NANOSCALE SURFACE ENGINEERING FOR CERAMIC FUEL CELLS

A DISSERTATION

SUBMITTED TO THE DEPARTMENT OF MECHANICAL ENGINEERING

AND THE COMMITTEE ON GRADUATE STUDIES

OF STANFORD UNIVERSITY

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

Young Beom Kim

August 2011

http://creativecommons.org/licenses/by-nc/3.0/us/

This dissertation is online at: http://purl.stanford.edu/sy490sp5158

© 2011 by Young Beom Kim. All Rights Reserved.

Re-distributed by Stanford University under license with the author.

This work is licensed under a Creative Commons Attribution-Noncommercial 3.0 United States License.

ii

I certify that I have read this dissertation and that, in my opinion, it is fully adequatein scope and quality as a dissertation for the degree of Doctor of Philosophy.

Friedrich Prinz, Primary Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequatein scope and quality as a dissertation for the degree of Doctor of Philosophy.

Thomas Kenny

I certify that I have read this dissertation and that, in my opinion, it is fully adequatein scope and quality as a dissertation for the degree of Doctor of Philosophy.

Xiaolin Zheng

Approved for the Stanford University Committee on Graduate Studies.

Patricia J. Gumport, Vice Provost Graduate Education

This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file inUniversity Archives.

iii

iv

ABSTRACT

Ceramic fuel cell (CFC) is an all-solid-state energy conversion device and usually refers

to fuel cells employing solid ceramic electrolytes. The present generation of ceramic fuel

cells can be classified into two types according to the electrolytes they use: oxygen ion

conducting fuel cells, or solid oxide fuel cells (SOFCs) and proton conducting fuel cells

(PCFC or PCOFC). CFCs usually have the highest operating temperature of all fuel cells

at about 600~1000oC for reasonably active charge transfer reactions at the electrode-

electrolyte interface and ion transport through the electrolyte. This high CFC’s operating

temperature has limited practical applications. The goal of my Ph.D. research is to

minimize the activation losses at the electrode/electrolyte interface by nanoscale

engineering to achieve decent performance of ceramic fuel cells at lower operating

temperatures (300~500oC). This dissertation has three main nanoscale surface

engineering approaches according to the fuel cell components: electrode structure,

composite electrolyte structures with thin interlayers, and the fabrication of three-

dimensional fuel cell membrane-electrode assemblies (MEAs).

We would call the first part of the dissertation as nanoscale electrode structure

engineering for ceramic fuel cells. It describes the fabrication and investigation of

morphologically stable model electrode structures with well-defined and sharp

v

platinum/yttria stabilized zirconia (YSZ) interfaces to study geometric effects at triple

phase boundaries (TPB), which is known as the actual electrochemical reaction site. A

nanosphere lithography (NSL) technique using monodispersed silica nanoparticles is

employed to deposit nonporous platinum electrodes containing close-packed arrays of

circular openings through the underlying YSZ surface. These nano-structured dense Pt

array cathodes exhibited better structural integrity and thermal stability at the fuel cell

operating temperature of 450~500oC when compared to porous sputtered Pt electrodes.

More importantly, electrochemical studies on geometrically well-defined Pt/YSZ sharp

interfaces demonstrated that the cathode impedance and cell performance both scale

almost linearly with aerial density of TPB length. These controlled experiments also

allowed for the estimation of the area of the electrochemical reaction zone. This

information can be used as a platform for designing the electrode structure to maximize

the performance of ceramic fuel cells.

The second part of the experiment is about electrolyte surface structure engineering by

fabricating composite electrolyte structures. This study describes, both theoretically and

experimentally, the role of doped ceria cathodic interlayers and their surface grain

boundaries in enhancing oxygen incorporation kinetics. Quantum mechanical simulations

of oxygen incorporation energetics support the experimental results and indicate a low

activation energy of only 0.07eV for yttria-doped ceria (YDC), while the incorporation

reaction on YSZ is activated by a significantly higher energy barrier of 0.38eV. For

experiments, epitaxial and polycrystalline YDC, gadolinia-doped ceria (GDC) thin films

vi

were grown by pulsed laser deposition (PLD) on the cathode side of 300µm-thick single

crystalline (100) and 100µm-thick polycrystalline YSZ substrates, respectively. For the

composite electrolyte sample with YDC interlayer, the Oxygen isotope exchange

experiment was conducted employing secondary ion mass spectrometry (SIMS) with

high spatial resolution (50nm). The surface mapping result of 18

O/16

O shows high activity

at surface grain boundary regions indicating that the grain boundary regions are

electrochemically active for oxygen incorporation reaction. Fuel cell current-voltage

behavior and electrochemical impedance spectroscopy measurements were carried out in

the temperature range of 350oC-450

oC on both single crystalline and polycrystalline

interlayered cells. Results of dc and ac measurements confirm that cathodic resistances of

cells with epitaxial doped-cerium oxides (GDC, YDC) layers are lower than that for the

YSZ-only control cell. This is attributed to the higher surface exchange coefficient for

doped-cerium oxides than for YSZ. Moreover, the role of grain boundary density at the

cathode side external surface was investigated on surface-engineered electrode-

membrane assemblies (MEA) having different doped-ceria surface grain sizes. MEAs

having smaller surface grain size show better cell performance and correspondingly

lower electrode interfacial resistance. Electrochemical measurements suggest that doped-

ceria grain boundaries at the cathode side contribute to the enhancement of oxygen

surface kinetics. These results provide an opportunity and a microstructure design

pathway to improve performance of LT-SOFCs by surface engineering with nano-

granular, catalytically superior thin doped-ceria cathodic interlayers.

vii

Thirdly, as a reaction surface engineering for SOFC, we investigated a novel method for

creating a three-dimensional (3-D) fuel cell architecture to enhance fuel cell performance

by increasing the area of the electrolyte membrane. The research describes the fabrication

and operation of a low temperature 3-D protonically conducting ceramic fuel cell

featuring a close packed and free standing crater patterned architecture achieved by

nanospherical patterning (NSP) and dry etching techniques. The cell employed conformal

layers of yttria-doped barium zirconate (BYZ) anhydrous electrolyte membrane

(~120nm) sandwiched between thin (~70nm) sputtered porous Pt electrode layers. The

fuel cell structure achieved the highest reported peak power densities up to 186 mW/cm2

at 450oC using hydrogen as fuel. To further investigate the proton conductivity of the

electrolyte, which is BYZ, we studied the effect of crystalline structures on proton

conductivity of BYZ thin films. The results showed that the grain boundaries impede the

proton transport through the grain boundary and cause extremely high resistance for ionic

transport in the film. This experimental result also can provide significant implications in

designing proton conducting ceramic fuel cells.

All these efforts and investigations were intended to enhance the ceramic fuel cell

performance at low operating temperatures (300–500oC) by improving

electrode/electrolyte interface electrochemical reactions. We expect to achieve further

enhancement when we combine the approaches each other. For example, fabrication of

three-dimensional fuel cells with doped-ceria interlayers and composite electrolyte

viii

structures with optimized electrode nano-structures. Investigations are on-going in our

laboratory as a future work.

ix

ACKNOWLEDGMENTS

First of all, I would like to thank all the members in Nanoscale Prototyping Laboratory at

Stanford. This work would have not been possible without their support throughout my

graduate career. Especially, I would like to express my great depth of gratitude to my

principal advisor, Prof. Fritz Prinz. There is a saying that “Water changes its shape

depending on the containers. Similarly, people vary depending on whom he/she with”

and I personally believe that. After joining Prof. Prinz’s research group, I can say that my

attitude toward my life and scientific research has been changed. He is a great leader and

motivator. His remarkable guidance, encouragement, and enthusiasm for scientific

research have truly helped me to make it this far. Also, I am truly grateful to Prof. Turgut

Gür for his thoughtful guidance and intellectual contributions throughout my research. I

would also like to thank my dissertation committee members for their insightful reviews,

valuable comments and time: Prof. Thomas Kenny, Prof. Xiaolin Zheng, and Prof. Sally

Benson.

I would like to thank especially to my close colleagues, Dr. Joon Hyung Shim, Dr.

Wonyoung Lee, Joong Sun Park, Jihwan An, Dr. Pei-Chen Su, Dr. Cheng-Chieh Chao,

Dr. Jason Komadina for their valuable discussions, advice, and dedicated collaborations.

I would also like to thank former NPL members who inspired me, Dr. Suk Won Cha, Dr.

Sangkyun Kang, Dr. Rainer Fasching, Dr. Won Hyung Ryu, Dr. Minhwan Lee, Dr. Tim

x

Holme, Dr. Rojana Pornpransertsuk, Dr. Seoung-jai Bai, Masayuki Sugawara, Dr. Kyle

Hammerick, Dr. Yu-Chi Chang, and Dr. Jeremy Cheng.

Also, I am grateful to all NPL members, Dr. Neil Dasgupta, Dr. Munekazu Motoyama,

Hee-Joon Jung, Hark Lee, Xu Tian, Orlando Trejo, Mike Langston, James Mack, Phil

Van Stockum, Andrei Iancu, John Xu, Zubin Huang, Zeng Fan, Takane Usui, Ushio

Harada, and Hitoshi Iwadate for their support and collaboration.

I must recognize all my friends in Korea for being always supportive and cheerful. They

have really helped me not to lose my focus. I thank all my KME and KCF friends for

their support and prayer.

In addition, I would like to thank my family members. I am deeply thankful to In-Soo

Kim, Heung-Soo Kim, and Sung-Duk Lee for their support throughout my life and also

thank to all my cousins. I would like to thank my in-laws, Sung-Min Choi, Young-Ja

Song, Bryan Cho, Woo-Ri Choi (Yuna and Yennie), and Dan-Bi Choi for their prayers.

Last but not least, I must thank my dad, Sung-Soo Kim, mom, Shin-Ae Moon, brother,

Young-Jun Kim (and Se-Hee Kim), my lovely wife, Ji-Soo Choi for their endless love,

support, and prayers.

xi

DEDICATION

The author wishes to dedicate this dissertation to his grandfathers, grandmothers, father,

mother, brother, wife, and everyone in his family.

xii

TABLE OF CONTENTS

List of tables .......................................................................................................................xv

List of figures ................................................................................................................... xvi

Chapter 1: Intdroduction ..................................................................................................1

1.1 Dissertation Overview .............................................................................................1

1.2 Outline......................................................................................................................3

1.3 Individual/Group Research Statement .....................................................................5

1.4 References ................................................................................................................6

Chapter 2: Fuel Cell Overview .........................................................................................8

2.1 Fuel Cell Basics .......................................................................................................8

2.2 Types of Fuel Cells ...............................................................................................11

2.3 Fuel Cell Performance and Loss Mechanisms .......................................................13

2.3.1 Thermodynamic Reversible Voltage ............................................................15

2.3.2 Activation Losses ..........................................................................................18

2.3.3 Ohmic Losses ................................................................................................21

2.3.4 Mass Transport Losses ..................................................................................23

2.4 Ceramic Fuel Cells .................................................................................................24

2.5 References ..............................................................................................................27

Chapter 3: Materials for Ceramic Electrolytes and A Fabrication Technique .........28

3.1 Ceramic Electrolyte Materials ...............................................................................28

3.1.1 Oxygen Ion Conducting Ceramics ................................................................28

3.1.2 Proton Conducting Ceramics ........................................................................32

3.2 Pulsed Laser Deposition ........................................................................................34

3.2.1 Background ...................................................................................................34

3.2.1 Example Materials of Pulsed Laser Deposition ............................................38

xiii

3.3 References ..............................................................................................................41

Chapter 4: Nano-pore Structured Platinum Electrode Arrays for LT-SOFCs .........43

4.1 Introduction ............................................................................................................43

4.2 Experimental ..........................................................................................................48

4.2.1 Spherical Silica Particle Fabrication .............................................................48

4.2.2 Nanosphere lothography (NSL) for Nano-Structured Electrodes .................49

4.3 Results and Discussion ..........................................................................................53

4.3.1 Thermal Stability of the Nano-pore Structured Electrode ............................53

4.3.2 Investigation of TPB Scaling Behavior ........................................................59

4.3.3 TPB Width Estimation for Pt/YSZ Interface ................................................65

4.4 Conclusion .............................................................................................................70

4.5 References ..............................................................................................................72

Chapter 5: Cathodic Surface Engineered LT-SOFCs ..................................................75

5.1 Introduction ............................................................................................................75

5.2 Experimental ..........................................................................................................80

5.2.1 Quantum Simulation of Oxygen Incorporation Energies .............................80

5.2.2 Oxygen Isotope Exchange and NanoSIMS...................................................81

5.2.3 Composite Electrolyte Fuel Cell Fabrication and Characterization .............83

5.3 Results and Discussion ..........................................................................................85

5.3.1 Surface Engineered SOFC with YDC Cathodic Interlayer...........................85

5.3.2 Surface Engineered SOFC with GDC Cathodic Interlayer.........................109

5.4 Conclusion ...........................................................................................................116

5.5 References ............................................................................................................117

Chapter 6: Three-dimensional Proton Conducting Fuel Cell Architecture with

Ultra Thin Ceramic Electrolyte ....................................................................................121

6.1 Introduction ..........................................................................................................121

6.2 Experimental ........................................................................................................124

xiv

6.3 Results and Discussion ........................................................................................127

6.4 Conclusion ...........................................................................................................139

6.5 References ............................................................................................................139

Chapter 7: Effect of Crystallinity on Proton Conductivity in Yttrium-doped

Barium Zirconate Thin Films .......................................................................................142

7.1 Introduction ..........................................................................................................143

7.2 Experimental ........................................................................................................145

7.3 Results and Discussion ........................................................................................148

7.4 Conclusion ...........................................................................................................165

7.5 References ............................................................................................................166

xv

LIST OF TABLES

Number Page

Table 2-1: Characteristics of five major fuel cell types ....................................................11

Table 3-1: Wavelength of excimer lasers .........................................................................36

Table 3-2: Example materials deposited by pulsed laser deposition and applications

of those materials .............................................................................................40

xvi

LIST OF FIGURES

Number Page

Figure 2-1: Schematic of a typical hydrogen-oxygen fuel cell ...........................................9

Figure 2-2: Schematic of a typical fuel cell current-voltage (I-V) curve .........................15

Figure 3-1: Schematic of PLD mechanism (Top) and a sample picture of PLD process:

deposition of BYZ on MgO (100) substrate ....................................................35

Figure 4-1: Synthesized spherical silica particles with different diameters: (a) 130nm,

(b) 300nm, (c) 650nm ......................................................................................49

Figure 4-2: Schematic of the Langmuir –Blodgett trough .................................................50

Figure 4-3: Fabrication process schematic of nano-pore structured electrode by using

nanosphere lithography (NSL) technique ........................................................51

Figure 4-4: SEM images of nano-pore structured dense Pt electrodes ..............................52

Figure 4-5: Schematic illustration of the probing station for electrochemical

characterization of fuel cell MEAs ..................................................................53

Figure 4-6: SEM images of (a) as sputtered porous Pt layer, (b) after short time (~

30mins) operation of fuel cell at elevated temperature (450oC), clearly

indicate a dramatic change in Pt morphology and a proportionate reduction

in the TPB density ............................................................................................54

Figure 4-7: Potentioamperometry data at 0.6V, comparing the behavior of SOFC MEA

with porous Pt electrode and SOFC with nano-pore structured Pt electrode.

Measurement was conducted for 12 hours continuously at 500oC. (a)

Absolute output current densities indicating severe degradation in

performance of porous Pt within a short time as opposed to stable behavior

of patterned dense Pt. (b) Normalized current densities plot showing

relative amounts of degradation from the initial performance .........................55

Figure 4-8: High resolution SEM images of nano-structured fresh Pt electrode before

running (top (a) and tilted (c) views), and after running

chronoamperometrically for 12 hours at 500oC (top (b) and tilted (d)

xvii

views). Clearly, nano-structured Pt does not show any major morphological

change and the TPBs are well conserved after long operation ........................57

Figure 4-9: V-I-P comparison of SOFC at 450oC employing DC sputtered porous Pt

electrode versus nano-pore structured Pt electrode at the cathode ..................58

Figure 4-10: Nano-pore structured dense Pt electrode with different final pore sizes

and TPB density, (a) 300nm, (b) 400nm. SEM images were taken with the

same magnification and the image with smaller initial particle size shows

denser pores, which relates to increased TPB density .....................................61

Figure 4-11: (a) Fuel cell I-V measurement of SOFC samples at 450oC with structured

electrode with different pore diameters. (b) V-I-P plot where current is

normalized by TPB density. The plots overlay in good registry especially in

the activation regime as expected, and indicates that fuel cell performance

scales with TPB density ...................................................................................62

Figure 4-12: EIS Nyquist spectra of SOFC samples featuring nano-pore structured

dense Pt cathodes with different TPB densities ..............................................64

Figure 4-13: (a) Schematic of the TPB. (b) Cross-section image showing TPB width .....65

Figure 4-14: Graphical estimation of electrode/electrolyte interface resistance as the

TPB density increases. Interface resistance will be decrease as we increase

the TPB density by decreasing the pore size since we are introducing more

electrochemical reaction site. If the TPB width overlap starts, the resistance

will not decrease anymore and will show saturation behavior ........................68

Figure 4-15: SEM images of nano-pore structured Pt electrodes on single crystalline

YSZ substrates. TPB linear density was estimated by measuring the pore

size and spacing between the pores. As we have smaller pore size, we have

larger TPB density. The electrode pore sizes are (a) 240nm, (b) 350nm, (c)

430nm, and (d) 570nm .....................................................................................69

Figure 5-1: (a) Change in energy as a function of height for atomic oxygen diffusing

into a vacancy in the first layer of YSZ. Height and energy is referenced to

the stable adsorbed state. Charge, plotted on the right axis, is the electron

density integrated within a Wigner Seitz sphere of radius 0.82Å around the

xviii

radius (a more positive value corresponds to higher electron density). (b)

Snapshots of atom structure as oxygen is incorporated. Oxygen atoms are

shown in red, Y in yellow, Zr in purple, and Pt in silver .................................86

Figure 5-2: (a) Change in energy as a function of height for atomic oxygen diffusing

into a vacancy in the first layer of YDC. Height and energy is referenced to

the stable adsorbed state. Charge, plotted on the right axis, is the electron

density integrated within a Wigner Seitz sphere of radius 0.82Å around the

radius. (b) Snapshots of atom structure as oxygen is incorporated. Oxygen

atoms are shown in red, Y in yellow, Ce in blue, and Pt in silver ...................87

Figure 5-3: XRD patterns of PLD YDC films deposited on single crystalline YSZ

(100) substrate. Spectra show only (100) peak up to the film thickness of

130nm, which indicates perfect epitaxial growth of YDC films .....................89

Figure 5-4: I-V performance of YDC interlayered SOFC and YSZ-only control sample

measured at 450oC. The plot shows gradual performance enhancement up

to about 50nm of YDC interlayer thickness, beyond which the fuel cell

performance remains unchanged with increasing YDC thickness ..................90

Figure 5-5: EIS data of YDC interlayered fuel cell measured at different cell voltage

conditions at 400oC. Two loops are observed. The high frequency loop

seems to be independent of cell voltage, indicating that this arc corresponds

to ionic transport through the electrolyte (Rohmic). In contrast, the low

frequency loop is dependent on cell voltage indicating that this arc

corresponds to the electrode interface resistance (Relectrode) .............................91

Figure 5-6: Extracted ohmic resistances of YDC interlayered SOFCs for different

YDC thicknesses at temperatures of 350oC~450

oC. Zero in the x-axis

indicates the bare YSZ sample with no interlayer. The plot shows no

discernable change in cell ohmic resistance with increasing interlayer

thickness up to 130nm .....................................................................................94

Figure 5-7: Electrode interface resistance values for the YDC interlayered SOFCs with

different thicknesses extracted from impedance measurements. The

resistance starts to drop immediately after the introduction of a thin layer

xix

YDC (<10nm). After forming a full covered YDC layer, the electrode

resistance reaches a plateau, and does not change with further increase in

YDC thickness .................................................................................................94

Figure 5-8: AFM scanned surface topography images of YDC interlayers with

different thicknesses. (Left) Bare single crystalline YSZ, (Center) ~14nm

YDC, (Right) ~80nm YDC on top of YSZ. It shows grain formation as the

YDC thickness increases..................................................................................96

Figure 5-9: Surface sensitive XPS analysis of surface modified samples with three

different YDC thickness within the binding energy regime of one of the

main Zr peaks. As the YDC interlayer thickness increases Zr peak

decreases and at the thickness about 26nm, almost no Zr peak is observed ....97

Figure 5-10: (a) Surface SEM image of YDC sintered pellet, where dashed line shows

clear grain boundaries. (b) 18

O/16

O concentration map of corresponding

YDC surface obtained from NanoSIMS. 18

O/16

O count ratio was observed

higher at grain boundary regions (dashed) than bulk regions indicating

oxygen isotopes were more populated in grain boundary regions .................100

Figure 5-11: AFM images of YDC surface additionally deposited on polycrystalline

YSZ substrate and post-annealed at different temperatures. (a) 750oC, (b)

1100oC, (c) 1300

oC, and (d) 1500

oC ..............................................................101

Figure 5-12: Average grain size of YDC interlayer as a function of post-annealing

temperature ....................................................................................................102

Figure 5-13: Current-Voltage (I-V) behavior of fuel cell MEAs measured at 400oC.

Fuel cells with smaller surface grain size show higher performance in terms

of peak power densities ..................................................................................104

Figure 5-14: Electrochemical impedance spectroscopy (EIS) data of 1500oC annealed

YDC/YSZ composite fuel cell sample measured at 350oC indicating three

loops. The two high frequency loops seem to be independent of cell voltage

conditions, indicating that these arcs correspond to ionic transport through

electrolyte and representing bulk (arc I) and grain boundary (arc II)..

whereas the low frequency loop shows dependence on cell voltage

xx

conditions indicating that this arc corresponds to the electrode interface

resistance ........................................................................................................105

Figure 5-15: A plot showing extracted electrode interface resistances (at 450oC, 0.6V)

as a function of estimated surface grain boundary densities. As expected,

the electrode resistance decreases as the surface grain boundary density

increases (lower grain sizes) ..........................................................................106

Figure 5-16: Exchange current densities for all measured samples with different grain

sizes were calculated at temperatures 350oC-450

oC. As the surface grain

size decreases (i.e., higher grain boundary density), the electrode interface

resistance decreases. This indicates that the surface grain boundaries

enhance oxygen surface kinetics at the cathode side .....................................108

Figure 5-17: X-ray diffraction patterns of (a) epitaxial and (b) fully developed

polycrystalline GDC films on single (100) and polycrystalline YSZ

substrates, respectively ..................................................................................110

Figure 5-18: Current-voltage (I-V) behavior of epitaxial GDC interlayered MEA,

measured at 450oC. The SOFC MEA with GDC interlayer shows about 2-

fold higher peak power density ......................................................................111

Figure 5-19: Atomic force microscopy (AFM) topography images of GDC surfaces,

annealed at (a) 750oC, (b) 1200

oC, and (c) 1450

oC. As the post-annealing

temperature increases, the grain size also increases ......................................112

Figure 5-20: I-V performance at 450oC of GDC/YSZ composite electrolyte MEAs

with different GDC surface grain sizes,. The smaller surface grain size

sample (lower annealing temperature), which corresponds to the higher

surface grain boundary density, shows higher peak power density ...............114

Figure 5-21: Arrhenius plot of cathodic interfacial resistances of MEAs with different

GDC surface grain sizes. As the post-annealing temperature increases (i.e.,

as the surface grain boundary density decreases), the electrode interfacial

resistance increases. MEAs with nano-granular GDC surface grains show

lower electrode interfacial resistances than those with larger surface grain

size .................................................................................................................115

xxi

Figure 6-1: Schematic illustration of the NSP processing sequence for the fabrication

of 3-D crater patterned freestanding fuel cell MEAs. (a) Si substrate. (b)

Silica particles with Al layer on Si wafer. (c) Al mask after removing the

particles. (d) Formation of trenches by RIE etching. (e) Silicon nitride

deposition. (f) Removal of Si template by KOH etching. (g) Depositing

BYZ and removal of the nitride layer by dry etching. (g) Sputtering of

porous Pt catalyst/electrode (dots) deposited on both sides (h) .....................125

Figure 6-2: SEM images of, (a) silicon nano-trenches after removal of spherical

particles, (b) silicon nano-trench structure created after gas phase etching,

(c) free-standing 3-D nitride template after removing the backside silicon

by KOH etching .............................................................................................128

Figure 6-3: SEM images of the crater patterned BYZ fuel cell MEA, (a) after BYZ

deposition on the 3-D nitride template, (b,c) after porous Pt electrodes are

coated on both sides of the membrane, and (d) finished 3-D BYZ MEA

taken from an angle of 52o from the top (d) ..................................................129

Figure 6-4: Cross sectional HRTEM images ((a) and (c)) showing the dense columnar

grain structure, and, (b) the SAD pattern indicating the fully developed

polycrystalline nature of the BYZ film .........................................................131

Figure 6-5: (a) Electrochemical impedance spectra at 400oC at cell voltages of 0.9V

and 0.7V, with inset showing the details of the high frequency region, and

(b) voltage-current-power density (V-I-P) behavior of 3-D crater patterned

BYZ fuel cells measured at 350-450oC using hydrogen fuel .......................133

Figure 6-6: Compositional depth profiles of the PLD BYZ film ....................................136

Figure 6-7: The high resolution C1s

spectra show two peaks at ~285.0eV and ~289.9eV

assigned to surface contamination and to CO32-

(possibly in the form of

BaCO3) environment, respectively ...............................................................136

Figure 7-1: Conductivity measurement setup with microcontacting probes connected

to the EIS software ........................................................................................147

Figure 7-2: XRD patterns of BYZ thin films grown on quartz substrates (Q) with

different deposition temperatures (a) 900oC, (b) 700

oC, and (c) 400

oC .......148

xxii

Figure 7-3: XRD patterns of BYZ thin films grown on MgO(100) substrates with

different deposition temperatures (a) 900oC, (b) 800

oC, (c) 700

oC, and (d)

600oC .............................................................................................................149

Figure 7-4: High resolution TEM image: {001} planes of BYZ (perovskite structure)

grows epitaxially on {001} type planes of MgO (rocksalt structure). A

yellow-dotted rectangle shows dimensional matching of each MgO and

BYZ unit cell .................................................................................................151

Figure 7-5: Selected area diffraction (SAD) patterns of BYZ films deposited by PLD

at 900 °C (a series at top) and at 600 °C (b series at bottom): Both a-1 and

b-1 SAD patterns were taken only from MgO for setting orientation

standard. Both a-2 and b-2 SAD patterns were taken from MgO and BYZ

to check orientation relationship of BYZ films to MgO substrate. BYZ film

deposited by PLD at 900 °C (a-2) shows epitaxial growth, unlike BYZ film

deposited by PLD at 600 °C, which illustrates slight disorientation. A

digitally 4X magnified SAD of 200 type spots (b-3) confirms orientation

mismatch of each film ...................................................................................152

Figure 7-6: Bright Field (BF), Dark Field (DF), High Resolution (HR) TEM images

and SAD pattern of BYZ films deposited by PLD at 600 °C on Quartz:

Both BF (a) and DF (b) images show visual orientation difference of each

BYZ grain. HRTEM (c) shows polycrystallinity of BYZ films with grains

and grain boundaries. And SAD pattern only taken from BYZ indicates

randomly-oriented of polycrystalline BYZ grains ........................................154

Figure 7-7: Measured Nyquist impedance plots and fitting curves to the parallel R//C

circuit model. (a) BYZ-MgO(100) sample deposited at 900oC and

measured at 200oC. Bias independence of the spectra indicates that the

semicircle is associated with electrolyte impedance. (b) BYZ film

deposited on quartz at 400oC and measured at 700

oC ..................................157

Figure 7-8: EIS data measured at different temperatures for BYZ-MgO(100) film

deposited at 900oC ........................................................................................158

xxiii

Figure 7-9: Arrhenius plot showing the conductivity values of BYZ thin films

deposited on MgO(100) and amorphous quartz substrates at different

deposition temperatures. In addition, both sets of data are compared with

the reference conductivity values from the literature, including bulk and

experimentally obtained non-epi references .................................................159

Figure 7-10: Variations in the SAD patterns obtained from a cross-section sample of

MgO (100)/BYZ film (deposited by PLD at 700°C) as the SAD aperture

position is moved from MgO into the BYZ film: SAD aperture positioned

on MgO only (a), 20nm into BYZ from MgO interface (b), 40nm into BYZ

from MgO interface (c), and 100nm across the entire BYZ (d). Digitally 4X

magnified SAD patterns of 101 spots from Figure 7-10-b, c & d are shown

in Figure 7-10-e, and indicate how the epitaxy in the BYZ film near the

MgO substrate gradually changes to increasing polycrystallinity with wider

divergence in orientation as the aperture moves towards upper regions of

the BYZ film. The ranges selected for SAD patterns are indicated on the

cross-sectional bright field TEM image of Figure 7-10-f using a color

scheme (white for a, blue for b, green for c, and orange for d) ....................162

Figure 7-11: Plot showing the conductivity versus degree of crystallinity of BYZ-

quartz samples with three different deposition temperatures. Error bars are

included for one measured temperature data since all three samples have

the same error bars. The plot indicates that as the deposition temperature

increases the degree of crystallinity and the conductivity increases .............165

1

CHAPTER 1. Introduction

1.1 Dissertation Overview

Due to the scarcity of fossil fuels and their environmental pollution from carbon dioxide

(CO2) emission, interest in renewable and clean energy resources of all forms is very

high. Renewable energy resources such as solar and wind energy have been considered

as strong candidates, but both sources have the inherent problem of irregularity. Using

hydrogen as an energy carrier, fuel cells can be considered as a renewable energy and a

good sustainable energy resource.

Fuel cells are environmentally friendly energy conversion and power generation devices,

and some of the most promising candidates as a zero-emission power sources. Among the

fuel cells, ceramic fuel cells such as solid oxide fuel cells (SOFCs) and proton conducting

oxide fuel cells (PCOFCs) have attracted recent attention due to their high energy

conversion efficiency. However, the ceramic fuel cells especially SOFCs usually have

high operating temperatures (800~1000oC) due to the nature of the high activation energy

(~1eV) of ionic transport in such solid ceramic electrolytes [1-2]. This high operating

temperature limits a wide range of practical applications. It is therefore desirable to

reduce the operating temperature of SOFCs to lower than 500oC. Unfortunately, in this

2

low temperature regime, two critical factors become much more pronounced and

adversely affect the performance of low-temperature SOFCs (LT-SOFC). One is

increased ohmic resistance due to slower ionic transport through the YSZ electrolyte at

these temperatures. The other is similarly increased activation losses mostly due to the

slower kinetics of the oxygen reduction reaction at the cathode interface, and it is known

that this increase in activation energy is more severe than the increased ohmic resistance

[1-2].

To compensate for the increased ohmic resistance, many studies have sought to develop

materials having high conductivity and to reduce the electrolyte thickness [3-8]. Although

the use of thin film electrolytes mitigated ohmic losses, the electrode polarization

process, which is highly related to the electrochemical surface reactions, still remains a

key challenge at these low operating temperatures (300~500oC) due to the high activation

energy (>1.5eV) of oxygen reduction reaction. Recently, nanoscale engineering has been

intensively investigated in such energy conversion devices, and it became available from

the development of technologies for fabrication and characterization of materials in

nanometer, or even smaller, scale. In nanoscale, the increase of surface-to-volume ratio is

dramatic, and the material properties can also be changed from the macro-scale.

Therefore, by nanoscale engineering superior and/or unique material properties, which

can enhance the surface reaction kinetics, we can significantly improve the performance

of energy conversion devices.

3

This dissertation discusses nanoscale surface engineering approaches to reduce the

remaining activation loss issue for improving the performance of low temperature

ceramic fuel cells. First, a novel fabrication method for nano-structured electrodes was

developed to understand the geometry of the electrochemically active zone for solid

oxide fuel cells and to maximize the reaction sites. Second, a composite electrolyte was

fabricated using cathodic interlayers, which have higher ionic conductivity and superior

surface activity to increase electrochemical surface reaction rate. The third approach is

the fabrication of a three-dimensional ceramic fuel cell structure to improve the

performance by increasing the effective surface reaction sites in a given area.

1.2 Outline

The main body of this thesis contains 6 chapters. The first two chapters provide

fundamental backgrounds for the works in the dissertation. The following 4 chapters

present the research focused on improving the performance of ceramic fuel cells by the

nanoscale engineering of electrochemical reaction surfaces.

Chapter 2 describes the scientific and theoretical overview of fuel cell operation,

fuel cell types, fuel cell performance characteristics, and ceramic fuel cell basics.

4

Chapter 3 describes basic materials science of ceramic electrolytes including the

fundamental ion transport mechanism and exemplary materials for each type of

ionic conductors. It also provides the basics of pulsed laser deposition as a

technique for depositing those materials.

Chapter 4 introduces a novel fabrication method for nano-pore structured metal

electrodes and the investigation of the electrochemically active zone for fuel cell

operation.

Chapter 5 describes the fabrication of YSZ-based composite electrolytes using

YDC and GDC cathodic thin interlayers to enhance surface oxygen kinetics for

low temperature SOFCs.

Chapter 6 introduces a method for fabricating three-dimensional proton

conducting ceramic fuel cell electrolyte structures to increase effective reaction

sites at a given area.

Chapter 7 describes the effect of crystallinity on proton conductivity in BYZ thin

films as proton conducting ceramic electrolytes.

5

1.3 Individual/Group Research Statement

Much of the work presented here was accomplished in collaboration with outstanding

group members in the Nanoscale Prototyping Laboratory (NPL). Professor Fritz B. Prinz,

who is the principal investigator (PI) of the research group, highly encourages a

cooperative research environment and teamwork-based scientific research to expedite the

progress. The group members, having their own specialties, are highly involved in one

research project. Therefore, the lively discussions and cooperation for experiments with

all NPL members have been priceless throughout the research in this dissertation.

The initial fabrication works for the nano-pore structured layer were performed in

collaboration with Dr. Steve Connor and Ching-Mei Hsu. Dr. Pei-Chen Su and I came up

with the idea of using that structure as fuel cell electrodes. I designed and performed the

series of experiments to study structural stability of the nano structured metal electrodes

and to investigate the electrochemical reaction sites for solid oxide fuel cells.

The PLD of BYZ thin film work was initiated by Prof. Joon Hyung Shim. For further

development of PLD BYZ and other materials used for the experiments in this thesis such

as YDC and GDC, I explored the deposition conditions and characterization of the thin

films. Based on the deposition conditions, I designed and performed the doped ceria

cathodic interlayered SOFC experiments (Chapter 5).

6

The DFT simulations for calculating the oxygen incorporation energy of Pt/YSZ, Pt/YDC

systems were designed and performed by Dr. Tim Holme. The oxygen isotope exchange

experiment YDC bulk substrate was performed by Joong Sun Park (Chapter 5).

Microscale three-dimensional (3-D) SOFC fabrication by the MEMS process was

initiated by Dr. Pei-Chen Su. On this 3-D fabrication concept, I designed a nanoscale 3-D

proton conducting ceramic fuel cell by developing a novel fabrication process using

nanosphere lithography (NSL). All the TEM works (Chapter 6 and 7) were performed by

Hee-Joon Jung.

In all ceramic fuel cell experiments, I fabricated the fuel cell samples and characterized

their electrochemical performances. Much of this work has been previously presented at

conferences and/or published in scientific journals. References to the work stemming

from this thesis are provided in context throughout the dissertation. Any inaccuracies or

errors in this dissertation are wholly my responsibility.

1.4 References

[1] B. C. H. Steel, A. Heinzel, Nature, 414, 345–352 (2001)

[2] N. P. Brandon, S. Skinner, B. C. H. Steele, Annu. Rev. Mater. Res., 33, 183–213

(2003)

7

[3] U. P. Muecke, D. Beckel, A. Bernard, A. Bieberle-Hutter, S. Graf, A. Infortuna, P.

Muller, J. L. M. Rupp, J. Schneider, L. J. Gauckler , Adv. Funct. Mater., 18, 3158–

3168 (2008)

[4] P.-C. Su, C.-C. Chao, J. H. Shim, R. Fasching, and F. B. Prinz, Nano Lett., 8, 2289

(2008)

[5] J. H. Shim, C.-C. Chao, H. Huang, and F. B. Prinz, Chem. Mater., 19, 3850 (2007)

[6] H. Huang, M. Nakamura, P. Su, R. Fasching, Y. Saito, and F. B. Prinz, J.

Electrochem.Soc., 154, B20 (2007)

[7] H. Huang, T. M. Gür, Y. Saito, and F. Prinz, Appl. Phys. Lett., 89, 143107 (2006)

[8] A. Evans, A. Bieberle-Hutter, J. L.M. Rupp, L. J. Gauckler, Journal of Power

Sources, 194, 119–129 (2009)

8

CHAPTER 2. Fuel Cell Overview

This chapter describes fundamental backgrounds and technological progresses of fuel

cells based on literatures [1, 2].

2.1 Fuel Cell Basics

A fuel cell is an electrochemical device that directly converts the chemical energy of

reactants into electrical energy. Electricity is generated from the electrochemical

reactions between reactants (fuels) and oxidants. Fuel cells are often compared with other

energy conversion devices such as batteries and combustion engines. Batteries store

electric energy chemically and generate electricity by similar electrochemical reactions to

fuel cells. One of the key differences between fuel cells and batteries is the source of

reactants. Batteries produce a certain amount of electricity based on the maximum

capacity of the battery materials and the designed systems. In contrast, fuel cells consume

a reactant from an external source and generate electricity as long as fuel (reactant) is

supplied. Also, heat engines inevitably suffer from heat losses through multiple

conversions during combusting fuels for electricity generation. However, fuel cells

extract electricity from fuels in the shortest and most efficient way.

9

Figure 2-1. Schematic of a typical hydrogen-oxygen fuel cell

Typically, a fuel cell is composed of three active components which are an electrolyte, an

anode (a fuel electrode), and a cathode (an oxidant electrode). Figure 2-1 shows a

schematic of fuel cell membrane electrode assembly (MEA) and illustrates the basic

operational principle of fuel cells for both proton and oxide ion conducting electrolytes.

To generate electricity for fuel cells, electrons can be extracted directly from fuels

through electrochemical reactions and flow through the external circuit load while ions,

either protons (H+) or negative oxide ions (O

2-), internally transport across the electrolyte.

Assuming that fuel cells run on hydrogen fuel using a proton conducting electrolyte

membrane, the following reactions take place during the operation. Hydrogen (H2) is

10

delivered to the anode and at the anode/electrolyte interface the hydrogen oxidation

reaction (HOR) happens as follows:

eHH 222 (2.1)

The generated protons are transported through the proton conducting electrolyte. At the

cathode side, transported protons and electrons react with the supplied oxygen forming

water:

OHeHO 22 222

1

(2.2)

This is a basic mechanism of electricity generation in proton conducting H2-O2 fuel cells.

11

2.2 Types of Fuel Cells

Table 2-1. Characteristics of five major fuel cell types [1].

PEMFC PAFC AFC MCFC CFC

Electrolyte Polymer

membrane

Liquid

H3PO4 Liquid KOH

Molten

Carbonate Ceramic

Charge

carrier H

+, H3O

+ H

+ OH

- CO3

2- H

+, O

2-

Operating

temperature 80

oC 200

oC 60-200

oC 650

oC 600-1000

oC

Fuel

compatibility H2, methanol H2 H2 H2, CH4 H2, CH4, CO

Fuel cells may be categorized or classified in a variety of different ways depending upon

the criteria used. Those are typically the parameters related to fuel cell design and

operation such as the type of electrolyte, the type of ion transferred thorough the

electrolyte, the type of reactants, and so on. Generally, fuel cells are categorized by the

type of electrolyte used since the material properties of the electrolyte usually determine

the properties of fuel cells, including the species of ionized charge carriers, the operation

principle and the design. There are five major types of fuel cells, differentiated from one

another by the electrolytes: the phosphoric acid fuel cell (PAFC), the polymer electrolyte

membrane fuel cell (PEMFC, this often refers to the proton exchange membrane fuel

cell), the Alkaline fuel cell (AFC), the molten carbonate fuel cell (MCFC), and the

ceramic fuel cell (CFC).

12

Fuel cells’ target operating conditions and their applications are critical factors for

determining the type of fuel cells to use. PAFC is a type of the acid-electrolyte fuel cells,

and, as the name implies, it uses phosphoric acid (H3PO4) as its electrolyte. A PAFC

normally operates at temperatures of around 170oC~210

oC. PEMFC employs the proton-

conducting polymer electrolyte membrane, usually the sulfonic acid polymer, or

NafionTM

. It has a fairly low operating temperature range, and thus it is suitable for

portable applications. However, there is a water management issue. The operating

temperature of the PEMFC is limited to 90oC or lower because the polymer membrane

must be hydrated with liquid water to maintain decent proton conductivity. The AFC

employs an aqueous potassium hydroxide (KOH) electrolyte. Depending upon the

concentration of KOH in the electrolyte, the AFC can operate at temperatures between

60oC and 220

oC. The MCFC operates at higher temperatures (~650

oC) than the fuel cells

described above. The MCFC uses a molten mixture of alkali carbonates as an electrolyte

material, and the carbonate ion (CO32-

) acts as the mobile charge carrier. The CFC

employs solid ceramic electrolytes, which can conduct an oxide ion (O2-

) or a proton (H+)

as a mobile charge carrier. The operating temperature of the SOFC is typically between

600oC and 1000

oC. As the CFCs are the subject of this dissertation, they will be

discussed in more detail below.

13

2.3 Fuel Cell Performance and Loss Mechanisms

The performance of a fuel cell device is usually evaluated with a graph of its current-

voltage characteristics. This graph is a fuel cell polarization behavior curve, or current-

voltage (I-V) curve, showing the voltage output of the fuel cell for a given current output.

The maximum voltage is determined by the difference between intrinsic chemical

potentials of the reactant and the oxidant. It is achieved when the fuel cell is operated

under the thermodynamically reversible condition. This maximum possible cell potential

is called reversible cell voltage or thermodynamic fuel cell voltage. As the current is

drawn from the fuel cell, the output voltage starts to decrease from the reversible cell

voltage. This voltage drop characterizes the irreversible losses in a fuel cell operation.

The more current is drawn, the greater these losses. There are three major types of fuel

cell losses, which give a fuel cell I-V curve its characteristic shape:

1) Activation loss (ηact) from the electrochemical reaction kinetics at the

electrode/electrolyte interface

2) Ohmic loss (ηohmic) from ionic transport thorough the electrolyte

3) Concentration loss (ηconc) due to the mass transport of fuels and oxidants to the

reaction sites

Therefore, the real voltage output for a fuel cell is the reversible cell voltage minus the

voltage drops due to these losses:

14

concohmicactthermoEV (2.3)

where V is the actual fuel cell voltage, Ethermo is the thermodynamic reversible voltage,

ηact is the activation overpotential, ηohmic is the ohmic overpotential, and ηconc is the

concentration overpotential.

Figure 2-2 shows a typical fuel cell I-V curve indicating each type of loss described

above. As shown in the figure, the activation losses mostly affect the low current region

of the curve. The ohmic losses are most apparent in the middle region of the curve and

the concentration losses are significant when a fuel cell draws a large amount of current.

The power density values can be calculated by simply multiplying the voltage and the

current density values.

15

Figure 2-2. Schematic of a typical fuel cell current-voltage (I-V) curve.

2.3.1 Thermodynamic Reversible Voltage

As previously mentioned, the maximum voltage which can be obtained for a given fuel

cell can be determined by the chemical potential difference of the species. Chemical

potential measures how the Gibbs free energy of a system changes as the chemistry of the

system changes. Chemical potential of a chemical species is expressed as follows:

iii aRT ln0 (2.4)

16

where 0

i is the reference chemical potential of species i at standard-state conditions, R

is the ideal gas constant, T is absolute temperature, and ia is the activity of species i. ia

characterizes relative concentration of species involved in a reaction of interest. The

chemical potential energy is interchangeable with a voltage as in (2.5), and changes in

Gibbs free energy for a system of i chemical species is expressed as in (2.6):

nFE (2.5)

i i

iiiii dnaRTdndG )ln( 0 (2.6)

Assume that we have a hydrogen-oxygen fuel cell system. Then, the fuel cell reaction of

the system is expressed as follows:

)()(2

1)( 222 lOHgOgH (2.7)

Using the relationship in (2.6), we can calculate the Gibbs free energy change by

inserting the chemical potentials of species involved in the reaction above and subtracting

the reactant terms from the product terms. Combining the result with equation (2.5)

allows us to estimate the thermodynamic reversible cell voltage as a function of the

chemical activity of the species:

17

2/1

0

22

2lnOH

OH

thermoaa

a

nF

RTEE (2.8)

where n is the number of charges transferred in a given reaction (n=2 in this case) and F

is Faraday constant. The term 0E contains standard state chemical potentials and refers

to the standard state reversible voltage, which is not affected by temperature.

This result is known as the Nernst equation. The Nernst equation outlines how reversible

electrochemical cell voltages vary as a function of species concentration, gas pressures,

and so on. For a general case, the Nernst equation is written as:

i

i

tsreac

products

a

a

nF

RTEE

tan

0 ln (2.9)

where i represents the stoichiometric coefficient of the activity of each species.

Therefore, if we know the operating temperature and partial pressure of the reactant

species we can calculate the thermodynamic reversible voltage. For instance, if we

operate a H2-O2 fuel cell at standard conditions (298K, using 1atm air at the cathode

side), the reversible voltage is 1.219V where 0E is 1.229V for the hydrogen-oxygen fuel

cell under standard state conditions. Hence, the theoretical maximum reversible voltage

that we can achieve from the H2-O2 fuel cell, running at 298K and ambient air condition,

is 1.219V.

18

2.3.2 Activation Losses

The activation losses associated with a fuel cell are mainly due to the overpotential

necessary to drive the electrochemical reactions occurring at the electrodes. These

activation losses usually come from the electrochemical reactions at the

electrode/electrolyte interface. These losses are referred to as ‘activation’ losses because

they are the losses (overpotentials) required to ‘activate’ or drive the chemical reactions

from their equilibrium state to the forward direction. The electrochemical reaction rate

depends on the probability of the surface reaction, and the probability ( RTGeP / ) is

exponentially dependent on the size of the activation barrier or the Gibbs energy barrier

( G ) and the temperature. The reaction rate can be expressed as follows:

RTG

R efcJ/

1

*

11

(2.10)

where J1 is the reaction rate (mol/s) in the forward direction (reactants → products), *

Rc is

the reaction surface concentration, and 1f is the decay rate to the products. The decay rate

characterizes the lifetime of the activated species and the possibility that it will convert to

a product instead of back to a reactant. The exchange current density (j) can be obtained

from the reaction rate J from the relationship of nFJj . Therefore, the forward current

19

density and the reverse current density (j2, products → reactants) can be expressed as the

following equations:

RTG

R efnFcj/

1

*

11

(2.11)

RTG

P efnFcj/

2

*

22

(2.12)

At thermodynamic equilibrium, the forward and reverse current densities must be in

balance ( 21 jj ), and there is no net current density. In other words, at equilibrium, we

have:

021 jjj (2.13)

This equilibrated current is called the exchange current density ( 0j ) for the reaction.

When a net current is produced from the electrode reaction, the electrode reaction

becomes irreversible and an imbalance of electron transfer exists. The net amount of

current flow to the electrode depends on the extent to which the potential at the electrode

differs from its equilibrium value. This electrode potential difference is defined as the

overpotential of the electrode reaction ( act ). This activation overpotential changes the

dependency of the current density on Gibbs free energy barrier:

20

)exp( 11

*

1RT

nFGfnFcj act

R

(2.14)

))1(

exp( 22

*

2RT

nFGfnFcj act

P

(2.15)

where the parameter α is called the transfer coefficient (or symmetry factor); theoretically

the value lies between 0 and 1. Using the exchange current density shown above, the

equations (2.14) and (2.15) can be expressed as:

)exp(01RT

nFjj act

(2.16)

))1(

exp(02RT

nFjj act

(2.17)

Therefore, the net current is:

)])1(

exp()[exp(021RT

nF

RT

nFjjjj actact

(2.18)

This is the Butler-Volmer equation, representing the general relation between the net

current density produced and the activation overpotential act . When the electrode

overpotential is large, the backward reaction is negligible compared with the forward

21

reaction because the second term in equation (2.18) becomes much smaller than the first

term. Hence, the Butler-Volmer equation can be reduced to

)exp(0RT

nFjj act

(2.19)

or

)ln(0j

j

nF

RTact

(2.20)

which is the well-known Tafel equation, one of the fundamental relations in

electrochemistry, representing the activation overpotential as a function of the current

density. Reducing the activation loss is one of the main objectives in this dissertation. As

shown in the equation (2.20), we can reduce the activation loss by increasing the

exchange current density ( 0j ), which is highly related to the electrochemical surface

reaction rate. Various approach and study results to enhance the surface reactions will be

introduced in later chapters (Chapter 4–6).

2.3.3 Ohmic Losses

22

The ohmic losses associated with the fuel cell operation are mainly due to the resistance

of ionic transport through the electrolyte component. They are simply governed by

Ohm’s law:

A

LiiRV (2.21)

where A is fuel cell reaction area, L is the length of the ionic transport path (normally,

thickness of the electrolyte), and σ is the ionic conductivity of the electrolyte material. In

this equation, the voltage V represents the voltage, which must be applied in order to

transport charge at a rate given by i. Thus, this voltage represents a loss. Generally,

current density value is used to compare fuel cell performance instead of current.

Therefore, it is reasonable to use area-normalized fuel cell resistance, which is known as

area-specific resistance (ASR). The equation (2.21) can be re-written using the ASR and

current density, Aij / :

L

jASRjohmic (2.22)

As shown in the equation above, we can decrease the ohmic loss either by reducing the

electrolyte thickness or by using an electrolyte material with high ionic conductivity.

23

2.3.4 Mass Transport Losses

Another source of loss in some fuel cells is known as the mass transport loss, or

concentration loss ( conc ). The concentration loss is caused by a number of processes that

hinder the transport of mass. Generally, the low solubility of reactants in the electrolyte

and the slow diffusion of reactants through the electrolyte constitute the major

contribution to the concentration loss. This can be characterized as a function of the

limiting current density ( Lj ) that represents the maximum current density drawn in the

case of consuming the full amount of supplied reactants at equilibrium:

jj

j

nF

RT

L

Lconc

ln (2.23)

And this limiting current density, Lj , can be expressed as follows:

0

Reff

L

cnFDj (2.24)

where effD is the effective diffusivity of the reactants at the electrode/electrolyte

boundaries, 0

Rc is the bulk reactant concentration, δ is the diffusion length through the

electrode. In designing a fuel cell, a large Lj helps save energy due to the mass transport.

24

To maximize the limiting current density, it is necessary to optimize electrode structure to

minimize the gas depletion effect so that the reactant is consistently high across an entire

fuel cell device as well as having an effective fuel delivery scheme (low δ).

2.4 Ceramic Fuel Cells

A ceramic fuel cell (CFC) is an all-solid-state energy conversion device and usually

refers to fuel cells employing solid ceramic electrolytes [3]. The present generation of

ceramic fuel cells can be classified into two types. One is based on oxygen ion

conducting electrolytes (SOFC), and the other one is based on proton conducting

electrolytes (PCFC or PCOFC). The main difference between the two CFC types is the

side in the fuel cell in which water is produced (the fuel side in SOFC and the oxidant

side in PCFC). Also, certain gases, such as CO, can be used as fuel in SOFCs but not in

PCFCs. CFCs usually have the highest operating temperature of all fuel cells at about

600~1000oC. It is mainly due to the significantly lower ionic conductivity of solid state

electrolytes (ceramic electrolytes) as compared to polymer membranes. Thus, the most

practical application of CFCs so far is stationary devices such as on-site electricity

generators. However, there have been many recent efforts to reduce this high operating

temperature by using new electrolyte materials or by having smartly engineered fuel cell

structures for a wider range of practical applications for CFCs. Since the CFCs usually

operate at high temperatures, liquid production is not a problem, unlike with other types

25

of fuel cells using polymer-based or liquid-based electrolytes. For this reason, CFC is

free of the severe chemical degradation of cells, which comes from corrosive liquid. This

eliminates the intermediate step of producing hydrogen through the costly reforming

process when using other types of fuels such as hydrocarbons. Also, the rigidity of the

ceramic components enables flexible design of three-dimensional fuel cell structures such

as tubular and corrugated SOFCs while other types of fuel cells are made mostly in the

two-dimensional planar format.

As previously mentioned, solid oxide fuel cells use oxygen ions as charge carriers and

therefore use an oxygen ion conducting ceramic as an electrolyte. The most popular

electrolyte material is Yttria-stabliized zirconia (YSZ), which is a doped cubic fluorite

type oxide. The dopant has different charges to its host cation, and for the charge

neutrality, it produces oxygen vacancies. Through these oxygen vacancies, oxygen ions

transport in the electrolyte. The oxide ion conducting electrolytes will be discussed in

detail in the next chapter (Chapter 3). At the cathode, the oxygen dissociates into oxygen

ions when combined with electrons from a connected external circuit. The ionized

oxygen ions transport through the electrolyte and form water by combining with

hydrogen at the anode side. The half reactions at the anode and cathode sides are:

2

2 22

1OeO (cathode) (2.25)

eOHOH 22

2

2 (anode) (2.26)

26

The cathode reaction is known as the oxygen reduction reaction (ORR) since the

electrodes are consumed by the reaction. The anode reaction is known as the hydrogen

oxidation reaction (HOR) since the electrons are released as a product. Generally, it is

agreed that the activation loss for SOFC mainly comes from the ORR due to its sluggish

process.

As previously mentioned, the high operating temperature of SOFC (600~1000oC) hinders

a wide range of practical applications, and there have been efforts to reduce the SOFC’s

operating temperature (400~600oC). In this respect, proton conducting oxide fuel cells

have attracted attentions due to higher ionic conductivity at a similar temperature regime

to oxygen ion conducting ceramics. This means that we can reduce the operating

temperature and expect similar ionic conductivity when using proton conducting oxides.

Several acceptor-doped perovskite oxides have shown proton conductivity in hydrogen-

containing environments [4]. Unlike the SOFCs, product water is generated at the

cathode side in PCOFCs. The half reactions of PCOFCs are expressed as:

eHH 222 (anode) (2.27)

)(22

12 22 gOHeOH

(cathode) (2.28)

27

Even though the half reactions are same as the polymer based fuel cells, the generated

water is gas phase due to high operating temperature, and it is free from the water

management at the cathode side. Another difference from the PEMFC is the proton

conducting mechanism. For polymer-type electrolyte membranes, the hydration of the

membrane is an important factor since protons can transport as a form of H3O+, combined

with water molecules. In the case of PCOFC, protons transport through the crystal lattice

without water. Thus, once the electrolyte membrane is sufficiently protonated, a PCOFC

can operate in dry environments [4]. The proton conduction through ceramic electrolytes

will be discussed further in Chapter 3.

2.5 References

[1] R. O’Hayre, S. Cha, W. Colella, F. B. Prinz, Fuel Cell Fundamentals, John Wiley and

Sons, New York (2006)

[2] X. Li, Principles of Fuel Cells, Taylor & Francis (2005)

[3] N. Q. Minh, T. Takahashi, J. Am. Ceram. Soc., 76, 563 (1993)

[4] K. D. Kreuer, Annu. Rev. Mater. Res., 33, 333 (2003)

28

CHAPTER 3. Materials for Ceramic Electrolytes and A

Fabrication Technique

In the previous chapter, we discussed the basic fuel cell fundamentals and fuel cell types,

especially ceramic fuel cells. This chapter describes ceramic electrolyte materials for both

oxygen ion and proton conducting electrolytes. The pulsed laser deposition (PLD)

technique will be presented as a deposition method of the ceramic electrolytes.

3.1 Ceramic Electrolyte Materials

3.1.1 Oxygen Ion Conducting Ceramics

There are numerous candidate materials for SOFC electrolytes. One of the major

requirements for the electrolyte is having conductivity for oxygen ions over a wide range

of O2 partial pressures without having electronic conductivity. Another requirement is

chemical stability since the electrolytes are exposed to strongly oxidizing and reducing

environments. Also, it requires mechanical stability of the membrane over high operating

temperatures.

29

Perhaps the most conventional fast oxygen ion conducting materials have crystal

structures of fluorite type AO2, where A is a tetravalent cation [1-3]. And the best known

fluorite type oxygen ion conductor is acceptor-doped ZrO2. Pure zirconia is not a good

ion conductor. However, once acceptor dopants are introduced onto the cation sublattice,

oxygen vacancies are generated for the charge neutrality. Through these oxygen

vacancies, oxygen ions can diffuse in the crystal structure. This can be expressed using

Krӧger-Vink notation:

2

'' ZrOVMOZrMO OZr

x

O

x

Zr (3.1)

2

'

32 222 ZrOVROZrOR OZr

x

O

x

Zr (3.2)

where M is a divalent cation (i.e. Ca), R is a trivalent cation (i.e. Sc, Y, Ln), and

OV is a

compensating oxygen vacancy [3-4]. The oxygen vacancy concentration is a critical

factor which affects the ionic conductivity of the oxygen ion conducting electrolyte:

RT

RTGDzFc act )/exp()( 0

2 (3.3)

in which c is the vacancy concentration, z is the charge of the carrier (-2 for the oxygen

ions), F is the Faraday constant, D0 is the oxygen self-diffusion coefficient of the

30

material, actG is the activation energy barrier for diffusion, R is the gas constant, and T

is the absolute temperature [1].

From the equation above, the ionic conductivity increases as the vacancy concentration

increases. However, there is an upper limit to the amount of doping and beyond this limit

the conductivity starts to decrease. As we have more and more vacancies, the electrostatic

interaction between dopants and vacancies also increases. This ultimately impedes

oxygen ion and oxygen vacancy mobility. Thus, there is an optimal dopant concentration

value which yields the maximum ionic conductivity. For yttria-stabilized zirconia (YSZ),

which is the most commonly used electrolyte for SOFC, the optimum dopant

concentration is around 8 mol% [1-2].

As stated above, YSZ is arguably the most commonly used electrolyte material for SOFC

due to its superior chemical stability in both oxidizing and reducing environments.

Despite the chemical stability, YSZ has low ionic conductivity, about 0.02 S/cm at 800oC

and 0.1 S/cm at 1000oC and has high activation energy [2]. This is the main reason that

SOFCs have such high operating temperatures. To make the use of YSZ suitable for

intermediate (500~700oC) and low (300~500

oC) operating temperature regimes, there

have been efforts to decrease the electrolyte thickness [5-7] because the ionic

conductivity is a function of membrane thickness (Chapter 2).

31

Another popular oxygen ion conducting electrolyte material is doped ceria (CeO2).

Typical dopants include gadolinia (Gd2O3) and yttria (Y2O3), producing GDC and YDC,

respectively. These doped ceria have much higher ionic conductivities than YSZ,

especially at lower temperature regimes (500~700oC) and lower activation energies. For

example, 10 mol% doped GDC (Ce0.9Gd0.1O1.95) has an ionic conductivity of 0.01 S/cm

at 500oC [8]. This is why the doped ceria materials are attractive for the use of lower

operating temperature SOFC electrolytes. In addition to the higher ionic conductivity, the

oxygen surface exchange coefficient is much higher than that of YSZ [9-10]. In overall

electrode kinetics, the rate of exchange of oxygen between gas phase and the oxide

electrolyte is considered as the rate limiting step. Thus, this rate of surface exchange can

determine the performance of oxide electrolyte and electro-catalytic materials [11].

However, doped ceria materials do have a significant disadvantage in SOFC electrolyte

applications. At high temperature (above 600oC) and low oxygen partial pressure

environment, the conductivity is not purely ionic. Under reducing conditions, Ce4+

partially reduces to Ce3+

and this induces n-type electronic conductivity, which may lead

to internal electronic short circuits. As we increase the operating temperature, this

problem increases. In response to this stability issue of doped ceria materials in a

reducing environment, composite electrolytes (YDC/YSZ, GDC/YSZ) have been

investigated. By utilizing doped ceria at the cathode side and YSZ at the anode side, we

can enhance the oxygen exchange kinetics and achieve chemical stability. Investigation

32

details of these YDC and GDC cathodic interlayered SOFCs will be further discussed in

chapter 5.

3.1.2 Proton Conducting Ceramics

Proton transport in several members of acceptor-doped perovskites of the general formula

ABO3 is reported to be fast, with activation energies of about 0.45eV [12], which makes

them of interest as potential solid electrolytes for next generation protonic devices such

as fuel cells, electrolyzers, hydrogen sensors, and gas reformers [13-17]. Proton transport

in these anhydrous oxides occurs via the Grotthuss mechanism through hydroxide defects

that are produced by water incorporation into oxide ion vacancies generated in the crystal

lattice upon extrinsic doping of the tetravalent site by the trivalent ion. In Krӧger-Vink

notation, this process can be written as:

O

x

OO OHOVOH 22 (3.4)

The detailed mechanism of charge transport in proton conducting oxides was reviewed in

recent publications [ref. Kreue rev, 17-19]. The proton in a hydroxide defect,

OOH ,

resides asymmetrically in an interstitial position near the oxide ion. Because of the

softness of this dynamic hydrogen bonding between the proton and its eight nearest

oxygen neighbors, the protons are highly mobile in these cubic oxide structures, giving

33

rise to high ionic conductivities with relatively low activation energies [18-19]. Charge

transport involves protons jumping from one oxide ion to an adjacent one in the oxide

sublattice along the edges of the BO6 octahedra. Each jump is followed by a rotation of

the proton around the oxide ion to reorient itself for the subsequent jump.

Initially, the proton conductivity in doped perovskites was investigated with alkaline-

earth cerates and zirconates where the trivalent Ce4+ sites were partially substituted by

trivalent dopant cations such as Y3+

, Gd3+

, or Nb3+

. Despite the high ionic conductivity of

doped cerate perovskites, they suffer from significant chemical instability in CO2

environments. Several studies reported that SrCeO3 and BaCeO3 easily turn into SrCO3

and BaCO3 through reaction, even with a small amount of CO2 [12]. Also, it is reported

that BaCeO3 easily decomposed into Ba(OH)2 in the presence of water. In contrast,

alkaline-earth zirconates, especially Y-doped BaZrO3 (BYZ) has been considered one of

the most promising electrolyte materials for proton conducting fuel cells for their high

proton conductivity and excellent chemical stability [20]. In a later chapter (Chapter 6),

we will present the results of the fuel cell performance test with three-dimensional fuel

cell architecture design using a BYZ electrolyte. Moreover, detailed investigations of

BYZ thin film properties will be discussed later in Chapter 7.

34

3.3 Pulsed Laser Deposition

3.3.1 Background

Pulsed laser deposition (PLD) is a unique physical vapor deposition (PVD) technique

which has gained considerable attention for its good reproducibility of target material

properties such as chemical compositions. The technique uses high power laser pulses to

melt, evaporate and ionize material from the surface of a target. The vaporized or ablated

material from the target travels through a generated plasma plume and is deposited on a

substrate. Figure 3-1 shows a simple schematic of PLD system.

35

Figure 3-1. Schematic of PLD mechanism (Top) and a sample picture of PLD process:

deposition of BYZ on MgO (100) substrate.

Most of the PLD systems use gas excimer and Nd3+

: YAG lasers. Generally, the useful

wavelength spectra for PLD thin film deposition are between 200nm and 400nm since

most materials for PLD show strong absorption in this spectral region. Therefore, Nd3+

:

36

YAG laser, which has an operating wavelength of 1064nm, uses a beam attenuator to

adjust frequency down to 266~355nm. Unlike Nd3+

: YAG lasers, gas excimer lasers emit

their radiation directly in the ultra violet region. Table 2-1 shows the gas excimer

wavelengths for commercial laser systems [21].

Table 3-1. Wavelength of excimer lasers

Excimer Wavelength (nm)

F2 157

ArF 193

KrCl 222

KrF 248

XeCl 308

XeF 351

For effective melting of targets and evaporation of particles, laser light should be strongly

absorbed on the target with minimum optical absorption. The thermal diffusion length

(Lt) is given by

2

1

2

mol

tcn

tL

(3.5)

where δt is the pulse duration of the laser, κ is the thermal conductivity of the target, c is

the molar heat capacity of the target, and nmol is the molar density of the target. The

37

values of Lt for oxides and metals are typically in micrometers. Considering the optical

absorption length of those targets is less than 10nm, the laser in PLD is effectively used

for evaporation with strong thermal absorption on the surface of the target [21-23].

There are several factors that influence PLD process: target-sample distance, laser

energy, background gas (pressure), and substrate temperature. Among them, the laser

energy density is considered the most critical factor. Since the laser beam focuses on a

small area (2-3mm2) with high energy (60-80mJ), the energy density is quite high (3-

4J/cm2). Therefore, temperature at the local target surface is very high (about 10000K)

which allows thermal evaporation and deposition of basically all kinds of materials. The

plume also interacts with the laser, and evaporated particles acquire high kinetic energy

up to 10eV for neutral species and 1000eV for ionized species. Due to their large kinetic

energy and mobility, those particles can form dense and highly organized films on the

substrate. If we use a single crystalline substrate, the particles tend to from a film in a

highly organized fashion, and we can achieve an epitaxial PLD layer. However, kinetic

energy acquired by laser is not preserved when the chamber is filled with process gases,

usually inert gas or oxygen for oxide film deposition. Bombardments between the

evaporated particles and ambient gas molecules decrease their kinetic energy and, as a

consequence, their mean free path is reduced from kilometers to microns. The

background gas and its pressure also affect the area and the characteristics of PLD films.

Therefore, the background gas is another critical factor for PLD [21, 23].

38

The PLD technique has significant benefits over other thin film deposition methods: 1)

The capability of exact stoichiometry transfer of material from target to substrate, i.e. it

can reproduce the chemical composition of the target even though it is a complex

material. 2) Since it has the exact stoichiometry transfer capability, PLD has a wide,

almost unlimited, range of materials selection. 3) PLD has relatively high deposition

rates, typically ~100Å /min, with precise deposited film thickness by turning on and off

the laser power. 4) The use of a carousel, which holds multiple targets, enables in-situ

deposition of multilayer films without breaking the vacuum when changing between

materials. Despite these significant advantages, the area of material deposition (about

1~2cm2) is quite small in comparison to that required for industrial applications (required

area coverage of 7.5 x 7.5cm2). There have been efforts to solve this problem and it has

been solved to a large extent by using line-focus laser spots.

3.3.1 Example Materials of Pulsed Laser Deposition

As previously mentioned, the strongest benefit of PLD is exact stoichiometry transfer of

materials. In other words, the stoichiometry of the target material is preserved in the films

since the intensive laser effectively evaporates all the components or ions on the surface

at almost the same rate. For this reason, target materials are required to possess

homogeneity in composition. This characteristic is beneficial especially for deposition of

complex compounds including defect-contained oxide materials such as

39

(Y2O3)0.08(ZrO2)0.92 or superconducting materials like YBa2Cu3O7 (YBCO). Generally,

the PLD process is known as a technique producing the best quality YBCO. Table 3-2

shows the example materials deposition using PLD as summarized by Chrisey et al. [21].

40

Table 3-2. Example materials deposited by pulsed laser deposition and applications of

those materials [21].

41

3.3 References

[1] R. O’Hayre, S. Cha, W. Colella, F. B. Prinz, Fuel Cell Fundamentals, John Wiley and

Sons, New York (2006)

[2] X. Li, Principles of Fuel Cells, Taylor & Francis (2005)

[3] L. Malavasi, C. A. J. Fisher, S. Islam, Chem. Soc. Rev., 39 4370 (2010)

[4] H. Tuller, A. S. Nowick, J. Electrochem. Soc., 122, 255 (1975)

[5] H. Huang, M. Nakamura, P. Su, R. Fasching, Y. Saito, and F. B. Prinz, J.

Electrochem.Soc., 154, B20 (2007)

[6] H. Huang, T. M. Gür, Y. Saito, and F. Prinz, Appl. Phys. Lett., 89, 143107 (2006)

[7] J. H. Shim, C.-C. Chao, H. Huang, and F. B. Prinz, Chem. Mater.,19, 3850 (2007)

[8] V. V. Kharton, F. M. B. Marques, A. Atkinson, Solid State Ionics, 174, 135 (2004)

[9] B. C. H. Steele, Solid State Ionics, 75, 175 (1995)

[10] B. C. H. Steele, K. M. Hori, S. Uchino, Solid State Ionics, 135, 445 (2000)

[11] B. C. H. Steele, J. A. Kilner, P. F. Dennis, A. E. McHale, Solid State Ionics, 18-19,

1038 (1986)

[12] K. D. Kreuer, Ann. Rev. Mat. Res., 33, 333 (2003)

[13] W. Münch, G. Seifert, K. D. Kreuer, J. Maier, Solid State Ionics, 97, 39-44 (1997)

[14] N. Bonanos, B. Ellis, M. N. Mahmood, Solid State Ionics, 44, 305-11 (1991)

[15] N. Kuwata, N. Sata, T. Tsurui, H. Yugami, Jpn. J. App. Phys., 44, 8613-8618

(2005)

42

[16] T. Hibino, A. Hashimoto, M. Suzuki, M. Sano, J. Electrochem. Soc., 149, A1503-8

(2002)

[17] H. Iwahara, H. Uchida, K. Morimoto, J. Electrochem. Soc., 137, 462-465 (1990)

[18] K. D. Kreuer, S. J. Paddison, E. Spohr, M. Schuster, Chem. Rev., 104, 4637 (2004)

[19] K. D. Kreuer, E. Schonherr, J. Maier, Solid State Ionics, 1, 70-71 (1994)

[20] H. Iwahara, T. Yajima, T. Hibino, K. Ozaki, H. Suzuki, Solid State Ionics, 61, 1-3

(1993)

[21] D. B. Chrisey, G. K. Hubler, Pulsed Laser Deposition of Thin Films, John Wiley &

Sons (1994)

[22] M. Ohring, The Materials Science of Thin Films, Academic Press (2001)

[23] D. B. Chrisey, J. S. Horwitz, P. C. Dorsey, J. M. Pond, Laser Focus World, p.155,

May (1995).

43

CHAPTER 4. Nano-pore Structured Platinum Electrode

Arrays for Low Temperature SOFCs

This study presents the fabrication and investigation of morphologically stable model

electrode structures with well-defined and sharp platinum/yttria stabilized zirconia (YSZ)

interfaces to study geometric effects at triple phase boundaries (TPB), which is the actual

electrochemical reaction site for fuel cells. These nano-structured dense Pt array cathodes

exhibited better structural integrity and thermal stability at the fuel cell operating

temperature of 450~500oC when compared to porous sputtered Pt electrodes. More

importantly, electrochemical studies on geometrically well-defined Pt/YSZ sharp

interfaces demonstrated that the cathode impedance and cell performance both scale with

aerial density of TPB length. By controlling the density of TPB we could optimize and

maximize the fuel cell performance.

4.1 Introduction

Solid oxide fuel cells (SOFC) are efficient energy conversion devices that are being

developed for practical applications. Due to large activation energies (~1eV) for oxide

ion transport in solid oxide electrolytes and relatively sluggish oxygen reduction reaction

44

at the cathode, SOFCs are usually operated at elevated temperatures (800~1000oC) to

obtain practically meaningful fluxes and fuel cell performance. Typically, an SOFC

element is made of a yttria-stabilized zirconia electrolyte layer, a mixed conducting

ceramic cathode such as La1-xSrxCo1-yFeyO3 (LSCF) and La1-xSrxMnO3-δ (LSM), and a

cermet anode such as Ni/YSZ.

Operation of SOFCs at elevated temperatures may be desirable for enhanced kinetics and

transport purposes, but pose serious challenges in microstructural and thermal stability,

seal integrity, aging and degradation, thermal cycling, and cost of materials and

fabrication. To mitigate some of these problems, recent efforts have been aimed towards

lowering the operating temperature of SOFCs to intermediate temperature regime of 600-

800oC, i.e., IT-SOFCs [1-4]. Although oxygen chemical diffusion in mixed conducting

electrodes such as LSCF and LSM is relatively fast at elevated temperatures [5-9], their

catalytic activities and transport rates decrease precipitously with temperature that leads

to increased activation losses at intermediate temperatures [1, 10-13].

In recent years, the thrust of our research effort has been aimed to further lower the

operating temperature of SOFCs to a regime between 300 and 500oC by employing thin

film structures of the YSZ electrolyte and electrodes [14-17]. Expectedly, use of mixed

conducting ceramic cathodes for such low temperature (300~500oC) SOFCs (LT-SOFCs)

would not be meaningful due to severe losses expected from large overpotentials at the

electrodes. Hence, platinum is selected as a model electrode material in this study not

45

only to enhance the fuel cell performance but also to provide a stable and geometrically

well-defined platform to examine the triple phase boundary effects in SOFCs. In this

relatively low temperature regime, Pt remains to be the best catalyst to enhance the rate

for the oxygen reduction reaction at the cathode that is critically important to minimize

the activation loss, and to improve the fuel cell performance of LT-SOFCs.

The primary purpose of this study is to control the geometry of the Pt/YSZ interface that

makes it possible to investigate TPB effects in a quantifiable manner. Although similar

patterning efforts have been reported in the literature, only a handful has considered the

Pt/YSZ system [18, 19], while most were intended for mixed oxide cathodes such as La-

Sr-Mn-O [20, 21].

The methodology developed in this study allowed fabrication of morphologically stable

Pt array electrodes that exhibit sharp and well-defined triple phase boundaries (TPB).

The strategy was to define the geometry of the three phase contact interfaces while

providing a dense and nonporous Pt layer that constrains the charge transfer reaction to

TPB only as well as eliminating porosity effects at the Pt/YSZ interface under the Pt

layer. This patterning approach opens up the opportunity for other researchers to

systematically study TPB effects in SOFCs. The results presented here successfully

demonstrate that controlled TPB density correlates well with cell performance.

46

The charge transfer reaction at the cathode involves reduction of oxygen at the TPB,

where the gas, catalytic cathode, and the electrolyte are all in physical contact. Due to

relatively high activation energy of >1.5eV for the oxygen reduction reaction [22-27], it

is generally agreed that the processes at the cathode govern the overall behavior of

SOFCs even at elevated temperatures [1]. Naturally, cathodic overvoltage becomes even

more pronounced at the low operating temperatures employed in this study. To mitigate

this effect and to enhance fuel cell performance, it is not only desirable to maximize the

TPB density but also to employ Pt catalyst at the cathode to improve the reaction kinetics

for oxygen reduction in LT-SOFCs. However, performance maximization is not the focal

point of this study at this point.

Typically, DC sputtering is employed to deposit randomly structured porous Pt electrodes

that increase TPB density for improved fuel cell performance. However, this technique

poses challenges to quantify the TPB geometry and to carry out systematic studies of

TPB effects in SOFCs because control, quantification, and stability of the Pt/YSZ

interface morphology are not trivial issues. It is difficult to characterize the

morphological details of the Pt/YSZ interface with sufficient precision [28-31]. Also,

time-dependent changes [32], such as Ostwald ripening of the sputtered porous Pt

electrodes that can lead to microstructural coarsening and degradation complicate the

problem further more. Indeed, recent work in our laboratory has shown that sputtered

porous pure Pt electrodes are not thermally stable even at the low operating temperatures

of LT-SOFCs [33]. Most importantly, however, it is difficult to quantify and define the

47

exact geometry, scale and nanostructure of the TPB. This makes quantitative

investigation of the rate processes at the TPB challenging.

It is for these reasons that the present study reports on a new non-lithographic patterning

technique for fabrication of dense Pt electrodes with easily tunable and well-defined TPB

geometry. The intent is not to achieve improved cell performance but rather, to establish

a methodology that allows systematic investigation of TPB characteristic and scaling

behavior.

This study reports a dense patterned Pt architecture that forms sharp, stable Pt/YSZ

interfaces with well-defined geometries, and hence, allows determination of TPB linear

density with good accuracy. Since the Pt electrode layer is nonporous, this architecture

also restricts the charge transfer reaction to this geometrically defined interface.

Furthermore, the present study demonstrates superior microstructural and thermal

stability of such patterned electrodes for improved and consistent SOFC performance.

The nanosphere patterning (NSP) method provides the ability to vary the TPB geometry,

which makes it possible to investigate TPB characteristics in a controlled manner and it

gives implications optimizing the LT-SOFC’s electrode/electrolyte interface structures.

48

4.2 Experimental

4.2.1 Spherical Silica Particle Fabrication

For the use of masking solution, monosize spherical silica particles were created by a

modified Stöber synthesis technique using tetraethyl orthosilicate (TEOS) [34-36]. The

TEOS molecules are easily converted to silicon dioxide via a series of condensation

reactions. The TEOS molecules are easily converted to silicon dioxide via a series of

condensation reactions.

(4.1)

The reaction rate is very sensitive to the presence of catalytic solutions. In this

experiment, a pH value of about 12 was employed in a mixture of 38% ammonium

hydroxide and ethanol as a basic catalyst for the reaction. Size of the spherical silica

particles can be controlled by varying the concentration of TEOS and the catalyst. Thus,

different sizes of silica nanopsphere particles were prepared and were used as mask for

subsequent patterning of the Pt cathode layers (Figure 4-1). After synthesis of the SiO2

particles, the surface of the particles was functionalized with aminopropyl

methyldiethoxysilane (APDES) to terminate them with positively charged amine group

and this step prevents aggregation of the particles.

49

Figure 4-1. Synthesized spherical silica particles with different diameters: (a) 130nm, (b)

300nm, (c) 650nm.

4.2.2 Nanosphere lithography (NSL) for nano-structured electrodes

Monosize spherical silica particles were transferred onto a surface of the YSZ substrate

using a Langmuir-Blodgett (LB) trough technique demonstrated by Cui and co-workers

[37]. This technique allows transferring the silica particles from the surface of water by

dipping a desired substrate (Figure 4-2). Initially, the LB trough was filled with DI water.

Then the prepared silica particles were introduced on to the surface of the water by slow

injection. Using the compression bars in the trough, a monolayer of nanospherical

particles was produced at the surface. The substrate was then dipped into the trough and

pulled out at a constant rate. In this experiment, the substrate is a 1cm by 1cm, 8 mol%

yttria-stabilized zirconia (YSZ) single and polycrystalline wafer obtained from

Marketech Inc. The transfer process involved slow insertion of the YSZ substrate

50

vertically into the LB trough and removing the adhered spherical particles as a film from

the liquid surface by steadily withdrawing the YSZ substrate from the liquid at a finite

rate. This method creates a close-packed monolayer of silica particles on the YSZ

substrate.

Figure 4-2. Schematic of the Langmuir –Blodgett trough

After forming a close-packed monolayer of spherical particles on the substrate,

anisotropic plasma etching with a gas mixture of O2 and CHF3 was used to uniformly

reduce the size of the silica particles in order to open up space between them for

depositing Pt through the interspacing between the particles. A dense Pt layer is then

deposited on the cathode side by DC sputtering under 100W plasma power and 1Pa of Ar

pressure for about 60 seconds at room temperature. This yielded a nonporous Pt film of

about 60nm in thickness [16]. Then the silica particles were removed mechanically from

51

the YSZ substrate using ultra sonication. On the anode side of the substrate, 80nm thick

porous Pt layer was deposited by DC sputtering under 50W power, 10Pa of Ar pressure

for 150 seconds at room temperature. With that, the fabrication of the SOFC membrane

electrode assembly (MEA) was complete. This novel process yields a patterned Pt

cathode layer with well-defined geometry that enables the study of TPB geometric effects

in a systematic and controlled manner (Figure 4-3 and 4-4). In addition, by appropriately

varying the silica particle size and the spacing between particles, it is possible to vary and

optimize the TPB density on the YSZ surface in a controlled manner.

Figure 4-3. Fabrication process schematic of nano-pore structured electrode by using

nanosphere lithography (NSL) technique.

52

Figure 4-4. SEM images of nano-pore structured dense Pt electrodes.

For the fuel cell performance characterization, we used a homemade experimental

chamber with a micro-manipulating probe station (Figure 4-5). During the measurements,

pure dry hydrogen is provided at the anode side as a fuel and air is used as the oxygen

source at the cathode side. Constant temperature is maintained by a temperature

controlling unit. For electrochemical characterization, a Gamry Potentiostats (Gamry

Instruments) unit is used for collecting both I-V characteristic data and Electrochemical

Impedance Spectra (EIS) in the frequency range of 300 kHz to 0.1 Hz with an AC

amplitude of 50mV.

53

Figure 4-5. Schematic illustration of the probing station for electrochemical

characterization of fuel cell MEAs.

4.3 Results and Discussion

4.3.1 Thermal stability of the nano-pore structured electrode

SOFCs that operate at elevated temperatures typically do not employ Pt electrodes for

obvious reasons. However, for LT-SOFCs the use of Pt electrodes is a near-necessity due

to sluggish reaction kinetics at these low temperatures. Micro-SOFCs utilizing 50 to 750

nm thick YSZ membranes and porous Pt electrodes have been studied and power

densities in the range 0.02 to 152mW/cm2 have been reported [38]. Comparatively,

similar work in our laboratory has demonstrated record power densities up to

861mW/cm2 at 450

oC [14-16]. Long-term stability of the sputtered Pt electrodes reported

in these studies, however, has not been fully documented or investigated. Similarly, there

is not a coherent understanding or establish process for consistent fabrication of stable Pt

54

electrode morphologies that also display optimum performance. Previously, evidence

from our laboratory has indicated that porous Pt electrodes fabricated by DC sputtering

undergo morphological changes during cell operation even at these low temperatures,

while stable morphology and performance was demonstrated with Pt-Ni alloy electrodes

[33]. Based on these observations, dense Pt electrodes with geometric openings to create

well-defined and stable TPB was pursued in the present study.

Figure 4-6. SEM images of (a) as sputtered porous Pt layer, (b) after short time (~

30mins) operation of fuel cell at elevated temperature (450oC), clearly indicate a dramatic

change in Pt morphology and a proportionate reduction in the TPB density.

55

Figure 4-7. Potentioamperometry data at 0.6V, comparing the behavior of SOFC MEA

with porous Pt electrode and SOFC with nano-pore structured Pt electrode. Measurement

was conducted for 12 hours continuously at 500oC. (a) Absolute output current densities

indicating severe degradation in performance of porous Pt within a short time as opposed

to stable behavior of patterned dense Pt. (b) Normalized current densities plot showing

relative amounts of degradation from the initial performance.

Initial performance of fuel cell sample using randomly porous Pt electrode is higher than

the sample using nano-pore (e.g. patterned with 400nm pores) structured electrode.

Although the fuel cell MEA with porous Pt electrodes outperformed that with patterned

Pt cathode in the beginning of cell testing, its performance dramatically decreased within

a short time due to electrode degradation. The change in the morphology of the DC

sputtered porous Pt electrodes before and after fuel cell testing at elevated temperature is

shown in Figures 4-6(a) and 4-6(b). It is clear that even at the moderately low

(400~450oC) operating temperature of LT-SOFC, significant changes take place in the

microstructure of the porous Pt electrode driven likely by its high surface energy.

However, due to the visibly evident changes in the Pt morphology and equally so at the

Pt/YSZ interface, reduction in the TPB density leads to a rapid degradation in the fuel

56

cell performance until it seems to stabilize at a significantly lower value. This is shown in

Figure 4-7, which compares the fuel cell performance of sputtered porous Pt electrodes

with patterned Pt electrodes under potentiostatic conditions. After completion of the I-V

experiments for both samples, the output currents of the two cells were monitored for 12

hours at 500oC at a constant voltage that corresponds to the peak power density value.

Despite the noise in the data that came from the measurement setup, particularly from the

on/off type temperature controller and the fume hood fan, Figure 4-7 clearly indicates a

stable performance by the nano-structured Pt electrode, while the current output from the

porous Pt cell clearly shows a rapid decay over time. In addition, a morphological

comparison provided in Figure 4-8 demonstrates the stability of the nano-structured Pt

electrode after running under potentiostatic conditions for 12 hours at 500oC. This

morphological degradation in porous Pt electrodes occurs even at temperature lower than

500oC and becomes observable within a short period of time. Unlike the porous Pt

cathodes, the patterned Pt morphology indicated no discernable change in the

microstructure at the test temperatures and under load in a SOFC configuration. Clearly,

the sharp interfacial edges of the Pt layer, representing the location of TPB, are uniformly

well preserved even after prolonged use. The results demonstrate that the nano-patterning

method employed in this study helped stabilize the morphology and provide thermal and

microstructural stability at fuel cell operating temperatures.

57

Figure 4-8. High resolution SEM images of nano-structured fresh Pt electrode before

running (top (a) and tilted (c) views), and after running chronoamperometrically for 12

hours at 500oC (top (b) and tilted (d) views). Clearly, nano-structured Pt does not show

any major morphological change and the TPBs are well conserved after long operation.

58

Figure 4-9. V-I-P comparison of SOFC at 450oC employing DC sputtered porous Pt

electrode versus nano-pore structured Pt electrode at the cathode.

Furthermore, nano-patterned Pt electrodes, when optimized, have the potential to provide

better cell performance due to controlled and stable TPB geometries that also offer ready

access to the gas phase. Indeed this is demonstrated in Figure 4-9, which compares the

performance of a SOFC MEA at 450oC featuring a nano-patterned Pt electrode of

approximately 400 nm diameter pore size with that of a MEA with DC sputtered porous

Pt electrodes measured after the cell performance is stabilized at the operating

temperature. It is clear that the cell featuring nano-patterned Pt performs better than the

one having porous sputtered Pt electrode. This result suggests either a higher density of

TPB for the nano-patterned Pt electrode, or possibly a significant amount closed porosity

in the case of the sputtered Pt electrode likely due to structural coarsening and

degradation, which hinders direct access from the gas phase, even though it might have

initially possessed a larger Pt/YSZ interfacial contact length. The maximum power

59

density values obtained with fuel cell MEAs with nano-patterned cathodes are still orders

of magnitude lower than previous reports from our laboratory on record performance of

LT-SOCFs using ultra-thin film YSZ electrolyte and porous Pt electrode [14-16]. This is

primarily due to two reasons, namely, that the TPB density of nano-patterned cathodes is

naturally and significantly lower than porous Pt cathodes, and also, polycrystalline

100µm thick bulk YSZ wafers were used as the electrolyte in the present study.

Therefore, it may not be meaningful to directly compare these performance values due to

the combined effect of the electrolyte thickness and deposition methods, which generate

different microstructures. Again, this study presents a fabrication methodology for nano-

pore structured electrodes not for the purpose of achieving the highest performance but

for investigation of electrochemical behavior of TPB geometry and possibly its

optimization.

4.3.2 Investigation of TPB scaling behavior

The sharp interfacial geometry of the open pores in a close-packed fashion makes it

possible to approximate the TPB density (cm/cm2) for a given patterned area. With the

help of SEM images, the TPB densities for the two patterned Pt cathodes were

approximated to be 8.93 x 104

cm/cm2 for cathodes with 300 nm opening and 5.07 x 10

4

cm/cm2 for cathodes with 400 nm opening (Figure 4-10). The SOFC MEA with the

smaller initial diameter of the silica particle showed about a factor of 1.8 higher TPB

60

density than the SOFC MEA with the larger size of the starting silica particles. Naturally,

the size of the pore opening as well as the interspacing between openings for a given

geometric area determines the TPB density. The results presented below correlate the

TPB scaling factor with cell performance and behavior.

The performance of fuel cells featuring two different nanopore sizes, or respective TPB

densities, is presented in Figure 4-11(a). As previously mentioned above, the SOFC

MEA with higher density TPB of about 8.93 x 104 cm/cm

2 of projected area (pore

diameter ~300nm) at the cathode shows better performance than the SOFC with coarser

(~400 nm) cathode openings with TPB density of about 5.07 x 104 cm/cm

2. These values

are larger than 15-150 cm/cm2 TPB densities reported for patterned Pt electrodes on YSZ

[19]. Peak power densities and open circuit voltages (OCV) of the patterned Pt cathodes

with pore openings of 300 nm and 400 nm pore size are 2.2mW/cm2

and 1.04 V, and

1.4mW/cm2 and 1.01 V, at 450

oC, respectively at about a cell operating voltage of

0.45V. Also the coarser TPB sample shows a larger exponential drop at low current

region, indicative of a larger activation loss than the denser TPB sample. However, the

differences in their relative performances vanish when these plots are normalized with

respect to the areal TPB densities of the two MEAs. This is expected since the charge

transfer reaction is restricted to TPB whose linear length per unit area is determined by

the masking process during patterning. The results are shown in Figure 4-11(b) where the

I-V-P plots of the 300 nm and 400 nm patterned MEAS overlay on top of each other in

61

almost perfect registry and agreement on the low current regime, where activation losses

and hence, the charge transfer along TPB dominates behavior.

Figure 4-10. Nano-pore structured dense Pt electrode with different final pore sizes and

TPB density, (a) 300nm, (b) 400nm. SEM images were taken with the same

magnification and the image with smaller initial particle size shows denser pores, which

relates to increased TPB density.

62

Figure 4-11. (a) Fuel cell I-V measurement of SOFC samples at 450oC with structured

electrode with different pore diameters. (b) V-I-P plot where current is normalized by

TPB density. The plots overlay in good registry especially in the activation regime as

expected, and indicates that fuel cell performance scales with TPB density.

Further characterization of the MEAs was carried out by electrochemical impedance

spectroscopy (EIS). Figure 4-12 compares the EIS spectra at 450oC of SOFC samples

with nano-patterned Pt cathodes of different pore opening sizes under 0.6 V DC biased

63

condition. From the EIS Nyquist plot of SOFC with YSZ electrolyte, it is well known

that the high frequency arc corresponds to the ohmic resistance of the electrolyte and the

low frequency arc corresponds to the electrochemical reaction at the electrolyte/electrode

interface [34]. For SOFCs, there is general agreement that cathode reaction kinetics is

much slower than that for the anode reaction, and hence, the cathode reaction dominates

the activation loss [3]. Assigning the low-frequency arc in EIS spectra to the ORR

(oxygen reduction reaction) process as suggested by others [39, 40], one can conclude

that the cathode impedance is reduced for the 300 nm pore size Pt electrode due to

increased TPB density as compared to the 400 nm pore size Pt having a lower TPB

density, and hence, a larger cathode impedance. Clearly, EIS impedance data provides

further support for faster or improved oxygen reduction kinetics at the cathode by

increasing the TPB density and the charge transfer reaction sites [41].

From the EIS spectra in Figure 4-12, it is possible to determine the corresponding

cathode resistance for the two samples. The electrode resistance of the coarser TPB MEA

(~2600 ohms) is a factor of 2.06 higher than the MEA with denser TPB (~1260 ohms),

corresponding to smaller pore openings. Although the scaling factor does not perfectly

match the ratio of 1.8 between the TPB densities of the two samples, it does suggest

within experimental error that the electrode resistance scales almost linearly with the TPB

density.

64

Figure 4-12. EIS Nyquist spectra of SOFC samples featuring nano-pore structured dense

Pt cathodes with different TPB densities.

Furthermore, a similar scaling behavior is observed between the TPB density and the

peak power densities of the two samples shown in the I-V curves of Figure 4-11. Indeed,

the ratio of 1.6 between the peak power densities of the two samples agrees within

experimental error with the TPB ratio of 1.8 between the two patterned cathode samples,

suggesting again the possibility of a linear scaling law.

Considering the fact that the linear length of the Pt/YSZ interface per circular opening in

these patterned cathodes is about 1260nm for the 400nm diameter opening, and

correspondingly 940nm for the 300nm diameter opening, the results above further

suggest that the TPB width under the test conditions is significantly smaller than its

65

length, possibly in the several nanometers to tens of nanometers range in general

agreement with estimated literature values of the same order of magnitude for this

operating condition [18]. This yields a near-linear scaling relationship and confirms the

argument that the TPB is actual electrochemical reaction site for SOFC.

4.3.3 TPB width estimation for Pt/YSZ interface

Figure 4-13. (a) Schematic of the TPB. (b) Cross-section image showing TPB width.

As previously mentioned, the actual electrochemical reaction site for charge transfer is

called TPB. This TPB is where electrolyte, reacting gas, and catalytic electrode are in

physical contact. The red lines in Figure 4-13(a) show schematic of TPB. Though the

TPB is shown as a line, it actually has its own width (Figure 4-13(b)). If the TPB width

overlaps each other, we are not fully utilizing the electrochemically active sites.

Therefore, it is very important to know the geometry of the TPB width. With the

66

information we can design the electrode structure to achieve high and optimized fuel cell

performance at a certain operating condition.

The width is determined by the electrolyte’s material properties and fuel cell operating

conditions such as operating temperature. TPB width can be expressed by following

equations [42]:

k

DwTPB (4.2)

where D is a diffusion coefficient of electrolyte material and k is the reaction rate at the

electrode/electrolyte interface. For electrochemical systems, the reaction rate constant (k)

may be expressed in terms of the current density of the electrochemical reaction, jrxn

(A/cm2):

o

rxn

c

jk (4.3)

And the reaction current density (jrxn) can be estimated from the Tafel approximation for

a given overpotential, η [42-43]:

b

rxn jj /

010 (4.4)

67

where j0 is the exchange current density (A/cm2), η is overpotential (mV), and the b is the

Tafel slope (mV). By combining the equations above, we could get an expression of TPB

width as following:

bTPBj

Dcw

/

0

0

10 (4.5)

As previously mentioned and can be seen from the equation above, the TPB width is

dependent on the material properties and the operating conditions.

For the nano-pore structured Pt electrode case, the periphery of the pore is where TPB is

located. Therefore, increasing the density of pores by decreasing the pore size in a given

reaction area, we can increase the fuel cell performance. However, since the TPB has a

certain width we will have TPB overlaps if the pore radius is smaller than TPB width. In

other words, we can estimate the TPB width by observing the saturation regime of

performance enhancement while we keep decreasing the electrode pore size. So, if we

consider the electrode interface resistance, first it will decrease as we increase the TPB

density since we have more reaction site. But when the TPB overlap happens the

electrode resistance will not decrease anymore and show a saturation behavior (Figure 4-

14). By measuring the radius of pores at this saturation regime, we can estimate the TPB

width at a certain operating condition of SOFC.

68

Figure 4-14. Graphical estimation of electrode/electrolyte interface resistance as the TPB

density increases. Interface resistance will be decrease as we increase the TPB density by

decreasing the pore size since we are introducing more electrochemical reaction site. If

the TPB width overlap starts, the resistance will not decrease anymore and will show

saturation behavior.

For the experiments, four different sizes of silica particles (about 300nm ~ 600nm) were

prepared. Using those particles, SOFC samples were fabricated having nano-pore

structured Pt electrodes with different pore sizes. During the fabrication process, the

spacing between the pores was consistent for all the electrodes in order to be able to

compare the TPB density by the pore sizes. Figure 4-15 shows the SEM images of the

structured electrodes. To estimate the TPB width during the fuel cell operation, other than

the material properties we can change the operating conditions such as temperature and

the overpotential. From the equation (4.5), since the operating temperature both affects

69

diffusion coefficient, D, and the exchange current density, j0, we chose the overpotential

value as a variable at a fixed operating temperature condition for easier approach.

Figure 4-15. SEM images of nano-pore structured Pt electrodes on single crystalline YSZ

substrates. TPB linear density was estimated by measuring the pore size and spacing

between the pores. As we have smaller pore size, we have larger TPB density. The

electrode pore sizes are (a) 240nm, (b) 350nm, (c) 430nm, and (d) 570nm.

YSZ grain boundary is known to have higher surface oxygen exchange rate and bulk

grain. This will be discussed more in the next chapter. To exclude the contribution on

electrode interface resistance which comes from randomly distributed grain boundaries,

70

we used single crystalline YSZ which has no surface grain boundary as an electrolyte

material. From the relationship shown in equation (4.5), higher overpotential (η = OCV –

cell voltage) yields smaller TPB width. Therefore, higher cell voltage yields lower

overpotential and it has larger TPB width than smaller cell voltage condition. The EIS

spectra were measured at 400oC with different cell operating voltage conditions. From

the data, the electrode interface resistances of the four samples were extracted. Figure 4-

16 shows extracted electrode interface resistances at 400mV and 800mV cell voltage

conditions as a function of the electrode pore sizes. As can be observed, 800mV case,

which has larger TPB width than 400mV, shows saturation behavior in electrode

resistance values as the pore size gets smaller while the 400mV case shows continuous

decrease in electrode resistance. This indicates that the TPB width is about 200nm when

we run the Pt/YSZ system SOFC at the operating conditions of 400oC and 800mV cell

voltage. This is the first demonstration of the TPB width estimation during the actual fuel

cell operation. And this will provide a significant implication in electrode structure

designing for maximizing LT-SOFC performance by fully utilizing the electrochemical

reaction sites.

4.4 Conclusion

In summary, the present study offers a process methodology for controlling the geometry

of the Pt/YSZ interface for the systematic study of TPB characteristics. The results

71

presented here point to several important and useful consequences. First is the near-linear

scaling law of cell impedance and, correspondingly, cell performance on the linear

density of TPB. The EIS spectra showed significant improvement in the oxygen

reduction kinetics with increased TPB density. The linear scaling behavior infers that the

charge-transfer reaction is constrained to TPB as expected. It also implies that the width

of TPB in our MEAs is negligibly small with respect to its length, which agrees with

literature. Second is the precision and versatility of the patterning method presented here

to fabricate dense electrodes with sharp interfaces and well-defined geometries. These

electrodes also exhibit superior stability compared to sputtered Pt electrodes and maintain

their morphology and microstructure intact after prolonged fuel cell operation. Third is

the opportunity of stable operation of IT to LT SOFCs that employ these dense and

patterned electrodes. Fourth is the ability to study TPB geometric effects in a controllable

and reproducible manner over a wide range of circular pore diameters using this powerful

patterning technique that does not require lithography and micromachining. The pore size

and TPB density are easily tunable by changing the initial size of the masking silica

particles and the spacing between them. By doing this, we could effectively estimate the

TPB width at desired fuel cell operating conditions. This provides significant implication

in designing the nano-electrode structure to optimize and maximize LT-SOFC

performance.

72

4.5 References

[1] J. Fleig, Annu. Rev. Mater. Res., 33, 361 (2003)

[2] Y. Jiang and A. V. Virkar, J. Electrochem. Soc., 150(7), A942 (2003)

[3] S. DeSouza, S. J. Visco, and L. C. DeJonghe, Solid State Ionics, 98, 57 (1997)

[4] E. D. Wacshman, ECS Trans., 25(2), 783 (2009)

[5] A. Belzner, T. M. Gür, and R. A. Huggins, Solid State Ionics, 57, 327 (1992)

[6] K. Nisancioglu and T. M. Gür, Solid State Ionics, 72, 199 (1994)

[7] S. Sunde, K. Nisancioglu, and T. M. Gür, J. Electrochem. Soc., 143, 3497 (1996)

[8] S. Diethelm, A. Closset, K. Nisancioglu, J. Van Herle, A. J. McEvoy, and T. M. Gür,

J. Electrochem. Soc., 146, 2606 (1999)

[9] R. Doshi, J. L. Routbort, and C. B. Alcock, Defect and Diffusion Forum, vol.127-128,

Scitec Publications, p.39 (1995)

[10] S. B. Adler, Chem. Rev., 104, 4791 (2004)

[11] S. Wang, T. Kato, S. Nagata, T. Kaneko, N. Iwashita, T. Honda, and M. Dokiya,

Solid State Ionics, 152-153, 477 (2002)

[12] B. C. H. Steele, Solid State Ionics, 134, 3 (2000)

[13] R. Doshi, V. L. Richards, J. D. Carter, X. Wang, and M. Krumpelt, J. Electrochem.

Soc., 146, 1273 (1999)

[14] P-C. Su, C-C. Chao, J. H. Shim, R. Fasching, and F. B. Prinz, Nano Lett., 8(8), 2289

(2008)

[15] J. H. Shim, C.-C. Chao, H. Huang, and F. B. Prinz, Chem. Mater., 19, 3850 (2007)

73

[16] H. Huang, M. Nakamura, P.-C. Su, R.Fasching, Y. Saito, F. Prinz, J. Electrochem.

Soc., 154(1), B20 (2007)

[17] H. Huang, T. M. Gür, Y. Saito, and F. Prinz, Appl. Phys. Lett., 89(14), 143107-1-3

(2006)

[18] J. L. Hertz, H. L. Tuller, Solid State Ionics, 178, 915-923 (2007)

[19] A. Mitterdorfer and L. J. Gauckler, Solid State Ionics 117, 203 (1999)

[20] R. Radhakrishnan, A. V. Virkar, and S. C. Singhal, J. Electrochem. Soc. 152(1),

A210 (2005)

[21] V. Brichzin, J. Fleig, H.-U. Habermeier, and J. Maier, Electrochem. Solid State Lett.

3(9), 403 (2000)

[22] E. Siebert, A. Hammouche, M. Kleitz, Electrochim. Acta 40, 1741 (1995)

[23] B.C.H. Steele., Solid State Ionics, 86-88, 1223 (1996)

[24] J. van Herle , A.J. McEvoy, K.R. Thampi, Electrochim Acta, 41, 1447 (1996)

[25] A. Mitterdorfer, L.K. Gauckler, Solid State Ionics, 111, 185 (1998)

[26] Y. Matsuzaki, I. Yasuda, Solid State Ionics, 126, 307 (1999)

[27] M. Juhl, S. Primdahl, C. Manon, M. Mogensen, J. Power Sources, 61, 173 (1996)

[28] J. Fleig, Annu. Rev. Mater. Res. 33, 361-82 (2003)

[29] E. Mutoro, S Günther, B. Luerssen, I. Valov, J. Janek, Solid State Ionics, 179 , 1835-

1848 (2008)

[30] E. Mutoro, B. Luerssen, S. Günther, J. Janek, Solid State Ionics, 179, 1214 (2008)

[31] F.S. Baumann, J. Fleig, H-U. Habermeier, J. Maier, Solid State Ionics, 177, 1071-

1081 (2006)

74

[32] T. M. Gür and R. A. Huggins, J. Appl. Electrochem. 17, 800 (1987)

[33] X. Wang, H. Huang, T. Holme, X. Tian, F. B. Prinz, J. Power Sources, vol. 175, pp.

75-81, (2008)

[34] G. H. Bogush, M. A. Tracy and C. F. Zukoski IV, J. Non-Crystalline Solids, 104,

95-106 (1988)

[35] W. Stöber, A. Fink and E. Bohn, J. Colloid and Interface Science, 26, 62-69 (1968)

[36] G.H. Bogush and C.F. Zukoski IV, J. Colloid and Interface Science, Vol. 142, No.1

(1991)

[37] C.-M. Hsu, S. T. Connor, M. Tang,

Y. Cui, Appl. Phy. Lett., 93, 133109 (2008)

[38] A. Evans, A. Bieberle-Hütter, J. Rupp, L.J. Gauckler, J. Power Sources, 194, 119-

129 (2009)

[39] E. Barsoukov and J. R. Macdonald, Impedance Spectroscopy: Theory, Experiment,

and Applications, 2nd Ed., John Wiley and Sons (2005)

[40] H. Huang, T. Holme, and F. B. Prinz, ECS Trans., 3(32), 31 (2007)

[41] A. Optiz, A. Schintlmeister, H. Hutter, J, Fleig, ECS Trans., 25(2) 2783-2792 (2009)

[42] R. O'Hayre, D. M. Barnett, F. B. Prinz, J. Electrochem. Soc., 152(2), A439 (2005)

[43] A. J. Bard and L. R. Faulkner, Electrochemical Methods, 2nd ed., John Wiley &

Sons, New York (2001)

75

CHAPTER 5. Cathodic Surface Engineered Low Temperature

Solid Oxide Fuel Cells

In the previous chapter, we investigated the actual electrochemical reaction site for SOFC

which is the interface of the electrode/electrolyte by developing well-defined and

controllable nano-pore structured electrodes.

In this chapter, we report both experimental and theoretical results of the role of doped-

cerium oxides surface modification layer on the oxygen reduction reaction and its grain

boundaries in enhancing the oxygen incorporation kinetics of low temperature solid oxide

fuel cells (LT-SOFCs).

5. 1 Introduction

Fuel cells offer opportunities to achieve efficient energy conversion, and solid oxide fuel

cells (SOFCs) are distinguished by fuel flexibility, high quality waste heat, and simpler

water management systems. Among the range of oxide ion conducting ceramic

electrolyte materials, yttria stabilized zirconia (YSZ) has been the most commonly

utilized for its high chemical stability over a wide regime of oxygen activities ranging

76

from severely reducing to highly oxidizing conditions. However, due to the highly

activated (~1eV) nature of ionic transport in such solid oxide electrolytes and the

sluggish rate for the oxygen reduction reaction at the cathode side, SOFCs are usually

operated at relatively high temperatures (800oC~1000

oC). Such high operating

temperatures pose serious challenges for practical applications in seal integrity, structural

and thermal stability, high fabrication and materials costs, and compatibility of fuel cell

components. [1, 2] Typically, a SOFC membrane electrode assembly (MEA) is made of a

YSZ electrolyte membrane, a mixed conducting ceramic cathode such as

La1−xSrxCo1−yFeyO3 (LSCF) and La1−xSrxMnO3-δ (LSM), and a cermet anode such as

Ni/YSZ.

In recent years, there have been efforts to reduce the SOFC operating temperature to

intermediate temperatures of 500-700oC [3-5]. Most of these studies employed thin film

techniques to reduce the YSZ membrane thickness down to 10-100µm, thereby

minimizing the ohmic loss and lowering the fuel cell operating temperature. Some work,

including that done in our laboratory, has focused on depositing YSZ films as thin as 0.1-

1µm to reduce the SOFC operating regime to even lower temperatures (LT) between

300oC and 500

oC [6-10]. A recent report provides a good review of various ultra thin film

micro-SOFC efforts [11].

A key obstacle to reducing the temperature regime for SOFC operation is the poor

catalytic activity and transport properties of mixed conducting oxide based cathode

77

materials for the electrochemical reduction of oxygen at these low temperatures [2]. For

this reason, platinum (Pt) is still considered as the best catalytic cathode material at these

temperatures to enhance the oxygen reduction reaction rate or reduce the activation loss,

which is crucial for improving the performance of LT-SOFCs.

Acceptor doped cerium oxides such as gadolinia doped ceria (GDC) and yttria doped

ceria (YDC) are known to have higher ionic conductivity than YSZ below 700oC, but

exhibit mixed ionic electronic conduction at higher temperatures and under reducing

conditions. Because of this, ceria-based electrolytes are considered primarily for

intermediate temperature solid oxide fuel cell applications [12-14], but they have been

investigated at high temperatures as composite electrolytes in combination with stable

electrolytes such as YSZ [15-17]. In addition to their high ionic conductivity, doped

cerium oxides are good catalysts and exhibit fast cathodic kinetics for oxygen reduction

at the triple phase boundary (TPB) [18, 19]. The surface exchange coefficient is a good

measure to assess oxygen ion incorporation. Indeed, Steele and co-workers reported that

the surface exchange coefficient has a positive relationship with the oxygen diffusivity of

a given material [20]. It is reported that surface exchange on doped ceria at 700oC is

several times faster than that for YSZ at this temperature [21, 22]. Accordingly, the

rationale for choosing doped ceria interlayer in the present study to modify the cathode

interface and reduce the associated activation loss is based on its superior catalytic

activity and high surface exchange and oxide ion transport rates.

78

In this study, we performed quantum simulations to calculate and compare the activation

energies for oxygen incorporation into the Pt/YSZ and Pt/YDC systems. To the best of

our knowledge, this constitutes the first attempt reported in the literature to estimate the

energetics of oxygen incorporation in YDC. In addition, we experimentally investigated

the catalytic effect of surface modification on YSZ that involved deposition of ultra thin

YDC and GDC films as an interlayer between YSZ electrolyte and the Pt cathode to

promote the oxygen reduction reaction. Previous work in our laboratory had indicated

that inserting nanoscale thin GDC interlayer at the cathode side of the YSZ substrate

significantly enhanced fuel cell performance by reducing the activation loss associated

with oxygen reduction reaction (ORR) at the cathode [9, 23]. Similarly, others

theoretically observed low activation overpotentials with thin cathodes and ionic

conducting YSZ electrolyte with small grain sizes [17, 24-25]. Accordingly, we

hypothesized that surface grain structure of the YDC and GDC cathodic interlayers

contributes or enhances fuel cell performance.

In order to circumvent complications from grain boundaries, we designed the

experiments with two different approaches for each cathodic interlayer material. First, it

was decided to employ single crystalline YDC and GDC interlayer to investigate material

effect itself on oxygen kinetics at the cathode side without any contributions from the

grain boundaries. For this approach, we successfully fabricated epitaxial YDC, GDC

cathodic interlayers on single crystalline YSZ substrate and achieved composite

electrolytes by using PLD. Second, we investigated the contribution of grain size and

79

hence, the surface grain boundaries of doped cerium oxides in improving the cathode

kinetics of ORR by providing both spectrometric and spectroscopic evidences. For this

grain boundary contribution study, we investigated spectrometric and spectroscopic

evidence indicating enhanced activity at surface grain boundaries. We demonstrated

successful fabrication of polycrystalline composite electrolytes also by PLD technique.

By controlling post-annealing temperatures, we have systematically varied the size of

surface grains of doped ceria interlayers deposited on the cathode side. Then, we studied

how the performances are affected by the size of surface grains, i.e., the grain boundary

density on the doped ceria external surface at the cathode side. For both bulk grain and

grain boundary experiment cases, the fuel cell performance was characterized by current-

voltage measurements in the temperature range of 350oC~450

oC while cathode interfacial

resistances were extracted from electrochemical impedance data. Experimental results

indicate that the fuel cell performance enhances by having doped ceria interlayer by

increasing the surface reaction rate at the cathode side. Moreover, the samples with nano

size surface grains further improve the fuel cell performance by enhancing the oxygen

kinetics. The findings of this study provide important implications for engineering the

cathode as well as the surface grain structure of YSZ composite electrolytes with doped

ceria interlayer in order to enhance SOFC performance.

80

5. 2 Experimental

5.2.1 Quantum simulation of oxygen incorporation energies

Quantum mechanical simulations of oxygen incorporation in YSZ and YDC were

performed using a periodic boundary condition method as implemented in VASP [26-29].

Electron wave functions were expanded in a plane wave basis set using projector

augmented waves up to a maximum energy cutoff of 400eV with an exchange-correlation

functional as parameterized by Perdew et al. [30, 31] Gaussian smearing was used with a

width of 0.2eV to determine partial occupancies. Energy was sampled on a 2x2x1

Monkhorst Pack grid [32]. Calculations were performed assuming non-spin polarized

systems.

The supercell consists of a Pt38 cluster placed on top of a 3x3x3 YSZ slab of

stoichiometry (Y2O3)2(ZrO2)23, giving two vacancies in the anion sublattice. One

vacancy was chosen to reside on the top surface of the slab to study incorporation in the

vacancy, the other vacancy position, and the positions of the four yttrium atoms, were

chosen randomly (see Figure 5-1(b)). Before a slab relaxation in vacuum, the bulk was

constructed and relaxed to find a lattice constant of 5.12Å . This structure was then

relaxed in 20Å of vacuum in the z-direction. The Pt38 QD was separately relaxed, then

placed on top of the slab and relaxed again. To mimic the equilibrium coverage of Pt of

¼ ML in atmosphere, eight oxygen atoms were adsorbed on the surface of Pt, one near

81

the surface vacancy site. To simulate YDC, bulk CeO2 was constructed and relaxed, then

four Ce atoms were replaced by Y atoms in the same position that they were chosen in

YSZ, and the slab was relaxed in vacuum (see the final geometry in Figure 5-2(b)). The

electrochemical reaction was simulated by adding two electrons to the supercell.

The electronic self-consistent charge density was computed to a tolerance of 10-4

eV for

geometry relaxations, and all atoms are allowed to relax (i.e. no atoms are fixed) to find

adsorbed and incorporated states (i.e. initial and final positions). To calculate reaction

pathways, a single oxygen atom located above a surface vacancy was moved down into

the vacancy by constraining the z-coordinate of that atom in successive steps, while

allowing all other coordinates of all other atoms to relax. Charge densities were

calculated by integration inside a Wigner Seitz sphere of radius 0.82Å for oxygen.

5.2.2 Oxygen isotope exchange and NanoSIMS

For spectrometric observation we performed surface microstructure and oxygen exchange

measurements on commercial (Japan Fine Ceramics) sintered YDC pellets 1cmx1cm in

size and 500µm in thickness. The second category of experiments involved fuel cell and

electrode impedance studies as a function of YDC surface grain size (i.e., grain boundary

density) on composite electrolyte samples having a thin YDC layer deposited on the

cathode sides of YSZ substrates.

82

The surface microstructure of commercially available YDC pellets was analyzed by FEI

XL30 Sirion scanning electron microscopy (SEM). These pellets were also employed for

oxygen isotope exchange experiments, where incorporation was carried under DC biased

conditions applied across porous platinum cathode and anode layers sputtered on both

faces of the 500μm-thick polycrystalline YDC pellets. Isotope exchange experiments

involved evacuation of the vessel to at least 10-6

Torr followed by the introduction of

research grade (>99%) 18

O2 gas at 150±1Torr, which is equivalent to the ambient oxygen

partial pressure. The YDC samples were annealed at 400°C for 3 hours in 18

O2 under 1V

of externally applied cathodic (negative) bias. Prior to 18

O2 exchange, the samples were

annealed in 150Torr of 16

O2 environment for at least three times the isotope exchange

time.18

O and 16

O ion counts were measured simultaneously by using high spatial

resolution SIMS (NanoSIMS-50L, Cameca, France). A primary Cs+ ion beam (16keV)

was applied to analyze the secondary ions emitted from the samples. As the primary ion

beam sputters the surface (1010m2 of rastered area and 256×256 pixel with dwell time

of 1ms/pixel), the concentration of ions was measured layer by layer as a function of

depth.

83

5.2.3 Composite electrolyte fuel cell fabrication and characterization

For the fabrication of YDC, GDC interlayers on YSZ substrates, we employed pulsed

laser deposition (PLD) technique. A sintered 1-inch diameter and 0.125-inch thick YDC,

GDC disk pellets (10 mol%, Kurt J. Lesker) were used as target materials to deposit thin

films on two different types of YSZ substrates. A single crystalline YSZ substrate (1cm x

1cm x 300µm, Marketech Inc.) was used to fabricate epitaxial thin interlayers having no

grain boundaries, and polycrystalline YSZ substrates (1cm x 1cm x100µm, BEANS

International Corp.) were used to fabricate polycrystalline thin films with different grain

sizes. A Lambda Physik 248nm KrF excimer laser with energy density of 1.5J/cm2 per

pulse was used for ablating the target in a 100mTorr background oxygen gas environment

in the deposition chamber. During deposition, the sample stage was maintained at 750oC

and the sample-to-target distance was 50mm. The growth rates for YDC and GDC were

~0.22Å /pulse and ~0.23Å /pulse, respectively. For the polycrystalline interlayer samples,

the films were post-annealed at 750-1450oC for 10 hours in ambient air to achieve

different grain sizes. Membrane-electrode assembly (MEA) fabrication was completed by

dc sputtering of porous catalytic platinum (Pt) electrode on both sides for 150secs under

10Pa Ar background pressure and 50W plasma power.

For characterization of the deposited films, X-ray diffraction (XRD) was conducted for

structure and crystallinity characterization of the YDC and GDC interlayer films using a

PANalytical X’Pert PRO XRD system (Cu Kα X-ray with λ=1.54Å ) utilizing the

84

symmetrical θ/2θ scan method for phase analysis. The surface grain size of the post-

annealed films was measured by atomic force microscopy (AFM) (XE-70, Park Systems

Inc.) operated in non-contact surface scanning mode.

For fuel cell performance measurement and characterization, we employed a custom

made fuel feeding chamber sitting on a temperature-controlled heating stage with micro-

manipulating probe stations as shown in previous chapter. During fuel cell measurement,

pure dry hydrogen was supplied at the anode side as fuel while the cathode side was

exposed to ambient air as the oxygen source. For fuel cell performance evaluation, A

Gamry Potentionstat (FAS2, Gamry Instruments, Inc.) was used to obtain current-voltage

(I-V) behavior as well as the electrochemical impedance spectroscopy (EIS) data. Fuel

cell performance was measured from 350-450oC and EIS was measured under various

fuel cell voltage conditions in the frequency range of 300 kHz~0.1Hz. ZView software

(Scribner Associate, Inc) was used to analyze the EIS spectra based on complex nonlinear

least-squares fitting [33].

85

5. 3 Results and Discussion

5.3.1 Surface engineered SOFC with thin YDC cathodic interlayer

Quantum simulations for Pt/YSZ and Pt/YDC systems

Using density functional theory (DFT) simulations, oxygen adsorption on Pt38 clusters on

YDC, YSZ, and YDC|YSZ surfaces as well as oxygen incorporation into vacancies in the

surface layer of the oxide from the Pt38 cluster were investigated. Oxygen incorporation

in YSZ was found to be slightly energetically favorable by 0.09eV as compared to the

state of atomic oxygen adsorbed on Pt (Figure 5-1(a)). The incorporation reaction is

activated by an energy barrier of 0.38eV. As the oxygen atom moves away from Pt,

energy goes uphill while the O-Pt bonds are broken, and the electron charge on the

oxygen atom initially decreases. Past the activated state, the oxygen atom begins to form

bonds with surface Zr atoms, the energy goes down, and the charge on the oxygen

increases. As the oxygen atom begins to move past the stable surface site down into a

subsurface layer, the charge increases further as well as the energy since there is no

vacancy below the diffusing oxygen in this geometry.

86

Figure 5-1. (a) Change in energy as a function of height for atomic oxygen diffusing into

a vacancy in the first layer of YSZ. Height and energy is referenced to the stable

adsorbed state. Charge, plotted on the right axis, is the electron density integrated within

a Wigner Seitz sphere of radius 0.82Å around the radius (a more positive value

corresponds to higher electron density). (b) Snapshots of atom structure as oxygen is

incorporated. Oxygen atoms are shown in red, Y in yellow, Zr in purple, and Pt in silver.

87

Figure 5-2. (a) Change in energy as a function of height for atomic oxygen diffusing into

a vacancy in the first layer of YDC. Height and energy is referenced to the stable

adsorbed state. Charge, plotted on the right axis, is the electron density integrated within

a Wigner Seitz sphere of radius 0.82Å around the radius. (b) Snapshots of atom structure

as oxygen is incorporated. Oxygen atoms are shown in red, Y in yellow, Ce in blue, and

Pt in silver.

88

Compared to YSZ, oxygen incorporation in YDC shows significantly lower activation

energy, 0.07eV, and the incorporated state is of even lower energy, 0.49eV below the

adsorbed state (see Figure 5-2(a)). The difference in oxygen reaction energetics between

YSZ and YDC suggests that the nature of subsurface layers may affect reaction

energetics, demonstrating that interactions beyond the first nearest neighbor may play an

important role in fluorite ceramics. For example, it is known that oxygen and yttrium in

YSZ tend to occupy second nearest neighbor positions [34, 35] whereas they prefer first

nearest neighbors in doped ceria [36]. Further research is required to better understand

the role of different oxidation states and to clarify the behavior of YSZ and YDC

regarding oxide ion incorporation.

Grain boundary free YDC cathodic interlayer

By utilizing PLD, epitaxial films of single crystalline YDC layers of various thicknesses

were deposited on single crystalline YSZ substrates (8-mol%, 1cm x 1cm x 300µm-

thick). The main objective of employing epitaxial YDC layers was to eliminate

contributions from grain boundary effects. Figure 5-3 shows XRD spectra taken on

epitaxial YDC thin films of different thicknesses grown on single crystalline YSZ (100)

substrates. Intensity axis (y-axis) is plotted in a log scale to make the lower intensity

peaks more clearly visible. Only a strong YDC (100) peak is visible throughout the

spectra, which indicates perfect epitaxy of YDC observed for thicknesses up to 130nm.

89

Also, we observed that the peak intensity ratio of YDC to YSZ increased (roughly about

1.9-4.3) as the YDC thickness was increased.

Figure 5-3. XRD patterns of PLD YDC films deposited on single crystalline YSZ (100)

substrate. Spectra show only (100) peak up to the film thickness of 130nm, which

indicates perfect epitaxial growth of YDC films.

90

Figure 5-4. I-V performance of YDC interlayered SOFC and YSZ-only control sample

measured at 450oC. The plot shows gradual performance enhancement up to about 50nm

of YDC interlayer thickness, beyond which the fuel cell performance remains unchanged

with increasing YDC thickness.

SOFC MEAs employing epitaxial YDC interlayers grown at the cathode side on single

crystalline YSZ substrates were fabricated with different YDC film thicknesses of ~8nm,

17nm, 50nm, and 130nm. After deposition of porous Pt electrodes on both sides, fuel cell

performance was measured at temperatures from 350oC to 450

oC. Cell performance

measured at 450oC is shown in Figure 5-4, which compares the polarization behavior for

the YSZ-only control MEA with those of the YDC/YSZ MEAs having different

thicknesses of the YDC interlayer. Open circuit voltage (OCV) values of the samples

were in the range of 1.02~1.10V for all measurement temperatures. As seen in Figure 5-

4, fuel cell performance improved with YDC interlayer thickness and the peak power

91

density increased more than 2-fold. It shows enhancement after introducing even ~8nm

thin YDC layer and the performance continually improved up to a certain interlayer

thickness. Considering the deposition mechanism, the observed behavior suggests that

PLD may not form a fully and uniformly covering YDC layer below 10nm range, but

possibly does so above 20nm. Interestingly, after about 50nm the cell performance does

not enhance any further and shows the same behavior with the thicker (130nm) interlayer

MEA. We speculate that the overall ohmic loss of the cell is dominated primarily by the

ionic resistance of the 300µm-thick YSZ bottom substrate, which is about three orders of

magnitude thicker than the YDC interlayer. This and the higher ionic conductivity of

YDC compared to YSZ at this measurement temperature result in a negligibly small

contribution by the YDC layer to the overall resistive loss in the cell.

Figure 5-5. EIS data of YDC interlayered fuel cell measured at different cell voltage

conditions at 400oC. Two loops are observed. The high frequency loop seems to be

independent of cell voltage, indicating that this arc corresponds to ionic transport through

the electrolyte (Rohmic). In contrast, the low frequency loop is dependent on cell voltage

indicating that this arc corresponds to the electrode interface resistance (Relectrode).

92

For further analysis of this surface engineering effect, EIS measurements were performed

for each sample at different measurement temperatures. Figure 5-5 shows a sample EIS

data for a 50nm thick YDC interlayer sample measured at 400oC. Two loops were

observed and a commonly used equivalent circuit model shown in the figure was adopted

to fit the spectra [14]. As shown in Figure 5-5, EIS measurements were made at several

DC bias voltages. While the electrode processes are highly affected by the magnitude of

cell voltage, the ionic transport resistance in the electrolyte is generally independent of

the cell voltage conditions. The high frequency loop shows no discernable changes for

three different cell voltage conditions, namely, OCV, 600mV, and 200mV, suggesting

that this arc corresponds to ionic transport through the electrolyte. However, voltage

dependence of the low frequency semi-circle suggests that this arc is associated with

electrode processes at the electrolyte/electrode interface [28-30]. Also, the capacitance

estimated for the low frequency arc is about 10-6

F/cm2, which is a typical value for

electrode processes, while the magnitude of the capacitance value for the high frequency

arc is in the order of 10-9

/cm2 [34].

For SOFCs, there is a general agreement that cathode reaction kinetics is considerably

more sluggish than the anode reaction such that the cathode reaction dominates overall

activation losses for the cell [4, 37]. Moreover, for our MEA fabrication, the reaction area

of anode side is about 5-fold larger than the active area of the cathode. Due to reasons

above, it is likely that the smaller anode loop is merged into the much larger cathode loop

93

resulting in only one semi-circle for the electrode process rather than two. By fitting the

obtained EIS data for all samples, individual contributions of electrolyte and electrode

interface impedances were determined.

As described above, we speculate that the incremental increase in electrolyte resistance

due to the additional YDC interlayer would be small and would not have an adverse

effect on fuel cell performance. Figure 5-6 shows the ohmic resistances of the pure YSZ

SOFC control sample and different thicknesses of YDC interlayered SOFCs measured at

temperatures from 350oC to 450

oC. As expected, there is no change in the ohmic

resistance due to the electrolyte and practically no dependence on the interlayer thickness

in this experimental regime. Clearly, the ohmic resistance for the surface modified SOFC

is primarily dominated by the thick YSZ substrate and the contribution from the YDC

interlayer thickness is negligible for all practical purposes.

94

Figure 5-6. Extracted ohmic resistances of YDC interlayered SOFCs for different YDC

thicknesses at temperatures of 350oC~450

oC. Zero in the x-axis indicates the bare YSZ

sample with no interlayer. The plot shows no discernable change in cell ohmic resistance

with increasing interlayer thickness up to 130nm.

Figure 5-7. Electrode interface resistance values for the YDC interlayered SOFCs with

different thicknesses extracted from impedance measurements. The resistance starts to

drop immediately after the introduction of a thin layer YDC (<10nm). After forming a

full covered YDC layer, the electrode resistance reaches a plateau, and does not change

with further increase in YDC thickness.

95

Figure 5-7 compares the measured electrode interface resistance for the YSZ-only control

MEA with YDC/YSZ MEAs for different interlayer thicknesses in the temperature

regime of 350oC~450

oC. During PLD, deposition is initiated with surface nucleation

followed by island formation on the substrate [38]. Hence, it is likely that full and

uniform surface coverage in films less than 10~20nm thick is difficult to achieve. To

examine the microstructure and verify the uniformity of surface coverage by the

interlayer, AFM topography scanning and surface sensitive XPS were conducted on YDC

interlayered samples. For AFM topography scanning, we picked two thickness points

from the data in Figure 5-7, one from the sloped region (~14nm) and one from the

saturated region (~80nm). Figure 5-8 shows the topography image of the three samples,

bare YSZ, ~14nm YDC, and ~80nm YDC, respectively. The images show clear

differences in topography and grain formation. However, due to the limitation of the

AFM tip aspect ratio in relation to the small grain sizes, it was difficult to confirm full

coverage of YDC interlayers on the YSZ substrate by AFM. For that reason, we also

employed angle resolved XPS measurements to determine the extent of surface coverage

of YDC interlayers. By tilting the sample stage of our XPS setup (15~30o) we were able

to limit the beam penetration depth to around 3nm or less. That arrangement excluded

signal detection from the underlying YSZ substrate and allowed signals only from the top

surface. Figure 5-9 shows the measured XPS data of interlayer samples with three

different thicknesses. The XPS spectra clearly show one of the main Zr peaks at ~186eV

96

for the sample containing ~8nm thin YDC layer indicating incomplete coverage. For the

17nm thick sample, the Zr peak intensity decreased as the YDC interlayer thickness

increases suggesting increasing coverage. Full surface coverage of YSZ by the YDC

interlayer is achieved possibly in the 30-50nm thickness range under the experimental

conditions employed in our study.

Figure 5-8. AFM scanned surface topography images of YDC interlayers with different

thicknesses. (Left) Bare single crystalline YSZ, (Center) ~14nm YDC, (Right) ~80nm

YDC on top of YSZ. It shows grain formation as the YDC thickness increases.

97

Figure 5-9. Surface sensitive XPS analysis of surface modified samples with three

different YDC thickness within the binding energy regime of one of the main Zr peaks.

As the YDC interlayer thickness increases Zr peak decreases and at the thickness about

26nm, almost no Zr peak is observed.

These results verify that for very thin YDC films possibly less than 30nm, the YSZ

substrate surface is not fully covered with the YDC interlayer but instead has patchy

coverage with open spaces between, where YSZ is exposed to the gas phase. With the

~26nm thick sample the Zr peak was practically indistinguishable from the background

suggesting that at this thickness the YDC film forms near full coverage. This helps

explain the reason why the cathode impedance initially displays a monotonic decrease

with increasing YDC thickness up to ~26nm. If on the other hand, the coverage at the

thinnest YDC layer were to result in full and uniform instead, then one would expect to

98

observe an abrupt drop in the electrode impedance in Figure 5-7 from the case featuring

bare YSZ cell to the case where the first YDC interlayer was introduced.

To observe the full effect of YDC interlayer on oxygen incorporation, therefore, complete

surface coverage of the YDC film on YSZ substrate is required. In Figure 5-7, a

noticeable drop in the interface resistance is clearly evident right after the introduction of

the first YDC interlayer. This trend continues as the YDC thickness is further increased

until it reaches a plateau value after about 50nm. This implies that initially the coverage

of the YDC layer is far from being complete, where the Pt electrode contacts directly the

YSZ surface in some regions. The cathode interfacial resistance decreases as the surface

coverage of the YDC film on YSZ increases until the entire active Pt cathode interface is

in full and direct contact with the YDC interlayer. As the XPS data shows, we expect to

have fully covering YDC layer after about 30nm and it would show same performance

behavior as the 50nm interlayer sample.

After forming a complete YDC layer, the interface resistance saturates for each

measurement temperature. Compared to the YSZ-only control sample, about a 2-fold

decrease in electrode interface resistance was achieved by employing single crystalline

YDC surface modified cathode interlayer. The results shown above clearly confirm that

the YDC interlayer itself enhances LT-SOFC performance by improving surface oxygen

kinetics.

99

Grain boundary and grain size contribution of YDC cathodic interlayer

To investigate YDC surface grain boundary activity, we first performed spectrometric

direct observation using isotope exchange method and secondary ion mass spectrometry

(SIMS) as a characterization tool. Figure 5-10(a) shows the SEM of the microstructure

and topography of the polycrystalline YDC pellet surface, indicating an average grain

size of 6+1µm. Sintered polycrystalline YDC pellets were also employed in oxygen

exchange experiments conducted under a cathodic DC bias of 1V while annealing in 18

O2

environment at 400°C. Surface activity for ORR was determined by the use of a high

spatial resolution SIMS (NanoSIMS) with a beam size less than 100nm. The NanoSIMS

image presented in Figure 5-10(b) as the 18

O/16

O ratio clearly indicates enhanced activity

along the grain boundary regions on the YDC surface. The 18

O/16

O count ratio in grain

boundary regions are about two times higher than that in bulk regions at this operation

condition. This result agrees well with previous work and observations in our laboratory

on YSZ surfaces, and confirming that this phenomenon is not unique to YSZ [39]. It also

provides spectrometric evidence that grain boundaries on the external surfaces of ionic

conducting oxides provide preferential sites for oxygen incorporation, possibly due to

higher concentration of vacancies in the grain boundary region [40].

100

Figure 5-10. (a) Surface SEM image of YDC sintered pellet, where dashed line shows

clear grain boundaries. (b) 18

O/16

O concentration map of corresponding YDC surface

obtained from NanoSIMS. 18

O/16

O count ratio was observed higher at grain boundary

regions (dashed) than bulk regions indicating oxygen isotopes were more populated in

grain boundary regions.

To study how grain size affects electrochemical behavior, YDC films were fabricated

with size-engineered surface grains at the cathode sides of the fuel cell elements. Size

engineering of grains was achieved by controlling the post annealing temperature for

these films. YDC surface microstructure of YDC/YSZ samples was determined by AFM.

Figure 5-11 shows the AFM topography images of YDC surfaces after post-annealing at

temperatures from 750oC to 1500

oC. As expected, lower annealing temperatures yield

smaller grain size in the sub-micrometer regime, namely, 55±15nm for 750oC (Figure 5-

11(a)) and 120±30nm for the 1100oC (Figure 5-11(b)) samples. Higher annealing

temperatures result in grain sizes 2.02±1.04µm for the 1300oC sample (Figure 5-11(c))

and 6.50±1.72µm for the 1500oC sample (Figure 5-11(d)) respectively. Figure 5-12

101

shows the resulting average grain sizes as a function of annealing temperature. This post

annealing process affects grain growth only in the thin YDC interlayer, but has no

significant effect on the grain size and microstructure of the underlying YSZ substrate. In

other words, the grain size of only the cathode interface has been varied in these

experiments while the grain size at bulk and the anode interface remained practically

unchanged. This was confirmed by our previous experiments using PLD YSZ thin

interlayers in our laboratory [39]. Therefore any significant variations in the

electrochemical behavior of these samples can be associated with the role of grain size at

the cathode interface.

Figure 5-11. AFM images of YDC surface additionally deposited on polycrystalline YSZ

substrate and post-annealed at different temperatures. (a) 750oC, (b) 1100

oC, (c) 1300

oC,

and (d) 1500oC.

102

Figure 5-12. Average grain size of YDC interlayer as a function of post-annealing

temperature.

The first set of experimental evidence supporting the role of grain size comes from fuel

cell measurements. After deposition of porous Pt electrodes on both sides of these

samples under identical sputtering conditions, their fuel cell performances were measured

at temperatures from 350oC to 450

oC using hydrogen as the fuel and air as the oxidant.

Open circuit voltage (OCV) values were in the range of 1.03V - 1.06V vs air for all

measured samples. Figure 5-13 shows the current-voltage (I-V) behavior of YDC/YSZ

composite SOFC samples measured at 400oC with varied grain sizes (or, grain boundary

densities) at the cathode side. The data show consistent improvement in the fuel cell

performance with decreasing grain size at the YDC surface. Indeed, the composite

sample having the smallest grain size at the YDC surface which corresponds to the

highest grain boundary density shows the highest peak power density and the lowest

103

activation loss. In other words, having more grain boundaries at the cathode interface

significantly enhanced the fuel cell performance in terms of peak power density by up to

4-fold and this enhancement is primarily due to improved cathode activation for the

oxygen reduction reaction.

In order to exclude the possibility of surface roughness effects contributing to increased

fuel cell performance, the average roughness of YDC surface was measured by AFM for

all samples. We measured multiple spots for each sample and found the root mean square

(RMS) roughness value was about 4-20nm. Based on our previous experimental results,

the contribution from such roughness values to the effective surface area is rather

marginal, with an estimated enhancement of only 1-3%. Therefore, one can exclude

surface area effects and conclude that the improvement in fuel cell performance with

decreasing grain size is primarily due to enhanced oxygen reduction kinetics, which is

consistent with SIMS results of Figure 5-10, and not from an increase in the effective

reaction site density due to surface roughness.

104

Figure 5-13. Current-Voltage (I-V) behavior of fuel cell MEAs measured at 400oC. Fuel

cells with smaller surface grain size show higher performance in terms of peak power

densities.

The second set of experimental evidence supporting the role of grain size (i.e., surface

grain boundary density) comes from electrochemical impedance spectroscopy (EIS)

measurements. Accordingly, the effect of the YDC grain boundary density at the cathode

side was studied by EIS for each sample at different temperatures. Figure 5-14 shows a

representative Nyquist plot for the 1500oC post-annealed composite sample measured at

350oC under various cell voltage conditions. The spectra indicates three arcs, where the

two high frequency arcs (arc I and arc II) showed no discernible change under three

different cell voltage conditions indicating that these two arcs are most likely associated

with ionic transport across bulk grains and grain boundaries, respectively [14]. The total

electrolyte resistance value, which is the sum of the two high frequency arcs, matches

well with the reference conductivity value for YSZ at the measured temperature [41-42].

105

Possible increase in electrolyte resistance due to the additional thin YDC layer can be

neglected since its contribution is merely 0.1% of the ohmic resistance of YSZ under this

experimental condition. The ionic conductivity of YDC is more than an order of

magnitude higher than that of YSZ at this temperature, and the thickness of added YDC

interlayer is almost three orders of magnitude smaller than the thickness of the YSZ

substrate. Thus, it can be assumed that ohmic resistance is primarily due to the YSZ bulk

substrate. This is also confirmed by recent experiments in our laboratory [43]. In contrast,

the low frequency arc is highly affected by the cell voltage conditions suggesting that this

arc corresponds to electrode processes, most likely associated with the cathode reaction

as discussed previously.

Figure 5-14. Electrochemical impedance spectroscopy (EIS) data of 1500oC annealed

YDC/YSZ composite fuel cell sample measured at 350oC indicating three loops. The two

high frequency loops seem to be independent of cell voltage conditions, indicating that

these arcs correspond to ionic transport through electrolyte and representing bulk (arc I)

and grain boundary (arc II).. whereas the low frequency loop shows dependence on cell

voltage conditions indicating that this arc corresponds to the electrode interface

resistance.

106

Figure 5-15. A plot showing extracted electrode interface resistances (at 450oC, 0.6V) as

a function of estimated surface grain boundary densities. As expected, the electrode

resistance decreases as the surface grain boundary density increases (lower grain sizes).

By using a representative equivalent circuit model, the values for electrolyte and

electrode (cathode) resistances are extracted for all measured samples. The representative

electrode interface resistance values (450oC at 0.6V cell voltage condition) were plotted

in Figure 5-15 as a function of surface grain boundary density estimated from values in

Figure 5-12. As expected, the electrode resistance decreases with increasing surface grain

boundary density. This clearly indicates that the YDC surface grain boundary enhances

the oxygen exchange rate at the cathode surface. In addition, exchange current densities,

which are highly related to the charge transfer reaction rate at the cathode, were

calculated from the measured EIS and I-V data. From the measured I-V fuel cell

107

performances, the exchange current density (j0) was extrapolated by fitting the activation

or polarization loss (ηact) and current density (j) values with the Tafel approximation [43]:

0

lnj

j

nF

RTact

(6.1)

where R is the ideal gas constant, α is the charge transfer coefficient, n is the number of

electrons involved in the electrode interface reaction, and F is the Faradic constant.

Figure 5-16 shows the results for all the YDC/YSZ composite samples measured in the

temperature regime of 350oC-450

oC and indicates activation energies of 0.62-0.66eV.

The composite 1500oC annealed sample having the largest grain size (5~7µm), i.e., the

lowest density of surface grain boundaries, has shown the lowest exchange current

density values. It is consistently observed that cathodic interface resistances of the

samples decrease as the grain size of the YDC interlayer decreases by lowering the post-

annealing temperature. The resistance values for the nano-grain size sample (40~70nm)

which was annealed at the deposition temperature of 750oC showed the lowest resistance

values, about 6~7-fold less compared to that of the largest grain sample, at each

measurement temperature. Results in Figure 5-16 and the EIS analysis indicate that the

exchange current density as well as the cathodic interface resistance scales with grain

size, more precisely, the grain boundary density on the YDC surface. Smaller surface

grains naturally generate higher grain boundary density at the cathode interface. The fuel

cell MEA with higher YDC surface grain boundary density (i.e. smaller grain size) shows

108

higher exchange current densities. This study clearly demonstrates that such nano-

granular surface microstructure gives rise to higher charge transfer rate at the cathode

side. Based collectively on the results presented in this study, we postulate that YDC

surface grain boundaries serve as active sites for enhanced oxygen exchange kinetics.

Figure 5-16. Exchange current densities for all measured samples with different grain

sizes were calculated at temperatures 350oC-450

oC. As the surface grain size decreases

(i.e., higher grain boundary density), the electrode interface resistance decreases. This

indicates that the surface grain boundaries enhance oxygen surface kinetics at the cathode

side.

109

5.3.2 Surface engineered SOFC with thin GDC cathodic interlayer

Upon the YDC cathodic interlayer experimental results shown above, we investigated

another dope ceria, which is gadolinia doped ceria, as a cathodic interlayer for LT-SOFC.

We adopted similar experimental scheme to study the role of GDC on oxygen kinetics.

As shown in previous section, we first studied the effect of GDC material itself as an

interlayer using grain boundary free GDC thin film. Then, we investigated the

contribution of GDC grain boundaries on oxygen kinetics.

110

Grain boundary free GDC cathodic interlayer

Figure 5-17. X-ray diffraction patterns of (a) epitaxial and (b) fully developed

polycrystalline GDC films on single (100) and polycrystalline YSZ substrates,

respectively.

Figure 5-17(a) shows the x-ray diffraction pattern of the first experimental set using a

single crystalline YSZ (100) substrate. About 60nm of a GDC layer is deposited on the

cathode side. In the spectra in Figure 5-17(a), only the strong (100) peaks of GDC and

YSZ are visible, indicating the epitaxially grown GDC layer on the YSZ (100) substrate.

The main objective of this first experimental set was to exclude grain boundary

contributions and to observe only the cathodic interlayer effect on surface oxygen

kinetics.

111

Figure 5-18. Current-voltage (I-V) behavior of epitaxial GDC interlayered MEA,

measured at 450oC. The SOFC MEA with GDC interlayer shows about 2-fold higher

peak power density.

After depositing porous Pt catalytic electrodes on both sides, SOFC MEA fabrication

employing epitaxial GDC cathodic interlayer was completed and the fuel cell

performance was measured at temperatures from 350oC to 450

oC. Figure 5-18 shows

representative current-voltage behavior data measured at 450oC, comparing an MEA

made of pure YSZ electrolyte with that employing epitaxial GDC interlayer electrolyte.

Open circuit voltages (OCVs) were 1.04V for the YSZ-only control cell and 1.05V for

the interlayered cell, respectively. The measured peak power density of the interlayered

cell is about 1.9 times higher than that of the control YSZ sample. From previous

experiments conducted in our laboratory we know that the increase in ohmic resistance

due to the thickness of the additional layer on the cathode side is almost negligible [44],

112

since the ionic conductivity of GDC is more than an order of magnitude higher in that

measurement temperature regime. Also, the interlayer thickness is more than three orders

of magnitude higher than the bottom substrate thickness (300µm). Therefore, the ohmic

resistance primarily due to the ionic resistance of the thick YSZ bottom substrate. Figure

5-18 suggests that losses for the YSZ control cell are larger than that of the GDC

interlayered cell. To find out why this is the case and to further investigate the effect of

the GDC interlayer on cathodic interface resistance, EIS measurements were performed

at same temperature range under different cell voltage conditions. Similar to the

experiment using single crystalline YDC interlayer in previous section, electrode

interface resistances were extracted by using an equivalent circuit model. It showed about

1.75times lower interface resistances for the epitaxial GDC layer cell than for the YSZ

control cell. This result suggests that GDC exhibits faster surface oxygen kinetics than

YSZ and it is consistent with YDC.

Figure 5-19. Atomic force microscopy (AFM) topography images of GDC surfaces,

annealed at (a) 750oC, (b) 1200

oC, and (c) 1450

oC. As the post-annealing temperature

increases, the grain size also increases.

113

To investigate the role of GDC surface grain boundaries on oxygen kinetics, interlayered

SOFC MEAs were fabricated with different grain sizes (i.e., grain boundary densities) at

the cathode side by a post-annealing process. Fully developed polycrystalline GDC thin

layers were observed on polycrystalline YSZ (Figure 5-17(b)). The microstructure of

cathodic GDC grains was determined by AFM. Figure 5-19 shows the AFM topography

images of the GDC interlayer surfaces annealed at temperatures from 750oC to 1450

oC.

The grain size increases as the post-annealing temperature increases. Estimated grain

sizes are about 61±11nm, 198 ± 22nm, and 5.74±1.56um for 750oC, 1200

oC, and 1450

oC

annealed samples, respectively (Figure 5-19(a), (b), and (c)). In previous section of this

chapter, we confirmed with YDC/YSZ composite electrolyte that this post-annealing

process changes only the grain sizes of the YDC cathodic interlayer with no significant

effect on the grain structure of the underlying thick YSZ substrate, including its bulk and

its anode side surface. Therefore, also in this case, the difference in the electrochemical

behavior of samples can be attributed to the effect of grain microstructure of the cathode

interface.

114

Figure 5-20. I-V performance at 450oC of GDC/YSZ composite electrolyte MEAs with

different GDC surface grain sizes,. The smaller surface grain size sample (lower

annealing temperature), which corresponds to the higher surface grain boundary density,

shows higher peak power density.

Figure 5-20 shows a representative I-V performance of GDC/YSZ composite electrolyte

fuel cells with different cathodic grain sizes operated at 450oC. The measured OCVs were

1.01V-1.06V for all fuel cell samples. As expected, the MEA with smaller GDC grain

size shows higher performance in terms of peak power density and reduced activation

loss. This supports previous experiment and simulation results conducted in our

laboratory that point to higher concentration of oxygen vacancies at grain boundary

regions, which provide more reaction sites for ORR [9, 36]. Hence, we postulate that

nano-granular GDC with increased grain boundary density improves cell performance by

enhanced surface oxygen exchange at the cathodic surface. This is experimentally

verified by systematic variation of grain structure of GDC/YSZ composite electrolyte.

115

Figure 5-21. Arrhenius plot of cathodic interfacial resistances of MEAs with different

GDC surface grain sizes. As the post-annealing temperature increases (i.e., as the surface

grain boundary density decreases), the electrode interfacial resistance increases. MEAs

with nano-granular GDC surface grains show lower electrode interfacial resistances than

those with larger surface grain size.

Similarly, EIS measurements at temperatures of 350oC-450

oC were conducted on MEAs

featuring different GDC surface grain sizes. Using an equivalent circuit model which was

mentioned for YDC case in previous section, the electrode and GDC interface resistances

were extracted. Figure 5-21 presents the Arrhenius behavior of electrode interfacial

resistance for cells with different GDC grain boundary densities. The GDC/YSZ

composite MEA annealed at 1450oC, which has the largest grain size (~6µm), in other

words, the lowest grain boundary density on the surface shows the largest cathodic

interface resistance. As the post-annealing temperature decreases, interfacial resistances

consistently decrease due to increased grain boundary density. The composite MEA

116

annealed at 750oC with the GDC surface grain size of about 65nm (i.e., large grain

boundary density) shows the lowest cathodic interfacial resistance, about 5-6 times lower

than the MEA with the largest grain size sample that was annealed at 1450oC. Moreover,

EIS analysis clearly indicates that cathodic interface resistance scales well with the size

of surface grains, i.e., the surface grain boundary density on the GDC surface. These

results also support both simulation and experimental studies previously conducted in our

laboratory that were also based on the hypothesis that surface grain boundaries enhance

oxygen reduction kinetics.

5.4 Conclusion

In this chapter, we investigated engineering effect of doped ceria (YDC and GDC)

interlayers at the cathode side of YSZ electrolyte and the role of doped ceria surface grain

boundaries on LT-SOFC performance. Quantum mechanical simulations demonstrated

reduced activation energy and a larger energetic driving force for oxygen incorporation in

YDC as compared to YSZ. Surface mapping of 18

O and 16

O ions by a high spatial

resolution NanoSIMS indicated preferential enrichment of 18

O along grain boundaries on

the YDC external surface. For fuel cell performance characterization, epitaxial doped

ceria thin interlayers (both YDC and GDC) on single crystalline YSZ (100) substrate

helped decrease cathodic interfacial resistance due to their faster surface exchange rate

than that of YSZ. Composite electrolyte fuel cell MEAs with different doped ceria’s grain

117

sizes (from about 40nm to 6µm) indicated that MEAs with smaller surface grain size (i.e.,

higher surface grain boundary density) showed superior fuel cell performance in terms of

higher peak power density and lower cathodic interface resistance. Moreover, the

exchange current densities were calculated and showed similar trend that higher surface

grain boundary density results the higher charge transfer rate. These results suggest that

grain boundary regions are electrochemically more active and nano-granular surface

grains enhance surface oxygen exchange rate.

This study successfully demonstrates the substantial effect of thin YDC, GDC cathode

interlayer and size of surface grains on cell behavior. Results of this study provide

significant implications in designing nano-granular YDC surfaces to achieve enhanced

LT-SOFC performance by improving oxygen surface kinetics.

5.5 References

[1] B. C. H. Steel, A. Heinzel, Nature, 414, 345–352 (2001)

[2] N. P. Brandon, S. Skinner, B. C. H. Steele, Annu. Rev. Mater. Res., 33, 183–213

(2003)

[3] X. Chen, N. J. Wu, L. Smith, A. Ignatiev, Appl. Phys. Lett., 84, 2700–2 (2004)

[4] S. de Souza, S. J. Visco, L. C.De Jonghe, Solid State Ionics, 98, 57–61 (1997)

118

[5] A.V. Virkar, Low-temperature Anode-supported High Power Density Solid Oxide

Fuel Cells with Nanostructured Electrodes, University of Utah (2003)

[6] U. P. Muecke, D. Beckel, A. Bernard, A. Bieberle-Hu tter, S. Graf, A. Infortuna, P.

Muller, J. L. M. Rupp, J. Schneider, L. J. Gauckler , Adv. Funct. Mater., 18, 3158–

3168 (2008)

[7] P.-C. Su, C.-C. Chao, J. H. Shim, R. Fasching, and F. B. Prinz, Nano Lett., 8, 2289

(2008)

[8] J. H. Shim, C.-C. Chao, H. Huang, and F. B. Prinz, Chem. Mater.,19, 3850 (2007)

[9] H. Huang, M. Nakamura, P. Su, R. Fasching, Y. Saito, and F. B. Prinz, J.

Electrochem.Soc., 154, B20 (2007)

[10] H. Huang, T. M. Gür, Y. Saito, and F. Prinz, Appl. Phys. Lett., 89, 143107 (2006)

[11] A. Evans, A. Bieberle-Hutter, J. L.M. Rupp, L. J. Gauckler, Journal of Power

Sources, 194, 119–129 (2009)

[12] B.C.H. Steele, Solid State Ionics, 129, 95 (2000)

[13] J. A. Kilner, Solid State Ionics, 129, 13-23 (2000)

[14] H.L. Tuller, Solid State Ionics, 131, 143-157 (2000)

[15] A.V. Virkar, J. Electrochem. Soc., 138, 1481–1487 (1991)

[16] K. Eguchi, T. Setoguchi, T. Inoue, H. Arai, Solid State Ionics, 52, 165–172 (1992)

[17] T. Tsai, S. A. Barnett, Solid State Ionics, 98, 191-196 (1997)

[18] T. Hibino, A. Hashimoto, T. Inoue, J. Tokuno, S. Yoshina, M. Sano, Science, 288,

2031 (2000)

[19] H. Uchida, M. Yoshida, and M. Watanabe, J. Phys. Chem., 99, 3282 (1995)

119

[20] B. C. H. Steele, Solid State Ionics, 75, 175 (1995)

[21] B. C. H. Steele, K. M. Hori, S. Uchino, Solid State Ionics, 135, 445 (2000)

[22] J. A. Lane, J. A Kilner, Solid State Ionics, 136-137, 927 (2000)

[23] H. Huang, T. Holme, F. B. Prinz, J. Fuel Cell Sci. Tech., 7, 1-5 (2010)

[24] C. W. Tanner, K. Z. Fung, A. V. Virkar, J. Electrochem. Soc., 144, 21–30 (1997)

[25] S. H. Chan, X. J. Chen, K. A. Khor, J. Electrochem. Soc., 151, A164–A172 (2004)

[26] G. Kresse and J. Hafner, Phys. Rev. B, 47, 558 (1993)

[27] G. Kresse and J. Hafner, Phys. Rev. B, 49, 14251 (1994)

[28] G. Kresse and J. Furthmüller, Comput. Mat. Sci., 6, 15 (1996)

[29] G. Kresse and J. Furthmüller, Phys. Rev. B, 54, 11169 (1996)

[30] G. Kresse, J. Joubert, Phys. Rev. B, 59, 1758 (1999)

[31] J.P. Perdew, J. A. Chevary, S. H. Vosko, K. A. Jackson, M. R. Pederson, D. J.

Singh, C. Fiolhais, Phys. Rev. B, 46, 6671 (1992)

[32] H. J. Monkhorst, J. Pack, Phys. Rev. B, 13, 5188 (1976)

[33] E. Barsoukov, J. R. Macdonald, Impedance Spectroscopy: Theory, Experiment, and

Applications, 2nd ed., Wiley, New York (2005)

[34] P. Li, I.W. Chen, J.E. Penner-Hahn, Phys. Rev. B, 48, 10074 (1993)

[35] G. Stapper, M. Bernasconi, N. Nicoloso, M. Parrinello, Phys. Rev. B, 59, 797 (1999)

[36] H. B. Lee, F. B. Prinz, W. Cai, Acta Materialia, 58, 2197–2206 (2010)

[37] T. Holme, R. Pornprasertsuk, F. Prinz, J. Electrochem. Soc., 157, B64-70 (2009)

[38] J. D. Ferguson, G. Arikan, D. S. Dale, A. R. Woll, J. D. Brock, Physical Review

Letters, 103(25): 256103 (2009)

120

[39] J. Shim, J. Park, T. Holme, K. Crabb, W. Lee, Y. B. Kim, X. Tian, T. M. Gür, F. B.

Prinz, Acta Materialia (2011) submitted.

[40] W. Lee, M. Lee, H.-J. Jung, F. B. Prinz, Meet. Abstr. Electrochemical Society, 1001,

724 (2010)

[41] S. M. Haile, Acta Materialia, 51, 5981-6000 (2003)

[42] B. C. H. Steele, Mat. Sci. and Eng., 79, B13 (2002)

[43] R. O’Hayre, S. W. Cha, W. Colella, F. B. Prinz, Fuel Cell Fundamentals, Wiley,

New York (2005)

[44] Y. B. Kim, T. Holme, T. M. Gür, F. B. Prinz, Adv. Func. Mater.,

10.1002/adfm.201101058 (2011)

121

CHAPTER 6. Three-Dimensional Proton Conducting Fuel Cell

Architecture with Ultra Thin Ceramic Electrolyte

As a reaction surface engineering for SOFC, we investigated a novel method for creating

three-dimensional (3-D) fuel cell architecture to enhance fuel cell performance by

increasing the area of electrolyte membrane. This chapter presents the fabrication and

operation of a low temperature 3-D protonically conducting ceramic fuel cell featuring a

close packed and free standing crater patterned architecture achieved by nanospherical

patterning (NSP). The cell employed conformal layers of yttrium-doped barium zirconate

(BYZ) anhydrous electrolyte membrane (~120nm) sandwiched between thin (~70nm)

sputtered porous Pt electrode layers. The fuel cell structure achieved the highest reported

peak power densities up to 186 mW/cm2 at 450

oC using hydrogen as fuel.

6.1 Introduction

Lowering the operating temperature of ceramic fuel cells is desirable to circumvent issues

related to stability of cell materials and performance. In this regard, there have been

studies in our laboratory [1, 2] and by others [3-6] that aim at achieving lower operating

122

temperatures primarily by employing thin film solid oxide fuel cell (SOFC) structures

with yttria stabilized zirconia (YSZ) electrolyte membranes that exhibit transport of oxide

ions in the oxygen sublattice with activation energies of ~1eV.

Alternatively, proton conducting anhydrous ceramic electrolytes offer opportunities to

lower the operating temperature of oxide-based fuel cells due to high ionic conductivity.

Proton transport in several members of acceptor-doped perovskites of the general formula

ABO3 is reported to be fast, with activation energies of about 0.45eV [7], which makes

them of interest as potential solid electrolytes for next generation protonic devices such

as fuel cells, electrolyzers, hydrogen sensors, and gas reformers [7-18]. Proton transport

in these anhydrous oxides occurs via the Grotthuss mechanism through hydroxide defects

that are produced by water incorporation into oxide ion vacancies generated in the crystal

lattice upon extrinsic doping of the tetravalent Zr+4

site by the trivalent Y+3

ion. Among

the anhydrous oxides, yttrium-doped BaZrO3 (BYZ) shows better chemical stability in

acidic gas environments, as well as higher proton conductivity than doped BaCeO3 in the

intermediate temperature regime [7, 16].

We have previously reported on the fabrication, properties, and performance of thin film

planar (2D) fuel cells employing both oxide ion conducting yttria-stabilized zirconia

(YSZ) [1] and proton conducting BYZ membrane electrode assemblies (MEAs) made by

MEMS processing [19]. A recent report provides an extensive review of previous studies

and cell performance results for micro-fuel cells with planar geometries [6].

123

This study reports on the first fabrication and operation of a three-dimensional (3-D)

close packed crater shaped thin film proton conducting ceramic fuel cell MEA using

nano-sphere patterning (NSP) towards achieving enhancement in the effective reaction

surface area. Equally important, such a 3-D architecture is expected to be mechanically

more compliant than planar thin films, although no mechanical testing to verify this

hypothesis was undertaken as part of this research. This paper also reports the fuel cell

performance of this 3-D BYZ MEA structure in the temperature range of 350~450oC.

Similar work in our laboratory aimed at enhancing mechanical stability as well as

increasing the electrochemical reaction area was recently reported for a 3-D solid oxide

fuel cell (SOFC) that featured cylindrical cup architecture with vertical walls [2]. In

comparison, the NSP method described in the present study offers simplicity in

fabrication and also yields slanted walls, which facilitate effective and conformal coating

of the side walls with porous electrode layers even using an inherently directional

physical vapor deposition (PVD) method such as sputtering. This opens up wide range of

material selection for electrolytes and electrodes.

124

6.2 Experimental

The sequence of NSP and MEMS processing depicted in Figure 6-1 produces a silicon

nano-trench structure upon which the BYZ fuel cell MEA is fabricated. First, spherical

silica particles with 700nm diameter were created by a modified Stöber synthesis

technique. Using a Langmuir-Blodgett (L-B) trough, the silica particles were transferred

onto a silicon (100) substrate to form a close-packed monolayer. Then, plasma etching is

used to uniformly reduce the size of the particles to open up sufficient spacing among the

particles for the deposition of the metal mask. A dense aluminum (Al) layer (~100nm) is

deposited by DC magnetron sputtering at a partial pressure of 1Pa argon (Ar) and 100W

power at room temperature (Figure 6-1(b)). The silica particles were mechanically

removed from the substrate by ultra sonication leaving behind a nano-pore structured Al

layer on the silicon substrate (Figure 6-1(c)). Using this metal layer as a mask and

reactive ion etching (RIE) in a gaseous environment of sulfur hexafluoride (SF6) and

chlorodifluoromethane (CHClF2), combined processes of isotropic and anisotropic

etching is carried out to form the silicon nano-trench structure (Figure 6-1(d)). 100nm

thick conformal low stress nitride layers are deposited on both sides of the wafer by low

pressure chemical vapor deposition (LPCVD) (Figure 6-1(e)). A square opening is

created on the backside of the Si wafer using a simple photolithography process for

solution etching. The backside of the silicon template is then etched in 30% potassium

hydroxide (KOH) solution at 85~90oC for 5 hours utilizing the top silicon nitride as an

125

etch stop layer (Figure 6-1(f)). A free-standing crater shaped silicon nitride layer is thus

created in the square opening with a projected dimension of 100µm x 100µm.

Figure 6-1. Schematic illustration of the NSP processing sequence for the fabrication of

3-D crater patterned freestanding fuel cell MEAs. (a) Si substrate. (b) Silica particles with

Al layer on Si wafer. (c) Al mask after removing the particles. (d) Formation of trenches

by RIE etching. (e) Silicon nitride deposition. (f) Removal of Si template by KOH

etching. (g) Depositing BYZ and removal of the nitride layer by dry etching. (g)

Sputtering of porous Pt catalyst/electrode (dots) deposited on both sides (h).

126

120nm thick BYZ electrolyte layer is grown on this 3-D nitride template at 400oC at a

rate of 0.28Å /pulse by pulsed laser deposition (PLD) (Figure 6-1(g)) using a sintered

BaZr0.8Y0.2O3-δ target (Praxair Inc.) separated by 65mm from substrate. A Lambda Physik

248nm KrF excimer laser with energy density of 1.8J/cm2 per pulse was used for ablating

the target in 100mTorr oxygen environment. Samples cooled down naturally in 300Torr

oxygen pressure. After removing the nitride layer by plasma etching, 70nm thick porous

platinum electrodes were deposited at room temperature by DC sputtering at 50kW and

10Pa Ar pressure on both sides of the BYZ membrane to complete fabrication of the 3-D

crater patterned BYZ fuel cell MEA (Figure 6-1(h)).

Fuel cell performance was measured inside a test chamber using a micro-manipulating

probe station developed earlier [1]. Pure dry H2 was supplied to the anode side while the

cathode was exposed to ambient air. Gamry Potentiostats and Echem Analyst software

(Gamry Instruments) were used for fuel cell testing. Electrochemical impedance

spectroscopy (EIS) measurements were performed at various cell voltages in the

frequency range of 300 kHz - 1Hz.

127

6.3 Results and Discussion

The SEM images along the fabrication sequence of the silicon nano-trench structure and

the crater shaped BYZ electrolyte fuel cell MEA are shown in Figures 6-2 and 6-3,

respectively. The resulting silicon nano-trenches have depths of about 550-600nm. The

shape of the trenches is 3-D trapezoidal rather than semispherical or conical, due to the

isotropic/anisotropic etching recipe employed in the fabrication process. Due to the

beveled (or slanted, as opposed to vertical) walls of the trenches, a conformal BYZ

coating about 120nm thick was possible to achieve with PLD (Figure 6-3). The cross-

sectional SEM image (Figure 6-3(d)) of the finished crater patterned MEA shows the Pt

electrode layers (~70nm) whose porosity was estimated to be around 30-40%. Naturally,

the purpose of having porous electrodes is to increase the electrochemically active region,

i.e., the triple phase boundary (TPB). As expected, the upper side of the BYZ layer is

slightly thicker than inside the trenches due to the directional deposition mechanism of

PVD. Nevertheless, conformality of both the BYZ and Pt layers were satisfactory. The

inside surfaces of the crater structure adopt a quadrangular pyramid shape that resemble a

crater. Using high resolution SEM images, the geometric enhancement achieved in the

surface area by this 3-D structure is estimated to be about a factor of 1.7-1.8 per projected

planar area.

128

Figure 6-2. SEM images of, (a) silicon nano-trenches after removal of spherical particles,

(b) silicon nano-trench structure created after gas phase etching, (c) free-standing 3-D

nitride template after removing the backside silicon by KOH etching.

129

Figure 6-3. SEM images of the crater patterned BYZ fuel cell MEA, (a) after BYZ

deposition on the 3-D nitride template, (b,c) after porous Pt electrodes are coated on both

sides of the membrane, and (d) finished 3-D BYZ MEA taken from an angle of 52o from

the top (d).

130

Further microstructural analysis of the BYZ grain structure was carried out using a

transmission electron microscopy (TEM). BYZ specimens were made using a focused ion

beam (FIB, FEI Strata 235DB dual-beam FIB/SEM) lift-out Omniprobe technique

employing a Ga ion beam at 30 keV. Due to difficulties in preparing TEM specimens

directly from the 3-D MEAs, a planar 120nm thin BYZ film was deposited under

identical PLD conditions on a silicon wafer with silicon nitride overcoat. Cross-sectional

high-resolution (HRTEM) images and selected area diffraction (SAD) patterns were

taken by an FEI Tecnai G2 F20 X-TWIN microscope operated at 200 kV. The HRTEM

images and SAD patterns of this surrogate sample shown in Figure 6-4 clearly indicate a

dense columnar BYZ film with clean grain boundaries extending vertically through the

film thickness, making a well-defined interface with silicon nitride. Although the HR-

TEM images in Figure 6-4 were not produced from the 3-D MEAs, the microstructure of

the surrogate sample is nevertheless expected to represent the BYZ film in the actual

MEAs.

Results of electrochemical impedance spectroscopy (EIS) and cell performance of the 3-

D MEA are presented in Figure 6-5(a) and 6-5(b) respectively, for the fuel cell,

Air, Pt/BYZ/Pt, pure H2 (6.1)

131

In this fuel cell the transporting specie is protons, and consequently, the oxidation

product H2O forms at the air electrode (i.e., cathode) by the net reaction

eOHOH 22

12 22 . This is in contrast to SOFC based on oxide ion conducting

YSZ electrolyte, where the net cathode reaction is the reduction of molecular oxygen to

oxide ions ( 2

2 22

1 OeO ) while the oxidation reaction to form H2O occurs at the

H2 electrode (i.e., anode).

Figure 6-4. Cross sectional HRTEM images ((a) and (c)) showing the dense columnar

grain structure, and, (b) the SAD pattern indicating the fully developed polycrystalline

nature of the BYZ film.

132

The EIS spectra taken at cell voltages of 0.7 and 0.9V indicate two semi-circles. The high

frequency semi-circle in Figure 6-5(a) shows no discernable changes with varying

voltage while the second semi-circle at lower frequencies is significantly affected. This

suggests that the first semicircle is due to the BYZ electrolyte, and indeed, the impedance

of the high frequency loop matches the expected ohmic resistance. The second semicircle

at the lower frequency regime is due to electrode processes, and to a large extent due to

the cathode because of significantly higher polarization at the cathode than at the anode

reported for similar ceramic-based protonic fuel cells [20-22].

133

Figure 6-5. (a) Electrochemical impedance spectra at 400oC at cell voltages of 0.9V and

0.7V, with inset showing the details of the high frequency region, and (b) voltage-

current-power density (V-I-P) behavior of 3-D crater patterned BYZ fuel cells measured

at 350-450oC using hydrogen fuel.

134

Cell polarization curves in Figure 6-5(b) indicate significant activation loss. Open circuit

voltages (OCV) and peak power densities, respectively, were 1.01V and 30mW/cm2

at

350oC, 0.87V and 69mW/cm

2 at 400

oC, and 0.85V and 186mW/cm

2 at 450

oC. The

performance of the 3-D MEAs demonstrated at 350-450oC in this study is superior to

other studies reported in the literature, even after accounting for the enhancement in the

surface area. For example, Peng et al. recently reported 25mW/cm2 at 800

oC with a

protonic fuel cell employing pure hydrogen as fuel and featuring a 10µm thick ZnO-

doped BYZ membrane with porous Pt electrodes, while the performance of a similar cell

with 1mm thick BYZ at the same temperature was less than 4mW/cm2 [23]. Similarly,

Traversa and coworkers have reported power densities of 7mW/cm2 at 700

oC for protonic

cell with 0.6mm thick BYZ [24] and more recently, 110mW/cm2 at 600

oC from a cell

employing pulsed laser deposited BYZ membrane [25]. More recently, Sun et al

employed 20µm thick BYZ membrane on NiO/BaZr0.1Ce0.7Y0.2O3 cermet anode support

with a Sm0.5Sr0.5CoO3–Ce0.8Sm0.2O2 composite cathode and achieved 170mW/cm2 at

700oC [26].

Although the performance values achieved in this study are arguably the highest reported

in the literature for BYZ based protonic fuel cells in this intermediate temperature

regime, they are nevertheless several fold inferior to other reports especially those from

our own laboratory that employed hydrogen fuel in thin film oxide ion conducting

SOFCs in the same temperature regime [1, 2]. This is intuitively puzzling, since proton

diffusion in BYZ is inherently orders of magnitude faster than oxide ion transport in YSZ

135

in the same temperature regime. We have several speculations for this lower performance

than expected. First, we suspected that the reason for inferior performance may be due to

the formation of a carbonate layer (i.e., BaCO3) at the air side on the BYZ surface, which

would then severely block both electron and proton transport. In this temperature regime,

Kreuer’s data suggest that BYZ would be susceptible to form a carbonate layer with CO2

in the ambient air that has nominally a CO2 concentration of about 380ppmv [7]. Indeed

XPS analysis of the BYZ films kept in a covered sample box in ambient air for a week

indicated clear presence of a carbonate formation.

To determine BYZ’s propensity for carbonate formation due to susceptibility to CO2 in

the ambient air, we deposited a BYZ thin film by PLD on a 100 nm low stress silicon

nitride film grown by LPCVD on a silicon wafer. The PLD process conditions and

thickness of this 2-D sample were maintained the same as that for the 3-D BYZ MEA.

The BYZ film sample was stored inside a sample box with the lid closed (but not air-tight

sealed). After one week of minimal exposure to ambient air in the laboratory at room

temperature, the surface composition of the BYZ film was characterized in x-ray

photoelectron spectroscopy (XPS) using Physical Electronics (PHI) Quantum 2000

scanning ESCA microprobe with monochromatic Al Kα X-ray source (h = 1486.6 eV).

The depth profile of the BYZ sample was obtained with 1 keV Ar+ ion sputtering. The

carbon content was 37% at the surface but decreased monotonically with the sputtering

depth.

136

0 10 20 30 400

10

20

30

40

50

60

70

Ato

mic

concentr

ation (

%)

Depth (nm)

C

O

Y

Zr

Ba

Figure 6-6. Compositional depth profiles of the PLD BYZ film.

Figure 6-7: The high resolution C1s

spectra show two peaks at ~285.0eV and ~289.9eV

assigned to surface contamination and to CO32-

(possibly in the form of BaCO3)

environment, respectively.

137

To analyze the bonding state of the carbon, high resolution XPS scans for C1s

peak were

performed at each depth (Figure 6-7). At the surface, two C1s

peaks were observed. The

one near 285.0eV is the signature for surface contamination commonly observed due to

species in the ambient, but the other C1s

peak at 280.9eV represents carbon in CO32-

state.

The C1s

peak around 280.9eV was strongest at the surface and decreased with depth. This

matches well with the diminishing carbon content below about 20nm as shown in Figure

6-6. Thus, the carbon in the BYZ film is identified as carbonate, providing strong

evidence to suggest carbonation of the BYZ sample, most likely into BaCO3.

Also, observed OCV values were somewhat lower than theoretically expected. We

suspect that this is most likely due to gas leakage through the Au-ring seals in the cell

holder, or chemical shorting across broken MEA windows on the chip. The latter is a

frequent problem with free-standing MEA windows due to the uncontrolled mechanical

force exerted by the microcontacting probe during electrochemical testing. Existence of

pinholes in the BYZ layer, not uncommon in thin PLD films, may also be responsible.

The possibility of electronic conductivity in BYZ is less likely at these temperatures.

Indeed, oxide ion transference number in (Ba,Ca)(Zr,Y)O3 is measured to be 0.97-0.99 in

dry atmosphere at high temperatures between 600 and 1000oC [20]. The balance is most

likely due to protonic conduction. A similar finding was reported by Bonanos [27] at

138

elevated temperatures. So electronic conduction in BYZ at 350-450oC may be negligibly

small, and most likely, would not account for low OCVs.

In addition, it is known that the proton conductivity of BYZ through the grain boundary

is very low due to extremely high resistance. PLD BYZ film on the silicon nitride formed

polycrystalline with nano size grains. This high grain boundary density might lower the

proton conductivity of the BYZ electrolyte. Details regarding the relationship between

crystal structure of BYZ and proton conductivity will be discussed in the next chapter.

Although quantitative comparison of 3-D BYZ cell performance reported in this study

with 2-D BYZ cells reported earlier [19] is not meaningful due to poor control of

reproducibility of the Pt/BYZ interfacial microstructure, 120mW/cm2 reported earlier for

the 2-D PLD BYZ cell at 450oC [19] scales with 186mW/cm

2 for the 3-D cell at this

temperature by a factor of ~1.5. Since the cathode largely dominates overall cell behavior

[20-22] this scaling factor is also representative of the relative cathodic losses observed

for the 3-D (this study) and planar BYZ geometries [19]. This is in general agreement

with 1.7-1.8 enhancement in the geometric area achieved by the 3-D architecture.

139

6.4 Conclusion

This study constitutes the first reporting of three-dimensional crater patterned proton

conducting MEAs that were successfully fabricated and operated in fuel cell mode. The

3-D fuel cell architecture, which consisted of conformal layers of 120 nm thick BYZ

electrolyte with 70nm porous Pt electrodes, was accomplished by employing a Langmuir-

Blodgett based nanospherical patterning technique combined with MEMS processing

methods. A factor of 1.7-1.8 enhancement in the surface area was obtained. The 3-D

MEAs were tested with the H2/air couple in the temperature range of 350-450oC, and

achieved a power density of 186mW/cm2 at 450

oC. This constitutes the best performance

reported in the literatures for BYZ-based protonic fuel cells in this intermediate

temperature regime. Though this thin film fuel cells show outstanding fuel cell

performance, the mechanical stability of such a thin film structure could be a challenge.

To resolve the issue, on-chip fabrication method is under development in our laboratory.

6.5 References

[1] H. Huang, M. Nakamura, P. Su, R. Fasching, Y. Saito, F. B. Prinz, J. Electrochem.

Soc., 154, B20 (2007)

[2] P. C. Su, C. C. Chao, J. H. Shim, R. Fasching, F. B. Prinz, Nano Lett., 8, 2289-2292

(2008)

140

[3] Y. Jiang, A. V. Virkar, J. Electrochem. Soc., 150(7), A942 (2003)

[4] S. DeSouza, S. J. Visco, L. C. DeJonghe, Solid State Ionics, 98, 57 (1997)

[5] E. D. Wachsman, ECS Trans., 25(2), 783 (2009)

[6] A. Evans, A. Bieberle-Hütter, J. Rupp, L. J. Gauckler, J. Power Sources, 194, 119-

129 (2009)

[7] K. D. Kreuer, Ann. Rev. Mat. Res., 33, 333 (2003)

[8] W. Münch, G. Seifert, K. D. Kreuer, J. Maier, Solid State Ionics, 97, 39-44 (1997)

[9] W. Münch, K. D. Kreuer, S. Adams, G. Seifert, J. Maier, Phase Transitions, 68, 567-

586 (1999)

[10] R. C. T. Slade, N. Singh, Solid State Ionics, Diffusion & Reactions, 46, 111-15

(1991)

[11] N. Bonanos, B. Ellis, M. N. Mahmood, Solid State Ionics, 44, 305-11 (1991)

[12] N. Kuwata, N. Sata, T. Tsurui, H. Yugami, Jpn. J. App. Phys., 44, 8613-8618

(2005)

[13] T. Hibino, A. Hashimoto, M. Suzuki, M. Sano, J. Electrochem. Soc., 149, A1503-8

(2002)

[14] H. Iwahara, H. Uchida, K. Morimoto, J. Electrochem. Soc., 137, 462-465 (1990)

[15] H. Iwahara, H. Uchida, K. Ono, K. Ogaki, J. Electrochem. Soc., 135, 529-533

(1988)

[16] H. Iwahara, T. Yajima, T. Hibino, K. Ozaki, H. Suzuki, Solid State Ionics, 61, 1-3

(1993)

[17] N. Ito, M. Iijima, M. Kimura, S. Iguchi, J. Power Source, 152, 200 (2005)

141

[18] U. Balachandran, T. H. Lee, B. Ma, S. E. Dorris, Mater. Res. Soc. Symp. Proc., 972,

AA01-09 (2007)

[19] J. H. Shim, J. S. Park, J. An, T. M. Gür, S. Kang, F. B. Prinz, Chem. Mater., 21,

3290-3296 (2009)

[20] H. Uchida, S. Tanaka, H. Iwahara, J. Appl. Electrochem., 15, 93 (1985)

[21] H. Yamaura, T. Ikuta, H. Yahiro, G. Okada, Solid State Ionics, 176, 269 (2005)

[22] H. Yahiro, H. Yamaura, M. Asamoto, Mater. Res. Soc. Symp. Proc., 972, AA01-06

(2007)

[23] C. Peng, J. Melnik, J. Li, J. Luo, A. R. Sanger, K. T. Chuang, J. Power Sources, 190,

447 (2009)

[24] A. D’Epifanio, E. Fabbri, E. Di Bartolomeo, S. Licoccia, E. Traversa, Fuel Cells,

08(1), 69 (2008)

[25] D. Pergolesi, E. Fabbri, E. Traversa, abstract submitted to the 218th

Meeting of The

Electrochemical Society, Las Vegas, October 10-15 (2010)

[26] W. Sun, L. Yan, Z. Shi, Z. Zhu, W. Liu, J. Power Sources, 195, 4727 (2010)

[27] N. Bonanos, Solid State Ionics, 53-56, 967 (1992)

142

CHAPTER 7. Effect of Crystallinity on Proton Conductivity in

Yttrium-doped Barium Zirconate Thin Films

In previous chapter, we demonstrated a novel fabrication process to create 3-D proton

conducting ceramic fuel cell architecture by employing nanosphere lithography technique

(NSL). Despite BYZ’s higher proton conductivity than the oxide ion conducting

electrolytes at LT-SOFC regime, the fuel cell performance was lower compared to the

fuel cell performance conducted previously in our laboratory using YSZ electrolyte. Even

though the bulk ionic conductivity of BYZ is very high, it is known that the total

conductivity is quite low due to the extremely high grain boundary resistance. In this

chapter, we investigated the effect of crystallinity on proton conductivity in BYZ thin

films grown 120nm in thickness on amorphous (quartz) and single crystal MgO(100)

substrates has been studied. The conductivity was measured in the temperature range of

150~350oC. By altering the film deposition temperature, varying degrees of

crystallization and microstructure were observed by x-ray diffraction and transmission

electron microscopy. The epitaxial BYZ film grown on MgO(100) substrate at 900oC

showed the highest proton conductivity among other samples with an activation energy of

0.45eV, whereas polycrystalline and amorphous BYZ films showed lower conductivities

due to grain boundaries in their granular microstructure.

143

7.1 Introduction

Due to their high proton conductivity [1-12], doped perovskite oxides have been widely

studied in recent years as proton conducting solid electrolytes for a variety of

electrochemical devices such as fuel cells, hydrogen sensors, electrolyzer, and hydrogen

pumps. Best known and intensively investigated examples of such anhydrous perovskites

are alkaline-earth cerates and zirconates. Despite high proton conductivity of barium

cerate based materials [13-15], their chemical susceptibility to reactions with acid gases

(e.g., CO2, SO2) and moisture makes them unsuitable electrolytes for most fuel cell

applications [16-18]. However, Y-doped BaZrO3 (BYZ) has been considered one of the

most promising electrolyte materials for protonic fuel cells for its significant proton

conductivity as well as excellent chemical stability [10].

In general, doped perovskites have oxygen vacancies that can absorb water molecules

which give rise to protonic defects via the reaction,

H2OVO OO

X 2(OH)O (7.1)

where,

VO denotes an oxygen vacancy in the oxygen sublattice in BYZ,

OOX represents

neutral oxygen in its normal lattice site in BYZ, and

(OH )O is the protonic defect

associated with a lattice oxygen in BYZ. Principal proton transport mechanism in doped

144

perovskites is generally described as a two-step Grotthuss-type diffusion mechanism

[19], which consists of fast rotational diffusion of the protonic defect followed by proton

transfer to the neighboring oxide ions, the latter most likely being the rate-limiting step

[1, 2, 20-21]. At the low-to-intermediate temperature regime, bulk ionic conductivity in

BYZ is higher than that of oxide ion conducting ceramics. However, most fuel cell

performance values reported for proton conducting fuel cells with BYZ is much lower

than the performance with oxide ion conducting ceramics such as YSZ. We speculate that

this can be due not only to the presence of high density of grain boundaries but also to the

degree of crystallinity and evolution of the crystal structure in BYZ. In this regard,

deviations from the cubic perovskite structure may impact the formation and mobility of

protonic charge carriers. The mobility of protonic defects in perovskites with structures

deviating from its parent cubic is significantly lower [22] and this effect has been

investigated by researchers in detail comparing structural and dynamical features of

protonic defects in Y:BaCeO3 and Y:SrCeO3 [3]. Also for doped BaZrO3, it is generally

agreed that structural distortions and grain separation lead to decreased proton mobility

and result in a large grain boundary resistance [2, 23-24]. Kreuer [2] describes grain

separation as highly distorted BYZ grains making reduced number of point contacts and

separated by a grain boundary region that exhibit high impedance. Particularly, the highly

refractory nature of barium zirconate compounds usually results in small grains and

consequently a high grain boundary density with considerable impedance for proton

transport [25]. As a result, the overall conductivity is significantly reduced [2, 13, 26-27].

Therefore, it is important to optimize the fabrication process and the resulting properties

145

in terms of structure, microstructure and density of grain boundaries in order to achieve

high performance in fuel cell and other protonic device applications.

In this work, we studied the effect of crystallinity on proton conductivity in BYZ using a

two-prong strategy. For one, the degree of crystallinity was varied between amorphous to

a fully crystalline structure. The second strategy involved the comparison between single

crystalline (epitaxial) BYZ and polycrystalline BYZ structures. To serve these purposes,

we have systematically fabricated BYZ films with varying degrees of structure and

microstructure formation and investigated proton conductivity in relation to structural

features. For these experiments, we chose two different substrates, namely, MgO(100)

and quartz. MgO is ideally suited for epitaxial growth of thin BYZ films, because it has a

cubic rock-salt structure with the lattice constant of 4.21Å that matches BYZ (4.19Å )

perfectly well. Amorphous quartz substrate is used for fabricating amorphous nano-

granular BYZ films to study evolution and formation of the crystalline films deposited at

different temperatures.

7.2 Experimental

BYZ deposition is carried out by utilizing the pulsed laser deposition (PLD) technique. A

sintered BaZr0.8Y0.2O3-δ pellet (Praxair Inc.) was used as a target material to deposit the

ultra thin BYZ films. A Lambda Physik 248nm KrF excimer laser with the energy

146

density of 1.8J/cm2 per pulse was used to ablate the target in 100mTorr background

oxygen gas environment in the deposition chamber and the sample-to-target distance was

maintained at 65mm during depositions. The deposition temperature, measured by a

thermocouple placed in the center of the substrate heater, was varied from 400oC to

900oC for quartz substrates and 600

oC to 900

oC for MgO(100) substrates. After

deposition, samples were naturally cooled down in the oxygen environment with the

camber pressure of 300Torr. The thickness of the BYZ films grown on quartz and

MgO(100) substrates were 130nm with a growth rate of ~0.3Å /pulse with a pulse

repetition rate of 8Hz.

Crystallinity and structural phase of the deposited films were analyzed by X-ray

diffraction (XRD) method for both quartz and MgO(100) samples using a PANalytical

X’Pert PRO XRD system (Cu Kα X-ray with λ=1.54Å ) utilizing the symmetrical θ/2θ

scan method for phase analysis. X’Pert HighScore Plus software (PANalytical) was used

to estimate the degree of crystallization. To confirm epitaxial growth and poly

crystallinity, cross-section image and electron diffraction patterns of the BYZ-MgO

specimens were obtained using transmission electron microscopy (TEM). Specimens

with a thickness of ~80 nm were made for TEM analysis using a focused ion beam (FIB,

FEI Strata 235DB dual-beam FIB/SEM) lift-out Omniprobe technique with a Ga ion

beam at 30keV. Cross-sectional high resolution transmission electron microscopy

(HRTEM) images and selected area diffraction (SAD) patterns were taken by an FEI

Tecnai G2 F20 X-TWIN operated at an accelerating voltage of 200 kV.

147

Figure 7-1. Conductivity measurement setup with microcontacting probes connected to

the EIS software.

For conductivity measurements, a planar cell geometry shown in Figure 7-1 was

employed. Platinum pads (600µm x 600µm) with a thickness of about 100nm were

deposited on the BYZ films in a planar fashion utilizing DC sputtering technique. The

cells are mounted on a temperature controlled heating station. AC impedance

spectroscopy data were obtained in ambient air (RH=30%) by using a home-made

mircomanipulating probe set-up. AC impedance measurements were made over a

frequency range from 100 kHz to 1Hz with signal amplitude of 100mV at temperatures

from 150 to 350oC for BYZ/MgO(100) samples. Impedance measurements were done in

at open circuit condition (0mV bias) or under 500mV dc bias. It is known that in this

temperature regime proton conduction in BYZ is dominant [25]. A Gamry Potentionstat

148

(Gamry Instruments, Inc) unit and ZView software (Scribner Associate, Inc) were used

for collecting and analyzing electrochemical impedance spectroscopy (EIS) data.

7.3 Results and Discussion

Figure 7-2. XRD patterns of BYZ thin films grown on quartz substrates (Q) with

different deposition temperatures (a) 900oC, (b) 700

oC, and (c) 400

oC.

XRD analysis indicated increase in the extent of crystallization of the BYZ film as the

deposition temperature is increased, irrespective of the substrate. Figure 7-2 shows the

XRD patterns of BYZ thin films deposited on amorphous quartz substrates with different

149

deposition temperatures. The BYZ film deposited at 400oC (Figure 7-2(c)) shows no

discernable peaks which indicates that the film has amorphous or nano-granular structure.

As the deposition temperature is increased, it shows crystallization and the formation of a

polycrystalline film (Figure 7-2(b) and 7-2(a)).

Figure 7-3. XRD patterns of BYZ thin films grown on MgO(100) substrates with

different deposition temperatures (a) 900oC, (b) 800

oC, (c) 700

oC, and (d) 600

oC.

Diffraction patterns of BYZ films grown on MgO(100) are shown in Figure 7-3. As

previously mentioned, the deposition temperature was varied from 600oC to 900

oC. At

the low temperature regime (600oC), the XRD pattern shows the evolution of a

polycrystalline structure rather than epitaxial. As the temperature is increased from 600oC

to 700oC, it shows the formation of a polycrystalline film (Figure 7-3(d) and 7-3(c)). We

150

speculate that initially it follows the guiding structure, which is (100), but as the film gets

thicker it relaxes and starts to form grains with different orientations. At 800oC the BYZ

film seems to have less crystalline peaks and at 900oC, only a strong BYZ(100) peak,

which indicates perfect epitaxy of BYZ, was observed (Figure 7-3(b) and 7-3(a),

respectively). Previously, epitaxial growth of BYZ film on MgO(100) was demonstrated

at around 800oC by Shim et al. [28]. The polycrystallinity as well as epitaxial growth of

BYZ thin films on MgO(100) substrates observed by XRD were further corroborated

with TEM cross-sectional images and diffraction patterns (Figure 7-4, 7-5, and 7-6).

151

Figure 7-4. High resolution TEM image: {001} planes of BYZ (perovskite structure)

grows epitaxially on {001} type planes of MgO (rocksalt structure). A yellow-dotted

rectangle shows dimensional matching of each MgO and BYZ unit cell.

152

Figure 7-5. Selected area diffraction (SAD) patterns of BYZ films deposited by PLD at

900 °C (a series at top) and at 600 °C (b series at bottom): Both a-1 and b-1 SAD patterns

were taken only from MgO for setting orientation standard. Both a-2 and b-2 SAD

patterns were taken from MgO and BYZ to check orientation relationship of BYZ films

to MgO substrate. BYZ film deposited by PLD at 900 °C (a-2) shows epitaxial growth,

unlike BYZ film deposited by PLD at 600 °C, which illustrates slight disorientation. A

digitally 4X magnified SAD of 200 type spots (b-3) confirms orientation mismatch of

each film.

153

Epitaxial growth of BYZ film deposited at 900 °C on MgO (100) substrate is clearly

visible in the high resolution TEM cross sectional image in Figure 7-4. The yellow-dotted

region in Figure 7-4 indicates a unit cell matching that of perovskite (BYZ) and rocksalt

(MgO) structures and clearly shows the lattice match between the BYZ film and the MgO

substrate. Figure 7-5 shows the SAD patterns of epitaxial and polycrystalline BYZ films

on MgO(100) using a 150 nm diameter aperture. The SAD patterns in Figure 7-5a-2

again corroborate the epitaxial growth based on that perovskite (001) type spots of BYZ

and they are completely matched to rocksalt (001) spots.

Unlike the epitaxial BYZ film deposited at 900°C on MgO(100) substrate, the BYZ film

deposited at 600°C on MgO(100) shows misorientation. As seen in Figure 7-5b-2, (110)-

related diffraction spots that can only be generated from the BYZ film shows an angle

divergence, which indicates lattice misorientation and polycrystallinity. Digitally 4X

magnified SAD pattern for (200) spots in the blue box region in Figure 7-5b-2, which

includes both MgO and BYZ diffraction spots, points out polycrystallinity of the BYZ

film with the BYZ {200} planes slightly misoriented from {200} planes of the MgO

substrate. However, because diffraction spots of same type planes of BYZ and MgO still

show very similar position (distance from center beam to diffraction spots), similarity of

lattice parameter of BYZ film and MgO substrate can be inferred.

154

20 nm20 nm 20 nm20 nm

BYZBYZ

a) Bright Field TEMa) Bright Field TEM b) Dark Field TEMb) Dark Field TEM

BYZBYZ

Grain boundaries

c) High Resolution TEMc) High Resolution TEM d) SAD (from BYZ)d) SAD (from BYZ)

Random orientation of grainsRandom orientation of grains2 nm2 nm

Figure 7-6. Bright Field (BF), Dark Field (DF), High Resolution (HR) TEM images and

SAD pattern of BYZ films deposited by PLD at 600 °C on Quartz: Both BF (a) and DF

(b) images show visual orientation difference of each BYZ grain. HRTEM (c) shows

polycrystallinity of BYZ films with grains and grain boundaries. And SAD pattern only

taken from BYZ indicates randomly-oriented of polycrystalline BYZ grains.

155

Figure 7-6 shows TEM cross-section images and SAD of BYZ thin film deposited on

amorphous quartz substrate at 600oC. Bright Field TEM (imaged only from centered

diffraction beam) and Dark Field TEM (imaged only from specific outer diffraction

beam) in Figure 7-6-a and b are showing direct visual of polycrystallinity of BYZ film.

The contrast difference in BFTEM and DFTEM stems from orientation difference of each

grain if other conditions like materials composition or thickness are reasonably same over

the imaging area. HRTEM image directly shows grain boundaries and orientation

difference of each columnar grain with lattice fringe. SAD pattern analysis also verifies

random orientation of those grains.

For most samples, the impedance spectra showed a single semicircle where the real-axis

intercept corresponds to the total ionic resistance of the electrolyte (Figure 7-7). The

spectrum was fitted to a parallel R//C circuit, where R denotes the ohmic resistance

obtained from the EIS spectra, and C is capacitance. In addition, we varied the dc bias

applied to the samples during ac impedance spectroscopy measurements. While the

electrode processes are highly affected by the magnitude of dc bias, the ionic transport

process across the grain boundary and within the bulk grain (or, intra-grain) is generally

independent of dc bias conditions. Previously, using applied dc biases up to 14V Guo et

al [29] reported voltage dependence of grain boundary impedance in oxide ion

conducting yttria-stabilized ceria (YDC) solid electrolyte. Unfortunately, they did not

156

provide independent evidence or experimental verification (e.g., using YDC samples with

different grain sizes leading to different grain boundary densities and impedances) that

their assignment of the grain boundary impedance to the second semicircle in their

impedance spectra (of Figure 2 of Guo et al [ref. 29]) was indeed correct and justified.

Assuming their analysis is correct nevertheless, their results indicate that bias dependence

of grain boundary impedance becomes discernable only at high bias values greater than

2-3V, whereas at 0V and 0.3V biases, their impedance spectra (in Figure 2 of ref. 29) did

not show evidence of bias dependence. This is also indicated in Figure 3 of ref. 29, where

at low biases over single boundaries dependence on voltage is quite weak, and perhaps

difficult to discern experimentally.

157

Figure 7-7. Measured Nyquist impedance plots and fitting curves to the parallel R//C

circuit model. (a) BYZ-MgO(100) sample deposited at 900oC and measured at 200

oC.

Bias independence of the spectra indicates that the semicircle is associated with

electrolyte impedance. (b) BYZ film deposited on quartz at 400oC and measured at

700oC.

158

This was indeed the case in the present study where no discernable changes are observed

in Figure 7-7(a) between 0mV and 500mV bias conditions and the equivalent fitting

curves are well matched. This was observed for both samples at all measured temperature

ranges (see Figure 7-8). This confirmed and verified that the measured EIS semicircle is

not from electrode but from electrolyte processes.

Figure 7-8. EIS data measured at different temperatures for BYZ-MgO(100) film

deposited at 900oC.

The total conductivity values were extracted from the measured electrolyte impedances at

different temperatures. However, this total conductivity value includes two contributions,

namely, intra-grain and grain boundary resistances. The latter contribution in

polycrystalline BYZ is known to be high [2, 23-24]. Indeed, the impedance values for the

epitaxially grown BYZ films are significantly smaller than those for polycrystalline

159

samples, as expected. Furthermore, the capacitance values estimated from the peak

frequency and the width of the real axis intercept is of the order of 10pF (~ 4–5pF in

Figure 7-8), a typical value for geometric capacitance.

Figure 7-9. Arrhenius plot showing the conductivity values of BYZ thin films deposited

on MgO(100) and amorphous quartz substrates at different deposition temperatures. In

addition, both sets of data are compared with the reference conductivity values from the

literature, including bulk and experimentally obtained non-epi references.

By using total conductivity values and the general Arrhenius relation, we obtained

activation energies for ionic conduction in each BYZ sample:

160

kT

ET aexp0 (7.2)

where Ea is the activation energy, k is the Boltzmann constant, T is the absolute

temperature, and σ0 is a pre-exponential constant. The Arrhenius plot of Figure 7-9 shows

the total conductivity data of thin BYZ films deposited on two types of substrates at

different temperatures. From the plot, the activation energies for the BYZ films deposited

on MgO(100) sample were calculated as 0.45eV for 800oC and 900

oC deposited films, in

good agreement with the reference value of 0.44eV [2, 28, 30] for bulk proton transport,

but showed slightly higher values (0.51-0.53eV) for films deposited at lower

temperatures of 600oC and 700

oC. In a similar study, Traversa and coworkers recently

reported an activation energy of 0.63eV for proton transport in 1m thick BYZ epitaxial

films also grown by PLD at 600oC on MgO(100) substrate [31]. The authors explained

this relatively high activation energy by sample-to-sample differences both in the

structure and protonation of the films.

Interestingly, there is significant enhancement in conductivity with the evolution and

formation of the crystalline structure. As illustrated in Figure 7-3, increasing the PLD

deposition temperature made it possible to move from a polycrystalline microstructure at

lower temperatures to a single crystalline epitaxial BYZ layer at 900oC. This implies

decreasing grain boundary density due to growing grain size as the deposition

temperature is increased, where finally the epitaxial film forms the single crystalline

161

structure. Therefore, the measured conductivity values at 900oC are indeed the same as

intra-grain reference values. This again verifies the hypothesis that grain separation and

grain boundaries in doped BaZrO3 films significantly impede ionic transport resulting in

high ionic resistance.

Hindrance to proton transport across BYZ grain boundaries is expected to result in a

higher activation energy than intra-grain transport. Indeed, the activation energy for the

grain boundary transport in BYZ is reported to be 0.71eV [25]. To explain the

significantly lower activation energies ranging from 0.45eV to 0.53eV depending the on

film deposition temperature, we speculate that even the polycrystalline films may possess

an epitaxial BYZ interlayer at the BYZ/MgO interface driven by the guiding MgO

structure. Given the cell geometry employed for conductivity measurements (see Figure

7-1), we hypothesize that this epitaxial interlayer may provide an alternative transport

pathway connected in parallel arrangement with the highly resistive pathway through

grain boundaries of the polycrystalline microstructure at the external surface region of the

BYZ film.

162

Fig 7-10. Variations in the SAD patterns obtained from a cross-section sample of MgO

(100)/BYZ film (deposited by PLD at 700°C) as the SAD aperture position is moved

from MgO into the BYZ film: SAD aperture positioned on MgO only (a), 20nm into

BYZ from MgO interface (b), 40nm into BYZ from MgO interface (c), and 100nm across

the entire BYZ (d). Digitally 4X magnified SAD patterns of 101 spots from Figure 7-10-

b, c & d are shown in Figure 7-10-e, and indicate how the epitaxy in the BYZ film near

the MgO substrate gradually changes to increasing polycrystallinity with wider

divergence in orientation as the aperture moves towards upper regions of the BYZ film.

The ranges selected for SAD patterns are indicated on the cross-sectional bright field

TEM image of Figure 7-10-f using a color scheme (white for a, blue for b, green for c,

and orange for d).

163

In order to test and verify this hypothesis, we have employed SAD analysis using TEM.

Figure 7-10 shows the cross-sectional image and corresponding SAD patterns of the BYZ

film deposited at 700oC on MgO(100). When the SAD aperture, which defines the

specific area for diffraction, is gradually moved from the MgO substrate through the

thickness of the BYZ film, the diffraction spots in Figure 7-10-b, 7-10-c & 7-10-d clearly

show increasing misalignment in the BYZ film deviating from the perfectly aligned

epitaxial diffraction array. As the figure shows, under the deposition condition BYZ

grows epitaxially up to ~20nm. Beyond that thickness, it seems like second nucleation

due to misorientation starts and forms hetero epitaxial BYZ layer. Especially the SAD

pattern of Figure 7-10-d, which was taken from the upper region of BYZ indicated in

Figure 7-10-f, clearly shows more randomly oriented diffraction spots. This trend is

confirmed in Figure 7-10-e, which shows a monotonic increase in the divergence angle in

digitally magnified 101 spots as larger fractions of the BYZ film microstructure are

utilized for SAD analysis.

Clearly, SAD analysis confirms that the epitaxial nature of the BYZ film immediately

next to the MgO surface changes gradually to a polycrystalline microstructure away from

this interface, as evidenced by increasingly misaligned orientation as the BYZ thickness

is traversed. These results verify the validity of our initial hypothesis that the epitaxial

BYZ interlayer shown in Figure 7-10-f most likely provides an alternative and parallel

transport pathway for proton transport that results in an activation energy of 0.51-0.53eV,

164

which is similar to that of epitaxial BYZ film, but slightly higher possibly due to

polycrystalline nature of the upper region of the BYZ film. However, it was not possible

to discern and estimate the individual or relative contributions to proton transport from

these two structurally different BYZ regions. The polycrystalline outlaying region of the

BYZ films that were deposited at lower temperatures naturally presents cross grain

boundaries to proton transport, thus significantly lowering the conductivity values while

only slightly increasing the activation energy. We believe the proposed parallel pathway

for proton transport in these thin BYZ films helps explain the low proton conductivity

values concurrent with activation energy comparable to that for single crystal BYZ.

In case of BYZ films grown on quartz substrates, a similar but a more gradual trend was

observed as shown in Figure 7-11. Although there is some inherent error in extracting

accurate values for degree of crystallinity from the XRD data, the figure nevertheless

indicates semi quantitatively that proton conductivity increases monotonically as the

BYZ film forms an increasingly more crystalline structure (see Figure 7-2 also). As

expected, the conductivity values for the amorphous and polycrystalline BYZ films

grown on amorphous quartz substrates are about four orders of magnitude lower than the

bulk values. Also, the activation energy is about 1.12eV, which is close to the reference

value of non-epitaxially grown BYZ [2]. We speculate that this high ionic resistance is

due to the formation of nano size grains and significant grain separation in films grown

on amorphous structures, which fail to provide a guiding structure for crystal habit

formation of the deposited film [2]. Nevertheless, both Figures 7-9 and 7-11 clearly

165

indicate an enhancement in conductivity with the formation of the BYZ crystalline

structure.

Figure 7-11. Plot showing the conductivity versus degree of crystallinity of BYZ-quartz

samples with three different deposition temperatures. Error bars are included for one

measured temperature data since all three samples have the same error bars. The plot

indicates that as the deposition temperature increases the degree of crystallinity and the

conductivity increases.

7.4 Conclusion

In summary, the relation between the evolution and formation of crystalline structure and

the resulting ionic conductivity in ultra thin epitaxial and polycrystalline BYZ films was

investigated. BYZ films were grown by PLD on MgO(100) and quartz substrates at

166

different deposition temperatures. XRD patterns showed that different deposition

temperatures lead to structures ranging from amorphous to polycrystalline to epitaxial

single crystal PLD BYZ films. TEM cross-section images and SAD patterns confirmed

epitaxial BYZ films deposited on MgO(100) substrate at 900oC. At lower deposition

temperatures, SAD analysis confirmed the formation of a BYZ epitaxial interlayer at the

interface due to the guiding structure of the MgO substrate, which gradually changes into

a polycrystalline microstructure towards the top surface. Experimentally obtained

conductivity values and the extracted activation energies were in good agreement with

reference values from the literature for both bulk and non-epitaxial BYZ films. The

results showed a clear trend of higher conductivity with increased crystallinity and less

grain boundary and grain separation. Therefore, the obtained results provide design

implications when using BYZ as an electrolyte material for ceramic fuel cell. It would be

beneficial to fabricate larger grain size or less grain boundary density electrolyte to

enhance the proton conductivity thorough the membrane.

7.5 References

[1] W. Münch, G. Seifert, K. D. Kreuer, J. Maier, Solid State Ionics, 97, 39-44 (1997)

[2] K. D. Kreuer, Ann. Rev. Mat. Res., 33, 333-359 (2003)

[3] W. Münch, K. D. Kreuer, S. Adams, G. Seifert, J. Maier, Phase Transitions, 68,

567-586 (1998)

167

[4] R. C. T. Slade, N. Singh, Solid State Ionics, Diffusion & Reactions, 46, 111-15 (1991)

[5] N. Bonanos, B. Ellis, M. N. Mahmood, Solid State Ionics, 44, 305-11 (1991)

[6] N. Kuwata, N. Sata, T. Tsurui, H. Yugami, Jpn. J. Appl. Phys., 44, 8613-8618 (2005)

[7] T. Hibino, A. Hashimoto, M. Suzuki, M. Sano, J. Electrochem. Soc., 149, A1503-8

(2002)

[8] H. Iwahara, H. Uchida, K. Morimoto, J. Electrochem. Soc., 137, 462-465 (1990)

[9] H. Iwahara, H. Uchida, K. Ono, K. Ogaki, J. Electrochem. Soc., 135, 529-533 (1988)

[10] H. Iwahara, T. Yajima, T. Hibino, K. Ozaki, H. Suzuki, Solid State Ionics, 61, 1-3

(1993)

[11] N. Ito, M. Iijima, K. Kimura, S. Iguchi, J. Power Sources, 152, 200 (2005)

[12] U. Balachandran, T. H. Lee, B. Ma, S. E. Dorris, Mater. Res. Soc. Symp. Proc., 972,

AA01-09 (2007)

[13] K. Katahira, Y. Kohchi, T. Chimura, H. Iwahara, Solid State Ionics, 138, 91-98

(2000)

[14] G. Ma, T. Chimura, H. Iwahara, Solid State Ionics, 110, 103-110 (1998)

[15] D. Shima, S. M. Haile, Solid State Ionics, 97, 443-455 (1997)

[16] N. Zakowsky, S Williamson, S. T. S. Irvine, Solid State Ionics, 176, 3019-3026

(2005)

[17] S. V. Bhide, A. V. Virkar, J. Electrochem. Soc., 146, 2038-2044 (1999)

[18] F. Chen, O. T. Sørensen, G. Meng, D. Peng, J. Mater. Chem., 7, 481-485 (1997)

[19] B. Merinov, W. J. Goddard III, Chem. Phys., 130, 194707 (2009)

168

[20] W. Münch, G. Seifert, K. D. Kreuer, J. Maier, Solid State Ionics, 86-88, 647-652

(1996)

[21] M. Pioinke, T. Mono, W. Schweika, T. Springer, T. Schober, Solid State Ionics, 97,

497-504 (1997)

[22] K. D. Kreuer, Solid State Ionics, 125, 285-302 (1999)

[23] S. M. Haile, G. Staneff, K. H. Ryu, J. Mater. Sci., 36, 5 (2001)

[24] P. Babilo, S. M. Haile, J. Am. Ceram. Soc., 88, 9 (2005)

[25] P. Babilo, T. Uda, S. M. Haile, J. Mater. Res., 22, 5 (2007)

[26] E. Fabbri, A. D’Epifanio, E. Di Bartolomeo, S. Licoccia, E. Traversa, Solid State

Ionics, 179, 558-564 (2008)

[27] A. D’Epifanio, E. Fabbri, E. Di Bartolomeo, S. Licoccia, E. Traversa, Fuel Cells, 8,

69-76 (2008)

[28] J. H. Shim, T. M. Gür, F. B. Prinz, Appl. Phys. Lett., 92, 253115 (2008)

[29] X. Guo, S. Mi, R. Waser, Electrochem Solid State Lett., 8, J1 (2005)

[30] H. G. Bohn, T. Schober, J. Am. Ceram. Soc., 83, 4 (2000)

[31] D. Pergolesi, E. Fabbri, A. D’Epifanio, E. Di Bartolomeo, A. Tebano, S. Sanna, S.

Licoccia, G. Balestrino, and E. Traversa, Nature Mater., 9, 846 (2010)