness [59, 60].

Recently this experiment result was verified by continuum-level electrochemical

model developed Duncan and Wachsman [60]. Fig. A-3 shows the resultant modeling fit

with author's experimental data by Duncan. For this modeling, the OCP model equation

is given as

1 z(u u)c, uc
oc th ioc K In + I (nU (A-1)
q \v ,L Z ee Z ( v),L u e

Detailed explanation and derivation of this model equation are in the previous work

by Duncan et. al [60].


163








































i.


~fi-F-WW4-~ 15~,


Figure A-1. Microstructures of Ni-GDC anode/GDC electrolyte/LSCF cathode SOFCs
with various electrolyte thicknesses.


164








1.00

0.95-

"l 0.90 A H
I
a0.85 o


S0.80-

c 0.75
Q. I 6500C
O 0.70- 6000C

A 5000C
0.65-
I i I i I i I i I i I i I
5 10 15 20 25 30 35
Electrolyte thickness (10GDC, pm)
Figure A-2. Experimental OCP values from electrochemical test at 500, 600, and 650
C as a function of GDC electrolyte thickness


165














0.9 .' / 650 C -





O 0.8 600 OC
O .


I'?

0.7 ,'
'I


0 10 20 30 40 50 60

Electrolyte thickness (10GDC, rnm)

Figure A-3. Fit of the OCP model (eq. A-1) to experimental data for OCP as a function
of electrolyte thickness.


166









Table A-1. Summary of sample description and OCP result
S # GDC Electrolyte OCP OCP OCP
Thickness(pm) at 650(oC) at 600(oC) at 500(C)
1 6.2 0.648 0.664
2 9.0 0.749 0.838
3 9.4 0.790
4 11.3 0.801 0.847 0.890
5 12.6 0.822 0.855 0.913
6 19.7 0.843 0.860 0.920
7 26.3 0.876 0.914 0.961
8 32.9 0.895 0.935 0.990


167









APPENDIX B
LONG TERM STABILITY FOR A SOFC WITH NI-GDC AFL

In this section, the result of the long term stability test of a SOFC with 60wt% NiO

in Ni-GDC AFL is described. This result is extended long term stability test data from

chapter 4. To see the effect of AFL on the SOFC performance of the function of time, a

potentiostatic test was conducted for 600 hrs at 650 C. As a testing condition, a voltage

of 0.379V where cell reached 98% of their MPD at initial I-V test was applied. The test

was carried out under the gas condition of the 90 sccm of H2 with 3% of H20 as a fuel

and 90 sccm of dry air as an oxidant at anode and cathode sides, respectively. Figure

B-1 shows the result of a 600 hours long term stability test for the 60 wt% NiO AFL cell.

Up to 200 hours, the effect of the AFL was retained with high power density of ~1.1

W/cm2. However, over 200 hours, the power density decreased with time. The

degradation rate was quite linear with time and measured at the time period from 250 to

600 hours with a linear fitting method. The estimated degradation rate was ~1.03

mW/cm2/hour.

To further investigate, current-voltage curves were measured before and after long

term test at the various intermediate temperature ranges. In figure B-2, the I-V

characteristics before and after long term tests compared in the temperature ranges

from 500 to 650 C with 50 C interval. The maximum power densities before long term

testing were 1.11, 0.63, 0.27, 0.12 W/cm2 at 650, 600, 550, and 500 C, respectively.

However, after the potentiostatic test for 600 hours the MPDs measured were

decreased as 0.78, 0.47, 0.21, 0.08 W/cm2 at 650, 600, 550, and 500 C, respectively. It

is clearly shown that the degradation of power densities mainly came from lower OCP

after long term test. It might be explained as the decreasing effective thickness of GDC


168









electrolyte due to gradual reduction of Ce4+ into Ce3+ under low Po2 condition, causing

increase of electronic current leakage through the electrolyte. In addition to leakage

current issue, this degradation is possibly explained that this performance degradation

phenomenon with time might come from the changing of microstructures of the anode

and cathode with time under applied current. For further investigation, electrochemical

impedance spectroscopy can be analyzed and the microstructural analysis of the long

term tested sample with SEM or 3D reconstruction using FIB/SEM dual beam system

will be required.


169






































0 59 100 150 2 290 30 3 BD s 430 40 9M0 5B 61D 6
Time (hours)

Figure B-1. Long term stability test of a SOFC with 60wt% NiO in Ni-GDC AFL. The
potentiostatic test was conducted at 650 C for 600 hours under an applied
voltage of 0.379 V.


170





























I (Amps/cm)
(a)


0 1 2 3
I (Amos/cm)
(b)


Figure B-2. Comparison of I-V plots of the testing sample between before long term test
and after long term test for 600 hours at 6500C(a), 6000C(b), 5500C(c),
500C(d).


171















0.75 After -
0.2






0.1
0.25



0 1 0
0 0.5 1.0 1.5 2.0
I (Amps/cm)
(C)

1.00 110.15
5000C

Before
0.75

After -0.10



U 00

0.05
0.25 -




0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
I /Amncwrr
(d)

Figure B-2. Continued


172










APPENDIX C
EXPERIMENTAL SETUP


0-1%
vJN
CD cO

Stu

:r
^I
4a-a "~


Cu




ra
w


I
.1-1
LU
0


N
2-%
r1
0


-0/ 1 -

o o o --

0 U %,



< C
Figure C-1. Schematic SOFC testing setup a button cell testing setup configuration
and I-V and EIS testing equipment
m m-m
E E .





Figure C-i. Schematic SOFC testing setup a button cell testing setup configuration
and I-V and EIS testing equipment


173


U

mo


- I
ar
E 0





m '.
r


-0
0

LJ
ro








-r
v_
oa
















































Figure C-2. Illustration of symmetric cell configuration for EIS test (top) and EIS testing
setup (bottom)










174









LIST OF REFERENCES

1. W. R. Grove, Philosophical Magazine and Journal of Science 14 (1839), p. 127.

2. N.Q. Minh, Journal of the American Ceramic Society 76 (1993) (3), p. 563.

3. S.M. Haile, Acta Materialia 51 (2003) (19), p. 5981.

4. E.D. Wachsman and S.C. Singhal, Electrochemical society interface 18 (2009)
(3), p. 38.

5. B.C.H. Steele, Journal of Materials Science 36 (2001) (5), p. 1053.

6. J.W. Fergus, Journal of Power Sources 162 (2006) (1), p. 30.

7. B.C.H. Steele, Solid State lonics 129 (2000) (1-4), p. 95.

8. S. deSouza, S.J. Visco and L.C. DeJonghe, Solid State lonics 98 (1997) (1-2), p.
57.

9. B.C.H. Steele and A. Heinzel, Nature 414 (2001) (6861), p. 345.

10. R.E. Williford, L.A. Chick, G.D. Maupin, S.P. Simner and J.W. Stevenson,
Journal of the Electrochemical Society 150 (2003) (8), p. A1067.

11. N. Ai, Z. Lu, K.F. Chen, X.Q. Huang, X.B. Du and W.H. Su, Journal of Power
Sources 171 (2007), p. 489.

12. N. Ai, Z. Lu, J.K. Tang, K.F. Chen, X.Q. Huang and W.H. Su, Journal of Power
Sources 185 (2008) (1), p. 153.

13. K.F. Chen, X.J. Chen, Z. Lu, N. Ai, X.Q. Huang and W.H. Su, Electrochimica
Acta 53 (2008) (27), p. 7825.

14. J.S. Ahn, H. Yoon, K.T. Lee, M.A. Camaratta and E.D. Wachsman, Fuel Cells 9
(2009) (5), p. 643.

15. B.C.H. Steele, Solid State lonics 75 (1995), p. 157.

16. T. Takahashi, T. Esaka and H. Iwahara, Journal of Applied Electrochemistry 7
(1977) (4), p. 299.

17. N.X. Jiang, E.D. Wachsman and S.H. Jung, Solid State lonics 150 (2002) (3-4),
p. 347.

18. E.D. Wachsman, P. Jayaweera, N. Jiang, D.M. Lowe and B.G. Pound, Journal of
the Electrochemical Society 144 (1997) (1), p. 233.


175









19. J.S. Ahn, M.A. Camaratta, D. Pergolesi, K.T. Lee, H. Yoon, B.W. Lee, D.W.
Jung, E. Traversa and E.D. Wachsman, Journal of the Electrochemical Society
157 (2010) (3), p. B376.

20. J.S. Ahn, D. Pergolesi, M.A. Camaratta, H. Yoon, B.W. Lee, K.T. Lee, D.W.
Jung, E. Traversa and E.D. Wachsman, Electrochemistry Communications 11
(2009) (7), p. 1504.

21. A. Hammouche, E.J.L. Schouler and M. Henault, Solid State lonics 28 (1988), p.
1205.

22. N.Q. Minh, Chemtech 21 (1991) (2), p. 120.

23. R.T. Dehoff, Thermodynamics in Materials Science, McGraw-Hill (1993).

24. P.J. Gellings and H.J.M. Bouwmeester, The CRC Handbook of Solid State
Electrochemistry, CRC press (1997).

25. J. Larminie and A. Dicks, Fuel Cell System Explanied, Wiley (2003).

26. M. Yashima, M. Kakihana and M. Yoshimura, Solid State lonics 86-8 (1996), p.
1131.

27. J.C. Boivin and G. Mairesse, Chemistry of Materials 10 (1998) (10), p. 2870.

28. D.W. Strickler and W.G. Carlson, Journal of the American Ceramic Society 47
(1964) (3), p. 122.

29. E.C. Subbarao and H.S. Maiti, Solid State lonics 11 (1984) (4), p. 317.

30. H. Yahiro, Y. Eguchi, K. Eguchi and H. Arai, Journal of Applied Electrochemistry
18 (1988) (4), p. 527.

31. K. Eguchi, T. Setoguchi, T. Inoue and H. Arai, Solid State lonics 52 (1992) (1-3),
p. 165.

32. H.A. Harwig, Zeitschrift Fur Anorganische Und Allgemeine Chemie 444 (1978)
(SEP), p. 151.

33. T. Takahashi, T. Esaka and H. Iwahara, Journal of Solid State Chemistry 16
(1976) (3-4), p. 317.

34. M.J. Verkerk, K. Keizer and A.J. Burggraaf, Journal of Applied Electrochemistry
10 (1980) (1), p. 81.

35. C.Z. Wang, X.G. Xu and B.Z. Li, Solid State lonics 13 (1984) (2), p. 135.

36. P. Duran, J.R. Jurado, C. Moure, N. Valverde and B.C.H. Steele, Materials
Chemistry and Physics 18 (1987) (3), p. 287.


176









37. E.D. Wachsman, G.R. Ball, N. Jiang and D.A. Stevenson, Solid State lonics 52
(1992) (1-3), p. 213.

38. H. Yahiro, Y. Baba, K. Eguchi and H. Arai, Journal of the Electrochemical Society
135 (1988) (8), p. 2077.

39. J.Y. Park and E.D. Wachsman, lonics 12 (2006) (1), p. 15.

40. J.Y. Park, H. Yoon and E.D. Wachsman, Journal of the American Ceramic
Society 88 (2005) (9), p. 2402.

41. Y.J. Leng and S.H. Chan, Electrochemical and Solid State Letters 9 (2006) (2), p.
A56.

42. E.D. Wachsman, Solid State lonics 152 (2002), p. 657.

43. T.H. Etsell and S.N. Flengas, Chemical Reviews 70 (1970) (3), p. 339.

44. M. Godickemeier, K. Sasaki, L.J. Gauckler and I. Riess, Solid State lonics 86-8
(1996), p. 691.

45. Z.P. Shao and S.M. Haile, Nature 431 (2004) (7005), p. 170.

46. C.W. Sun and U. Stimming, Journal of Power Sources 171 (2007) (2), p. 247.

47. E. Ramirez-Cabrera, A. Atkinson and D. Chadwick, Solid State lonics 136
(2000), p. 825.

48. T. Suzuki, Z. Hasan, Y. Funahashi, T. Yamaguchi, Y. Fujishiro and M. Awano,
Science 325 (2009) (5942), p. 852.

49. M.F. Liu, D.H. Dong, R.R. Peng, J.F. Gao, J. Diwu, X.Q. Liu and G.Y. Meng,
Journal of Power Sources 180 (2008) (1), p. 215.

50. H. Koide, Y. Someya, T. Yoshida and T. Maruyama, Solid State lonics 132
(2000) (3-4), p. 253.

51. J. Will, A. Mitterdorfer, C. Kleinlogel, D. Perednis and L.J. Gauckler, Solid State
lonics 131 (2000) (1-2), p. 79.

52. J.W. Kim, A.V. Virkar, K.Z. Fung, K. Mehta and S.C. Singhal, Journal of the
Electrochemical Society 146 (1999) (1), p. 69.

53. J.J. Haslam, A.Q. Pham, B.W. Chung, J.F. DiCarlo and R.S. Glass, Journal of
the American Ceramic Society 88 (2005) (3), p. 513.

54. D. Stover, H.P. Buchkremer and S. Uhlenbruck, Ceramics International 30 (2004)
(7), p. 1107.


177









55. S.D. Kim, S.H. Hyun, J. Moon, J.H. Kim and R.H. Song, Journal of Power
Sources 139 (2005) (1-2), p. 67.

56. E. Wanzenberg, F. Tietz, P. Panjan and D. Stover, Solid State lonics 159 (2003)
(1-2), p. 1.

57. K.T. Lee, H.S. Yoon, M.A. Camarratta, N.A. Sexson, J.S. Ahn and E.D.
Wachsman, to be submitted.

58. M.D. Gross, J.M. Vohs and R.J. Gorte, Journal of the Electrochemical Society
154 (2007) (7), p. B694.

59. X. Zhang, M. Robertson, C. Deces-Petit, W. Qu, O. Kesler, R. Maric and D.
Ghosh, Journal of Power Sources 164 (2007) (2), p. 668.

60. K.L. Duncan and E.D. Wachsman, Journal of the Electrochemical Society 156
(2009) (9), p. B1030.

61. L.C.R. Schneider, C.L. Martin, Y. Bultel, D. Bouvard and E. Siebert,
Electrochimica Acta 52 (2006) (1), p. 314.

62. J.R. Wilson and S.A. Barnett, Electrochemical and Solid State Letters 11 (2008)
(10), p. B181.

63. N.P. Brandon, S. Skinner and B.C.H. Steele, Annual Review of Materials
Research 33 (2003), p. 183.

64. K.T. Lee, N.J. Vito, C.A. Mattehw, H.S. Yoon and E.D. Wachsman, ECS
transactions to be submitted (2010).

65. N. Shikazono, Y. Sakamoto, Y. Yamaguchi and N. Kasagi, Journal of Power
Sources 193 (2009) (2), p. 530.

66. J.R. Wilson, W. Kobsiriphat, R. Mendoza, H.Y. Chen, J.M. Hiller, D.J. Miller, K.
Thornton, P.W. Voorhees, S.B. Adler and S.A. Barnett, Nature Materials 5 (2006)
(7), p. 541.

67. D. Gostovic, J.R. Smith, D.P. Kundinger, K.S. Jones and E.D. Wachsman,
Electrochemical and Solid State Letters 10 (2007), p. B214.

68. J.R. Smith, A. Chen, D. Gostovic, D. Hickey, D. Kundinger, K.L. Duncan, R.T.
DeHoff, K.S. Jones and E.D. Wachsman, Solid State lonics 180 (2009) (1), p. 90.

69. J.R. Wilson, A.T. Duong, M. Gameiro, H.Y. Chen, K. Thornton, D.R. Mumm and
S.A. Barnett, Electrochemistry Communications 11 (2009) (5), p. 1052.

70. J.R. Wilson, M. Gameiro, K. Mischaikow, W. Kalies, P.W. Voorhees and S.A.
Barnett, Microscopy and Microanalysis 15 (2009) (1), p. 71.


178









71. J.R. Wilson, J.S. Cronin, A.T. Duong, S. Rukes, H.Y. Chen, K. Thornton, D.R.
Mumm and S. Barnett, Journal of Power Sources 195 (2010) (7), p. 1829.

72. N. Shikazono, D. Kanno, K. Matsuzaki, H. Teshima, S. Sumino and N. Kasagi,
Journal of the Electrochemical Society 157 (2010) (5), p. B665.

73. A. Bieberle, L.P. Meier and L.J. Gauckler, Journal of the Electrochemical Society
148 (2001) (6), p. A646.

74. H. Inaba and H. Tagawa, Solid State lonics 83 (1996) (1-2), p. 1.

75. V.V. Kharton, E.N. Naumovich and V.V. Samokhval, Solid State lonics 99 (1997)
(3-4), p. 269.

76. A. Jaiswal, C.T. Hu and E.D. Wachsman, Journal of the Electrochemical Society
154 (2007) (10), p. B1088.

77. M. Camaratta and E. Wachsman, Journal of the Electrochemical Society 155
(2008) (2), p. B135.

78. A.L. Patterson, Physical Review 56 (1939) (10), p. 978.

79. A.M. Azad, S. Larose and S.A. Akbar, Journal of Materials Science 29 (1994)
(16), p. 4135.

80. N.X. Jiang and E.D. Wachsman, Journal of the American Ceramic Society 82
(1999) (11), p. 3057.

81. M. Hrovat, A. Bencan, J. Holc, T. Rojac and M. Kosec, Journal of Materials
Research 18 (2003) (6), p. 1297.

82. V. Gil, J. Tartaj, C. Moure and P. Duran, Ceramics International 33 (2007) (3), p.
471.

83. D.W. Jung, K.L. Duncan and E.D. Wachsman, Acta Materialia 58 (2010) (2), p.
355.

84. E.P. Murray, T. Tsai and S.A. Barnett, Solid State lonics 110 (1998) (3-4), p. 235.

85. S.P. Yoon, J. Han, S.W. Nam, T.H. Lim, I.H. Oh, S.A. Hong, Y.S. Yoo and H.C.
Lim, Journal of Power Sources 106 (2002) (1-2), p. 160.

86. S.P. Jiang, Journal of Power Sources 124 (2003) (2), p. 390.

87. C.C. Kan, H.H. Kan, F.M. Van Assche, E.N. Armstrong and E.D. Wachsman,
Journal of the Electrochemical Society 155 (2008) (10), p. B985.

88. C.C. Kan and E.D. Wachsman, Journal of the Electrochemical Society 156
(2009) (6), p. B695.


179









89. T. Kenjo, S. Osawa and K. Fujikawa, Journal of the Electrochemical Society 138
(1991) (2), p. 349.

90. E.P. Murray and S.A. Barnett, Solid State lonics 143 (2001) (3-4), p. 265.

91. C.W. Tanner, K.Z. Fung and A.V. Virkar, Journal of the Electrochemical Society
144 (1997) (1), p. 21.

92. M. Juhl, S. Primdahl, C. Manon and M. Mogensen, Journal of Power Sources 61
(1996) (1-2), p. 173.

93. Y.J. Leng, S.H. Chan, K.A. Khor and S.P. Jiang, Journal of Solid State
Electrochemistry 10 (2006) (6), p. 339.

94. J.L. Li, S.R. Wang, Z.R. Wang, R.Z. Liu, X.F. Ye, X.F. Sun, T.L. Wen and Z.Y.
Wen, Journal of Power Sources 188 (2009) (2), p. 453.

95. N.M. Sammes, G.A. Tompsett, H. Nafe and F. Aldinger, Journal of the European
Ceramic Society 19 (1999) (10), p. 1801.

96. B.A. Boukamp, I.C. Vinke, K.J. Devries and A.J. Burggraaf, Solid State lonics 32-
3 (1989), p. 918.

97. B.A. Boukamp, Solid State lonics 136 (2000), p. 75.

98. J.C. Boivin, C. Pirovano, G. Nowogrocki, G. Mairesse, P. Labrune and G.
Lagrange, Solid State lonics 113 (1998), p. 639.

99. I.C. Vinke, K. Seshan, B.A. Boukamp, K.J. Devries and A.J. Burggraaf, Solid
State lonics 34 (1989) (4), p. 235.

100. M. Dumelie, G. Nowogrocki and J.C. Boivin, Solid State lonics 28 (1988), p. 524.

101. C.R. Xia, Y. Zhang and M.L. Liu, Applied Physics Letters 82 (2003) (6), p. 901.

102. M. Camaratta and E. Wachsman, Solid State lonics 178 (2007), p. 1411.

103. M. Camaratta and E. Wachsman, Solid State lonics 178 (2007) (19-20), p. 1242.

104. J.L. Li, S.R. Wang, Z.R. Wang, R.Z. Liu, T.L. Wen and Z.Y. Wen, Journal of
Power Sources 179 (2008) (2), p. 474.

105. J.L. Li, S.R. Wang, X.F. Sun, R.Z. Liu, X.F. Ye and Z.Y. Wen, Journal of Power
Sources 185 (2008) (2), p. 649.

106. Z.Y. Jiang, L. Zhang, K. Feng and C.R. Xia, Journal of Power Sources 185
(2008) (1), p. 40.

107. J. Li, S. Wang, R. Liu, T. Wen and Z. Wen, Fuel Cells 9 (2009) (5), p. 657.


180









108. J.L. Li, S.R. Wang, Z.R. Wang, R.Z. Liu, T.L. Wen and Z.Y. Wen, Journal of
Power Sources 194 (2009) (2), p. 625.

109. Z.Y. Jiang, L. Zhang, L.L. Cai and C.R. Xia, Electrochimica Acta 54 (2009) (11),
p. 3059.

110. Z.Y. Jiang, C.R. Xia, F. Zhao and F.L. Chen, Electrochemical and Solid State
Letters 12 (2009) (6), p. B91.

111. J.L. Li, S.R. Wang, Z.R. Wang, J.Q. Qian, R.Z. Liu, T.L. Wen and Z.Y. Wen,
Journal of Solid State Electrochemistry 14 (2010) (4), p. 579.

112. Z. Jiang, Z. Lei, B. Ding, c. Xia, F. Zhao and F. Chen, International Journal of
Hydrogen Energy in press (2010).

113. Q.S. Zhang, A. Hirano, N. Imanishi, Y. Takeda and K. Yamahara, Journal of Fuel
Cell Science and Technology 6 (2009) (1).

114. P. Shuk, H.D. Wiemhofer, U. Guth, W. Gopel and M. Greenblatt, Solid State
lonics 89 (1996) (3-4), p. 179.


181









BIOGRAPHICAL SKETCH

Kang Taek Lee was born in Seoul Korea in 1976. After graduating from Saehwa

High School in Seoul, Korea, He started his campus life at Yonsei University in Korea

with Ceramic Engineering major. During his studies, he was enlisted in the Republic of

Korea Army for a general soldier in the department of biological and chemical weapon.

After that, he came back to the Yonsei University and finished his course work and

received his Bachelor of Science degree, in August 2002. His enthusiasm for materials

science drove him to enter the graduate school at Korea Institute of Advanced Science

and Engineering (KAIST) in Korea. After 2 years, he received the master's degree and

worked at LG electronics for one and half years. In 2006, he moved to the U.S. and

entered University of Florida at the department of materials science and engineering.

He joined in Dr. Wachsman's group, and finally received his Ph.D in August of 2010.


182





PAGE 1

1 C OMPREHENSIVE D EVELOPMENT OF HIGH P ERFORMANCE SOLID OXIDE FUEL CELLS FOR INTERMEDIATE AND LOW TEMPERATURE APPLICATIONS By KANG TAEK LEE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTI AL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Kang Taek Lee

PAGE 3

3 To my mom and wife

PAGE 4

4 ACKNOWLEDGMENTS First of all, I glorify God (and Jesus Christ), who is my savior, for this disser tation. During last four year journey I could not complete anything without His guidance and protection. I strongly wish that this dissertation and all my works in the future would be a confession of my faith i n Him. For this dissertation, I wrote six exp erimental notes. On the first pages for all these notes, it is written that Do not be deceived. God cannot be mocked. A man reaps what he sows. (Galatians 6:7). Relying on these words, I always did try to do my best for all works. Moreover, I should say that this work would not have been possible were it not for the support of many people. I would like to thank my advisor, Prof. Eric D. Wachsman, for his support and guidance. His encouragement helped me to reach a higher level of success and expand my pot ential. I also would like to thank Prof. Juan C. Nino, Prof. Simon Phillpot, Prof. Wolfgang Sigmund, Prof. Mark Orazem and Prof. Valentin Craciun for their advice, guidance and constructive comments. I also thank Dr. Heesung Yoon who taught me most of the processes of SOFC fabrication. I also wish to acknowledge other former and current group members; Dr. Keith Duncan, Dr. Takkeun Oh, Dr. Sean Bishop, Dr. Dongjo Oh, ByungWook Lee, Eric Armstrong, Dr. Bryan Black Burn, Eric Ma cam and other members for provi ding me with an excellent research environment and helpful comments. I would especially like to thank Dr. Dohwon Jung, Dr. Matthew Camaratta, Dr. Jin Soo Ahn and Nick Vito for their sincere friendship and coworking partnership. I truly want to give my th anks to Kwanjeong Educational Foundation for unconditional financial support during my entire doctoral research with a very honorable scholarship.

PAGE 5

5 It was also truly helpful and enjoyable for me to share the time with all my Timothy group friends at the Ko rean Baptist Church of Gainesville. All worship services and activities were great encouragement and support for me. I would like to thank my family members Most of all, I want to thank my wifes parents Dae Heun Kang and Jung Suk Kim for entrusting their precious and only daughter to me and welcoming me into their family. I am also thoroughly grateful to my father, Sam Soo Lee who is a great supporter of mine and who prayed for me through this process I also strongly pray to God for restoring his healt h. At this moment, I strongly thank and miss my brother, Kang Yong Lee who is my mentor and supporter with his endless trust and encouragement I d like to also mention his wife, Ji Eun Kim and their precious daughter for their support Finally, I dedicat e this dissertation to my mom and wife. My mother Yang Soon Kim has dedicated herself to me for over 30 years. She always trusted and encouraged me in any circumstance. My beautiful and lovely wife, Yoo Jin Kang moved to here, leaving her family and friend s to be with me. She was by my side during my struggles with endless patience and love. I love you so much! Thank you.

PAGE 6

6 TABLE OF CONTENTS ACKNOWLEDGMENTS .................................................................................................. 4 page LIST OF TABLES ............................................................................................................ 9 LIST OF FIGURES ........................................................................................................ 10 LIST OF ABBREVIATIONS ........................................................................................... 14 ABSTRACT ................................................................................................................... 16 CHAPTER 1 INTRODUCTION .................................................................................................... 19 2 BACKGROUND ...................................................................................................... 23 2.1 Basic Principle of SOFC Operation ................................................................... 23 2.2 Actual SOFCs Operation .................................................................................. 24 2.2.1 Opencircuit P otential (OCP) and T ransference N umber (ti) ................... 25 2.2.2 Irreversible Losses .................................................................................. 27 2.2.2.1 Activation p olarization l osses ......................................................... 27 2.2.2.2 Leakage current p olarization l osses ............................................... 28 2.2.2.3 Ohmic polarization ......................................................................... 29 2.2.2.4 Concentration polarization ............................................................. 29 2.3. Materials and Design ....................................................................................... 30 2.3.1 Stabilized Zirconia Electrolytes ................................................................ 30 2.3.2 Aliovalent Cations Doped Ceria Electrolytes ........................................... 31 2.3.3 Stabilized Bismuth Oxide Electrolytes ..................................................... 32 2.3.4 Bilayered E lectro lyte C oncept for H igh P erformance IT SOFCs ............. 34 2.3.4.1 Ceria / Zirconia bilayer electrolyte .................................................. 34 2.3.4.2 Ceria / Bismuth oxide bila yer electrolyte ........................................ 35 3 INTERGRATING NANO AND MICRO STRUCTURED ANODE FUNCTIONAL LAYERS FOR IMPROVED IT SOFC PERFORMANCE ......................................... 44 3.1 Introduction ....................................................................................................... 44 3.2 Experimental ..................................................................................................... 47 3.3 Result and Discussion ...................................................................................... 49 3 .4 Conclusions ...................................................................................................... 54 4 EFFECT OF NI GDC AFL COMPOSITION ON PERFORMANCE OF IT SOFCS 62 4.1 Introduction ....................................................................................................... 62 4.2 Experimental ..................................................................................................... 64

PAGE 7

7 4.2.1 Cell F abrication ....................................................................................... 64 4.2.3 Characterization ...................................................................................... 65 4.3 Results and Discussion ..................................................................................... 66 4.3.1 Microstructural Analysis ........................................................................... 66 4.3.2 Effect of A FL Composition on Power Density .......................................... 67 4.3.2.1 I V characteristics at 650 C ........................................................... 67 4.3.2.2 Temperature d ependence .............................................................. 70 4.3.2.3 Long t erm stability .......................................................................... 71 4.3.2.4 Effect of AFL composition on ASR ................................................. 71 4.4 Conclusions ...................................................................................................... 72 5 COMPREHENSIVE QUANTIFICATION OF NIO GDC ANODE FUNCTIONAL LAYER MICROSTRUCTURE BY THREEDIMENSIONAL RECONSTRUCTION USING FIB/SEM ..................................................................................................... 84 5.1 Introduction ....................................................................................................... 84 5.2 Experimental ..................................................................................................... 85 5.3 Results and Discussion ..................................................................................... 87 5.4 Conclusions ...................................................................................................... 93 6 HIGH PERFORMANCE IT SOFC WITH CERIA/BISMUTH OXIDE BILAYERED ELECTROLYTES FABRICATED BY A SIMPLE COLLOIDAL ROUTE USING NANO SIZED ESB POWDER .............................................................................. 107 6.1 Introduction ..................................................................................................... 107 6.2 Experimental Procedure ................................................................................. 110 6. 2.1 ESB Powder Fabrication ....................................................................... 110 6. 2.2 Fuel C ell F abrication .............................................................................. 111 6. 2. 3 Characterization .................................................................................... 112 6.3 Result and Discussion .................................................................................... 113 6.3.1 Powder C haracterization ....................................................................... 113 6 .3.2 Effect of S intering T emperature on ESB/GDC B ilayered E lectrolyte ..... 114 6.3. 3 Microstructure of a F ull Button Cell with ESB/GDC B ilayered E lectrolyte ................................................................................................... 117 6. 3.4 Performance of a B utton C ell with ESB/GDC Bilayered E lectrolyte ...... 118 6.4 Conclusions .................................................................................................... 121 7 HIGH PERFORMANCE LSM BASED CATHODE BOOSTED BY STABILIZE D BISMUTH OXIDE FOR LOW TO INTERMEDIATE TEMPERATURE SOFCS ..... 131 7.1 Introduction ..................................................................................................... 131 7.2 Experimental ................................................................................................... 133 7.2.1 Sample F abrication ................................................................................ 133 7.2.2 Characterization .................................................................................... 135 7.3 Result and Discussion .................................................................................... 136 7.3.1 Impedance S pectroscopy for S ymmetric C ells ...................................... 136 7.3.2 I V C haracterization for B utton C ells ..................................................... 140

PAGE 8

8 7.4 Conclusions .................................................................................................... 144 8 CONCLUSIONS ................................................................................................... 158 APPENDIX A DEPENDENCE OF OCP ON GDC ELECTROLYTE THICKNE SS ....................... 162 B LONG TERM STABILITY FOR A SOFC WITH NI GDC AFL ............................... 168 C EXPERIMENTAL SETUP ..................................................................................... 173 LIST OF REFERENCES ............................................................................................. 175 BIOGRAPHICAL SKETCH .......................................................................................... 182

PAGE 9

9 LIST OF TABLES Table page 2 1 Calculated Po2 and Nernst voltage at opencircuit condition (T=500~700oC) ..... 43 2 2 Conductivity Data for Stabilized ZrO2 Doped with RareEarth Oxides ................ 43 3 1 Detailed OCP, MPD and ASR values of the fuel cell samples with N+C AFL, C AFL, no AFL at 600oC. .................................................................................... 61 4 1 Detailed ASR values of the testing cells with var ious NiO contents in AFL ........ 83 5 1 3D reconstruction dimension and total volume fractions of Ni, GDC and pore phase and solid volume fractions of Ni and GDC ............................................. 106 5 2 Summary of quantification of microstructural features of AFL with various compositions ..................................................................................................... 106 6 1 Calcination condition and crystallite size of ESB powders synthesis by co precipitation and solid state route. .................................................................... 130 6 2 Comparison of specification and electrochemical performance of the studied cells. ................................................................................................................. 130 7 1 Detailed total, ohmic, and electrode ASR values for Cell 1 and Cell 2 at 650 oC ..................................................................................................................... 157 A 1 Summary of sample description and OCP result .............................................. 167

PAGE 10

10 LI ST OF FIGURES Figure page 2 1 S chematic diagram of reactions in SOFCs based on oxygenion conductors .... 37 2 2 SOFC curre nt voltage behavior indicating relative polarization losses ............... 38 2 3 Variation of ionic conductivity of stabilized ZrO2 with dopant concentration (T=807oC) ........................................................................................................... 39 2 4 Conductivities of selected electrolyte materials .................................................. 40 2 5 ESB conductivity versus Po2 in purified argon atmosphere ............................... 41 2 6 Conceptual representation of a bilayer electrolyte showing the effect of relative thickness on interfacial oxygen partial pressure (Po2 ) ........................... 41 2 7 Bulk electrolyte ASR a t 500 oC as a function of relative (t = LESB/LSDC) and total thickness for bilayers .................................................................................. 4 2 3 1 Schematic illustration of the proposed N+C AFL structure on anodesupported SOFC and effect of N+C AF L on expending TPB length. Yellow triangles represent TPBs in conventional AFL (C AFL) and red triangles. ......... 55 3 2 SEM micrographs of the anode surface after deposition and presintering (a, c, e) and after full sintering followed by simulated testing atmospheric conditions (b, d, f) for samples with no AFL (a, b), C AFL (c, d) ....................... 56 3 3 Comparison of I V characteristics for the fuel cell samples with N+C AFL, CAFL, and no AFL at 600 C. (a) I V plots at the temperature ranging from 650 to 500 oC for N+C AFL (b), C AFL (c), and no AFL (d). ..................................... 57 3 4 Electrochemical impedance spectra of the testing samples with N+C AFL, C AFL and no AFL at various temperature; 650 C (a), 600 C (b), 550 C (c), and 500 C (d). ................................................................................................... 59 3 5 MPD (a) and ASR plots (b) for the different samples tested between 500 and 650 C. ............................................................................................................... 60 4 1 Backscattered images showing a cross sectional view of anodesupported SOFCs with different NiO content in the anode functional layers; no AFL(a), 4 0wt% (b), 50wt%(c), 60wt%(d), 6 5 wt%(e), and 80wt%(f) NiO. ......................... 74 4 2 Magnified microstructures of the anode or AFLs with different NiO content no AFL(a), 40wt% (b), 50wt%(c), 60wt%(d), 65wt%(e), a nd 80wt%(f) NiO. Backscattering mode provides better contrast to distinguish Ni (dark gray) ...... 75

PAGE 11

11 4 3 I V plots of fuel cells with various AFL compositions at 650oC; 40( ), 50( ), 60( ), 65( ) and 80( )wt% of NiO in AFL, and no AFL( ) The gas condition was 90sccm of air and 3% of wet hydrogen on the anode ................ 76 4 4 Open circuit potential of the fuel cells wit h various NiO contents in NiO GDC AFL. Solid line (red) shows linear fit of the measured data (square) .................. 77 4 5 MPD (Red square) and total ASRIV estimated from IV curves (blue star) are plotted with NiO contents in AFL. The open symbols represent no AFL cell. ..... 78 4 6 Maximum power densities of fuel cells with various AFL compositions at the temperature range from 450 to 650 C. Open s ymbols represent MPD of no AFL cell at each temperature. ............................................................................. 79 4 7 Long term stability test of fuel cell with 60wt% of NiO in the AFL and the no AFL cell for 200 hrs at 650 C. Potentiostatic tes ts were conducted with an applied voltage of 0.379 V for the NiO 60wt% AFL cell and 0.380 V ................ 80 4 8 Impedance spectra with various AFL compositions (a), and total, electrode, and ohmic ASRs of fuel cells with different NiO content (b) calculated from impedance spectra (a). Open symbols represent no AFL results. ...................... 81 4 9 MPD plots with electrode ASR shows a linear relationship Red line is linear fitting of the measured data (black dots). ............................................................ 82 5 1 Schematic diagram of FIB/SEM dual beam system with sample (a) and 3D reconstruction process (b) .................................................................................. 95 5 2 3D reconstruction of Ni GDC anode (a), and AFLs with initial composition of 50 (b), 60 (c), 65 (d), and 80 (e)wt% NiO nearby at anode(or AFL)/electrolyte interface. ............................................................................................................. 96 5 3 Individually reconstructed phases from the 3D reconstruction of AFL with 65 wt% NiO ; GDC (a), Ni (b), Pore (c), and combination of Ni and Pore phases ... 97 5 4 Phase gradient of reconstructed samples with no AFL (a), 50 (b), 60 (c), 65 (d), and 80 (e)wt% NiO in Ni GDC AFL .............................................................. 98 5 5 Volume fraction of Ni, GDC and pore phase in total volume (a), and volume fraction of Ni and GDC in solid volume of AFLs with various compositions. Open symbols represent theoretical values. ..................................................... 101 5 6 Effective particle diameters of Ni (rectangular), GDC (circle), and pore (trian gle) phase of AFLs with various compositions .......................................... 102

PAGE 12

12 5 7 Schematic diagram of TPB length calculation from 3D reconstruction. A rectangular parallelepiped represents a voxel in a 3D reconstruction and each one is labeled as one of phases; Ni, GDC, or Pore phase.. ..................... 103 5 8 Plot of quantified surface area and TPB density of AFL with various NiO contents. Dotted lines are only for guide purpose. ............................................ 104 5 9 (a) TPB density and electrode ASR with various AFL compositions (Dotted lines are only for guide purpose.) (b) plot of 1 over electrode ASR with TPB density. A red line represents linear fit for the plot showing inverse. ................ 105 6 1 XRD diffraction pattern of ESB powders synthesized by coprecipitation route (red line) and solid state route (black line) (a). The magnified XRD diffraction pattern of the (111) peak is shown at the 2 range from 27 to 29o (b). ............ 123 6 2 SEM of ESB powders synthesized by wet chemical coprecipitation method (a) and solidstate route (b). ............................................................................. 124 6 3 Evolution of ESB layers on GDC electrolyte at various sintering temperatures using ss ESB (a,c,e,g) and cpESB (b,d,f,h). It is noted that the magnification of images for ESB ele ctrolyte using cpESB powder is higher. ........................ 125 6 4 Cross sectional view of GDC electrolyte under ss ESB layer after sintering at 900 oC, which is the magnified image from Fig. 53g. In backscattering mode, ESB (white), GDC (light gray), and NiO (dark gray) phases are well. ............... 126 6 5 Cross sectional SEM image of a full button cell with ESB/GDC bilayered electrolyte. EDX line scan was conducted along the straight base line (yellow) and the intensity of each elements are presented as red (Bi) ............. 126 6 6 SEM image of cross sectional view of a single GDC electrolyte cell (a) and ESB/GDC bilayered cell (b). Surface views of GDC electrolyte (c) and ESB electrolyte (d) are shown. ................................................................................. 127 6 7 IV Characteristics (a) and impedance spectra (b) of ESB/GDC bilayer (red square) and GD C single layer (blue triangle) cell ............................................. 129 7 1 XRD pattern of LSM, ESB, and LSM+ESB (50:50wt%) powers before and after annealing at 900 oC 50 hrs ....................................................................... 145 7 2 SEM images of LSM ESB cathode on GDC electrolyte (a) and ESB electrolyte (b). The insets are backscattered images. ...................................... 146 7 3 Impedance spectra of the LSM ESB cathode on ES B and GDC pellets at the temperature ranges from 500 to 700 oC ........................................................... 147 7 4 Electrode ASRs of LSM ESB cathode on GDC (blue circles) and ESB (red squares) electrolytes, and ASR reduction rate (black stars). ............................ 148

PAGE 13

13 7 5 Comparison of the electrode polarization resistance of LSM bismuth oxide cathodes at IT ranges. LBSM is short for La0.74Bi0.10Sr0.16MnO3. .................... 149 7 6 Long term stability test of LSM ESB cathode on ESB electrolyte at 700 oC for 100 hours. ........................................................................................................ 150 7 7 SEM images of cross sectional views for cell 1(a) and cell 2(b), and surface views for cell1 ( c) and cell 2 ( d ) ....................................................................... 151 7 8 I V characteristics (a) and impedance spectroscopy (b) of cell 1 and cell 2 at 650 oC ............................................................................................................... 153 7 9 I V characteristics at various temperature for cell 1 (a) and cell 2 (b) .............. 154 7 10 Maximum power density improvement (black stars) of cell 2 (red squares) at various temperatures compared to cell 1 (blue circles) .................................... 155 7 11 Comparison of maximum power density of SOFCs using LSM bismuth oxide composite cathodes at IT ranges ..................................................................... 156 A 1 Microstructures of Ni GDC anode/GDC electrolyte/LSCF cathode SOFCs with various electrolyte thicknesses. ................................................................. 164 A 2 Experimental OCP values from electrochemical test at 500, 600, and 650 oC as a function of GDC electrolyte thickness ....................................................... 165 A 3 Fit of the OCP model (eq. A 1) to experimental data for OCP as a function of electrolyte thickness. ........................................................................................ 166 B 1 Long term stability test of a SOFC with 60wt% NiO in Ni GDC AFL. The potentiostatic test was conducted at 650 oC for 600 hours under an applied voltage of 0.379 V. ............................................................................................ 170 B 2 Comparison of I V plots of the testing sample between before long term test and after long term test for 600 hours at 650oC(a), 600oC(b), 550oC(c), 500oC(d). .......................................................................................................... 171 C 1 Schematic SOFC testing setup a button cell testing setup configuration and I V and EIS testing equipment .......................................................................... 173 C 2 Illustration of symmetric cell configuration for EIS test (top) and EIS testing setup (bottom) .................................................................................................. 174

PAGE 14

14 LIST OF ABBREVIATIONS AFL Anode functional layer ASR Area specific resistance BET Brunauer Emmett Teller method BRO Bismuth ruthenate BSCF Barium strontium cobalt ferrite DBP Dibutyl phthalate EDX Energy dispersive X ray analysis EIS Electrochemical impedance spectroscopy ESB Erbium stabilized bismuth oxide FIB Focused ion beam FWHM Full width at half maximum GDC Gadolina doped ceria IT Intermediate temperature LMIS Liquid metal or ganic ion source LSCF Lanthanum strontium cobalt ferrite LSM Lanthanum strontium manganite MIEC Mixed ionic electronic conductor MPD Maximum power density OCP Open circuit potential PLD Pulsed laser deposition PVB Polyvinyl butyral ROI Region of interest S cSZ Scandia stabilized zirconia SDC Samaria doped ceria

PAGE 15

15 SEM Scanning electron microscope SOFC Solid oxide fuel cell TLD Throughlens detector XRD X ray diffraction YSB Yttrium stabilized bismuth oxide YSZ Yttrium stabilized zirconia

PAGE 16

16 Abstract of Dissertati on Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy C OMPREHENSIVE D EVELOPMENT OF HIGH P ERFORMANCE SOLID OXIDE FUEL CELLS FOR INTERMEDIATE AND LO W TEMPERATURE APPLICATIONS By Kang Taek Lee August 2010 Chair: Eric. D. Wachsman Major: Materials Science and Engineering In order to develop high performance solid oxide fuel cells (SOFCs) operating at low to intermediate temperatures the three main SOFC components --the anode, electrolyte, and cathode--were comprehensively studied. In order to lower anodic polarization losses in anodesupported SOFCs, a novel composite anode functional layer (AFL) having bimodal ( nano/micro) structure was developed. Application of this AFL involved a simple process where a precursor solution was coated onto a conventional submicron sized colloidally deposited Ni GDC AFL. Cells prepared in this manner yielded maximum power densities (MPD) of 1.29, 1.16, 0.7 and 0.38 W/ cm2 at 650, 600, 550 and 500 oC, respectively. Electrochemical impedance results showed a striking decrease in both ohmic and nonohmic area specific resistances (ASRs) for these cells compared to those with either no AFL, or a conventional AFL. In additi on, the effect composition of the conventional submironsized AFL on performance was examined. The highest MPD ( 1.15 W/cm2 at 650 C ) was achieved at a composition of 60wt% NiO. This composition had the best performance over the intermediate temperature range ( 450 to 650 C ) For the potentiostatic test, t he cell

PAGE 17

17 exhibited stable performance over 200 hrs of operation at 1.1 W/cm2. It was also revealed that electrode ASR has an inverse linear relationship with maximum power density at 650oC. To better under stand the effect of AFL composition, m icrostructural features of AFLs having various Ni GDC compositions were quantified by a 3 dimensional (3D) reconstruction technique using a FIB/SEM dual beam system. Of the compositions tested, t he highest triple phase boundary ( TPB) density was achieved at 60wt% NiO which corresponds to a 1:1 volume ratio of Ni to GDC phase. The quantified TPB density showed an inverse proportionality to electrode ASR. Using a wet chemical co precipitation method, nanosized ESB particles were successfully synthesized at temperatures as low as ~ 500 oC. Due to the high sinterability of this powder, a dense erbia stabilized bismuth oxide (ESB) layer was successfully formed on a gadolinia doped ceria (GDC) electrolyte by a simple colloidal coating method. A systematic study on the sintering behavior of ESB was conducted to determine the optimum sintering conditions for these materials IV measurement a cell using this bilayered electrolyte sytem showed a high power density ( ~ 1.5 W/cm2) at 650 oC due to an enhancement in OCP and a significant reduction in ASR when compared to a GDC single cell. T he performance of conventional (La0.80Sr0.20)MnO3(LSM) cathodes were dramatically improved at the IT range by combining it with a highly conductive ESB phase. The electrode ASR measured from a symmetric cell consisting of LSM ESB cm2 at 700oC which is ~60% lower than that of LSM cm2). This exemplifies the

PAGE 18

18 synerget ic effect the ESB phase has both in the cathode bulk and at the electrolyte/electrode interface. The MPDs of the anodesupported SOFCs with LSM ESB cathodes on ESB/GDC bilayered electrolytes were ~ 836 mW/cm2 at 650 oC which is the highest value reported f or SOFCs using LSM bismuth oxide composite cathodes

PAGE 19

19 CHAPTER 1 INTRODUCTION A f uel cell is an energy conversion device which directly produces electrical energy from the chemical energy contained in various fuels by electrochemical reactions. In 1838, the basic principle of the fuel cell was written in one of the scientific magazines of the time by German chemist, Christian Friedrich Schnbein. One month later, Sir William Grove reported the first functional fuel cell in 1839. He used a dilute sulfuric aci d solution as an electrolyte at room temperature, which produced water and electricity [1] However the history of SOFCs began much later in 1899, with the discovery of the solid oxide electrolyte by Nernst and followed with the first SOFC invented by Baur and Peris in 1937 [2] Since that time, and especially in the last several decades, tremendous effort and progress has been made to commercialize SOFCs. For instance, Siemens Westinghouse has successfully developed and operated a 100 k W system for over 20,000 h without significant deterioration in performance [3] R ecently the Solid State Energy Conversion Alliance (SECA) the fuel cell program under United States Department of energy (DOE) announced their road map, including the development of a prototype SOFC stack with megawatt capability and fuel flexiblity by 2015 [4] O ne of the biggest challenges to SOFC commercialization is to reduce the operation temperature while maintaining high power densities. At intermediate temperatures (IT, 500~700 oC), the system cost can be significantly reduced by allo wing the use of cheap stainless steel for the bipolar plates and the balanceof plant, as well as the use of high temperature gaskets rather than rigid glass based seals, which can also enhance mechanical stability and life time [5]

PAGE 20

20 Conventional SOFCs with yttria stabilized zirconia ( YSZ) electrolyte s operate at high temperatures ( over ~1000oC ) due to its thermally activated ionic conduction, and thus have unacceptable system cost and slow start up times [4 7] T wo main strategies have been studied to reduce ohmic losses in the electrolyte at reduced temperatures F irst is the thin electrolyte approach--the electrolyte resistance is inversely proportional to the electrolyte thickness [8] To accommodate thin electrolyte films anode supported cell s has been developed [5, 9] In this conf iguration however, anodic polarization can limit performance due to the relatively high anode thickness. I t is believed that most fuel oxidation reactions take place at near the anode/electrolyte interface, indicating that most anodic losses occur in this region [10] Therefore, engineering of the interfacial region has received much attention as a way to reduce losses at the anode [1114] The other approach is to use materials with enhanced ionic conductivity. For example, erbia stabilized bismuth oxide (ESB) and gadolinia doped ceria (GDC) have one to two orders of magnitude higher ionic conductivity in the IT range than YSZ [15] However, these two materials have disadvantages including thermodynamic instability at the low Po2 conditions experienced at the anode side of fuel cell system s [3, 16, 17] To overcome these limitations a bismuth oxide/ ceria bilayer electrolyte concept has been proposed [18] In order to produce high power densities in the low to intermediate temperature range, one can combine the bilayer electrolyte concept with a thin film approach. Using this concept with a thin and dense ESB (~ bilayered electrolyte, the author and colleagues recently demonstrated a cel l having an exceptionally high power density of ~ 2 W/cm2 at 650oC [19, 20] In that study the

PAGE 21

21 dense ESB layer was deposited by pulsed laser deposition (PLD), which is not a viable technique for mass production. Therefore, a more simple and cost effective fabrication process is necessary. The focus of this dissertation is the development of SOFCs producing high power densities in the IT range and prepared with practical and cost effective fabrication processes. Each component of the SOFC -the anode, electrolyte, and cathode--was investigated to reduce its major polarization losses. In order to control anodic polarization losses, a novel AFL was developed at the anode/electrolyte interface by integrating nanoand mi cronparticle structures. For further improvement of the NiGDC AFL, the effect of composition was carried out. Microstructural features of the AFLs were quantified using a stateof theart 3D reconstruction technique by a FIB/SEM dual beam system. From t his work, the understanding of the relationship between electrochemical performance and microstructures was enhanced. In order to improve electrolyte performance, the ESB/GDC bilayered electrolyte system was investigated. C ost effective fabrication of dens e ESB electrolytes was achieved by a simple colloidal deposition technique. In order to accomplish this nanosized ESB particles with high sinterability were synthesized by a wet chemical coprecipitation method. The r eproducibility of the high performanc e exhibited by these bilayered electrolyte cells was carefully demonstrated. In addition, an (La0.8Sr0.2)0.9MnO3( LSM ) ESB composite cathode was studied as an alternative cathode for low to intermediate temperatures. The use of conventional LSM cathodes has been limited to high temperature SOFCs due to its low ionic conduction at reduced temperatures [21] In this work LSM was mixed with the fast ion

PAGE 22

22 conductor, ESB. The performance of the LSM ESB cathode was investigated in the IT range. This work demonstrated that coupled to an ESB electrolyte, the performance of LSM ESB was stable and significantly better than that of the same cathode on conventional GDC or YSZ electrolytes below 650 oC.

PAGE 23

23 CHAPTER 2 BACKGROUND 2.1 Basic Principle of SOFC O peration In principle, overall SOFC reaction is expressed as a simple reaction formula; O H H O2 2 2 2 1 (2 1) In order to complete this reaction in the actual fuel cell operation, the reaction is divided into two half cell reacti ons [2, 3, 9] Fig. 2 1 shows a schematic diagram of a general SOFC structure and half cell reactions at anode and cathode side [22] I n the cathode side, oxygen which generally comes from air is reduced to O2with electrons provided from outside fuel cell which written as; 2 22 2 1 O e O (2 2) For this half reaction, a cathode conducts adsorption of oxygen molecules and dissociation of adsorbed oxygen. This is followed by formation of oxygen ion by electron transfer and charge transfer (O2and electron) at the triple phase boundar ies (TPBs) between gas, ionic, and electronic conducting phases. Therefore, a cathode should be a high catalysis to dissociate molecules and have high ionic and electronic conduction with good compatibility to the electrolyte as well. Related work with high catalytic cathode for IT SOFC has be done at ch. 7 in this dissertation Transferred Oxygen ions move to anode via electrolyte, for which high ionic conductivity of the electrolyte is necessary. A driving force of the ion migration is the Nernst potential due to PO 2 difference between cathode and anode. Therefore, the electrolyte should be a good barrier between the air side and the fuel side to maintain the low PO2 at the anode, which can be achieved by highly dense electrolyte. Highly

PAGE 24

24 dense electrolyte with a special design for high stability and conductivity was researched at ch. 6. At the anode, the fuel is oxidized with the migrated oxygen ions. e O H O H 22 2 2 (2 3) e CO O CO 22 2 (2 4) e n O H n nCO O n H Cn n) 2 6 ( ) 1 ( ) 1 3 (2 2 2 2 2 (2 5) Each equation presents the oxidation reaction for different fuels; eq. (23) for H2 eq. (24) for CO, and eq. (25) for hydr ocarbon fuel. In this case, the catalytic properties of anode is important, which takes place at TPBs. Therefore, concentration and spatial distribution of TPB can be a key factor to improve anode performance for SOFCs. In this dissertation, control and mechanism of TPB extension in anode nearby the anode and electrolyte interface was intensively studied (ch. 3 ~ 5 ). 2.2 Actual SOFCs O peration In actual SOFCs operation condition, the performance of SOFCs is commonly measured by voltage out as a function of applied current density. Fig 2 2 shows a representative voltagecurrent plot of SOFC [4] As mentioned above, the Nernst potential by P o2 difference between anode and cathode produces a driving force to operate SOFC s. However, this ideal voltage can not be maintained under applied current due to various irreversible polarization mechanisms. The actual operational cell voltage ( E ) as a function of current density can be written as; conc ohm act OCPE E (2 6)

PAGE 25

25 where EOCP is open circuit potential including leakage current and act, ohm and conc represent the activ ation, ohmic, and concentration polarization, respectively. Detailed polarization mechanisms are explained in following subsections. 2.2.1 Open c ircuit P otential (OCP) and T ransference N umber (ti) The voltage is generated across a cell by various gas mixtures with two different oxygen partial pressures (Po2). The opencircuit potential for the oxygen potential gradient cells is given by the well known Nernst equation in a cell [3] ln 42 2Po Po F RT E (2 7) where Po2 and Po2 are the equilibrium partial pressure of oxygen at the two sides of the cell, R is the ideal gas constant, F is Faraday s constant and T is the absolute temperature. The reaction for oxygen is, 2 2 2 4 ) ( O e g O (2 8) If one knows the oxygen partial pressure of the reference electrode (Po2) and measures the OCP at a given temperature, the equilibrium value of oxygen pressure at the working electrode may be determined from eq. ( 2 7 ). Generally, low oxygen pressure can be easily obtained under CO/CO2 or H2/H2O mixture [23] However, previous results showed that in CO/CO2 gas mixtures, the equilibrium was not readily attained, while H2/H2O gas mixtures showed equilibr ation for low Po2. In this study, H2/H2O gas mixtures w ere used to maintain low oxygen partial pressure on the anode side. The hydrogen w as bubbled with 3% of H2O through a membrane submersed in water. The condition of a controlled oxygen partial pressure can be obtained via thermodynamic relations at equilibrium. At high temperature, H2 and H2O gases react with traces of oxygen as following.

PAGE 26

26 O H O g H2 2 2 2 1) ( (2 9) And the Gibbs free energy is 2 12 2 2ln ) (O H O H oP P P RT TG G (2 10) where 0G (T) is the standard Gibbs free energy of the reaction, R is ideal gas constant, and T is temperature. From the thermodynamic data [23] ) ( ) ( 7 44 )( 000 242 ) ( K T K mol J mol J T Go (2 11) At equilibrium, 0 G 2 12 2 2ln ) (O H O H oP P P RT T G (2 12) 2 2) ) ( exp(2 2 2 RT T G P P Po H O H O (2 13) Therefore, with the reference oxygen potential temperature, and H2/H2O partial pressures, the controlled oxygen partial pressure can be calculated. For example, the experimental condition can be maintained with reference oxygen partial pressure of 0.21 atm on the cathode side (1 atm air) and 3% of H2O with H2 on the anode side. Based on these experimental conditions, the theoretical Nernst voltage was calculated by eq. ( 2 7 ) to ( 2 13) over the temperature range of 500 to 700 oC in 50 oC increments as shown in Table 2 1. The transference number (or transport number) ti, is defined as the fraction of the total conducti vity due to each charged species;

PAGE 27

27 total i it (2 14) The ionic transference number, ti, is equal to 1 for a purely ionic conductor, such as YSZ. Approaching unity of ti means that there is no significant electron or hole conduction. Any electronic conductivity causes an internal short circuit in the electrolyte of a fuel cell. The ratio between the measured and the theoretical OCP or Nernst voltage is tr ansference number ( ti); al theorectic measured iOCP OCP t (2 15) The transference number will be evaluated from the experimentally measured OCP values with the theoretical values in Ta ble 2 1. 2.2.2 Irreversible Losses 2.2.2.1 Activation p olarization l osses At low current density condition, the slow reaction kinetics at the cathode and the anode can cause the activation polarization. In other words, the excessive energy to overcome a energy barrier for electrode reaction such as the oxygen reduction and hydrogen oxidation produces the voltage drop which is increased with the current density drawn. T he phenomena can be formulated by the Butler Volmer equation [24] ; RT nF RT nF i ia act a c a act a a a , 0exp exp (2 16) RT nF RT nF i ic act c c c act c a c , 0exp exp (2 17)

PAGE 28

28 w here i0,a, i0,c are the exchange current densities of the anode and cathode, respectively, a,a and c,a are the anodic (i>0) and cathodic (i<0) charge transfer coefficients of the anode, and a,c and c, c are the anodic and cathodic charge transfer coefficients of the cathode. The charge transfer coefficients depend on the electrocatalytic reac tion mechanism and usually ~0.5 is used for SOFCs It is noted that as shown above, the Butler Volmer equation should be applied for each electrode separately This non linear Butler Volmer equation can be simplified as; i i i nF RTc act 0 0ln (2 18) In this case, this equation only considers on forward direction reaction, that is, reduction at the cathode side and oxidation at the anode side. Surprisingly, this simplified form was already predicted in empirical equation by Tafel in 1905 [25] which is written as; i i aact 0ln log (2 19) where a is a called the Tafel slope. 2.2.2.2 Leakage c urrent p olarization l osses In principle, voltage measurement of SOFC under open circuit condition (OCP) should show the theoretical Nernst voltage at the testing temperature. However, t he actual measurement of OCP usually has some deviation from the theoretical voltage. Even YSZ known as purely ionic conductor still sometimes shows some OCP deviation, which might be comes from gas leaks across the electrolyte itself or poor seal. For mix ed ionic and electronic conductors (MIECs), such as dopeceria, the partial electronic conduction causes OCP drop from theoretical one. For example, the

PAGE 29

29 theoretical OCP at 650 oC is ~1.14V but the reported OCPs of the GDC electrolyte cells are 0.7 ~ 0.8V [7] As explained above, the ratio of measured OCP to theoretical one is expressed as trans ference number, ti. 2.2.2.3 Ohmic polarization The ohmic polarization is the losses due to total electrical resistances from electrodes, electrolyte and lead wires. This polarization loss simply follows Ohm s law (V=IR). Therefore, ohmic polarization whic h is function of current can be expressed by ) (contact e electrolyt electrode ohmR R R I (2 20) where Relectrode, Relectrolyte, and Rcontact are resistances from electrode (both cathode and anode), electrolyte and electrode ele ctrolyte contact, respectiv ely, and I is current density. For SOFC structure, it has been generally accepted that most of the ohmic resistance comes from electrolyte due to much slow conduction process of ion migration rather than that of electrons. In thi s case, comparison of resistance between electrolytes is difficult due to its thickness dependence. Moreover, performance of SOFC is measured as a function of current density ( i ) not current ( I ). Therefore, ohmic polarization of electrochemical devices is generally expressed by; tot tot tot ohm ASR i R A i R I (2 21) where ASRtot is total area specific resistance. 2.2.2.4 Concentration polarization Concentration polarization is generally observed at high current densi ty regime of I V curves due to restriction to the transport of the fuel gas molecules to the anodic reaction site. At high current density, excess water byproduct can block the reaction sites. Therefore significant deactivation of reaction sites can be occ urred. The

PAGE 30

30 concentration polarization can be alleviated by higher gas pressure to drive out excess water from the reaction sites, reduction of anode thickness to shorten the distance to electrolyte, or higher porosity formation with same reason. 2.3. Mat erials and Design 2.3.1 Stabilized Zirconia Electrolytes F o r high temperature SOFCs, stabilized zirconia, such as yittriastabilized zironia (YSZ) has been most widely used as an electrolyte due to its high stability, reasonable ionic conductivity at high temperature (> ~900 oC), and relatively inexpensive cost [6] Although pure zirconia (Zr O2) is also chemically stable in both oxidizing and reducing conditions, it has not been chosen as a solid electrolyte due to its poor ionic conductivity. In addition to low conduction, pure ZrO2 shows phase tran sition from monoclinic to tetragonal and from tetragonal to cubic fluorite at 1170 and 2370 oC, respectively, accompanying unacceptable volume change (3 ~ 5%) in the fabrication temperature ranges [26] However, it has been known that some aliovalent cations, such as cations of Ca, Y, Mg and Sc, can stabilize the ZrO2 phase as cubic fluorite structure from room temperature to high temperature [27] Moreover, this aliovalent cation doping in ZrO2 produces higher vacancy concentration, leading higher ionic conduction at the wider Po2 ranges. This aliovalent dopant effect on vacancy concentration can be explained by Kroger Vink notat ion in which the negative charge produced by substituting a dopant is indicated by prime or a superscript dot if it is positive. The amount of charge is indicated by the number or prime or dot. For neutrality after substituting, it is marked as the supersc ript x For example, the incorporation reaction between trivalent dopant and ZrO2 can be written as [3] ;

PAGE 31

31 V O M O Mx o Zr ZrO3 223 2 (2 22) where the M is trivalent dopant and V is vacancy, indicating that two M dopants produce one oxygen vacancy. Various tri and divalent dopants has been studied to m ake the stabilize ZrO2 with high ionic conduction. It was shown that there is certain dopant concentration to give maximum conductivity For yttria stabilized zirconia (YSZ), as Y dopant concentration increases, the conductivity is increased upto 8 mol %, w hile over 8 mol % it shows degradation [28] As show n in Fig 2 3 Most of the stabilized zirconia shows similar trend [2] It is explained that at higher dopant concentration, defect ordering or vacancy clustering occurs leading the reduction of total number of active vacancy [29] Among the dopants, Y is most widely used due to cost and stability, while highest conductivity has been reported for Sc dopant ( Table 2 2 ) [29] The ionic conductivity of YSZ is strongly depends on the concentration and mobi lity of ions, which is known to a thermally activated process. Therefore, conductivity of YSZ suffers significant conductivity reduction at low temperature, which limits it operational temperature ~1000 oC. 2.3.2 Aliovalent CationsDoped Ceria Electrolytes Recently, aliovalent cations doped ceria (CeO2) has been given much attention as a potential solid electrolyte because of its higher ionic conductivity over a range of high to intermediate temperature. The i onic conductivity of ceria is considerably incr eased by aliovalent cation doping which increases the oxygen vacancy concentration in ceria [30, 31] The magnitude of electrical conductivity and the stability under reducing conditions for ceria based oxides depend greatly on the kind and quantity of doping elements. Alkaline earth oxides(e.g. CaO a nd SrO) and rare earth oxides(e.g. Gd2O3 and Sm2O3)

PAGE 32

32 are highly soluble in the ceria sublattice. Among these, Sm and Gddoped ceria shows the highest electrical conductivity in ceria based oxides with 10~20% of dopant concentration [31] It is co nsidered because of the similar ionic radii of Sm+3 and Gd+3 to that of Ce+4. Since Steele calmed the ionic conductivity of Gd0.1Ce0.9O1.95(10GDC) is the highest among various Gd and Sm dopant concentrations at a temperature range of 500~700oC, 10GDC has been paid attention as one of the most suitable candidates for IT SOFCs electrolyte [7] Mo reover, as shown in Fig. 24 the ionic conductivity of dopedceria is approximately one to two order of magnitude greater than that of stabilized zirconia, which is most widely used electrolyte material up to present [3] It is considered because Ce4+ (0.87 ) ion has larger ionic radius that Zr4+ (0.72 ) causing easier oxygen ion migration through a more open structure. However, one big drawback of ceriabased electrolyte makes us hesitate to select it as a best electrolyte material for IT SOFCs despite of its high ionic conductivity. When ceriabased oxides are reduced at low oxygen partial pressures (<1014 atm), Ce4+ transfers into Ce3+ leading significant n type electronic conduction with a P(O2)1/4 dependence [7] This phenomenon reduces the ionic transference number ( ti) and the open circuit potential (OCP), thereby making ceria less efficient for application as an ITSOFCs. To increase the electrolyte domain and to preserve the ionic conductivity of the doped ceria by any means is important. 2.3.3 Stabilized Bismuth Oxide Electrolytes Various polymorphism in bismuth oxide based materials have been identified with [32] Even though the cubic phase at high temperature (> 729 oC) shows an excellent ionic conductivity, attributed to the presence of such a large concentration of oxygen vacancies, it is unstable and transforms into a monoclinic phase with below 729 oC resulting in a discontinuous decrease in conductivity [33]

PAGE 33

33 However, when a solid solution of bismuth oxide is formed with erbia(Er2O3) or several other rareearth oxides, bismuth oxide is known to remain stable in the cubic phase. 20% erbia stabilized oxide (ESB) has the excellent conductivity among the stabilizedbismuth oxides. The greater conductivity of stabilizedbismuth oxide electrolytes has tremendous potential for lower operating temperature, thus considerably growing the number of applications for SOFCs [34] .Despite the high conductivity of stabilized bismuth oxide electrolytes, they have not been used in solid electrochemical devices such as SOFCs due to their thermodynamic instability. Takahashi et al. indicated that the critical PO2 value below which stabilized bismuth oxide would decompose is the equilibrium oxygen pressure of a Bi/Bi2O3 mixture [16] The decomposition process may be simplified as: ) ( 2 3 ) ( 22 3 2g O s Bi O Bi (2 -23) The opencircuit potential from galvanic cells w ith an air cathode and metal/metal oxide anode were stable in oxygen partial pressures above 1013.1 atm a t 600oC. This result showed that there was no contribution of electronic conduction to the total conductivity above the equilibrium oxygen potential o f Bi/Bi2O3 mixture. Therefore, they concluded that the minimum oxygen partial pressure at 600oC is 1013.1atm. On the other hand, Wang and other researches reported the ionic conductivity of Bi2O3 could be measured without critical decomposition of Bi/Bi2O3 under an H2/H2O atmosphere [35, 36] Wachsman et al reported that the measured conductivit y of ESB was independent of Po2 over the range 1 to 1022 atm under O2/Ar atmosphere as shown in Fig. 2 5 while ESB was decomposed at low Po2(1021atm) with H2, which was also confirmed by XRD [37] It was considered that the stability of ESB in the Ar/O2

PAGE 34

34 atmosphere is probably due to slow heterogeneous kinetics in the absence of an active reducing agent, such as H2. F rom these results, it can considered that bismuth oxide based electrolytes may be kinetically stable in the absence of contact with an active reducing agent. Therefore, in order to use bismuth oxide as an ITSOFC electrolyte, exposure to the reducing environment of the fuel gases must be prevented due to decomposition of the bismuth oxide. 2.3.4 Bilayer ed E lectrolyte C oncept for H igh P erformance ITSOFCs 2.3.4.1 Ceria / Zirconia bilayer electrolyte Bilayered electrolytes have been proposed as an alternative of overcoming the decomposition by the thermodynamic instability of highly conductive oxides. Yahiro et al. demonstrated that a thin and dense layer of YSZ on the fuel side of ceria avoided the effect of reduction of electrolytes by blocking the electronic conduction and consequently increased the OCP and power density [38] Alternatively, some researchers have suggested placing the YSZ layer, which has a low electronic conductivity, on the air ( i.e., oxidizing) side of the SOFC where its function is only to block electronic flux (thereby increasing the efficiency of the SOFC) [23] Of the two approaches the latter has been the most successful. However, i n both cases the YSZ layer could not be made thin enough for the total ionic conductance of the bilayer to be high enough for efficient power generation at low temperatures. Generally, a YSZ/SDC or SDC/YSZ bilayer electrolyte has no intrinsic advantage over just a thin YSZ electrolyte itself, other than providing a nonporous substrate for YSZ deposition, due to the relatively low conductivity of YSZ [18]

PAGE 35

35 2. 3 4 2 Ceria / Bismuth oxide bilayer electrolyte Wachsman et al. proposed a bi layer electrolyte consisting of a layer of erbiastabilizaed bismuth oxide (ESB) on the oxidizing side and a layer of SDC or GDC on the reducing side[9] In this arrangement, the ceria layer would protect the bismuth oxide layer from decomposing by shieldi ng it from very low P o2 and the ESB layer would serve to block electronic flux. They demonstrated concepts of bilayered electrolytes by using a system, which consists of SDC (Ce0.8Sm0.2O1.5) on the reducing side and ESB (Bi0.8Er0.2O1.5) on the oxidizing si de [39] A s shown in Fig. 2 6 it was considered that the gas phase on either side is fixed by the gradient of oxygen partial pressure and the relative electronic conductivity relies on the local oxygen activity. In this bilayer structure, the SDC layer prevents the ESB layer from decomposing at very low P o2. That is, in bilayered bismuth/ceria electrolytes, thermodynamics stability of the bismuth oxide electrolytes can be elevated and ceria can be act as both an electrolyte and anode depending on local oxygen partial pressure. As a result, higher OCP can be obtainable in the bilayered bismuth/ceria electrolytes, since transference number of bismuth oxide electrolyte is unity [18] Based on the gradient oxygen partial pressure and vacancy transport theory, modeling results showed that the interfacial oxygen partial pressure can be mainly determined by the relative thickness ratio between two oxide electrolyte layers[24]. These results implies that the relative thickness ratio is the key parameter to the electrochemical performance of bilayer electrolytes, since the PO 2 at the ESB/SDC interface can be controlled by the thickness ratio of SDC and ESB layers. Recently Park et al. reported study result on SDC/ESB bilayer electrolyte [39, 40] In this paper, they successfully deposited thin ESB layers with pulsed laser deposition (PLD) technique and a dip coating method on 1.7mm thick SDC pellet and

PAGE 36

36 showed that there is no interfacial phase formation. Such a formation can lower the total electrolyte conductivity and cause the ESB layer to be ineffective in blocking electron flow from ceria electrolyte causing higher opencircuit potential. Although the studied relative thickness ratio(LESB/LSDC) of the bismuth/ceria bilayer electrolytes was up to 102 level, it was expected that the higher relative thickness ratio can obtain higher electron open circuit potential without loss of high ionic c onductivity on electrolyte. Leng et al. also demonstrated the bilayer concept through YDB (yittriadoped bismuth)/ GDC electrolyte system [41] Their results showed not only higher opencircuit potential but also lowering electrode polarization effect due to the bismuth oxide interlayer. In this study they used relatively higher thickness ratio of 0.3, which implies the possibility of thermodynamic stability of thin film bilayer electrolyte even in higher ratio. However, o nly one thickness ratio was used and the cathode (Pt) was different from previous studies (Au). The thickness ratio effect cannot be directly compared with other results. Recently, modeling of the transport in ESB/SDC bilayer electrolytes has shown that the thickness of the bismuth oxide layer can be increased relative to the ceria layer, due to the increase in the electrolytic domain with decreasing operation temperature [42] As shown in Fig. 2 7 it has been expected that, at higher the ESB/SDC electrolyte thickness ratio, the total electrolyte resistance can be dominantly influenced by higher conductivity bismuth oxide layer, from which we can expect a significantly lower ASR (area specific resistance) as well as very high OCP.

PAGE 37

37 Figure 21. S chematic diagram of reactions in SOFCs based on oxygenion conductors [22]

PAGE 38

38 Figure 22. SOFC current voltage behavior indicating relative polarization losses [4]

PAGE 39

39 Figure 23. Variation of ionic conductivity of stabilized ZrO2 with dopant concentration (T=807oC) [2]

PAGE 40

40 Fig ure 2 4 Conductivities of selected electrolyte materials [3]

PAGE 41

41 Figure 2 5 ESB conductivity versus Po2 in purified argon atmosphere [37] Fig ure 26 Conceptual representation of a bilayer electrolyte showing the effect of relative thickness on interfacial oxygen partial pressure (P o 2 ) [39]

PAGE 42

42 0.04 0.06 0.08 0.1 0.12 0.001 0.01 0.1 1 10 100 1000 Thickness Ratio (LESB/LSDC) ASR ( cm2) Thicker ESB Thicker SDC T=500C Total thickness=10 m Fig ure 27 Bulk electrolyte ASR at 500 o C as a function of relative (t = LESB/LSDC) and total thickness for bilayers

PAGE 43

43 Tabl e 2 1. Calculated Po2 and Nernst voltage at opencircuit condition (T=500~700oC) Temperature (oC) 0G (T) (J/mol) Po2 (atm) Nernst Voltage; OCPtheoretical (V) Experimental Conditions 700 198507 4.641x10 25 1.1415 2 2H O HP P =0.0309 Po2,ref=0.21 atm 650 200742 1.816 x10 26 1.1473 600 202977 4.900 x10 28 1.1531 550 205212 8.526 x10 30 1.1588 500 207447 8.784 x10 32 1.1646 Table 22. Conductivity Data for Stabilized ZrO2 Doped with RareEarth Oxides [43] D opant Composition Conductivity (1000 o C) Activation eneragy (M2O3) (mol% M2O3) (X 102 1cm1) (kJ/mol) Y2O3 8 10 96 Nd2O3 15 1.4 104 Sm2O3 10 5.8 92 Yb2O3 10 11 82 Sc2O3 10 25 62

PAGE 44

44 C HAPTER 3 INTERGRATING NANO AND MICRO STRUCTURED ANODE FUNCTIONAL LAYERS FOR IMPROVED IT SOFC PERFORMANCE 3 .1 Introduction The commercialization of solid oxide fuel cells (SOFCs) as highly efficient low pollution power sources can be realized by lowering operational temperatures [3] As Steele claimed, in the lower temperature operation regime ( < ~ 700 oC ), the system cost can be significantly reduced with the use of cheap stainless steel for the bipolar plates and the balance of the plant, combined with the use of high temperature gaskets rather than rigid glass based seals The lower operation temperature can also enhance the SOFC s mechanical stability and life time [ 5] However, c ritical points to achieve this goal are to alleviate the significantly increased ohmic and activation polarizations at reduced temperatures due to their thermally activated nature [9, 44] An anodesupported designs for low (~ 500 oC) to intermediate (~ 600 oC) temperature SOFCs have recently received much attention since they can accommodate very thin electrolytes (< ~ 10 m), thereby eliminating a large fraction of the cells ohmic resistanc e [5] Until now, many anodesupported SOFCs with stateof theart thi n, highly conductive electrolytes, and highly electrocatalytic cathodes have been shown to obtain high power density at intermediate temperatures (IT) For example, Shao and Haile reported a high performance of ~ 1.01 W/cm2 at 600 oC with a barium doped perovskite cathode (Ba0.5Sr0.5Co0.8Fe0.2O3, BSCF ) accompanied with thin doped ceriaelectrolyte [45] Recent study also demonstrated a nickel and gadolinia doped ceria (Ce0.9Gd0.1O1.95, GDC) composite anodethick of bismuth/ceria bilayer e d electrolyte coupled to a bismuthruthenate cathode that achieved a high maximum power density (MPD) of ~ 1.95 W/cm2 at 650 oC [20] In the

PAGE 45

45 anodesupported design, however, anodic polarization at the interface betw een electrolyte and anode can dominat e due to its relatively high volume fraction compared to the electrolyte and cathode. Generally the SOFC anode provides the conducting phase for charge transfer as well as reaction sites for the electrochemical oxidati on of the fuel [2] For example with hydrogen as the fuel, the following reaction occurs e O H O H 22 2 2 ( 3 1) For low temperature SOFC application, Ni GDC cermet anode has been widely used due to high electrocatalytic effect of Ni both for the direct oxidation of hydrogen and for steam reforming of methane [46] The GDC in the cermet anode, which high ionic conductivity at low temperature, extends the reaction zone and compatibility with GDC electrolyte in addition to preventing Ni sintering. Moreover, it has been reported that doped ceria showed high resistance to carbon deposition in hydrocarbon fuel [47] I n this composite structure, to achieve higher anode performance, the anode microstructures should be carefully controlled. Recent research by Suzuki et al. showed that high MPD of ~ 1.1 and 0.5 W/cm2 at 600 and 550 oC with a anodesupported microtubular fuel cell using ~3 m thick Sc doped zirconia (ScSZ) and a GDC interlayer as electrolyte [48] I n this study, they demonstrated that ITSOFC performance greatly depends on anode microstructural factors. For example, porosity and size of Ni particle s in anode influenced concentration polarization and amount of triple phase boundary (TPB) which is believed as anodic reaction sites between the gas phase, ionic and electronic conduction phases. Moreover, it is generally accepted that most of the fuel oxidation reaction, such as eq.( 3 1), take place in a limited zone inside ~ 10 m

PAGE 46

46 thickness of anode adjacent to the anode/electrolyte interface [10] For this reason, int erfacial a node functional layers (AFLs) have been explored to increase the TPB length. T o date many studies have successfully demonstrated that graded AFL interlayers with submicron sized NiO particles at the anode/electrolyte interface can effectively ext end active TPB site reducing anodic polarization and give higher mechanical and chemical stability [11, 49] Most of these AFLs were fabricated by conventional colloidal slurry deposition. In the previous work author and coworkers introduced a new method for fabricating AFLs by dispersing a GDC precursor at the interface between the anode and the electrolyte [14] The resulting nanosized particles formed a smooth interfacial region between the anode and electrolyte by filling in pores and crevices and also extend ing TPBs. This resulted in an improvement in electrolyte deposition quality as well as higher electrochemical performance. Based on this study, the author developed a novel AFL which combines a conventional particle size (~ < 1 m) AFL applied by colloidal deposition and a nanosized NiGDC applied by a precursor solution coating. Due to the nature of the precursor solution, the nanoscale Ni GDC can penetrate into the AFL and be well distributed. As shown in Fig. 3 1 w e expect two major benefits from this bimodally integrated AFL concept. First, the TPB length of the AFL can be significantly higher than that of the conventional AFL because very fine Ni GDC particles surround the submicronsize NiGDC AFL particles. Secondly, precursor solutions form very fine particl es which fill submicron sized pores at the interfacial region and thereby increase the actual 2dimensional contact area with the electrolyte (2phase boundaries) so that

PAGE 47

47 it is in closer agreement with the nominal measured contact area, thus reducing the i nterfacial ohmic resistance. In addition to higher electrochemical performance, this novel AFL is very attractive for mass production of ITSOFCs because all materials are commercially available (NiO and GDC powder, and metal cation nitrates) and this simp le process is readily applicable to most anodesupported fuel cell designs with precise control. In this chapter the cost effective and high performance novel AFL for IT SOFCs is described. To investigate the effect of the novel AFL on SOFC performance, three cells all from the same anode tape were compared. These include no AFL, conventional AFL ( referred to as C AFL), and one sprayed precursor onto the conventional AFL to form a nano/micro composite AFL (referred to as N+C AFL). T hin GDC electrolyte and La0.6Sr0.4C0.2F0.8O3(LSCF) GDC composite cathodes were used for the balance of the button cells The microstructural evolution was analyzed and the electrochemical performance of the SOFC with this novel AFL were measured and characterized. 3 2 Experi mental The NiO GDC anode support was fabricated by tapecasting using a 65:35 (wt %) mixture of NiO (micron scale, Alfa Aesar) and GDC ( Rhodia) powders. Based on the ethanol solvent, an appropriate binder system was prepared with Solsperse, di n butyl phthalate (DBP) and poly vinyl butyral (PVB) with mixing ratio of 5.9 : 44.1 : 49.9 wt%, as the dispersant, plasticizer, and binder respectively. For a homogeneous slurry with proper viscosity and strength before and after tape casting, the binder system was m ixed with the powder mixture and ball milled for 24 hrs After a de airing step to avoid cracks or defects caused by air bubbles during the tapecasting process, a Procast tape casting system (DHI, Inc) produced a NiO GDC anode tape from the slurry. To mak e a

PAGE 48

48 button type fuel cell, the dried tape was punched out into a circular shape with a 32 mm diameter and presintered at 900 C for 2 hrs Fabrication of the N+C AFL on anode substrate involved two steps. First, a NiO GDC colloidal slurry containing subm icron size NiO (JT Baker) and GDC (Rhodia) (65:35 w t %) mixed with the proper binder system was deposited on one side of anode support by spin coating. The binder system consisted of Solsperse, di n butyl phthalate (DBP) and poly vinyl buteral (PVB) with mi xing ratio of 52.2 : 28.2 : 19.6 wt% as the dispersant, plasticizer, and binder respectively. In this work, the C AFL thickness was maintained by coating the support two times at 1500 rpm for 15 s. Subsequent heat treatment at 400 oC was carried out for the removal of the binder system. Next, the Ni GDC nitrate precursor was coated onto the first AFL. A 1 M solution of NiGDC precursor having the same mole ratio of each element as the anode substrate was synthesized by dissolving Ni(NO3)26H2O, Gd(NO3)36H2O and Ce(NO3)36H2O in e thyl alcohol using ultrasonication for 30 min. The precursor solution was transferred to a spray gun (Excell), sprayed onto the C AFL surface and presintered at 900 oC for 1 hr. Thin and even GDC electrolytes were deposited by spin coating with a GDC colloidal slurry. For the GDC colloidal slurry the Rhodia GDC powder was ball milled for 48 hrs with a binder system based on an ethanol solution. As a binder system, for 10 g of GDC powder, 0.5g of Solsperse (dispersant), 0.3g of PVB (binder) and 0.2g of DBP (plasticizer) were used with 70cc of ethanol The spin coating was conducted at 1500 rpm for 15s for each deposition. A f ter drying at room temperature for 10 hrs the anode/N+C AFL/electrolyte multilayer structure was sintered at 1450 oC for 4 hrs using a 3 oC /minute ramp rate in air.

PAGE 49

49 A La0.6Sr0.4Co0.2Fe0.8O3( LSCF) GDC composite cathode was prepared and applied on the GDC electrolyte surface. Cathode inks were synthesized by mixing LSCF (Praxair) and GDC (Rhodia) at 50:50w t %. For a solvent, Alphaterpiniol and ethanol were used. DBP and PVB were used as the plastisizer and binder, respectively. After mixing and grinding the cathode ink for 1 hour, the ink was brushpainted onto the GDC electrolyte evenly. The first layer of cathode ink was dried in an oven for 1 hour at 120 C, and a second layer of the same cathode ink was brush painted onto the first layer. The active cathode area was ~0.4 cm2. The cathode was fired at 1100 C for 1 hour. Ag mesh and Pt wire were bonded ont o both electrode surfaces using Pt paste for current collecting and then fired at 900 C for 1 hour. The AFL m icrostructures on Ni GDC anode supports were observed using scanning electron microscope (SEM, JEOL 6400 / 6335F). For electrochemical performance, fuel cell samples were loaded in sealed fuel cell testing apparatuses Current voltage (I V) characteristics were conducted by a Solartron 1287 using 3 0 sccm of dry air on the cathode side and 3 0 sccm of humidified (3 vol% H2O) hydrogen on the anode side In addition to the I V measurement, 2point electrochemical impedance analysis was carried out under open circuit condition using a Par stat 2273 (Princeton Applied Research) with a frequency range of 100 KHz to 100 m Hz. 3.3 Result and D iscussion Fig 3 2 shows the microstructure s of the anode substrate surfaces after application of the different AFL types and the evolution of these microstructures after various heat treatments T he AFL surfaces were compared after deposition and presintering at 900 oC ( Fig. 3 2 a, c, e) in order to inspect and compare the i nitial morphological state of N+CAFL deposition on the anode. As seen in Fig. 3 2 a, the

PAGE 50

50 substrate with no AFL has numerous micronsized pores. Such a microstructure can cause poor mechanical contact between the anode and electrolyte and lead to micro cracking during high temperature operation [11] In contrast, the microstructure of the N+CAFL covered surface (Fig. 3 2 e) exhibits little if any micron sized porosity and the surface particles are fine and well distributed. The surface coated with C AFL still exhibits a degree of micron size por osity (Fig. 3 2 c). Next, we sought to observe the microst ructure of the N+CAFL coating after testing. However, due to the presence of the electrolyte on the surface, it is difficult to observe the reduced anode surface after electrochemical test ing Therefore, in order to simulate the effect of testing, the thr ee anode samples were sintered at 1450 oC with no electrolyte coating and were reduced at 650 oC under simulated operational gas flow conditions for 10 hrs Fig. 3 2 b shows again that the bare anode forms big pores on the interfacial surface, compounded by the reduction of large NiO particles into Ni metal The conventional C AFL exhibits a much finer particulate microstructure ( Fig. 3 2 d ) T he characteristic microstructure of the N+C AFL is presented in Fig. 3 2 f. Compared with Fig. 3 2 d, the very fine Ni and GDC particles (marked as dotted circles) appear to b e better distributed between pores of the submicron size Ni GDC AFL network structure, indicating likelihood for extended TPB length s and 2 phase contact area with the electrolyte. Fig. 3 2 g exh ibits a characteristic N+C AFL structure magnified from Fig. 3 2 f. In this figure it is clearly shown that a nanosized particle is necking with submicron or micron sized AFL particles, which corresponds well with the schematic diagram of N+CAFL in Fig. 3 1 b. Fig. 3 2 h shows the cross sectional view of actual anode/N+C -

PAGE 51

51 AFL/electrolyte multilayered structure after cell testing, indicating the very fine AFL structure is well constructed with dense electrolyte and porous anode. Fig 3 3 shows the I V characteristics and power densities for the three different types of samples at the temperature ranges from 500 to 6 50 oC. In Fig. 3 3 a, I V plots of 3 cells were compared at 600 oC and the detailed values are tabulated in Table 3 1. The C AFL sample exhibited a maximum power density (MPD) of 681 mW/cm2, an i ncreas e of 127% compared to the sample which used no AFL which measured only 300mW/cm2. In addition, t he open circuit potential (OCP) of increased from 0.82 V for the cell without an AFL to 0.86 V for th e cell using the conventional AFL. This supports the theory that particle size graded anode functional layer s may improve the quality of GDC electrolyte deposition by partial filling the interfacial por osity [14] Although the OCP of the NiGDC AFL sample was about 0.04 V higher than the sample with no AFL, as sh own I V plot in Fig. 3 3 a, the major contribution to the improvement in electrochemical performance comes from its lower area specific resistance (ASR) For the novel N+C AFL cell, the MPD reached 1160 mW/cm2a 287% increase compared to sample with no AF L and 70% higher than that of sample using a C AFL while its OCP (0.85 V) was comparable to that of the C AFL sample (0.86 V). Again the highly improved performance is caused by the further reduction of the polarization loss es This result indicate that the precursor solution penetrated into the graded anode functional layer (Ni GDC AFL) and formed nanoparticles with proper percolati on and distribution of both GDC and Ni particles thus increasing active TPB length effectively at the interface between anode and GDC electrolyte.

PAGE 52

52 To further investigate the effect of the N+C AFL on performance enhancement, electrochemical impedance testing was conducted under open circuit condition on each sample at the same temperature range as used for I V measurement (Fig. 3 4). Comparison of the nyquist plots of the three samples at 600 oC are shown in Fig. 3 4 b T he total, ohmic, and electrode ASR values at 600 oC were extracted and are given in T able 3 1. As expected from the I V curves, the total ASR value of the N +C AFL sample (0.206 cm2) was much less than that of the C AFL sample (0.387 cm2) and the sample with no AFL (0.607 cm2) showing and ASR reduction of 47% and 66% respectively, which came largely from reduced electrode polarization drops The electrode ASR of the N + C AFL sample was reduced by 52% and 70% respectively, compared to C AFL and AFLless samples, suggesting again that the use of the N+CAFL did have a positive effect on extending TBP lengths In addition, it should be noted that the ohmic ASR (0.070 cm2) of the N+C AFL cell was also reduced by 31% and 53% respectively, compared with C AFL and no AFL cells. Based on previously reported ionic conductivity of GDC, ohmic ASR of ~10 m thick GDC electrolyte is 0.054 cm2, which is close to that of N+C AFL, while the ohmic ASRs of no AFL (0.149 cm2) and C AFL (0.102 cm2) cell shows much higher values. As reported by Koide et. a l, these additional IR resistance can be caused by higher interfacial contact resistance between anode and electrolyte [50] Therefore, this ohmic ASR drop of the novel AFL cell indicates that the N+C AF L effectively improved the interfacial wetting and expended the active contact area of two phase (electrolyte and anode) boundaries at the GDC electrolyte/Ni GDC anode interface, causing the lower resistance to the flow of

PAGE 53

53 oxygen ions through the electroly te to anode, as well as a two phase boundary area that more closely matches the nominal active area used in power density calculations T his ASR analysis implies two additional points with respect to anodesupported SOFCs. Firs t the anodic polarization in this SOFC design is a very large portion of the total electrode polarization, since modifying the anode alone using N+C AFL reduced electrode ASR by 70% Second, the fact that only ~ 5 m of N+C AFL lowered the total ASR by 66% shows that most of this anodic polarization occurs near the electrolyte/anode interfacial region (~ m of depth) which can be mitigated by the use of a proper anode functional layer at the interface. Fig. 3 5 summarizes MPD obtained from I V plots shown in Fig. 3 3 and total ASR va lues given in Fig. 4 of SOFCs with N+C AFL, C AFL, and no AFL at temperatures ranging from 650 oC down to 500 oC. The MPD of the N+CAFL cell reached 1296, 697, and 380 mW/cm2 at 650, 550, and 500 oC (Fig. 3 5 a). Thus, the improvement in performance becomes even greater at lower temperatures the power density of the N+CAFL cell compared to that of the cell with no AFL increased by 107% (625 to 1296 mW/cm2) at 650 oC and by 407% (75 to 380 mW/cm2) at 500 oC. The same trend is observed for the total ASR va lues as shown in Fig. 3 5 b. The total ASRs of the N+C AFL were 0.091, 0.497 and 0.654 cm2 at 650, 550 and 500 oC, respectively, showing total ASR reduction of 67.2 and 80.2% at 650 and 500 oC. This implies that at the low end of the IT range (below 650 oC), the ability of the N+C AFL to reduce the ASR and improve MPD was confirmed and the improvements were significant. On the other hand, for all samples, the portion of electrode ASR in total ASR is getting greater as decreasing temperature (Fig. 3 4). Ev en N+C AFL showed electrode ASR fraction

PAGE 54

54 increase in total ASR from 54.8% at 650 oC to 75.8% at 500 oC. We believe this is because the sluggish oxygen reduction reaction occurred at the conventional perovskite cathode due to high activation energy at the l ower temperature. Coupled with recently reported highly catalytic cathodes, even higher performance of this SOFC is expected at low temperature. 3.4 Conclusions In conclusion, bimodally integrated nano/micron composite AFL was developed by simple spray coating a precursor solution into conventional submicron sized Ni GDC in a functional layer. This combined structure produced a novel N+C AFL. Microstructural analysis revealed that very fine Ni and GDC particles were homogeneously distributed into the co nvent ional AFL and formed network structures in 3D, leading to a significant increase in TBP length. A SOFC using this novel AFL exhibited a MPD of 1.16 W/cm2 at 600oC. Due to its characteristic structure, N+C AFL reduced both electrode and ohmic ASR. Co mpared with the performance of a SOFC without AFL, the cell using the N+CAFL showed a 287% increase in power density as well as a 66% reduction in total ASR This effect was observed throughout the IT range tested, indicating the N+CAFL is a n excellent structure for use in high performance ITSOFCs.

PAGE 55

55 Submiro sized NiO particles Micro sized NiO particles GDC particles Nano sized Ni cations in solution step1 step2 Ni GDC precursor solution GDC electrolyte N+C AFL Ni-GDC anodeNi GDC Ni GDC Figure 3 1 Schematic illustration of the proposed N+C AFL structure on anodesupported SOFC and effect of N+C AFL on expending TPB length. Yellow triangles represent TPBs in conventional AFL (C AFL) and red triangles represent TPBs by N+C AFL.

PAGE 56

56 (a) (b) (d) (c) (e) (f)5 m 5 m 5 m 5 m 5 m 5 m 1 m 1 m 1 m 1 m 1 m 1 m 100nm 100nm (g) (h) 5 m Electrolyte N+C AFL Anode Figure 32. SEM micrographs of the anode surface after deposition and presintering (a, c, e) and after full sintering followed by simulated testing atmospheric conditions (b, d, f) for samples with no AFL ( a, b), C AFL (c, d) and N+C AFL (e, f). Highly magnified image shows the characteristic N+C AFL structure (g). The cross sectional image presents actual multilayered fuel cell with N+C AFL(h).

PAGE 57

57 0 1 2 3 4 0.0 0.2 0.4 0.6 0.8 1.0 Current Density (A/cm2)Potential (V)0.0 0.2 0.4 0.6 0.8 1.0 1.2 Power Density (W/cm2) 600oC N+CAFL C AFL no AFL (a) 0 1 2 3 4 0.0 0.2 0.4 0.6 0.8 1.0 Current Density (A/cm2)Potential (V)0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Power Density (W/cm2) N+CAFL 650oC 600oC 550oC 500oC ( b) Figure 33 Comparison of I V characteristics fo r the fuel cell samples with N + C AFL, C AFL, and no AFL at 600 C. (a) I V plots at the temperature ranging from 650 to 500 o C for N+CAFL (b), C AFL (c), and no AFL (d).

PAGE 58

58 0 1 2 3 4 0.0 0.2 0.4 0.6 0.8 1.0 Current Density (A/cm2)Potential (V)0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Power Density (W/cm2) C AFL 650oC 600oC 550oC 500oC (d) 0 1 2 3 0.0 0.2 0.4 0.6 0.8 1.0 Current Density (A/cm2)Potential (V)0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Power Density (W/cm2) no AFL 650oC 600oC 550oC 500oC (c) Figure 33. Continued

PAGE 59

59 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10 -Z'' ( cm2) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.0 0.1 0.2 -Z'' ( cm2) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.0 0.2 0.4 -Z'' ( cm2) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.0 0.5 1.0 Z' ( cm2)-Z'' ( cm2) 650oC 600oC 550oC 500oC N+C AFL C AFL no AFL(a) (b) (c) (d) Figure 3 4. Electrochemical impedance spectr a of the testing samples with N+C AFL, C AFL and no AFL at various temperature; 650 C (a), 600 C (b), 550 C (c), and 500 C (d).

PAGE 60

60 500 550 600 650 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Maximum Power Density (W/cm2)Temperature (oC) N+C-AFL C-AFL no AFL (a) 500 550 600 650 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 N+C-AFL C-AFL no AFL Area Specific Resistance ( cm2)Temperature (oC) (b) Figure 3 5. MPD (a) and ASR plots (b) for the different samples tested between 500 and 650 C.

PAGE 61

61 Table 3 1. De tailed OCP, MPD and ASR values of the fuel cell samples with N+C AFL, C AFL, no AFL at 600oC. Cell Type OCP MPD Total ASR Ohmic ASR Electrode ASR unit V mW/cm 2 2 2 2 no AFL 0.82 300 0.607 0.149 0.458 C AFL 0.86 682 0.387 0.102 0.285 N+C AFL 0.85 1156 0.206 0.070 0.136

PAGE 62

62 CHAPTER 4 EFFECT OF NI GDC AFL COMPOSITION ON PERFORMANCE OF IT SOFCS 4.1 Introduction Wider commercialization of solid oxide fuel c ells (SOFCs) can be achieved by lowering the operation temperature [5] Critical points to achieve this goal are to alleviate the significantly increased ohmic and activation polarizations at reduced temperatures due to their thermally activated nature [5, 7, 44] Anodesupported designs for solid oxide fuel cells operating at intermediate temperatures (IT, 500 ~ 650 C) are widely used due to their ability to utilize a very thi eliminating a large fraction of the ohmic polarization loss [51, 52] Currently state of the art thin electrolytes and highly electrocatalytic cathodes are applied on anodesupported cells, showing high performance at IT ranges. For example, we reported that anode supported SOFCs using a Ni/ Ce0.9Gd0.1O1.95 ( GDC) composite anode, bismuth oxide/GDC bilayered electrolyte and bismuth/ruthenate composite cathode achieved an exceptionally high maximum power density (MPD) of ~ 2 W/cm2 at 650 C [20] In the anodesupported design, however, anodic polarization at the interface can dominate due to its relatively high fraction of volume compared to the electrolyte and cathode. Generally, the anodesupport is fabricated using submicronsized N iO particles with poreformer to increase porosity in the anode [11, 53] Even though sufficient porosity is readily achieved, the large sized pores at the electrolyte interface cause a large interfacial anodic loss and low mechanical strength due to the poor quality of electrolyte deposition [11] To overcome these problems, interfacial anode functional layers have been developed. Up to this date, many researchers have reported that thin anode functional layers (<10 former effectively reduce

PAGE 63

63 activation polarization by increasing triple phase boundary (TPB) density, strengthening mechanical properties and lowering ohmic losses by improving the quality of electrolyte deposition due to significantly reducing interfacial porosity [11, 5356] For the last several years, we have concentrated on developing high performance IT SOFCs with a tape cast Ni GDC anode using large micron sized NiO particl es [14] Although this anode design provided sufficient mec hanical properties and porosity without additional poreformer, it still suffered from low performance The low performance is due to the coarse microstructure with large size pores at the electrolyte interface, causing poor electrolyte deposition. Recentl y, Ahn et al introduced a new method to establish AFLs by dispersing a GDC precursor solution at the anode/electrolyte interface by simple spray coating. This result ed in ultra fine particles forming a much smooth er interfac e, leading to extended TPB dens ity [14] Base d on this study, Lee et al developed a novel AFL integrating a nanosized Ni GDC particles into a submicronsize d AFL by a simple precursor solution coating [57] Because of the nature of the precursor solution, very fine NiGDC p articles were well distributed and formed a bimodally structured AFL. The AFL led to a high performance of ~1.3 W/cm2 at 650 C with a 10 We have shown how microstructural changes due to different particle sizes influence the electrochemical performance for SOFCs with respect to the surface porosity and anode active reaction sites, TPBs. However, the spatial distrib ution, amount of porosity and TPBs in two phase composite anodes can be significantly affected by the amount of each phase [12, 58] In this study, we fabricated and investigated ITSOFCs having AFLs of various compositions using submicronsized NiO and GDC particles at the electrolyte interface. Electrochemical

PAGE 64

64 performance studies were conducted by current volage ( I V ) and impedance testing. The optimal composition of this AFL was investigated, and the relationships between composition and performance were analyzed. 4.2 Experimental 4.2.1 Cell F abrication Flat NiO GDC anode supports were fabricated by the tapecasting method. In order to provide a sufficient gas channel without the aid of poreformer, micron sized l arge NiO particles (Alfa Aesar) were mixed with nano sized GDC from Rhodi a. In this study, the composition of the anode support was fixed at 65 to 35wt% of NiO to GDC. Based on the ethanol solvent, an appropriate binder system was prepared with Solsperse, di n butyl phthalate (DBP) and poly vinyl butyral (PVB) as the dispersant, plasticizer, and binder respectively. For a homogeneous slurry with proper viscosity and strength before and after tape casting, the binder system was mixed with the powder mixture and ball milled for 24 hrs After a de airing step to avoid cracks or defects caused by air bubbles during the tapecasting process, a Procast tape casting system (DHI, Inc) produced a NiO GDC anode tape from the slurry. To make a button type fuel cell, the dried tape was punched out into a circular shape with a 32 mm diameter and presintered at 900 C for 2 hrs For a finer and graded AFL structure, submicronsized NiO (JT Baker) was mixed with nanosized GDC (Rhod i a). To investigate the effect of the c omposition of the AFL, various NiO GDC AFL content s, from 40wt% to 80wt% of NiO, were fabricated on the anode surface by spin coating. Similar to the tape casting slurry, a proper binder system was added to each AFL powder, leading to a well dispersed coll oidal slurry. During the spin coating process, the thickness of the NiGDC AFL layer was controlled by the

PAGE 65

65 number of depositions at the same spin speed and time. In this work, we produced the same thickness of AFL by applying 3 coats at 1500 rpm for 15s. A fter deposition, the AFL layer was presintered at 900 C for 1 hour to remove the binder system. For the electrolyte, the same GDC powder for the anode and AFL was ball milled for 48 hrs with a binder system based on an ethanol solution. As a binder system, Solsperse (dispersant), PVB (binder) and DBP (plasticizer) were used. The thin GDC electrolyte was coated by the spin coating method with same process as the AFL deposition. After deposition, samples were dried at room temperature for 10 hrs After dry ing the multilayer anode/AFL/electrolyte structure was sintered at 1450 C for 4 hrs A La0.6Sr0.4Co0.2Fe0.8O3( LSCF) GDC composite cathode was prepared and applied on the GDC electrolyte surface. Cathode inks were synthesized by mixing LSCF (Praxair) and GDC (Rhodia) at 50:50w t %. For a solvent, Alphaterpiniol and ethanol were used. DBP and PVB were used as th e plastisizer and binder, respectively. After mixing and grinding the cathode ink for 1 hour, the ink was brushpainted onto the GDC electrolyte evenly. The first layer of cathode ink was dried in an oven for 1 hour at 120 C, and a second layer of the sam e cathode ink was brush painted onto the first layer. The active cathode area was ~0.4 cm2. The cathode was fired at 1100 C for 1 hour. Ag mesh and Pt wire were bonded onto both electrode surfaces using Pt paste for current collecting and then fired at 900 C for 1 hour. 4.2.3 Characterization For electrochemical performance, the prepared cells were loaded on a fuel cell testing set up. In order to obtain the gas tight sealing, the edge of the cell and testing tube were covered with a mixture of two parts ceramic sealant using ceramabond517 (Aremco). I V tests were carried out by a Solartron 1407E with 3% wet hydrogen as a

PAGE 66

66 fuel on the anode side and dry air as an oxidant on the cathode side at various temperatures. The gas flow rate was controlled by a mas s flow controller (MKS 647C). In addition to current voltage characteristics, electrochemical impedance spectroscopy with a two point probe was measured by a Solartron 1400 using a frequency range of 100 kHz to 100 mHz under the same gas and temperature c onditions. For microstructural analysis, the tested fuel cells were fractured and the cross sections of the multilayered structures were observed using a scanning electron microscope (SEM, JEOL 6400 / 6335F) with the back scattering mode. 4.3 Results and Discussion In this study, five kinds of fuel cells with AFLs containing 40, 50, 60, 65, and 80 wt% of NiO were prepared in order to investigate the effect of AFL composition on the microstructure and electrochemical performance. At the same time, a cell without AFL was tested as a reference cell. 4.3.1 Microstructural Analysis Figure 4 1 shows the microstructures from a cross sectional view of the samples with different AFL compositions after reducing and testing under 90 sccm of air on the cathode side and hydrogen on the anode side. In order to directly compare the effect of the AFL, the thickness of the electrolyte, anode, cathode, and active cathode area were designed to be the same between the different samples. As seen in Figure 4 1 a~f the dense GDC electrolyte for each cell shows an even thickness of 17 to 19 m with a few closed pores. For the other anode and cathode, the thicknesses were measured by SEM and found to be nearly identical between the samples at 290 m to 300 m for anode and ~30 m for cathode. For the AFLs, each AFL thickness measured was about 10 m. In addition, it is clearly shown that interfaces between the AFLs and GDC

PAGE 67

67 electrolyte (Fig 4 1 b~f) look very flat and continuous while the interface between the GDC electrolyte and anode support (no AFL sample, Fig 4 1 a ) shows a very irregular and discontinuous morphology due to large pores caused by reduction of large size NiO into Ni. In Fig 4 2, the magnified microstructures of the anode/AFLs nearby the GDC electrolyte after reducing and testing are presented. In c omparing Fig 4 2 a to 4 2 b~f, particulate size differences between the AFLs and anode are clearly shown indicating that submicronsized NiO successfully formed a very fine NiGDC AFL structure. Using backscatte red mode the Ni (dark gray), GDC (white), and pore (black) phases are well distinguishable due to contrast difference. As shown in Fig 4 2 b~f, it is observed that the amount of both Ni phase and porosity due to reduction of NiO into Ni during the oper ation increased with higher NiO wt% in AFL. 4.3.2 Effect of AFL Compositi on on Power Density 4.3.2.1 I V c haracteristics at 650 C Fig 4 3 shows I V characteristics of fuel cells with various compositions of AFL at 650 C. The open circuit potential of the test ed cells were 0.806, 0.834, 0.838, 0.814, 0.807, and 0.801 V for t he cells with no AFL, 40, 50, 60, 65, and 80wt% NiO AFL. It has been commonly reported that formation of AFL at the interface between the anode and thin electrolyte enhances OCP [13, 14, 57] Generally the theoretic al Nernst voltage for thin electrolytes below 10 m can be lowered by gas permeation through electrolyte which is caused by various structural defects such as open pores and microcracks in addition to inherent internal shorting of a mixed ionic and electronic conductor (MIEC) such as doped ceria. In that case, AFL can help to enhance the quality of the electrolyte deposition by reducing the possibility of crack and pore formation at the interface due to

PAGE 68

68 its finer and reduced surface porosity [14, 57] In addition, as Chen et al reported, AFL itself can act as a barrier to gas per meation along with the electrolyte although this can cause concentration polarization [13] In this study, however, the OCP does not increase only by the existence of the AFL as shown between the AFL cell (0.806V) and 65% wt NiO AFL cell (0.807 V). It is considered that t he electrolyte thickness (~20um) was much larger than the size of bare anode surface pore (below 5 m), suggesting a similar gas permeability in all electrolytes regardless of AFL existence. Instead, the observed OCP seems to be influenced by AFL composition. Fig 4 4 shows the plot of OCP versus NiO composition of the AFL or anode at the anode/electrolyte interface. Higher OCP is observed with decreasing NiO wt% in the AFL. SEM observation of the anode and AFL structure before reduction reveals that the anode without poreformer and AFL are relatively dense with little porosity. Therefore, it is considered that the total pore volume in the anode or AFL after testing increases proportionally with NiO content because the only source to produce porosity is the reduction of NiO under operational conditions. This indicates that a n AFL with lower NiO content forms less porosity after reduction and is a better barrier of gas permeation, leading to a higher OCP. In addition, the higher Ni content can cause lower OCP because of greater electronic conductivity. It has been reported that for thin MIEC electrolyte, OCP greatly depends on electrolyte thickness [59, 60] Therefore it is noted that for the ultra thin electrolyte system, the OCP change with AFL composition can be much greater and have a significant impact on cell performance.

PAGE 69

69 T he maximum power densities (MPD s) of cells with no AFL, 40, 50, 60, 65, and 80 wt% NiO AFL were obtained from the power density curves in Fig. 4 3 as 882, 732, 1033, 1147, 1077, and 711 mW/cm2, respectively. In addition to MPDs, total area specific resistance (ASR) value for each cell also estimated from the IV curves near OCP region. Resultant MPDs and total ASRs are plotted in Fig. 4 5. Compared to the no AFL cell, cells with 50, 60, and 65wt% NiO AFL showed significant improvement in MPD at 650 C. While the OCPs of these cells ( Fig 4 4) shows a different trend and relatively trivial change (~0.03V), the resultant performance enhancement mostly comes from decreasing polarization with AFL as shown in Fig 4 5 This implies that the finer AFL structures ( Fig 4 2 c,d,e) effectively inc reased TPB density at the anode/ electrolyte interface. In contrast, MPDs of cells with 40 and 80wt% NiO AFL compared to that of cell with no AFL were decreased. In this case, it is considered that although the AFLs formed very fine particulate structures ( Fig 4 2 b,f), excessively high content of NiO or GDC can reduce the connectivity of the TPBs (Ni, GDC, and pore) leading to inactive regions. Therefore, the AFL composition should be considered as an important factor in terms of SOFC performance. As ment ioned above, the highest MPD was obtained at the AFL with 60wt% of NiO (1147 mW/cm2) showing a 30% increase compared to no AFL cell (882 mW/cm2), while higher OCPs w ere observed at higher NiO wt% AFL s. Assuming full reduction of NiO into Ni during operation, the Ni to GDC volume ratio of this AFL is easily calculated with the densities of Ni, NiO and GDC. The results is 48.6vol% Ni to 51.4vol% GDC, that is almost 1:1 volume ratio. This result is in good agreement with a recent study of NiO SDC (samarium dope d ceria) AFL system by Ai et al [12] It was also predicted by

PAGE 70

70 previous modeling work. Schnieder et al. conducted analytical modeling for estimating TPB length of composite electrode using the discrete element method and showed that th e TPB length is maximized at 50vol% of the ionic conducting phase in the composit e anode [61] On the other hand, another recent study by Wilson and Barnett reported the lowest electrode ASR with the highest TPB length at 50wt% NiO in a NiO YSZ(yttria stabilized zirconia) active layer, which is a 1:2 volume ratio of Ni to YSZ [62] However, it is reasonably considered that the ionic conductivity of stabilized zirconia is much lower than that of doped ceria at the IT range ( 500 C ~ 800 C ) T he NiO particle size (~2.5 m) used in that study was much larger compared to this studys NiO size (< 1 m). Therefore, the lowest resistance with the highest TPB length can be achieved at l ower G DC content and higher NiO wt% leading to higher porosity. For further understanding, the quantification of microstructural properties will be needed to find relationship with electrochemical performances. 4.3.2.2 Temperature d ependence In order to verify the validity of th e optimal AFL composition of 60wt% NiO at the IT range, a current voltage measurement for each cell was conducted at various temperatures from 450 to 6 50 C with a 50 C interval. The resultant MPDs with AFL composition are plotted in Fi g 4 6 including the MPD s of the no AFL cell separately marked as open symbols For all tem peratures tested, cells with 60wt% NiO in the AFL shows highest MPDs with 1147, 626, 271, 118, and 45 mW/cm2 at 650, 600, 550, 500, and 450 C, respectively. This r esult shows that the effect of AFL optimal composition is valid through the intermediate to low temperature ranges.

PAGE 71

71 4.3.2.3 Long t erm s tability To see the effect of optimal composition AFL on high performance, a potentiostatic test under an applied voltage where cells reached 98% of their MPD was done for each cell. The applied voltages were 0.379 V for the AFL cell and 0.380 V for the no AFL cell. Fig 4 7 shows preliminary results of a 200 hrs long term stability test for the no AFL and 60wt% NiO AFL cells under the 90 sccm of H2/Air condition. While the no AFL cell shows initial degradation of power density and stabilized behavior, the effect of the optimal composition AFL was retained for 200 hrs with high power density of ~1.1 W/cm2. For practical appli cation of this AFL, however, further long term testing under various temperatures and gas conditions should be conducted. 4.3.2.4 Effect of AFL c omposition on ASR In order for further investigation, electrochemical impedance tests were carried out for all samples. Fig 4 8 a shows the impedance spectra of each AFL composition at 650 C for which an I V test was conducted. From the high and low frequency complex plane intercepts of the impedance spectr um with the real axis, the ohmic, electrode, and total ASR values were calculated while normalizing the resistance according to cathode area. T he detailed values are tabulated in Table 4 1 and plotted in Fig 4 8 b. As shown in the Table, in this study, total ASR values from electrochemical impedance (ASREIS) ar e within 4% deviation from the ASR at IV curves (ASRIV) in Fig 4 5. As expected from the previous section, total ASRs from AFL composition show a similar trend to the MPD T he lowest total ASR (0.188 2) at 650 C was achieved at 60wt% NiO AFL, which was decreased by 25.7% from that of the no AFL cell (0.253 2). The ohmic ASRs are similar for all samples because the samples tested in this study have similar thicknesses of electrolytes with simila r densities as supported by

PAGE 72

72 microstructures shown in Fig 4 1. Therefore, most of the ASR drop comes from the electrode ASR drop, which is considered as anodic polarization reduction since the cathodes should be identical. From this result it is expected that different AFL compositions d irectly influence a change in TPB density affecting anodic polarization. Most likely t he TPB density is the highest at 60wt % of NiO in AFL, which is at 1:1 volume ratio of Ni to GDC after reduction. Based on these results the relationship between electrode ASR and MPD with various AFL compositions are examined. As shown in Fig 4 5, the performance enhancement of power density reflects the ASR trend. In Fig 4 9 it is clearly shown that electrode ASR has an almost linear relationship with MPD. However, there are still some deviations from the linearity. At this point it is noted that the ASR s were measured at the open circuit condition, whereas the MPDs occurred at higher current densit ies which have complex contribution s from various polarizations, such as activation, ohmic and concentration polarization as shown in I V curves in Fig 4 3. Therefore, for a more precise study, the ASRs measured under applied currents will be done in the future. In addition, to completely understand electrode ASR, comprehensive microstructural features should be considered, such as surface area, porosity tortuosity and TPB density 4.4 Conclusions In this study, the effect of AFL composition on the electrochemical performance was investigat ed for IT SOFCs. For this, the various AFLs with composition from 40 to 80wt% NiO were fabricated. The fine and welldistributed AFL structures with different NiO amounts were confirmed by microstructural analysis. The optimal AFL composition was achieved at 1:1 volume ratio of Ni to GDC in the AFL, which is 60wt% NiO. The

PAGE 73

73 effect of the optimal AFL composition on MPD was valid at intermediate to low temperature ranges. In addition, a preliminary longterm stability test showed the possibility of practical a pplication of this optimal composition. The measured MPD and ASR show a linear relationship implying that the performance enhancement greatly depends on the AFL composition, which might be caused by microstructural features such as TPB density porosity, tortuosity of pores and surface area. For further understanding of the AFL effect on electrochemical performances, the quantitative analysis of AFL microstructures should be done.

PAGE 74

74 GDC electrolyte LSCF-GDC cathode Ni -GDC anode (a) (b) (d) (c) (e) (f) Anode functional layer Figure 4 1 Backscattered images showing a cross sectional view of anodesupported SOFCs with different NiO content in the anode f unctional layers; no AFL(a), 40wt% (b), 50wt%(c), 60wt%(d), 6 5 wt%(e), and 80wt%(f) NiO.

PAGE 75

75 Ni GDC Pore Ni GDC Pore (a) (b) (d) (c) (e) (f) Figure 4 2. Magnified microstructures of the anode or AFLs with different NiO content no AFL(a), 40wt% (b), 50wt%(c), 60wt%(d), 65wt%(e), and 80wt%(f) NiO. Backscattering mode provides better contrast to distinguish Ni (dark gray), GDC (white), and pore (black) phases.

PAGE 76

76 0 1 2 3 4 5 6 0.0 0.2 0.4 0.6 0.8 1.0 1.2 no AFL 40:60 AFL 50:50 AFL 60:40 AFL 65:35 AFL 80:20 AFLCurrent Density (A/cm2)Potential (V)0.0 0.2 0.4 0.6 0.8 1.0 1.2 Power Density(W/cm2) Figure 4 3. I V plots of fuel cells with various AFL compositions at 650oC; 40( ), 50( ), 60( ), 65( ) and 80( )wt% of NiO in AFL, and no AFL( ) The gas condition was 90sccm of air and 3% of wet hydrogen on the anode and cathode side, respectively.

PAGE 77

77 20 30 40 50 60 70 80 90 100 0.79 0.80 0.81 0.82 0.83 0.84 0.85 OCP (V)NiO in AFL (wt%) Figure 4 4. Open circuit potential of the fuel cells with various NiO contents in NiO GDC AFL. Solid line (red) shows linear fit of the measured data (square)

PAGE 78

78 40 50 60 70 80 200 400 600 800 1000 1200 1400 NiO in AFL (wt%)MPD (mW/cm2)0.10 0.15 0.20 0.25 0.30 Total ASRiv ( cm2) Figure 4 5 MPD (Red square) and total ASRIV estimated from IV curves (blue star) are plotted with NiO contents in AFL. The open symbols represent no AFL cell.

PAGE 79

79 600oC 550oC 500oC 450oC 650oC 40 50 60 70 80 0 200 400 600 800 1000 1200 MPD (mW/cm2)NiO in AFL (wt %) Figure 4 6 Maximum power densities of fuel cells with various AFL compositions at the temperature range from 450 to 650 C. Open symbols represent MPD of no AFL cell at each temperature.

PAGE 80

80 Figure 4 7 Long term stability test of fuel cell with 60wt% of NiO in the AFL and the no AFL cell for 200 hrs at 650 C. Potentiostatic tests were conducted with an applied voltage of 0.379 V for the NiO 60wt% AFL cell and 0.380 V for the no AFL cell, at which the cells showed 98% of MPD. The gas condition was 90sccm of air and 3% of wet hydrogen on the anode and cathode side, respectively.

PAGE 81

81 650oC 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10 0.15 0.20 0.25 0.30 no AFL 40:60 AFL 50:50 AFL 60:40 AFL 65:35 AFL 80:20 AFL -Z'' ( cm2)Z' ( cm2) (a) Electrode ASR Ohmic ASR Total ASR 40 50 60 70 80 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 Area Specific Resistance ( cm2)NiO in AFL (wt%) (b) Figure 4 8 Impedance spectra with various AFL compositions (a), and total, electrode, and ohmic ASRs of fuel cells with different NiO content (b) calculated from impedance spectra (a). Open symbols represent no AFL results.

PAGE 82

82 0.14 0.16 0.18 0.20 0.22 700 800 900 1000 1100 1200 Electrode ASR ( cm2) MPD (mW/cm2) 60:40 AFL 65:35 AFL 50:50 AFL no AFL 40:60 AFL 80:20 AFL Figure 4 9 MPD plots with electrode ASR shows a linear relationship Red line is linear fitting of the measured data (black dots).

PAGE 83

83 Table 4 1. Detailed ASR values of the testing c ells with various NiO contents in AFL AFL NiO (wt%) Total ASR iv ( cm2) Total ASR EIS ( cm 2 ) Ohmic ASR 2 ) Electrode ASR 2 ) 40:60 40 0.259 0.260 0.071 0.189 50:50 50 0.227 0.231 0.070 0.161 60:40 60 0.182 0.188 0.055 0.133 65:35 65 0.196 0.203 0.061 0.142 80:20 80 0.286 0.281 0.076 0.20 5 no AFL 65 0.253 0.253 0.074 0.179

PAGE 84

84 CHAPTER 5 COMPREHENSIVE QUANTI FICATION OF NIOGDC ANODE FUNCTIONAL LAYER MICROSTRUCTURE BY THREEDIMENSIONAL RECONSTR UCTION USING FIB/SEM 5.1 Introduction The i nevitable demand on lowering solid oxide fuel cell ( SOFC) operational temperatures has been fueled by commercialization of SOFCs [3, 5, 7, 9, 63] Currently the anode supported SOFC design has been widely studied due to possible accommodation of ultra thin electrolyt e [8, 20, 55] For this design, however, the anode polarization is possibly dominant due to the anode having the largest volume fraction in the SOFC To solve this problem, the electrode polarization losses at the a node should be effectively reduced. It has been well known that the microstructures of an electrode greatly influence the electrochemical properties in a SOFC. For example, the volume fraction of the each component in the electrode can modify microstructur es in an anode or cathode[62] In previous work, it was shown that tailoring different compositions of the anode functional layer changed the electrochemical performance and power density of the SOFC [64] In the work, 60wt% of NiO in NiO GDC AFL, which is NiGDC AFL after reduction, showed highest maximum power density with 1.15 W/cm2 at 650 C. Reducing the electrode ASR is the major factor for increase performance, and the ASR was found to have an optimal NiO composition (60wt%) of AFL. Moreover, the electrode ASR and AFL composition change showed an inverse linear relation to maximum power density. It was expected that at the optimal composition the number of reaction sites, that is, triple phase boundaries (TPBs) which consisted of an electronic conductor (Ni), ionic conductor (GDC), and gas diffusion path (pore) is highest. However, the evaluation of TPB density is not easily achi eved with the conventional two dimensional

PAGE 85

85 (2D) SEM analysis. Some studies were conducted for estimating TPB length or density by stereological methods using analysis of 2D SEM images [62, 65] However, accuracy of active TPB length estimation was limited due to phase interconnectivity issues, which is threedimensional (3D) property. Recently, 3D reconstruction techniques w ere employed to deal with microstructural analysis of SOFCs using FIB/SEM dual beam system [66 72] Using this technique, more accurate and realistic quantifications of the SOFC electrode microstructure has been available. Moreover, quantified microstructural features were attempted to link directly with electrochemical properties. For example, TPB density was linked to charge transfer and adsorption properties and the tortuosity was used to describe concentration polarization[68] However, most of the studies were conducted for composite cathodes and Ni YSZ a node for intermediate to high temperature (> ~700 oC) SOFC application s. In this study, the microstructural features of Ni GDC AFLs, which is widely used for intermediate to low temperature (400 ~ 700 oC) SOFCs, with different NiO contents were investigated by 3D reconstruction technique using a FIB/SEM dual beam system. After reconstruction, the comprehensive quantification for various properties of the studied samples was conducted and the values were analyzed. Finally, the active TPB density was calculat ed and link ed with electrochemical performances 5. 2 Experimental In the work four AFL samples were reconstructed and quantified, which include cells with 50, 60, 65, and 80wt% NiO AFL. As a reference, a sample without an AFL was also reconstructed. The det ailed fabrication process and electrochemical performance test of the samples were described in previous work [64] First of all, all samples were mounted in an epoxy supporter using a Struers EpoVac System. At the

PAGE 86

86 same time, this epoxy inf iltrates into the samples and fill s open pores inside porous Ni GDC anode/AFL structures, providing better contrast for SEM imaging. After the mounting process, the sample stub was grinded and polished usi ng sandpaper and exposing anodeelectrolyte interface. The automated sectioning and imaging were carried out with a FIB/SEM duel beam system (FEI Strata DB 235). In this system, th e electron beam pole (SEM) against the ion beam pole (FIB) leans at a 52o angle. ( Fig. 5 1 a ) The image at each slice was captured by SEM. To get better contrast difference among the phases, a t hroughlens detector (TLD) in backscatter mode was utilized. The FIB was used to create a trench around the region of interest (ROI). The slicing distance (z axis resolution) was 60 nm. To avoid charging effect during SEM process and protect from ion damage, protective platinum layers were deposited with an insitu liquid metal organic ion source (LMIS). This repeated imaging and slicing processes were automatically controlled using the Auto Slice and View software system (FEI Company). After collecting the cross sectional images for each sample, the alignment, segment, cropping and labeling for the three dimensional reconstruction were conducted by Amira software ResolveRTTM (ver 4.0, Mercury Computer System Inc.). Fig. 51 b shows the schematic diagram of 3D reconstruction process. Amira was also utilized to quanti fy the various microstructural features from the 3D reconstructions of the samples, such as each phase (Ni, GDC, and pore) volume and surface area, phase gradient, and tortuosity of pores.

PAGE 87

87 5. 3 Results and Discussion Fig. 5 2 shows the 3D reconstructions of the cells, which allows for direct qualitative comparisons between the anode/AFL structures. The z axis dimension reconstructed for each sample was approximated at the dista the analyzed AFL depth in this study covers almost full cross section of each AFL [64] The detailed dimention for reconstruc ted samples are summariezed in Table 5 1. Compared to the no AFL sample (Fig. 5 2 a), cells with AFLs (Fig. 5 2 b,c,d,e) show much smaller particulate structures. Among the AFLs, the change of amount of Ni (green i n 3D reconstruction) and GDC (red in 3D reconstructions) is clear as NiOwt% increases. This result is in good agreement with 2D SEM image observation discussed in the previous section. Moreover, this 3D reconstruction allows the separation of each phase, or to combine only two phases. As illustrated in Fig. 5 3, each phase of GDC, Ni, pore, and combination of Ni pore phases were reconstructed individually. From this phase separation, the quantification of the phase gradient, volume fraction, and surface a rea for each phase is possible. For the spatial distribution of each phase in AFLs, the phase gradient for each AFL sample was plotted in Fig. 5 4 It is evident in Fig. 5 4 that from b to e, the level of Ni and pore phase is getting higher, while the amount of GDC phase is decreased. In addition, the all AFLs (Fig. 5 4 b,c,d,e) have a very narrow transition zone (~ below 500nm) which is a region for significant increase of the GDC phase nearby GDC electrolyte, compared to over 2000 nm for bare anode (Fig. 5 4 a). This result indicates a very fine microstructure in the AFL and its features such as high surface area and large amount of reaction sites (TPBs) are well retained at the interface of the AFL/electrolyte. The anode without AFL lose many reaction sit es and gas

PAGE 88

88 diffusion paths due to penetration of the GDC electrolyte into large anode pores at the interfacial zone which is known to be the most important active reaction zone for the hydrogen oxidation. For Ni, GDC and pore phases, the phase volume frac tions were quantified using AmiraTM tissue statistics module. In this module, the number of voxels for each phase were counted and based on this result the numerical volume fraction for each phase is easily able to be calculated. This result can be one of the basic criteria for the practical confirmation of the credibility of the resultant 3D reconstructions. It is because there are theoretical values of Ni reduced from NiO, GDC, and pore volume fraction based on initial AFL composition, which can be calcul ated from the material properties such as Ni, NiO, and GDC densities when assuming full reduction of NiO into Ni. The theoretical and measured values of total volume fraction of each phase for different AFL are tabulated in Table 5 1 and plotted in Fig. 5 5 a. As observed in phase gradient graphs (Fig. 5 4), the Ni contents and porosity in AFL increases with increasing NiOwt%, while GDC is invers ely proportional to initial NiO wt% in AFL. It is clearly shown that the Ni and GDC volume fractions extracted fro m the 3D reconstruction are well matched with theoretical values. For example, the volume fractions of Ni were 29.1 1.0 37.8 5.9 43.2 6.7 and 51.1 6.4vol% for 50, 60, 65, and 80wt% NiO AFL samples, respectively, which are very close to the theoretical values of 30.4, 36.3, 39.2, and 47.7vol%. However, porosity is little below the theoretical value, which might be an attributable to sampling resolution. In addition to the total volume fraction, the solid phase volume fraction between GDC and Ni was also calculated (Fig. 5 5 b). As expected fr om the theoretical value, at 60wt% NiO AFL the Ni to GDC volume ratio was

PAGE 89

89 almost 1:1 with measured value of 48.1 6.7 and 51.9 6.7, at which ratio the highest performance among studied samples were shown in previous section. At this point, it is noted that these volume fraction results from the 3D reconstructions were based on around 1 5 0 slices of 2D SEM images and the graded phase plots in Fig. 5 4 showed the continuous fluctuation of each phase along the dis tance from the electrolyte. Therefore, to quantify the structural analysis of the composite SOFC anodes, the 3D reconstruction is more reliable and necessary compared to 2D SEM image analysis method. The total surface area values of the AFL structures with different NiO contents were calculated using Amira tissue statistics module. The resultant values are normalized by the total volume of the region of interest (ROI) (Table 5 2) The effective particle or pore diameters ( d ) for each AFL composition were calculated with a Brunauer EmmettTeller (BET) method using a general formula written as; S V d 6 ( 5 1) where V and S are the volume and surface area of the each phase, respectively. The Amira tissue statistics program was utilized for the calculation of the phase volume and surface area of Ni, GDC and pore in each sample. The effective diameters of Ni ( dNi) phase in 50, 60, 65 and 80 wt% NiO AFL co rrespond to 683, 771, 917, and 1120 nm, respectively, showing the expected trend. The complete data set is tabulated in T able 5 2 and plotted in Fig. 5 6. For all three phases, the linear relationship between effective diameter and AFL composition is shown, which is the proportionality for Ni and porosity, and an inverse proportionality for GDC. This result implies that the high NiO content in the AFL produce a structure with large Ni particles and very small GDC

PAGE 90

90 particles, which can cause poor connectivity of the smaller particle phase (GDC), increasing the deactivated TPB sites. In contrast, for low NiO AFL, same situation for Ni phase can occur but worse due to lower porosity, which interferes with the fast gas diffusion. Therefore, it is expected that at the medium point, around 55~60wt% NiO AFL can have optimal structure for highest TPB density. In this case, particle size control should be considered with other factors, such as composition and porosity. The tortuosity was estimated using the moment of i nertia module in the Amira software. Using this module, tracking of the center of the open pore phase through sample from the beginning of the AFL to the GDC electrolyte interface is possible, which allows us to measure the accumulated 3D Euclidian distanc e through the region. The accumulated Euclidean distance divided by AFL thickness yielded the tortuosity of the each sample (Table 5 2). For the 50, 60, 65, and 80wt% NiO AFL, the tortuosity values were 2.77, 1.91, 1.91, and 1.69, respectively. This result is in good agreement with the general theory that the higher porosity with larger pore diameter provides less complicated gas diffusion paths thorough the open pores, which means low tortuosity. This tortuosity concept can be combined with volume fraction of porosity to estimate effective diffusion coefficient, which is directly related to concentration polarization [68] Therefore, for further analysis of this property and cell performance the deconvolution of the electrochemical impedance under applied current is in progress. The TPB density for each AFL composition was quantified. The TPB site is where the three phases such as Ni, GDC, and pore meet at the same place. However, the TPB works only when these phases are properly connected. For example, the G DC phase of a TPB site should be connected to the GDC electrolyte to make a path for

PAGE 91

91 oxygen ions from anode(AFL)/ GDC electrolyte interface to the active site. At the same time, the Ni phase and pores should have percolation to the interconnect (or current collector) and fuel side (outside anode), respectively. The reasons are to bring hydrogen gas from the fuel source to TPB sites, to conduct charge transfer during hydrogen oxidation at the TPB, and to complete the extraction of electrons from the reaction site outside circuit. Therefore, counting the active TPB and removing the inactive (dead) TPB from the total TPB measurement is a critical issue [70, 72] In this study, the TPB density was calculated based on the 3D reconstruction. For 3D reconstruction, every voxel was label one of the phases including GDC, Ni, and pore. Generally one edge is shared by 4 voxels and if an edge is shared by all 3 kinds of phases, then it is counted as TPB length (Fig. 5 7). In order to estimate the actual working TPB density, the TPB sites were classified into 3 categories, which are active, inactive, and unknown TPB. For this, connectivity of each phase at the TPB was traced along xy, yz, and zx plane. If the Ni, GDC and pores are connected across the AFL from GDC electrolyte to end of the AFL, it was counted as active TPB. If one of the phases connected to the TPB site was isolated inside the reconstructed region, which was referred to inactive TPB. O ther cases are sorted as unk nown TPBs. The total TPB length (LTPB) was estimated by summation of the length of the voxel edges counted as the TPB. The TPB density ( TPB) is calculated by total TPB TPBV L ( 5 2) where the Vtotal 2 3). For unknown TPBs, it was assumed that the same connectivity of the each phase exists out of the ROI and the unknown TPBs might have the same por tion of the active

PAGE 92

92 and dead TPB in known region. Based on this assumption, the total active TPB density ( TPB, active, total) was estimated by dead TPB active TPB unknown TPB total active TPB total active TPBL L L V L, , ,1 ( 5 3) where the LTPB,active, LTPB,dead, and LTPB,u nknown are active, dead, and unknown TPB length in a measured volume, respectively. The calculated total active TPB densities were 8.5 1.4, 15.6 5.0, 14.2 3.4, and 6.3 2.3 2, respectively (Table 5 2). As expected from total surface area and effective particle size analysis discussed above, the highest TPB dens ity was achieved in AFL with 60wt% NiO, which is 1:1 volume ratio of Ni to GDC. Moreover this is in good agreement with the electrochemical impedance analysis in previous chapter. However, t he standard deviation of TPB is greater for higher values. This result might reflect that the AFL with larger TPB density has higher randomness and complicity of the structure. This scattering of the TPB density can be reduced by introducing computational simulation method [72] or morphological correction factor [71] Fig. 5 8 shows the TPB result compared to surface area. The surface area for each sample does not show any trend with AFL composition contrast to TPB density It does not seem to be in agreement with results from other studies that the higher surface area produces higher probability of TPB density as previous shown previous studies [65, 67] Howeve r, it should be considered that in this study the effective particle size and porosity was controlled by compositional change of AFL accompanying volume fraction change between phases, while other studies the phase composition of the electrodes were fixed. Fig. 5 9 a shows the TPB an d electrode ASR with various AFL compositions. For the electrode ASR, the result was taken from the previous study [64] In that study the electrode ASR measured from the AFL with

PAGE 93

93 different NiO contents were presumable assumed that the cathode ASRs were the same value due to same process of the cathode fabrication for each cell. Therefore, the electrode ASR trend can show the anodic ASR change. As shown in Fig. 5 9 a TPB density shows the inverse trend of electrode (anode) ASR change. Previously, Bieberie et. al reported that the main electrode process directly depends on the TPB length [73] In that research, using a Ni pattern anode as a model electrode, they showed that the electrode conductivity under open circuit condition is correlated to TPB length, which is linear relationship between inverse resistance and TPB length. Fig. 5 9 b shows the plotting of one over ASR corresponds to TPB density, also showing the linear relationship. However, some deviation from the linearity is shown. This can be occurred by scattering of the TPB length from the 3D reconstruction accuracy or the different electrochemical mechanism dependence of compositional change of AFL on electrochemical polarization. Previous ly Smith et. al reported that for the LSM YSZ cathode the charge transfer and adsorption have a different dependence on TPB length, which was evaluated by deconvolution of impedance[68] Therefore, further stud y about the effect of TPB density on anode polarization loss will be conducted through the deconvolution of anode impedance under applied current. 5. 4 Conclusions In this work, microstructural properties of Ni GDC anode functional layer s for IT SOFCs were quantified by stateof theart 3D reconstruction technique. For 3D reconstruction, each sample was automatically sectioned and each sectional image was acquired using FIB/SEM dual beam system. After labeling of each image to give phase separation, the seri es of sectioned images were incorporated and reconstructed in 3D utilizing the Amira software. From this 3D reconstruction, the phase gradient through the

PAGE 94

94 sample depth and volume fraction, effective diameter, surface area for each phases (Ni, GDC, and pore), and pore tortuosity were quantified. This result showed that the volume fraction was well matched with theoretical value showing each sample was well reduced. In addition to volume fraction, the graded phase plot showed that the actual AFL/ anode struct ure have some degree of deviation of the each phase volume fraction, so the accurate quantification microstructural properties could be achieved by bulk analysis As one of the most important features in the anode microstructure, the active TPB density was evaluated with the algorithm of checking the connectivity of voxels for each phase. The highest TPB length was found at the 1:1 volume ratio of the Ni to GDC in AFL. Moreover, the TPB densities showed a linear relation to the inverse of electrode ASR. For the more accurate TPB estimation, the mathematical adjustment of rectangular voxel will be needed. In addition, the detailed analysis of anode reaction mechanism by deconvolution of impedance spectrum should be conducted for the direct relation to electrochemical properties with quantified microstructural features.

PAGE 95

95 (a) Y X Z 50nm Y X Z 50nm 10 m Alignment, Cropping, Segmentation 3D reconstruction Serial sectioning -2D images (b) Figure 5 1. Schematic diagram of FIB/S EM dual beam system with sample (a) and 3D reconstruction process (b)

PAGE 96

96 (a) (b) (c) ( d) (e) Figure 5 2. 3D reconstruction of Ni GDC anode (a), and AFLs with initial composition of 50 (b), 60 (c), 65 (d), and 80 (e) wt% NiO nearby at anode(or AFL)/electrolyte interface.

PAGE 97

97 (a) (b) (c) (d) Figure 5 3. Individually reconstructed phases from the 3D reconstruction of AFL with 65 wt% NiO ; GDC (a), Ni (b), Pore (c), and combination of Ni and Pore phases

PAGE 98

98 0 2000 4000 6000 8000 10000 0 20 40 60 80 100 Phase Volume (%)Distance from electrolyte (nm) Pore GDC Ni (a) 0 2000 4000 6000 8000 10000 0 20 40 60 80 100 Distance from Electrolyte (nm)Phase Volume (%) Pore Ni GDC (b) Figure 5 4. Phase gradient of reconstructed samples with no AFL (a), 50 (b), 60 (c), 65 (d), and 80 (e) wt% NiO in NiGDC AFL

PAGE 99

99 0 2000 4000 6000 8000 10000 0 20 40 60 80 100 Pore Ni GDCPhase Volume (%)Distance from Electrolyte (nm) (c) 0 2000 4000 6000 8000 10000 0 20 40 60 80 100 Distance from Electrolyte (nm)Phase Volume (%) Pore Ni GDC (d) Figure 5 4. Continued

PAGE 100

100 0 2000 4000 6000 8000 1000 0 20 40 60 80 100 Distance from Electrolyte (nm)Phase Volume (%) Pore Ni GDC (e) Figure 5 4. Continued

PAGE 101

101 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Volume Fraction (vol %)NiO in AFL (wt %) Ni GDC Pore (a) 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Solid Volume Fraction (vol %)NiO (wt %) Ni GDC Pore (b) Figure 5 5. Volume fraction of Ni, GDC and pore phase in total volume (a), and volume fractio n of Ni and GDC in solid volume of AFLs with various compositions. Open symbols represent theoretical values.

PAGE 102

102 50 60 70 80 400 600 800 1000 1200 Diameter (nm)NiO in AFL (wt %) Ni GDC Pore Figure 5 6. Effective particle diameters of Ni (rectangular), GDC (circle), and pore (triangle) phase of AFLs with various compositions

PAGE 103

103 GDC Ni Pore TPB F igure 5 7. Schematic diagram of TPB length calculation from 3D reconstruction. A rectangular parallelepiped represents a voxel in a 3D reconstruct ion and each one is labeled as one of phases; Ni, GDC, or Pore phase. A edge which is shared by all three phases is counted as a TPB length.

PAGE 104

104 50 60 70 80 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Surface Area per Volume ( m-1) NiO (wt %)0 5 10 15 20 25 30 35 40 TPB density ( m-2) Figure 58. Plot of quantified surface area and TPB density of AFL with various NiO contents. Dotted lines are only for guide purpose.

PAGE 105

105 50 60 70 80 0 10 20 30 40 NiO in AFL (wt %)TPB Density (1/ m2)0.00 0.05 0.10 0.15 0.20 0.25 Electrode ASR ( cm2) (a) 0 5 10 15 20 25 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 1/(Electrode ASR) ( -1-cm-2)TPB Density ( m-2) 60:40 AFL 65:35 AFL 50:50 AFL 80:20 AFL (b) Figure 59. (a) TPB density and electrode ASR with vari ous AFL compositions (Dotted lines are only for guide purpose.) (b) plot of 1 over electrode ASR with TPB density. A red line represents linear fit for the plot showing inverse relationship between TPB and electrode ASR

PAGE 106

106 Table 5 1. 3D reconstruction dimension and total volume fractions of Ni, GDC and pore phase and solid volume fractions of Ni and GDC Initial AFL Composition NiO (wt%) 50 60 65 80 GDC (wt%) 50 40 35 20 3D Reconstruction Dimension X ( m) 12.95 8.75 10.45 10.42 10.22 6.06 7.95 7.88 8.90 10.64 10.76 9.32 Total Reconstructed v olume 3) 1177.91 564.21 893.91 765.26 Phase Volume After Reduction Ni in Total volume(%) Theoretical 30.4 36.3 39.2 47.7 Measured(SD) 29.1(~1.0) 37.8(~5.9) 43.2(~6.7) 5 1.1(~6.4) GDC in total volume(%) Theoretical 48.3 38.4 33.4 19.0 Measured(SD) 54.6(~3.0) 40.6(~4.8) 35(~5.5) 20.0(~5.9) Pore in total volume(%) Theoretical 21.2 25.3 27.4 33.3 Measured(SD) 16.3(~2.9) 21.6(~3.0) 21.8(~3.2) 28.9(~2.9) N i in solid volume(%) Theoretical 38.6 48.6 53.9 71.6 Measured(SD) 34.8(~1.0) 48.1(~6.7) 55.1(~7.3) 71.9(~8.0) GDC in solid volume(%) Theoretical 61.4 51.4 46.1 28.4 Measured(SD) 65.2(~1.0) 51.9(~6.7) 44.9(~7.3) 28.1(~8.0) Table 5 2. S ummary of quantification of microstructural features of AFL with various compositions Initial NiO composition in AFL wt% 50 60 65 80 Effective Ni diameter, d Ni nm 683 771 917 1120 Effective GDC diameter,d GDC nm 890 642 650 316 Effective Pore diameter,d pore nm 381 452 489 560 Surface area per volume, SA/V m1 2.4 2.9 2.4 2.6 Tortuosity, N/A 2.77 1.91 1.91 1.69 TPB density,TPB(SD) m2 8.5(~1.4) 15.6(~5.0 ) 14.2(`3.4 ) 6.3(~2.3) *Electrode ASR at 650oC -cm2 0.161 0.133 0.142 0.205

PAGE 107

107 CHAPTER 6 HIGH PERFORMANCE IT SOFC WITH CERIA/BISM UTH OXIDE BI LAYERED ELECTROLYTES FABRICATED BY A SIMPLE COLL OIDAL ROUTE USING NANO SIZED ESB POWDER 6 .1 Introduction Solid oxide fuel cells (SOFCs) have been widely accepted and studied as a next generation energy conversion device. They produces electricity by elect rochemically combining fuel and oxidant across a ceramic ionic conductor, i.e., a solid state electrolyte [2] The efficiency of SOFCs is not limited by theoretical Carnot efficiency unlike that of combustiontype systems is limited. The fuel to electrical efficiency c an reach approximately 45 to 60%. Considering utilization of the by product heat in cogeneration or bottoming cycles, the projected system efficiency can exceed 80 %. In addition to high efficiency, SOFCs are attractive because of their reduced production of SOx and NOx and significantly lower green house gas emissions compared to combustion engines [3] H igh system cost is the one of the largest barriers to the commercialization of SOFC technology. I onic conduction in the solid electrolytes is a thermally activated process, leading convent ional SOFCs to operate in a temperature range from 900 oC to 1000 oC. Consequently this high operation temperature requires the use of ceramic interconnects, high temperature seals, and supper alloy based balanceof plant components, resulting in prohibiti ve system costs [9] As Steele claimed, in lower temperature operation (500~700 oC), the system cost can be significantly reduced by allowing for the use of cheap stainless steel for the bipolar plates and the balanceof plant, combined with the use of high temperature gaskets rather than rigid glass based seals, which can also enhance its mechanical stability and lifetime [5]

PAGE 108

108 Recent research efforts aim to develop alternative SOFC materi als that operate in the 500 to 800 C range. There are two major issues for lower temperature operation of SOFCs. The first is that the reduced oxygen ion conduction in ceramic electrolytes causes a significant increase in ohmic polarization at low and int ermediate temperatures. The second is that electrode polarization is significant due to reduced cathode activity at the electrode and electrolyte interface. Therefore, development of electrolytes with high ionic conductivity in the IT range (500 ~ 700 oC) has been widely investigated by many researchers [5, 7] Among the studied electrolyte materi als, dopedceria and stabilized bismuth oxide have been reported to have 1~2 orders of magnitude of higher ionic conductivity than conventional yttria stabilized zirconia electrolytes [74] However, doped ceria shows lower open circuit potenti al (0.7~0.8 V) than the theoretical Nernst voltage due to its mixed ionic electronic conductivity (MIEC) causing electronic leakage and low power density [31] Moreover, although stabilized bismuth oxide has very high oxygen ionic conductivity its inherent thermodynamic instability under reducing conditions makes it a poor choice by itself as an electrolyte for SOFC s [37] To overcome these problems, Wachsman et al suggested a bismuth oxide/ceria bilayer electrolyte consisting of a layer of stabilized bismuth oxide on the oxidizing side and a layer of doped ceria on the reducing side [18] In this arrangement, the ceria layer can improve the thermodynamic stability of the bismuth oxide layer by shielding it from very low Po2 and the bismuth oxide layer can serve to bl ock electronic flux from doped ceria in reducing atmos pheres theoretically yielding high OCP s approaching the Nernst potential at the IT range. For over a decade, several researches have demonstrated

PAGE 109

109 that the OCP can be effectively improved using the bism uth oxide/ceria bilayered electrolyte concept [18, 3941] However, high power density at low temperature was not readily achieved due to high ohmic losses from thick electrolytesupported cell designs and the high reactivity of bismuth oxide with conventional perovskite cathodes, such as La0.6Sr0.4Co0.2Fe0.8O3(LSCF) due to weak metal oxide bonds [75] Therefore, recent research has focused on development of thin erbia stabilized bismuth oxide (ESB) / gadolinia doped ceria (GDC) bilayered electrolytes and ESB compati ble cathodes. Previously, Park and Wachsman reported that an ultra thin (~0.2 m) and dense ESB layer can be deposited on Sm doped ceria(SDC) pellet s by pulsed laser deposition (PLD) [39, 40] In addition, to overcome the reactivity of ESB with conventional perovskite cathodes, a bismuth ruthenate (Bi2Ru2O7, BRO7 ) ESB composite cathode was developed [76] A r ecent optimization study by Camaratta et al. reported that BRO7ESB composite cathodes exhibited very low ASR (0.73 and 0.03 cm2 at 500 and 700 oC, respectively) [77] Based on these studies, we demonstrated an impressively high performance of ~1.94 W/cm2 at 650 oC using a thin ESB/GDC electrolyte on an anodesupported cell with highly optimized BRO7 ESB cathodes [19, 20] To accommodate dense and thin electrolytes on porous anodesupport s, we integrated a recently developed a GDC anode functional layer (AFL) between the anode and GDC electrolyte by spreading of a precursor solution [14] In this study thin (~4 m) and relatively dense ESB was successfully deposited by PLD technique on 10 m thick GDC film. T he obtained total ASR was only 0.079 cm2, showing a ~40% decrease compared to a cell with a singl e GDC electrolyte.

PAGE 110

110 Up to now, these various studies have soundly proven that the bilayered electrolyte concept is highly encouraging in practical SOFC applications for low temperature operation. One of the key factors to fabricate ESB/GDC bilayered electr olyte s is to obtain a dense and thin ESB layer on the sintered GDC electrolyte. Although PLD can be good for demonstration purposes in lab scale experiments a more cost effective and practical fabrication process is necessary for future application of the ESB/GDC bilayered electrolyte. I n this study, we fabricated a thin and dense ESB/GDC bilayered electrolyte on anodesupported SOFCs by a simple and cost effective colloidal deposition process. To obtain a dense ESB layer, nanosized ESB particles used for the colloidal coating slurry were synthesized by a wet chemical coprecipitation method. In order to optimize the sintering conditions of the ESB layer on dense GDC, the evolution of the ESB microstructure with sintering temperature was investigated. To g auge the performance of the developed ESB/GDC electrolyte, current voltage characteristics and electrochemical impedance test s were carried out. 6 .2 Experimental Procedur e 6 2.1 ESB Powder Fabrication To synthesis very fine ESB powder, a coprecipitation r oute was employed. Pure Bi nitrate and Er nitrate were used as starting raw materials. They were weighed in stoichiometric proportions and dissolved in 70% nitric acid to produce a solution. An excess ammonia solution (Acros Organics, 28 30% of NH3 soluti on in water) was added to the stirred solution to increase the pH value to 12. The addition of the ammonia solution resulted in the formation of a yellowishbrown color precipitate. The precipitate was filtered, and then subsequently dried at 80 C for 12 h rs The

PAGE 111

111 agglomerated powder was then ground into fine particles using a mortar and pestle. The powder was then calcined at 900 C for 10 h rs in air. F o r comparison purpose, the ESB powder was synthesis by the conventional solid state route. A stoichiometr ic mixture of Bi2O3 (99.9995% pure) and Er2O3 (99.99% pure), from Alfa Aesar, were mixed and ball milled with zirconia ball media in a highdensity polyethylene bottle for 24 hrs After drying, the mixed powders of ESB were calcined at 800 C for 16 hrs. A gglomerated powders were ground using mortar and pestle and sieved using a 325 m mesh. 6 2.2 Fuel C ell F abrication The SOFC fabrication involved GDC spin coating on tapecast anodes followed by ESB colloidal drop coating. The anode support was prepared by tapecasting 65 wt% of NiO (Alfa Aesar) and 35 wt % of GDC (Rhodia) with an appropriate amount of solvents and organic compounds. The anode tapes were presintered at 900 C for 2 hrs. The GDC AFL was deposited by spraying GDC precursor solution on presintered anode surface. Subsequent heat treatment at 9 00 oC was carried out for the removal of the binder system. Detailed preparation and fabrication of GDC AFL on tapecast anode were described the previous study [14] Thin and uniform GDC electrolytes were deposited by spin coating with a GDC colloidal slurry. The Rhodia GDC powder was ball milled for 24 hrs with solsperse (dispersant) in ethanol. PVB (binder) and DBP (plasticizer) were added after the first ball milling step and the solution was ball milled for an additional 24 hrs For spin coating, the anode subs trates were fixed on the vacuum chuck of the spin coater. The t hickness of the GDC electrolyte was controlled by the number of depositions with same spin speed. After deposition, samples were dried at room temperature for 10 h rs After

PAGE 112

112 drying, the multilay er anode/electrolyte structure was sintered at 1450 oC for 4 hrs using a ramp rate of 3 oC /m in air. For the ESB/GDC bilayered electrolyte, the ESB layer was colloidally deposited by drop coating with coprecipitated ESB (cpESB) powder. To make a ESB col loidal slurry, co precipitated ESB powder was mixed with a binder system which consisted of Solsperse, DBP and PVB as the dispersant, plasticizer, and binder respectively. This mixture was ball milled in ethanol for 24 hrs and dropcoated onto the sinter ed GDC electrolyte surface. The dropcoating was repeated until a desired thickness was achieved. To see the effect of sintering temperature on formation of ESB layer on dense GDC electrolyte, as deposited ESB/GDC bilayered cell was divided into several pi eces by a diamond saw and each sample was sintered at 700, 800, and 900 oC for 4 hrs using a 400 C 1 h binder burnout step, and a 5 C/min ramp rate. For comparison, the same experiment s were repeated using solid state ESB powder. Two different composite cathodes were used for this study -LSCF GDC (50:50 wt%) on GDC electrolyte and A BRO7ESB (50:50 wt%) on ESB/GDC bilayered electrolyte. The cathode development and cathoding procedure can be found earlier work [19, 77] 6 2. 3 Characterization The phase and size of the crystallites of as calcined ESB powders were investigated by means of X ray diffraction analysis (XRD, Philips APD 3720). Microstructures of ESB powders and fuel cell structures with ESB/GDC bilayered electrolytes were observed using scanning electron microscopy (SEM, JEOL 6400 / 6335F) Qualitative elemental analysis of the fuel cell structure was conducted by e nergy dispersive X ray spectroscopy (EDX).

PAGE 113

113 For electrochemical performance measurement fue l cell samples were loaded in a fuel cell testing set up. Current voltage (I V) characteristics were conducted by a Solartron 1287 potentiostat The gas condition used was 30 sccm of dry air and 90 sccm of wet hydrogen to the cathode and anode side, respec tively. To avoid gas leakage, the edge of cell and testing alumina tube were covered with ceramic sealant using the mixture of two part ceramabond517 (liquid and power, Aremco). After the I V measurement, twopoint probe impedance analysis was carried out under open circuit condition using a Par stat 2273 (Princeton Applied Research) over a frequency range of 100 KHz to 100 mHz. 6 .3 Result and Discussion 6 .3.1 Powder C haracterization Fig 6 1 shows the XRD result of ESB powders synthesized by c o precipitation (cp) and solidstate (ss) route. For cpESB, the precursor powder was calcined at 500 oC for 4 hrs, while ss ESB powder was synthesized by calcined at 800 oC for 16 h rs. As shown in Fig. 6 1 a, both cpESB and ss ESB show a cubic fluorite structure of dopedbismuth oxide without any other phases. This result indicates that the wet chemical coprecipitation method for synthesis of ESB powder is a greatly effective way to reduce the calcination temperature and processing time compared to the conventional solid state route. Due to less thermal energy input during cubic fluorite phase formation, we can expect lowered grain growth and significantly reduced crystallite size of cpESB powder. Generally the crystallite size of the material can be calculated by Scherrer equation [78] ; cos 2 L K B ( 6 1 )

PAGE 114

114 where K is the shape factor, is the x ray wavelength, typically 1.54 B is the line broadening at half the maximum intensity (FWHM) in radians, is the Bragg angle, and L is the crystallite length. Fig 6 1 b shows the magnified x ray diffraction pattern of the (111) peak from Fig 6 1 a at the theta range from 27 to 29 degree. Using this graph, we calculated the crystallite size of the both ESB powers by eq. (1) and they are summarized in T able 6 1 The crystallite size of c p ESB was 1/ 3 the size of ssESB, indicating a high possibility of smaller particle sizes of cp ESB powder The microstructural analysis by SEM was carried out to investigate the size and morphology of the ESB powder. Fig 6 2 shows SEM images of the ESB prepared from different s ynthesis methods, that is, conventional solid state and wet chemical route using coprecipitation. As shown in Fig. 6 2 a, t he particle size of c p ESB powder is much less than 5 m and each particle consists of soft agglomeration of nanosized rodshaped particulates with high aspect ratio. In contrast the ssESB powder in Fig. 6 2 b shows particle size over ~5 m and each particle appears to be hard agglomerated. This result indicates that the coprecipitation process successfully reduced the resultant ESB particle size and significantly enlarged surface area, allowing for much higher sinterability. 6 .3. 2 Effect of S intering T emperature on ESB /GDC B ilayered E lectrolyte In order t o investigate the effect of particle size and sintering temperature on the fo rmation of an ESB layer on a GDC electrolyte, ESB/GDC bilayered electrolytes were fabricated using cpESB and ss ESB powder at various sintering temperatures ( 700, 800, and 900 oC ) The SEM images in Fig 6 3 show crosssectional views of the bilayered str uctures on anode supports fired at various temperatures. Using backscatter imagery the

PAGE 115

115 different phases ESB (white), GDC (light gray), and NiO (dark gray) are easily distinguished due to differences in elemental contrast It should be noted that the SEM i mages for each ESB layer (Fig. 6 3 a, c, e, g for ss ESB and b, d, f, h for cpESB) were taken at different magnifications. From these microstructures it is clear that both cpand ss ESB layers increase in density with increasing sintering temperature. H owever, the ESB layer which was prepared from ss ESB powder and fired at 900 oC is not fully dense. Instead, this ss ESB electrolyte looks less dense when fired at 900 oC (Fig. 6 3 g) than when fired at 800 oC (Fig. 6 3 e) due to the formation of large por es. It is known that the bismuth oxide melting temperature is ~ 825 oC [79] As Jiang and Wachsman reported, however, doped bismuth oxides have much higher melting temperature of over 2000 oC and differential thermal analysis (DTA) also found no melting endotherm for the stabilized bismuth oxide up to 1100 oC [80] Therefore, we believe that this phenomenon is possibly caus ed by the sublimation of bismuth oxide phase over 825oC. Moreover, the sintering of ESB layer is processed on a highly densified GDC substrate which was sintered at a much higher temperature (~1450 oC). In this case, the sintering mechanism of ESB powder i s primarily dependent on vertical shrinkage without lateral shrinkage. Therefore, ss ESB powder might not establish a dense layer below 800 oC due to its insufficient surface area leading low sinterability Additionally, ss ESB powder will be porous when f ired above 900 oC due to the sublimation of bismuth oxide phase. This result indicates that micronsized ESB powder prepared by conventional solid state synthesis is not appropriate for colloidal deposition of the ESB layer in ESB/GDC bilayered electrolytes.

PAGE 116

116 On the other hand, as shown in Fig. 6 3 b, d, f, h, when cp ESB powder was used, the ESB layer was highly dense when fired at 800 oC. SEM powder analysis (Fig. 6 3 a), of the cpESB powder shows that it has a very high surface area due to the nanosized nature of the partic les Therefore, it is possible for cpESB powder to produce sufficient surface energy at lower sintering temperatures for the particles to neck and densify by surface and lattice diffusion. Unlike the layer prepared from ss ESB powder the cp ESB layer sintered at 900 oC (Fig. 6 3 h) looks very dense. However the ESB thickness was reduced compared to the same layer sintered at 800 oC (Fig. 6 3 f) Since 900 oC is above the pure bismuth oxide melting temperature, as discussed above, thi s indicates that partial sublimation of bismuth oxide phase or penetration of bismuth oxide into GDC layer have possibly occurred. It should be noted that as seen in Fig 6 3 g, h, both ss and cpESB layers sintered at 900 oC show diffusion of ESB phase i nto ceria layer and segregation of it into GDC grain boundaries. This is also clearly shown in Fig. 6 4 which is a magnified image of Fig. 6 3 g. It is considered that the over melting temperature of pure bismuth oxide (~825 oC) the partial pressure or fugacity of bismuth oxide might be high, which can cause higher activity of bismuth oxide or bismuth in ESB lattice. Therefore, this high activity can accelerate motion of bismuth oxide phase, leading diffusion of bismuth oxide phase along the GDC gr ain b oundaries which generally has a high surface energy. It has also been reported that bismuth oxide can be soluble in CeO2 forming a solid solution of Bi1xCexO2xO2x/2 with a cubic fluorite structure [81] Gil et al recently reported that the solid solubility limit of bismuth oxide in doped ceria is ~ 0.8wt% [82] Park and Wachsman predicted the possible existence of solid solutioning at the

PAGE 117

117 interface between the stabilized bismuth oxide and doped ceria bilayered electrolytes [40] In this study, the conductivity of 2 mm thick Sm doped ceria (SDC) was lower than that of ESB (~ 0.2 m)/SDC (~ 2 mm) bilayered electrolyte due to the high grain boundary conductivity of ESB/SDC. This phenomenon was explained as the formation of a solid solution of ESB into the grain boundary of SDC, which can lower the activation energy for oxygen ion transfer in SDC grain boundar ies by a scavenging effect on impurity phases. Therefore, we can expect a positive effect on lowering ohmic polarization in the observed integrated ESB/GDC structure. On the other hand, i t is also shown that the fracture mode of GDC electrolyte with ESB sintered at 900 oC changed from inter granular to intragranular (Fig. 6 4), compared to GDC cross sections near ESB layers sintered at lower temperature (Fig. 6 3 a ~ f). This might cause deteriorat ion in the mechanical strength of the GDC electrolyte due to highly segregated bismuth oxide phase along the GDC grain boundary. Therefore, this phenomenon should be carefully controlled and further investigation will be needed. Consequently, we could obtain dense ESB/GDC bilayered electrolytes by a simple colloidal deposition using fine ESB powders made by a co precipitat ion method. The optimal sintering temperature for synthesizing dense ESB layers by colloidal deposition was ~800 oC, which lim ited bismuth oxide sublimation and penetration into the GDC electrolyte. 6 .3. 3 Microstructure of a F ull Button Cell with ESB/GDC B ilayered E lectrolyte For comparison of electrochemical performance, two kinds of button cells were prepared a single GDC el ectrolyte cell and an ESB/GDC bilayered electrolyte cell based on above study. To obtain similar electrolyte thickness, the colloidal deposition

PAGE 118

118 process for each layer was repeated. For microstructural analysis before electrochemical performance, another button cell with ESB/GDC bilayered electrolyte was fabricated with the similar process. An ESB layer was deposited using colloidal slurry containing cp ESB powder and sintered at 800 oC for 4 hours. Fig. 6 5 shows the crosssectional SEM image of NiO GDC an ode/ ESB GDC bilayered electrolyte/ BRO7 ESB composite cathode. As shown in this figure, a dense ESB/GDC layer was obtained and every layer is clearly distinguishable with good interfacial contact. To examine the possible interdiffusion between layers, an energy dispersed x ray (EDX) line scan analysis was conducted along the base line (yellow) in Fig. 6 5. Three elements were traced along the line, which includes Bi (red line), Ce (blue line), and Ni (white line). The resultant data are overlapped in Fig. 6 5 and no solid state reaction or interdiffusion between layers was observed. Fig. 6 6 shows the m icrostructures of cross section and surface of anode supported cells with a single GDC electrolyte and an ESB/GDC electrolyte after electrochemical performance test. Both cross section and surface view of the GDC layer (Fig. 6 6 a,c) shows very high density without pores. In contrast, the ESB surface (Fig. 6 6 d) shows some porosity and lower density compared to the GDC surface, while the crosssection of ESB (Fig. 6 6 b) looks dense. For further densification, it needs to be opmisized. As shown in Fig. 6 6, each electrolyte thickness was measured from the SEM image. T h e GDC electrolyte thickness was identical ~9 m for each cells. The ESB electrolyte thickness was ~4 m. 6. 3.4 Performance of a B utton C ell with ESB/GDC Bi layered E lectrolyte Fig. 6 7 a shows the I V characteristics of cells with GDC single layer and ESB/GDC bilayered electrolyte at 650 oC. The meas urement was conducted under 90

PAGE 119

119 sccm of 3 % wet hydrogen to anode side and 30 sccm of dry air to cathode side. The maximum power density (MPD) of the ESB/GDC electrolyte reached ~1.47 W/cm2 with a 69% increase in MPD compared to a GDC single layer (~ 0.87 W/ cm2). The OCPs of the single GDC electrolyte cell and ESB/GDC bilayered electrolyte cell were 0.75 and 0.80 V, respectively. This OCP improvement with bilayered electrolyte can be evidence that the ESB layer in ESB/GDC bilayered electrolyte may effectively block the electronic conduction from the GDC due to reduction of Ce4+ to Ce3+ at low Po2 region. Moreover, we previously reported similar OCP enhancement with a thin ESB/GDC bilyaered electrolyte. At that time we fabricated a dense ESB layer of ~ 4 m by PLD technique on ~ 10 m thick GDC electrolyte by spray coating method, resulting OCP increase from 0.72 to 0.77 V [20] It is noted that in this study we obtained higher OCP for both single and bilayered electrolyte cells compared to previous research results. We believe that this is caused by higher GDC density due to spin coating method instead spraying coating for GDC electrolyte in the previous study. Therefore, this result demonstrates that thin and dense ESB/GDC bilayered electrolytes can be achieved by conventional colloidal process by control of initial powder morphology and sintering condition. Although the ~ 0.05 V increase in OCP was achieved in a dense bilayered electrolyte, the major contribution of this outstanding enhancement in MPD came from ASR drop. The es timated total ASRs from IV curves near OCP region were 0.084 cm2 and 0.164 cm2 for cells with ESB/GDC bilayer and GDC single layer, respectively. For the further analysis electrochemical impedance test was conducted and the result is shown in Fig. 6 7 b In these N yquist plots the total and ohmic ASR were extracte d from the low and high frequency intercepts of the impedance spectra with the

PAGE 120

120 real axis, respectively and the electrode ASR was calculated from the difference between these two ASR values. The detailed values are tabulated in Table 6 2 The total ASR of ESB/GDC bilayered cell (0. 088 2) was decreased by 4 7 6 % compared to that of cell with GDC single layer (0. 1 68 2). In addition, in this study the total ASRs from this AC impedance tests are identical to the ASRs from the I V curves (DC impedances) within 5% deviation showing the good validity of the results. T he electrode ASR of the bilayered electrolyte cell was decreased by 62.6% compared to that of GDC single cell This ASR drop (0.062 2) is expected to mostly come from the cathodic polarization reduction at the interface between high catalytic BRO7ESB cathode and ESB electrolyte, considering the both cells utilized the same anode material using identical procedures This result has been also well supported by the former studies [19, 77] The ohmic ASR was also decreased ( 2 6.1%, 0.0 18 2) at 650 oC. Previously, it was shown that the ohmic ASR ( ASRelectrolyte) of a bilyared electrolyte can be estimated by a simple modeling for two series resistors written as GDC ESB GDC e electrolytL ASR 1 ( 6 2) where LGDC is thickness of GDC electrolyte, is thickness ration of ESB to GDC, and ESB and GDC are the conductivit ies of ESB and GDC, respectively [19, 42] Based on known ESB and GDC conductivit ies [83] a similar or slightly higher ohmic ASR of the ESB (~4 m)/GDC (~9 m) bilayered electrolyte than that of ~a single 9 m thick GDC can be expected by e q (2) du e to much higher conductivity of ESB than GDC at IT ranges. However, previous studies repeatedly showed that bilayer electrolyte can effectively reduce the electrolyte ASR even when we used same GDC thickness with a single GDC layer and an additional ESB l ayer was deposited [19, 20, 39] Therefore, we

PAGE 121

121 believe that this phenomenon is related to penetration and segregation of bismuth oxide phase at the GDC grain boundary, even though it was not defective in EDX analysi s. Further discussion about ASR reduction of ESB/GDC bilayered electrolyte was exposited elsewhere [19, 20] However, this MPD result was somewhat lower compared to the previous highest MPD with thin ESB by PLD an d GDC bilayered electrolyte cell (~1.95 W/cm2) [20] As ohmic ASR results suggested, the density of cpESB layer might be still less dense compared to PLD ESB. Moreover, in this study, we used both solid state BRO7 and ESB powder for the cathode, while in the previous stud y microstructurally optimized BRO7 ESB cathode was used. Optimization of BRO7ESB was described in detail by Camaratta el. a l [77] Therefore, we can expect even higher MPD result by further improvement of ESB density and cathode performance. 6. 4 Conclusions We fabricated thin and dense ESB/GDC bilayered electrolytes by a simple colloidal deposition process. For producing high ly sinterable ESB particle s, a wet chemical co precipitation route was introduced. Using this nanoscale ESB powder with high surface area, a dense and very thin (~ 4 m) ESB layer was established on a GDC electrolyte by simple colloidal drop coating. The optimal sintering temperature of the ESB layer was found to be ~800 oC due to its poor densification at lower temperatures and evaporation at higher temperatures Finally, an anodesupported SOFC with the thin and dense ESB/GDC bilayered electrolyte coupled to a BRO7 ESB cathode produced a very high maximum power density of ~ 1.5 W/cm2 at 650 oC This was possible due to the effect of the bilayered electrolyte on signi ficantly decreasing eletrode ASR as well as ohmic ASR. This study demonstrated that the high performance of

PAGE 122

122 ESB/GDC bilayered electrolyte can be reproducible by the cost effective and practical colloidal deposition. Based on this study, fabrication of large scale planar cell s with bilayered electrolytes for stack cell application is underway

PAGE 123

123 20 30 40 50 60 70 80 (420) (331) (400) (222) (311) (220) (200) (111) 800oC-16h Coprecipitation Solid state route 500oC-4h Intensity (arbitrary unit) 2 angle (111) (200) (220) (311) (222) (400) (331) (420) 800oC -16h 500oC -4h Coprecipitation Solid state (a) 27 28 29 Solid state (800oC for 16h) (111) Intensity (arb. unit)2 (angle) Coprecipitation (500oC for 4h) 2 angle Intensity (arbitrary unit) (111) Coprecipitation (500oC 4h) Solid State (800oC 16h) (b) Figure 6 1. XRD diffraction pattern of ESB powders synthesized by coprecipitation route (red line) and solid state route (black line) (a). The magnified XRD diffraction pattern of the (111) peak is shown at the 2 range from 27 to 29o (b).

PAGE 124

124 5 m (a) 5 m (b) Figure 6 2. SEM of ESB powders synthesized by wet chemical co precipitation method (a) and solid state route (b).

PAGE 125

125 Figure 6 3. Evolution of ESB layers on GDC electrolyte at various sintering temperatures usi ng ss ESB (a,c,e,g) and cpESB (b,d,f,h). It is noted that the magnification of images for ESB electrolyte using cpESB powder is higher that that of ss ESB powder.

PAGE 126

126 5 mESB GDC Ni Figure 6 4. Cross sectional view of GDC electrolyte under ss ESB layer after sintering at 900 oC, which is the magnified image from Fig. 53g. In backscattering mode, ESB (white), GDC (light gray), and NiO (dark gray) phases are well distinguishable, showing ESB penetration into GDC grain boundaries. GDC ESBBilayered Electrolyte BRO7 ESB Cathode NiOGDC Anode Bi Ce Ni 20 m Figure 6 5. Cross sectional SEM image of a full button cell with ESB/GDC bilayered electrolyte. EDX line scan was conducted along the straight base line (yellow) and the intensity of each elements are presented as red (Bi), blue (Ce), and white (Ni) lines

PAGE 127

127 20 m LSCF GDC GDC NiO GDC 9 m ( a ) 9 m 4 m 20 m GDC NiO GDC BRO7 ESB ESB (b) Figure 6 6. SEM im age of cross sectional view of a single GDC electrolyte cell (a) and ESB/GDC bilayered cell (b). Surface views of GDC electrolyte (c) and ESB electrolyte (d) are shown.

PAGE 128

128 20 m ( c) 20 m 20 m (b) Figure 6 6. Continued

PAGE 129

129 0 1 2 3 4 5 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Current Density ( cm2)Potential (V)0.0 0.5 1.0 1.5 Power Density (W/cm2) ESB/GDC GDC (a) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.00 0.02 0.04 0.06 -Z" ( cm2)Z' ( cm2) ESB/GDC GDC (b) Figure 6 7. IV C haracteristics (a) and i mpedance spectra (b) of ESB/GDC bilayer (red square) and GDC single layer (blue triangle) cell

PAGE 130

130 Table 6 1. Calcin ation condition and crystallite size of ESB powders synthesis by co precipitation and solid state route. Synthesis Ca lcine condition FWHM (o2 ) Crystallite size () Co precipitation 500oC for 4 h 0.48 171 Solid State 800oC for 16h 0.16 512 Table 6 2. Comparison of specification and electrochemical performance of the studied cells. Cell Type GDC thickness ESB thickness OCP MPD Total ASRIV Total ASREIS Oh m ic ASR Electrode ASR Ref V W/cm2 2 2 2 2 GDC 9 0.75 0.87 0.164 0.168 0.069 0.099 this study cp ESB/GDC 9 4 0.80 1.47 0.084 0.088 0.051 0.037 this study PLD ESB/GDC 10 4 0.77 1.94 0.075 0.079 0.046 0.033 [20]

PAGE 131

131 CHA PTER 7 HIGH PERFORMANCE LSM BASED CATHODE BOOSTED BY STABILIZE D BISMUTH OXIDE FOR LO W TO INTERMEDIATE TEMPERATURE SOFCS 7 .1 Introduction La1xSrxMnO3(LSM) based cathodes are some of the most widely used cathodes for high temperature (over 800 oC) solid oxide fuel cells (SOFCs) due to their high thermal and chemical stability, electrical conductivity, as well as their catalytic activity for oxygen reduction. In addition, LSM based cathodes exhibit good compatibility with conventional electrolyte materials, such as stabilized zirconia and doped ceria [2, 8486] However, despite these advantages, LSM based cat hodes are not a popular choice for reduced temperature SOFCs. This is due to the fact that LSM has negligible ionic conductivity and a high activation energy for oxygen reduction due to slow oxygen incorporation reaction into the solid lattice, leading to significantly deteriorated electrochemical catalytic effect at low temperatures [7, 63, 87, 88] To overcome this problem, dual phase composite cathode systems have been explored. Conventional ionic conducting phas es such as yttria stabilized zirconia (YSZ) and godoliadoped ceria (GDC) have been mixed with LSM leading to significantly enhanced electrochemical performance [84, 8994] For instance, Murray and Barnett reported that the area specific resistance (ASR) of pure LSM was lowered cm2 to cm2 for LSM cm2 for LSM GDC [90] The improved performance is a result of the increased number of three phase boundary (TPB) reaction sites between gas, electronic and ionic conducting phases, and by the addition of a pathway for ionic species to be transported thru the cathode. Thus it is expected that dual phase cathodes which incorporate materials with an ionic conductivity will yield further improvements in performance.

PAGE 132

132 Stabilized bismuth oxide has been reported to have one of the highest known ionic conductivities one or two orders of magnitude higher than the conventional material YSZ at intermediate temperatures (IT) [17, 95] In additi on to its high ionic conductivity, bismuthoxide enhances surface oxygen exchange rate as well as charge transfer and oxygen dissociation, which is believed to be the rate limiting step in the oxygen reduction reaction at the cathode [96100] Therefore, stabilized bismuth oxide is a good candidate for the ionic conducting phase in composite cathodes For example, the addition of yttria or erbia stabilized bismuth oxide (YSB or ESB) into a metallic cathode silver (Ag ) improve d the performance of the cathode both by increasing TPBs and forming an ionic conduction path to the electrolyte [101103] However, the long term stability of Ag based cathodes is poor [103] Recently, we developed a bismuthruthenate Bi2Ru2O7 (BRO7) ESB composite cathode [76, 77] Using this cathode on an ESB/GDC bilayered electrolyte, we demonstrated exceptionally high power density ( ~ 1.94 W/cm2 at 650 oC ) [19, 20] It was suggested in the above study that cathode performance can be promoted not only by the addition of ESB into the cathode but also by using an ESB electrolyte at the cathode interface due to effect of ESB on improvement of oxygen dissociation followed by reduction of cathode polarization losses Recently, in order to overcome limitations of LSM based cathodes, the utilization of stabilized bismuth oxides as the ionic conducting phase has received much attention [104113] Jiang et al. reported the performance of LSM YSB composite cathodes, showing an interfacial resistance of ~1.0 cm2 at 600 oC, which is 11 times lower than reported for LSM YSZ cathodes [106] Moreover, various fabrication techniques we re

PAGE 133

133 used in order to obtain increased TPB lengths by microstructural evolution [106, 107, 112] Although these efforts progressively reduced the cathode ASR, the reported power densities of SOFCs utilizing LSM bismut h ox ide cathodes are still relatively low at IT (~600 mW/cm2 at 650oC ) [107] To date, most of the studies of LSM bismuth oxide composite cathodes have been conducted on zirconiabased electrolytes which exhibit significant ohmic resistance at intermediate temperatures due to its high activation energy for ionic conduction. Even though ultra thin YSZ electrolytes successfully reduce total ohmic polariz ation losses at reduced temperature [8] significant deterioration of ionic conduction and oxygen dissociation rate at the cathode/electrolyte interface is inevitable due to the abrupt change in ionic conduction phase from bismuth oxide in the cathode bulk to the zirconia electrolyte. Therefore, modific ation of the interface between the LSM bismuth oxide cathode and the electrolyte is needed in order to achieve high performance at lower temperatures In this study, to overcome this, we introduced an (Er2O3)0.20(Bi2O3)0.80 (ESB) electrolyte layer coupled to the (La0.80Sr0.20)MnO3(LSM)ESB composite cathode. To gauge the effect of the ESB electrolyte on the cathode performance, the interfacial resistance of the cathode on ESB and on GDC was measured on symmetric cells. The actual performance improvement of the LSM ESB cathode on an ESB electrolyte layer was evaluated on anodesupported cells by I V characterization. 7.2 Experimental 7.2.1 Sample F abrication The LSM ESB composite electrode was prepared by simple powder mixing. Co mmercial LSM powder with a surface area of 5.6 m2/g was purchased from Fuel Cell Materials. The ESB powder for the cathode was synthesis by the conventional solid-

PAGE 134

134 state method. A stoichiometric mixture of Bi2O3 (99.9995% pure), and Er2O3 (99.99% pure), from Alfa Aesar, were mixed and ball milled with zirco nia ball media in a highdensity polyethylene bottle for 24 hrs After drying, the mixed powders of ESB were calcined at 800 C for 16 hrs Agglomerated powders were ground using a mortar and pestle and sieved using a 325 m mesh to get uniform particle si zes. For the LSM ESB electrode ink LSM powder and ESB powder of the same weight ratio (50:50wt%) were mixed with a binder system, which consists of alpha terpineol (Alfa Aesar), Di n butyl phthalate ( DBP) and ethanol as binder, plasticizer and solvent, re spectively To make symmetric cells, we prepared both ESB and GDC electrolyte pellets by uniaxial pressing and sintering at 890 C for 16 hrs and 1500 C for 10 hr s, respectively. Once an appropriate viscosity was reached, the electrode slurry was applied to both sides of the electrolyte substrates by brush painting. After drying the symmetric cells at 120 oC for 1 h rs a second coat of electrode slurry was applied to the electrolyte substrates. The doubly coated cells were then sintered at 800 oC for 2 hrs in air. After sintering the electrode, silver mesh current collectors and platinum lead wires were pressed against the samples in a quartz reactor using a ceramic screw andbolt assembly. For full button cell fabrication, the anodesupported cell structur es with thin electrolytes were employed. NiO YSZ (65:35 wt%) anodesupports and anode functional layers (AFLs) consisting of NiO GDC (65:35wt%) were tapecasted and attached to each other by uniaxial press on heating substrate and presintered at 900 oC. Thi n and uniform GDC electrolytes were prepared by spin coating. In order to investigate the effect of the ESB electrolyte on LSM ESB cathode, ESB/GDC bilayered electrolyte was prepared. To fabricate a thin and dense ESB layer on a GDC electrolyte, nanosized

PAGE 135

135 ESB powder was synthesized by wet chemical coprecipitation method. Using a colloidal solution containing nanosized ESB particles, the ESB layer was deposited by spin coating. A detailed fabrication and discussion of ESB/GDC bilayered electrolytes was des cribed in the previous work [19] The multilayer of ESB/GDC/AFL/NiO GDC anode structures is co sintered at 1450 oC at 4 hrs. After sintering, the LSM ESB cathode was applied in two coats on both ESB and GDC electrolyte surfaces by brush painting and sintered at 800 oC for 2hrs. 7.2.2 Characterization The phase identification of LSM, ESB, and LSM ESB composite cathodes were investigated by means of X ray diffraction analysis (XRD, Philips APD 3720). Microstructures of fuel cells with LSM ESB cathodes were obtaining by scanning electron microscopy (SEM, JEOL 6400 / 6335F). The interfacial polarization resistance of the LSM ESB electrodes on symmetric cells was conducted through two point probe electrochemical impedance spectroscopy (EIS) using a Solartron 1260. The measurement condition of impedance was an AC voltage amplitude of 50 mV over the frequency range of 0.1 MHz to 0.1 Hz in air. The frequency response analyzer was used in standalone mode and interfaced to a computer using Zplot software. Measurements were made from 500 to 700 oC with an interval of 50 oC. For electrochemical performance of the button cells, samples were loaded in the testing setup and sealed with cerambond sealant system. Current voltage (I V) characterist ics were conducted by a Solartron 1407E under 90 sccm of dry air and 3% wet hydrogen to the cathode and anode side, respectively.

PAGE 136

136 7.3 Result and Discussion 7.3.1 Impedance S pectroscopy for S ymmetric C ells In order to confirm the compatibility of ESB with LSM, a LSM and ESB powder mixture (50 to 50wt%) was annealed at 900oC for 50 hrs. Fig. 7 1 shows the resultant X ray diffraction (XRD) patterns of LSM, ESB powder and mixtures of both powders before and after heat treatment. After annealing, the XRD patter n of LSM and ESB mixture contains only the peaks of cubic fluorite structure from ESB and perovskite structure from LSM without any other peaks. This result indicates no inter phase formation, suggesting that LSM and ESB are chemically stable even at 900 oC. This result of compatibility between LSM and ESB is in good agreement with previously reported results [106, 108, 113] In contrast, it has been known that the pervoskite La1xSrxCo1 yFeyO3 (LSCF) based cathode, which is widely used as the conventional cathode for IT SOFCs, is highly reactive with ESB electrolyte due to weak metal oxygen bond of bismuth oxide [19, 77] Fig. 7 2 shows a cross sectional view of the symmetric cells with LSM ESB cathodes on ESB and GDC electrolyte pellets. In both cases, the composite cathodes show good adhesion to electrolyte substrates at the cathode/electrolyte interface. In order to distinguish each phase, backscatter ed SEM images were also observed, which are shown as insets in Fig. 7 2. It is clearly shown that relatively large ESB particles are welldistributed in a very fine LSM particle matrix. However, the majority in LSM ESB cathode seems to be LSM particles, ev en though we mixed 50: 50wt % of LSM to ESB powders. It is noted that in this study the ESB powder in the composite cathode is large As many studies reported, microstructural

PAGE 137

137 optimization, such as particle size and spatial distribution can modify the connectivity and increase the surface area of the composite cathodes to produce more active TPB sites. Therefore, at this point, the per formance of LSM ESB composite cathode might not be optimal. Fig. 7 3 is the resultant impedance spectra measured under open circuit condition from 500 to 700 oC with 50 oC interval for LSM ESB/ESB/LSM ESB and LSM ESB/GDC/LSM ESB symmetric cells. For direct comparison of electrode ASR on the different electrolytes, all of the impedance spectra have been ohmic resistancecorrected. That is, the high frequency intercept at the real axis of each spectrum, which corresponds to bulk electrolyte, electrodesheet, and leadcontact resistance, has been subtracted from each data point. The electrode ASRs of the LSM ESB cathode on GDC at 500, 550, 600, 650, and 700 oC were 10.56, 3.31, 1.11, 0.44, and 0.19 cm2, respectively. This result is very comparable with the ASRs of the LSM YSB cathode on Sm doped ceria (SDC) electrolytes reported by Li et al. In their study, the electrode cm2 at 600 and 700 oC, respectively. This suggests that the performance of the LSM ESB cathode in this study is reasonable. It should be noted, however, that despite the fact that ESB has higher ionic conductivity than YSB [114] the ASR of LSM ESB on GDC in the present study is slightly higher than for LSM YSB on SDC. This might be due to microstructural features associated with infiltrated nanoscale YSB particles into LSM backbones, while in this study, LSM ESB cathode was fabricated by a simple mechanical mixing method with micronsized ESB powders. As mentioned above, further improv ements in electrochemical performance of this LSM ESB system can be expected by microstructural optimization. On the other hand, the

PAGE 138

138 electrode ASRs of the LSM ESB on ESB substrate were substantially lower --4.18, 1.29, 0.43, 0.19, and 0.08 at 500, 550, 600, 650, and 700 oC, respectively. In Fig. 7 4, the electrode ASR values from the impedance spectra in Fig. 7 3 are plotted with various temperatures (right Y axis). In addition, the percent electrode ASR reduction on ESB relative to that on GDC was calculated at each temperature (left Y axis in Fig. 7 4), and a ~ 60% reduction in ASR was retained for all testing temperatures. This result indicates that the reduced interfacial resistance of LSM ESB cathode on the ESB electrolyte remained in effect throughout t he IT range. We believe that this dramatic electrode ASR drop using the ESB electrolyte is caused by the high oxygen dissociation rate of the bismuth oxide and the continuous fast ionic conduction pathways from the cathode to the electrolyte without an abr upt change at the LSM ESB cathode and ESB electrolyte interface. As illustrated in Fig. 7 5, the measured cathodic resistances of LSM ESB on both ESB and GDC electrolyte are compared to that of LSM GDC and LSM bismuth oxide composite cathodes on various el ectrolytes in recent literature studies. As expected, all LSM bismuth oxide cathodes including the results in this study showed lower electrode polarization losses than those of LSM GDC cathodes which were reported by Murray et al [90] This result indicates that bismuth oxide phase integrated into LSM basedcathodes increases the effective TPB length due to its higher ionic conduction. However, even though doped ceria such as GDC and SDC has higher ionic conductivity than YSZ (or ScSZ) electrolytes at IT, the reported LSM bismuth oxide cathodes (and the LSM ESB cathode in this study) on GDC or SDC show a similar level of electrode ASRs compared to cathodes using zirconiabased electrolytes, while LSM GDC

PAGE 139

139 cathodes have shown lower ASR on GDC electrolytes compared to YSZ electrolytes. It is considered that due to the superior ionic conductivity of ESB or YSB in the composite cathode, the effect of the GDC electrolyte at the cathode/electrolyte interface against the YSZ electrolyte can be negligible. Instead, the performance of the LSM bismuth oxide cathode on GDC or YSZ seems to depend on bismuth oxide phase fraction and cathode microstructures as shown in previous studies. On the other hand, it is clearly shown that the LSM ESB cathode on an ESB electrolyte exhibits significantly lower electrode polarization losses than that of any other LSM based cathode in all measured temperature ranges from 500 to 700 oC. This outstanding result can be explained that the synergetic effect of the ESB phases both in cathode bulk and at the electrolyte/cathode interface on enhancing oxygen dissociation rate and accelerating ionic conduction leading to a dramatic reduction of activation polarization for oxygen reduction in the cathode. Previously we reported a similar effect of the ESB electrolyte on reducing electrode ASR for BRO7ESB composite cathode on ESB, showing 26% reduction of the ASR relative to that on GDC electrolyte [19] Therefore, this study again demonstrates that the ESB electrolyte has a highly significant effect on boosting the SOFC cathode reaction rate at the IT range. Meanwhile, the activation energy of LSM ESB cathodes on GDC and ESB was estimated from ASR plots in Fig. 7 5 using an Arrhenius relationship. The calculated activation energies for both cells were the same, ~1.24 eV, which is in good agreement with low end of other reported values for LSM bismuth oxide cathodes (1.23 ~ 1.5 eV) which depend on microstructure and composition of the composite cathodes [104, 106, 108, 112] In addition, it is been reported that the activation energy for the pure LSM on

PAGE 140

140 SDC electrolyte is ~ 1.5 eV [106] This result shows that the addition of ESB phase into the cathode bulk or at the electrode/electrolyte interface might change the mechanism of the oxygen reduction at cathode. In addition, this estimated activation energy implies that the surface oxygen exchange reaction of which activation energy is known to ~ 1.25 eV at low temperature might be a ratelimiting step for the cathode reaction, providing a good reason for use of ESB with high oxygen surface exchange rate to reduce the electrode polarization [45] In order to gauge the longterm stability of the LSM ESB cathode, the electrode ASR of the LSM ESB cathode on an ESB electrolyte pellet was measured at 700 oC. Fig. 7 6 is the resultant plot of the electrode ASR at 700oC for 100 hrs. The ASR of the LSM ESB cathode maintained a constant value for 100 hrs cm2, indicating no initial degradation in electrode performance. To verify the stability of the LSM ESB system for IT SOFCs, however, further long term testing in the lower temperature ranges and under various applied current conditions s hould be carried out. 7.3.2 I V C haracterization for B utton C ells To further investigate the effect of LSM ESB cathodes coupled to an ESB electrolyte on the actual ITSOFC performance, current voltage measurements were conducted on anodesupported button c ells. In order to obtain an ESB electrolyte, an ESBGDC bilayered electrolyte cell design was utilized due to the thermodynamic instability of the ESB electrolyte at low Po2 conditions [16, 18] Further details regarding the ESBGDC bilayered electrolyte are available in the recent studies [19] In this study two cells were fabricated, which included LSM ESB (cathode) / GDC (electrolyte) /Ni GDC and LSM ESB/ESBGDC/Ni GDC.

PAGE 141

141 Fig. 7 7 shows SEM images of the two cells after testing. As shown in the cross sectional views (Fig. 7 7 a,b), the structure of two cells are identical except a thin ( ~ 3 is observed in ESB/GDC bilayer cell between the LSM ESB cathode and GDC electr olyte (Fig. 7 7 b). Fig. 7 7 c,d show the surface views of the GDC and ESB electrolyte for the GDC single electrolyte cell and GDC/ESB bilayered electrolyte cell, respectively, and some cracks and pores can be observed. This indicates that the densities of both electrolytes were not optimal. This issue will be discussed later in this section. The current voltage measurement results of both cells at 650oC are plotted in Fig. 7 8 a. The maximum power densities (MPDs) obtained were 658 and 836 mW/cm2 for LSM E SB on GDC and LSM ESB on ESB/GDC, respectively. Even though the OCP of the cell with LSM ESB on ESB/GDC is slightly lower than that of LSM ESB on GDC, the MPD of LSM ESB on ESB/GDC was increased by 27% compared to that of GDC single layer cell. The increased power density is due to the significant reduction in the total ASR calculated from the IV curves in Fig. 7 8 a, which were reduced by ~ 54% cm2 for cell cm2 for LSM ESB on ESB/GDC. In this study we used the same fabrication procedure and confirmed identical microstructures for the two cells with the exception of the ESB interlayer in ESB/GDC bilayer structure as sh own in Fig. 7 7 a,b. Thus, we believe that MPD enhancement of the LSM ESB on ESB/GDC cell mostly came from the effect of ESB electrolyte on reducing the electrode (cathodic) ASR at the LSM ESB cathode/ ESB electrolyte interface. For further analysis, two points probe impedance tests were conducted. Fig. 7 8 b illustrates the resultant nyquist plots for each cell under open circuit condition with the

PAGE 142

142 same gas condition for I V test at 650 oC. From the high frequency intercepts (RH) and low frequency intercepts (RL) at the real axis of the complex plane, the ohmic and total ASR values were estimated, respectively. The electrode ASR including anode and cathode polarization resistance was calculated using the equation written as; H L electrodeR R ASR ( 7 1) The detailed values are tabulated in Table 7 1. In this study, the total ASR values obtained using DC measurement from IV (ASRIV) and electrochemical impedance (ASREIS) showed less t han 1% deviation indicating high reliability of the resultant data. As expected from symmetric cell measurements, the electrode ASR was significantly reduced for the bilayered electrolyte cm2), showing ~ 64% reduction compared to single layer cm2). This result is quite consistent with the cathodic ASR change for the symmetric cell measurement in the previous section. These I V and impedance results demonstrate that the effect of the ESB electrolyte on LSM ESB cathode polarization losses directly influences the SOFCs performance at IT, showing reasonably high maximum power density of over 850 mW/cm2 at 650 oC. At this point it is noted that the GDC electrolyte densities in the testing cells were not high and have many small pores and the ESB layer also showed poor density affected by GDC layer as shown in Fig. 7 7. It is well known that the density and thickness of the electrolyte strongly influence ohmic resistance and SOFC performance [14] Therefore, the MPD of the LSM ESB cathode can be much higher if denser and thinner GDC and ESB ele ctrolytes are used. Fig. 7 9 shows IV curves of both cells at various temperature ranging from 450 to 650 oC. The MPDs of the cell with LSM ESB on GDC were 25, 72, 182, 389, and 658

PAGE 143

143 mW/cm2 at 450, 500, 550, 600, and 650 oC, respectively. In case of the bil ayered cell, higher MPDs were exhibited and are reported to be 50, 124, 275, 533, and 836 mW/cm2 at 450, 500, 550, 600, and 650 oC, respectively. The improvement in MPD from the cell with LSM ESB on GDC to the cell with LSM ESB on ESB at each temperature w as calculated and plotted with actual MPD values for each cell in Fig. 7 10. It is noted that in order to give better visualization of the MPDs at low temperature, a log scale was used for the power density ( left axis in Fig. 7 10). In this plot, it is cle arly shown that the enhancement in MPD linearly increases as temperature decreases. For instance, the MPD of ESB/GDC bilayered cell compared to that of GDC single layer increased by less than 30% at 650 oC but 51 and 100% at 550 and 450 oC, respectively. This indicates that the effect of the ESB electrolyte on LSM ESB performance is valid and even greater at low temperature. It is believed that the portion of cathodic polarization losses is larger at lower temperature due to its thermally activated nature [7] Therefore, the beneficial effect of the ESB electrolyte layer on cathode performance s hould be emphasized on using LSM ESB cathodes for low temperature SOFC application. Fig. 7 1 1 shows a comparison of MPDs for SOFCs with various LSM bismuth oxide cathodes at low to intermediate temperatures. For all temperatures, the LSM ESB cathode coupled to ESB electrolyte shows highest MPD. To my knowledge, the MPDs ( LSM ESB on ESB/GDC electrolyte ) in this study are the highest for any SOFCs using LSM bismuth oxide composite cathodes reported to date. Furthermore, this LSM ESB cathode on an ESB electrol yte can be expected to produce much higher power density at intermediate and even lower temperatures through microstructural tailoring, including cathode structure as well as electrolyte density and thickness control.

PAGE 144

144 7.4 Conclusions Conventional LSM cathodes for high temperature SOFCs were prepared for low temperature SOFC application by pairing it with ESB in a composite cathode. Due to the inherent high conductivity and fast oxygen exchange rate of ESB, the LSM ESB composite cathode showed much lower cat hodic polarization losses than any other LSM based cathode at the IT range. Moreover, when ESB was used as an electrolyte, the electrode ASR was reduced further (by ~60% ) compared to that of the same LSM ESB cathode on a GDC electrolyte. Using the LSM ESB cathode on a ESBGDC bilayered electrolyte, the MPD produced at 650 oC was ~ 865 mW/cm2, which is the highest reported MPD to date for SOFCs using LSM bismuth oxide cathodes. This study demonstrated that the performance of LSM ESB cathodes can be effectively boosted when using ESB electrolytes T his is a very promising finding for development of SOFCs which operate at low to intermediate temperatures.

PAGE 145

145 20 30 40 50 60 70 80 (c) LSM+ESB before annealing (d) LSM+ESB after annealing (a) Pure LSM Intensity (arb. unit)2 (degree)(b) 20ESB LSM+ESB after annealing LSM+ESB before annealing 20ESB Pure LSM 2 (degree) Intensity (arbitrary unit) Figure 7 1 XRD pattern of LSM, ESB, and LSM+ESB (50:50 wt%) powers before and after annealing at 900 oC 50 hrs

PAGE 146

146 ESB LSM GDC 20 m (a) ESB LSM 20 m (b) Figure 7 2 SEM images of LSM ESB cathode on GDC electrolyte (a) and ESB electrolyte (b) The insets are backscattered images.

PAGE 147

147 0 2 4 6 8 10 12 0 1 2 3 -Z'' ( cm2) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.0 0.5 1.0 1.5 -Z'' ( cm2) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.1 0.2 0.3 0.4 0.5 -Z'' ( cm2) 0.0 0.1 0.2 0.3 0.4 0.5 0.00 0.05 0.10 0.15 0.20 -Z'' ( cm2) 0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.02 0.04 0.06 0.08 0.10 -Z'' ( cm2)Z' ( cm2) (a) 500oC (b) 550oC (c) 600oC (d) 650oC (e) 700oC LSM ESB on ESB LSM ESB on GDC Figure 7 3 Impedance spectra of the LSM ESB cathode on ESB and GDC pellets at the temperature r anges from 500 to 700 oC

PAGE 148

148 500 550 600 650 700 0 2 4 6 8 10 12 Temperature (oC)Electrode ASR ( -cm2)0 10 20 30 40 50 60 70 80 90 100 ASR Decrease (%) Figure 7 4 Electrode ASRs of LSM ESB cathode on GDC (blue circles) and ESB (red squares) electrolytes and ASR reduction rate (black stars).

PAGE 149

149 [a] [b] [c] [d] [e] [f] [g] [h] 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30 0.1 1 10 LSM-GDC(50wt%) on YSZ LSM-GDC (50wt%) on GDC LSM-20ESB (15wt%) on YSZ LBSM-30ESB(50wt%) on ScSZ LSM-30ESB (50wt%)on ScSZ LSM-YSB(50wt%) on YSZ LSM-YSB(50wt%)on SDC LSM-YSB(77wt%) on YSZ LSM-20ESB(50wt%) on GDC, this study LSM-20ESB(50wt%) on ESB, this study Electrode ASR ( cm2)1000/T (K-1) Figure 7 5 Comparison of the electrode polarization resistance of LSM bismuth oxide cathodes at IT ranges. LBSM is short for La0.74Bi0.10Sr0.16MnO3. Ref : a,b [90] c [111] d [104] e [108] f[109] g [106] h [112]

PAGE 150

150 0 20 40 60 80 100 0.00 0.02 0.04 0.06 0.08 0.10 ASR ( cm2)Time (h) Time (hours) Electrode ASR ( -cm2) Figure 7 6 Long term stability test of LSM ESB cathode on ESB electrolyte at 700 oC for 100 hours.

PAGE 151

151 LSM ESB GDC Ni GDC 20 m 20 m (a) LSM ESB ESB GDC Ni GDC 25 m 3 m 50 m 500 m 20 m 20 m (b) Figure 7 7 SEM images of cross sectional views for cell 1(a) and cell 2(b), and surface views for cell1 ( c) and cell 2 ( d )

PAGE 152

152 GDC 20 m 20 m (c) ESB 20 m 20 m (d) Figure 7 7. Continued

PAGE 153

153 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0.0 0.2 0.4 0.6 0.8 1.0 Current Density (A/cm2)Potential (V)0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Power Density (W/cm2) 650oCCell 1 (LSMESB on GDC) Cell 2 (LSMESB on ESB) (a) Cell 2 Cell 1 650oC 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.00 0.05 0.10 0.15 0.20 -Z" ( -cm2)Z' ( -cm2) (b) Figure 7 8 I V characteristics (a) and impedance spectroscopy (b) of cell 1 and cell2 at 650 oC

PAGE 154

154 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0.0 0.2 0.4 0.6 0.8 1.0 Current Density (A/cm2)Potential (V)0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Power Density (W/cm2) Cell1 (LSM ESB on GDC) 650oC 600oC 550oC 500oC 450oC (a) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0.0 0.2 0.4 0.6 0.8 1.0 Current Density (A/cm2)Potential (V)0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Power Density (W/cm2) Cell2 (LSM ESB on ESB) 650oC 600oC 550oC 500oC 450oC (b) Figure 7 9 I V characteristics at various temperature for cell 1 (a) and cell 2 (b)

PAGE 155

155 450 500 550 600 650 10 100 1000 Temperature (oC)MPDs (mW/cm2)0 20 40 60 80 100 MPD Improvement (%) Figure 7 10. Maximum power density improvement (black stars) of cell 2 (red squares) at various temperatures compared to cell 1 (blue circles)

PAGE 156

156 450 500 550 600 650 700 0 100 200 300 400 500 600 700 800 900 1000 1100 MPDs (mW/cm2)Temperature (oC) LSM-20ESB(50wt%) on ESB, this study LSM-20ESB(50wt%) on GDC, this study LSM-YSB(50wt%) on SDC LSM-YSB(75wt%) on YSZ LSM-30ESB(50wt%) on YSZ LBSM-30ESB(50wt%) on ScSZ LSM-YSB(50wt%) on YSZ [a] [b] [c] [d] [e] Figure 7 11 Comparison of maximum power density of SOFCs using LSM bismuth oxide composite cathodes at IT ranges Ref : a [107] b [106] c [112] d [108] e [111] f[109]

PAGE 157

157 Table 7 1 Detailed total, ohmic, and electrode ASR values for Cel l 1 and Cell 2 at 650 oC (unit : cm2) C athode Electrolyte Total ASR IV Total ASR EIS O hm ic ASR Electrode A SR Cell -1 LSM -ESB GDC 0.566 0.569 0.119 0.450 Cell -2 LSM -ESB ESB-GDC 0.263 0.261 0.097 0.164

PAGE 158

158 CHAPTER 8 CONCLUSIONS In this dissertation, several works have been conducted to reduce polarization losses for the anode, electrolyte and cathode at reduced temperatures. In all, t he goal was to develop high performance solid oxide fuel cells running at low to intermediate temperatures. A novel AFL was developed to dramatically improve S OFC performance (Chapter 3). The bimodally integrated nano/micron composite AFL was fabricated by simpl y spray coating a precursor solution into a conventional submicron Ni GDC functional layer. The composition used for the anode substrate, the colloida l AFL (C AFL) and the precursor integrated AFL (N + C AFL) was NiGDC. T he electrolyte composition was GDC, a system that has received much attention for its potential for use in ITSOFCs. A systematic study comparing cells using no AFL, a C AFL and the newly developed N + C AFL was conducted. It showed that the N+CAFL sample exhibited a maximum power density of 1160 mW/cm2 at 600 C a 287% increase compared to the sample with no AFL and a 70% increase compared to the sample using a conventional AFL. Both ohmic and nonohmic losses were lowered, suggesting that the 2D interfacial region between the anode and the electrolyte was enhanced as well as TBP lengths. Additionally, it was shown that the fractional improvement in power increases with decreasing temperature, a critical point for reduced operating temperatures. These findings are very encouraging not only based on dramatic improvement in performance, but also in the simplicity of the technique itself. It is further believed that this N+C AFL technique is versatile enough to be applied to other SOFC systems for similar gains in performance, and will be a milestone in the field of reduced temperature SOFCs.

PAGE 159

159 In addition to the effect of particle size in AFL, the effect of AFL composition on the electrochem ical performance was investigated using submicronsized Ni GDC AFL (Chapter 4) For this, AFLs with composition ranging from 40 to 80 wt% NiO were fabricated. Microstructural analysis confirmed these functional layers to have phases which are fine and well d istributed. The optimal AFL composition was achieved at 1:1 volume ratio of Ni to GDC, which corresponds to 60 wt% NiO. This composition exhibited the higest MPD over the intermediate to low temperature range. In addition, a preliminary longterm stability test showed the possibility of using this system in practical SOFC applications The measured MPD and ASR show a n inverse linear relationship implying that the performance enhancement greatly depends on AFL composition. To better understand the relation ship between microstructure and electrochemical performance, the effect composition on the m icrostructures of Ni GDC AFLs was investigated (Chapter 5) AFLs with various Ni GDC compositions (50 80 wt% NiO before reduction) were quantified by a three dimensional (3D) reconstruction technique using a FIB/SEM dual beam system. Each AFL sample was automatically sectioned into 150 slices with 60 nm intervals. The Amira software package allowed for alignment, segmentation, and reconstruction of the 2D images to a 3D image. From these reconstructions, the volume fraction, effective particle size, phase gradient, and surface area for Ni, GDC, and pore phases as well as pore tortuosity for each sample were quantified. The estimated phase volume fraction was well mat ched to the theoretical ly calculated value. The optimal effective particle size and phase fraction were seen at 60 wt % NiO. The active TPB densities were calculated based on the connectivity of voxels

PAGE 160

160 labeled with each phase. The highest TPB density achiev ed was 2 and corresponded to 60wt% NiO The TPB density show ed an inverse proportionality to electrode ASR. This result implies that the quantified microstructural values, which are controllable, can be directly applied to predict the electrochemical performance of SOFCs In Chapter 6, a discussion was given on the fabrication of a thin and dense bilayer electrolyte consisting of erbium stabilized bismuth oxide (ESB) and gadolinium doped ceria (GDC) applied on a tapecast anodesupported SOFC using a practical and cost effective colloidal deposition process. Using a wet chemical coprecipitation method, nanosized ESB particles were successfully synthesized at temperatures as low as ~ 500 oC, which is much lower than those needed for powders prepar ed by the conventional solid state route (~ 800 oC). Due to the high sinterability of this powder, a dense erbia stabilized bismuth oxide (ESB) layer was successfully formed on a gadolinia doped ceria (GDC) electrolyte by a simple colloidal coating method. A systematic study on the sintering behavior of ESB revealed that at higher sintering temperatures, bismuth oxide can sublime or penetrate into the GDC sublayer. SEM and EDX analysis of a full button cell with an ESB/GDC bilayer sintered at 800 oC showed no visible interfacial diffusion between each layer. I V measurement of the cell showed high power density ( ~ 1.5 W/cm2) at 650 oC due to an enhancement in OCP and a significant reduction in ASR when compared to a GDC single cell. This result demonstrates that this ESB/GDC bilayer electrolyte is practical for high performance SOFCs at low operational temperature.

PAGE 161

161 In Chapter 7, an alternative high performance composite cathode for low to intermediate temperature SOFCs was described. U sing a highly conducti ve ESB phase, the performance of conventional (La0.80Sr0.20)MnO3(LSM) cathodes was dramatically improved. The ESB phase was utilized not only as the ionconducting phase in the LSM ESB composite cathode, but also as an electrolyte coupled to LSM ESB cat hode. The electrode ASR measured from a symmetric cell consisting LSM ESB electrodes on cm2 at 700oC, which is ~60 % lower than that of LSM cm2). This exemplifies the synergetic effect the E SB phase has both in the cathode bulk and at the electrolyte/electrode interface. The ESB phase is presumed to increase the effective TPB length as well as enhance the oxygen surface exchange reaction. This effect was shown to occur at all temperatures tes ted, from 500 to 700 oC. The MPDs of the anodesupported SOFCs with LSM ESB cathodes on ESB/GDC bilayered electrolytes were 50, 124, 275, 533, and 836 mW/cm2 at 450, 500, 550, 600, and 650 oC, respectively. These are to date the highest reported MPDs for S OFCs using LSM bismuth oxide composite cathodes and demonstrate that the LSM ESB composite cathode boosted by an ESB electrolyte is very promising for lowto intermediate SOFC applications.

PAGE 162

162 APPENDIX A DEPENDENCE OF OCP ON GD C ELECTROLYTE THICKNESS In t his dissertation, the GDC electrolyte has been used for all works. Contrast to the conventional YSZ electrolyte with pure ionic conductivity, SOFCs with doped ceria electrolyte have shown lower OCP from theoretical Nernst voltage due to electronic leakage current and oxygen permeation of GDC [60] It has been also reported that these leakage current and oxygen permeation properties are function of the electrolyte thickness [59, 60] In this appendix, the dependence of the OCP on GDC electrolyte thickness on Ni GDC anodesupported SOFCs was briefly studied. In order to gauge the effect of GDC electrolyte thickness on OCP of SOFCs in the IT range, eight anodesupported SOFCs with different electrolyte thicknesses were fabricated. The NiO GDC anode supports were made by a tapecasting process. For the better electrolyte deposition, submicronsized Ni GDC AFL (50wt% NiO) was colloidally deposited. Next, GDC elect rolyte was coated on the AFL surface by a spin coating method. Thickness of the electrolyte was controlled by the number of repeated coating process in the fixed spin speed and time condition. Detailed fabrication process of SOFCs with thin GDC electrolyte on the anodesupport was described on Chapter 4. Fig. A 1 shows the SEM images of the cross sectional microstructures of the tested samples. From these micrographs, it was observed that different GDC electrolyte thicknesses of the prepared samples were obtained using the control of spin coating process. T he measured electrolyte thicknesses were varied from 6.2 to 32.9 m. The detailed values of electrolyte thickness measured were tabulated in Table A 1. The OCP was measured under 90 sccm of hydrogen with 3% water as a fuel at the anode side and 30 sccm of dry air as an oxidant at the cathode side in the

PAGE 163

163 temperature range from 500 to 650 oC. The resultant OCP values at 500, 600, and 650 oC for each sample with electrolyte thickness were plotted in Fig. A 2, and summarized in Table A 1. At all temperatures, the OCP was decreased as electrolyte thickness decreased. These ex perimental results demonstrate the theory that the electronic leakage current and oxygen permeation in GDC electrolyte is greater at thinner electrolyte thickness [59, 60] Recently this experiment result was verif ied by continuum level electrochemical model developed Duncan and Wachsman [60] Fig. A 3 shows the resultant modeling fit with author s experimental data by Duncan. For this modeling, the OCP m odel equation is given as oc thoc kBT q 1 zV ln cv 0cv L De DvzvDv zeDe ln zvue uv cv 0 uecazvue uv cv L ueca (A 1) Detailed explanation and derivation of this model equation are in the previous work by Duncan et. al [60]

PAGE 164

164 32.9 m(h) LSCF GDC cathode GDC electrolyte Ni GDC anode 6.2 m(a) 9.0 m(b) 9.4 m(c) 11.3 m(d) 12.6 m(e) 19.7 m(f) 26.3 m(g) Figure A 1. Microstructu res of Ni GDC anode/GDC electrolyte/LSCF cathode SOFCs with various electrolyte thicknesses.

PAGE 165

165 650oC 600oC 500oC 5 10 15 20 25 30 35 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 Electrolyte thickness (10GDC, m)Open circuit potential (V) 1 Figure A 2. Experimental OCP values from electrochemical test at 500, 600, and 650 oC as a function of GDC electrolyte thickness

PAGE 166

166 0.7 0.8 0.9 1 0 10 20 30 40 50 60OCV (V)L ( m)500 oC 650 oC 600 oC OCP Electrolyte thickness (10GDC, m) Figure A 3. Fit of the O C P model ( eq. A 1 ) to experimental data for OCP as a function of electrolyte thickness.

PAGE 167

167 Table A 1. Summary of sample description and OCP result Sample# GDC Electrolyte Thickness( m) OCP at 650(oC) OCP at 600(oC) OCP at 500(oC) 1 6.2 0.648 0.664 2 9.0 0.749 0.838 3 9.4 0.79 0 4 11.3 0.801 0.847 0.890 5 12.6 0.822 0.855 0.913 6 19.7 0.843 0.860 0.920 7 26.3 0.876 0.914 0.961 8 32.9 0.895 0.935 0.990

PAGE 168

168 APPENDIX B LONG TERM STABILITY FOR A SOFC WITH NI GDC AFL In this section, the result of the long term stability test of a SOFC with 60wt% NiO in NiGDC AFL is de s cribed. This result is extended long term stability test data from chapter 4. To see the effect of AFL on the SOFC performance of the function of time, a potentiostatic test was conducted for 600 hrs at 650 oC. As a testing condition, a voltage of 0.379V where cell reached 98% of their MPD at initial I V test was applied. The test was carried out under the gas condition of the 90 sccm of H2 with 3% of H2O as a fuel and 90 sccm of dry air as an oxidant at anode and cathode sides, respectively. Figure B 1 shows the result of a 6 00 hours long term stability test for the 60 wt% NiO AFL cell Up to 200 hours, the effect of the AFL was retained with high power density of ~1.1 W/cm2. However, over 200 hours, the power density decreased with time. Th e degradation rate was quite linear with time and measured at the time period from 250 to 600 hours with a linear fitting method. The estimated degradation rate was ~ 1.03 mW/cm2/hour. To further investigate, current voltage curves were measured before and after long term test at the various intermediate temperature ranges. In figure B 2, the I V characteristi cs before and after long term tests compared in the temperature ranges from 500 to 650 oC with 50 oC interval. The maximum power densities before long term testing were 1.11, 0.63, 0.27, 0.12 W/cm2 at 650, 600, 550, and 500 oC, respectively. However, after the potentiostatic test for 600 hours the MPDs measured were decreased as 0.78, 0.47, 0.21, 0.08 W/cm2 at 650, 600, 550, and 500 oC, respectively. It is clearly shown that the degradation of power densities mainly came from lower OCP after long term test. It might be explained as the decreasing effective thickness of GDC

PAGE 169

169 electrolyte due to gradual reduction of Ce4+ into Ce3+ under low Po2 condition, causing increase of electronic current leakage through the electrolyte. In addition to leakage current issue, this degradation is possibly explained that this performance degradation phenomenon with time might come from the changing of microstructures of the anode and cathode with time under applied current. For further investigation, electrochemical impedance s pectroscopy can be analyzed and the microstructural analysis of the long term tested sample with SEM or 3D reconstruction using FIB/SEM dual beam system will be required.

PAGE 170

170 Figure B 1. Long term stability test of a SOFC with 60wt% NiO in NiGDC AFL. The potentiostatic test was conducted at 650 oC for 600 hours under an applied voltage of 0.379 V.

PAGE 171

171 0 2.5 5.0 7.5 0 0.25 0.50 0.75 1.00I (Amps/cm2)E (Volts) 0 0.5 1.0Power (W/cm2) Before After 650oC (a) 0 1 2 3 4 0 0.25 0.50 0.75 1.00I (Amps/cm2)E (Volts) 0 0.2 0.4 0.6Power (W/cm2) Before After 600oC (b) Figure B 2. Comparison of I V plots of the testing sample between before long term test and after long term test for 600 hours at 650oC(a), 600oC(b), 550oC(c), 500oC(d).

PAGE 172

172 0 0.5 1.0 1.5 2.0 0 0.25 0.50 0.75 1.00I (Amps/cm2)E (Volts) 0 0.1 0.2 0.3Power (W/cm2) Before After 550oC (c) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.25 0.50 0.75 1.00I (Amps/cm2)E (Volts) 0 0.05 0.10 0.15Power (W/cm2) Before After 500oC (d) Figure B 2. Continued

PAGE 173

173 APPENDIX C EXPERIMENTAL SETUP Figure C 1. Schematic SOFC testing setup a button cell testing setup configuration and I V and EIS testing equipment

PAGE 174

174 Figure C 2 Illustration of symmetric cell configuration for EIS test (top) and EIS testing setup (bottom) Tube furnaceSolatron1260 Z -plot software Quartz Plates Quartz Plates Quartz Plates

PAGE 175

1 75 LIST OF REFERENCES 1. W.R. Grove, Philosophical Magazine and Journal of Science 14 (1839), p. 127. 2. N.Q. Minh, Journal of the American Ceramic Society 76 (1993) (3), p. 563. 3. S.M. Haile, Acta Materialia 51 (2003) (19), p. 5981. 4. E.D. Wachsman and S.C. Singhal, Electrochemical society interface 18 (2009) (3), p. 38. 5. B.C.H. Steele, Journal of Materials Science 36 (2001) (5), p. 1053. 6. J.W. Fergus, Journal of Power Sources 162 (2006) (1), p. 30. 7. B.C.H. Steele, Solid State Ionics 129 (2000) (14), p. 95. 8. S. deSouza, S.J. Visco and L.C. DeJonghe, Solid State Ionics 98 (1997) (12), p. 57. 9. B.C.H. Steele and A. Heinzel, Nature 414 (2001) (6861), p. 345. 10. R.E. Williford, L.A. Chick, G.D. Maupin, S.P. Simner and J.W. Stevenson, Journal of the Electrochemical Society 150 (2003) (8), p. A1067. 11. N. Ai, Z. Lu, K.F. Chen, X.Q. Huang, X.B. Du and W.H. Su, Journal of Power Sources 171 (2007), p. 489. 12. N. Ai, Z. Lu, J.K. Tang, K.F. Chen, X.Q. Huang and W.H. Su, Journal of Power Sources 185 (2008) (1), p. 153. 13. K.F. Chen, X.J. Chen, Z. Lu, N. Ai, X.Q. Huang and W.H. Su, Electrochimica Acta 53 (2008) (27), p. 7825. 14. J.S. Ahn, H. Yoon, K.T. Lee, M.A. Camaratta and E.D. Wachsman, Fuel Cells 9 (2009) (5), p. 643. 15. B.C.H. Steele, Solid State Ionics 75 (1995), p. 157. 16. T. Takahashi, T. Esaka and H. Iwahara, Journal of Applied Electrochemistry 7 (1 977) (4), p. 299. 17. N.X. Jiang, E.D. Wachsman and S.H. Jung, Solid State Ionics 150 (2002) (34), p. 347. 18. E.D. Wachsman, P. Jayaweera, N. Jiang, D.M. Lowe and B.G. Pound, Journal of the Electrochemical Society 144 (1997) (1), p. 233.

PAGE 176

176 19. J.S. Ahn, M.A. Camaratta, D. Pergolesi, K.T. Lee, H. Yoon, B.W. Lee, D.W. Jung, E. Traversa and E.D. Wachsman, Journal of the Electrochemical Society 157 (2010) (3), p. B376. 20. J.S. Ahn, D. Pergolesi, M.A. Camaratta, H. Yoon, B.W. Lee, K.T. Lee, D.W. Jung, E. Traver sa and E.D. Wachsman, Electrochemistry Communications 11 (2009) (7), p. 1504. 21. A. Hammouche, E.J.L. Schouler and M. Henault, Solid State Ionics 28 (1988), p. 1205. 22. N.Q. Minh, Chemtech 21 (1991) (2), p. 120. 23. R.T. Dehoff, Thermodynamics in Materials Science McGraw Hill (1993). 24. P.J. Gellings and H.J.M. Bouwmeester, The CRC Handbook of Solid State Electrochemistry CRC press (1997). 25. J. Larminie and A. Dicks, Fuel Cell System Explanied Wiley (2003). 26. M. Yashima, M. Kakihana and M. Yoshimura, Solid State Ionics 868 (1996), p. 1131. 27. J.C. Boivin and G. Mairesse, Chemistry of Materials 10 (1998) (10), p. 2870. 28. D.W. Strickler and W.G. Carlson, Journal of the American Ceramic Society 47 (1964) (3), p. 122. 29. E.C. Subbarao and H.S. Mai ti, Solid State Ionics 11 (1984) (4), p. 317. 30. H. Yahiro, Y. Eguchi, K. Eguchi and H. Arai, Journal of Applied Electrochemistry 18 (1988) (4), p. 527. 31. K. Eguchi, T. Setoguchi, T. Inoue and H. Arai, Solid State Ionics 52 (1992) (13), p. 165. 32. H.A. Harwig, Zeitschrift Fur Anorganische Und Allgemeine Chemie 444 (1978) (SEP), p. 151. 33. T. Takahashi, T. Esaka and H. Iwahara, Journal of Solid State Chemistry 16 (1976) (34), p. 317. 34. M.J. Verkerk, K. Keizer and A.J. Burggraaf, Journal of Applied E lectrochemistry 10 (1980) (1), p. 81. 35. C.Z. Wang, X.G. Xu and B.Z. Li, Solid State Ionics 13 (1984) (2), p. 135. 36. P. Duran, J.R. Jurado, C. Moure, N. Valverde and B.C.H. Steele, Materials Chemistry and Physics 18 (1987) (3), p. 287.

PAGE 177

177 37. E.D. Wachsman, G.R. Ball, N. Jiang and D.A. Stevenson, Solid State Ionics 52 (1992) (13), p. 213. 38. H. Yahiro, Y. Baba, K. Eguchi and H. Arai, Journal of the Electrochemical Society 135 (1988) (8), p. 2077. 39. J.Y. Park and E.D. Wachsman, Ionics 12 (2006) (1), p. 15. 40. J.Y. Park, H. Yoon and E.D. Wachsman, Journal of the American Ceramic Society 88 (2005) (9), p. 2402. 41. Y.J. Leng and S.H. Chan, Electrochemical and Solid State Letters 9 (2006) (2), p. A56. 42. E.D. Wachsman, Solid State Ionics 152 (2002), p. 657. 43. T.H. Etsell and S.N. Flengas, Chemical Reviews 70 (1970) (3), p. 339. 44. M. Godickemeier, K. Sasaki, L.J. Gauckler and I. Riess, Solid State Ionics 868 (1996), p. 691. 45. Z.P. Shao and S.M. Haile, Nature 431 (2004) (7005), p. 170. 46. C.W. Sun and U. Stimming, Journal of Power Sources 171 (2007) (2), p. 247. 47. E. Ramirez Cabrera, A. Atkinson and D. Chadwick, Solid State Ionics 136 (2000), p. 825. 48. T. Suzuki, Z. Hasan, Y. Funahashi, T. Yamaguchi, Y. Fujishiro and M. Awano, Science 325 (2009) (5 942), p. 852. 49. M.F. Liu, D.H. Dong, R.R. Peng, J.F. Gao, J. Diwu, X.Q. Liu and G.Y. Meng, Journal of Power Sources 180 (2008) (1), p. 215. 50. H. Koide, Y. Someya, T. Yoshida and T. Maruyama, Solid State Ionics 132 (2000) (34), p. 253. 51. J. Will, A. Mitterdorfer, C. Kleinlogel, D. Perednis and L.J. Gauckler, Solid State Ionics 131 (2000) (12), p. 79. 52. J.W. Kim, A.V. Virkar, K.Z. Fung, K. Mehta and S.C. Singhal, Journal of the Electrochemical Society 146 (1999) (1), p. 69. 53. J.J. Haslam, A.Q. Pha m, B.W. Chung, J.F. DiCarlo and R.S. Glass, Journal of the American Ceramic Society 88 (2005) (3), p. 513. 54. D. Stover, H.P. Buchkremer and S. Uhlenbruck, Ceramics International 30 (2004) (7), p. 1107.

PAGE 178

178 55. S.D. Kim, S.H. Hyun, J. Moon, J.H. Kim and R.H. Song, Journal of Power Sources 139 (2005) (1 2), p. 67. 56. E. Wanzenberg, F. Tietz, P. Panjan and D. Stover, Solid State Ionics 159 (2003) (1 2), p. 1. 57. K.T. Lee, H.S. Yoon, M.A. Camarratta, N.A. Sexson, J.S. Ahn and E.D. Wachsman, to be submitted. 58. M.D. Gross, J.M. Vohs and R.J. Gorte, Journal of the Electrochemical Society 154 (2007) (7), p. B694. 59. X. Zhang, M. Robertson, C. Deces Petit, W. Qu, O. Kesler, R. Maric and D. Ghosh, Journal of Power Sources 164 (2007) (2), p. 668. 60. K.L. Duncan and E.D. Wachsman, Journal of the Electrochemical Society 156 (2009) (9), p. B1030. 61. L.C.R. Schneider, C.L. Martin, Y. Bultel, D. Bouvard and E. Siebert, Electrochimica Acta 52 (2006) (1), p. 314. 62. J.R. Wilson and S.A. Barnett, Electrochemical and Solid State Letters 11 (2008) (10), p. B181. 63. N.P. Brandon, S. Skinner and B.C.H. Steele, Annual Review of Materials Research 33 (2003), p. 183. 64. K.T. Lee, N.J. Vito, C.A. Mattehw, H.S. Yoon and E.D. Wachsman, ECS transactions to be submitted (2010). 65. N. Shikazono, Y. Sakamoto, Y. Yamaguchi and N. Kasagi, Journal of Power Sources 193 (2009) (2), p. 530. 66. J.R. Wilson, W. Kobsiriphat, R. Mendoza, H.Y. Chen, J.M. Hiller, D.J. Miller, K. Thornton, P.W. Voorhees, S.B. Adler and S.A. Barnett, Nature Materi als 5 (2006) (7), p. 541. 67. D. Gostovic, J.R. Smith, D.P. Kundinger, K.S. Jones and E.D. Wachsman, Electrochemical and Solid State Letters 10 (2007), p. B214. 68. J.R. Smith, A. Chen, D. Gostovic, D. Hickey, D. Kundinger, K.L. Duncan, R.T. DeHoff, K.S. J ones and E.D. Wachsman, Solid State Ionics 180 (2009) (1), p. 90. 69. J.R. Wilson, A.T. Duong, M. Gameiro, H.Y. Chen, K. Thornton, D.R. Mumm and S.A. Barnett, Electrochemistry Communications 11 (2009) (5), p. 1052. 70. J.R. Wilson, M. Gameiro, K. Mischaiko w, W. Kalies, P.W. Voorhees and S.A. Barnett, Microscopy and Microanalysis 15 (2009) (1), p. 71.

PAGE 179

179 71. J.R. Wilson, J.S. Cronin, A.T. Duong, S. Rukes, H.Y. Chen, K. Thornton, D.R. Mumm and S. Barnett, Journal of Power Sources 195 (2010) (7), p. 1829. 72. N. Shikazono, D. Kanno, K. Matsuzaki, H. Teshima, S. Sumino and N. Kasagi, Journal of the Electrochemical Society 157 (2010) (5), p. B665. 73. A. Bieberle, L.P. Meier and L.J. Gauckler, Journal of the Electrochemical Society 148 (2001) (6), p. A646. 74. H. In aba and H. Tagawa, Solid State Ionics 83 (1996) (12), p. 1. 75. V.V. Kharton, E.N. Naumovich and V.V. Samokhval, Solid State Ionics 99 (1997) (3 4), p. 269. 76. A. Jaiswal, C.T. Hu and E.D. Wachsman, Journal of the Electrochemical Society 154 (2007) (10), p. B1088. 77. M. Camaratta and E. Wachsman, Journal of the Electrochemical Society 155 (2008) (2), p. B135. 78. A.L. Patterson, Physical Review 56 (1939) (10), p. 978. 79. A.M. Azad, S. Larose and S.A. Akbar, Journal of Materials Science 29 (1994) (16), p 4135. 80. N.X. Jiang and E.D. Wachsman, Journal of the American Ceramic Society 82 (1999) (11), p. 3057. 81. M. Hrovat, A. Bencan, J. Holc, T. Rojac and M. Kosec, Journal of Materials Research 18 (2003) (6), p. 1297. 82. V. Gil, J. Tartaj, C. Moure and P Duran, Ceramics International 33 (2007) (3), p. 471. 83. D.W. Jung, K.L. Duncan and E.D. Wachsman, Acta Materialia 58 (2010) (2), p. 355. 84. E.P. Murray, T. Tsai and S.A. Barnett, Solid State Ionics 110 (1998) (34), p. 235. 85. S.P. Yoon, J. Han, S.W. Nam, T.H. Lim, I.H. Oh, S.A. Hong, Y.S. Yoo and H.C. Lim, Journal of Power Sources 106 (2002) (12), p. 160. 86. S.P. Jiang, Journal of Power Sources 124 (2003) (2), p. 390. 87. C.C. Kan, H.H. Kan, F.M. Van Assche, E.N. Armstrong and E.D. Wachsman, Journal of the Electrochemical Society 155 (2008) (10), p. B985. 88. C.C. Kan and E.D. Wachsman, Journal of the Electrochemical Society 156 (2009) (6), p. B695.

PAGE 180

180 89. T. Kenjo, S. Osawa and K. Fujikawa, Journal of the Electrochemical Society 138 (1991) (2), p. 349. 90. E.P. Murray and S.A. Barnett, Solid State Ionics 143 (2001) (34), p. 265. 91. C.W. Tanner, K.Z. Fung and A.V. Virkar, Journal of the Electrochemical Society 144 (1997) (1), p. 21. 92. M. Juhl, S. Primdahl, C. Manon and M. Mogensen, Journal of Power S ources 61 (1996) (12), p. 173. 93. Y.J. Leng, S.H. Chan, K.A. Khor and S.P. Jiang, Journal of Solid State Electrochemistry 10 (2006) (6), p. 339. 94. J.L. Li, S.R. Wang, Z.R. Wang, R.Z. Liu, X.F. Ye, X.F. Sun, T.L. Wen and Z.Y. Wen, Journal of Power Sourc es 188 (2009) (2), p. 453. 95. N.M. Sammes, G.A. Tompsett, H. Nafe and F. Aldinger, Journal of the European Ceramic Society 19 (1999) (10), p. 1801. 96. B.A. Boukamp, I.C. Vinke, K.J. Devries and A.J. Burggraaf, Solid State Ionics 323 (1989), p. 918. 97. B.A. Boukamp, Solid State Ionics 136 (2000), p. 75. 98. J.C. Boivin, C. Pirovano, G. Nowogrocki, G. Mairesse, P. Labrune and G. Lagrange, Solid State Ionics 113 (1998), p. 639. 99. I.C. Vinke, K. Seshan, B.A. Boukamp, K.J. Devries and A.J. Burggraaf, Solid State Ionics 34 (1989) (4), p. 235. 100. M. Dumelie, G. Nowogrocki and J.C. Boivin, Solid State Ionics 28 (1988), p. 524. 101. C.R. Xia, Y. Zhang and M.L. Liu, Applied Physics Letters 82 (2003) (6), p. 901. 102. M. Camaratta and E. Wachsman, Solid State I onics 178 (2007), p. 1411. 103. M. Camaratta and E. Wachsman, Solid State Ionics 178 (2007) (1920), p. 1242. 104. J.L. Li, S.R. Wang, Z.R. Wang, R.Z. Liu, T.L. Wen and Z.Y. Wen, Journal of Power Sources 179 (2008) (2), p. 474. 105. J.L. Li, S.R. Wang, X.F Sun, R.Z. Liu, X.F. Ye and Z.Y. Wen, Journal of Power Sources 185 (2008) (2), p. 649. 106. Z.Y. Jiang, L. Zhang, K. Feng and C.R. Xia, Journal of Power Sources 185 (2008) (1), p. 40. 107. J. Li, S. Wang, R. Liu, T. Wen and Z. Wen, Fuel Cells 9 (2009) (5) p. 657.

PAGE 181

181 108. J.L. Li, S.R. Wang, Z.R. Wang, R.Z. Liu, T.L. Wen and Z.Y. Wen, Journal of Power Sources 194 (2009) (2), p. 625. 109. Z.Y. Jiang, L. Zhang, L.L. Cai and C.R. Xia, Electrochimica Acta 54 (2009) (11), p. 3059. 110. Z.Y. Jiang, C.R. Xia, F. Zhao and F.L. Chen, Electrochemical and Solid State Letters 12 (2009) (6), p. B91. 111. J.L. Li, S.R. Wang, Z.R. Wang, J.Q. Qian, R.Z. Liu, T.L. Wen and Z.Y. Wen, Journal of Solid State Electrochemistry 14 (2010) (4), p. 579. 112. Z. Jiang, Z. Lei, B. Ding, c Xia, F. Zhao and F. Chen, International Journal of Hydrogen Energy in press (2010). 113. Q.S. Zhang, A. Hirano, N. Imanishi, Y. Takeda and K. Yamahara, Journal of Fuel Cell Science and Technology 6 (2009) (1). 114. P. Shuk, H.D. Wiemhofer, U. Guth, W. Gopel and M. Greenblatt, Solid State Ionics 89 (1996) (34), p. 179.

PAGE 182

182 BIOGRAPHICAL SKETCH Kang Taek Lee was born in Seoul Korea in 1976. After graduating from Saehwa High School in Seoul, Korea, He started his c ampus life at Yonsei University in Korea with Ceramic Engineering major. During his studies, he was enlisted in the Republic of Korea Army for a general soldier in the department of biological and chemical weapon. After that, he came back to the Yonsei Uni versity and finished his course work and received his Bachelor of Science degree, in August 2002. His enthusiasm for material s science drove him to enter the graduate school at Korea Institute of Advanced Science and Engineering (KAIST) in Korea. After 2 y ears, he received the m aster s degree and worked at LG electronics for one and half years. In 2006, he moved to the U.S. and entered University of Florida at the department of materials science and engineering. He joined in Dr. Wachsman s group, and finall y received his Ph.D in August of 2010.