Electrochemically Deposited Ceria Struct

Electrochemically Deposited Ceria Structures for Advanced Solid Oxide Fuel Cells - CaltechTHESIS
CaltechTHESIS
A Caltech Library Service
About
Browse
Deposit an Item
Instructions for Students
Electrochemically Deposited Ceria Structures for Advanced Solid Oxide Fuel Cells
Citation
Brown, Evan Cornell
(2011)
Electrochemically Deposited Ceria Structures for Advanced Solid Oxide Fuel Cells.
Dissertation (Ph.D.), California Institute of Technology.
doi:10.7907/93NE-TG09.
Abstract
As the pursuit towards emissions reduction intensifies with growing interest and nascent technologies, solid oxide fuel cells (SOFCs) remain an illustrious candidate for achieving our goals. Despite myriad advantages, SOFCs are still too costly for widespread deployment, even as unprecedented materials developments have recently emerged. This suggests that, in addition to informed materials selection, the necessary power output—and, thereby, cost-savings—gains must come from the fuel cell architecture. The work presented in this manuscript primarily investigates cathodic electrochemical deposition (CELD) as a scalable micro-/nanoscale fabrication tool for engineering ceria-based components in a SOFC assembly. Also, polymer sphere lithography was utilized to deposit fully connected, yet fully porous anti-dot metal films on yttira-stabilized zirconia (YSZ) with specific and knowable geometries, useful for mechanistic studies. Particular attention was given to anode structures, for which anti-dot metal films on YSZ served as composite substrates for subsequent CELD of doped ceria. By tuning the applied potential, a wide range of microstructures from high surface area coatings to planar, thin films was possible. In addition, definitive deposition was shown to occur on the electronically insulating YSZ surfaces, producing quality YSZ|ceria interfaces. These CELD ceria deposits exhibited promising electrochemical activity, as probed by A.C. Impedance Spectroscopy. In an effort to extend its usefulness as a SOFC fabrication tool, the CELD of ceria directly onto common SOFC cathode materials without a metallic phase was developed, as well as templated deposition schemes producing ceria nanowires and inverse opals.
Item Type:
Thesis (Dissertation (Ph.D.))
Subject Keywords:
Fuel Cells, Electrochemical Deposition, Thin Films, Ceria
Degree Grantor:
California Institute of Technology
Division:
Engineering and Applied Science
Major Option:
Materials Science
Thesis Availability:
Public (worldwide access)
Research Advisor(s):
Haile, Sossina M.
Thesis Committee:
Haile, Sossina M. (chair)
Greer, Julia R.
Goddard, William A., III
Rossman, George Robert
Defense Date:
20 May 2011
Record Number:
CaltechTHESIS:05262011-123505549
Persistent URL:
DOI:
10.7907/93NE-TG09
Default Usage Policy:
No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:
6451
Collection:
CaltechTHESIS
Deposited By:
Evan Brown
Deposited On:
27 May 2011 20:38
Last Modified:
09 Oct 2019 17:10
Thesis Files
Preview
PDF (full thesis)
- Final Version
See Usage Policy.
33MB
Preview
PDF (Chapter 1/Intro)
- Final Version
See Usage Policy.
558kB
Preview
PDF (Chapter 2)
- Final Version
See Usage Policy.
5MB
Preview
PDF (Chapter 3)
- Final Version
See Usage Policy.
9MB
Preview
PDF (Chapter 4)
- Final Version
See Usage Policy.
5MB
Preview
PDF (Chapter 5)
- Final Version
See Usage Policy.
5MB
Preview
PDF (Chapter 6/Conclusion/Refs)
- Final Version
See Usage Policy.
7MB
Repository Staff Only:
item control page
CaltechTHESIS is powered by
EPrints 3.3
which is developed by the
School of Electronics and Computer Science
at the University of Southampton.
More information and software credits
Electrochemically Deposited Ceria Structures for
Advanced Solid Oxide Fuel Cells

Thesis by

Evan C. Brown

In Partial Fulfillment of the Requirements
For the Degree of
Doctor of Philosophy

California Institute of Technology
Pasadena, California

2011
(Defended May 20, 2011)

Evan C. Brown

iii

Acknowledgements
Foremost, I would like to dedicate this work to the One who loved me first. Truly,
there is no greater love than one who lays down his life for another.
I would like to thank my wife for giving me such support and encouragement.
Those words are easily written, but they were never given lightly by her. As for the rest
of my family, I would not have accomplished much without all of their love and their
crazy; especially my parents, who inspire me and have always reminded me who I am.
I also want to thank Sossina Haile for being my advisor and my advocate—
without her help, this manuscript would be 100 times longer with half as much insight.
She has shown me how compassion and strength can blend together in true leadership.
And I want to thank my labmates and science friends for the community and fellowship
we have shared. Particularly, I want to thank Dr. William Chueh for too many things to
count (but specifically for analysis, measurements, and insights pertaining to Chapter 4),
Dr. Yong Hao (Chapter 4, microfabrication), Dr. David Boyd for his analytical expertise,
and Stephen Wilke (Chapter 2, fabrication and SEM assistance). Also, to Taesik Oh,
Aron Varga, Mary Louie, Carol Garland (TEM), Jane Liu (FIB), Aminy Ostfeld, and
Sylvia Sullivan for invaluable discussions and technical assistance. I would also like to
acknowledge the Global Climate and Energy Program in conjunction with Stanford
University for funding.

Abstract
As the pursuit towards emissions reduction intensifies with growing interest and nascent
technologies, solid oxide fuel cells (SOFCs) remain an illustrious candidate for achieving
our goals. Despite myriad advantages, SOFCs are still too costly for widespread
deployment, even as unprecedented materials developments have recently emerged. This
suggests that, in addition to informed materials selection, the necessary power output—
and, thereby, cost-savings—gains must come from the fuel cell architecture. The work
presented in this manuscript primarily investigates cathodic electrochemical deposition
(CELD) as a scalable micro-/nanoscale fabrication tool for engineering ceria-based
components in a SOFC assembly. Also, polymer sphere lithography was utilized to
deposit fully connected, yet fully porous anti-dot metal films on yttira-stabilized zirconia
(YSZ) with specific and knowable geometries, useful for mechanistic studies. Particular
attention was given to anode structures, for which anti-dot metal films on YSZ served as
composite substrates for subsequent CELD of doped ceria. By tuning the applied
potential, a wide range of microstructures from high surface area coatings to planar, thin
films was possible. In addition, definitive deposition was shown to occur on the
electronically insulating YSZ surfaces, producing quality YSZ|ceria interfaces. These
CELD ceria deposits exhibited promising electrochemical activity, as probed by A.C.
Impedance Spectroscopy. In an effort to extend its usefulness as a SOFC fabrication tool,
the CELD of ceria directly onto common SOFC cathode materials without a metallic
phase was developed, as well as templated deposition schemes producing ceria nanowires
and inverse opals.

Table of Contents
Acknowledgements………………………………..……………………………… iii
Abstract…………………………………………………………………………… iv
Table of Contents…………………………………………………………………. v
List of Figures…………………………………………………………………….. x
List of Tables……………………………………………………………………… xvi
List of Symbols and Notations…………………………………………………… xviii

1 Introduction and Background…………………………………………………. 1
1.1 A Global Perspective…………………………………………………………... 1
1.2 SOFC Introduction…………………………………………………………….. 3
1.2.1 SOFC Basics……………………………………………………………. 3
1.2.2 Materials Selection: Samaria-Doped Ceria (SDC) ……………………... 7
1.2.3 Cell Architecture………………………………………………………... 10
1.3 Anti-Dot Substrates: A New Design Framework……………………………… 12
1.4 Three-Dimensional Structures and Their Fabrication by CELD………………. 14
1.4.1 SOFC Fabrication Method/Morphology Non-Negotiables…………..…. 14
1.4.2 Cathodic Electrochemical Deposition (CELD)…………………………. 16
2 Anti-Dot Substrates…………………………………………………………….. 19

2.1 Polymer Sphere Lithography Background and Summary……………………... 19
2.2 Experimental Details……………………………………………………………23
2.2.1 Substrate Preparation……………………………………………………. 23
2.2.2 Nanosphere Deposition…………………………………………………. 23
2.2.3 Microsphere Deposition………………………………………………… 23
2.2.4 Metal Deposition………………………………………………………... 24
2.2.5 Microstructure Analysis………………………………………………… 24
2.2.6 High Temperature Stability……………………………………………... 25
2.3 Results and Discussion………………………………………………………… 25
2.3.1 Nanosphere Lithography Results……………………………………….. 25
2.3.2 Microsphere Lithography Results………………………………………. 26
2.3.3 Microstructure Fidelity………………………………………………….. 31
2.3.4 Thermal Stability………………………………………………………... 36
3 Cathodic Electrochemical Deposition of Undoped and Doped Ceria……….. 38
3.1 Introduction………………………………………………………….………… 38
3.2 Experimental Details……………………………………………………………40
3.2.1 Substrate Definition……………………………………………………... 40
3.2.2 Experimental Setup……………………………………………………... 42
3.2.3 Characterization Details………………………………………………… 45
3.3 Results…………………………………………………………………………. 46
3.3.1 Bulk……………………………………………………………………... 46
3.3.2 High Surface Area (HSA) Coatings…………………………………….. 52
3.3.3 Thin Films………………………………………………………………. 58

vii

3.4 Discussion……………………………………………………………………… 64
3.4.1 General Deposition Overview…………………………………………... 64
3.4.2 The Physical Deposition Picture………………………………………... 69
3.4.3 Deposition on Non-Conducting Parts of the Substrate…………………. 71
3.4.4 HSA and Thin Film Transients…………………………………………. 79
4 The Electrochemical Activity of CELD Ceria Structures……………………. 84
4.1 Introduction, Methods, and Background…………………………………......... 84
4.1.1 A.C. Impedance Spectroscopy (ACIS) Introduction……………………. 84
4.1.2 Experimental Approach…………………………………………………. 87
4.1.3 System Precedence……………………………………………………… 90
4.2 Arc Identification: PLD Films vs. CELD Coatings………………………......... 91
4.2.1 Representative Spectra…………………………………….…………… 91
4.2.2 Origin of the Single Arc in the Metal-Sandwich Configuration….……. 94
4.2.3 Origin of the HF Arc in Embedded Metal Configurations….………….. 97
4.2.4 Origin of the LF Arc in Embedded Metal Configurations..……………. 102
4.3 The SDC|Gas Interface Arc: A Closer Look……………………..………......... 104
4.3.1 Platinum Strips………………………………………………………….. 104
4.3.2 Nickel Anti-Dot Films…………………………………………………... 115
5 Sundry Specialized CELD Microstructures…………………………………... 121
5.1 Anodic Aluminum Oxide (AAO) Templated Nanowires……………………… 121
5.1.1 AAO Template Formation Mechanism and Background………………..121
5.1.2 AAO Fabrication Experimental Details………………………………… 124

viii

5.1.3 AAO Template Results…………………………………………………. 126
5.1.4 Ceria Nanowire Growth………………………………………………… 133
5.2 Inverse Opals……………………………………………………………………135
5.2.1 Inverse Opal Definition and Background……………………………….. 135
5.2.2 Inverse Opal Fabrication Details…………………………………………136
5.2.3 Inverse Opal Results…………………………………………………….. 136
5.3 Oxidation Protection Coatings…………………………………………………. 138
5.3.1 Experimental Details……………………………………………………. 138
5.3.2 Results…………………………………………………………………... 139
5.4 CELD Ceria Grown Directly on MIEC SOFC Cathode Substrates…………… 140
5.4.1 Substrate Preparation Details…………………………………………… 141
5.4.2 CELD Results and Discussion………………………………………….. 143
6 Summary and Conclusions…………………………………………………….. 150
Appendix A: ImageJ Analysis Details…………………………………………… 152
Appendix B: Additional Images……….………………………………………… 155
B.1

Additional CELD Images………………….………………………………. 155

B.2

Additional AAO Images…………………………………… ………………158

B.3

Additional Inverse Opal Images…………………………………………… 160

B.4

Additional MIEC Substrate Images……………………………………….. 160

B.5

Additional Oxidation Protection Coating Images…………………………. 161

Appendix C: Alternate SOFC Microstructure Fabrication Routes…………… 162

C.1

Solution Impregnation into AAO Templates……………………................. 162

C.2

Copper Nanowire Synthesis……………………………………………….. 165

References………………………………………………………………………… 169

List of Figures
1.1

SOFC schematic………………………………..………………………….. 4

1.2

SOFC polarization curve………..…………………………………………..6

1.3

Electrolyte materials’ ionic conductivity comparison…………..…………. 8

1.4

3PB and 2PB in powder-processed electrodes schematics……….…..……. 10

1.5

Anti-dot film and templated electrode microstructure schematics…..…….. 15

2.1

SEM images of the polymer sphere lithography process……………..…… 20

2.2

Representative anti-dot film SEM images………..………………….…….. 22

2.3

SEM images of nanosphere short- and long-range coverage……..………...26

2.4

SEM images of nanosphere coverage for 500, 680, and 790 nm spheres..... 27

2.5

SEM images of multilayers and void regions from microsphere
deposition...................................................................................................... 28

2.6

Optical photographs of multilayers and void regions…………..………….. 28

2.7

Optical photographs of the water wash method………..………………….. 30

2.8

Optical and SEM images of one spin coat with the water wash method…... 30

2.9

SEM images of microsphere coverage for 2 and 3.2µm spheres………..….31

2.10

Pore diameter histograms of Cu anti-dot films…………………………….. 33

2.11

2PB area fraction histograms of Cu anti-dot films………………………… 33

2.12

SEM and AFM images of Ni anti-dot films before and after thermal
treatment…………………………………………………………………… 37

3.1

Pourbaix diagram for the Ce-H2O-H2O2 system from ref [86]……………. 40

3.2

SEM images of porous metal network on YSZ substrates used for CELD... 41

3.3

CELD setup schematic with corresponding potential electrode values…..... 44

3.4

XRD and EDS analyses of undoped and doped bulk CELD ceria……..….. 47

3.5

Raman analyses of undoped and doped bulk CELD ceria…..……………... 48

3.6

FT-IR analyses of undoped and doped bulk CELD ceria…………..……… 51

3.7

TGA analyses of undoped and doped bulk CELD ceria………..………….. 51

3.8

SEM images of undoped HSA ceria on YSZ/Pt strips…..………………… 53

3.9

SEM images of doped HSA ceria on various substrates…..………………. 54

3.10

SEM images of HSA ceria deposited from the doped + H2O2 electrolyte.... 55

3.11

SEM images of the HSA microstructure’s thermal stability………………. 56

3.12

SEM images of a crack-free HSA structure after annealing……………….. 57

3.13

SEM images of as-deposited thin film morphologies……………………… 58

3.14

AFM scans of as-deposited thin films from the doped and doped + H2O2
electrolytes…………………………………………………………………. 59

3.15

SEM cross-sectional images of thin film morphologies………………….... 60

3.16

SEM images of the deposits from the doped + acetic electrolyte………….. 62

3.17

SEM images showing the thermal stability of thin films on YSZ/Pt strips... 63

3.18

CV scan for the doped and doped + H2O2 electroltyes…….………………. 68

3.19

SEM images of as-deposited HSA CELD ceria growth on exposed YSZ
surfaces…………………………………………………………………….. 73

3.20

SEM images of as-deposited planar CELD ceria growth on exposed YSZ
surfaces…………………………………………………………………….. 75

3.21

SEM images of equivalent HSA growth on YSZ and Pt strip regions….…. 76

3.22

SEM images of as-deposited planar growth on the 3PB region of a

xii

YSZ/Pt strips substrate……………………………………………………. 78
3.23

TEM and HRTEM images of HSA CELD ceria…………………………... 78

3.24

HSA voltage transients and corresponding chronological SEM images…... 80

3.25

Thin film current transients…………………………………………………82

4.1

Representative Nyquist plot for a PLD/Pt strips exposed configuration….. 85

4.2

Schematics showing 2PB reaction pathways for lithographically
defined substrates, metal-embedded and metal-sandwich configurations for
PLD and CELD samples for ACIS studies……….………………………... 89

4.3

SEM images of PLD and CELD metal-embedded samples…..…………… 90

4.4

Representative Nyquist plots for PLD and CELD metal-embedded
samples………………………………………………………………….…. 92

4.5

Nyquist plots with hydrogen and water partial pressure dependencies for a
representative CELD/Ni anti-dot-embedded sample………………………. 93

4.6

Representative Nyquist plots for metal-sandwich configurations………… 94

4.7

Hydrogen partial pressure dependence of the single arc from metal-exposed
and metal-sandwich configurations…………..………………..…..………. 95

4.8

SEM images of the deleterious phenomena associated with CELD/metalsandwich samples…………………………………………………………...96

4.9

Hydrogen partial pressure dependence of the HF arc from metal-embedded
configurations……………………………………………………………… 98

4.10

Hydrogen and water partial pressure dependence, as well as 3PB and metal
spacing dependencies, of the HF arc for large pattern sized PLD samples…99

xiii

4.11

TEM images showing voids in the HSA CELD deposit near the exposed
metal surfaces…………………………………………………………….. 101

4.12

Hydrogen partial pressure dependence of the LF arc from metal-embedded
and the single arc from metal-exposed configurations……………………. 103

4.13

Pt pattern size effect on the SDC|gas interfacial ASR partial pressure
dependencies………………………………………………………………. 106

4.14

SEM images of undoped CELD ceria on 5-5 µm and 20-20 µm Pt patterns
on YSZ…………..……………………………………………………….... 106

4.15

Undoped CELD deposition time effect on the SDC|gas interfacial ASR
partial pressure dependencies……………..………………………………. 108

4.16

SEM images of undoped CELD ceria samples deposited for 5 and 10
minutes…………………………………………………………………….. 108

4.17

Doped CELD deposition time effect on the SDC|gas interfacial ASR partial
pressure dependencies……………….……………………….……………. 109

4.18

SEM images of doped CELD ceria samples deposited for 5, 10, and 20
minutes……………………………………………………………………... 109

4.19

Doping effect for 5 minute deposits on the SDC|gas interfacial ASR partial
pressure dependencies……………..………………………………...…….. 110

4.20

SEM images of doped and undoped CELD ceria samples deposited for 5
minutes……………………………………………………………………... 111

4.21

Doping effect for 10 minute deposits on the SDC|gas interfacial ASR partial
pressure dependencies….…………………………………………………. 111

xiv

4.22

SEM images of doped and undoped CELD ceria samples deposited for 10
minutes……………………………………………………………………... 111

4.23

Consecutive depositions effect on the SDC|gas interfacial ASR partial
pressure dependencies……………………………….……………………. 113

4.24

SEM images of consecutive depositions following thermal treatment…...... 114

4.25

SEM image comparison of doped CELD/Ni anti-dot-embedded samples
deposited for 5, 10, and 20 minutes……………………………………….. 117

4.26

Deposition time effect for doped CELD/Ni anti-dot-embedded samples on
the SDC|gas interfacial ASR partial pressure dependencies……………….. 118

4.27

SEM image comparison of two doped CELD HSA samples and one doped
CELD planar sample………………………………………………………. 119

4.28

SDC|gas interfacial ASR partial pressure dependencies comparison
between two HSA and one planar doped CELD samples…………………. 120

5.1

AAO structure schematic………..…………………………………………. 122

5.2

SEM images of AAO templates grown from Al foil……..………………... 127

5.3

SEM image comparison of phosphoric acid pore diameter etching times.....128

5.4

SEM images of AAO templates grown from sputtered Al thin films……....129

5.5

Current transients for AAO templates grown from various Al thin film
samples, whose optical photographs are also shown……………………… 130

5.6

SEM images of the barrier layer from AAO templates grown from Al foil

and sputtered Al thin films………………………………………………... 132
5.7

SEM images of as-deposited CELD ceria nanowires in the pores of AAO.. 134

5.8

SEM images of AAO ceria nanowires after thermal treatments…..………..134

5.9

SEM images of ceria inverse opal structures on YSZ/Pt strips and Ni anti-dot
substrates grown via CELD………………………………………………... 137

5.10

SEM images of difficulties encountered during the inverse opal fabrication
process……………………………………………………………………... 138

5.11

SEM images of the oxidative protection coating activity of CELD ceria
coatings on Ni anti-dot films……...………………………………………. 140

5.12

SEM images of the depositing surface of porous BSCF substrates that has been
planarized via abrasive paper………………………………………………. 143

5.13

SEM images of as-deposited undoped CELD ceria grown on dense BSCF..144

5.14

SEM images of as-deposited doped CELD ceria grown on porous BSCF… 145

5.15

CV scan comparison between Ni and BSCF substrates for the doped
electrolyte………………………………………………………………….. 145

5.16

SEM images of thin films of CELD ceria grown on porous BSCF at
non-standard and open working potentials.....……………………………. 147

5.17

SEM images of various CELD ceria structures deposited near the meniscus
area of a dense BSCF sample……………………………………………... 149

A.1

The ImageJ analysis process……………...………………………………... 154

B.1

Additional CELD HSA SEM images…...…………………………………. 155

xvi

B.2

Additional CELD thin film cracking SEM images………………………… 156

B.3

Additional CELD TEM images……………………………………………. 157

B.4

Additional optical and SEM images of AAO templates…………………… 158

B.5

Additional CELD ceria nanowires SEM images…………………………... 159

B.6

Additional CELD inverse opal SEM images….…………………………… 160

B.7

Additional CELD on MIEC substrate SEM images……………………….. 160

B.8

Additional oxidation protection coating SEM images…………………….. 161

C.1

SEM images of unaided solution phase impregnated ceria nanowires…….. 163

C.2

SEM images of sonicated-assisted impregnated ceria nanowires…………. 163

C.3

SEM images of stirring-assisted impregnated ceria nanowires……………. 163

C.4

SEM images of sonicated- and stirring-assisted impregnated ceria
nanowires…………………………………………………………………... 164

C.5

SEM image of an un-etchable AAO template after thermal treatment……. 164

C.6

SEM images of CuO nanowires thermally grown from Cu foil…………… 166

C.7

SEM images of CuO nanowires thermally grown from thin films of Cu on
SDC and porous Cu films after harsh hydrogen plasma treatment………… 167

C.8

SEM images of Cu nanowires resulting from reduction via a hydrogen
plasma at moderate power densities……………………………………….. 168

List of Tables
2.1

Comparison of theoretical and experimental 3PB length areal density and
percent 2PB exposure for different initial PS sphere diameters…………… 34

xvii

3.1

CELD liquid electrolyte compositions…………………………………….. 43

xviii

List of Symbols and Notations

number of electrons

Faraday’s constant

Nernstian voltage

oxygen non-stoichiometry

Ce3+/4+

dissociated aqueous cerium ions of a particular cerium valence

Ce(III/IV)

precipitated/solid cerium species of a particular cerium valence

𝑡ℎ𝑒𝑜
𝜌3𝑃𝐵

theoretical 3PB areal density

𝜌3𝑃𝐵

𝑒𝑥𝑝

experimental 3PB areal density

𝑡ℎ𝑒𝑜
𝑓2𝑃𝐵

theoretical 2PB area fraction

𝑓2𝑃𝐵

𝑒𝑥𝑝

experimental 2PB area fraction

𝜙𝑖

initial PS sphere diameter

𝜙𝑓

final PS sphere diameter

crystallite size

XRD x-ray wavelength

adjusted full-width half max

XRD diffracting angle

complex impedance

resistance

capacitance

frequency

√−1

xix

𝑍�

complex impedance normalized by total deposited area

𝑍� ∗

complex impedance normalized by the projected area of the exposed SDC

𝑅�∗

resistance associated with a Nyquist arc normalized by the projected area

𝑅�

resistance associated with a Nyquist arc normalized by total deposited area

surface

of the exposed SDC surface

Abbreviations
SOFC

solid oxide fuel cell

CELD

cathodic electrochemical deposition

YSZ

yttria-stabilized zirconia

OCV

open circuit voltage

SDC

samaria-doped ceria

GDC

gadolinia-doped ceria

MIEC

mixed ionic-electronic conductor

3PB

three-phase boundary

2PB

two-phase boundary

PLD

pulsed-laser deposition

CVD

chemical vapor deposition

AELD

anodic electrochemical deposition

polystyrene

SEM

scanning electron microscopy

AFM

atomic-force microscopy

HSA

high surface area

SCE

standard calomel electrode

XRD

x-ray diffraction

FT-IR

Fourier transform infrared

cyclic voltammetry

EDS

x-ray energy dispersive spectroscopy

TEM

transmission electron microscopy

ACIS

A.C. impedance spectroscopy

ASR

area-specific resistance

low frequency

high frequency

AAO

anodic aluminum oxide

BSCF

Ba0.5Sr0.5Co0.8Fe0.2O3-δ

SCN

SrxCoyNbzO3-δ

Chapter 1

Introduction and Background
1.1

A Global Perspective

Eventually, the world will run out of fossil fuels, period. This simple fact necessarily
motivates an intensive search for alternatives. As if to underscore the immediacy of such
a quest, geopolitical tensions and complications have again and again proven to disrupt
what people love most about fossil fuels—they are consistently available, relatively easy
to use, and, above all else, cost little to do so. Finding a (host of) suitable replacement
candidate(s) is difficult, owing to the plethora of pros to using fossil fuels. Indeed,
societies worldwide have in many cases developed around their day-to-day use, making
widespread adoption of anything else a nearly overwhelming task: humans are loathe to
radically change. Nevertheless, the pioneer views this picture as ripe with opportunity,
and science has historically cast itself as a trail blazer of progress.
There is a finite amount of energy that is available for power generation, in any
form. And since thermodynamics dictates that energy cannot be created or destroyed, we
are limited to options such as solar, wind, nuclear, hydroelectric, tidal, biomass, and
geothermal forms of energy. Of these, solar energy is far and away the most abundant,
and, therefore, the most practical to develop. Even as all of the so-called “renewable
energy” technologies are considered, two of the most attractive, solar and wind, suffer
from intermittency issues— the sun only shines during the day, and inclement weather
can be prohibitive; wind is notoriously temperamental. Energy storage media are
necessary to complement a system that relies solely on these renewable energy

technologies for power generation. Energy that is converted from solar or wind could be
used at a later time, for instance, when the electricity demand exceeds the supply ability,
like at night or when the wind isn’t blowing. Chemical bonds remain the most efficient
energy storage method, although significant gains have been made in batteries and
supercapacitors [1-4]. But once a fuel is made, there is the question of how one extracts
the stored energy. Humans have almost entirely relied upon combustion of fossil fuels to
do so, but the by-products invariably add to the growing amount of greenhouse gases in
the earth’s atmosphere. With the daunting prospect of global climate change, a better fuel
(and way of extracting its stored energy) is desperately needed.
Fuel cells have tremendous promise to address these concerns. A fuel cell is an
energy conversion device that relies upon electrochemical driving forces to extract
energy from a fuel as electricity, rather than the familiar, but Carnot-restricted
combustion cycles. This allows more of the chemical potential in a fuel to be converted
into useful work, with calculated efficiencies in excess of 80% for combined heat and
power systems [5]. Fuel cells operating at higher temperatures can run off of a wide range
of fuels, from standard, already-in-use fossil fuels to pure hydrogen. This flexibility is a
pragmatic necessity for bridging the current addiction to greenhouse-gas-producing fuel
to a “clean”, carbon-free source. A number of future scenarios can be imagined, but a
particularly compelling vision for the power generation of the future is to utilize solar
energy to split water into hydrogen and oxygen, where the hydrogen is stored until power
is needed. The hydrogen could then be utilized as the fuel in a fuel cell, producing
electricity. The by-product of such a process is water, which can be fed back to the
original input stream.

Challenges undoubtedly remain. Chief among those are economic—fuel cells are
~10-100 times too expensive to be competitive [5-6]. To ameliorate this issue, better
performing and cheaper materials/fabrication processes need to be developed.
This manuscript concentrates on combining modern, high-performance materials
with advanced architectural designs of solid oxide fuel cells (SOFCs), all to achieve the
ultimate goal of dramatically increasing their power output. Two fairly well-established
fabrication methods with little to no prior demonstration of actual application in a fuel
cell are utilized here for SOFCs, namely, polymer sphere lithography [7-8] for substrate
preparation and cathodic electrochemical deposition [9-10] for oxide material deposition.
Extensive modifications and further development was needed to appropriately adapt
them, which are the subjects of Chapters 2 and 3. Chapter 4 details activity analyses of
various SOFC components made with these fabrication methods, and Chapter 5 involves
the fabrication of specialized microstructures. First, however, a broad introduction to
SOFC operational basics is presented in Section 1.2, and the necessary linkage of,
applicability towards, and motivation for utilizing polymer sphere lithography and
cathodic electrochemical deposition in SOFC fabrication is subsequently established in
sections 1.3 and 1.4, respectively.

1.2

SOFC Introduction

1.2.1

SOFC Basics

A fuel cell consists of three main components: an electrolyte sandwiched between two
electrodes, the anode and cathode. The electrolyte is an ionically conducting material,
allowing ions, but not electrons, to migrate through it. Fuel cells are typically categorized

→ 2e-

1/ O + 2e- → O22 2

Cathode

Anode

Electrolyte

← O2-

H2 + O2- → H2O + 2e-

← O2Fig. 1.1. A schematic of a generalized SOFC, showing each electrode’s half-reactions and the
migration directions of each mobile species.

by their mobile ionic species and temperature of operation. In this manuscript, solid oxide
fuel cells are the focus. They are solid-state devices (meaning no liquid electrolytes) and
typically conduct oxygen ions through metal oxide constituents, although some protonconducting SOFCs exist [11-12]. Each electrode is responsible for facilitating transport of
electrons, oxygen ions, and gaseous reactants to surface reaction sites, where the
appropriate half-cell reaction occurs. A schematic of a generalized SOFC is shown in Fig.
1.1. On the anode side, fuel is introduced, where it reacts with oxygen ions supplied from
the cathode that have migrated through the solid electrolyte, producing water vapor and
electrons, according to the half-reaction in Eqn. 1.1.
𝐻2 (𝑔) + 𝑂2− → 𝐻2 𝑂(𝑔) + 2𝑒 −

(1.1)

Driven by the need to maintain overall charge neutrality, the negatively charged

electrons travel through an externally connected circuit to the cathode, effectively
offsetting the dearth of negative charge left by migrating oxygen ions. These incoming

electrons then react with atmospheric oxygen, producing oxygen ions according to the
half-reaction:

𝑂2 (𝑔) + 2𝑒 − → 𝑂2−

(1.2)

The two electrode half-reactions combine to yield the overall reaction given in

Eqn 1.3, from which the ΔGrxn can be calculated and then converted to a Nernstian
voltage (Eqn. 1.4), measured as the open circuit potential (OCV), where n is the number
of participating electrons and F is Faraday’s constant. This is the potential at which no
net current is flowing through the cell. For the high temperatures of SOFCs and pure
oxygen/hydrogen atmospheres, typical OCVs are ~1.1 V.
𝐻2 (𝑔) + 12𝑂2 (𝑔) → 𝐻2 𝑂(𝑔)
𝐸𝑁 =

∆𝐺𝑟𝑥𝑛
𝑛𝐹

(1.3)
(1.4)

Various deleterious phenomena decrease the operating voltage from the

theoretical Nernstian value, as depicted in the polarization curve of Fig. 1.2. A cell’s
power output is defined as the operating voltage multiplied by the drawn current,
meaning that these processes lower SOFCs’ power output. At open circuit conditions,
leaks in the sealing that separate the anodic and cathodic compartments, as well as holes
in the solid electrolyte, can allow fuel cross-over, which immediately lowers the
operating voltage. Also, non-zero electronic conductivity in the solid electrolyte has the
same effect. Once current is drawn from the cell, three so-called overpotentials further
decrease the operating voltage. Activation overpotentials are related to the finite-rate
electrode reaction kinetics, and typically dominate the voltage losses. Ohmic
overpotentials originate from conductivity resistances encountered when charged species

Fig. 1.2. A visualization of the overpotential losses typically
encountered in SOFCs and the associated power density output of such a
cell.

migrate throughout the cell. Concentration overpotentials arise when not enough
reactants are supplied to the half-reaction sites, most often caused by mass transfer
limitations in the gas phase, but these effects are only seen at very high current densities
beyond practical operating conditions.
Species’ transport through the crystal structure of metal oxides is generally
thermally activated, and electrode kinetics are enhanced as temperature increases;
therefore, high temperatures are desirable as they increase conductivity and reaction
rates. Standard SOFC operating temperatures are anywhere from 700 – 1000 °C [5-6].
These high temperatures enforce strict requirements for component materials, even
making choice of the interconnect material, which conducts the electrons to and from the
respective electrodes, a complicated matter. In fact, the lack of cheap, viable options for

high temperature interconnects has largely motivated the move toward intermediate
operating temperatures, i.e., 500-650 °C. This is the point at which stainless steel and its
derivatives can resist prohibitive oxidation, and could therefore conceivably be used for
interconnects [13]. Furthermore, thermal cycling can lead to significant wear and tear due
to differences in thermal coefficients of expansion, although it is less severe at lower
temperatures.
Manufacturing scalability and its cost is a perpetual concern. Low-throughput,
expensive fabrication processes cannot be a part of the final solution, although they can
be useful toward more fundamental understanding. Similarly, catalytic materials can be
used to impact and define sluggish reaction kinetic pathways, but they often consist of
expensive, rare precious metals such as platinum or palladium [14]. Even though much
lower operating temperatures can be achieved, this strategy is not viable on a large scale.
With so many aspects to SOFC technology, a methodical approach is needed to
gain fundamental insights and elucidate the rate-limiting steps, eventually contributing to
an informed, optimized design. From the brief overview above, two design focal points
emerge—materials selection and cell architecture.

1.2.2

Materials Selection: Samaria-Doped Ceria (SDC)

Cerium(IV) oxide (or, ceria—CeO2-δ) has a cubic fluorite structure, capable of large
oxygen non-stoichiometry (δ) via oxygen vacancies. In the moderate oxygen partial
pressure atmospheres experienced by the SOFC electrolyte (known as the electrolytic
regime), the oxygen vacancy concentration in ceria is extrinsically pinned down by a

Temperature [ C]
800 700 600

400

300

Bi2O3

(Bi,Y)2O3
La
0.9 S
r0

-1

.1 Ga
0.8 M

0.8 G

-2
-3

2 )0

aZr Y
0.9 0.1 O

3-δ

.9 (S

3 )0

) 0.1
aO

) 0.1
O3
(Y
) 0.9
rO
(Z

-4

0.2 O
3-δ
0.2 O
1.9 B

(Zr
) 0.8
rO
(Z

Log(σ) [Ω-1cm-1]

500

-5

1.0

1.2

1.4

1.6

1.8

-1

1000/T [K ]
Fig. 1.3. A through-plane ionic conductivity comparison for
common electrolyte materials taken from [13].

strictly 3+ cation dopant, such as samarium (SDC) or gadolinium (GDC). A samarium
doping example is written here in Kröger-Vink notation:
𝑆𝑚2 𝑂3 + 2𝐶𝑒𝐶𝑒
+ 4𝑂𝑂𝑋 ↔ 2𝑆𝑚𝐶𝑒
+ 𝑉𝑂∙∙ + 3𝑂𝑂𝑋 + 2𝐶𝑒𝑂2

(1.5)

This induces significant ionic conductivity at intermediate temperatures, garnering much
interest for doped ceria as the SOFC electrolyte component [15-17]. Fig. 1.3 shows a
conductivity comparison between common SOFC electrolyte materials, including the
traditional favorite, yttria-stabilized zirconia (YSZ)[14]. A generally accepted benchmark
for electrolyte conductivity is ~0.01 S cm-1, above which a candidate is deemed suitable.
According to this metric, ceria-based electrolytes could potentially operate from 500 –
650 °C, without sacrificing performance, as would be the case with YSZ.

Additionally, under the high temperature reducing conditions typically seen in a
SOFC anode, intrinsic oxygen vacancies form spontaneously via the oxidation of lattice
oxygen, according to [1]:
𝑂𝑂𝑋 ↔ 12𝑂2 (𝑔) + 𝑉𝑂∙∙ + 2𝑒 ′

(1.6)

These vacancies are charge compensated by electrons, which subsequently cause the
cerium cations to change valence from nominally all 4+ to mixed 4+/3+. This gives rise
to a non-trivial electronic conductivity via polaron hopping, making ceria a so-called
mixed ionic-electronic conductor (MIEC). Although MIEC perovskite-type metal oxides
are commonly employed as cathodes [18-20], there are few that are stable under the
anode’s high temperature reducing conditions, and those that are have low conduction
and/or slow hydrogen electrooxidation kinetics [21-23].
Due to the lack of available MIECs, a traditional SOFC anode is typically
composed of a random, three-dimensional amalgamation of an electronically conducting
phase, e.g., nickel, an ionically conducting phase, e.g., YSZ, and a gas-permeable
“phase,” e.g., a network of pores [24-25]. The intersection of these three phases is termed
the three phase boundary (3PB), shown schematically in Fig. 1.4a. The 3PB density (Fig.
1.4b) defines the number of reaction sites per projected electrolyte area, as the anode
half-reactions can only take place at this intersection. This is in stark contrast to a MIEC,
where electrochemical reactions can theoretically take place at nearly any point along its
exposed surface, or the two phase boundary (2PB), as in Fig. 1.4c. There has been a
significant effort to establish and quantify the anodic electrochemical activity of ceriabased 2PBs, even in the absence of a closely-adjacent, purely electronically conducting
phase [26-32]. Therein, it is definitively shown that the surface of doped ceria alone is,

(b)

(a)

H2
H2O

2e-

metal

YSZ
(ion conductor)

O2-

10 nm

1 µm

(d)
(c)

H2
H2O

metal

2e-

SDC
(ion and electron conductor)

O2-

10 nm

1 µm

Fig. 1.4. Schematic diagrams of (a) the three-phase boundary (3PB) region where gas, metal, and yttriastabilized zirconia (YSZ) phases intersect; (b) the 3PB density for a powder-processed anode; (c) the twophase boundary (2PB) region of a mixed ionic-electronic conductor like samaria-doped ceria (SDC); and
(d) the 2PB density for the same anode as in (b), but with SDC instead of YSZ. Light blue areas indicate
electrochemically active regions.

itself, electrocatalytically active, and its bulk electronic conductivity is sufficient to place
a current collector up to several microns away from a reaction site. Comparing the
visualized anodes in Fig. 1.4b and d, it can be seen that a simple materials switch from
YSZ to SDC affords a much greater reaction site density, owing to 2PB dominance over
3PB. In this way, materials selection paves the way for an architectural design paradigm,
one where 2PB microstructures, rather than more restrictive 3PB microstructures, are
possible.

1.2.3

Cell Architecture

Returning to the polarization curve of Fig. 1.2, there are three general design guidelines
related to the three overpotentials outlined above. First, to reduce ohmic losses, all
conduction pathways should be kept as short as possible. The primary culprit of ohmic

loss is oxygen ion transport through the solid electrolyte—the conclusion here is simple:
make the electrolyte layer thin (µm scale). Second, for a given electrode reaction rate,
maximizing the number of active reaction sites will increase the current density, on the
basis of the projected area of the cell. For a MIEC anode like SDC, this effectively
translates into maximizing the active surface area (nm scale). Third, one must ensure easy
gas phase access by highly porous, non-tortuous electrodes, although this is less of a
concern than the previous two (µm and perhaps nm scale).
The ideal cell design must balance µm and nm length scales, which also means
that new fabrication approaches must accommodate both. As SOFCs are high
temperature devices, care should be taken to ensure stability of any as-fabricated
nanometer-sized features. Despite the obvious need for feature size reduction, a general
hierarchical structure is desirable for aspects like electronic current collection—electrons
cannot be expected to only travel through nanometer-sized metal films or multiplemicron-lengths of SDC without incurring severe resistance penalties.
State-of-the-art powder processing methods that produce Ni/YSZ cermet anodes
(as in Fig. 1.4b) are cheap and scalable, but offer limited structural tunability and little
fundamental insight into the details of SOFC operation [25]. This is primarily due to the
randomized nature of the electrode geometry—key features like 3PB (or 2PB in the case
of a MIEC), pore size, conduction pathway lengths, and so on are all ill-defined. Even if
these parameters are determined post production (and probably using a destructive
method), the sample-to-sample variation is relatively high for randomized structures [33].
On the other hand, cell architectures with specifically engineered and well-defined
geometries offer dual advantages of physically-correlated diagnostic analyses and the

subsequent ability to alter the design in accordance with the results. For instance,
knowing the relationship between a SDC anode’s 2PB and its impedance spectra (c.f.
Chapter 4) could lend valuable insight into which design knob to turn, and how much.
In summary, most operational voltage loss mechanisms in today’s SOFCs inform
an overall feature size reduction of every component of the cell architecture. This move
should be done intelligently, so as not to incidentally incur other penalties, e.g., gas
diffusion limitations and bulk transport resistances, while at the same time maintaining
manufacturability, scalability, and the ability to produce large total footprint cells.
Furthermore, trending towards defined, as opposed to randomized, geometries can help
link performance to tunable features.
As such, there is tremendous need and potential for entirely new SOFC design
schemata, as well as complimentary fabrication techniques.

1.3

Anti-Dot Substrates: A New Design Framework

In recognition of the need to examine geometrically well-defined structures, some recent
mechanistic studies have employed two-dimensional electrodes patterned onto the
electrolyte of interest [18, 34-36]. This approach has begun to bear fruit and mechanistic
models have begun to be developed [37-38]; however, challenges in understanding ‘real’
fuel cells remain because the two-dimensional patterns have a substantially lower areal
density of 3PBs (defined as the 3PB length per unit of projected electrolyte area) than the
systems they represent. Specifically, conventional photolithographic techniques with a
minimum feature size of about 5 µm can attain a maximum areal 3PB density of 2,000
cm cm-2 [34]. In contrast, typical fuel cell electrodes boast values as high as 800,000 cm

cm-2 [33]. Such significant microstructural differences can plausibly induce differences in
reaction pathways. Accordingly, there is a pressing need to obtain geometrically defined
electrode structures with tunable feature sizes that are more relevant to SOFC
electrocatalysis.
Demonstrated below is a facile fabrication strategy, known as polymer sphere
lithography, in which monodisperse polymer spheres serve as sacrificial templates to
construct anti-dot metal films (see Fig. 1.5a), permitting access to 3PB areal densities
over an enormous range, from 2,000 to 43,500 cm cm-2. Though not previously explored
in the fuel cell context, the anti-dot structure is ideal for advancing the aforementioned
fundamental studies for this reason.
When these porous, metal films are overlaid onto an electrolyte substrate such as
YSZ or SDC, the fraction of exposed electrolyte area and the 3PB are concurrently and
specifically known, true for all two-dimensional lithographic processes. This enables
electrocatalysis studies for the underlying electrolyte material, particularly as it pertains
to 3PBs (and 2PBs for MIECs). The accessible 3PB regime here is previously untouched
by conventional lithography, moving much closer to actually-in-use 3PB densities. Use in
conjunction with diagnostic tools such as A.C. Impedance Spectroscopy (ACIS) allows
definitive relationships between 3PBs/2PBs and various electrochemical activity-related
materials characterization parameters to be established, e.g. rate limiting processes’
resistances, capacitances, etc. And when combined with traditional lithography
techniques, an extremely wide range of 3PBs can be sampled. Although tempting, such a
geometry as-is, however, is not actually a suitable electrode candidate because of issues
like high electronic resistance through the relatively thin anti-dot metal film.

An even higher number of reaction sites can be achieved by moving from a planar
to a three-dimensional structure, and these anti-dot films are a good starting point to get a
variety of well-defined three-dimensional electrode structures.
Chapters 2 and 5 present the fabrication of the anti-dot structure and its derivatives,
and Chapter 4 discusses the performance of its related SOFC electrodes.

1.4

Three-Dimensional Structures and Their Fabrication by CELD

Using the anti-dot structure as a starting point for the fabrication of high surface area
three-dimensional structures, several specific, more optimized architectures can be
considered, as in Fig. 1.5. Cathodic electrochemical deposition (CELD) is an ideal
candidate to produce template-free high surface area structures, as well as templated
frameworks like inverse opals (Fig. 1.5c and d) and nanowires/tubes (Fig. 1.5e). As a
testimony to their flexibility, anti-dot based substrates can also accommodate new and
old approaches such as screen printing [39], pulsed-laser deposition (PLD) [40], chemical
vapor deposition (CVD) [41], and CELD (c.f. Chapter 3).

1.4.1

SOFC Fabrication Method/Morphology Non-Negotiables

Up to this point, only general SOFC materials/architectural design guidelines have been
discussed, without reference to a particular method to produce such schemes. This section
is devoted to the assessment of new fabrication techniques and their associated asproduced morphologies, to aid in their development.
Before any new fabrication method/morphology is adopted for SOFCs, a few nonnegotiable requirements must be met. First, the fabrication method must be able to

(a)

(b)

1 µm

(c)

(d)

1 µm

SDC
(ion and electron conductor)

1 µm

(e)

SDC
(ion and electron conductor)

11µm
µm

Fig. 1.5. (a) A schematic of a metal anti-dot network; (b) a cross-sectional depiction of the anti-dot film in
(a) replacing metal powder as a current collector and thereby increasing the 2PB density; and examples of
potential templated electrodes with tunable geometries like inverse opals (c) and (d), and nanowires (e).

consistently produce the desired materials composition. Keeping large-scale
manufacturability in mind, basic repeatability is absolutely necessary. Second, the asdeposited morphology/microstructure cannot be adversely affected by SOFC operating
conditions, e.g., high temperatures, oxidizing/reducing atmospheres, etc. This includes,
for instance, cracking in electrolytes and agglomeration of small features in electrodes.
Third, continuous and accessible migration pathways to and from surface reaction sites
must exist in the electrodes. Of course, low resistance pathways are desirable, rather than
only connected ones.

In this manuscript, cathodic electrochemical deposition is evaluated as a
components fabrication tool for a SDC-based, intermediate temperature SOFC.

1.4.2

Cathodic Electrochemical Deposition (CELD)

CELD is a liquid-based, low temperature fabrication technique that is able to produce
ubiquitous and conformal metal oxide/hydroxide coatings of tunable surface area at low
capital and operational costs [42-43]. The experimental setup is straightforward (see Fig.
3.2): three electrodes are immersed in a liquid electrolyte—electrons flow out of the
anode and into the cathode through the external circuit, and the reference electrode
measures the cell potential but does not allow any current to flow through it. A working
potential is applied, and the appropriate electrochemical reactions occur.
Being liquid-based makes CELD scalable as a batch process, and allows easy
control of large substrates, even if irregularly shaped: appropriate operating
configurations ensure uniform deposition on protruding and porous substrates alike.
Furthermore, cation doping in liquid systems is simple [44-47], while the low operating
temperatures diminishes the incorporation of undesirable impurities. Low temperatures
and open, ambient conditions also reduce the experimental complexity, especially in
regard to otherwise stringent substrate requirements. Other common metal oxide
fabrication methods, such as CVD and PLD, typically involve in situ high temperatures
with a background atmosphere of oxygen—prime conditions for unwanted oxidation of
metallic substrate components, and risky due to the potential for impurity incorporation
into the oxide phase. CELD is also favorable as a manufacturing process as deposition
times are on the order of minutes, rather than hours or days. In addition to being explored

for general SOFC applications [48-52], CELD ceria has been previously studied for
corrosion protective coatings [46, 53-56], superconductor buffer layers [44], powder
synthesis for increased sinterability [57], and nanowire/tube fabrication [49, 58-59].
Aside from ceria, other SOFC-relevant materials have been produced using this method,
such as BaTiO3, Nb2O5, ZrO2, LaMnO3 [42], and Y2O3 [60].
There are two general categories of oxide/hydroxide electrochemical deposition,
defined by which electrode experiences the desired deposition, known as the working
electrode. Anodic electrochemical deposition (AELD) directly oxidizes Ce3+(aq) ions to
insoluble Ce(IV) [54, 61-62]. A stabilizing ligand must be added to the electrolyte
solution to ensure that Ce(III) species do not precipitate prematurely. A fundamental
limitation of this technique is that Ce3+ ions must contact a surface that can conduct
electrons away; as CeO2 is generally insulating, AELD should only be able to deposit
extremely thin films, on the order of tens of nanometers.
Cathodic electrochemical deposition, on the other hand, proceeds by a two-step
process. First, the electrolyte solution becomes progressively basic as electrochemical
reduction reactions of various electrolytic species occur due to the applied cathodic
potential at the cathode|electrolyte interface. This is widely referred to as
electrogeneration of base. Second, the newly-formed base induces chemical precipitation
of Ce(III/IV) species, e.g. Ce(OH)3 or hydrated CeO2, which are finally oxidized to the
desired fluorite CeO2 phase. It is somewhat surprising that a Ce(IV) deposit on the
cathode could result from a nominally Ce3+ electrolyte—this is a testament to the purely
chemical nature of the deposition step. Restated, the nucleation and growth process here
is non-Faradaic. One example reaction each from the electrogeneration of base and

precipitation steps is given in Eqns. (1.7) and (1.8), respectively, although a myriad of
possibilities exist (see Chapter 3 for an in-depth discussion).
𝑂2 + 2𝐻2 𝑂 + 4𝑒 − → 4𝑂𝐻 −
𝐶𝑒 3+ + 3𝑂𝐻 − → 𝐶𝑒(𝑂𝐻)3

(1.7)
(1.8)

Detailed investigation of and results from the CELD of ceria microstructures are

presented in Chapters 3 and 5. Performance analyses of CELD ceria-based SOFC anode
structures are presented in Chapter 4.

Chapter 2

Anti-Dot Substrates
2.1

Polymer Sphere Lithography Background and Summary

Polymer sphere lithography and, in particular, nanosphere lithography have gained recent
attention for a wide range of applications ranging from novel nanofabrication techniques
and photonic crystals to superhydrophobic surfaces [8, 63-64]. Most often, because the
polymer spheres are typically not treated prior to metal deposition, the resulting patterned
film is limited to isolated locations corresponding to the interstices between the template
beads [7, 65-66]. With control of the fabrication process, however, the film may form a
fully interconnected, yet fully porous network, acquiring what has been termed an ‘antidot’ configuration [67-68]. Specifically, an ordered layer of monodisperse polystyrene
(PS) spheres is first applied to the surface of a SOFC electrolyte material; afterwards, the
spheres are radially etched in an oxygen plasma, so as to create vias between them. Metal
is deposited using a line-of-sight deposition method that enables the still-round PS to
serve as a lithographic mask. Upon removal of the polymer template, the desired anti-dot
porous structure is obtained. This process is illustrated stepwise in Fig. 2.1.
The periodicity provided by polymer sphere self-assembly is not critically
important for SOFC studies; however, sufficient knowledge of microstructural
parameters is required, as is high temperature stability. Accordingly, both factors are
evaluated below.
Essential to the success of the polymer sphere lithographic technique is achieving
a single layer of the polymer spheres across the entirety of the substrate. Several

(a)

(b)

(c)

(d)

Fig 2.1. The polymer lithography process, all on YSZ: (a) a monolayer of 500 nm polystyrene (PS)
spheres; (b) diameter of spheres reduced via oxygen plasma etching; (c) metal (Cu) deposited by thermal
evaporation; (d) PS spheres removed.

approaches for monolayer deposition have been pursued in the literature, with varying
degrees of complexity and experimental constraints. The most common methods are
combined sedimentation plus evaporation [69-70]; spin-coating plus evaporation [71];
and controlled evaporation in combination with gradual substrate withdrawal from the
solution (dip-coating) [72]. More exotic methods include electrophoretic assembly
(suitable only to conducting substrates), and high pressure infusion in combination with
ultrasonication [73-75]. While sedimentation, dip-coating, and spin-coating are relatively
straightforward methods that produce structures with regularity sufficient for
electrocatalysis studies, they suffer from the tendency of the processes to yield regions
with multiple layers and others entirely devoid of the polymer spheres. Furthermore,

achieving adequate control of the evaporation step for the former two can require
excessive processing times. With the exception of a dip-coating setup encased to provide
strict humidity control [72], these methods have difficulty spanning the nanospheremicrosphere range; that is, any given polymer sphere deposition method works with
either nanospheres or microspheres, but not both (note: PS spheres less than 1 µm in
diameter are herein referred to as nanospheres, whereas spheres greater than 1 µm are
referred to as microspheres).
Here, spin coating is employed as a facile means of obtaining the desired
monolayers on YSZ and SDC substrates, where a slight variation of the standard spinning
approach is necessary for microspheres. Utilization of electronically insulating substrates
precludes electrodeposition as a means of subsequent growth of the metallic film,
motivating the two-step process pursued here of bead etching and vapor phase metal
deposition. For the plasma-etched spheres to serve as effective templates, it is necessary
for film growth to be limited to line-of-sight methods that avoid deposition in the void
space on the underside of the round beads. Thermal evaporation has been employed in
this work for conventional metals (copper, nickel, titanium, titanium/gold, aluminum),
whereas electron-beam evaporation has been used for platinum (due to its high melt
temperature), with equal effectiveness in all cases. A representative selection of the types
of anti-dot electrode structures obtained in this work is presented in Fig. 2.2. The ability
to fabricate anti-dot structures from a range of metals on multiple electrolyte materials is
essential for ultimate fundamental electrochemical studies.

(a)

(b)

(c)

(d)

(e)

(f)

Fig 2.2. Selection of representative copper anti-dot metal films on YSZ showing a range of feature sizes
achieved using polymer sphere lithography: (a) 500 nm initial bead size; (b) 790 nm initial bead size;
(c) 2 µm initial bead size, heavily etched; (d) 2 µm initial bead size, lightly etched; (e) 3.2 µm initial
bead size, heavily etched; and (f) 3.2 µm initial bead size, lightly etched.

2.2

Experimental Details

2.2.1

Substrate Preparation

The YSZ substrates (MTI Corporation) used are (100) single-crystals, and the SDC
substrates are epitaxially deposited thin films on single-crystal YSZ via pulsed laser
deposition [32]. For subsequent use for nanosphere deposition, the substrates were
exposed to an oxygen plasma for 5 minutes at 75 W and 250 mTorr (Technics Planar
Etch II) to enhance hydrophilicity. In contrast, as-purchased or as-fabricated substrates
were directly used for microsphere deposition.

2.2.2

Nanosphere Deposition

Monolayers were spun using a Laurell, WS-400B-6NPP/LITE spin coater, with a 10 wt%
suspension of PS nanospheres, surface functionalized with carboxyl groups (Bangs
Laboratories, Inc.™). Before spin coating on the substrate, the as-received PS suspension
is sonicated to ensure the beads are homogeneously dispersed. Exactly 35 µL of the
suspension was manually spread over the entire 1 cm x 1 cm substrate before spinning.
The final RPM of the spin coater was 3000 RPM, with varying accelerations depending
on the starting PS diameter. The PS monolayer was radially etched in the same oxygen
plasma system, but at elevated powers (75 – 200 W). In this step, the beads do not move
from their original positions. Typical etching times were anywhere from 5 – 20 minutes.

2.2.3

Microsphere Deposition

Non-functionalized PS beads were used for the larger diameters (Thermo Scientific), as
10 wt %. 35 µL of the PS suspension was manually spread over the entire substrate and

spun as before. The spin coater was spun at 4000 RPM. A standard laboratory spray
bottle was used to employ the water-wash method, described in detail in Section 2.3.2. If
the spin-wash-dry cycle was repeated too many times, immovable multilayers would
form. For the 2 µm spheres, the cycle was repeated 3 times; for the 3.2 µm spheres, the
cycle was repeated 6 times.

2.2.4

Metal Deposition

An in-house constructed thermal evaporation system was used to deposit copper, nickel,
titanium, titanium/gold, or aluminum (Alfa Aesar, 99.98+%) at 10-5 Torr. Platinum
networks were evaporated using an electron beam evaporator (re-manufactured CHA
MK-40). The now covered PS beads were removed with an acetone-soaked cotton swab;
regardless of the metal deposited, the surface became lustrous after wiping repeatedly,
indicating the PS was gone.

2.2.5

Microstructure Analysis

Optical photos were taken using a Nikon SMZ1500 stereomicroscope. Electron
micrographs were taken on an LEO 1550VP Field Emission SEM. Atomic force
microscopy (AFM) images were collected using a Park Systems XE-70 AFM. Image
analyses were performed using ImageJ 1.41o freeware. Statistical data pertaining to antidot structural features was collected from a series of SEM photos that captured a little
over 1% of the total substrate area, constituting 100-400 photos, depending on the
magnification used. The number of pores evaluated per sample was 60,000-650,000: the

pore areas were assumed to be perfectly circular, and the diameters were calculated from
the individual pore areas.

2.2.6

High Temperature Stability

Nickel networks 200 nm thick were brought to 600 ºC under flowing 98.7% H2 and 1.3%
H2O and held there for 50 hours.

2.3

Results and Discussion

2.3.1

Nanosphere Lithography Results

Monolayers of as-purchased, carboxyl-functionalized polystyrene spheres with diameters
less than 1 µm (specifically, 500, 680, and 790 nm) were prepared by spin-coating onto a
hydrophilic surface (Fig. 2.3), where the optimal spin acceleration, ultimate spin rate, and
dwell time were each found to depend on the sphere diameter. Non-functionalized beads
displayed insufficient attraction to one another and, consequently, spin-coating resulted in
large areas devoid of the template, despite exhaustive attempts at optimizing the spinning
conditions. The next step, etching of polystyrene by oxygen plasma treatment, is wellknown and was readily applied here [68]. It was observed that short treatments generate
contacts between the spheres, presumably as a consequence of softening of the polymer.
This undesirable ‘necking’ was avoided by longer treatments that remove at least ¼ of
the original sphere diameter. Finally, in order to ensure the removal of the template
without damage to the desired pattern, the film thickness is limited to approximately ½
the diameter of the etched spheres, with the further constraint that a minimum thickness
of about 150 nm is required in order to attain acceptable electron transport properties in

Fig 2.3. SEM images of etched PS beads of 500 nm initial diameter covering large areas of the substrate
from a single spin coat step.

the porous film. These considerations preclude fabrication of useful anti-dot electrodes
with PS spheres of less than 500 nm in diameter. Fig. 2.4 shows that qualitatively similar
coverage is achievable with a variety of different PS sphere diameters.

2.3.2

Microsphere Lithography Results

In contrast to the deposition of nanospheres, no set of conditions could be identified for
the preparation of a comprehensive monolayer of PS microspheres using a single spincoating step. Under all accessible spinning conditions, functionalized PS microspheres
assembled into irreversible multilayers (see SEM images in Fig. 2.5) that could not be
modified for further use due to the line-of-sight nature of the metal deposition step. If
metal deposition was pursued regardless of the presence of multilayers, unacceptable

(a)

(b)

(c)

Fig 2.4. Substrate coverage via a single spin-coat for polystyrene nanospheres on YSZ: (a) 500 nm
diameter; (b) 680 nm diameter; and (c) 790 nm diameter. Each image contains ~1200 beads.

levels of irregularity in the anti-dot films resulted, as in Fig. 2.5b. In contrast, at high
spinning rates, non-functionalized PS microspheres formed monolayers, but with only
partial coverage (see Fig. 2.5c), whereas at lower spinning rates multilayer regions
emerged (particularly towards the edge of substrate), without elimination of the void
areas. A low magnification optical image of such a dilemma is shown in Fig. 2.6, where
multilayer regions (white portions) mark the border of the substrate, as well as covering a
little less than half of the remaining interior; and void regions (dark areas, more easily
seen in Fig. 2.6b) litter the entirety of the interior. The thick multilayer border region is of
particular concern, as it can constitute up to 10% of the total substrate.

(a)

(b)

(c)

(d)

Fig 2.5. (a) Irreversible multilayers formed after a single spin coat of functionalized PS microspheres; (b)
the resulting metal film porosity suffers from such multilayers; and (c) and (d) a single spin coat of nonfunctionalized PS microspheres formed monolayers on most parts of the substrate, but with a significant
number of voids.

(a)

(b)

Fig 2.6. Optical photographs of the entire 1 x 1 cm YSZ substrate (a) and a zoomed in view of one
corner (b) with a single spin coating of PS microspheres. The white regions are multilayers, the light
gray regions are monolayers, and the dark regions are voids.

As an alternative to a single-step monolayer deposition procedure, a process was

developed employing multiple spin-coating steps of the non-functionalized PS beads,
depicted in Fig. 2.7. Specifically, a substrate for which the first deposition has yielded a
mixture of void regions, monolayer regions and multilayer regions is gently rinsed with
water to remove the excess layers in the multilayer regions and the spin-coating is
repeated to induce deposition in the void regions. The process is repeated multiple times
until the void regions constitute less than about 10% of the substrate area, beyond which
multilayer regions cannot be removed by a gentle rinse with water. In a final step, a small
amount of the PS suspension is directly applied to the substrate and allowed to dry,
eliminating the remaining void regions but with unordered sphere arrangement, as
opposed to the relatively periodic arrangement produced by spin coating. For this reason,
the spin coating step is repeated as many times as possible before this last step is applied,
as disorder in the PS monolayer undoubtedly affects the 3PB/2PB densities. Fig. 2.8
exhibits the most extreme case of utilizing only one spin coat run, resulting in a large
portion of the substrate being disordered.
Somewhat fortuitously, the build-up of multilayers causes the otherwise bleakly
opaque PS monolayer regions to have a progressively whiter hue—this enables trouble
areas to be identified on-the-spot and water-washed more thoroughly, allowing the
majority of the void areas to be replaced by an orderly arranged PS monolayer via spincoating. As will be explicitly shown in Section 2.3.3, the water-wash method produces
sufficient coverage and ordering so as to give confidence in the predictability of the
theoretical values of 3PB density and 2PB area fraction. As a consequence of these
adaptations, microsphere anti-dot substrate preparation is easily repeatable with a high
degree of accuracy.

(a)

(b)

Fig 2.7. (a) The water-wash method utilizes sequential spin coats with water-washing in-between steps. 2
µm beads shown here on a standard 1 x 1 cm YSZ substrate; (b) an optical photograph showing
comprehensive monolayer coverage over the entirety of the substrate after the final multilayers are washed
off, and a magnified SEM image of a monolayer of 2 µm beads.

Fig 2.8. Optical and SEM images showing the result of utilizing only one spin coat step in the water-wash
method. The darker regions in the SEM image correspond to relatively ordered arrangements of PS beads
from the spin coat step; alternatively, the lighter regions correspond to unordered arrangements from the
final evaporation step. The substrate shown here is 1 x 1 cm YSZ with gold metal.

(b)

(a)

Fig 2.9. Monolayer substrate coverage via multiple spin and wash cycles of polystyrene microspheres on
YSZ: (a) 2 µm, and (b) 3.2 µm. Both images contain ~1200 beads.

By this method, it was possible to prepare comprehensive monolayers of PS beads
up to 3.2 µm in diameter with the same coverage quality as the nanospheres (compare the
nanosphere coverage of Fig. 2.4 to the microsphere coverage of Fig. 2.9). After the
microsphere monolayer deposition is complete, the subsequent plasma treatment, metal
deposition and template removal steps then proceed as described for the nanosphere
lithography, where, again, a minimum of ¼ of the bead diameter must be removed in
order to prevent necking during oxygen plasma treatment.

2.3.3

Microstructural Fidelity

Given the importance of three-phase boundaries for SOFC electrocatalysis, the 3PB areal
density is a key parameter describing the microstructural features of these or any fuel cell
electrode. A further important parameter in the case of the two-dimensional electrodes
prepared here is the metal coverage, or inversely, the fraction of exposed electrolyte area,
i.e., the 2PB areal density. With knowledge of these two parameters and an ability to tune
them over a wide range, it becomes possible to achieve the goal of deconvoluting
microstructural and compositional influences on electrocatalysis rates.

For a perfect micro-/nanosphere lithographic process in which the template beads
display ideal periodicity over the entirety of the substrate, both the 3PB areal density,
𝜌3𝑃𝐵 , and the 2PB area fraction, 𝑓2𝑃𝐵 , can be computed from knowledge of the starting

bead size and the extent of size reduction induced upon plasma etching. The theoretical
values of these two quantities are given in equations 2.1 and 2.2, respectively, as
functions of the initial (𝜙𝑖 ) and final (𝜙𝑓 ) diameters of the PS beads.
𝑡ℎ𝑒𝑜
𝜌3𝑃𝐵

𝑡ℎ𝑒𝑜
𝑓2𝑃𝐵

2𝜋 𝜙𝑓 1
� �
√3 𝜙𝑖 𝜙𝑖

2√3

𝜙𝑓 2

� �
𝜙𝑖

(2.1)
(2.2)

For an imperfect fabrication process, many kinds of defects exist—disordered

regions of PS beads, multilayer and void areas (areal defects); grain boundaries between
ordered regions (line defects); and singly missing PS beads within an ordered region
(point defects). To assess the influence of these random structural elements, a continuous
string of scanning electron microscopy (SEM) images (typically numbering from 100400 images per sample) was collected from border to border for four representative films.
From each image the following parameters were determined: the number and diameter of
the pores, the metal|substrate interface length, i.e., 3PB length, and the exposed
electrolyte area fraction, i.e., 2PB area fraction. The film characteristics and measured
results are summarized below in Table 2.1 and Figures 2.10 and 2.11.
The distribution of pore diameters (Fig. 2.10) in the films was found to be rather
narrow, with a Gaussian peak width that is ~ 2% of the mean diameter. This distribution
largely reflects the size distribution in the as-purchased PS beads, also about 2%, as
oxygen plasma treatment was observed to remove material from the beads in a spatially
uniform fashion, both radially and from bead to bead (Fig. 2.1b) [76]. In addition to the

(a) 160000

(b) 12000
10000

Counts

120000
80000
40000

8000
6000
4000
2000

0.0 0.5 1.0 1.5 2.0 2.5 3.0
Effective Pore Diameter / µm

12000

(c)

5000

8000

Counts

Effective Pore Diameter / µm

(d) 6000

10000
6000
4000
2000

0.0 0.5 1.0 1.5 2.0 2.5 3.0

4000
3000
2000
1000

0.0 0.5 1.0 1.5 2.0 2.5 3.0
Effective Pore Diameter / µm

Fig 2.10. Pore diameter histograms of copper networks on YSZ, reflecting the different starting PS bead
sizes: (a) 500 nm etched to 300 nm; (b) 2 µm etched to 1.29 µm; (c) 2 µm etched to 1.58 µm; (d) 3.2 µm
etched to 1.72 µm.

(b) 20

(a) 40

Counts

30
20
10

20 40 60 80 100
2PB Area Fraction (%)

(d) 16
14
Counts

Counts

20 40 60 80 100
2PB Area Fraction (%)

12
10

Fig 2.11. 2PB area fraction histograms of copper networks on YSZ, reflecting different starting PS bead
sizes: (a) 500 nm etched to 300 nm; (b) 2 µm etched to 1.29 µm; (c) 2 µm etched to 1.58 µm; (d) 3.2 µm
etched to 1.72 µm. Solid and dashed lines indicate the average and theoretical values, respectively.

pores represented in the histograms of Figure 2.10, a small number of pores with large
diameters, > 3 µm, was also observed. These are taken to reflect regions in which
multilayers of PS beads occurred, which prevents metal deposition over larger areas (c.f.
Fig. 2.5b). For the PS with an initial diameter of 2 µm, the number of these multilayerinduced pores is less than 1% of the total; for the 3.2 µm initial diameter spheres, the
number is less than 0.5%; and for the 500 nm spheres, the number is less than 0.1%.
Aside from their statistical insignificance, the contribution of these large diameter pores
to the overall 𝜌3𝑃𝐵 is confirmed to be small, as indicated by the good agreement between

the theoretical and experimental values of this parameter, Table 2.1, and they are omitted
from the plotted range for clarity.
The image-to-image variation in the 2PB area fraction (Fig. 2.11) shows that the

variability in 𝑓2𝑃𝐵 is more significant than the pore diameter variability. The widest

distribution in 𝑓2𝑃𝐵 is evident for the film prepared using 500 nm PS beads, where the

standard deviation is 12% of the substrate area (i.e., 𝑓2𝑃𝐵 is 31.5 ± 12.0%). Moreover, in
all cases, the observed 2PB area was lower than that computed from the initial and final

Table 2.1. Comparison of theoretical and experimental 3PB length areal density and percent 2PB exposure
for different initial PS bead diameters.
Initial bead

Final pore

Pore diameter

Theoretical

Experimental

Theoretical

Experimental

2PB exposure

diameter,

Gaussian

3PB length

percent 2PB

standard

𝝓𝒊 /µm

width/µm

density/m cm-2

exposure

deviation

0.5

𝜙𝑓 /µm
0.3

0.06

435

406

32.6%

31.5%

12.02

1.29

0.07

117

112.8

37.7%

33.4%

6.70

1.58

0.08

143

137.3

56.6%

49.7%

4.69

3.2

1.72

0.10

57.6

26.2%

23.5%

3.60

PS sphere sizes. This can be attributed to the occurrence of point and line defects in the
PS two-dimensional crystals, as well as the presence of disordered regions in the
monolayer. The statistics surrounding the two films prepared using PS beads with an
initial diameter of 2 µm suggest that line and point defects become increasingly important
as the extent of etching is minimized. In the case of the film obtained from lightly etched
𝑒𝑥𝑝

𝑡ℎ𝑒𝑜
PS beads (𝜙𝑓 = 1.58 µm) there is a large difference between 𝑓2𝑃𝐵
and 𝑓2𝑃𝐵 (56.6 vs.

49.7 %). When the beads are more heavily etched (𝜙𝑓 = 1.29 µm), the difference
𝑒𝑥𝑝

𝑡ℎ𝑒𝑜
and 𝜌3𝑃𝐵 and the distribution of pore
decreases, whereas the difference between 𝜌3𝑃𝐵

sizes for the two films are essentially the same. This behavior can be understood as

follows. In the case of the lightly etched film, isolated missing beads (both point defects
and dislocations behave as isolated, absent beads in a two-dimensional crystal) become a
significant portion of the open area available for metal deposition, and, in this manner,
such defects increasing in number dominate the coverage features.
The histograms of 2PB area display significant numbers of occurrences outside of
what is roughly the main peak. As already indicated, regions with 2PB area below the
mean occur as a consequence of defects in the two-dimensional crystals, i.e., voids in the
PS bead array, whereas regions with higher fractions of 2PB area occur because
multilayers form during the PS bead deposition process, i.e., excessive coverage of the
substrate with PS beads. The 500 nm diameter nanospheres generate films in which
slightly less than 10% of the regions display significant PS bead void areas, whereas 4%
display multi-layered areas. In contrast, for both sizes of microspheres (2 and 3.2 µm) the
regions affected by voids in the PS bead array are less than 5%, indicating that the
multiple deposition process has more comprehensively filled the monolayer. The

occurrence of multilayer regions for the 2 µm microspheres accounts for 11% of the
regions imaged, whereas for the 3.2 µm it is only 1%, suggesting that multi-layer removal
becomes facile as the bead size increases.
Overall, despite the imperfection of the monolayer deposition process, the
theoretical and experimental values, respectively, of 𝑓2𝑃𝐵 and of 𝜌3𝑃𝐵 agree quite well

with one another, indicating that the fabrication is, in fact, rather controlled. Indeed, all of
the experimental 𝜌3𝑃𝐵 values are within 93% of the theoretical, and this was found to

hold irrespective of substrate employed or metal deposited. Accordingly, the geometric
features of any sample prepared by the methodology presented here can, within a
reasonable degree of certainty, be predicted from knowledge of 𝜙𝑖 and 𝜙𝑓 .
2.3.4

Thermal Stability

An additional important characteristic of model fuel cell electrodes is thermal stability.
That is, the 3PB length and 2PB area must not change during the course of a hightemperature electrochemical measurement. To evaluate thermal stability, nickel anti-dot
networks were annealed at 600 ºC for over 48 hours under humidified hydrogen, typical
operating conditions for the anode of an intermediate temperature SOFC. SEM images
reveal no discernable microstructural evolution as a consequence of the heat treatment
(Fig. 2.12ab), whereas slight changes are visible in the atomic force microscopy (AFM)
images (Fig. 2.12cd). Specifically, the nickel surface roughens, from a root mean square
roughness of approximately 8 to 14 nm, and the grains undergo slight growth, in a
direction limited largely to the surface normal. No other metal networks were subjected
to high temperatures.

(a)

(b)

(c)

(d)

Fig 2.12. Images of an anti-dot porous nickel network (a) and (c) before thermal treatment at 600 ºC under
hydrogen (pH2 = 0.1 atm); and (b) and (d) after thermal treatment. (a) and (b) are top-down SEM images;
(c) and (d) are AFM images.

Chapter 3

Cathodic Electrochemical Deposition of
Undoped and Doped Ceria
3.1

Introduction

There are numerous reports in the literature regarding the CELD of ceria [45-46, 53, 55,
57, 77-83], including a handful that generally list SOFCs as potential applications, but
with limited demonstration [48-52]. In addition, the contribution of electrogeneration of
base to CELD is well-documented [42-43]. However, insight related to the crucial SOFC
design criteria outlined in Section 1.4.1 is incomplete as most reports focus on asdeposited composition and characterization, which is a broad area of study by itself due
to the large parameter space. Aside from grain growth evolution and brief mention in a
few studies, high temperature data are largely missing [49-51, 78, 80, 83]. Perhaps most
importantly, the vast majority of reports utilize non-porous, purely metallic substrates,
which violate the continuous pathway for ionic species requirement. To the best of the
author’s knowledge, a composite conducting/non-conducting substrate is mentioned only
once as a part of a larger study, in which the CELD of ceria was performed on a
nickel/yttria-stabilized zirconia cermet, but was not explored in detail [45]. Therefore,
this study assesses CELD according to (1) its compositional control, (2) the high
temperature behavior of its coatings, (3) its ability to meet minimum SOFC
configurational requirements, and (4) its potential for wide-ranging microstructural
optimization. In so doing, the electrogeneration of base/chemical precipitation
mechanism is validated with the assistance of others’ previous works [55, 82, 84-86], and

new insights are gained regarding the roles of the working potential and the depositing
species to the resulting microstructure, allowing a predictive, instead of haphazard,
approach to future CELD work.
In order to explore the flexibility of CELD as a fabrication tool, a range of
deposition conditions were examined, and a correspondingly wide range of reproducible
morphologies were obtained. Rather than describe the entirety of those results, two
primary types of morphologies are reported here—high surface area (HSA) coatings and
thin, planar films. These two microstructures are evaluated in the context of the HSA
coatings’ ability to be used as electrode components and the thin, planar films’ ability to
be used as electrolyte components.
As a visual aid to understand the general deposition mechanism and to distinguish
between the two experimental conditions probed in this chapter, a Ce-H2O-H2O2
Pourbaix diagram is shown in Fig. 3.1, adopted from reference [85]. Pourbaix diagrams
are pictorial representations of thermodynamic stabilities in the potential-pH parameter
space, although they do not contain any kinetics information. Initially, with no applied
potential, the pH sits in the range 2.5 – 4 (note that this diagram was constructed versus
the natural hydrogen electrode (NHE), whereas the working potentials in this manuscript
are referenced versus the standard calomel electrode (SCE), or +0.25 V vs. NHE). Once a
working potential is applied, the state of the system moves along a straight horizontal line
to the right, indicating that the electrolyte is becoming more basic. The black arrow on
the right-hand side vertical axis of Fig. 3.1 shows the working cathodic potential for the
HSA coating and the gray arrow indicates the working potential for thin films. The

Fig. 3.1. Pourbaix diagram for the Ce-H2O-H2O2 system, reprinted with permission from [86]. The
black arrow on the right-hand side, vertical axis represents a typical HSA depositing potential (-0.8 V
vs. SCE); the gray arrow represents the thin films’ depositing potential (-0.55 V vs. SCE). Final
interfacial pH values are ~10.5.

applied potentials are used for the two morphologies, regardless of the composition of the
electrolyte solution.

3.2

Experimental Details

3.2.1

Substrate Definition

The primary type of substrates used was a composite substrate, comprised of a supporting
YSZ base, on top of which various kinds of porous metal networks are overlaid. The

(a)

(b)

(c)

Fig. 3.2. Interconnected nickel anti-dot network (a), photolithographically defined platinum strip network
(b), and platinum paste network (c) on single-crystal YSZ supporting substrates. The darker regions of each
image are the exposed YSZ.

pores in the metal films are necessary to allow for oxygen ion flux, and connectedness in
the metal networks is necessary to provide electronic conduction.
All reagents obtained were research grade. YSZ single-crystals (MTI Corp.), 1 cm
x 1 cm x 0.5 mm, oriented (100) are used as the supporting substrates. On the surface of
the YSZ, two primary porous metal network configurations are used. One, 400 nm thick
nickel anti-dot films are made via polymer sphere lithography (Fig. 3.2a), the details of
which are described in Chapter 2 of this manuscript. Two, 200 nm thick parallel platinum
strips are made via conventional photolithography (Fig. 3.2b), obtained from Dr. Yong
Hao, whose work is described in detail elsewhere [31]. The strips are electrically
connected to one another by a platinum border near the edge of the YSZ substrate. Here,
the widths of the platinum strips are made identical to each other and to the open spacing

between them, denoted by the shorthand 5-5µm and 10-10µm, indicating that the widths
are 5 and 10 µm, respectively. It should be mentioned that these lithographic networks
allow both the metal and exposed YSZ surface areas to be specifically known, to a high
degree of accuracy.
One additional, thicker metal network configuration is used, but only to test the
high temperature annealing behavior of the deposits (Fig. 3.2c). Platinum paste
(Engelhard 6082) is spread across the entire YSZ surface and allowed to dry for two
hours, and then heat treated at 400 °C for 1 hour and 900 °C for 2 hours at 1 °C min-1 to
remove residual organics and sinter the platinum particles together. This results in a
spider-web-like network of platinum with feature sizes on the order of microns, necessary
to prevent metal coarsening at higher temperatures from damaging the deposits.
To obtain large amounts of the deposits for bulk studies, 0.25 mm thick nickel foil
substrates are used with depositions performed at 0.8 mA cm-2 for 1-2 hours. The
powdery deposits are subsequently scraped off of the nickel foil and gathered for
analysis. Also, in order to image cross-sections of the thin film deposits, 350 – 400 nm
thick nickel films are thermally evaporated onto 1 x 1 cm silicon substrates.

3.2.2

Experimental Setup

A traditional three-electrode cell, like the one schematically shown in Fig. 3.3a, is used
with a standard calomel electrode (SCE) for a reference electrode, and a carbon rod for
the counter (anodic) electrode, using a Solartron 1286 Electrochemical Interface for
potentio/galvanostatic control.

Four different liquid electrolyte solution compositions are used, but they are all
nitrate-based, with a total cation concentration held constant at 0.05 M. The first
electrolyte solution contains undoped ceria with no additional additives, where the cerium
nitrate concentration is 0.05 M—referred to as “undoped.” The second is samarium
doped with no additional additives, where [Sm3+] + [Ce3+] = 0.05 M, and the relative
samarium content ranges from 4 to 50% of the cerium content—referred to as “Smdoped.” The third is samarium doped, but also contains 0.025 M hydrogen peroxide as an
additive—referred to as “Sm-doped + H2O2.” Hydrogen peroxide is reported to help with
adhesion and promote Ce(IV) precipitation over Ce(III) [45, 48]. The fourth is samarium
doped, but also contains 0.05 M acetic acid as an additive—referred to as “Sm-doped +
acetic.” Acetic acid is most commonly employed in AELD as a stabilizing ligand, which
helps prevent unwanted Ce(III)-based precipitation [54]. In the CELD case, the
stabilizing ligand action of acetic acid has been previously utilized to simply retard the
overall deposition rate, with the hypothesis that slower deposition rates would lead to
denser, more adherent coatings [46]. This electrolyte solution is only briefly investigated,
primarily in Section 3.3.3. The electrolyte solution compositions are summarized in Table
3.1.
Table 3.1. CELD liquid electrolyte compositions.

Electrolyte Solution Name

[Ce(NO3)3]

[Sm(NO3)3]

[Additive]

Undoped

0.05 M

n/a

Sm-doped

0.0498 M

0.002 M

n/a

Sm-doped + H2O2

0.0498 M

0.002 M

0.025 M H2O2

Sm-doped + acetic acid

0.0498 M

0.002 M

0.05 M acetic acid

ee- Mn+
Mn+
ee- Mn+
eMn+
eee- Mn+
ecathode

potential

reference
(standard
calomel
electrode
[SCE])

anode

SCE

cathode

anode

Fig. 3.3. Schematic representation (a) of the standard three-electrode liquid electrochemical cell used for
CELD; and (b) the corresponding relative potential values.

The unadjusted, initial pH of the undoped and Sm-doped electrolyte solutions is
around 4, whereas the Sm-doped + H2O2 electrolyte solution is 2.5 – 3, and the Sm-doped
+ acetic electrolyte solution is 2.5. All electrolyte solutions are allowed to naturally aerate
before each deposition, ensuring that an adequate measure of dissolved oxygen is
incorporated. The depositions are conducted at room temperature, and over a metal
surface area roughly equal to 0.5 cm2.
First, high surface area (HSA) coatings are obtained in galvanostatic mode, at 0.8
– 2 mA cm-2, deposited for 1 – 60 minutes, which corresponds to 0.5 – 20 µm thick
coatings. The effective operating voltages for the HSA coatings are approximately -0.7 to
-1.0 V vs. SCE. Second, thin, planar films are obtained in potentiostatic mode, at -0.5 to 0.55 V vs. SCE, deposited for 0.2 – 60 minutes, which corresponds to 30 – 300 nm thick
films.

3.2.3

Characterization Details

Both as-deposited and annealed deposits are analyzed, with typical annealing
temperatures ranging from 650 – 1000 °C for 10 – 24 hours, in 0.1% H2 in Ar, or in
ambient air. “Bulk” characterization results from deposits that are scraped off of nickel
foil substrates (not patterns), to eliminate convolution of substrate effects. These coatings
are deposited for relatively longer periods at the HSA working potential, but are identical
in every other way to their thinner counterparts. X-ray diffraction patterns (XRD) are
obtained using a Phillips X’Pert Pro powder x-ray diffractometer using Cu Kα radiation
(45 kV, 40 mA). Raman spectra are obtained with a Renishaw Ramascope (532 nm diode
pumped laser) equipped with a Leica DMLM microscope, and FT-IR spectra are obtained
using a Durascope Nicolet ATR system (KBr beam splitter). Thermogravimetric analysis
is performed with a Netzsch STA 449 C. The electrolyte solutions are characterized using
cyclic voltammetry (CV) at 50 mV s-1 from 0 to -1.25 V vs. SCE, using the same
Solartron 1286. The morphology of the coatings are imaged using scanning electron
microscopy (SEM), with two different systems, a Zeiss 1550VP FE SEM equipped with
an Oxford INCA x-ray energy dispersive spectrometer (EDS) and a Hitachi S-4100 FE
SEM. Atomic-force microscopy (AFM) is used to measure the thin, planar films’
roughness with a Park Systems XE-70. Transmission electron microscopy (TEM) is
performed on a FEI Tecnai F30UT operated at 300 kV, with the lift-out performed on an
Omniprobe Autoprobe 200 (the lift-out procedural details are summarized elsewhere)
[87-88].

3.3

Results

3.3.1

Bulk

Fig. 3.4 shows the as-deposited and annealed XRD patterns for typical CELD ceria
deposits, both for the undoped and Sm-doped electrolyte solutions. These particular
deposits were obtained at 0.8 mA cm-2, approximately corresponding to -0.8 V vs. SCE
(the Sm-doped, Sm-doped + H2O2, and Sm-doped + acetic electrolyte solutions give
qualitatively identical XRD patterns). In all cases, both the as-deposited and annealed
deposit patterns show a cubic fluorite structure, indicating that CeO2-δ is the primary
phase at this working potential, regardless of the temperature history or annealing
atmosphere. Similar behavior has been observed in the literature [53, 77]. The asdeposited patterns are shifted to slightly lower diffracting angles, indicating some level of
Ce(III) content that is afterwards oxidized to Ce(IV) upon annealing. The lattice
constants of the doped films post-annealing can be used to determine how much
samarium is incorporated into the ceria structure. The measured samarium doping levels
in the deposits are compared to the nominal samarium doping levels in the electrolyte
solution, and are plotted in the inset of Fig. 3.4. Also shown are the EDS compositional
analyses. It is evident that the concentration of samarium in the films is greater than the
concentration of samarium in the electrolyte solutions. From these results, a solution
samarium relative concentration of 4.6% is chosen in order to obtain a target deposit
composition of ~12%, which is a desirable doping level in terms of optimal oxygen ion
conductivity.
Some ambiguity in the literature exists with regard to the as-deposited crystal
structure, as both crystalline Ce2O3 and Ce(OH)3 are possibilities, although some of the

Measured Doping
Level / %

1800
1600

Intensity / a.u.

1400

doped 700°C

1200

Nominal
XRD
EDS

50
40
30
20
10

Nominal Doping Level / %

1000
800

doped as-dep

600

undoped 700°C

400

undoped as-dep

200

Position / °2θ
Fig. 3.4. As-deposited and annealed XRD patterns for the undoped and doped electrolytes, deposited at
0.8 mA cm-2 (~ -0.8 V vs. SCE). The annealed patterns are identical in both 0.1% H2 in Ar and ambient
air annealing atmospheres. Also shown is the doping level measured in the coatings by XRD and EDS
vs. the doping level in the liquid electrolyte (inset).

variance can be explained by differing operational parameters that greatly affect the
precipitating species. Ce2O3 has distinct XRD peaks from CeO2, and it is clear that those
peaks are absent here, in contrast to a report where traces were detected [59]. On the
other hand, Ce(OH)3 shares some large-intensity peak positions with CeO2, differing by
less than one degree of each other, which convolutes shifting that is solely due to mixed
valency in the CeO2 phase. However, among other missing peaks, a particularly highlydiffracting Ce(OH)3 peak at ~39.5° is missing in all of the as-deposited patterns, strongly
suggesting that Ce(OH)3 is not the diffracting phase. Furthermore, the fact that the

(a)

Ce-O8 stretching mode

2.5

Intensity / a.u.

2.0

oxygen
vacancies

nitrate ions

doped
H2O2

1.5

1.0

doped

0.5

undoped
0.0
250

500

750

1000

1250

-1

HWHM / cm-1

(b)

(c)

0.20

Intensity / a.u.

Wavenumber / cm

0.15

Saitzek 2008
Weber 1993
Kosacki 2002
HWHM Lorentz

0.0

0.1

d-1 / nm-1

0.2

0.3

700°C anneal

0.10
0.05
0.00
3000

as-dep
1000°C anneal
3500

Wavenumber / cm-1

4000

Fig. 3.5. Raman spectra (a). Top group: doped + H2O2 electrolyte; solid line is as-deposited, dashed line is
700 °C annealed, dotted line is 1000 °C annealed. Middle group: doped electrolyte; solid line is asdeposited, dashed line is 700 °C annealed, dotted line is doped ceria reference powder. Bottom group:
undoped electrolyte; solid line is as-deposited, dashed line is 700 °C annealed, dotted line is undoped ceria
reference powder. (b) HWHM vs. crystallite size for the undoped electrolyte with literature comparison. (c)
Hydration peaks for the doped electrolyte– the relative intensity decreases after annealing. All annealing is
under ambient air conditions.

annealed patterns are simply shifted versions of their as-deposited counterparts reinforces
this notion.
The as-deposited peaks are broader than the annealed peaks in every case,
indicating grain growth at high temperatures. Using the general Scherrer equation (D =
0.9λ/βcosθ), where D is the crystallite size, λ is the x-ray wavelength, β is the adjusted
full-width half max of the peak at position 2θ, the as-deposited crystallite size is
determined to be approximately 6 nm, which increases to 15 – 20 nm after annealing at
700 °C for 10 hours in either reducing or oxidizing atmospheres. This is in good
agreement with others’ CELD of ceria findings [78, 83]. The undoped results are
summarized in Fig. 3.5b, where the half-width at half-max of the Raman peak at ~466
cm-1 is plotted against the inverse of the crystallite size. The sizes obtained in this work
even correspond well with undoped ceria obtained by other fabrication means [89-91].
Fig. 3.5a shows the Raman spectra obtained for all three electrolyte solutions asdeposited and annealed at 700 °C, as well as undoped and Sm-doped reference spectra,
and one scan of a Sm-doped + H2O2 sample annealed at 1000 °C. All of the spectra share
a main Ce-O stretching mode peak centered at ~466 cm-1 [89-91], which shifts upon
samarium doping [46, 50]. As can be seen, the annealed undoped and Sm-doped main
peaks agree well with their respective references. This band can also reportedly shift up
to ~10 wavenumbers due to nanoscale crystallite size effects [58, 91-92], which could
partly explain why the Sm-doped and Sm-doped + H2O2 as-deposited peaks are shifted
from one another. The exact doping level in the deposits could also be slightly different
for these two electrolyte solutions. A band at ~600 cm-1 indicates oxygen vacancies in the
ceria lattice [46], which is seen in all of the Sm-doped and Sm-doped + H2O2 scans, as

well as the as-deposited undoped sample. The origin of this peak is the samarium doping
in the Sm-doped electrolyte solutions’ cases, and partial Ce(III) content in the asdeposited undoped case, whose peak disappears upon annealing. These data agree well
with the XRD results. Also, oxygen vacancies have been known to shift the main Ce-O
peak, perhaps explaining the lack of movement from as-deposited to annealed samples
with the Sm-doped electrolyte solution. The bands centered at ~740 and 1049 cm-1 are
attributed to nitrate ions [53], and they disappear upon annealing, as expected.
In the 3000 – 4000 cm-1 range, a broad multi-peak exists for all of the electrolyte
solutions’ as-deposited samples, depicted for the Sm-doped electrolyte solution in Fig.
3.5c. A peak in this range indicates some O-H inclusion in the deposit, either as H2O or
hydroxides. The distinction between the two, particularly for xH2O·CeO2 and Ce(OH)4 is
subtle [56, 92], and unimportant for the ultimate aim of this work. What this peak does
indicate, however, is that there is some hydrated content in the as-deposited samples,
even though the XRD patterns exhibit the cubic fluorite structure. After annealing to 700
°C, these Raman peaks decrease in relative intensity and appear to reach a steady-state,
possibly indicating some grain boundary and/or unavoidable powder surface hydration.
Selected FT-IR spectra are shown in Fig. 3.6. All depositions give qualitatively
identical spectra, so only the Sm-doped case is shown. These results show further
evidence of hydration of the as-deposited material. The as-deposited scan (Fig. 3.6a)
exhibits two peaks attributed to OH stretching (~3600 cm-1, broad) and bending (~1640
cm-1, sharp) modes, that are shared with liquid water, shown for reference in Fig. 3.6 as
the dotted line. Additionally, carbonate (~1450 cm-1) and nitrate (~1300 and 1040 cm-1)
peaks are identified [53]. Fig. 3.6c and 3.6d are spectra from Sm-doped deposits annealed

0.35
nitrate

water stretching mode

0.30

carbonate

Intensity / a.u.

0.25
0.20
0.15

(a)

water bending mode

0.10

nitrate

(b)

0.05

(c)

0.00
4000

(d)
3500

3000

2500

2000

1500

1000

-1

Wavenumber / cm

Fig. 3.6. FT-IR spectra for the doped electrolyte. (a) As-deposited (solid line) and liquid water droplet
(dotted line) overlaid; (b) doped ceria powder reference; (c) 700 °C annealed; and (d) 1000 °C annealed.
1.05

Mass Loss / %

1.00

0.95

0.90

0.85

doped + H2O2
undoped

0.80

0.75

doped

200

400

600

800

1000

Temperature / °C
Fig. 3.7. TGA weight loss for three electrolytes. Final weight loss ranges from 13-21%.

at 700 and 1000 °C, respectively, showing all peaks dramatically losing intensity,
consistent with the XRD and Raman results. The spectra from the annealed deposits
correlate well with a SDC reference spectrum (Fig. 3.6b). TGA measurements underscore
these data, showing significant weight loss (~13-21%) upon heating (Fig. 3.7) [48, 78].
From the previous results, this is probably due to a combination of hydration, nitrate, and
carbonate removal.

3.3.2

High Surface Area (HSA) Coatings

Fig. 3.8 and Fig. 3.9 show representative SEM images of a HSA coating deposited at 0.8
mA cm-2 on various substrates with the undoped and Sm-doped electrolyte solutions,
respectively. The microstructure appears to consist of an overlapping, intersecting
combination of needle-like and nano-sheet growth emerging from the substrate base, with
widths varying from 10 – 50+ nm. Similar microstructures have been reported in refs [46,
53, 77, 80]. There is a slight morphological change between the undoped and Sm-doped
electrolyte solutions, as can be seen by comparing Figures 3.8 and 3.9, but is minimal
compared to the difference toggled by changing the depositing potential. The coating
thickness depends on the deposition time, but typically ranges from slightly less than 1
µm up to tens of microns.
The operational current density was found to alter the overall morphology and,
particularly, the sizes of the finer features, with 0.8 mA cm-2 being the most ideal current
density because of its large-scale deposition uniformity and its nanoscale features.
Extensive SEM analysis of the deposits obtained from a variety of experimental
conditions revealed that smaller current densities tend toward lower surface area,

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 3.8. SEM images of the as-deposited HSA microstructure deposited on YSZ/5-5 µm platinum
pattern substrates at 0.8 mA cm-2 with the undoped electrolyte for 5 minutes (a), (b), and (c); and 10
minutes (d), (e), and (f).

(a)

(b)

(c)

(d)

(e)

Fig. 3.9. SEM images of the as-deposited HSA microstructure deposited at 0.8 mA cm-2 with the
doped electrolyte. Nickel-only substrates used in (a) and (b) with a 5 minute deposition; a nickel antidot network on a YSZ substrate used in (c) with a 5 minute deposition; and a 5-5 µm platinum strip
network on YSZ used in (d, top-down view) and (e, cross-section) with a 10 minute total time
deposition. A platinum strip is seen in (e) directed out of the page.

featureless coatings, and larger current densities tend to encourage different growth rates
in different depositing areas, producing bush-like regions at the expense of less
developed regions. 0.8 mA cm-2 approximately corresponds to -0.8 V vs. SCE;
alternatively, these microstructures can be fabricated potentiostatically, but galvanostatic
mode is more consistent run-to-run for the HSA morphology. Despite the fairly
randomized nanoscale features, the deposition is ubiquitous and uniform, even up to
several cm2. The morphology evolves from nicely adherent to largely cracked as the
deposition time and, hence, coating thickness increase. The HSA microstructure is
maintained, even when metal network substrates are used (c.f. Fig. 3.8, 3.9c, d, and e).
An analogous, but not identical HSA morphology can be obtained using the Sm-

(a)

(b)

(c)

(d)

Fig. 3.10. SEM images of the as-deposited HSA microstructure obtained with the doped + H2O2 electrolyte
at 0.8 mA cm-2.

doped + H2O2 electrolyte solution, shown in Fig. 3.10, but the finer features are larger in
size and fewer in number than the Sm-doped electrolyte solution and thereby less ideal
for surface area enhancement. The HSA structure was not attempted for the Sm-doped +
acetic electrolyte solution, as the acetic acid addition is intended for dense, thin film
growth.
To probe the high temperature stability of the HSA microstructure, HSA
deposition was performed on platinum paste substrates (c.f. Fig. 3.2c), and subsequently
annealed at temperatures of 800, 1000, and 1100 °C (Fig. 3.11a, b, and c, respectively)
for 10 hours in ambient air. As can be seen, the nanoscale features stay intact up to 800
°C, at which temperature some coarsening begins and then worsens as the temperature

(a)

(b)

(c)

(d)

Fig. 3.11. Doped electrolyte HSA morphology high temperature stability with platinum paste networks on
YSZ substrates: (a) annealed for 10 hours in ambient air at 800 °C; (b) 1000 °C; and (c) 1100 °C. Also,
cracks can form after annealing, shown in (d) for the doped electrolyte on a YSZ/nickel anti-dot substrate,
but are healed by a subsequent deposition (d, inset).

increases up to 1100 °C, at which point the finer features have nearly coarsened away.
This coarsening behavior is reasonable, considering the melting temperature of ceria is
~2400 °C. Although not always the case, some microscale shrinkage that leads to
significant cracking can occur even after lower temperature annealing, i.e., 650 – 700 °C
(Fig. 3.11d). This problem is particularly pronounced for thicker coatings. However,
subsequent depositions can heal cracks that have formed, at least as far as the SEM can
image, as shown in the inset of Figure 3.11d. Additionally, for relatively thick coatings,
the microstructure can emerge from the electrolyte solution cracked as-deposited, even
before any heat treatment—without exception, previous studies on the HSA
microstructure only report cracked as-deposited coatings [46, 53, 77, 80]. It is unclear
whether this cracking is due to drying, or the deposition process itself. For both the
deposition-related and annealing-related cracking issues, the comparatively thinner, more
conformal depositions on porous metal networks on YSZ are more resistant to cracking
than metal-only substrates. The in-plane degrees of freedom of the porous metal networks
could provide an avenue to relieve the two crack-inducing driving forces. In fact, it is
entirely possible to produce crack-free, annealed HSA coatings, as can be seen in Fig.
3.12.

Fig. 3.12. As-annealed, crack-free HSA morphology deposited from a Sm-doped electrolyte solution at 0.8
mA cm-2 for 5 minutes onto a YSZ/Ni anti-dot substrate and thermally treated at 600 °C in 0.1% H2 in Ar
for 10 hours.

(a)

(b)

(c)

(d)

Fig. 3.13. As-deposited thin film morphologies. (a) Doped electrolyte (deposited at -0.525 V vs.
SCE), highly magnified view of the thin film deposit closing a pore of the nickel anti-dot network; (b)
top-down view of the characteristically featureless doped + H2O2 electrolyte thin film; (c) cracks form
at thicknesses above 200 nm with the doped + H2O2 electrolyte; (d) globular, three-dimensional
growth occurs at thicknesses above 250 nm with the doped + H2O2 electrolyte. The samples in (b) –
(d) are deposited at an applied potential of -0.55 V vs. SCE.

3.3.3

Thin Films

Planar, thin films were obtained potentiostatically at operating voltages from -0.5 to -0.55
V vs. SCE. These films are practically featureless, as in Fig. 3.13b, and nearly atomically
smooth—even after annealing at 650 °C for 24 hours in 0.1% H2 in Ar, the average
surface roughness measured by AFM is 1.6 nm for the Sm-doped electrolyte solution and
1.3 nm for the Sm-doped + H2O2 electrolyte solution (Fig. 3.14). For all four electrolyte
solutions, crack-free, planar growth occurs up to a point, when undesirable three-

(a)

(b)

Fig. 3.14. AFM roughness scans for thin film morphologies from the doped and doped + H2O2 electrolytes
with 1.6 and 1.3 nm roughness values, respectively. Both samples are deposited on thin film nickel on
silicon substrates and subsequently annealed at 650 °C for 24 hours in 0.1% H2 in Ar. (a) is deposited at
-0.525 V vs. SCE for 30 minutes and (b) is deposited at -0.55 V vs. SCE for 0.5 minutes.

dimensional growth and cracking begin. It might be possible to utilize multiple sequential
depositions to increase the thickness range for crack-free growth as in ref [48].
For the Sm-doped electrolyte solution, planar growth occurs up to ~50 nm (Fig.
3.15a), after which three-dimensional growth predominates; although, unlike the HSA
microstructure, its sharp features are not on the nanoscale. If deposition is continued,
cracking occurs when the planar plus three-dimensional growth thickness reaches about 1
µm. Light-brown coloration can be seen overlaying the lustrous metallic regions of the
substrate. This thin film growth is enough to begin to close the pores of the nickel antidot network, as in Fig. 3.13a.
For the Sm-doped + H2O2 electrolyte solution, planar growth occurs up to ~250
nm (Fig. 3.15b), after which three-dimensional growth predominates. Unlike the Smdoped electrolyte solution, cracking occurs before the three-dimensional growth period,
at thicknesses around 200 nm (Fig. 3.13c). This electrolyte solution’s three-dimensional
growth is clustered and also lacking finer features (Fig. 3.13d). Resembling the Smdoped electrolyte solution, a light-brown coloration is observed for films 50 – 100 nm
thick, and deep blue-green and red-orange colors occur for thicker films.

(a)

(b)

Fig. 3.15. SEM images of as-deposited thin film cross sections deposited on thin film nickel on silicon
substrates from (a) the doped electrolyte at -0.5 V vs. SCE for 1 hour, with a thickness of ~40 nm (image
taken at a 75° angle); and (b) the doped + H2O2 electrolyte at -0.55 V vs. SCE for 5 minutes, with a
thickness of ~200 nm (image taken at a 90° angle).

The Sm-doped + acetic electrolyte solution yields deposits very similar to the Smdoped electrolyte solution (see Fig. 3.16ab). This electrolyte solution poses some
difficulty in maintaining uniform two-dimensional growth across the entirety of the
substrate, however. As an example, Fig. 3.16c is shown, which is an image taken from
the same sample as in Fig. 3.16a and b. Although limited to a few trials, adjusting the
initial pH by NaOH addition significantly alters the deposited morphology, as can be seen
in Fig. 3.16d, e, and f for an initial pH adjusted from 2.5 to 5. With higher pH, rounded,
gumdrop-like islands nucleate and grow, although it is unclear if they are completely
connected, and there are uncontrollable sections of three-dimensional growth.
Sm-doped + H2O2 ceria-coated platinum strips are shown as-deposited (Fig.
3.17a) and annealed in 0.1% H2 in Ar (Fig. 3.17b). Practically no distinction can be made
via SEM before and after annealing. Fig. 3.17c shows two platinum strips after annealing,
where the left strip is coated and the right is uncoated. Significant surface roughness
differences can be seen between the two, noting that the exposed platinum has begun to
coarsen, whereas the coated platinum appears to be physically prevented from doing so.
That neither the as-deposited nor the annealed films are cracked is consistent with a

report that defined a critical crystallite size of 28 nm, via XRD determination, only above
which cracking in thin ceria films occur, albeit for anodic depositions [93]. From the
XRD analysis above, the as-deposited and annealed crystallite sizes in this work are
below this critical value.

To summarize the CELD findings thus far, depositing conditions were identified
that produced both HSA and thin, planar film morphologies, across a variety of
electrolyte solution compositions. It appears as though an additive-free electrolyte
solution maximizes the apparent surface area for the HSA deposits. Also, hydrogen
peroxide allows thicker films to be deposited, as compared to the additive-free Sm-doped
electrolyte solution. In all cases and for each morphology, there is no discernable
microstructural evolution at fuel cell operating temperatures. Desirable levels of
samarium doping have also been incorporated into the deposits.

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 3.16. SEM images of as-deposited thin films from the doped + acetic electrolyte deposited on thin film
nickel on silicon substrates. The sample shown in (a), (b), and (c) is deposited at -0.55 V vs. SCE for 1
hour, with an initial pH of 2.5; the sample shown in (d), (e), and (f) is deposited at -0.5 V vs. SCE for 1
hour, with an initial pH of 5.

(a)

(b)

(c)

Fig. 3.17. Three SEM images taken from the same platinum strip on YSZ sample, with a thin ceria coating
deposited from the doped + H2O2 electrolyte at -0.55 V vs. SCE for 0.5 minutes: (a) as-deposited, coated
platinum strip before annealing; (b) a coated platinum strip after annealing at 650 °C for 24 hours in 0.1%
H2 in Ar, with no discernable microstructural evolution of the coating or the platinum; (c) different
coarsening behavior is observed for coated (c, left) and uncoated (c, right) platinum strips after annealing.
Identical results are observed for the doped electrolyte as well.

3.4

Discussion

3.4.1

General Deposition Overview

Recall from Section 1.4.2 that the CELD of ceria via electrogeneration of base proceeds
in two distinct steps—electrochemical reduction of electrolyte solution species, and
subsequent chemical precipitation of cerium species. Note: Ce3+/4+ herein specifically
refers to dissociated aqueous ions, whereas Ce(III/IV) refers to precipitated/solid species
of a particular cerium valence state.
During the electrochemical reduction step, either acidic species are consumed or
basic species are produced, and both quickly increase the interfacial pH. The primary
species that are reduced are dissolved oxygen, hydronium ions, water, nitrate ions, and
hydrogen peroxide (if present) [80, 86]. Although not exhaustive, the following list of
equations describes their reduction behavior:
𝑂2 + 2𝐻2 𝑂 + 4𝑒 − → 4𝑂𝐻 −

(3.1)

2𝐻3 𝑂+ + 2𝑒 − → 𝐻2 + 2𝐻2 𝑂

(3.3)

𝑂2 + 2𝐻2 𝑂 + 2𝑒 − → 2𝑂𝐻 − + 𝐻2 𝑂2

(3.2)

2𝐻2 𝑂 + 2𝑒 − → 𝐻2 + 2𝑂𝐻 −

(3.4)

𝑁𝑂3− + 7𝐻2 𝑂 + 8𝑒 − → 𝑁𝐻4+ + 10𝑂𝐻 −

(3.5)

𝐻2 𝑂2 + 2𝑒 − → 2𝑂𝐻 −

(3.7)

𝑁𝑂3− + 𝐻2 𝑂 + 2𝑒 − → 𝑁𝑂2− + 2𝑂𝐻 −

(3.6)

As an applied cathodic potential is made more negative, interfacial pH values

measured in situ increase to and stabilize at ~10.5 at around -0.4 V vs. SCE, until the
potential reaches -1.0 V vs. SCE, at which point the pH jumps upwards of 12 [86]. This
corresponds well with calculated interfacial pH values in the range of 10.5 – 10.8 at -0.85
V vs. SCE [82]. Addition of nitrate ions to the electrolyte solution only affect the pH
values at potentials more negative than -1.0 V vs. SCE [86], outside of the operational

range for this work; therefore, the effect of nitrate ions on the pH can be disregarded.
Note that for all of these reductions, available electrons are required at the surface of the
cathode in order for base to continue to be electrogenerated.
Once the electrolyte solution has become sufficiently basic, or, equivalently,
enough hydroxide ions have been produced, chemical precipitation begins. There are two
precipitation pathways thought to occur. The first is through Ce(III) (Eqn. 3.8), and the
second is through Ce(IV) species, but only if H2O2 is present (Eqn. 3.9-3.11):
𝐶𝑒 3+ + 3𝑂𝐻 − → 𝐶𝑒(𝑂𝐻)3

(3.8)

𝐶𝑒(𝑂𝐻)2+
2 + 2𝑂𝐻 → 𝐶𝑒(𝑂𝐻)4 ↓

(3.10)

2𝐶𝑒 3+ + 2𝑂𝐻 − + 𝐻2 𝑂2 → 2𝐶𝑒(𝑂𝐻)2+

(3.9)

𝐶𝑒(𝑂𝐻)2+
2 + 2𝑂𝐻 → 2𝐻2 𝑂 + 𝐶𝑒𝑂2 ↓

(3.11)

If Ce(III/IV) hydroxides are the precipitating species, they are readily oxidized to CeO2 in

the presence of O2:
4𝐶𝑒(𝑂𝐻)3 + 𝑂2 → 4𝐶𝑒𝑂2 + 6𝐻2 𝑂
𝑂2
𝐶𝑒(𝑂𝐻)4 �� 𝐶𝑒𝑂2 + 2𝐻2 𝑂

(3.12)
(3.13)

Recognizing that naturally aerated electrolyte solution s contain a non-trivial amount of
dissolved oxygen, these chemical oxidations can proceed to CeO2 at any point during the
deposition, and definitively occur once the deposit is taken out of the liquid electrolyte
and exposed to the ambient air. The previous XRD and Raman results suggest that this
oxidation process is fast, even at room temperature. This explains how, even for
conditions that should yield strictly Ce(III) precipitation, some Ce(IV) is detected in situ
via XANES [82].
The Pourbaix diagram from Fig. 3.1. is a helpful aid to understand which cerium
species are involved during the chemical precipitation step. As can be seen in the

diagram, for most pH values and mild potentials, Ce3+ is the predominant ion in solution.
Recall that the initial state of the solution is with no applied potential and at a pH in the
range 2.5 – 4. As the state of the system moves to the right of the diagram, meaning that
the electrolyte solution is becoming more basic, stability lines are crossed, which
specifically indicates which species are involved in the precipitation step.
Consider the case for the additive-free undoped and Sm-doped electrolyte
solutions. As a reminder, the black arrow on the right-hand side vertical axis of Fig. 3.1
shows the working cathodic potential for the HSA coatings. As the electrochemical
reduction reactions proceed and the interfacial pH increases, it is clearly seen that the
Ce3+|Ce(OH)3 stability line is crossed, described by Eqn. 3.8. Intermediates of the type
+(3−𝑥)

𝐶𝑒(𝑂𝐻)𝑥

exist, where x ranges from 0 to 2, but reported precipitation tests

concluded that the kinetics are fast and continuous in progressing from Ce3+ to Ce(OH)3
[84]. The final pH is around 10.5, at which point Ce(OH)3 is no longer aqueous, but
precipitates out of solution. The gray arrow in Fig. 3.1 indicates the working cathodic
potential for thin films in the undoped and Sm-doped electrolyte solutions. Similar to the
HSA working potential, the Ce3+|Ce(OH)3 stability line is crossed, indicating that
Ce(OH)3 is the stable precipitating species for both morphologies in electrolyte solutions
where H2O2 is not explicitly added.
The addition of hydrogen peroxide to the electrolyte solution greatly influences its
chemistry, in particular by chemically inducing the formation of Ce(IV) aqueous species
before any potential is applied, according to Eqn. 3.9. This is observed experimentally,
manifest by the electrolyte solution appearance changing from transparent to slightly
yellow, which is a characteristic color of the Ce4+ valence state. Similarly to the above

67
+(4−𝑥)

discussion, Ce4+ intermediates exist of the type 𝐶𝑒(𝑂𝐻)𝑥

, where x ranges from 0 to

3, but, compared to 𝐶𝑒(𝑂𝐻)+2
2 , the others are either thermodynamically unfavorable in

the pH range above 2.5, or kinetically unfavorable shown by precipitation tests [84-85].
Once the requisite hydroxide ions are produced, Ce(IV) precipitates form according to
either Eqn. 3.10 or 3.11. As mentioned previously, the difference between small
crystallites of xH2O·CeO2 and Ce(OH)4 is minor, especially considering that the
oxidation reaction of Ce(OH)4 to CeO2 (Eqn. 3.13) is spontaneous and sufficiently fast,
according to the XRD results in Section 3.3.1, which unambiguously show the final
crystal structure to be that of cubic fluorite CeO2.
One other H2O2-related pathway to precipitation is possible, where Ce3+ ions are
directly oxidized to Ce(IV) precipitates, without the 𝐶𝑒(𝑂𝐻)+2
2 intermediate, according
to:

2𝐶𝑒 3+ + 𝐻2 𝑂2 + 6𝑂𝐻 − → 2𝐶𝑒(𝑂𝐻)4 ↓

2𝐶𝑒 3+ + 𝐻2 𝑂2 + 6𝑂𝐻 − → 4𝐻2 𝑂 + 2𝐶𝑒𝑂2 ↓

(3.14)
(3.15)

Considering the number of species involved in these reactions, however, this scheme
seems less kinetically likely than the intermediate-involved pathway proposed above.
Either precipitation mechanism involving hydrogen peroxide can be understood in light
of the Pourbaix diagram by recognizing that hydrogen peroxide increases the effective
solution potential, changing the initial and final potential-pH states of the system [84].
One must also consider the in situ production of hydrogen peroxide via the
electrochemical reduction of dissolved oxygen through the two-electron pathway (Eqn.
3.2). Consequently, some of the produced Ce(IV) species will have been the result of the
𝐶𝑒(𝑂𝐻)+2
2 path discussed above, even in nominally non-H2O2 electrolyte solutions.

However, given that the concentration of dissolved oxygen is on the order of µM [82],
the effect on the morphology should be much less than when mM of hydrogen peroxide
is explicitly added.

Cathodic Current Density / A cm-2

0.014
0.012
0.010
0.008
0.006
0.004

doped + H2O2

0.002

doped

0.000
-0.002

0.0

-0.2

-0.4

-0.6

-0.8

-1.0

-1.2

-1.4

Voltage / V
Fig. 3.18. CV scans for the doped (solid line) and doped + H2O2 (dotted line) electrolytes, taken at 50
mV s-1. Gray arrows indicate scan direction.

In order to further distinguish between the electrochemistries of the Sm-doped
and Sm-doped + H2O2 electrolyte solutions, their CV scans are shown in Fig. 3.18. The
two scans are qualitatively similar to one another except the small peak around -0.4 V vs.
SCE in the Sm-doped + H2O2 case. The shared, large current leg on the more negative
potential side of the scan is primarily due to oxygen reduction, hydrogen evolution, and
water reduction [86]. The additional peak with H2O2 could be related to either the

reduction of H2O2 itself (Eqn. 3.7), or the reduction of Ce(IV) species that had been
previously oxidized by H2O2. Using KNO3 + H2O2 as a reference electrolyte solution
without any cerium species, the resulting CV scan exhibits no additional peak (not
shown), indicating that its presence is related to the cerium species. Therefore, the peak’s
likely origin is an electrochemical reduction of the type:
3+
𝐶𝑒(𝑂𝐻)2+
+ 2𝑂𝐻 −
2 + 𝑒 → 𝐶𝑒

(3.16)

The hysteresis associated with this peak is related to the deposition that occurs

during the more negative section of the scan, which covers the electrode. The fact that
there is no additional peak observed in the Sm-doped electrolyte solution CV even after
multiple scans reinforces the notion that, although there is some in situ H2O2 production
via the reduction of dissolved oxygen, the amount is small enough to not impact the
chemistry; otherwise, an additional peak would be observed in the Sm-doped electrolyte
solution eventually.

3.4.2

The Physical Deposition Picture

A fundamental question remains unanswered. How does deposition of an insulating metal
oxide (as CeO2 is under these conditions) continue after an initial film is formed? To
address this issue, the following argument is presented.
Recall that electrons are needed at the cathode surface to reduce the various
available species, producing hydroxide ions. For 0.8 mA cm-2 at -0.8 V vs. SCE through a
100 nm film, an electronic conductivity of 10-8 Ω-1 cm-1 is required. Using the room
temperature mobility for electrons in undoped CeO2 [94], ~10-8 cm2 V-1 s-1, this gives an
electron concentration of ~1017 cm-3. However, using thermodynamic data to calculate

the electron concentration of undoped CeO2 at room temperature [32], a value of ~10-26
cm-3 is obtained, far below what is needed. Therefore, neither CeO2, nor the nonconducting Ce(OH)3/Ce(OH)4 are electronically conducting enough to facilitate continual
CELD. Consequently, the deposit must remain somewhat porous throughout, constantly
allowing molecular access to the metal|electrolyte solution interface, where the requisite
electroreductions occur. A 65% dense CeO2 thin film is reported on Hastelloy substrates
measured by ellipsometry and x-ray reflectivity, although the deposition was anodic [54].
Also, cathodically-produced CeO2 nanotubes in the aligned pores of an anodic alumina
template showed small cracks and holes by TEM and SEM, particularly in areas furthest
from the working electrode [49]. In the case of highly-cracked coatings deposited over
lengthy times, there will be even easier access to the metal|electrolyte solution
interface—this allows the observed growth up to and even beyond 20 µm thick, although
there will be a reasonable thickness limitation before spallation occurs.
Building this physical deposition picture up to the nano-/microscale gives
explanation to the clear morphological difference seen in the HSA coatings and thin film
microstructures, as well as trends in the Sm-doped and Sm-doped + H2O2 electrolyte
solutions. The former is related to the rate of base electrogeneration, which is dictated by
the applied potential—faster rates (more negative potentials) encourage the HSA
morphology (Fig. 3.9 and 3.8) and slower rates (less negative potentials) encourage thin
film growth (Fig. 3.13 and 3.15). Regardless of applied potential, the Sm-doped
electrolyte solution promotes sharper featured growth, whereas the Sm-doped + H2O2
electrolyte solution generally promotes more spherical, globular growth. This is related to
the difference in precipitating species for the two electrolyte solutions—Ce(OH)3 and

Ce(IV) species, respectively. Hydrogen bonding between as-produced Ce(OH)3
molecules is thought to emphasize elongated, needle-like growth, in stark contrast to the
spherical growth of the CeO2 phase [53, 77, 95]. This could explain the origin of the HSA
morphology, and why the Sm-doped + H2O2 electrolyte solution has a more difficult time
producing comparatively fine nanoscale features at standard HSA working potentials. A
counter theory for the HSA morphology asserts that hydrogen evolution during the
deposition acts as a dynamic template [83]. However, at HSA depositing potentials (-0.7
to -1.0 V vs. SCE), a non-trivial amount of hydrogen evolution is visually apparent for
the platinum-based substrates, but none is observed for the nickel-based substrates. This
suggests that hydrogen evolution is not the origin of the HSA microstructure, as deposits
from both platinum and nickel exhibit the same general features.

3.4.3

Deposition on Non-Conducting Parts of the Substrate

Up to this point, it has been shown that CELD consistently produces undoped and Smdoped CeO2 in a predictable fashion; that both HSA and thin film morphologies are stable
at high temperatures; and that the wide parameter space allows for CELD microstructural
tunability. In surveying the fabrication non-negotiables of Section 1.4.1, one remains
aloof—maintaining continuous pathways for all mobile species. The oxygen ion pathway
is of particular concern for CELD, which no doubt requires an electronically conducting
surface, even if it is only one part of a composite metal/metal oxide substrate (like the
anti-dot substrates of Chapter 2). In fact, one might be tempted to consider CELD a
bottom-up approach, where growth begins from the electronically conducting surface and
burgeons outward. This would make forming oxygen ion pathways difficult, as there

would be no inherent means for establishing interfaces between the ceria deposit and the
oxide portion of the substrate. Top-down approaches like CVD or PLD are not plagued
with such a concern, as their depositions are fairly substrate-material-independent—other
issues such as substrate temperature and line-of-sight positioning determine whether
deposition occurs. What is demonstrated below, however, is that unambiguous and
significant ceria deposition occurs via CELD on non-conducting and conducting parts of
composite substrates. This phenomenon can be understood in light of the two-step CELD
mechanism, discussed in Sections 3.4.1 and 3.4.2: electrogeneration of base is followed
by chemical precipitation.
To assist the above explanation, it is helpful to draw a distinction between the
(linguistically similar) cathodic electrodeposition of metals and cathodic electrochemical
deposition of oxides. Indeed, classical cathodic electrodeposition of metals on conducting
substrates gives misleading insight into the CELD process described in this manuscript.
In that scheme, a positive metal cation in solution combines with available electrons on
the depositing surface, i.e., the charge-transfer step, reduces its valence to zero, and
becomes solid as a consequence (see Fig. 3.3). Because the deposited metal coating is
itself electronically conducting, this process can continue indefinitely, as electrons are
able to travel to the newly formed deposit|electrolyte solution interface. Once reduced,
the adsorbed metal atom is not typically mobile on a micron scale, and therefore does not
reposition itself to nearby non-conducting surfaces. In contrast, the charge-transfer step in
the CELD of oxides is separate from the deposition step; in fact, the species reduced in
the charge-transfer step are themselves distinct from the metal cation species (compare
Eqns. 3.1-3.7 and 3.8-3.11). Therefore, there is no restriction to the depositing surface in

(a)

(b)

(c)

(d)

Fig. 3.19. SEM images of as-deposited HSA CELD ceria growth on exposed YSZ regions. (a) and (b) show
the coated platinum regions in the top right of the images, as well as evidence of deposition on the YSZ
regions to the bottom left of the images. (c) shows a highly magnified view of a disconnected ceria deposit
surrounded by the extremely smooth YSZ surface. (d) shows deposition proceeding distinctly from the
exposed YSZ region inside of a pore of the nickel anti-dot network. All depositions are performed at 0.8
mA cm-2 for 5 minutes in either the undoped (a-c) or doped (d) electrolytes.

CELD like there is in metal electrodeposition. On that note, there is strong evidence that
preference is initially given to the oxide component of a composite substrate. Beyond the
visual evidence given in this chapter, the strong electrochemical activity characteristics
covered in Chapter 4 alleviate any lingering concerns regarding continuous mobile
species’ pathways.
To reference the specific composite substrates used here, the ceria coating is not
restricted to the platinum strips or the nickel anti-dot networks, but can also form on the
adjacent, non-conducting YSZ surfaces. Fig. 3.19a and b show a magnified and zoomedout view, respectively, of a HSA coating deposited on a platinum strip/YSZ substrate,

where the platinum region is toward the top-right and the exposed YSZ is toward the
bottom left in both images. Deposition is clearly observed on areas that are adjacent to,
but not in direct contact with, the electronically conducting platinum, even as far away as
10 µm. It should be stressed that YSZ is completely electronically insulating at room
temperature, so no pathway for electrons exists in these regions. A highly magnified view
from the same sample shows an entirely disconnected island of ceria deposited on the
exposed YSZ surface (Fig. 3.19c), which definitively debunks the notion that growth
initiates on the metal surface and then proceeds outward to the YSZ regions. Fig. 3.19d
shows deposit growth distinctly proceeding from the pores of the nickel anti-dot
substrate, again, where the YSZ is exposed to the electrolyte solution.
Both the HSA and thin film morphologies can be deposited on the YSZ surface.
Fig. 3.20 shows cascading images of a YSZ surface exhibiting clear, ubiquitous thin film
deposition. The roughness seen up close in Fig. 3.20a, b, and c is not the original YSZ
surface, as the single-crystal substrates are received highly polished to sub-nanometer
roughness, as measured by AFM (not shown). It should be mentioned that the sample
imaged in Fig. 3.20 was deposited at 0.8 mA cm-2, which is typically associated with the
HSA microstructure, but the cerium concentration in the undoped electrolyte solution was
0.01 M, below the usual 0.05 M. This explains the planar growth. Fig. 3.21 shows the
HSA microstructure effectively grown on the YSZ spacing in between platinum strips,
with practically no distinction between the coating’s surface features over the platinum

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 3.20. SEM images from a single sample exhibiting as-deposited planar CELD ceria growth on a
YSZ/platinum strip substrate. Deposition was performed with the undoped electrolyte, but with a lower
(0.01 M) cerium nitrate concentration, and at 0.8 mA cm-2 for 5 minutes. Although this current density is
typically associated with HSA structures, the lower cerium concentration leads to planar growth in this
case. The flat ceria deposit on the YSZ regions appears to extend microns away from the platinum strips.

(a)

(b)

(c)

Fig. 3.21. SEM images from a single sample exhibiting as-deposited HSA CELD ceria growth on a
YSZ/platinum strip substrate. Deposition was performed with the doped electrolyte at 0.8 mA cm-2 for 10
total minutes. Effective HSA growth on the YSZ regions is clearly seen.

and that over the YSZ. Furthermore, Fig. 3.21 shows quality connectivity between the
deposit grown on the YSZ and that grown on the platinum—this is the crucial feature that
ultimately provides oxygen ion pathways from surface reaction sites on the deposit that is
located on top of the metal regions. Somewhat surprisingly, the thin films are also able to
establish such a linkage, as shown in Figure 3.22.
TEM cross-sectional images from a Sm-doped and annealed HSA sample
highlight the expected intersecting nano-sheet and needle-like growth on both metal and
YSZ areas (Fig. 3.23a); HRTEM images in Fig. 3.23c and 3.23d show qualitatively
similar Pt|CeO2 and YSZ|CeO2 interfaces, with no voids or other horribly distortional
artifacts visible. This suggests that any porosity that existed as-deposited has been

sufficiently removed by annealing at 650 °C for 2 hours in air, not surprising given the
small initial crystallite size [57]. Furthermore, the annealed grain size seen in Fig. 3.23b
and the selected-area electron diffraction pattern (see Appendix B) correspond well with
the XRD results above.
To have what is shown here in significant spatial deposition of ceria onto the metaladjacent YSZ surface, with a well-adhered and void-free interface is an all-but-certain
requirement for facile oxygen ion migration from the YSZ fuel cell electrolyte solution to
the ceria anode surface. Indeed, if the CELD of ceria only coated the metal, or if the
YSZ|CeO2 interface was poor, the conduction pathway for oxygen ions would be either
non-existent or highly resistive.

(a)

(b)

(c)
Fig. 3.22. As-deposited SEM images
from the sample shown in Fig. 3.21,
where connectivity between the thin
film deposit grown on the YSZ regions
and that grown on the platinum surface
is seen.

(a)

YSZ

(b)

CeO2
(c)

CeO2

(d)

YSZ

CeO2

Fig. 3.23. TEM images showing definitive deposition on the exposed YSZ areas as a cross-sectional view
(a); the polycrystalline nature of the deposit with annealed grain sizes ~15 – 20 nm (b); and HRTEM
Pt|CeO2 (c) and YSZ|CeO2 (d) interfaces. This sample’s deposition was performed at 0.8 mA cm-2 for 10
minutes with the doped electrolyte and annealed at 650 °C for 2 hours in air.

3.4.4

HSA and Thin Film Transients

To understand how CELD deposits evolve over time, voltage and current transients of the
HSA and thin film morphologies are shown in Fig. 3.24 and 3.25, respectively. For a
given electrolyte solution composition and concentration, the steady-state current value
depends on the interplay between electrochemical reduction reactions, whose rates
collectively give the current density, and any blockages that cover the metal electrode
surface, which reduce the number of reduction reaction sites available. These blockages
include hydrogen bubbles that persist on the metal surface and any depositing nuclei, as
they, too, are electronically insulating.
Fig. 3.24 shows typical voltage transients for the HSA morphology at an applied
current density of 0.8 mA cm-2 on both platinum strip and nickel anti-dot metal network
configurations. All of the platinum strip substrates exhibit a double-plateau voltage
response. Using the chronological SEM images as guides (Fig. 3.24, I through IV), the
first plateau appears to be mostly related to deposition on the YSZ regions. A more
negative working potential is required to maintain the constant applied current density
once deposition begins to cover the platinum surface (Fig. 3.24, II, III, and IV).
Comparing the voltage response of Sm-doped 5-5µm (Fig. 3.24b) and Sm-doped 1010µm (Fig. 3.24c) patterns shows that the platinum pattern sizes do not affect the
transients much. There is, however, a significant difference between the undoped
electrolyte solution (Fig. 3.24a) and the Sm-doped electrolyte solution (Fig. 3.24b and
3.24c)—the undoped sample has a much shorter dwell time on the first plateau. This
suggests that the samarium doping partially inhibits the precipitation kinetics, requiring

-0.6

doped, Ni anti-dot
-0.7

Voltage / V

(d)

III IV

doped, Pt 10-10 µm

-0.8

doped, Pt 5-5 µm
(a)

undoped, Pt 5-5 µm
-0.9

100

150

200

250

300

Time / s

II
Pt

YSZ

Pt
YSZ

III
YSZ

IV
Pt

YSZ
Pt

YSZ
YSZ

Fig. 3.24. HSA voltage transients taken at 0.8 mA cm-2 for various electrolyte/substrate configurations: (a)
undoped, platinum 5-5 µm; (b) doped, platinum 5-5 µm; (c) doped, platinum 10-10 µm; and (d) doped,
nickel anti-dot. Also, chronological SEM images of doped, platinum 5-5 µm samples taken at the times
indicated by (I, II, III, and IV) on the transient plot.

longer times to deposit on the same YSZ area size. Also, the potential value for the
second plateau in the Sm-doped case is less negative than the undoped case, suggesting
that the Sm-doped deposit blocks the reduction reactions less, and is, therefore, more
porous. Ultimately, ubiquitous deposition occurs on the platinum and YSZ surfaces alike;
however, the SEM evidence in Fig. 3.19, 3.23, and 3.24 indicates that the deposit prefers
the YSZ to the metal surface initially. Judging from the relative amount of deposit seen to
coat the metal after ~5 minutes (Fig. 3.24, IV), it seems that once enough nuclei have
formed on the metal surface, continual deposition on those nuclei are preferred to the
distant YSZ surface nuclei.
The nickel anti-dot metal network substrate voltage transient has a qualitatively
different shape than its platinum strip counterpart. Consider that all HSA working
potentials, regardless of the substrate used, are relatively more negative, where both
oxygen reduction and hydrogen evolution are major contributors to the current density
[86]. However, less hydrogen evolution is observed for nickel-based substrates, meaning
that a more negative initial voltage is needed to induce enough ion motion in the liquid
electrolyte solution to satisfy the applied current density value. Eventually, the voltage
decreases and reaches a steady-state value less negative than that of the platinum strips,
probably because there are less hydrogen bubbles blocking reduction reaction sites.
The current transients for the thin film morphologies are shown in Fig. 3.25,
again, for platinum strips and nickel anti-dot networks. In this working potential region,
oxygen reduction is dominant [86], and can occur via a four-electron pathway (Eqn. 3.1)
or a two-electron pathway (Eqn. 3.2). The four-electron pathway has a higher steady-state
current than the two-electron pathway [82]. For both Sm-doped and Sm-doped + H2O2

Cathodic Current Density / µA cm-2

3000

2500

doped + H2O2, Pt

2000

1500

doped + H2O2, Ni

doped, Pt

1000

500

doped, Ni

Time / s
Fig. 3.25. Thin film current transients taken at -0.55 V vs. SCE for various substrate/electrolyte
configurations. Here, Pt refers to platinum strip networks and Ni refers to nickel anti-dot networks.

electrolyte solutions, the steady state current for platinum substrates is higher than for
nickel substrates, possibly indicating a difference in the oxygen reduction pathway for the
two metals—platinum follows the four-electron pathway for acidic solutions and then at
pH 7 switches to 80% four-electron, 20% two-electron [96]. Comparably definitive
literature could not be found for nickel. Other possibilities are that the film is more dense
on nickel, or there still is a non-trivial hydrogen evolution-related current for platinum
but not nickel, even at these low cathodic potentials.
Higher steady-state current densities are seen in the Sm-doped + H2O2 electrolyte
solution than in the Sm-doped. This can be explained by recalling the Sm-doped + H2O2

CV scan in Fig. 3.18, where an appreciable current related to the reduction of Ce(IV)
intermediates is present. For the Sm-doped and Sm-doped + H2O2 electrolyte solutions,
typical thin film deposition rates are roughly 1.6 nm min-1 and 200 nm min-1,
respectively. These wildly disparate deposition characteristics should be distinguished
from and recognized as unrelated to their current transients; rather, the quickly forming
films with H2O2 addition indicate fast kinetics for the 𝐶𝑒(𝑂𝐻)+2
2 route precipitation (Eqn.
3.9 – 3.11), as compared to the Ce(OH)3 route (Eqn. 3.8).

These time-dependent characteristics underscore the previously discussed notions
that unmistakable deposition occurs on the non-conducting YSZ regions, and that the asdeposited coatings must be porous in order for a non-zero steady-state current to exist.

Chapter 4

The Electrochemical Activity of CELD
Ceria Structures
4.1

Introduction, Methods, and Background

4.1.1

A.C. Impedance Spectroscopy (ACIS) Introduction

Any good device design philosophy seeks to identify the area(s) of worst performance
and address the appropriate issues. In this way, maximum gains can be efficiently
accomplished. As fuel cells involve electrochemical reactions and current flow, both
inherently rate-related, the primary inhibiting process is referred to as the “rate-limiting
step.”
A.C. Impedance Spectroscopy (ACIS) is an invaluable tool for this effort, as it is
able to separate relevant processes, such as oxygen ion conduction and electrochemical
surface reaction rates, as well as elucidate their associated impedances. Processes are
distinguished by probing in the frequency domain, where differing characteristic
relaxation frequencies are expected for each process. In brief, a small voltage
perturbation is applied to a cell, and the phase-shifted current response is recorded, from
which the impedance can be ascertained. A detailed treatment of ACIS as an
electrochemical analysis tool can be found in references [94, 97]. For the purposes of this
chapter, three basic impedance responses need to be known. As an aid, the complex
impedances (Z) of a resistor and a capacitor are given below, where R is resistance, C is
capacitance, ω is frequency, and j is √−1.

pH2 = 0.04 atm, pH2O = 0.005 atm, 650 °C

-Im Z / Ω cm2

0.09 Hz

Re Z / Ω cm2

Fig. 4.1. A representative Nyquist plot for a PLD film of SDC deposited on single crystal YSZ, on top of which is laid
5-35 µm Pt strip patterns. The equivalent circuit for such a spectra is also given. These data and the equivalent circuit
are taken with permission from [1].

𝑍𝑟𝑒𝑠𝑖𝑠𝑡𝑜𝑟 = 𝑅
𝑍𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑜𝑟 =

(4.1)

𝑗𝜔𝐶

(4.2)

Most often, the complex impedances obtained via ACIS are represented as

Nyquist plots, where the positive real component is plotted on the x-axis, and the
negative imaginary component is plotted on the y-axis. These plots are parametric with
frequency, where data points on the right hand side of the plots are the lower frequencies,
and those on the left hand side are higher frequencies. For a purely resistive process, with
no associated capacitance (or, with a capacitance that cannot be resolved within
experimental limitations), there will be no imaginary component, so the impedance will
simply be a point on the x-axis (Eqn. 4.1). For a non-diffusion-related, resistive process
with an associated capacitance, a semi-circular, symmetric arc manifests in the Nyquist

plot, owing to the frequency dependence of the capacitive impedance (see Eqn. 4.2). This
case is shown in Fig. 4.1 for a YSZ substrate with SDC layers on either side of it,
deposited by PLD, and on top of which is a 5-35 µm Pt strip pattern. These data are taken
with permission from [1]. Each arc that manifests indicates one resistive process; hence,
in Fig. 4.1, there is exactly one process that is probed. The resistance of such a process is
easily extracted from the Nyquist representation of the complex impedance—it is simply
the breadth, or diameter, of the arc on the real (x) axis, or ~16 Ω cm2 for the process in
Fig. 4.1. This “ohmic offset” is subtracted for ease of comparison for all subsequent plots.
Also note that the arc is offset from the origin—this means that the resistive process is in
series with a simple resistor. Its origin is the electronic resistance in the wires connecting
the cell to the voltage supply, as well as the ohmic resistance of the oxygen ions in the
supporting YSZ substrate. Finally, for a diffusion-related, resistive process, a half teardrop shaped arc manifests, with a near 45° angled feature at higher frequencies.
Equivalent circuits (with R and C elements, among others) are used to represent
experimentally measured spectra. As an example, the physically-derived equivalent
circuit for the spectra in Fig. 4.1 is shown as an inset. This derivation can be quite
complicated, however, these equivalent circuits are not unique, meaning there are
essentially an infinite number of equivalent circuits that accurately map to a given
measured spectrum. To alleviate (but not eliminate, necessarily) this concern, the work in
this chapter employs a physically-derived model to establish the equivalent circuit (see
ref [26]).
The strategy used here is to correlate the processes measured via ACIS to the
physical geometry of the cell. The reason this is useful is because ACIS probes the most

resistive process for the least resistive serial pathway—this allows the rate-limiting step
to be identified. This is done by probing the evolution (or lack, thereof) of ACIS spectra
with surrounding atmospheric partial pressure changes and geometric changes. An
example of altering the cell’s geometry is as follows: if it appears that the rate-limiting
process is related to the migration of oxygen ions from point A to point B, then the
physical distance from A to B could be doubled, and the impedance response measured
again to see if it follows suit. Once a robust correlation is established, systematic
architectural changes can be made to maximize electrochemical activity. In addition, the
response of the CELD coatings will be compared with that of PLD films deposited on
identical substrates.
A quick note on notation is necessary. Electrode impedances (or resistances) that
have been normalized by the total deposited area are given as 𝑍� (or 𝑅�), whereas

impedances (or resistances) that have been normalized by the projected area of the
exposed SDC surface are given as 𝑍� ∗ (or 𝑅�∗ ). Accordingly, a resistance value extracted

from the Nyquist plots is referred to as an ASR, or area-specific resistance.

4.1.2

Experimental Approach

This chapter is concerned with the evaluation of the ceria-based, template-free HSA
microstructures discussed in Chapter 3 as a suitable anode candidate, shown
schematically in Fig. 4.2d. To do so, symmetric cells are constructed, where identical
electrode configurations exist on both sides of a single-crystal YSZ supporting substrate.
The first two porous metal networks mentioned in Chapter 3, namely platinum
strips and nickel anti-dot films, are used as conducting substrates during CELD and

current collectors during ACIS probing. Recall that the 3PB, 2PB, and metal/YSZ areas
are well-defined for both metal networks. The reader is referred to Chapter 3 for detailed
characteristics of these current collectors, as the ones used here are identical.
Three different cell configurations are examined, listed here with increasing
complexity.

Metal-exposed configuration: this configuration is the model configuration
explored in [1], where a PLD SDC layer is situated underneath Pt patterns.
Both the metal network and the PLD SDC under-layer are fractionally parts
of the total exposed surface area, as in Fig. 4.2a.

Metal-embedded configuration: PLD/CELD coatings are overlaid onto
metal networks on YSZ as in Fig. 4.2b and c. Fig. 4.3 shows SEM images
comparing a PLD top-layer (4.3ab) and a CELD top-layer (4.3cd).

Metal-sandwich configuration: first, a dense, 1 µm thick PLD layer is
deposited onto a bare YSZ substrate; second, the porous, metal network is
laid down as before; and third, PLD/CELD top-layer coatings are overlaid
onto both the metal and exposed PLD-SDC, as in Fig. 4.2d and e.

The CELD HSA coatings are deposited with the Sm-doped electrolyte solution and at 0.8
mA cm-2, a la Chapter 3. The PLD films are deposited at 300 mJ and 5 mtorr pO2, at a
substrate temperature of ~650 °C. These samples were provided by Dr. William Chueh
and Dr. Yong Hao, and the details of the PLD film deposition procedure can be found
elsewhere [1, 31].
Once made, the cells are evaluated in a symmetric gas configuration, meaning the
same atmospheric conditions (e.g., flow rate, partial pressures) are experienced on both

89
H2

(a)

H2O

metal

2eSDC

O2-

YSZ

(c)

(b)
metal

2e-

PLD SDC

2e-

YSZ

(d)
metal

PLD SDC

2e2e-

metal

2e-

CELD SDC

2e-

YSZ

(e)
metal

1st PLD SDC

CELD SDC

2e2e-

1st PLD SDC

YSZ

Fig. 4.2. Schematics showing surface reaction locations for two-phase boundary (2PB), mixed ionicelectronic conducting substrates (a); PLD (b) and CELD (c) embedding Pt strips; and PLD (d) and CELD
(e) sandwich configurations. In (b) and (c) electronic conduction across the deposited layer that lies on top
of the metal is inhibited and the field lines are accordingly confined to the nominal 3PB region; in (d) and
(e), the same inhibition exists as in (b) and (c), but the field lines are free to readjust themselves to
accommodate the newly available underside of the metal. The left column is PLD films, and the right
column is CELD coatings. The rows indicate identical configurations.

sides of the YSZ substrate. The primary reason for this is convenience—no sealing is
required, and the gas delivery system is simpler.
The electrochemical characterization system consisted of a vertical furnace tube
reactor system, through which gas was continually flowed at a total flow rate of ~101
sccm, controlled by MKS PR 400 controllers and MKS mass-flo controllers. Three gas
lines were used to achieve specific hydrogen and water partial pressures: a dry, pure
hydrogen line; a dry, pure argon line; and a humidified 0.1% hydrogen in argon line.
Humidification was achieved by passing the 0.1% H2 in Ar line through a variable

(a)

(b)
polycrystalline

single crystal

PLD SDC

thin Pt strip

(c)

PLD SDC

YSZ

(d)

Fig. 4.3. SEM images showing PLD top-layer (a and b) and CELD top-layer (c and d) metal-embedded
configurations. Depositions are performed at the standard conditions given in the corresponding text.

temperature bubbler. Impedance data were collected using a Solartron 1260A frequency
response analyzer at zero bias with a 20 mV perturbation amplitude via an in-house
Labview program. Four platinum wires were used to minimize inductance loops.
Platinum was chosen for its inertness in oxidative/reductive atmospheres.

4.1.3

System Precedence

Previous detailed work has been reported on an analogous, symmetric cell configuration
consisting of PLD thin film SDC on YSZ substrates overlaid with lithographically
patterned metal current collectors, as in Fig. 4.2a [1, 26, 31-32].
In those experiments, only one semi-circular arc was manifested (see Fig. 4.1),
and was unambiguously determined to be related to the SDC|gas interface. This

necessarily means that for this model system, the surface reaction that occurs at the
SDC|gas interface is rate-limiting. This simple system is used to provide insight into
interpretation of the following data obtained for more complex systems, namely those
given schematically in Fig. 4.2.

4.2

Arc Identification: PLD Films vs. CELD Coatings

4.2.1

Representative Spectra

A representative sampling of ACIS spectra taken from metal-embedded cells (Fig. 4.2b
and c) under identical conditions is shown in the Nyquist plot of Fig. 4.4. The three cells
shown are PLD embedding platinum strips (open circles), CELD embedding platinum
strips (open triangles), and CELD embedding a nickel anti-dot film (open squares). Each
sample was probed in different gas environments, an example of which is shown for the
CELD embedding a nickel anti-dot film in Fig. 4.5. As indicated by the notation, the
impedance is normalized by the total deposited area, irrespective of the metal network
geometry. Significantly, all spectra of the embedded metal geometry exhibited two
similar arcs, even though the SDC deposition techniques are different, and, in the CELD
case, different metal networks are used. Recall that frequency is swept from low to high,
which is right to left in the Nyquist plots; accordingly, the arc on the right-hand side is
referred to as the LF (low frequency) arc, and the arc on the left is the HF (high
frequency) arc. From these spectra, it is evident that PLD films have larger LF arcs than
the CELD coatings. Also, the HF arc is similarly sized for the PLD/Pt strip and CELD/Pt
strip samples, but is slightly smaller for the CELD/Ni anti-dot sample.

(a)

-Im Z / Ω cm2

CHF

pH2 = 0.04 atm, pH2O = 0.005 atm, 650 °C
PLD/Pt strips
CELD/Pt strips
CELD/Ni anti-dot

RHF

RLF

0.19 Hz

385 Hz

Re Z / Ω cm2

(b)

0.8

0.4
0.3
3.44 Hz

0.2
0.1
0.0

3.8

4.0

4.2

Re Z / Ω cm2

4.4

-Im Z / Ω cm2

CLF

0.6
0.4

3.44 Hz

0.2
0.0
2.2

2.4

2.6

2.8

3.0

Re Z / Ω cm2

3.2

3.4

Fig. 4.4. (a) Representative Nyquist plots exhibiting the two arc behavior for PLD embedding 5-5 µm Pt
strips (open circles), CELD embedding 5-5 µm Pt strips (open triangles), and HSA CELD embedding a Ni
anti-dot network with 1.4 µm pores (open squares); (b) a magnified view of the CELD/Pt strips’ LF arc; (c)
a magnified view of the CELD/Ni anti-dot network’s LF arc. Solid lines are the results from the fits to the
equivalent circuit shown as an inset in (a); and the dotted lines are simulations of the arcs as a guide to the
eyes. The normalization is the entire deposited area for all three cases here.

(a)

1.5
1.0

45% H2
29% H2
13% H2
4% H2

pH2O = 0.002 atm, 650 °C

0.5
0.0

(c)

Re Z / Ω cm2

-Im ~
Z / Ω cm2

0.75
0.50
0.25

0.00
5.75 6.00 6.25 6.50 6.75 7.00 7.25 7.50

Re Z / Ω cm2

(b)

1.5

0.114% H2O
0.066% H2O
0.05% H2O
0.021% H2O

pH2 = 0.04 atm, 650 °C

1.0
0.5
0.0

(d)
~ / Ω cm2
-Im Z

1.00

-Im ~
Z / Ω cm2

-Im Z / Ω cm2

Re Z / Ω cm2

1.00
0.75
0.50
0.25
0.00
5.75 6.00 6.25 6.50 6.75 7.00 7.25 7.50

Re ~
Z / Ω cm2

Fig. 4.5. Representative hydrogen (a) and water (b) partial pressure dependence of a Ni anti-dot-embedded
CELD sample. The anti-dot pores are 1.4 µm and the CELD is deposited at the HSA conditions. The HF
arc is relatively static with partial pressure changes, whereas the LF arc strongly depends on the gas
atmosphere. The normalization is the entire deposited area. The plots in (c) and (d) are magnified views of
the LF arcs in (a) and (b), respectively.

Analogously, a representative sampling of ACIS spectra taken from metalsandwich cells under identical conditions is shown in the Nyquist plot of Fig. 4.6, again
with the resistance normalized by the total deposited area. The two samples shown here
are PLD sandwiching platinum strips (c.f. Fig. 4.2d; seen in Fig. 4.6 as open squares),
and CELD sandwiching platinum strips (c.f. Fig. 4.2e; seen in Fig. 4.6 as open triangles).
Only one arc can be seen for both PLD- and CELD-platinum sandwich samples, which
manifests at lower frequencies. Although the sandwich configuration has a more complex
fabrication, the response is simpler, so it is considered first in the analysis below.

8 pH2 = 0.04 atm, pH2O = 0.005 atm, 650 °C
0.09 Hz

-Im Z / Ω cm2

PLD/Pt strips sandwich
CELD/Pt strips sandwich

Re Z / Ω cm2

Fig. 4.6. Representative Nyquist plot of a PLD/Pt strips sandwich (open squares); and a CELD/Pt strips
sandwich (open triangles). In both cases, there is only one arc. Solid lines are the results from the fits to the
equivalent circuit from Fig. 4.1. The normalization is the entire deposited area.

4.2.2

Origin of the Single Arc in the Metal-Sandwich Configuration

Recall that the sandwich configuration for both PLD and CELD samples yields a single
impedance arc (c.f. Fig. 4.6). Also recall that the model system of a SDC PLD film with
Pt strips exposed yields a single arc (c.f. Fig. 4.1). The hydrogen gas dependence of the
resistance values from these three samples are compared in Fig. 4.7—namely, the PLD/Pt
strips-exposed configuration (black squares), the PLD/Pt strips-sandwich configuration
(red circles), and the CELD/Pt strips-sandwich configuration (blue triangles).

100

pH2O = 0.005 atm, 650 °C

R* / Ω cm2

0.1

PLD/Pt strips exposed
PLD/Pt strips sandwich
CELD/Pt strips sandwich
0.01

0.1

H2 Content / atm

Fig. 4.7. Hydrogen partial pressure dependence of the solitary arc resistance values for PLD/Pt strips
exposed (black squares), PLD/Pt strips sandwich (red circles), and CELD/Pt strips sandwich (blue
triangles) configurations. The normalization is the projected area of the exposed ceria surface. Solid lines
are guides to the eyes.

Both the absolute values and gas dependence of the ASRs for each sample are
comparable. This strongly suggests that the single arc in the model PLD/Pt strips exposed
system is the same as the single arc in the PLD and CELD/Pt strips sandwich
configurations. Consequently, it can be concluded that this arc is due to the interface
between the SDC surface and the surrounding gas, for each sample compared in Fig. 4.7.
One notable peculiarity remains, however. When CELD is used as the top coating
in a platinum-sandwich configuration, the absolute ASR values of the SDC|gas arc do not
decrease as would be expected from the higher surface area accessed by the HSA CELD
coating. Upon ex situ SEM analysis of the CELD/Pt sandwich sample, it can be seen that
the CELD coating on the exposed SDC only slightly enhances the surface area, in

(a)

(b)

(c)

Fig. 4.8. SEM images of a CELD/Pt strip-sandwich configuration after testing at 650 °C for ~24 hours: (a)
the top-layer CELD does not significantly enhance the surface area over the exposed PLD SDC regions,
where some cracking is observed; (b) highly angled view of the phenomena in (a); and (c) the deposit lying
on top of the Pt strips appears disconnected from the deposit on the exposed SDC regions, and can be seen
here uncovering the Pt strip altogether.

contrast with CELD coatings on YSZ (compare Figure 4.8ab to Figure 4.2cd). Also seen
in Figure 4.8, the part of the CELD coating that lies on top of the metal network appears
disconnected from the coating on top of the SDC. This is likely due to the volume
reduction that inevitably happens when electrochemically deposited oxide material is
annealed—this adverse effect is exacerbated for thicker samples like the one shown here.
The CELD deposits on Pt/YSZ surfaces also exhibit this behavior, but restricted to a
small area in the vicinity of the liquid electrolyte meniscus. A break of this fashion is
tantamount to completely nullifying the activity of the SDC above the metal, as no
continuous pathway for oxygen ions exists. Furthermore, non-trivial cracking in the first

PLD layer is observed when the second layer is deposited via CELD, but not PLD. This
could be due to thin film stresses induced by a slight lattice expansion, arising from the
reduction of Ce4+ to Ce3+ in the underlying SDC PLD layer during the cathodic
electrochemical deposition. These factors undoubtedly affect the measured ASR for
CELD/metal-embedded samples.

4.2.3

Origin of the HF Arc in Embedded Metal Configurations

Now consider only the HF arc for the metal-embedded configurations shown in Fig. 4.4.
Analogously to the previous section, the hydrogen partial pressure dependence of the HF
ASR values extracted from the Nyquist plots is shown in Fig. 4.9 for the same three
samples from Fig. 4.4—namely, PLD/Pt strips-embedded (black open squares), CELD/Pt
strips-embedded (blue open triangles), and CELD/Ni anti-dot-embedded (red open
circles) configurations. Note the nearly flat response of the metal-embedded
configurations’ HF ASRs to changing hydrogen partial pressure. This is true regardless of
deposition technique or the metal network underneath. The water dependence is similarly
weak (not shown). The apparent lack of partial pressure and fabrication technique
dependence of the HF arc suggests a configurational origin, one that is shared between
PLD and CELD samples. To further investigate the origin of the HF arc, embedded Pt
strip samples of large pattern sizes are analyzed.
Four exotic platinum patterns are utilized to tease out robust dependencies related
to the HF arc—50-100 µm, 50-500 µm, 50-900 µm, and 50-1300 µm. When the metal
spacing is increased beyond 100 µm, the HF ASR values show appreciable
hydrogen/water partial pressure dependence, eventually exhibiting equal but opposite

100

pH2O = 0.005 atm, 650 °C

~ / Ω cm2
R*

0.1

PLD/Pt strips-embedded HF
CELD/Ni anti-dot-embedded HF
CELD/Pt strips-embedded HF

0.01

0.1

H2 Content / atm

Fig. 4.9. Hydrogen partial pressure dependence of the HF arc resistance values for PLD/Pt strips-embedded
(black open squares), CELD/Pt strips-embedded (blue open triangles), and CELD/Ni anti-dot-embedded
(red open circles) configurations. The normalization is the projected area of the exposed ceria surface.
Solid lines are guides to the eyes.

slopes in the log-log plots of Fig. 4.10a and b. This suggests that the HF arc is an
electron-related process, as the electronic carrier concentration is oxygen partial pressure
dependent [27, 31, 94]. For the same pattern sizes, the ASRs also show a strong
relationship to the nominal 3PB length and lateral metal spacing (Fig. 4.10c and d). These
dependencies, like the oxygen partial pressure dependence, decrease in intensity as the
pattern sizes are made smaller. Also of note is the fact that these HF arcs are decidedly
semi-circular, as opposed to half tear-drop shaped, suggesting that neither oxygen ion nor
electron diffusion are rate limiting, even on length scales approaching millimeters. These
data imply the SDC-metal interface as the origin of the HF arc.

+0.11
+0.12

~ / Ω cm2
R*

+0.12

+0.07

1E-3

-0.12
-0.12

-0.13
-0.09

50-100 µm
50-500 µm
50-900 µm
50-1300 µm

(b)
R* / Ω cm2

(a)

0.01

50-100 µm
50-500 µm
50-900 µm
50-1300 µm

0.01

H2 Content / atm

H2O Content / atm

(c)

0.1

(d)

50-1300 µm

50-900 µm
50-500 µm

50-100 µm
0.57% H2O
0.29% H2O
0.15% H2O
0.08% H2O

0.1

3PB / m cm-2

+0.85

50-100 µm
13% H2
4% H2
1.2% H2
0.4% H2

13% H2
4% H2
1.2% H2
0.4% H2

50-500 µm

-1.0

R* / Ω cm2

~ / Ω cm2
R*

50-900 µm

100

0.57% H2O
0.29% H2O
0.15% H2O
0.08% H2O

1000

Metal Spacing / µm

Fig. 4.10. HF resistance water (a) and hydrogen (b) partial pressure dependence of PLD embedding Pt
strips of large pattern sizes taken at 650 °C; deviation from an explicit oxygen partial pressure dependence
can be seen in the smaller patterns. HF resistance three-phase boundary (c) and metal spacing width (d)
dependencies are shown for the same samples as in (a) and (b); again, deviation from the strong
dependencies can be seen for the smaller pattern sizes. Solid lines are guides to the eyes.

100

Consider the SEM image in Fig. 4.3b and the corresponding schematic in Fig.
4.2b. SDC deposited via PLD on top of single-crystal YSZ produces epitaxial, singlecrystal growth [1, 31]. However, SDC deposited via PLD on top of the platinum strips is
polycrystalline and columnar. The columns are oriented perpendicular to the platinum
surface, which would cause in-plane electronic migration to be highly resistive. This
forces the electron migration paths from a surface reaction site that lies above YSZ,
where they are generated, to the nominal 3PB region (see Fig. 4.2b). The coalescence of
the field lines to the sides of a 200 nm thick platinum strip manifests as an additional arc
at higher frequencies. Electrons that are generated directly above the metal are not
restricted in this way, since they only have to travel along the columns’ length to access
the metal.
CELD samples also exhibit a HF arc, indicating that a similar restriction occurs,
albeit for a different reason. Figure 4.11a shows a TEM image of a CELD deposit on top
of a platinum surface, which has been annealed at 650 °C for 2 hours in air. Across the
entire metal surface, a layer approximately 10 nm thick of dense SDC can be seen. Above
this initial layer, sheets and needles of SDC randomly intersect with one another on both
the metal and YSZ areas. However, gaseous hydrogen evolution occurs on the metal
surface during CELD, making the deposit more porous on the metal than on the YSZ (c.f.
Chapter 3). This is schematically shown in Figure 4.2c and confirmed by TEM
imaging—a partial view of the inevitable voids above the metal area can be seen in a
matrix of SDC in Figure 4.11a. Similar to the PLD case, lateral electronic migration is
particularly resistive through the SDC that lies on the metal. An analogous restricting

101

(a)

(b)

SDC
Pt

void

3PB
region

Fig. 4.11. TEM images of an annealed (650 °C in ambient air for 2 hours) CELD SDC deposit on a Pt
surface, showing the continuous 10 nm thick layer, off of which tortuous nanosheets/needles grow, but with
significant voids, particularly above the Pt surface (a) and in the 3PB regions (b). The slight texturing in the
SDC deposit is FIB damage incurred during the lift-out process. Images obtained with assistance from
Carol Garland, Caltech.

effect occurs for the electronic field lines in the CELD deposits, producing the familiar
HF arc. This effect is possibly further exacerbated by an incomplete CELD coating,
particularly in the immediate region surrounding the 3PB. Figure 4.9b shows such a void,
whose effect would be to reduce the accessible SDC|metal interface even more.
Consistent with this picture, the absolute value difference of the HF ASRs for the
CELD/Pt strips-embedded and CELD/Ni anti-dot-embedded samples (see Fig. 4.9) can
be explained by their respective metal network thickness difference. The Pt strips are 200
nm thick, whereas the Ni anti-dot network is 400 nm thick.
This electronic pathway restriction theory is confirmed by the metal-sandwich
configuration impedance response, namely that the deleterious HF arc disappears and the
remaining, lone arc is identical to the single arc of the PLD/Pt strips exposed
configuration. The disappearance of the HF arc can be explained as follows. Although the
electronic migration through the SDC above the metal areas for both the PLD and CELD

102

top layers is highly resistive, the presence of a PLD SDC under-layer allows the
electronic field lines to be redistributed to the entirety of the underside of the metal,
shown schematically in Fig. 4.2d and e.

4.2.4

Origin of the LF Arc in Embedded Metal Configurations

Consider now the LF ASRs of the metal-embedded configurations (c.f. Fig. 4.2c and d),
whose hydrogen partial pressure dependencies are shown in Fig. 4.12—specifically, the
PLD/Pt strips-embedded (black squares), the CELD/Pt strips-embedded (blue triangles),
and the CELD/Ni anti-dot-embedded (red circles) configurations are compared. Also
included is the data for the single arc in the PLD/Pt strips exposed model system (green
upside-down triangles). Comparing the PLD/Pt strips exposed data to the PLD/Pt stripsembedded data, the absolute values and dependencies are almost identical. According to
[1], this is within an expected level of variation for the same surface-dominated process.
From this comparison, a confident connection can be made between the single arc in
metal-exposed configurations and the LF arc in metal-embedded configurations—their
origins are from the same surface-dominated, SDC|gas interfacial resistive process.
The context is now complete for understanding the LF arc behavior of
CELD/metal-embedded samples. Using Fig. 4.12 as a guide, it can be seen that both the
CELD/Pt strips-embedded and CELD/Ni anti-dot-embedded samples exhibit similar
hydrogen gas dependencies as the PLD/Pt strips-embedded sample. However, their
absolute values are 25 – 50x smaller. Taking the gas dependence similarity to be an
indication that the LF arc for the CELD/metal-embedded samples is also surface-related,
a simple surface area argument can explain the absolute value difference between these

103

100

pH2O = 0.005 atm, 650 °C

R* / Ω cm2

0.1

PLD/Pt strips-embedded LF
CELD/Pt strips-embedded LF
CELD/Ni anti-dot-embedded LF
PLD/Pt strips exposed single arc

0.01

0.1

H2 Content / atm

Fig. 4.12. Hydrogen partial pressure dependence of the LF arc resistance values for PLD/Pt stripsembedded (black squares), CELD/Pt strips-embedded (blue triangles), and CELD/Ni anti-dotembedded (red circles) configurations. Also for comparison, the single arc gas dependence for the
PLD/Pt strips exposed configuration is shown (green upside-down triangles). The normalization is the
projected area of the exposed ceria surface. Solid lines are guides to the eyes.

samples. Indeed, Fig. 4.3a and c show a surface area increase on that order obtained per
projected area when using CELD as the top coating. Despite the overall microstructural
similarity in the CELD deposits, there is some sample-to-sample surface area variation,
which manifests itself as the difference in ASRs between the two CELD samples. These
observations establish the LF arc in CELD/metal-embedded configurations to be the
SDC|gas interface.

104

4.3

The SDC|Gas Interface Arc: A Closer Look

The following is a brief summary of the results from Section 4.2. It was determined that
the single arc that manifested for the metal-sandwich configurations and the LF arc for
the metal-embedded configurations are both due to the SDC|gas interface. Hereafter, this
arc is referred to as the “SDC|gas interface arc.” This arc is scrutinized in greater detail in
this section, with the goal of reducing its associated ASR. Because of the undesirable
ASR increase that occurs when utilizing the CELD/metal-sandwich configuration, this
section utilizes the CELD/metal-embedded configuration, despite the presence of the HF
arc. The SDC|gas interface arc is taken to accurately represent the potential for CELD as
a fabrication method for producing highly active electrode structures, and, as such, is the
only arc examined from here on out. The rest of the chapter is organized by the metal
network that is used.

4.3.1

Platinum Strips

Four experimental parameters are varied to ascertain their effect on the SDC|gas
interfacial ASR. For each, the gas dependence of the ASR is evaluated, and an attempt to
correlate the observed behavior to the deposit geometry via SEM is made. The impedance
is normalized by the projected area of the exposed SDC surface.

1. Platinum pattern size effect: 5-5 µm vs. 20-20 µm.
Undoped HSA ceria is deposited on two platinum pattern sizes, 5-5 µm and 20-20 µm.
Their interfacial reaction resistance gas dependencies are shown in Fig. 4.13, where it can
be seen that the 5-5 µm pattern ASR absolute values are 2 – 3x smaller than their 20-20

105

µm counterpart. Examining the SEM images in Fig. 4.14 reveals a stark difference
between the total deposition coverage for the two pattern sizes. Both patterns have
qualitatively similar deposition on the platinum strip portion of the substrate, but the 2020 µm pattern only has disconnected, island-like growth on the exposed YSZ portion of
the substrate. In contrast, the 5-5 µm pattern has well-connected, HSA growth on
seemingly all portions of the substrate, platinum and YSZ alike. Despite the fact that both
patterns have nominally 50% platinum and 50% YSZ exposed surfaces, the lack of
quality deposition on the YSZ portions lying greater than a few microns away from the
base generating metal portions of the substrate causes a decrease in performance. This
defines a lateral metal spacing limitation; consequently, only patterns with feature sizes
equivalent to (or below, for the nickel anti-dot films) 5-5 µm are subsequently used.

106

(b)
pH2 = 0.45 atm, 650 °C

0.1
1E-3

pH2O = 0.005 atm, 650 °C

R* / Ω cm2

(a)

5-5 µm
20-20 µm

0.01

0.1

5-5 µm
20-20 µm

0.1

H2 Content / atm

H2O Content / atm

Fig. 4.13. Water (a) and hydrogen (b) partial pressure dependencies of the SDC|gas interfacial ASR for 5-5
µm (squares) and 20-20 µm (circles) CELD/Pt strips-embedded samples deposited for 10 minutes with the
undoped electrolyte.

(a)

(b)

(c)

(d)

Fig. 4.14. SEM images of the samples probed in Figure 4.13: (a) and (b) are top-down views of the 20-20
µm sample; (c) and (d) are angled views of the 5-5 µm sample. The inset in (b) is an isolated HSA ceria
growth in the center of a 20 µm wide YSZ region, pictured in low magnification in the rest of (b).

107

2. CELD deposition time effect: 5 vs. 10 (vs. 20) min.
Both CELD undoped and Sm-doped HSA ceria microstructures are evaluated at different
deposition times—5 and 10 minutes for the undoped ceria, and 5, 10, and 20 minutes for
the Sm-doped ceria.
Figure 4.15 shows the gas dependencies for undoped ceria deposited for 5 and 10
minutes. The only real difference is under water partial pressure change, where the 5
minute sample exhibits a sharper slope. Comparing SEM images for the two samples at
equal magnification shows a slight qualitative decrease in apparent surface area for the 10
minute sample (Fig. 4.16). The deposit that lies on top of the platinum strips in the 10
minute sample has less wispy features than that grown on the YSZ surfaces, and as
compared to all areas of the deposit grown for 5 minutes. This observation holds true for
multiple samples and multiple configurations—prolonged depositions tend to reduce the
apparent surface area of the coatings grown on metal surfaces.
Analogously, Figure 4.17 shows the partial pressure dependencies for Sm-doped
ceria deposited for 5, 10, and 20 minutes. Similarly to the undoped comparison, the ASRs
are essentially the same, although the 10 minute sample performed slightly better in
water. The SEM images in Figure 4.18 show a slight deposition coverage increase in
going from 5 to 10 minutes of deposition, and the familiar chunky morphology is
observed for the 20 minute sample. Recall from Chapter 3 that the deposition rate for Smdoped ceria is slower than that of undoped ceria. 5 minutes of Sm-doped ceria CELD is
not enough to cover the sample with the HSA morphology, hence its higher ASRs as
compared to the 10 minute sample. Just like the undoped case, however, chunky coatings
over the platinum areas due to prolonged deposition times also increase the ASR.

108

These results indicate an optimal deposition time—approximately 5 minutes for
undoped ceria and approximately 10 minutes for Sm-doped ceria.

(b)
pH2 = 0.04 atm, 650 °C

~ / Ω cm2
R*

R* / Ω cm2

(a)

0.1

0.01
1E-3

5 min
10 min

0.01

pH2O = 0.0021 atm, 650 °C

0.1

0.01

5 min
10 min

0.1

H2 Content / atm

H2O Content / atm

Fig. 4.15. Water (a) and hydrogen (b) partial pressure dependencies of the SDC|gas interfacial ASR for 5
minute (squares) and 10 minute (circles) depositions for 5-5 µm CELD/Pt strips-embedded samples with
the undoped electrolyte.

(a)

(b)

Fig. 4.16. SEM images of the undoped samples probed in Fig. 4.15: (a) a 5 minute deposition on a 5-5 µm
pattern; (b) a 10 minute deposition on a 5-5 µm pattern.

109

(b)
pH2 = 0.04 atm, 650 °C

0.1

0.01
1E-3

10 pH O = 0.0021 atm, 650 °C

R* / Ω cm2

(a) 10

5 min
10 min
20 min

0.01

5 min
10 min
20 min

0.1

H2 Content / atm

H2O Content / atm

Fig. 4.17. Water (a) and hydrogen (b) partial pressure dependencies of the SDC|gas interfacial ASR for 5
minute (squares), 10 minute (circles), and 20 minute (triangles) depositions for 5-5 µm CELD/Pt stripsembedded samples with the doped electrolyte.

(a)

(b)

(c)

Fig. 4.18. SEM images of the doped samples probed in Fig. 4.17: (a) a 5 minute deposition on a 5-5 µm
pattern; (b) a 10 minute deposition on a 5-5 µm pattern; and (c) a 20 minute deposition on a 5-5 µm pattern.

110

3. Cation doping effect: undoped vs. Sm-doped ceria.
The doping effect for samples deposited for 5 minutes can be seen in Figure 4.19. There
appears to be little to no impact introduced by samarium doping. Considering the
similarities in microstructure seen in Figure 4.20, and the short distance charged species
must travel, this result is not surprising. Indeed, samarium is not expected to aid the
surface reaction kinetics much, and is instead introduced to increase the oxygen vacancy
concentration, as in Eqn. 1.5. Conversely, there is a difference in the performance of
undoped and Sm-doped samples deposited for 10 minutes, as seen in Figure 4.21, but
judging from the SEM images of Figure 4.22, it is likely that this difference is due to
slight surface area differences, rather than surface reaction kinetics. Similar to the
preceding section, the deposition rate difference between the undoped and Sm-doped
samples appears to be the primary culprit here.
Despite the minimal difference for samples deposited for 5 minutes, Sm-doped
ceria is preferred over undoped, if for no other reason than compatibility with a Smdoped ceria electrolyte solution.

(b)

(a)

10 pH O = 0.0021 atm, 650 °C

pH2 = 0.04 atm, 650 °C

R* / Ω cm2

0.01
1E-3

0.1

undoped
doped

0.01

H2O Content / atm

0.01

undoped
doped

0.1

H2 Content / atm

Fig. 4.19. Water (a) and hydrogen (b) partial pressure dependencies of the SDC|gas interfacial ASR for
undoped (squares) and doped (circles) electrolyte depositions for 5 minutes for 5-5 µm CELD/Pt stripsembedded samples.

111

(b)

(a)

Fig. 4.20. SEM images of the 5 minute deposition, 5-5 µm pattern samples probed in Fig. 4.19: (a) with the
undoped electrolyte; and (b) with the doped electrolyte.

(b) 10

pH2 = 0.04 atm, 650 °C

R* / Ω cm2

(a) 10

0.1

0.01
1E-3

undoped
doped

0.01

pH2O = 0.0021 atm, 650 °C

0.1

0.01

undoped
doped

0.1

H2 Content / atm

H2O Content / atm

Fig. 4.21. Water (a) and hydrogen (b) partial pressure dependencies of the SDC|gas interfacial ASR for
undoped (squares) and doped (circles) electrolyte depositions for 10 minutes for 5-5 µm CELD/Pt stripsembedded samples

(a)

(b)

Fig. 4.22. SEM images of the 10 minute deposition, 5-5 µm pattern samples probed in Fig. 4.21: (a) with
the undoped electrolyte; and (b) with the doped electrolyte.

112

4. Consecutive deposition effect: 5 vs. 5+5 minutes.
Section 3.3.2 outlined the high temperature behavior of the HSA coatings. Therein,
undesirable cracking issues were defined, and subsequent depositions were investigated
as a possibility to “healing” cracks induced either by the deposition process itself, or a
later annealing step. To assess the electrochemical activity impact of consecutive
depositions on a single sample, a Sm-doped sample deposited for 5 minutes was
subjected to an annealing step at 600 °C for 10 hours. The un-cracked, as-deposited
morphology is shown in Fig. 4.24a, and the cracked, as-annealed morphology is shown in
Fig. 4.24b. The sample was then probed for the first time via ACIS. Immediately
following the testing step, a second 5 minute deposition was performed on the same
sample, pictured in Fig. 4.24c. Then, the sample was probed for the second time via
ACIS, and is pictured in its final post-second-testing state in Fig. 4.24d.
As can be seen in Figure 4.23, there is practically no change in activity between
the 5 and 5+5 minute samples. This is a telling result, indicating that cracking in the HSA
morphology is not a significant concern. Fig. 4.24cd exhibits the expected deposition in
the former cracks of the sample, yet no impact on the ASR is detected. Although cracking
seems intuitively counterproductive, the cracking observed in these samples is on a
length scale so as to not prevent the migration of charged species. In fact, in-plane
cracking propagated perpendicularly to the length of the platinum strips will not impact
movement of charged species in the same perpendicular direction, as is the case for this
electrode configuration. If there was significant in-plane cracking along the length of the
platinum strips, then there would be an insurmountable barrier for oxygen ions to
traverse, in order to access surface reaction sites on top of the metal strips.

113

From a different perspective, the possibility exists to create even more surface
area by consecutive depositions due to concentration on the cracked areas. For this
reason, and because it is now known to not have an adverse effect, consecutive
depositions are still pursued in practice.

(b)
pH2 = 0.04 atm, 650 °C

R* / Ω cm2

(a)

0.01
1E-3

pH2O = 0.0021 atm, 650 °C

0.1

5 min
5+5 min

0.01

H2O Content / atm

0.01

5 min
5+5 min

0.1

H2 Content / atm

Fig. 4.23. Water (a) and hydrogen (b) partial pressure dependencies of the SDC|gas interfacial ASR for a 55 µm CELD/Pt strips-embedded sample that underwent two consecutive 5 minute depositions, with a high
temperature step in between: the first 5 minute deposition (squares) and the second 5 minute deposition,
referred to as 5+5 minute (circles), with the doped electrolyte are shown.

114

(a)

(b)

(c)

(d)

Fig. 4.24. Chronological SEM images of the 5-5 µm pattern sample probed in Fig. 4.23: (a) the asdeposited, un-cracked morphology after the first 5 minute deposition; (b) the cracked morphology after
annealing at 600 °C for 10 hours in air; (c) the as-deposited morphology after the second 5 minute
deposition, healing the cracks; and (d) the cracked morphology post-testing for the second time.

The following is a summary of the insights gained from the above experiments.
First, the furthest non-conducting surface distance away from the base electrogenerating
metal features should not exceed 3 microns. Prolonged depositions tend to lead to
chunky, lower surface area coatings on the metal portions of the substrate. This motivates
a reduction in the exposed metal area fraction, but is limited in practice by thermal
stability of micro-/nano-sized metal features. 5 minute depositions are preferred for
undoped samples, and 10 minute depositions are preferred for Sm-doped samples—the
difference is due to the deposition rate influence of cation doping. Sm-doped samples are
preferred due to their inherent transport properties, but surface kinetics do not appear to

115

be impacted at all. Healing cracks by consecutive depositions does not harm the electrode
activity, but could lead to a further enhancement of surface area; also, cracking does not
appear to be a critical concern in this electrode configuration.
Using the trends with partial pressures, these data can be extrapolated to 97% H2,
3% H2O, at 650 °C. The best samples give SDC|gas interfacial ASRs in the range 1.3 –
3.7 mΩ cm2, far below the state-of-the-art at the time of this writing [24-25].

4.3.2

Nickel Anti-Dot Films

Two experimental parameters are investigated for the CELD/Ni anti-dot-embedded
configuration, analogous to the treatment of platinum strip samples given above.

1. CELD deposition time effect: 5 vs. 10 vs. 20 + 2.5 minutes.
The SDC|gas interfacial ASRs of three nickel anti-dot samples are compared in Fig. 4.26,
with different depositions times—namely, 5 + 2.5 minutes, 10 + 2.5 minutes, and 20 +
2.5 minutes samples. A 2.5 minute consecutive deposition is used in all three cases. Not
much difference can be discerned from the partial pressure dependence plot, although
there is a slight ranking under changing pH2O, with the 10 minute besting the 5 minute
and 20 minute samples, in that order. A microstructure investigation revealed some
offsetting issues.
All three samples suffer from asymmetric deposition on what should be identical
sides of the cell. This is due to an experimental shortcoming. When a sample is dipped
into the liquid electrolyte (c.f. Fig. 3.2), one side of the cell is directly facing the counter
electrode, whereas the other is facing away from it. The side facing the counter electrode

116

experiences faster deposition rates than the side facing away. This is shown in Fig. 4.25,
where each row of images is a different sample, and the first three images of the first
column, i.e., 4.25a, c, and e, represent the side facing away from the counter electrode,
and the first three images of the second column, i.e., 4.25b, d, and f, represent the side
directly facing the counter electrode. Clear deposition maturity differences can be seen
between the two sides of each sample. Longer deposition times develop slat-like growth
on top of the original HSA morphology, seen in Fig. 4.25d, f, and h; however, in-between
the slats lay ideal HSA morphology (Fig. 4.25g). The 20 minute sample even sees HSA
growth on top of the slats, although the slats begin to warp and fold, disconnecting
themselves from the underlying layer (Fig. 4.25h).
These features notwithstanding, the critical characteristic that appears to dominate
the ultimate performance is overall coverage. The 5 minute sample left innumerable
micro-sized regions devoid of the HSA morphology, as in Fig. 4.25a. The 20 minute
sample had large areas completely uncovered by any deposit, HSA or not. The 10 minute
sample, on the other hand, was the most consistently covered over large and small scales
with the desired morphology, and consequently exhibits the best performance. Neither
cracking nor slat-growth inhibits performance as much as basic coverage issues.

117

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

Fig. 4.25. SEM images of the doped Ni anti-dot samples probed in Fig. 4.26: the first row is the 5 + 2.5
minute sample, where (a) is taken from the side facing away from the counter electrode and (b) is from the
side facing toward it; the second row is the 10 + 2.5 minute sample, where (c) is the side facing away, (d) is
the side facing toward; and the third row is the 20 + 2.5 minute sample, where (e) is the side facing away,
(f) is the side facing toward. (g) is the morphology lying in between the slats in (d); and (h) shows
overgrown slats disconnecting from the substrate from (f).

118

(b)
doped, Ni anti-dot
pH2 = 0.04 atm, 650 °C

0.1

0.01
1E-3

10 doped, Ni anti-dot

pH2O = 0.0021 atm, 650 °C

R* / Ω cm2

~ / Ω cm2
R*

(a) 10

5 min
10 min
20 min

0.01

H2O Content / atm

0.01

5 min
10 min
20 min

0.1

H2 Content / atm

Fig. 4.26. Water (a) and hydrogen (b) partial pressure dependencies of the SDC|gas interfacial ASR for
CELD/Ni anti-dot-embedded samples deposited with the doped electrolyte for 5 + 2.5 minutes (squares),
10 + 2.5 minutes (circles), and 20 + 2.5 minutes (triangles). The initial PS bead size was 2 µm and etched
to 1.4 µm.

2. CELD morphology effect: planar vs. HSA(1) and HSA(2).
As discussed in Chapter 3, both HSA and planar film morphologies are possible with
CELD ceria. These are compared in Fig. 4.28. There is a clear difference in activity,
owing to the surface area difference between HSA samples (see Fig. 4.27ab, HSA(1) and
4.27cd, HSA(2)) and planar samples (see Fig. 4.27ef). Sample-to-sample variability is
again seen here in the two HSA samples. This difference appears to correlate well with
morphological differences apparent in the SEM images of Fig. 4.27.

119

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 4.27. Post-testing SEM images of the doped Ni anti-dot samples probed in Fig. 4.28: the first row is
HSA(1), deposited for 10 + 2.5 minutes; the second row is HSA(2), deposited for 10 + 2.5 minutes; and the
third row is the planar sample, deposited with the doped + H2O2 electrolyte at -0.55 V vs. SCE for 0.5 + 0.5
minutes.

120

(a) 100

(b) 100

pH2 = 0.45 atm, 650 °C

R* / Ω cm2

0.1

0.01
1E-3

pH2O = 0.005 atm, 650 °C

HSA (1)
HSA (2)
planar

0.01

0.1

0.01

H2O Content / atm

HSA (1)
HSA (2)
planar

0.1

H2 Content / atm

Fig. 4.28. Water (a) and hydrogen (b) partial pressure dependencies of the SDC|gas interfacial ASR for
CELD/Ni anti-dot-embedded samples deposited with the doped electrolyte for 10 + 2.5 minutes at 0.8 mA
cm-2 (HSA (1), squares), 10 + 2.5 minutes at 0.8 mA cm-2 (HSA (2), circles), and with the doped + H2O2
electrolyte for 0.5 + 0.5 minutes at -0.55 V vs. SCE (triangles). The initial PS bead size in each case was 2
µm and etched to 1.3-1.4 µm.

In summary, consistent HSA coverage is critical, and some sample-to-sample
variation exists. The deposition time should be long enough to cover the anti-dot
network, but not so long that slat-like growth begins, which effectively covers portions of
the enhanced surface area.
Extrapolating the best CELD/Ni anti-dot-embedded configuration data to 97% H2,
3% H2O, and 650 °C, the best LF ASR is 6.8 mΩ cm2.

It should be stressed that to make use of these promising results, the issues
associated with the CELD/metal-sandwich configuration need to be alleviated. The
reader should keep in mind that a HF arc still exists for the results presented in the
preceding section.

121

Chapter 5

Sundry Specialized CELD
Microstructures
5.1

Anodic Aluminum Oxide (AAO) Templated Nanowires

Due to their small sizes and potential for controllable features, nanowires/tubes represent
an attractive option for increasing the surface area of ceria-based anodes. Templated
approaches are currently the most flexible routes to fabricating nanowires/tubes, as only a
suitable filling method needs be developed for a particular material composition, be it
metallic, semiconducting, or ceramic, and so on. Anodic aluminum oxide (AAO)
templates are utilized here for their uniformity and tunability. The AAO template
fabrication process is first discussed, and then the CELD of ceria into its pores is
investigated.

5.1.1

AAO Template Formation Mechanism and Background

Aluminum naturally forms a thin oxide layer at room temperature and ambient
atmospheres. Given an extra driving force and a much different environment, this oxide
layer can take on a regular porous structure. Specifically, anodically oxidizing aluminum
metal in a liquid acidic electrolyte can produce an oxide layer with in-plane, hexagonally
arranged and vertically aligned pores (see the schematic in Fig. 5.1 and SEM images in
Fig. 5.2). The pores are straight and extend throughout the entire thickness of the oxide,
with the exception of a small “barrier layer” lying at the bottom of the pores, at the
metal|oxide interface. The geometric ratios of the structural features are primarily

122

Al2O3

Al3+

O2-

Al substrate
Fig. 5.1. A schematic showing a cross-sectional view of the hexagonally arrayed, vertically aligned porous
oxide template produced by anodically oxidizing aluminum metal. Aluminum ions generally travel toward
the liquid acidic electrolyte and oxygen ions from solution toward the positively biased aluminum metal.
An oxide barrier layer exists at the bottom of each pore.

controlled by the applied voltage—the diameter, inter-pore spacing, and even barrier
layer thickness all linearly depend on the voltage. In addition, different electrolyte
solution compositions provide different feature size ranges and, hence, different operating
voltage ranges. The three most common electrolyte solutions in ascending feature size
order are sulfuric acid, oxalic acid, and phosphoric acid.
Since the initial investigations began in the 1940’s [98-99], attempts have been
made to explain the seemingly anomalous behavior of the formation of porous alumina.
Recently, a comprehensive picture has been given that adequately explains the formation
mechanism and describes the steady-state growth conditions, mainly by uniting the early
theories [100-103]. The conclusions presented in the preceding references are
summarized below. First, the two governing reactions are given—aluminum ion

123

generation at the metal|oxide interface, and oxygen ion deposition at the oxide|electrolyte
solution interface.
𝐴𝑙 → 𝐴𝑙3+ + 3𝑒 −

𝐻2 𝑂 → 𝑂2− (𝑜𝑥) + 2𝐻 + (𝑎𝑞)

(5.1)
(5.2)

Electric species migration is no doubt happening—the oxide layer continues to

thicken and the aluminum metal is consumed as long as a voltage is applied. Although
the exact origin of the localized effects that ultimately lead to pore formation is not wellunderstood, it is now known that compressive stresses at the oxide|electrolyte solution
interface induce significant steady-state lateral flow in the oxide, in directions
perpendicular to the applied electric field. These stresses appear to come from
competitive adsorption of oxygen ions, as in Eqn. 5.2, and anions from the electrolyte
solution. The adsorption step highly regulates the current density, as well as impacts the
lateral Newtonian flow of the oxide layer, accounting for the extreme dependence of the
template geometry on electrolyte solution composition. This also explains why certain
acids, such as boric acid, produce dense, planar oxide films under identical conditions as
AAO template-producing acidic electrolyte solutions. The viscous flow induced by local
stress concentration acts to push oxide material radially away from the pore bottoms, and
then up the pore walls, and is ultimately restricted by volume expansion and charge
conservation. The cascading effect of the applied voltage on the current density,
adsorption rate, local stress formation, and lateral flow of the oxide explains its robust
relationship to the geometric feature sizes of the resulting template.
Historically, it has been theorized that the pores were formed simply as a result of
the applied voltage leading to local electric field-assisted dissolution of the oxide layer by

124

Joule (or, resistive) heating. Although this theory alone is unable to explain the
invariance of the barrier layer thickness with time, it is a mechanistic factor, as it can lead
to inhomogeneity in oxide formation rates for different regions of a single sample [103].
For bulk aluminum samples, Joule heating effects are unnoticeable. However, for thin
film aluminum samples, which contain a buried gold or platinum under-electrode, rapid
oxide dissolution arising from poor heat conduction of the supporting substrate can
expose the gold/platinum to the liquid electrolyte. At operating voltages of 20 – 100 V,
this translates into violent gaseous evolution catastrophically destroying the fragile thin
film configuration.

5.1.2

AAO Fabrication Experimental Details

A two-electrode setup is used, with an HP 6002A DC Power Supply as the voltage
source, aluminum metal (exposed surface area 0.5 – 2 cm2) as the anode, and a carbon
rod cathode, all immersed in an oxalic acid (0.3 – 0.6 M) liquid electrolyte. All
anodizations are conducted at 40 V and room temperature. Under these conditions, the
as-fabricated pore diameters are 15-25 nm, the inter-pore distance is ~90 nm, and the
barrier layer thickness is ~25 nm. To open up the pore diameters, an etching solution of
10 wt % phosphoric acid is used for 10-30 minutes at 30 °C, giving final pore diameters
of 40 – 80 nm.
Immediately upon applying a potential, the aluminum is anodically polarized, and
aluminum oxide spontaneously and continuously forms until the voltage is turned off.
This naturally consumes some depth of the aluminum metal, and for bulk aluminum
samples, AAO templates of almost arbitrary thickness can be achieved. The template

125

thickness naturally depends on the anodization time, with a formation rate of 3 – 4 nm
sec-1. Free-standing templates can be produced by anodizing for a couple of hours, which
gives a template thickness in the tens of microns range. To separate the oxide from the
metal, a saturated solution of HgCl2 is used to selectively attack the remaining aluminum
metal, while the AAO template is undisturbed.
The oxalic acid electrolyte solution does etch the alumina and the aluminum metal
during anodization, so a protective quick-drying coating (nail polish) is applied to the
meniscus area, as the dissolution activity is greatly enhanced there. In some
circumstances without this protective coating, a 0.25 mm thick aluminum foil sample
could be entirely etched through, after only a couple hours of anodizing.
Although not a strict requirement for the simple goal of surface area
enhancement, an ordered arrangement of pores is desirable, as a more accurate
determination of the specific increase in surface area can be calculated. To accomplish
this, a two-step anodization process is employed. First, an initial AAO layer is produced
at 40 V for 10 minutes. This layer is subsequently removed in a combination of chromic
(1.5 wt %) and phosphoric (6 wt %) acid at 60 °C, typically for 1 – 2 hours, depending on
the AAO layer thickness. This leaves indentations in the newly-exposed aluminum
surface. A second anodization at 40 V is then conducted, which produces the desired
ordered arrangement, utilizing the existing indentations as nucleation points. Finally, the
pore diameters are etched as before.
Conveniently, there is little variation in the produced structure when forming an
AAO template over small or large areas, and there is obvious symbiotic potential for
utilizing this liquid, electrochemical technique with other similar methods for ultimately

126

depositing nanowires, such as the CELDs of Chapter 3. Ultimately, the goal is to grow
ceria nanowires onto an existing two-dimensional porous metal network, as in the antidot films of Chapter 2. A gradual, graded experimental approach was taken to gather
critical criteria for successful AAO template growth in simple systems, before more
complex systems were attempted. For the former, 0.25 mm thick bulk aluminum foil is
used; for the latter, a thin film of aluminum (0.5 – 2 µm) is either thermally evaporated or
sputtered onto a non-porous thin film of gold/titanium, which itself is thermally
evaporated onto a robust substrate such as quartz, silicon, or YSZ. The titanium layer is
only 10 nm thick, and is used strictly for adhesion improvement of the gold thin film.
Because significant difficulties were encountered with this intermediate system, limited
attempts were made to fabricate thin AAO template layers on anti-dot films.

5.1.3

AAO Template Results

The twice-anodized AAO template process is illustrated in Figure 5.2 for a bulk
aluminum foil sample. As can be seen in Fig. 5.2a, the initial pore formation in the first
anodization is highly irregular. As the anodization continues, however, a more
homogeneous arrangement emerges. The first anodization should be conducted long
enough to reach this homogeneous state—10 minutes is sufficient for the conditions
given above. Fig. 5.2b shows the result of this evolution, where the first oxide layer has
been removed, and a hexagonal configuration of indentations can be seen on the surface
of the exposed aluminum metal. The chromic/phosphoric acid mixture used to remove
the first anodization layer does not appreciably attack the aluminum over the etching time
scales used here. Fig. 5.2cd show the ordered template after the second anodization

127

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 5.2. SEM images showcasing the two-step anodization procedure (see text for experimental details):
(a) a top-down view of the as-fabricated irregular pore growth after the first anodization; (b) after the first
anodized oxide layer is removed, periodic depressions can be seen on the newly exposed aluminum metal
surface; (c) and (d) top-down views of the second anodization layer after pore opening in phosphoric acid;
(e) a cross-sectional view of the pores; and (f) the barrier layers at the bottoms of the pores of the second
anodization oxide layer.

128

and pore etching steps, where the indentation effect from the first anodization can be
clearly seen. A cross-sectional view of a thick template is seen in Fig. 5.2e, with an inset
displaying some branched pores alongside perfectly aligned ones. Fig. 5.2f shows the
unmistakable barrier layer at the bottom of each pore.
No inherent limitation to the thickness of the AAO template was encountered;
nonetheless, relatively conservative thicknesses were investigated here, as maintaining
contact between as-produced nanowires and the substrate is imagined to be more difficult
the larger the aspect ratios are (the aspect ratio is the ratio of the length to the diameter).
The other factor affecting the aspect ratio is the pore diameter—a corresponding
comparison of different pore diameter etching times is shown in Figure 5.3. This etching

(a)

(b)

(c)

Fig. 5.3. SEM images of a twice-anodized template layer with different etching times in 10 wt %
phosphoric acid: (a) 10 minutes, (b) 20 minutes, and (c) 30 minutes.

129

action also removes the barrier layer lying at the bottom of the pores. Without its
removal, the barrier layer would cover all of the base electrogeneration sites, preventing
the CELD of ceria nanowires.
Similar results to bulk aluminum are shown for sputtered aluminum thin film
samples in Figure 5.4. The slight roughness pictured in Fig. 5.4b is a result of the
sputtering process, which sometimes deposits large chunks of aluminum, instead of the
expected flat and featureless film. The pores can be seen to run right down to the
underlying gold surface. Either gold or platinum must be used as the buried electrode,
because other non-precious metals would be too easily oxidized, as aluminum is.

(a)

(b)

Fig. 5.4. SEM images of an AAO template grown from a sputtered aluminum thin film on a gold/titanium
buried electrode on a silicon substrate. The pore opening was performed for 20 minutes.

Two prohibitive difficulties were encountered for the aluminum thin film
samples. The first is inhomogeneous current densities arising from poor heat dissipation
by the supporting substrate, as discussed in Section 5.1.1. Because of this issue,
successful thin film AAO template fabrication was less than 1 in 10. Figure 5.5 compares
three different quartz substrate thin film samples, with a corresponding current transient
taken during each anodization. The first sample contains no buried gold electrode under

130

Current / mA

1000

100

10
Al on quartz (a)
Al on Au/Ti on quartz (b)
Al on Au/Ti on quartz (c)

0.0

0.2

0.4

0.6

0.8

1.0

Time/Total Time

(a)

(c)

(b)

Fig. 5.5. Post-fabrication optical images of
thin film AAO template fabrication
attempts from a sputtered aluminum thin
film on a transparent quartz substrate: (a)
with no buried Au/Ti electrode; (b) with a
buried Au/Ti electrode; and (c) with a
buried Au/Ti electrode and in a
sequestered area defined on all sides. Each
sample had nail polish to protect the
meniscus area during fabrication. The
associated current transients are also
given.

131

the sputtered aluminum thin film, but still experienced complete oxidation, resulting in
the transparent window pictured in the top down view of Fig. 5.5a. Partial oxidation of
the aluminum around the border of the sectioned area can be seen manifesting by colorful
thin film interference patterns. This sample’s current transient is a standard anodization
response: after the initial charging current spike, a slight dip in the current is followed by
a flat steady-state value, until there is no more aluminum to oxidize and the current drops
to zero. In contrast, the sample pictured in Fig. 5.5b contains a buried gold electrode,
where violent gaseous evolution can be seen to have ripped apart portions of the film,
leaving the gold under-electrode completely exposed in some regions. The corresponding
current transient shows orders of magnitude difference in the absolute current values, as
well as undesirable spiking. The added current is coming from electrolytic oxidation
reactions taking place at exposed gold surfaces. Finally, a successful buried gold
electrode sample is shown in Fig. 5.5c, where the sectioned area for anodization is a
bronze color, in stark contrast to the protected, un-oxidized aluminum surrounding it.
This sample’s current transient bears more resemblance to the first, although some
variability still exists, largely due to the lower right-hand corner of the sample not being
protected enough. Similar to the sample with no buried gold electrode, the current drops
significantly when there is no more aluminum to oxidize.
A similar difficulty was reported in ref [103], utilizing a buried gold electrode thin
film configuration. To alleviate the problem, the anodization was carried out at ~5 °C,
and a pulsed voltage profile was employed, to allow time for any heat intensity to
dissipate. An analogous approach could be taken here.

132

The second issue is related to the barrier layer, shown for a bulk aluminum foil
sample in Fig. 5.6a and for a thin film sample in Fig. 5.6b. Note the thickness difference
in the barrier layers—the thin film sample has a much thicker barrier layer. Even after
etching, this barrier layer does not appear to be completely removed, which precludes
subsequent ceria CELD into the pores. This could be related to the non-zero final current
value recorded for the sample pictured in Fig. 5.5c. The stray current could be oxidizing
some gold metal, producing what appears to be a thicker barrier layer. Nevertheless, this
should not be an insolvent issue, as there are many reports of successful nanowire
depositions in this configuration, as discussed below.

(a)

(b)

Fig. 5.6. SEM images comparing barrier layer thicknesses: (a) ~30 nm from a bulk aluminum foil sample;
and (b) ~70 nm from a thin film sputtered aluminum sample.

133

5.1.4

Ceria Nanowire Growth

Aside from ceria [47, 49, 58-59, 81], a wide range of metals and semiconductors have
been grown in the pores of AAO templates, from nickel to titania [104-110]. Despite
numerous reports of successful electrochemical nanowire deposition into thin film AAO
on a conducting substrate, the problems mentioned in the previous section have not been
solved here, as of yet. To circumvent those two issues entirely and demonstrate a proofof-concept, nickel metal was sputter-coated onto one side of a free-standing AAO
template (Whatman Anodisc 47) with ~200 nm pore diameters. This template was then
attached to a glass slide via conducting copper tape and sealed with nail polish, leaving
~1 cm2 of an exposed surface. After allowing the electrolyte solution to naturally seep
into the AAO pores for ~30 minutes, CELD was performed galvanostatically at 1.6 mA
cm-2, for 30 minutes. After the deposition was complete, the alumina was etched in stages
in 3 M NaOH for 0.5 – 5 minutes. The resulting SEM images are shown in Figure 5.7.
The proportion of pores that are filled with CELD ceria is extremely high—it was
difficult to find a pore that is not filled. After etching, the nanowires appear to fall onto
each other, clumping up in sections. Although this result is not ideal for pure surface area
enhancement, some cross-linking of nanowires could have a beneficial effect for assisting
the lateral migration of active species, rather than requiring electrons and oxygen ions to
always traverse the length of the nanowire during fuel cell operation.
The entire assembly was also subjected to thermal treatment, to observe the
morphological stability of the nanowires. Figure 5.8 shows identically magnified images
of the nanowires after annealing at 300 and 500 °C for 5 hours at 5 °C min-1, revealing

134

that there is little to no evolution in the nanowires’ structure at these temperatures. This is
consistent with the high temperature results from Chapter 3, where much smaller features
were shown to be stable up to ~800 °C.

(a)

(b)

(c)

(d)

Fig. 5.7. SEM images of CELD nanowires grown in the pores of a free-standing AAO template after
etching in 3 M NAOH: (a) for 30 seconds and showing a top-down view; and (b-d) for 5 minutes, showing
angled views of various magnifications.

(a)

(b)

Fig. 5.8. SEM images of CELD nanowires grown in the pores of a free-standing AAO template after
annealing for 5 hours in air at (a) 300 °C and (b) 500 °C.

135

5.2

Inverse Opals

5.2.1

Inverse Opal Definition and Background

Nanowires are an effective way to increase surface area, but they are limited to
reasonable aspect ratios due to both the ability of the nanowires to stand relatively
upright, and the challenges of depositing several microns of aluminum metal that would
be required to make AAO templates thick enough to guide such growth. Indeed, the
potential electrochemically active space is generally assumed to be on the order of 10
microns perpendicularly away from the electrolyte layer in SOFCs. A large portion of
this volume would essentially remain unused in a nanowire-only design.
Another possibility is the inverse opal structure, schematically shown in Fig.
1.5cd. An opal structure is comprised of a three-dimensional close-packing of
monodisperse spheres, akin to a micro-sized face-centered cubic structure. An inverse
opal is the volume inverse of this close-packed monolith. Typically, one begins with an
opal structure, deposits the desired material in the interstices of the opal, and then
removes the original opal, leaving a well-defined inverse opal. Spheres with different
diameters can be utilized to create a range of three-dimensional surface area
enhancement, analogous to the two-dimensional variability shown for anti-dot metal
films in Chapter 2 (c.f. Fig. 2.2). Polystyrene (PS) spheres are used here, although both
poly(methyl methacrylate) and silica spheres are commonly used.
Any number of fluid based methods can be utilized to fill the interstices of the
sacrificial opal structure, from gaseous CVD processes to liquid precursor infiltration
techniques [111]. Perhaps the most popular infiltration route for metallic inverse opals is
electrodeposition [112-117]; for ceria, most reports utilize techniques based upon

136

common ceria sol-gel precursors such as alkoxides and chlorides [118-121]. To the
author’s knowledge, no reports exist at the time of this writing for ceria inverse opal
preparation by CELD.

5.2.2

Inverse Opal Fabrication Details

In keeping with CELD and SOFC configurational requirements, the familiar porous metal
networks on YSZ are used as substrates. To establish a PS opal on the substrate, ~45 µL
of a 10 wt % PS suspension is drop-cast onto a metal network/YSZ substrate and allowed
to naturally dry for ~1 hour. Although this produces a randomly arranged opal, ordering
is not crucial for surface area enhancement, just as with the CELD ceria nanowires of
Section 5.1. After the water from the PS suspension is completely evaporated away, the
resulting PS opal adheres nicely to the substrate. The entire assembly is then immersed
into an undoped liquid electrolyte in the same CELD system used for the depositions of
Chapter 3. Before the depositing potential is applied, ~30 minutes is given to allow the
electrolyte solution to adequately infiltrate all interstices of the opal, and afterwards ceria
is electrochemically deposited as before for 10-30 minutes. The PS spheres are initially
removed by extensive water washing post-deposition, and can be completely removed by
immersion in a toluene solution for 30 minutes, or thermal treatment at ~400 °C.

5.2.3

Inverse Opal Results

Figure 5.9 shows ceria inverse opal structures grown by CELD on platinum strip/YSZ
(Fig. 5.9a-c) and nickel anti-dot/YSZ substrates (Fig. 5.9d) with HSA deposition
conditions, i.e., 0.8 mA cm-2 with the undoped electrolyte solution. An advantage of

137

(a)

(b)

(c)

(d)

Fig. 5.9. SEM images of CELD inverse opal structures grown at 0.8 mA cm-2 with the undoped electrolyte
on (a-c) a 20-20 µm YSZ/Pt strip sample for 10 minutes and (d) on a YSZ/Ni anti-dot substrate for 30
minutes. Immersion in a toluene solution for 30 minutes removed all PS spheres.

depositing at the standard HSA potentials is nano-sized surface area enhancement in
addition to the micron-sized templated pores. The familiar nano-sheet/needle features can
be seen embedded in the inverse opal walls in Fig. 5.9b and the inset of 5.9c. A 10 minute
deposition produced 1-2 inverse opal layers for the platinum strip sample, and a 30
minute deposition produced 3-4 layers for the nickel anti-dot sample.
Two adhesion-related difficulties require further investigation. One, the PS opal
consistently sticks to the substrate if unperturbed; however, when immersing the
substrate/opal assembly into the CELD electrolyte solution, some of the opal becomes
detached. The result is non-templated deposition, as in Fig. 5.10a. Two, the adhesion of

138

(b)

(a)

Fig. 5.10. SEM images of difficulties encountered with the CELD inverse opal structures from Fig. 5.9: (a)
the PS opal structure can be accidentally removed when dipped into the liquid CELD electrolyte, and no
inverse opal results; and (b) few contact points between the CELD inverse opal and the substrate can lead
to spallation.

the ceria inverse opal to the substrate can be weak due to limited points of contact
resulting from the template action of the PS spheres. Fig. 5.10b shows such a situation,
where some of the inverse opal structure on a nickel anti-dot substrate spalled during the
relatively gentle post-deposition water washing step. As is the case with the nanowire
challenges posed in Section 5.1, these problems are solvable.

5.3

Oxidation Protection Coatings

Because of its ability to conformally coat irregularly shaped metallic substrates, CELD of
ceria has been previously investigated as a corrosion inhibitor [46, 53-56]. Specifically
relevant to SOFC fabrication techniques, ceria coatings on metal substrates that protect
against unwanted oxidation are investigated below.

5.3.1

Experimental Details

A nickel anti-dot film with initial and final PS diameters of 2 µm and 1 µm, respectively,
was deposited onto one side of a YSZ single crystal, as before (c.f. Chapter 2). This

139

substrate was vertically dipped halfway into a Sm-doped ceria CELD electrolyte solution,
and a potential was applied at -0.525 V vs. SCE for 1 hour. A thin film of ceria coated
half of the cell, whereas the other half still had the nickel anti-dot film exposed. This
sample was placed into a PLD system, where a 1 – 2 µm ceria film was deposited onto
both its coated and uncoated areas. The PLD operated with a background oxygen partial
pressure of 5 mtorr and at a substrate temperature of ~650 °C. SEM images are used to
evaluate the oxidative state for the coated and uncoated regions.

5.3.2

Results

The CELD coated nickel anti-dot film before the PLD deposition is shown in Fig. 5.11a.
Smooth and relatively crack-free, the deposition is nearly conformal, even over the
occasional trapped PS sphere. The PLD deposition over the coated region appears
similarly smooth, as in Fig. 5.11b, although there are more faceted features characteristic
of crystalline ceria. Of particular note is the flatness of the area lying directly above the
nickel anti-dot film. The cross section of the coated region is also shown in Fig. 5.11c.
The border area between the CELD coated and uncoated regions after the PLD deposition
is shown in Fig. 5.11c. There is a clear difference between the two, where the coated
region is flat, like the original morphology, and the uncoated region has volumeexpanded, noticeable by the smaller pore diameter. This volume expansion is from the
nickel metal oxidizing to NiO during PLD operation.

140

(a)

(b)

(c)

(d)

Fig. 5.11. SEM images of a thin film CELD ceria coating on a YSZ/Ni anti-dot substrate: (a) as-deposited;
(b) the subsequent PLD top coating on the region that was previously coated by CELD ceria; (c) a crosssectional view of the YSZ substrate/Ni anti-dot layer/CELD ceria layer/ PLD top coating layer; and (d) the
PLD top coating showing the border between the previously coated region (left) and the uncoated region
(right).

From these results, it can be concluded that the CELD ceria layer effectively
prevents significant oxidation of the nickel metal. CELD could be used to treat metallic
substrates before they are subjected to high temperature oxidizing conditions commonly
used in SOFC fabrication techniques.

5.4

CELD Ceria Grown Directly on MIEC SOFC Cathode Substrates

In contrast to their anode counterparts, SOFC cathodes are commonly comprised of
porous MIEC monoliths, typically with pervoskite-based crystal structures. Two such
MIEC materials are Ba0.5Sr0.5Co0.8Fe0.2O3-δ (BSCF) and SrCo1-xNbxO3-δ (SCN) [19, 122].

141

The mixed conduction is dominated by p-type electronic conduction in both material
systems, and both exhibit non-trivial conductivity at room temperature and ambient
pressure. This last characteristic, in particular, opens wide the possibility of utilizing
BSCF and SCN as conducting substrates in a variety of configurations for CELD ceria.
Most tantalizing is the possibility to make an entire cathode-electrolyte-anode SOFC
assembly, beginning with the cathode. One can imagine depositing a thin film of ceria by
CELD onto a self-supported, porous BSCF substrate, where densification of the ceria thin
film could allow it to perform as the electrolyte in a SOFC. The top surface of this newly
formed ceria electrolyte could then act as the substrate for subsequent anode fabrication,
like the model electrodes from Chapters 2 and 3. This fuel cell fabrication scheme has the
advantages of combining processing techniques that are naturally integrated with one
another, as well as not requiring the characteristic high processing temperatures
associated with more classical approaches.
Accordingly, the CELD of ceria onto BSCF substrates is investigated, with
analogous (but not shown) results for SCN substrates.

5.4.1

Substrate Preparation Details

Dense and porous BSCF substrates were prepared for CELD studies. A standard powder
process is employed for both, with a pore former used for the latter. BSCF powders are
prepared by a nitrate-based sol-gel method, which utilizes EDTA and citric acid as
complexing agents. The mixing bath is kept at temperatures around 80 °C, which induces
gelation. The resulting sticky gel is heat treated at ~250 °C to remove residual organics.
The blackened powders are then fully calcined at 950 °C to produce the desired pure

142

perovskite crystal phase. For dense substrates, this calcined powder is crushed by hand
with a mortar and pestle, pressed into a pellet isostatically at 350 MPa, and then
ultimately sintered at ~1050 °C. For porous substrates, the calcined powder is mixed in a
60:40 volumetric ratio with Cs2SO4, which acts as a pore former and is later removed by
water immersion, and the two powders are crushed together in the mortar and pestle to
ensure adequate mixing. This powder mixture is also isostatically pressed into a pellet,
but sintered at ~800 °C, owing to the low melting temperature of the cesium sulfate salt
(~1000 °C). The resulting pore sizes are on the order of a few microns.
The dense substrates are polished, using up to an 800 grit abrasive paper, and can
be simply immersed into the CELD electrolyte solution as before. However, to encourage
bridging of the pores by a thin ceria CELD layer and to discourage deposition inside the
pores, the porous BSCF substrates are mounted to a glass slide by conducting copper
tape, and infiltrated by viscous nail polish. The spontaneous capillary forces are sufficient
for full infiltration of the pores. After the polish has hardened, the top surface is gradually
planarized by 800 grit abrasive paper, and ultimately smoothed by 2 µm abrasive cloth
(Scientific Instrument Services micro mesh cloth, 12000 grit). The result is a nanometerscale smooth surface, with seamless interfaces between the nail polish-filled pores and
the surrounding BSCF matrix, as in Fig. 5.12. Electrical contact with the external power
supply is made via the underlying copper tape, a portion of which extends out from
underneath the pellet. All conducting surfaces except the desired smoothed BSCF one are
insulated from deposition by nail polish.

143

(b)

(a)

Fig. 5.12. SEM images of a porous BSCF substrate that has been infiltrated by viscous nail polish, which
has since hardened. These images are taken after planarization by abrasive cloth. The isolated, dark regions
are the hardened nail polish, surrounded by the lighter regions of sold BSCF.

5.4.2

CELD Results and Discussion

Figure 5.13 shows successful undoped ceria HSA deposition at 1.5 mA cm-2 for 5
minutes onto a dense BSCF substrate. The CELD undoped electrolyte solution used for
this sample is slightly altered from the familiar composition—0.1M cerium nitrate with
0.1 M H2O2. The voltage response was abnormal in that it reached values of -2.2 V vs.
SCE, which is much more negative than on regular metallic substrates. Highly uniform
and slightly cracked, the familiar nano-needle/sheet growth can be seen. Fig. 5.13c and d
reveal a cross-sectional piece of the deposit that was upended during subsequent handling
of the sample. The 5 minute deposition produced a multiple-microns-thick coating on the
BSCF, likely due to the high current density. The porous nature of the HSA deposit is
clearly visualized in these images.
Deposition onto the porous BSCF substrate assembly is shown in Figure 5.14.
Fig. 5.14a shows the sample immediately following deposition, with the nail polish still
inside the BSCF pores. Fig. 5.14b-d show the sample after the nail polish has been

144

(a)

(b)

(c)

(d)

Fig. 5.13. As-deposited SEM images of undoped CELD ceria grown on a dense BSCF substrate at 1.5 mA
cm-2 for 5 minutes, with a slightly more concentrated (0.1 M cerium nitrate) electrolyte with 0.1 M H2O2:
(a) and (b) are top-down views; and (c) and (d) show an upended cross-section.

removed by acetone washing. Some of the smaller pores are successfully bridged by the
thin ceria deposit, but as many of the BSCF pores are greater than 1 µm, they remain
open and, therefore, gas permeable. This deposition is using the standard Sm-doped
electrolyte solution with no hydrogen peroxide additive. Curiously, the applied voltage is
only

-0.5 V vs. SCE, which should produce a thin, planar film, a la the results obtained

in Chapter 3. Furthermore, at a deposition time of only 1 minute without hydrogen
peroxide, it is surprising that there is a discernable coating at all, much less one with the
HSA microstructure.
To investigate this peculiarity, CV scans for nickel and BSCF substrates in the
standard Sm-doped ceria electrolyte solution at 50 mV s-1 are compared in Figure 5.15.

145

(a)

(b)

(c)

(d)

Current Density / mA cm-2

12
10

Current Density / mA cm-2

Fig. 5.14. SEM images of doped CELD ceria grown on a porous BSCF substrate at -0.5 V vs. SCE for 1
minute: (a) the as-deposited morphology with the nail polish still intact, and a closer view in the inset; (b)
same as in (a), but with the nail polish removed; and (c) and (d) cross-sectional images showing some
bridging of larger pores.

BSCF
Ni

0.4
0.3
0.2
0.1
0.0

-0.2 -0.4 -0.6
Voltage / V

-2

0.0

-0.2

-0.4

-0.6

-0.8

Voltage / V

-1.0

-1.2

-1.4

Fig. 5.15. CV scans for nickel and BSCF substrates in the standard doped electrolyte, with a scanning rate
of 50 mV s-1. Inset is a magnified view for the less negative potential range.

146

As previously discussed, the large current leg for the nickel substrate corresponds to the
reduction of dissolved molecular oxygen and hydrogen gas evolution. As is clearly seen,
BSCF does not exhibit such currents at the same voltages. However, BSCF does have a
non-zero current at less negative voltages, significantly higher than the nickel substrate,
shown in the inset of Fig. 5.15. With such a non-zero current, deposition should be able
to occur at these lower voltages.
Indeed, Fig. 5.16a and c show top-down SEM images of films deposited in the
Sm-doped ceria electrolyte solution for 30 seconds on porous BSCF at -0.2 V vs. SCE
and no applied potential, respectively. When these porous BSCF substrates are removed
from the CELD electrolyte solution, a distinctive deep blue/purple film appears,
confirming deposition. Close inspection of the images in Figure 5.16 reveal slight cracks
or tears in the thin films, even on top of nail polish-filled BSCF pores. After removing the
nail polish by acetone washing and annealing at 700 °C for 10 hours under ambient air,
some of the bridged areas are maintained, while others are punctured, as in Fig. 5.16b.
There appears to be a simple correlation between the BSCF pore size and the ability of
CELD coatings to bridge the pore. In an effect to boost the bridging capability of these
thin films, eleven 30 second-long, subsequent depositions are performed at -0.65 V vs.
SCE in the Sm-doped ceria electrolyte solution onto porous BSCF. The as-deposited
results are shown in Fig. 5.16d, where some evidence of multiple depositions can be seen
by layered tearing in certain areas. This effort does not sufficiently strengthen the
deposited film, however, and the nail polish removal step introduced significant holes in
the ceria film, as before.

147

(a)

(b)

(c)

(d)

Fig. 5.16. SEM images of thin films of CELD ceria grown onto porous BSCF with the doped electrolyte:
(a) and (b) at -0.2 V vs. SCE for 30 seconds, (a) as-deposited with the nail polish still intact (dark regions)
and (b) after nail polish removal and annealing at 700 °C for 10 hours in air; (c) at no applied potential for
30 seconds, shown as-deposited; and (d) after eleven 30 second-long consecutive depositions at -0.65 V vs.
SCE, shown as-deposited. The inset in (d) shows layered tearing, confirming multiple depositions.

A plausible explanation for this anomalous behavior is the oxygen
electroreduction action of BSCF. The oxygen vacancy defect chemistry for an electron
hole/oxygen ion MIEC can be depicted by the following equation:
𝑂𝑂𝑋 + 2ℎ∙ ↔ 12𝑂2 (𝑔) + 𝑉𝑂∙∙

(5.3)

As the temperature increases, BSCF thermodynamically loses lattice oxygen to the
atmosphere; as the temperature decreases, the opposite occurs—BSCF incorporates
oxygen from its surroundings into its lattice. However, around room temperature, the
kinetics of such incorporation are slow, meaning BSCF that has experienced any high
temperatures will be essentially meta-stable at low temperatures seen later in time. This

148

incorporation reaction could be catalyzed when the BSCF is immersed into a CELD
electrolyte solution, and connected to an electronic circuit. Noticing that the non-zero
current in the BSCF CV of Fig. 5.15 is cathodic, the direction of current flow is
consistent with what would be induced by Eqn. 5.3. However, this explanation offers no
insight into why or how ceria is deposited. If it is assumed that the available oxygen
vacancies on the surface of BSCF are isolated, i.e. it is unlikely that two vacancies are
directly adjacent to each other in the crystal lattice, then only one oxygen atom would be
incorporated at a time. For molecular oxygen, this would leave another atom available to
aqueous hydrogen ions roaming in the acidic environment to produce a hydroxide ion.
Then, the precipitation of cerium species could continue, as before.
There is also scant evidence of BSCF substrates catalyzing anomalous ceria
structural growth. Microstructures reminiscent of the so-called nanosheaves of ref [123]
and feathery, self-assembling fractal scaffoldings are found near the meniscus area of the
sample from Figure 5.13. Some representative structures are pictured in Figure 5.17.

149

(a)

(b)

(c)

(d)

Fig. 5.17. SEM images of various structures deposited near the meniscus area for the sample from Fig.
5.13.

150

Chapter 6

Summary and Conclusions
Two fabrication techniques were investigated as they pertain to the assembly of
advanced solid oxide fuel cells.
Polymer sphere lithography has been utilized to create two-dimensional metallic
networks on fuel cell electrolyte materials. Although the fabrication process involves
somewhat imprecise, random elements, the experimental variation from the expected
geometries is extremely small. Under fuel cell operating conditions, the structures, and,
hence, the 3PB and 2PB area fraction values, exhibit remarkable high temperature
stability. These well-defined and well-behaved electrode structures access a wide range
of 3PB regimes, lending themselves to future mechanistic studies on electrolyte-electrode
material systems, as well as providing a strong experimentally correlated basis for
computational modeling. Beyond mechanistic studies, these anti-dot structures have
served as platforms for fabrication of three-dimensional electrodes.
The cathodic electrochemical deposition of undoped and Sm-doped ceria has been
developed in templated and template-free configurations to produce a variety of tunable
anode microstructures. The strictly chemical nature of the deposition step allows these
electronically insulating coatings to deposit onto non-conducting areas of substrates,
insofar as they are close enough to an exposed metal surface. The end result is ubiquitous
CeO2 coatings on thin, porous metallic networks overlaid onto YSZ/porous metal
substrates, with quality metal|CeO2 and YSZ|CeO2 interfaces, which are morphologically

151

stable at high temperatures and reducing atmospheres. Deposition was also definitively
shown to occur on two MIEC, fuel cell cathode materials—BSCF and SCN.
To probe the activity of CELD Sm-doped ceria anodes, detailed, morphologicallydriven ACIS analyses were conducted, revealing two co-dominant, resistive processes for
metal network embedded configurations. The LF arc was determined to be surfacerelated; the HF arc was determined to be configurationally related, in particular to the
resistance of electron migration through the SDC deposit on top of the metal regions, and
the resulting restriction of the field lines to the nominal 3PB region. The LF arc was
therefore taken to represent the true measure of surface activity for CELD ceria. The
lowest extrapolated ASR values for this arc were shown to be in the range of 1.3 – 6.8
mΩ cm2 at 650 °C in 97% H2 and 3% H2O.

152

Appendix A

ImageJ Analysis Details
The following describes the analytical approach to identifying and characterizing the
pores of the anti-dot networks using the ImageJ software described in Chapter 2. The
process is briefly illustrated in Fig. A.1. First, an as-taken grayscale SEM image is
imported into ImageJ (Fig. A.1a); then, the data bar region is cropped and the rest of the
image is converted into a true black-and-white image (Fig. A.1b). The ImageJ user can
define a grayscale threshold cutoff value, above which the associated pixels are converted
to purely black, and below which the pixels are converted to purely white. Consequently,
the ideal SEM image to be analyzed is one where there is significant grayscale contrast
between the circular pores exposing the electrolyte surface, and the metal network lying
on top. Secondary electron imaging mode was chosen owing to its inherent contrast
associated with topographical features (recall that the metal network is 200-400 nm
thick). Back-scattered mode, which provides elemental materials contrast, added
anywhere from 5-15% areal error due to pore shading; and in-lens mode, known for its
high contrast imaging ability, was found to provide inconsistencies related to charging
effects from the non-conducting YSZ. Care was taken to provide qualitatively consistent
contrast in the SEM images across the entirety of the substrate, to ensure accurate and
uniform threshold application.
The resulting image (Fig.A.1b) is now a mixture of connected and disconnected
black objects, which ImageJ can identify automatically. Problems arise, however, due to
dark pixels that are not pore-related. The smaller dark objects can be automatically
removed, and the image consequently cleaned up, as in Fig. A.1c. However, the messy,

153

larger unwanted dark objects remain—these are originally void areas left uncovered
during the PS deposition process and result in planar metal regions after the thermal
evaporation step. Fortunately, these areas are never circular, so a “circularity” filter can
be applied when identifying objects. This filter is applied between Fig. A.1c and d.
ImageJ defines a circularity of 1 to be a perfect circle, and 0 to be a straight line. In this
way, fractal objects like these metal regions can be removed from the counting. Fig. A.1d
is the final pictorial output of the ImageJ process, and shows which objects have been
identified. Pore area (2PB), perimeter (3PB), and total pore coverage are all automatically
enumerated. Depending on the pore size to be evaluated, different magnification was
necessary to ensure accuracy—it was found that no more than 1500 pores could be
evaluated from one image and retain acceptable levels of accuracy.

154

(a)

(b)

(c)

(d)

Fig A.1. The ImageJ image analysis process: (a) an as-taken SEM image; (b) cropping and conversion
to black-and-white; (c) image clean-up; and (d) final image identifying the pores.

155

Appendix B

Additional Images
B.1

Additional CELD Images

(a)

(b)

(c)

(d)

(e)

(f)

Fig. B.1. Angled (45°) SEM images of the HSA microstructure deposited at 0.8 mA cm-2 with a 0.05 M
doped electrolyte for 5 minutes (a through d); and a 0.1 M undoped electrolyte for 5 minutes (e and f).

156

(a)

(b)

(c)

(d)

(e)

(f)

Fig. B.2. SEM images of CELD thin films of ceria deposited at -0.55 V vs. SCE with the doped + H2O2
electrolyte on thin films of Ni on silicon substrates for different times: (a) – (d) 5 minutes; (e) and (f) 10
minutes. (a) shows an as-deposited crack that forms for thicknesses greater than 300 nm ceria films. (b) –
(f) are images taken after annealing in an Ar atmosphere at 600 °C for 10 hours. The white strips are crack
areas that originally formed as-deposited as in (a), but the exposed Ni metal has oxidized to NiO and
volume-expanded out of the crack. The higher degree of cracking for the thicker film in (e) and (f) is easily
visualized.

157

The following TEM images are taken from the same sample shown in Fig. 3.24. (a)
is a bright-field image, and (b) through (d) are corresponding dark-field images taken at
different tilting angles to highlight various grain orientations, which consequently appear
white. (e) is the selected-area diffraction pattern, labeled with approximate lattice
parameters. (f) is a HRTEM view of the polycrystalline deposit.

(a)

(b)

(c)

(d)

(e)

(f)

Fig. B.3. TEM images taken from the same sample as in Fig. 3.24. See description above.

158

B.2 Additional AAO Images
(a)
(b)

(c)

(d)

Fig. B.4. Various AAO images: (a) optical image of thin film interference patterns resulting from the
complete anodic oxidation of sputtered Al on a glass slide; (b) when the sputtered Al is sectioned off,
complete anodic oxidation results in a transparent window, seen here on a YSZ single crystal substrate
1 x 1 cm; (c) if Al metal is left in the oxalic acid electrolyte too long, crystallographic etching occurs;
and (d) when the AAO template is etched in chromic and phosphoric acid, incomplete template removal
results in peculiar structures.

159

(a)

(b)

(c)

(d)

Fig. B.5. CELD ceria nanowires from the same sample shown in Fig. 5.7. (a) – (c) are after etching in 3 M
NaOH for 2.5 minutes. (d) is as-deposited, showing the scale-like overgrowth of ceria once deposition was
complete in the entire pore lengths of the AAO template.

160

B.3

Additional Inverse Opal Images

(a)

(b)

(c)

(d)

Fig. B.6. Various SEM images of inverse opal structures from the same samples as in Fig. 5.9.

B.4

(a)

Additional MIEC Substrate Images

(b)

Fig. B.7. SEM images of HSA ceria grown on BSCF at 0.8 mA cm-2 for: (a) 1 minute, showing a good
interface between ceria and BSCF; and (b) 5 minutes, showing HSA bridging of a pore in the underlying
BSCF filled with nail polish (dark area).

161

B.5

Additional Oxidation Protection Coating Images

(a)

(b)

(c)

(d)

(e)

(f)

Fig. B.8. SEM images of the oxidative protection coating action of CELD ceria, as in Fig. 5.11: (a) and (b)
are the Ni anti-dot areas uncovered by CELD; (c) is the area covered by CELD; (d) is the border between
the covered and uncovered CELD regions; (e) and (f) are cross-sectional views of the CELD covered
regions. All images have a PLD top coating.

162

Appendix C

Alternate SOFC Microstructure
Fabrication Routes
C.1

Solution Impregnation into AAO Templates

A straightforward solution-phase approach to filling the pores of AAO templates was
reported in [124]. Briefly, an AAO template is immersed into a 2.5 M cerium nitrate bath
for 4 hours, dried at 50 °C for 4 hours, and then thermally treated from 150 – 500 °C to
solidify the nanowires/tubes. In an attempt to mimic this approach, an identical cerium
nitrate solution was employed with unaided impregnation (Fig. C.1), sonication-assisted
impregnation (Fig. C.2), stirring-assisted impregnation (Fig. C.3), and a combination of
stirring- and sonication-assisted impregnation (Fig. C.4). The last approach worked best,
in terms of filling fraction of the AAO pores. However, any solution-phase route to
making nanowires suffers from a common drawback—there is no inherent attachment to
an underlying substrate. Attempts were made to thermally sinter nanowires made from
these methods to a YSZ underlying substrate while they were still held in place by the
surrounding AAO matrix, but thermal treatment of the assembly has the undesirable sideeffect of crystallizing the alumina into an un-etchable form. Prolonged treatment in acids,
e.g., chromic and phosphoric, and bases, e.g. NaOH, had zero effect, as can be seen in
Fig. C.5. Accordingly, this method was abandoned.

163

(a)

(b)

Fig. C.1. Ceria nanowires partially filling the pore of an AAO template after unaided solution phase
impregnation.

(a)

(b)

Fig. C.2. Ceria nanowires partially filling the pore of an AAO template after sonicated solution phase
impregnation.

(a)

(b)

Fig. C.3. Ceria nanowires partially filling the pore of an AAO template after stirred solution phase
impregnation.

164

(a)

(b)

(c)

(d)

Fig. C.4. Ceria nanowires partially filling the pore of an AAO template after sonicated/stirred solution
phase impregnation.

Fig. C.5. SEM image of an AAO template after thermal treatment at 1100°C for 5 hours in air, and after
unsuccessful, repeated attempts to etch the template in chromic/phosphoric acid mixtures and NaOH.

165

C.2

Copper Nanowire Synthesis

It has been known since the 1960’s that copper oxide nanowires form spontaneously on
the outer surface of the oxide scale during thermal treatment of copper metal at
temperatures exceeding 400 °C [125-129]. The aspect ratio and number density can be
altered by changing the growth temperature and surrounding atmosphere. This approach
works well with bulk copper foil substrates, and also works on copper thin films grown
on a supporting substrate. However, the copper metal thin films must be greater than 500
nm in order to produce an appreciable amount of CuO nanowires. The as-produced CuO
nanowires can be subsequently reduced in a hydrogen plasma to copper metal [130].
Higher power density plasmas can significantly alter the original CuO morphology, but
lower power density plasmas can completely reduce the CuO to Cu without much
morphological evolution. Below are selected images from this approach. The
combination of copper metal thin film thickness limitations and the inconsistencies of the
process lead to its abandonment.

166

(a)

(b)

(c)

(d)

(e)

(f)

Fig. C.6. SEM images of CuO nanowires grown at ~500 °C for a couple of hours in ambient air from a
0.25 mm Cu foil.

167

(a)

(b)

(c)

(d)

(e)

(f)

Fig. C.7. SEM images of CuO nanowires grown at ~500 °C for a couple of hours in ambient air from
a thin film of Cu thermally evaporated onto polycrystalline SDC pellets (a) – (d); (e) and (f) show the
highly evolved morphology of CuO nanowires to a porous Cu film after they have been reduced in a
high power density hydrogen plasma for ~5 minutes.

168

(a)

(b)

(c)

(d)

(e)

(f)

Fig. C.8. SEM images of Cu nanowire structures resulting from a moderate power density hydrogen
plasma reduction of the CuO nanowires picture in Fig. C.6. (a) and (b) are treated with the plasma for
a couple of minutes; (c) – (f) have been treated for greater than 5 minutes, and show some texturing
on the nanoscale as a result.

169

References
1.

Chueh, W.C., Electrochemical & Thermochemical Behavior of CeO2-δ, in
Materials Science. 2011, California Institute of Technology.

Tarascon, J.M. and M. Armand, Issues and challenges facing rechargeable
lithium batteries. Nature, 2001. 414(6861): p. 359-367.

Zhao, X., et al., The role of nanomaterials in redox-based supercapacitors for
next generation energy storage devices. Nanoscale, 2011. 3(3): p. 839-855.

Zhang, H., G.P. Cao, and Y.S. Yang, Carbon nanotube arrays and their
composites for electrochemical capacitors and lithium-ion batteries. Energy &
Environmental Science, 2009. 2(9): p. 932-943.

Carrette, L., K.A. Friedrich, and U. Stimming, Fuel cells: Principles, types, fuels,
and applications. Chemphyschem, 2000. 1(4): p. 162-193.

Ryan O'Hayre, S.-W.C., Whitney Colella, Fritz B. Prinz, Fuel Cell Fundamentals.
2006: John Wiley & Sons.

Hulteen, J.C. and R.P. Vanduyne, Nanosphere Lithography - a Materials General
Fabrication Process for Periodic Particle Array Surfaces. Journal of Vacuum
Science & Technology a-Vacuum Surfaces and Films, 1995. 13(3): p. 1553-1558.

Haynes, C.L. and R.P. Van Duyne, Nanosphere lithography: A versatile
nanofabrication tool for studies of size-dependent nanoparticle optics. Journal of
Physical Chemistry B, 2001. 105(24): p. 5599-5611.

Ho, K.C. and J. Jorne, Electrochemical Impregnation of Nickel-Hydroxide - Flowthrough Vs Stagnant Electrodes. Journal of the Electrochemical Society, 1990.
137(1): p. 149-158.

10.

Indira, L. and P.V. Kamath, Electrogeneration of Base by Cathodic Reduction of
Anions - Novel One-Step Route to Unary and Layered Double Hydroxides (Ldhs).
Journal of Materials Chemistry, 1994. 4(9): p. 1487-1490.

11.

Yamazaki, Y., R. Hernandez-Sanchez, and S.M. Haile, High Total Proton
Conductivity in Large-Grained Yttrium-Doped Barium Zirconate. Chemistry of
Materials, 2009. 21(13): p. 2755-2762.

170

12.

Magraso, A., et al., Development of Proton Conducting SOFCs Based on LaNbO4
Electrolyte - Status in Norway. Fuel Cells, 2011. 11(1): p. 17-25.

13.

Tucker, M.C., Progress in metal-supported solid oxide fuel cells: A review.
Journal of Power Sources, 2010. 195(15): p. 4570-4582.

14.

Haile, S.M., Fuel cell materials and components. Acta Materialia, 2003. 51(19):
p. 5981-6000.

15.

Eguchi, K., et al., ELECTRICAL-PROPERTIES OF CERIA-BASED OXIDES
AND THEIR APPLICATION TO SOLID OXIDE FUEL-CELLS. Solid State
Ionics, 1992. 52(1-3): p. 165-172.

16.

Tsai, T.P. and S.A. Barnett, Effect of mixed-conducting interfacial layers on solid
oxide fuel cell anode performance. Journal of the Electrochemical Society, 1998.
145(5): p. 1696-1701.

17.

Steele, B.C.H., Appraisal of Ce1-yGdyO2-y/2 electrolytes for IT-SOFC operation
at 500 degrees C. Solid State Ionics, 2000. 129(1-4): p. 95-110.

18.

Fleig, J., Solid oxide fuel cell cathodes: Polarization mechanisms and modeling of
the electrochemical performance. Annual Review of Materials Research, 2003.
33: p. 361-382.

19.

Shao, Z.P. and S.M. Haile, A high-performance cathode for the next generation of
solid-oxide fuel cells. Nature, 2004. 431(7005): p. 170-173.

20.

Yu, H.C., et al., Electrochemical characterization and performance evaluation of
intermediate temperature solid oxide fuel cell with La0.75Sr0.25CUO2.5-delta
cathode. Journal of Power Sources, 2005. 152(1): p. 22-26.

21.

Pena-Martinez, J., et al., Performance of XSCoF (X = Ba, La and Sm) and LSCrX
' (X ' = Mn, Fe and Al) perovskite-structure materials on LSGM electrolyte for ITSOFC. Electrochimica Acta, 2007. 52(9): p. 2950-2958.

22.

Zhang, P., et al., Sr2CoMoO6 anode for solid oxide fuel cell running on H-2 and
CH4 fuels. Journal of Power Sources, 2011. 196(4): p. 1738-1743.

23.

Tao, S.W. and J.T.S. Irvine, Synthesis and characterization of
(La0.75Sr0.25)Cr0.5Mn0.5O3-delta, a redox-stable, efficient perovskite anode for
SOFCs. Journal of the Electrochemical Society, 2004. 151(2): p. A252-A259.

171

24.

Atkinson, A., et al., Advanced anodes for high-temperature fuel cells. Nature
Materials, 2004. 3(1): p. 17-27.

25.

Jiang, S.P. and S.H. Chan, A review of anode materials development in solid
oxide fuel cells. Journal of Materials Science, 2004. 39(14): p. 4405-4439.

26.

Ciucci, F., et al., Surface reaction and transport in mixed conductors with
electrochemically-active surfaces: a 2-D numerical study of ceria. Physical
Chemistry Chemical Physics, 2011. 13(6): p. 2121-2135.

27.

Chueh, W.C., W. Lai, and S.M. Haile, Electrochemical behavior of ceria with
selected metal electrodes. Solid State Ionics, 2008. 179(21-26): p. 1036-1041.

28.

Gross, M.D., J.M. Vohs, and R.J. Gorte, A strategy for achieving high
performance with SOFC ceramic anodes. Electrochemical and Solid State Letters,
2007. 10(4): p. B65-B69.

29.

Lu, C., et al., A comparison of Cu-ceria-SDC and Au-ceria-SDC composites for
SOFC anodes, in Solid Oxide Fuel Cells Viii, S.C. Singhal and M. Dokiya,
Editors. 2003, Electrochemical Society Inc: Pennington. p. 773-780.

30.

Marina, O.A., et al., A solid oxide fuel cell with a gadolinia-doped ceria anode:
preparation and performance. Solid State Ionics, 1999. 123(1-4): p. 199-208.

31.

W.C. Chueh, Y.H., S.M. Haile, High Electrochemical Activity of the Oxide Phase
in Model Ceria-Pt and Ceria-Ni Composite Anode. Submitted 2011.

32.

Chueh, W.C. and S.M. Haile, Electrochemical studies of capacitance in cerium
oxide thin films and its relationship to anionic and electronic defect densities.
Physical Chemistry Chemical Physics, 2009. 11(37): p. 8144-8148.

33.

Wilson, J.R., et al., Three-dimensional reconstruction of a solid-oxide fuel-cell
anode. Nature Materials, 2006. 5(7): p. 541-544.

34.

Mizusaki, J., et al., Preparation of Nickel Pattern Electrodes on Ysz and Their
Electrochemical Properties in H2-H2o Atmospheres. Journal of the
Electrochemical Society, 1994. 141(8): p. 2129-2134.

35.

Bieberle, A., L.P. Meier, and L.J. Gauckler, The electrochemistry of Ni pattern
anodes used as solid oxide fuel cell model electrodes. Journal of the
Electrochemical Society, 2001. 148(6): p. A646-A656.

172

36.

Baumann, F.S., et al., Quantitative comparison of mixed conducting SOFC
cathode materials by means of thin film model electrodes. Journal of the
Electrochemical Society, 2007. 154(9): p. B931-B941.

37.

Bieberle, A. and L.J. Gauckler, State-space modeling of the anodic SOFC system
Ni, H-2-H2O vertical bar YSZ. Solid State Ionics, 2002. 146(1-2): p. 23-41.

38.

Goodwin, D.G., et al., Modeling Electrochemical Oxidation of Hydrogen on NiYSZ Pattern Anodes. Journal of the Electrochemical Society, 2009. 156(9): p.
B1004-B1021.

39.

Piao, J.H., et al., A study of process parameters of LSM and LSM-YSZ composite
cathode films prepared by screen-printing. Journal of Power Sources, 2008.
175(1): p. 288-295.

40.

Beckel, D., et al., Thin films for micro solid oxide fuel cells. Journal of Power
Sources, 2007. 173(1): p. 325-345.

41.

De Jonghe, L.C., C.P. Jacobson, and S.J. Visco, Supported electrolyte thin film
synthesis of solid oxide fuel cells. Annual Review of Materials Research, 2003.
33: p. 169-182.

42.

Therese, G.H.A. and P.V. Kamath, Electrochemical synthesis of metal oxides and
hydroxides. Chemistry of Materials, 2000. 12(5): p. 1195-1204.

43.

Zhitomirsky, I., Cathodic electrodeposition of ceramic and organoceramic
materials. Fundamental aspects. Advances in Colloid and Interface Science,
2002. 97(1-3): p. 279-317.

44.

Bhattacharya, R., et al., Electrodeposited biaxially textured buffer layer for
YBa2Cu3O7-delta (YBCO) superconductor oxide films. Journal of the
Electrochemical Society, 2006. 153(5): p. C273-C276.

45.

Zhitomirsky, I. and A. Petric, Electrochemical deposition of ceria and doped
ceria films. Ceramics International, 2001. 27(2): p. 149-155.

46.

Kamada, K., N. Enomoto, and J. Hojo, Optimization of electrochemical synthesis
conditions for dense and doped ceria thin films. Electrochimica Acta, 2009.
54(27): p. 6996-7000.

173

47.

Bocchetta, P., M. Santamaria, and F. Di Quarto, Electrosynthesis of Ce-Co mixed
oxide nanotubes with high aspect ratio and tunable composition. Electrochemical
and Solid State Letters, 2008. 11(3): p. K27-K30.

48.

Zhitomirsky, I. and A. Petric, Electrolytic and electrophoretic deposition of CeO2
films. Materials Letters, 1999. 40(6): p. 263-268.

49.

Zhang, C.J., et al., Heterogeneous films of ordered CeO2/Ni concentric
nanostructures for fuel cell applications. Physical Chemistry Chemical Physics,
2010. 12(17): p. 4295-4300.

50.

Zivkovic, L., et al., Effect of samarium addition and annealing on the properties
of electrodeposited ceria thin films. Thin Solid Films, 2011. 519(11): p. 35383543.

51.

Lair, V., et al., Synthesis and characterization of cerium oxide by electrochemical
methods. Physica Status Solidi C - Current Topics in Solid State Physics, Vol 5,
No 11 2008, 2008. 5(11): p. 3492-3495

135.
52.

Elbelghiti, H., et al., Electrodeposition of Ceria on Stainless Steel for SOFC
Interconnect Applications. Solid Oxide Fuel Cells 10 (Sofc-X), Pts 1 and 2, 2007.
7(1): p. 2391-2397

2646.
53.

Hamlaoui, Y., et al., Cathodic electrodeposition of cerium-based oxides on
carbon steel from concentrated cerium nitrate solutions Part I. Electrochemical
and analytical characterisation. Materials Chemistry and Physics, 2009. 113(23): p. 650-657.

54.

Kulp, E.A., et al., Electrodeposition of nanometer-thick ceria films by oxidation of
cerium(III)-acetate. Solid State Ionics, 2007. 178(11-12): p. 749-757.

55.

Li, F.B., R.C. Newman, and G.E. Thompson, In situ atomic force microscopy
studies of electrodeposition mechanism of cerium oxide films: Nucleation and
growth out of a gel mass precursor. Electrochimica Acta, 1997. 42(16): p. 24552464.

174

56.

Balasubramanian, M., C.A. Melendres, and A.N. Mansour, An X-ray absorption
study of the local structure of cerium in electrochemically deposited thin films.
Thin Solid Films, 1999. 347(1-2): p. 178-183.

57.

Zhou, Y.C., R.J. Phillips, and J.A. Switzer, Electrochemical Synthesis and
Sintering of Nanocrystalline Cerium(Lv) Oxide Powders. Journal of the American
Ceramic Society, 1995. 78(4): p. 981-985.

58.

Inguanta, R., S. Piazza, and C. Sunseri, Template electrosynthesis of CeO2
nanotubes. Nanotechnology, 2007. 18(48): p. -.

59.

Bocchetta, P., M. Santamaria, and F. Di Quarto, From ceria nanotubes to
nanowires through electrogeneration of base. Journal of Applied
Electrochemistry, 2009. 39(11): p. 2073-2081.

60.

Zhitomirsky, I. and A. Petric, Electrochemical deposition of yttrium oxide.
Journal of Materials Chemistry, 2000. 10(5): p. 1215-1218.

61.

Wang, A.Q. and T.D. Golden, Anodic electrodeposition of cerium oxide thin films
- I. Formation of crystalline thin films. Journal of the Electrochemical Society,
2003. 150(9): p. C616-C620.

62.

Golden, T.D. and A.Q. Wang, Anodic electrodeposition of cerium oxide thin films
- II. Mechanism studies. Journal of the Electrochemical Society, 2003. 150(9): p.
C621-C624.

63.

Mahajan, S., et al., Tuning plasmons on nano-structured substrates for NIR-SERS.
Physical Chemistry Chemical Physics, 2007. 9(1): p. 104-109.

64.

Shiu, J.Y., et al., Fabrication of tunable superhydrophobic surfaces by
nanosphere lithography. Chemistry of Materials, 2004. 16(4): p. 561-564.

65.

Boneberg, J., et al., The formation of nano-dot and nano-ring structures in
colloidal monolayer lithography. Langmuir, 1997. 13(26): p. 7080-7084.

66.

Burmeister, F., et al., Colloid monolayers as versatile lithographic masks.
Langmuir, 1997. 13(11): p. 2983-2987.

67.

Spada, E.R., et al., Homogeneous growth of antidot structures electrodeposited on
Si by nanosphere lithography. Journal of Applied Physics, 2008. 103(11): p. -.

175

68.

Haginoya, C., M. Ishibashi, and K. Koike, Nanostructure array fabrication with a
size-controllable natural lithography. Applied Physics Letters, 1997. 71(20): p.
2934-2936.

69.

Dimitrov, A.S. and K. Nagayama, Steady-State Unidirectional Convective
Assembling of Fine Particles into 2-Dimensional Arrays. Chemical Physics
Letters, 1995. 243(5-6): p. 462-468.

70.

Onoda, G.Y. and E.G. Liniger, Experimental-Determination of the RandomParking Limit in 2 Dimensions. Physical Review A, 1986. 33(1): p. 715-716.

71.

Mihi, A., M. Ocana, and H. Miguez, Oriented colloidal-crystal thin films by spincoating microspheres dispersed in volatile media. Advanced Materials, 2006.
18(17): p. 2244-+.

72.

Dimitrov, A.S. and K. Nagayama, Continuous convective assembling of fine
particles into two-dimensional arrays on solid surfaces. Langmuir, 1996. 12(5): p.
1303-1311.

73.

Trau, M., D.A. Saville, and I.A. Aksay, Field-induced layering of colloidal
crystals. Science, 1996. 272(5262): p. 706-709.

74.

Hayward, R.C., D.A. Saville, and I.A. Aksay, Electrophoretic assembly of
colloidal crystals with optically tunable micropatterns. Nature, 2000. 404(6773):
p. 56-59.

75.

Park, S.H., D. Qin, and Y. Xia, Crystallization of mesoscale particles over large
areas. Advanced Materials, 1998. 10(13): p. 1028-+.

76.

Scientific, T. NIST Traceable Particle Size Standards.

77.

Hamlaoui, Y., et al., Cathodic electrodeposition of cerium based oxides on carbon
steel from concentrated cerium nitrate. Part II: Influence of electrodeposition
parameters and of the addition of PEG. Materials Chemistry and Physics, 2010.
120(1): p. 172-180.

78.

Arurault, L., B. Daffos, and F.X. Sauvage, Nanocrystallized ceria-based coatings
prepared by electrochemistry on TA6V titanium alloy. Materials Research
Bulletin, 2008. 43(4): p. 796-805.

176

79.

Zhou, Y.C. and J.A. Switzer, Growth of cerium(IV) oxide films by the
electrochemical generation of base method. Journal of Alloys and Compounds,
1996. 237(1-2): p. 1-5.

80.

Arurault, L., et al., Electrochemical preparation of adherent ceria coatings on
ferritic stainless steel. Thin Solid Films, 2004. 466(1-2): p. 75-80.

81.

Bocchetta, P., M. Santamaria, and F. Di Quarto, Cerium oxyhydroxide nanowire
growth via electrogeneration of base in nonaqueous electrolytes. Electrochemical
and Solid State Letters, 2008. 11(9): p. K93-K97.

82.

Aldykiewicz, A.J., A.J. Davenport, and H.S. Isaacs, Studies of the formation of
cerium-rich protective films using x-ray absorption near-edge spectroscopy and
rotating disk electrode methods. Journal of the Electrochemical Society, 1996.
143(1): p. 147-154.

83.

Li, G.R., et al., Microstructural evolution of CeO2 from porous structures to
clusters of nanosheet arrays assisted by gas bubbles via electrodeposition.
Langmuir, 2008. 24(8): p. 4254-4259.

84.

Hayes, S.A., et al., The phase stability of cerium species in aqueous systems - I. EpH diagram for the Ce-HClO4-H2O system. Journal of the Electrochemical
Society, 2002. 149(12): p. C623-C630.

85.

Yu, P., et al., The phase stability of cerium species in aqueous systems - II. The
Ce(III/IV)-H2O-H2O2/O-2 systems. Equilibrium considerations and pourbaix
diagram calculations. Journal of the Electrochemical Society, 2006. 153(1): p.
C74-C79.

86.

Nobial, M., et al., The nitrate reduction process: A way for increasing interfacial
pH. Journal of Electroanalytical Chemistry, 2007. 600(1): p. 87-94.

87.

Giannuzzi, L.A. and F.A. Stevie, A review of focused ion beam milling techniques
for TEM specimen preparation. Micron, 1999. 30(3): p. 197-204.

88.

Jennings, A.T., M.J. Burek, and J.R. Greer, Microstructure versus Size:
Mechanical Properties of Electroplated Single Crystalline Cu Nanopillars.
Physical Review Letters, 2010. 104(13): p. -.

177

89.

Saitzek, S., et al., Nanostructured ceria: a comparative study from X-ray
diffraction, Raman spectroscopy and BET specific surface measurements. Physica
Status Solidi a-Applications and Materials Science, 2008. 205(7): p. 1534-1539.

90.

Weber, W.H., K.C. Hass, and J.R. Mcbride, Raman-Study of Ceo2 - 2nd-Order
Scattering, Lattice-Dynamics, and Particle-Size Effects. Physical Review B, 1993.
48(1): p. 178-185.

91.

Kosacki, I., et al., Raman scattering and lattice defects in nanocrystalline CeO2
thin films. Solid State Ionics, 2002. 149(1-2): p. 99-105.

92.

Balasubramanian, M., C.A. Melendres, and A.N. Mansour, X-ray absorption
spectroscopy study of the local structure of heavy metal ions incorporated into
electrodeposited nickel oxide films. Journal of the Electrochemical Society, 1999.
146(2): p. 607-614.

93.

Phok, S. and R.N. Bhattacharya, Effect of samarium doping on electrodeposited
CeO2 thin film. Physica Status Solidi a-Applications and Materials Science, 2006.
203(15): p. 3734-3742.

94.

Lai, W. and S.M. Haile, Impedance spectroscopy as a tool for chemical and
electrochemical analysis of mixed conductors: A case study of ceria. Journal of
the American Ceramic Society, 2005. 88(11): p. 2979-2997.

95.

Chang, H.Y. and H.I. Chen, Morphological evolution for CeO2 nanoparticles
synthesized by precipitation technique. Journal of Crystal Growth, 2005. 283(34): p. 457-468.

96.

Sepa, D.B., M.V. Vojnovic, and A. Damjanovic, Reaction Intermediates as a
Controlling Factor in the Kinetics and Mechanism of Oxygen Reduction at
Platinum-Electrodes. Electrochimica Acta, 1981. 26(6): p. 781-793.

97.

M. E. Orazem, B.T., Electrochemical Impedance Spectroscopy. 2008: John Wiley
& Sons.

98.

Edwards, J.D. and F. Keller, Formation of Anodic Coatings on Aluminum.
Transactions of The Electrochemical Society, 1941. 79(1): p. 135-144.

99.

Hunter, M.S. and P. Fowle, Factors Affecting the Formation of Anodic Oxide
Coatings. Journal of the Electrochemical Society, 1954. 101(10): p. 514-519.

178

100.

Houser, J.E. and K.R. Hebert, The role of viscous flow of oxide in the growth of
self-ordered porous anodic alumina films. Nature Materials, 2009. 8(5): p. 415420.

101.

Skeldon, P., et al., A tracer study of porous anodic alumina. Electrochemical and
Solid State Letters, 2006. 9(11): p. B47-B51.

102.

Garcia-Vergara, S.J., et al., Compositional evidence for flow in anodic films on
aluminum under high electric fields. Journal of the Electrochemical Society,
2007. 154(9): p. C540-C545.

103.

Matefi-Tempfli, S. and M. Matefi-Tempfli, Vertically Aligned Nanowires on
Flexible Silicone using a Supported Alumina Template prepared by Pulsed
Anodization. Advanced Materials, 2009. 21(40): p. 4005-+.

104.

Fan, Z.Y., et al., Electrical and photoconductive properties of vertical ZnO
nanowires in high density arrays. Applied Physics Letters, 2006. 89(21): p. -.

105.

Yasui, N., A. Imada, and T. Den, Electrodeposition of (001) oriented CoPt L1(0)
columns into anodic alumina films. Applied Physics Letters, 2003. 83(16): p.
3347-3349.

106.

Fusil, S., et al., Nanolithography based contacting method for electrical
measurements on single template synthesized nanowires. Nanotechnology, 2005.
16(12): p. 2936-2940.

107.

Sander, M.S., et al., Template-assisted fabrication of dense, aligned arrays of
titania nanotubes with well-controlled dimensions on substrates. Advanced
Materials, 2004. 16(22): p. 2052-+.

108.

Yang, Y., et al., Anodic alumina template on Au/Si substrate and preparation of
CdS nanowires. Solid State Communications, 2002. 123(6-7): p. 279-282.

109.

Vlad, A., et al., Nanowire-decorated microscale metallic electrodes. Small, 2008.
4(5): p. 557-560.

110.

Chong, M.A.S., et al., Combinational template-assisted fabrication of
hierarchically ordered nanowire arrays on substrates for device applications.
Applied Physics Letters, 2006. 89(23): p. -.

111.

Colvin, V.L., From opals to optics: Colloidal photonic crystals. Mrs Bulletin,
2001. 26(8): p. 637-641.

179

112.

Yu, X.D., et al., Filling fraction dependent properties of inverse opal metallic
photonic crystals. Advanced Materials, 2007. 19(13): p. 1689-+.

113.

Chung, Y.W., et al., Fabrication of various nickel nanostructures by manipulating
the one-step electrodeposition process. Journal of the Electrochemical Society,
2007. 154(6): p. E77-E83.

114.

Bartlett, P.N., T. Dunford, and M.A. Ghanem, Templated electrochemical
deposition of nanostructured macroporous PbO2. Journal of Materials Chemistry,
2002. 12(10): p. 3130-3135.

115.

Bartlett, P.N., et al., Optical properties of nanostructured metal films. Faraday
Discussions, 2004. 125: p. 117-132.

116.

Yan, H.W., et al., Fabrication of 2D and 3D ordered porous ZnO films using 3D
opal templates by electrodeposition. Electrochemistry Communications, 2005.
7(11): p. 1117-1121.

117.

Zhu, R., et al., Controlling the electrodeposition of mesoporous metals for
nanoplasmonics. Nanoscale, 2009. 1(3): p. 355-359.

118.

Holland, B.T., et al., Synthesis of highly ordered, three-dimensional, macroporous
structures of amorphous or crystalline inorganic oxides, phosphates, and hybrid
composites. Chemistry of Materials, 1999. 11(3): p. 795-805.

119.

Wang, T.W., et al., Preparation of a large Mesoporous CeO2 with crystalline
walls using PMMA colloidal crystal templates. Colloid and Polymer Science,
2006. 285(1): p. 1-9.

120.

Waterhouse, G.I.N., et al., Physical and optical properties of inverse opal CeO2
photonic crystals. Chemistry of Materials, 2008. 20(3): p. 1183-1190.

121.

Zhao, J.P., et al., Fabrication of three-dimensionally ordered macroporous
gadolinia-doped ceria films. New Journal of Chemistry, 2008. 32(6): p. 10141019.

122.

Zhou, W., et al., A novel efficient oxide electrode for electrocatalytic oxygen
reduction at 400-600 degrees C. Chemical Communications, 2008(44): p. 57915793.

123.

Kim, S., et al., Anomalous Electrical Conductivity of Nanosheaves of CeO2.
Chemistry of Materials, 2009. 21(7): p. 1182-1186.

180

124.

Yu, K.L., et al., Convenient synthesis of CeO2 nanotubesl. Materials Science and
Engineering B-Solid State Materials for Advanced Technology, 2007. 139(2-3):
p. 197-200.

125.

Gulbransen, E.A., T.P. Copan, and K.F. Andrew, Oxidation of Copper between
250-Degrees-C and 450-Degrees-C and the Growth of Cuo Whiskers. Journal of
the Electrochemical Society, 1961. 108(2): p. 119-123.

126.

Jiang, X.C., T. Herricks, and Y.N. Xia, CuO nanowires can be synthesized by
heating copper substrates in air. Nano Letters, 2002. 2(12): p. 1333-1338.

127.

Kumar, A., et al., The effect of growth parameters on the aspect ratio and number
density of CuO nanorods. Journal of Physics-Condensed Matter, 2004. 16(47): p.
8531-8543.

128.

Zhang, K.L., et al., Synthesis of large-area and aligned copper oxide nanowires
from copper thin film on silicon substrate. Nanotechnology, 2007. 18(27): p. -.

129.

Chopra, N., B. Hu, and B.J. Hinds, Selective growth and kinetic study of copper
oxide nanowires from patterned thin-film multilayer structures. Journal of
Materials Research, 2007. 22(10): p. 2691-2699.

130.

Qin, Y., T. Staedler, and X. Jiang, Preparation of aligned Cu nanowires by roomtemperature reduction of CuO nanowires in electron cyclotron resonance
hydrogen plasma. Nanotechnology, 2007. 18(3): p. -.

Electrochemically Deposited Ceria Structures for
Advanced Solid Oxide Fuel Cells

Thesis by

Evan C. Brown

In Partial Fulfillment of the Requirements
For the Degree of
Doctor of Philosophy

California Institute of Technology
Pasadena, California

2011
(Defended May 20, 2011)

Evan C. Brown

iii

vii

viii

Additional CELD Images………………….………………………………. 155

B.2

Additional AAO Images…………………………………… ………………158

B.3

Additional Inverse Opal Images…………………………………………… 160

B.4

Additional MIEC Substrate Images……………………………………….. 160

B.5

Additional Oxidation Protection Coating Images…………………………. 161

Appendix C: Alternate SOFC Microstructure Fabrication Routes…………… 162

C.1

Solution Impregnation into AAO Templates……………………................. 162

C.2

Copper Nanowire Synthesis……………………………………………….. 165

References………………………………………………………………………… 169

List of Figures
1.1

SOFC schematic………………………………..………………………….. 4

1.2

SOFC polarization curve………..…………………………………………..6

1.3

Electrolyte materials’ ionic conductivity comparison…………..…………. 8

1.4

3PB and 2PB in powder-processed electrodes schematics……….…..……. 10

1.5

Anti-dot film and templated electrode microstructure schematics…..…….. 15

2.1

SEM images of the polymer sphere lithography process……………..…… 20

2.2

Representative anti-dot film SEM images………..………………….…….. 22

2.3

SEM images of nanosphere short- and long-range coverage……..………...26

2.4

SEM images of nanosphere coverage for 500, 680, and 790 nm spheres..... 27

2.5

SEM images of multilayers and void regions from microsphere
deposition...................................................................................................... 28

2.6

Optical photographs of multilayers and void regions…………..………….. 28

2.7

Optical photographs of the water wash method………..………………….. 30

2.8

Optical and SEM images of one spin coat with the water wash method…... 30

2.9

SEM images of microsphere coverage for 2 and 3.2µm spheres………..….31

2.10

Pore diameter histograms of Cu anti-dot films…………………………….. 33

2.11

2PB area fraction histograms of Cu anti-dot films………………………… 33

2.12

SEM and AFM images of Ni anti-dot films before and after thermal
treatment…………………………………………………………………… 37

3.1

Pourbaix diagram for the Ce-H2O-H2O2 system from ref [86]……………. 40

3.2

SEM images of porous metal network on YSZ substrates used for CELD... 41

3.3

CELD setup schematic with corresponding potential electrode values…..... 44

3.4

XRD and EDS analyses of undoped and doped bulk CELD ceria……..….. 47

3.5

Raman analyses of undoped and doped bulk CELD ceria…..……………... 48

3.6

FT-IR analyses of undoped and doped bulk CELD ceria…………..……… 51

3.7

TGA analyses of undoped and doped bulk CELD ceria………..………….. 51

3.8

SEM images of undoped HSA ceria on YSZ/Pt strips…..………………… 53

3.9

SEM images of doped HSA ceria on various substrates…..………………. 54

3.10

SEM images of HSA ceria deposited from the doped + H2O2 electrolyte.... 55

3.11

SEM images of the HSA microstructure’s thermal stability………………. 56

3.12

SEM images of a crack-free HSA structure after annealing……………….. 57

3.13

SEM images of as-deposited thin film morphologies……………………… 58

3.14

AFM scans of as-deposited thin films from the doped and doped + H2O2
electrolytes…………………………………………………………………. 59

3.15

SEM cross-sectional images of thin film morphologies………………….... 60

3.16

SEM images of the deposits from the doped + acetic electrolyte………….. 62

3.17

SEM images showing the thermal stability of thin films on YSZ/Pt strips... 63

3.18

CV scan for the doped and doped + H2O2 electroltyes…….………………. 68

3.19

SEM images of as-deposited HSA CELD ceria growth on exposed YSZ
surfaces…………………………………………………………………….. 73

3.20

SEM images of as-deposited planar CELD ceria growth on exposed YSZ
surfaces…………………………………………………………………….. 75

3.21

SEM images of equivalent HSA growth on YSZ and Pt strip regions….…. 76

3.22

SEM images of as-deposited planar growth on the 3PB region of a

xii

YSZ/Pt strips substrate……………………………………………………. 78
3.23

TEM and HRTEM images of HSA CELD ceria…………………………... 78

3.24

HSA voltage transients and corresponding chronological SEM images…... 80

3.25

Thin film current transients…………………………………………………82

4.1

Representative Nyquist plot for a PLD/Pt strips exposed configuration….. 85

4.2

4.3

SEM images of PLD and CELD metal-embedded samples…..…………… 90

4.4

Representative Nyquist plots for PLD and CELD metal-embedded
samples………………………………………………………………….…. 92

4.5

Nyquist plots with hydrogen and water partial pressure dependencies for a
representative CELD/Ni anti-dot-embedded sample………………………. 93

4.6

Representative Nyquist plots for metal-sandwich configurations………… 94

4.7

Hydrogen partial pressure dependence of the single arc from metal-exposed
and metal-sandwich configurations…………..………………..…..………. 95

4.8

SEM images of the deleterious phenomena associated with CELD/metalsandwich samples…………………………………………………………...96

4.9

Hydrogen partial pressure dependence of the HF arc from metal-embedded
configurations……………………………………………………………… 98

4.10

Hydrogen and water partial pressure dependence, as well as 3PB and metal
spacing dependencies, of the HF arc for large pattern sized PLD samples…99

xiii

4.11

TEM images showing voids in the HSA CELD deposit near the exposed
metal surfaces…………………………………………………………….. 101

4.12

Hydrogen partial pressure dependence of the LF arc from metal-embedded
and the single arc from metal-exposed configurations……………………. 103

4.13

Pt pattern size effect on the SDC|gas interfacial ASR partial pressure
dependencies………………………………………………………………. 106

4.14

SEM images of undoped CELD ceria on 5-5 µm and 20-20 µm Pt patterns
on YSZ…………..……………………………………………………….... 106

4.15

Undoped CELD deposition time effect on the SDC|gas interfacial ASR
partial pressure dependencies……………..………………………………. 108

4.16

SEM images of undoped CELD ceria samples deposited for 5 and 10
minutes…………………………………………………………………….. 108

4.17

Doped CELD deposition time effect on the SDC|gas interfacial ASR partial
pressure dependencies……………….……………………….……………. 109

4.18

SEM images of doped CELD ceria samples deposited for 5, 10, and 20
minutes……………………………………………………………………... 109

4.19

Doping effect for 5 minute deposits on the SDC|gas interfacial ASR partial
pressure dependencies……………..………………………………...…….. 110

4.20

SEM images of doped and undoped CELD ceria samples deposited for 5
minutes……………………………………………………………………... 111

4.21

Doping effect for 10 minute deposits on the SDC|gas interfacial ASR partial
pressure dependencies….…………………………………………………. 111

xiv

4.22

SEM images of doped and undoped CELD ceria samples deposited for 10
minutes……………………………………………………………………... 111

4.23

Consecutive depositions effect on the SDC|gas interfacial ASR partial
pressure dependencies……………………………….……………………. 113

4.24

SEM images of consecutive depositions following thermal treatment…......114

4.25

SEM image comparison of doped CELD/Ni anti-dot-embedded samples
deposited for 5, 10, and 20 minutes……………………………………….. 117

4.26

Deposition time effect for doped CELD/Ni anti-dot-embedded samples on
the SDC|gas interfacial ASR partial pressure dependencies……………….. 118

4.27

SEM image comparison of two doped CELD HSA samples and one doped
CELD planar sample………………………………………………………. 119

4.28

SDC|gas interfacial ASR partial pressure dependencies comparison
between two HSA and one planar doped CELD samples…………………. 120

5.1

AAO structure schematic………..…………………………………………. 122

5.2

SEM images of AAO templates grown from Al foil……..………………... 127

5.3

SEM image comparison of phosphoric acid pore diameter etching times.....128

5.4

SEM images of AAO templates grown from sputtered Al thin films……....129

5.5

Current transients for AAO templates grown from various Al thin film
samples, whose optical photographs are also shown……………………… 130

5.6

SEM images of the barrier layer from AAO templates grown from Al foil

and sputtered Al thin films………………………………………………... 132
5.7

SEM images of as-deposited CELD ceria nanowires in the pores of AAO.. 134

5.8

SEM images of AAO ceria nanowires after thermal treatments…..………..134

5.9

SEM images of ceria inverse opal structures on YSZ/Pt strips and Ni anti-dot
substrates grown via CELD………………………………………………... 137

5.10

SEM images of difficulties encountered during the inverse opal fabrication
process……………………………………………………………………... 138

5.11

SEM images of the oxidative protection coating activity of CELD ceria
coatings on Ni anti-dot films……...………………………………………. 140

5.12

SEM images of the depositing surface of porous BSCF substrates that has been
planarized via abrasive paper………………………………………………. 143

5.13

SEM images of as-deposited undoped CELD ceria grown on dense BSCF..144

5.14

SEM images of as-deposited doped CELD ceria grown on porous BSCF… 145

5.15

CV scan comparison between Ni and BSCF substrates for the doped
electrolyte………………………………………………………………….. 145

5.16

SEM images of thin films of CELD ceria grown on porous BSCF at
non-standard and open working potentials.....……………………………. 147

5.17

SEM images of various CELD ceria structures deposited near the meniscus
area of a dense BSCF sample……………………………………………... 149

A.1

The ImageJ analysis process……………...………………………………... 154

B.1

Additional CELD HSA SEM images…...…………………………………. 155

xvi

B.2

Additional CELD thin film cracking SEM images………………………… 156

B.3

Additional CELD TEM images……………………………………………. 157

B.4

Additional optical and SEM images of AAO templates…………………… 158

B.5

Additional CELD ceria nanowires SEM images…………………………... 159

B.6

Additional CELD inverse opal SEM images….…………………………… 160

B.7

Additional CELD on MIEC substrate SEM images……………………….. 160

B.8

Additional oxidation protection coating SEM images…………………….. 161

C.1

SEM images of unaided solution phase impregnated ceria nanowires…….. 163

C.2

SEM images of sonicated-assisted impregnated ceria nanowires…………. 163

C.3

SEM images of stirring-assisted impregnated ceria nanowires……………. 163

C.4

SEM images of sonicated- and stirring-assisted impregnated ceria
nanowires…………………………………………………………………... 164

C.5

SEM image of an un-etchable AAO template after thermal treatment……. 164

C.6

SEM images of CuO nanowires thermally grown from Cu foil…………… 166

C.7

SEM images of CuO nanowires thermally grown from thin films of Cu on
SDC and porous Cu films after harsh hydrogen plasma treatment………… 167

C.8

SEM images of Cu nanowires resulting from reduction via a hydrogen
plasma at moderate power densities……………………………………….. 168

List of Tables
2.1

Comparison of theoretical and experimental 3PB length areal density and
percent 2PB exposure for different initial PS sphere diameters…………… 34

xvii

3.1

CELD liquid electrolyte compositions…………………………………….. 43

xviii

List of Symbols and Notations

number of electrons

Faraday’s constant

Nernstian voltage

oxygen non-stoichiometry

Ce3+/4+

dissociated aqueous cerium ions of a particular cerium valence

Ce(III/IV)

precipitated/solid cerium species of a particular cerium valence

𝑡ℎ𝑒𝑜
𝜌3𝑃𝐵

theoretical 3PB areal density

𝜌3𝑃𝐵

𝑒𝑥𝑝

experimental 3PB areal density

𝑡ℎ𝑒𝑜
𝑓2𝑃𝐵

theoretical 2PB area fraction

𝑓2𝑃𝐵

𝑒𝑥𝑝

experimental 2PB area fraction

𝜙𝑖

initial PS sphere diameter

𝜙𝑓

final PS sphere diameter

crystallite size

XRD x-ray wavelength

adjusted full-width half max

XRD diffracting angle

complex impedance

resistance

capacitance

frequency

√−1

xix

𝑍�

complex impedance normalized by total deposited area

𝑍� ∗

complex impedance normalized by the projected area of the exposed SDC

𝑅�∗

resistance associated with a Nyquist arc normalized by the projected area

𝑅�

resistance associated with a Nyquist arc normalized by total deposited area

surface

of the exposed SDC surface

Abbreviations
SOFC

solid oxide fuel cell

CELD

cathodic electrochemical deposition

YSZ

yttria-stabilized zirconia

OCV

open circuit voltage

SDC

samaria-doped ceria

GDC

gadolinia-doped ceria

MIEC

mixed ionic-electronic conductor

3PB

three-phase boundary

2PB

two-phase boundary

PLD

pulsed-laser deposition

CVD

chemical vapor deposition

AELD

anodic electrochemical deposition

polystyrene

SEM

scanning electron microscopy

AFM

atomic-force microscopy

HSA

high surface area

SCE

standard calomel electrode

XRD

x-ray diffraction

FT-IR

Fourier transform infrared

cyclic voltammetry

EDS

x-ray energy dispersive spectroscopy

TEM

transmission electron microscopy

ACIS

A.C. impedance spectroscopy

ASR

area-specific resistance

low frequency

high frequency

AAO

anodic aluminum oxide

BSCF

Ba0.5Sr0.5Co0.8Fe0.2O3-δ

SCN

SrxCoyNbzO3-δ

Chapter 1

Introduction and Background
1.1

A Global Perspective

1.2

SOFC Introduction

1.2.1

SOFC Basics

→ 2e-

1/ O + 2e- → O22 2

Cathode

Anode

Electrolyte

← O2-

H2 + O2- → H2O + 2e-

← O2Fig. 1.1. A schematic of a generalized SOFC, showing each electrode’s half-reactions and the
migration directions of each mobile species.

(1.1)

Driven by the need to maintain overall charge neutrality, the negatively charged

electrons travel through an externally connected circuit to the cathode, effectively
offsetting the dearth of negative charge left by migrating oxygen ions. These incoming

electrons then react with atmospheric oxygen, producing oxygen ions according to the
half-reaction:

𝑂2 (𝑔) + 2𝑒 − → 𝑂2−

(1.2)

The two electrode half-reactions combine to yield the overall reaction given in

∆𝐺𝑟𝑥𝑛
𝑛𝐹

(1.3)
(1.4)

Various deleterious phenomena decrease the operating voltage from the

Fig. 1.2. A visualization of the overpotential losses typically
encountered in SOFCs and the associated power density output of such a
cell.

1.2.2

Materials Selection: Samaria-Doped Ceria (SDC)

Temperature [ C]
800 700 600

400

300

Bi2O3

(Bi,Y)2O3
La
0.9 S
r0

-1

.1 Ga
0.8 M

0.8 G

-2
-3

2 )0

aZr Y
0.9 0.1 O

3-δ

.9 (S

3 )0

) 0.1
aO

) 0.1
O3
(Y
) 0.9
rO
(Z

-4

0.2 O
3-δ
0.2 O
1.9 B

(Zr
) 0.8
rO
(Z

Log(σ) [Ω-1cm-1]

500

-5

1.0

1.2

1.4

1.6

1.8

-1

1000/T [K ]
Fig. 1.3. A through-plane ionic conductivity comparison for
common electrolyte materials taken from [13].

(1.5)

(1.6)

(b)

(a)

H2
H2O

2e-

metal

YSZ
(ion conductor)

O2-

10 nm

1 µm

(d)
(c)

H2
H2O

metal

2e-

SDC
(ion and electron conductor)

O2-

10 nm

1 µm

1.2.3

Cell Architecture

1.3

Anti-Dot Substrates: A New Design Framework

1.4

Three-Dimensional Structures and Their Fabrication by CELD

1.4.1

SOFC Fabrication Method/Morphology Non-Negotiables

(a)

(b)

1 µm

(c)

(d)

1 µm

SDC
(ion and electron conductor)

1 µm

(e)

SDC
(ion and electron conductor)

11µm
µm

In this manuscript, cathodic electrochemical deposition is evaluated as a
components fabrication tool for a SDC-based, intermediate temperature SOFC.

1.4.2

Cathodic Electrochemical Deposition (CELD)

(1.7)
(1.8)

Detailed investigation of and results from the CELD of ceria microstructures are

presented in Chapters 3 and 5. Performance analyses of CELD ceria-based SOFC anode
structures are presented in Chapter 4.

Chapter 2

Anti-Dot Substrates
2.1

Polymer Sphere Lithography Background and Summary

(a)

(b)

(c)

(d)

(a)

(b)

(c)

(d)

(e)

(f)

2.2

Experimental Details

2.2.1

Substrate Preparation

2.2.2

Nanosphere Deposition

2.2.3

Microsphere Deposition

Non-functionalized PS beads were used for the larger diameters (Thermo Scientific), as
10 wt %. 35 µL of the PS suspension was manually spread over the entire substrate and

2.2.4

Metal Deposition

2.2.5

Microstructure Analysis

pore areas were assumed to be perfectly circular, and the diameters were calculated from
the individual pore areas.

2.2.6

High Temperature Stability

Nickel networks 200 nm thick were brought to 600 ºC under flowing 98.7% H2 and 1.3%
H2O and held there for 50 hours.

2.3

Results and Discussion

2.3.1

Nanosphere Lithography Results

Fig 2.3. SEM images of etched PS beads of 500 nm initial diameter covering large areas of the substrate
from a single spin coat step.

2.3.2

Microsphere Lithography Results

(a)

(b)

(c)

Fig 2.4. Substrate coverage via a single spin-coat for polystyrene nanospheres on YSZ: (a) 500 nm
diameter; (b) 680 nm diameter; and (c) 790 nm diameter. Each image contains ~1200 beads.

(a)

(b)

(c)

(d)

(a)

(b)

As an alternative to a single-step monolayer deposition procedure, a process was

(a)

(b)

(a)

Fig 2.9. Monolayer substrate coverage via multiple spin and wash cycles of polystyrene microspheres on
YSZ: (a) 2 µm, and (b) 3.2 µm. Both images contain ~1200 beads.

2.3.3

Microstructural Fidelity

𝑡ℎ𝑒𝑜
𝑓2𝑃𝐵

2𝜋 𝜙𝑓 1
� �
√3 𝜙𝑖 𝜙𝑖

2√3

𝜙𝑓 2

� �
𝜙𝑖

(2.1)
(2.2)

For an imperfect fabrication process, many kinds of defects exist—disordered

(a) 160000

(b) 12000
10000

Counts

120000
80000
40000

8000
6000
4000
2000

0.0 0.5 1.0 1.5 2.0 2.5 3.0
Effective Pore Diameter / µm

12000

(c)

5000

8000

Counts

Effective Pore Diameter / µm

(d) 6000

10000
6000
4000
2000

0.0 0.5 1.0 1.5 2.0 2.5 3.0

4000
3000
2000
1000

0.0 0.5 1.0 1.5 2.0 2.5 3.0
Effective Pore Diameter / µm

(b) 20

(a) 40

Counts

30
20
10

20 40 60 80 100
2PB Area Fraction (%)

(d) 16
14
Counts

Counts

20 40 60 80 100
2PB Area Fraction (%)

12
10

variability in 𝑓2𝑃𝐵 is more significant than the pore diameter variability. The widest

distribution in 𝑓2𝑃𝐵 is evident for the film prepared using 500 nm PS beads, where the

standard deviation is 12% of the substrate area (i.e., 𝑓2𝑃𝐵 is 31.5 ± 12.0%). Moreover, in
all cases, the observed 2PB area was lower than that computed from the initial and final

Table 2.1. Comparison of theoretical and experimental 3PB length areal density and percent 2PB exposure
for different initial PS bead diameters.
Initial bead

Final pore

Pore diameter

Theoretical

Experimental

Theoretical

Experimental

2PB exposure

diameter,

Gaussian

3PB length

percent 2PB

standard

𝝓𝒊 /µm

width/µm

density/m cm-2

exposure

deviation

0.5

𝜙𝑓 /µm
0.3

0.06

435

406

32.6%

31.5%

12.02

1.29

0.07

117

112.8

37.7%

33.4%

6.70

1.58

0.08

143

137.3

56.6%

49.7%

4.69

3.2

1.72

0.10

57.6

26.2%

23.5%

3.60

𝑡ℎ𝑒𝑜
PS beads (𝜙𝑓 = 1.58 µm) there is a large difference between 𝑓2𝑃𝐵
and 𝑓2𝑃𝐵 (56.6 vs.

49.7 %). When the beads are more heavily etched (𝜙𝑓 = 1.29 µm), the difference
𝑒𝑥𝑝

𝑡ℎ𝑒𝑜
and 𝜌3𝑃𝐵 and the distribution of pore
decreases, whereas the difference between 𝜌3𝑃𝐵

sizes for the two films are essentially the same. This behavior can be understood as

with one another, indicating that the fabrication is, in fact, rather controlled. Indeed, all of
the experimental 𝜌3𝑃𝐵 values are within 93% of the theoretical, and this was found to

Thermal Stability

(a)

(b)

(c)

(d)

Chapter 3

Cathodic Electrochemical Deposition of
Undoped and Doped Ceria
3.1

Introduction

applied potentials are used for the two morphologies, regardless of the composition of the
electrolyte solution.

3.2

Experimental Details

3.2.1

Substrate Definition

The primary type of substrates used was a composite substrate, comprised of a supporting
YSZ base, on top of which various kinds of porous metal networks are overlaid. The

(a)

(b)

(c)

3.2.2

Experimental Setup

Electrolyte Solution Name

[Ce(NO3)3]

[Sm(NO3)3]

[Additive]

Undoped

0.05 M

n/a

Sm-doped

0.0498 M

0.002 M

n/a

Sm-doped + H2O2

0.0498 M

0.002 M

0.025 M H2O2

Sm-doped + acetic acid

0.0498 M

0.002 M

0.05 M acetic acid

ee- Mn+
Mn+
ee- Mn+
eMn+
eee- Mn+
ecathode

potential

reference
(standard
calomel
electrode
[SCE])

anode

SCE

cathode

anode

Fig. 3.3. Schematic representation (a) of the standard three-electrode liquid electrochemical cell used for
CELD; and (b) the corresponding relative potential values.

3.2.3

Characterization Details

3.3

Results

3.3.1

Bulk

Measured Doping
Level / %

1800
1600

Intensity / a.u.

1400

doped 700°C

1200

Nominal
XRD
EDS

50
40
30
20
10

Nominal Doping Level / %

1000
800

doped as-dep

600

undoped 700°C

400

undoped as-dep

200

(a)

Ce-O8 stretching mode

2.5

Intensity / a.u.

2.0

oxygen
vacancies

nitrate ions

doped
H2O2

1.5

1.0

doped

0.5

undoped
0.0
250

500

750

1000

1250

-1

HWHM / cm-1

(b)

(c)

0.20

Intensity / a.u.

Wavenumber / cm

0.15

Saitzek 2008
Weber 1993
Kosacki 2002
HWHM Lorentz

0.0

0.1

d-1 / nm-1

0.2

0.3

700°C anneal

0.10
0.05
0.00
3000

as-dep
1000°C anneal
3500

Wavenumber / cm-1

4000

0.35
nitrate

water stretching mode

0.30

carbonate

Intensity / a.u.

0.25
0.20
0.15

(a)

water bending mode

0.10

nitrate

(b)

0.05

(c)

0.00
4000

(d)
3500

3000

2500

2000

1500

1000

-1

Wavenumber / cm

Mass Loss / %

1.00

0.95

0.90

0.85

doped + H2O2
undoped

0.80

0.75

doped

200

400

600

800

1000

Temperature / °C
Fig. 3.7. TGA weight loss for three electrolytes. Final weight loss ranges from 13-21%.

3.3.2

High Surface Area (HSA) Coatings

(a)

(b)

(c)

(d)

(e)

(f)

(a)

(b)

(c)

(d)

(e)

(a)

(b)

(c)

(d)

Fig. 3.10. SEM images of the as-deposited HSA microstructure obtained with the doped + H2O2 electrolyte
at 0.8 mA cm-2.

(a)

(b)

(c)

(d)

(a)

(b)

(c)

(d)

3.3.3

Thin Films

(a)

(b)

(a)

(b)

(a)

(b)

(c)

(d)

(e)

(f)

(a)

(b)

(c)

3.4

Discussion

3.4.1

General Deposition Overview

(3.1)

2𝐻3 𝑂+ + 2𝑒 − → 𝐻2 + 2𝐻2 𝑂

(3.3)

𝑂2 + 2𝐻2 𝑂 + 2𝑒 − → 2𝑂𝐻 − + 𝐻2 𝑂2

(3.2)

2𝐻2 𝑂 + 2𝑒 − → 𝐻2 + 2𝑂𝐻 −

(3.4)

𝑁𝑂3− + 7𝐻2 𝑂 + 8𝑒 − → 𝑁𝐻4+ + 10𝑂𝐻 −

(3.5)

𝐻2 𝑂2 + 2𝑒 − → 2𝑂𝐻 −

(3.7)

𝑁𝑂3− + 𝐻2 𝑂 + 2𝑒 − → 𝑁𝑂2− + 2𝑂𝐻 −

(3.6)

As an applied cathodic potential is made more negative, interfacial pH values

(3.8)

𝐶𝑒(𝑂𝐻)2+
2 + 2𝑂𝐻 → 𝐶𝑒(𝑂𝐻)4 ↓

(3.10)

2𝐶𝑒 3+ + 2𝑂𝐻 − + 𝐻2 𝑂2 → 2𝐶𝑒(𝑂𝐻)2+

(3.9)

𝐶𝑒(𝑂𝐻)2+
2 + 2𝑂𝐻 → 2𝐻2 𝑂 + 𝐶𝑒𝑂2 ↓

(3.11)

If Ce(III/IV) hydroxides are the precipitating species, they are readily oxidized to CeO2 in

the presence of O2:
4𝐶𝑒(𝑂𝐻)3 + 𝑂2 → 4𝐶𝑒𝑂2 + 6𝐻2 𝑂
𝑂2
𝐶𝑒(𝑂𝐻)4 �� 𝐶𝑒𝑂2 + 2𝐻2 𝑂

(3.12)
(3.13)

𝐶𝑒(𝑂𝐻)𝑥

exist, where x ranges from 0 to 2, but reported precipitation tests

67
+(4−𝑥)

discussion, Ce4+ intermediates exist of the type 𝐶𝑒(𝑂𝐻)𝑥

, where x ranges from 0 to

3, but, compared to 𝐶𝑒(𝑂𝐻)+2
2 , the others are either thermodynamically unfavorable in

2𝐶𝑒 3+ + 𝐻2 𝑂2 + 6𝑂𝐻 − → 2𝐶𝑒(𝑂𝐻)4 ↓

2𝐶𝑒 3+ + 𝐻2 𝑂2 + 6𝑂𝐻 − → 4𝐻2 𝑂 + 2𝐶𝑒𝑂2 ↓

(3.14)
(3.15)

However, given that the concentration of dissolved oxygen is on the order of µM [82],
the effect on the morphology should be much less than when mM of hydrogen peroxide
is explicitly added.

Cathodic Current Density / A cm-2

0.014
0.012
0.010
0.008
0.006
0.004

doped + H2O2

0.002

doped

0.000
-0.002

0.0

-0.2

-0.4

-0.6

-0.8

-1.0

-1.2

-1.4

Voltage / V
Fig. 3.18. CV scans for the doped (solid line) and doped + H2O2 (dotted line) electrolytes, taken at 50
mV s-1. Gray arrows indicate scan direction.

(3.16)

The hysteresis associated with this peak is related to the deposition that occurs

3.4.2

The Physical Deposition Picture

3.4.3

Deposition on Non-Conducting Parts of the Substrate

(a)

(b)

(c)

(d)

(a)

(b)

(c)

(d)

(e)

(f)

(a)

(b)

(c)

(a)

(b)

(c)
Fig. 3.22. As-deposited SEM images
from the sample shown in Fig. 3.21,
where connectivity between the thin
film deposit grown on the YSZ regions
and that grown on the platinum surface
is seen.

(a)

YSZ

(b)

CeO2
(c)

CeO2

(d)

YSZ

CeO2

3.4.4

HSA and Thin Film Transients

-0.6

doped, Ni anti-dot
-0.7

Voltage / V

(d)

III IV

doped, Pt 10-10 µm

-0.8

doped, Pt 5-5 µm
(a)

undoped, Pt 5-5 µm
-0.9

100

150

200

250

300

Time / s

II
Pt

YSZ

Pt
YSZ

III
YSZ

IV
Pt

YSZ
Pt

YSZ
YSZ

Cathodic Current Density / µA cm-2

3000

2500

doped + H2O2, Pt

2000

1500

doped + H2O2, Ni

doped, Pt

1000

500

doped, Ni

Chapter 4

The Electrochemical Activity of CELD
Ceria Structures
4.1

Introduction, Methods, and Background

4.1.1

A.C. Impedance Spectroscopy (ACIS) Introduction

pH2 = 0.04 atm, pH2O = 0.005 atm, 650 °C

-Im Z / Ω cm2

0.09 Hz

Re Z / Ω cm2

𝑍𝑟𝑒𝑠𝑖𝑠𝑡𝑜𝑟 = 𝑅
𝑍𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑜𝑟 =

(4.1)

𝑗𝜔𝐶

(4.2)

Most often, the complex impedances obtained via ACIS are represented as

impedances (or resistances) that have been normalized by the projected area of the
exposed SDC surface are given as 𝑍� ∗ (or 𝑅�∗ ). Accordingly, a resistance value extracted

from the Nyquist plots is referred to as an ASR, or area-specific resistance.

4.1.2

Experimental Approach

89
H2

(a)

H2O

metal

2eSDC

O2-

YSZ

(c)

(b)
metal

2e-

PLD SDC

2e-

YSZ

(d)
metal

PLD SDC

2e2e-

metal

2e-

CELD SDC

2e-

YSZ

(e)
metal

1st PLD SDC

CELD SDC

2e2e-

1st PLD SDC

YSZ

(a)

(b)
polycrystalline

single crystal

PLD SDC

thin Pt strip

(c)

PLD SDC

YSZ

(d)

Fig. 4.3. SEM images showing PLD top-layer (a and b) and CELD top-layer (c and d) metal-embedded
configurations. Depositions are performed at the standard conditions given in the corresponding text.

4.1.3

System Precedence

4.2

Arc Identification: PLD Films vs. CELD Coatings

4.2.1

Representative Spectra

(a)

-Im Z / Ω cm2

CHF

pH2 = 0.04 atm, pH2O = 0.005 atm, 650 °C
PLD/Pt strips
CELD/Pt strips
CELD/Ni anti-dot

RHF

RLF

0.19 Hz

385 Hz

Re Z / Ω cm2

(b)

0.8

0.4
0.3
3.44 Hz

0.2
0.1
0.0

3.8

4.0

4.2

Re Z / Ω cm2

4.4

-Im Z / Ω cm2

CLF

0.6
0.4

3.44 Hz

0.2
0.0
2.2

2.4

2.6

2.8

3.0

Re Z / Ω cm2

3.2

3.4

(a)

1.5
1.0

45% H2
29% H2
13% H2
4% H2

pH2O = 0.002 atm, 650 °C

0.5
0.0

(c)

Re Z / Ω cm2

-Im ~
Z / Ω cm2

0.75
0.50
0.25

0.00
5.75 6.00 6.25 6.50 6.75 7.00 7.25 7.50

Re Z / Ω cm2

(b)

1.5

0.114% H2O
0.066% H2O
0.05% H2O
0.021% H2O

pH2 = 0.04 atm, 650 °C

1.0
0.5
0.0

(d)
~ / Ω cm2
-Im Z

1.00

-Im ~
Z / Ω cm2

-Im Z / Ω cm2

Re Z / Ω cm2

1.00
0.75
0.50
0.25
0.00
5.75 6.00 6.25 6.50 6.75 7.00 7.25 7.50

Re ~
Z / Ω cm2

8 pH2 = 0.04 atm, pH2O = 0.005 atm, 650 °C
0.09 Hz

-Im Z / Ω cm2

PLD/Pt strips sandwich
CELD/Pt strips sandwich

Re Z / Ω cm2

4.2.2

Origin of the Single Arc in the Metal-Sandwich Configuration

100

pH2O = 0.005 atm, 650 °C

R* / Ω cm2

0.1

PLD/Pt strips exposed
PLD/Pt strips sandwich
CELD/Pt strips sandwich
0.01

0.1

H2 Content / atm

(a)

(b)

(c)

4.2.3

Origin of the HF Arc in Embedded Metal Configurations

100

pH2O = 0.005 atm, 650 °C

~ / Ω cm2
R*

0.1

PLD/Pt strips-embedded HF
CELD/Ni anti-dot-embedded HF
CELD/Pt strips-embedded HF

0.01

0.1

H2 Content / atm

+0.11
+0.12

~ / Ω cm2
R*

+0.12

+0.07

1E-3

-0.12
-0.12

-0.13
-0.09

50-100 µm
50-500 µm
50-900 µm
50-1300 µm

(b)
R* / Ω cm2

(a)

0.01

50-100 µm
50-500 µm
50-900 µm
50-1300 µm

0.01

H2 Content / atm

H2O Content / atm

(c)

0.1

(d)

50-1300 µm

50-900 µm
50-500 µm

50-100 µm
0.57% H2O
0.29% H2O
0.15% H2O
0.08% H2O

0.1

3PB / m cm-2

+0.85

50-100 µm
13% H2
4% H2
1.2% H2
0.4% H2

13% H2
4% H2
1.2% H2
0.4% H2

50-500 µm

-1.0

R* / Ω cm2

~ / Ω cm2
R*

50-900 µm

100

0.57% H2O
0.29% H2O
0.15% H2O
0.08% H2O

1000

Metal Spacing / µm

100

101

(a)

(b)

SDC
Pt

void

3PB
region

102

4.2.4

Origin of the LF Arc in Embedded Metal Configurations

103

100

pH2O = 0.005 atm, 650 °C

R* / Ω cm2

0.1

PLD/Pt strips-embedded LF
CELD/Pt strips-embedded LF
CELD/Ni anti-dot-embedded LF
PLD/Pt strips exposed single arc

0.01

0.1

H2 Content / atm

104

4.3

The SDC|Gas Interface Arc: A Closer Look

4.3.1

Platinum Strips

105

106

(b)
pH2 = 0.45 atm, 650 °C

0.1
1E-3

pH2O = 0.005 atm, 650 °C

R* / Ω cm2

(a)

5-5 µm
20-20 µm

0.01

0.1

5-5 µm
20-20 µm

0.1

H2 Content / atm

H2O Content / atm

(a)

(b)

(c)

(d)

107

108

These results indicate an optimal deposition time—approximately 5 minutes for
undoped ceria and approximately 10 minutes for Sm-doped ceria.

(b)
pH2 = 0.04 atm, 650 °C

~ / Ω cm2
R*

R* / Ω cm2

(a)

0.1

0.01
1E-3

5 min
10 min

0.01

pH2O = 0.0021 atm, 650 °C

0.1

0.01

5 min
10 min

0.1

H2 Content / atm

H2O Content / atm

(a)

(b)

Fig. 4.16. SEM images of the undoped samples probed in Fig. 4.15: (a) a 5 minute deposition on a 5-5 µm
pattern; (b) a 10 minute deposition on a 5-5 µm pattern.

109

(b)
pH2 = 0.04 atm, 650 °C

0.1

0.01
1E-3

10 pH O = 0.0021 atm, 650 °C

R* / Ω cm2

(a) 10

5 min
10 min
20 min

0.01

5 min
10 min
20 min

0.1

H2 Content / atm

H2O Content / atm

(a)

(b)

(c)

110

(b)

(a)

10 pH O = 0.0021 atm, 650 °C

pH2 = 0.04 atm, 650 °C

R* / Ω cm2

0.01
1E-3

0.1

undoped
doped

0.01

H2O Content / atm

0.01

undoped
doped

0.1

H2 Content / atm

111

(b)

(a)

Fig. 4.20. SEM images of the 5 minute deposition, 5-5 µm pattern samples probed in Fig. 4.19: (a) with the
undoped electrolyte; and (b) with the doped electrolyte.

(b) 10

pH2 = 0.04 atm, 650 °C

R* / Ω cm2

(a) 10

0.1

0.01
1E-3

undoped
doped

0.01

pH2O = 0.0021 atm, 650 °C

0.1

0.01

undoped
doped

0.1

H2 Content / atm

H2O Content / atm

(a)

(b)

Fig. 4.22. SEM images of the 10 minute deposition, 5-5 µm pattern samples probed in Fig. 4.21: (a) with
the undoped electrolyte; and (b) with the doped electrolyte.

112

113

(b)
pH2 = 0.04 atm, 650 °C

R* / Ω cm2

(a)

0.01
1E-3

pH2O = 0.0021 atm, 650 °C

0.1

5 min
5+5 min

0.01

H2O Content / atm

0.01

5 min
5+5 min

0.1

H2 Content / atm

114

(a)

(b)

(c)

(d)

115

4.3.2

Nickel Anti-Dot Films

Two experimental parameters are investigated for the CELD/Ni anti-dot-embedded
configuration, analogous to the treatment of platinum strip samples given above.

116

117

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

118

(b)
doped, Ni anti-dot
pH2 = 0.04 atm, 650 °C

0.1

0.01
1E-3

10 doped, Ni anti-dot

pH2O = 0.0021 atm, 650 °C

R* / Ω cm2

~ / Ω cm2
R*

(a) 10

5 min
10 min
20 min

0.01

H2O Content / atm

0.01

5 min
10 min
20 min

0.1

H2 Content / atm

119

(a)

(b)

(c)

(d)

(e)

(f)

120

(a) 100

(b) 100

pH2 = 0.45 atm, 650 °C

R* / Ω cm2

0.1

0.01
1E-3

pH2O = 0.005 atm, 650 °C

HSA (1)
HSA (2)
planar

0.01

0.1

0.01

H2O Content / atm

HSA (1)
HSA (2)
planar

0.1

H2 Content / atm

121

Chapter 5

Sundry Specialized CELD
Microstructures
5.1

Anodic Aluminum Oxide (AAO) Templated Nanowires

5.1.1

AAO Template Formation Mechanism and Background

122

Al2O3

Al3+

O2-

123

generation at the metal|oxide interface, and oxygen ion deposition at the oxide|electrolyte
solution interface.
𝐴𝑙 → 𝐴𝑙3+ + 3𝑒 −

𝐻2 𝑂 → 𝑂2− (𝑜𝑥) + 2𝐻 + (𝑎𝑞)

(5.1)
(5.2)

Electric species migration is no doubt happening—the oxide layer continues to

124

5.1.2

AAO Fabrication Experimental Details

125

126

5.1.3

AAO Template Results

127

(a)

(b)

(c)

(d)

(e)

(f)

128

(a)

(b)

(c)

Fig. 5.3. SEM images of a twice-anodized template layer with different etching times in 10 wt %
phosphoric acid: (a) 10 minutes, (b) 20 minutes, and (c) 30 minutes.

129

(a)

(b)

Fig. 5.4. SEM images of an AAO template grown from a sputtered aluminum thin film on a gold/titanium
buried electrode on a silicon substrate. The pore opening was performed for 20 minutes.

130

Current / mA

1000

100

10
Al on quartz (a)
Al on Au/Ti on quartz (b)
Al on Au/Ti on quartz (c)

0.0

0.2

0.4

0.6

0.8

1.0

Time/Total Time

(a)

(c)

(b)

131

132

(a)

(b)

Fig. 5.6. SEM images comparing barrier layer thicknesses: (a) ~30 nm from a bulk aluminum foil sample;
and (b) ~70 nm from a thin film sputtered aluminum sample.

133

5.1.4

Ceria Nanowire Growth

134

(a)

(b)

(c)

(d)

(a)

(b)

Fig. 5.8. SEM images of CELD nanowires grown in the pores of a free-standing AAO template after
annealing for 5 hours in air at (a) 300 °C and (b) 500 °C.

135

5.2

Inverse Opals

5.2.1

Inverse Opal Definition and Background

136

common ceria sol-gel precursors such as alkoxides and chlorides [118-121]. To the
author’s knowledge, no reports exist at the time of this writing for ceria inverse opal
preparation by CELD.

5.2.2

Inverse Opal Fabrication Details

5.2.3

Inverse Opal Results

137

(a)

(b)

(c)

(d)

138

(b)

(a)

5.3

Oxidation Protection Coatings

5.3.1

Experimental Details

A nickel anti-dot film with initial and final PS diameters of 2 µm and 1 µm, respectively,
was deposited onto one side of a YSZ single crystal, as before (c.f. Chapter 2). This

139

5.3.2

Results

140

(a)

(b)

(c)

(d)

5.4

CELD Ceria Grown Directly on MIEC SOFC Cathode Substrates

141

5.4.1

Substrate Preparation Details

142

143

(b)

(a)

5.4.2

CELD Results and Discussion

144

(a)

(b)

(c)

(d)

-0.5 V vs. SCE, which should produce a thin, planar film, a la the results obtained

145

(a)

(b)

(c)

(d)

Current Density / mA cm-2

12
10

Current Density / mA cm-2

BSCF
Ni

0.4
0.3
0.2
0.1
0.0

-0.2 -0.4 -0.6
Voltage / V

-2

0.0

-0.2

-0.4

-0.6

-0.8

Voltage / V

-1.0

-1.2

-1.4

Fig. 5.15. CV scans for nickel and BSCF substrates in the standard doped electrolyte, with a scanning rate
of 50 mV s-1. Inset is a magnified view for the less negative potential range.

146

147

(a)

(b)

(c)

(d)

(5.3)

148

149

(a)

(b)

(c)

(d)

Fig. 5.17. SEM images of various structures deposited near the meniscus area for the sample from Fig.
5.13.

150

Chapter 6

151

152

Appendix A

153

154

(a)

(b)

(c)

(d)

Fig A.1. The ImageJ image analysis process: (a) an as-taken SEM image; (b) cropping and conversion
to black-and-white; (c) image clean-up; and (d) final image identifying the pores.

155

Appendix B

Additional Images
B.1

Additional CELD Images

(a)

(b)

(c)

(d)

(e)

(f)

156

(a)

(b)

(c)

(d)

(e)

(f)

157

(a)

(b)

(c)

(d)

(e)

(f)

Fig. B.3. TEM images taken from the same sample as in Fig. 3.24. See description above.

158

B.2 Additional AAO Images
(a)
(b)

(c)

(d)

159

(a)

(b)

(c)

(d)

160

B.3

Additional Inverse Opal Images

(a)

(b)

(c)

(d)

Fig. B.6. Various SEM images of inverse opal structures from the same samples as in Fig. 5.9.

B.4

(a)

Additional MIEC Substrate Images

(b)

161

B.5

Additional Oxidation Protection Coating Images

(a)

(b)

(c)

(d)

(e)

(f)

162

Appendix C

Alternate SOFC Microstructure
Fabrication Routes
C.1

Solution Impregnation into AAO Templates

163

(a)

(b)

Fig. C.1. Ceria nanowires partially filling the pore of an AAO template after unaided solution phase
impregnation.

(a)

(b)

Fig. C.2. Ceria nanowires partially filling the pore of an AAO template after sonicated solution phase
impregnation.

(a)

(b)

Fig. C.3. Ceria nanowires partially filling the pore of an AAO template after stirred solution phase
impregnation.

164

(a)

(b)

(c)

(d)

Fig. C.4. Ceria nanowires partially filling the pore of an AAO template after sonicated/stirred solution
phase impregnation.

165

C.2

Copper Nanowire Synthesis

166

(a)

(b)

(c)

(d)

(e)

(f)

Fig. C.6. SEM images of CuO nanowires grown at ~500 °C for a couple of hours in ambient air from a
0.25 mm Cu foil.

167

(a)

(b)

(c)

(d)

(e)

(f)

168

(a)

(b)

(c)

(d)

(e)

(f)

169

References
1.

Chueh, W.C., Electrochemical & Thermochemical Behavior of CeO2-δ, in
Materials Science. 2011, California Institute of Technology.

Tarascon, J.M. and M. Armand, Issues and challenges facing rechargeable
lithium batteries. Nature, 2001. 414(6861): p. 359-367.

Zhao, X., et al., The role of nanomaterials in redox-based supercapacitors for
next generation energy storage devices. Nanoscale, 2011. 3(3): p. 839-855.

Zhang, H., G.P. Cao, and Y.S. Yang, Carbon nanotube arrays and their
composites for electrochemical capacitors and lithium-ion batteries. Energy &
Environmental Science, 2009. 2(9): p. 932-943.

Carrette, L., K.A. Friedrich, and U. Stimming, Fuel cells: Principles, types, fuels,
and applications. Chemphyschem, 2000. 1(4): p. 162-193.

Ryan O'Hayre, S.-W.C., Whitney Colella, Fritz B. Prinz, Fuel Cell Fundamentals.
2006: John Wiley & Sons.

Ho, K.C. and J. Jorne, Electrochemical Impregnation of Nickel-Hydroxide - Flowthrough Vs Stagnant Electrodes. Journal of the Electrochemical Society, 1990.
137(1): p. 149-158.

10.

11.

Yamazaki, Y., R. Hernandez-Sanchez, and S.M. Haile, High Total Proton
Conductivity in Large-Grained Yttrium-Doped Barium Zirconate. Chemistry of
Materials, 2009. 21(13): p. 2755-2762.

170

12.

Magraso, A., et al., Development of Proton Conducting SOFCs Based on LaNbO4
Electrolyte - Status in Norway. Fuel Cells, 2011. 11(1): p. 17-25.

13.

Tucker, M.C., Progress in metal-supported solid oxide fuel cells: A review.
Journal of Power Sources, 2010. 195(15): p. 4570-4582.

14.

Haile, S.M., Fuel cell materials and components. Acta Materialia, 2003. 51(19):
p. 5981-6000.

15.

Eguchi, K., et al., ELECTRICAL-PROPERTIES OF CERIA-BASED OXIDES
AND THEIR APPLICATION TO SOLID OXIDE FUEL-CELLS. Solid State
Ionics, 1992. 52(1-3): p. 165-172.

16.

Tsai, T.P. and S.A. Barnett, Effect of mixed-conducting interfacial layers on solid
oxide fuel cell anode performance. Journal of the Electrochemical Society, 1998.
145(5): p. 1696-1701.

17.

Steele, B.C.H., Appraisal of Ce1-yGdyO2-y/2 electrolytes for IT-SOFC operation
at 500 degrees C. Solid State Ionics, 2000. 129(1-4): p. 95-110.

18.

Fleig, J., Solid oxide fuel cell cathodes: Polarization mechanisms and modeling of
the electrochemical performance. Annual Review of Materials Research, 2003.
33: p. 361-382.

19.

Shao, Z.P. and S.M. Haile, A high-performance cathode for the next generation of
solid-oxide fuel cells. Nature, 2004. 431(7005): p. 170-173.

20.

21.

22.

Zhang, P., et al., Sr2CoMoO6 anode for solid oxide fuel cell running on H-2 and
CH4 fuels. Journal of Power Sources, 2011. 196(4): p. 1738-1743.

23.

171

24.

Atkinson, A., et al., Advanced anodes for high-temperature fuel cells. Nature
Materials, 2004. 3(1): p. 17-27.

25.

Jiang, S.P. and S.H. Chan, A review of anode materials development in solid
oxide fuel cells. Journal of Materials Science, 2004. 39(14): p. 4405-4439.

26.

27.

Chueh, W.C., W. Lai, and S.M. Haile, Electrochemical behavior of ceria with
selected metal electrodes. Solid State Ionics, 2008. 179(21-26): p. 1036-1041.

28.

Gross, M.D., J.M. Vohs, and R.J. Gorte, A strategy for achieving high
performance with SOFC ceramic anodes. Electrochemical and Solid State Letters,
2007. 10(4): p. B65-B69.

29.

30.

Marina, O.A., et al., A solid oxide fuel cell with a gadolinia-doped ceria anode:
preparation and performance. Solid State Ionics, 1999. 123(1-4): p. 199-208.

31.

W.C. Chueh, Y.H., S.M. Haile, High Electrochemical Activity of the Oxide Phase
in Model Ceria-Pt and Ceria-Ni Composite Anode. Submitted 2011.

32.

33.

Wilson, J.R., et al., Three-dimensional reconstruction of a solid-oxide fuel-cell
anode. Nature Materials, 2006. 5(7): p. 541-544.

34.

Mizusaki, J., et al., Preparation of Nickel Pattern Electrodes on Ysz and Their
Electrochemical Properties in H2-H2o Atmospheres. Journal of the
Electrochemical Society, 1994. 141(8): p. 2129-2134.

35.

172

36.

Baumann, F.S., et al., Quantitative comparison of mixed conducting SOFC
cathode materials by means of thin film model electrodes. Journal of the
Electrochemical Society, 2007. 154(9): p. B931-B941.

37.

Bieberle, A. and L.J. Gauckler, State-space modeling of the anodic SOFC system
Ni, H-2-H2O vertical bar YSZ. Solid State Ionics, 2002. 146(1-2): p. 23-41.

38.

Goodwin, D.G., et al., Modeling Electrochemical Oxidation of Hydrogen on NiYSZ Pattern Anodes. Journal of the Electrochemical Society, 2009. 156(9): p.
B1004-B1021.

39.

Piao, J.H., et al., A study of process parameters of LSM and LSM-YSZ composite
cathode films prepared by screen-printing. Journal of Power Sources, 2008.
175(1): p. 288-295.

40.

Beckel, D., et al., Thin films for micro solid oxide fuel cells. Journal of Power
Sources, 2007. 173(1): p. 325-345.

41.

De Jonghe, L.C., C.P. Jacobson, and S.J. Visco, Supported electrolyte thin film
synthesis of solid oxide fuel cells. Annual Review of Materials Research, 2003.
33: p. 169-182.

42.

Therese, G.H.A. and P.V. Kamath, Electrochemical synthesis of metal oxides and
hydroxides. Chemistry of Materials, 2000. 12(5): p. 1195-1204.

43.

Zhitomirsky, I., Cathodic electrodeposition of ceramic and organoceramic
materials. Fundamental aspects. Advances in Colloid and Interface Science,
2002. 97(1-3): p. 279-317.

44.

Bhattacharya, R., et al., Electrodeposited biaxially textured buffer layer for
YBa2Cu3O7-delta (YBCO) superconductor oxide films. Journal of the
Electrochemical Society, 2006. 153(5): p. C273-C276.

45.

Zhitomirsky, I. and A. Petric, Electrochemical deposition of ceria and doped
ceria films. Ceramics International, 2001. 27(2): p. 149-155.

46.

Kamada, K., N. Enomoto, and J. Hojo, Optimization of electrochemical synthesis
conditions for dense and doped ceria thin films. Electrochimica Acta, 2009.
54(27): p. 6996-7000.

173

47.

48.

Zhitomirsky, I. and A. Petric, Electrolytic and electrophoretic deposition of CeO2
films. Materials Letters, 1999. 40(6): p. 263-268.

49.

Zhang, C.J., et al., Heterogeneous films of ordered CeO2/Ni concentric
nanostructures for fuel cell applications. Physical Chemistry Chemical Physics,
2010. 12(17): p. 4295-4300.

50.

Zivkovic, L., et al., Effect of samarium addition and annealing on the properties
of electrodeposited ceria thin films. Thin Solid Films, 2011. 519(11): p. 35383543.

51.

135.
52.

Elbelghiti, H., et al., Electrodeposition of Ceria on Stainless Steel for SOFC
Interconnect Applications. Solid Oxide Fuel Cells 10 (Sofc-X), Pts 1 and 2, 2007.
7(1): p. 2391-2397

2646.
53.

54.

Kulp, E.A., et al., Electrodeposition of nanometer-thick ceria films by oxidation of
cerium(III)-acetate. Solid State Ionics, 2007. 178(11-12): p. 749-757.

55.

174

56.

57.

Zhou, Y.C., R.J. Phillips, and J.A. Switzer, Electrochemical Synthesis and
Sintering of Nanocrystalline Cerium(Lv) Oxide Powders. Journal of the American
Ceramic Society, 1995. 78(4): p. 981-985.

58.

Inguanta, R., S. Piazza, and C. Sunseri, Template electrosynthesis of CeO2
nanotubes. Nanotechnology, 2007. 18(48): p. -.

59.

Bocchetta, P., M. Santamaria, and F. Di Quarto, From ceria nanotubes to
nanowires through electrogeneration of base. Journal of Applied
Electrochemistry, 2009. 39(11): p. 2073-2081.

60.

Zhitomirsky, I. and A. Petric, Electrochemical deposition of yttrium oxide.
Journal of Materials Chemistry, 2000. 10(5): p. 1215-1218.

61.

Wang, A.Q. and T.D. Golden, Anodic electrodeposition of cerium oxide thin films
- I. Formation of crystalline thin films. Journal of the Electrochemical Society,
2003. 150(9): p. C616-C620.

62.

Golden, T.D. and A.Q. Wang, Anodic electrodeposition of cerium oxide thin films
- II. Mechanism studies. Journal of the Electrochemical Society, 2003. 150(9): p.
C621-C624.

63.

Mahajan, S., et al., Tuning plasmons on nano-structured substrates for NIR-SERS.
Physical Chemistry Chemical Physics, 2007. 9(1): p. 104-109.

64.

Shiu, J.Y., et al., Fabrication of tunable superhydrophobic surfaces by
nanosphere lithography. Chemistry of Materials, 2004. 16(4): p. 561-564.

65.

Boneberg, J., et al., The formation of nano-dot and nano-ring structures in
colloidal monolayer lithography. Langmuir, 1997. 13(26): p. 7080-7084.

66.

Burmeister, F., et al., Colloid monolayers as versatile lithographic masks.
Langmuir, 1997. 13(11): p. 2983-2987.

67.

Spada, E.R., et al., Homogeneous growth of antidot structures electrodeposited on
Si by nanosphere lithography. Journal of Applied Physics, 2008. 103(11): p. -.

175

68.

Haginoya, C., M. Ishibashi, and K. Koike, Nanostructure array fabrication with a
size-controllable natural lithography. Applied Physics Letters, 1997. 71(20): p.
2934-2936.

69.

Dimitrov, A.S. and K. Nagayama, Steady-State Unidirectional Convective
Assembling of Fine Particles into 2-Dimensional Arrays. Chemical Physics
Letters, 1995. 243(5-6): p. 462-468.

70.

Onoda, G.Y. and E.G. Liniger, Experimental-Determination of the RandomParking Limit in 2 Dimensions. Physical Review A, 1986. 33(1): p. 715-716.

71.

Mihi, A., M. Ocana, and H. Miguez, Oriented colloidal-crystal thin films by spincoating microspheres dispersed in volatile media. Advanced Materials, 2006.
18(17): p. 2244-+.

72.

Dimitrov, A.S. and K. Nagayama, Continuous convective assembling of fine
particles into two-dimensional arrays on solid surfaces. Langmuir, 1996. 12(5): p.
1303-1311.

73.

Trau, M., D.A. Saville, and I.A. Aksay, Field-induced layering of colloidal
crystals. Science, 1996. 272(5262): p. 706-709.

74.

Hayward, R.C., D.A. Saville, and I.A. Aksay, Electrophoretic assembly of
colloidal crystals with optically tunable micropatterns. Nature, 2000. 404(6773):
p. 56-59.

75.

Park, S.H., D. Qin, and Y. Xia, Crystallization of mesoscale particles over large
areas. Advanced Materials, 1998. 10(13): p. 1028-+.

76.

Scientific, T. NIST Traceable Particle Size Standards.

77.

78.

Arurault, L., B. Daffos, and F.X. Sauvage, Nanocrystallized ceria-based coatings
prepared by electrochemistry on TA6V titanium alloy. Materials Research
Bulletin, 2008. 43(4): p. 796-805.

176

79.

Zhou, Y.C. and J.A. Switzer, Growth of cerium(IV) oxide films by the
electrochemical generation of base method. Journal of Alloys and Compounds,
1996. 237(1-2): p. 1-5.

80.

Arurault, L., et al., Electrochemical preparation of adherent ceria coatings on
ferritic stainless steel. Thin Solid Films, 2004. 466(1-2): p. 75-80.

81.

82.

83.

Li, G.R., et al., Microstructural evolution of CeO2 from porous structures to
clusters of nanosheet arrays assisted by gas bubbles via electrodeposition.
Langmuir, 2008. 24(8): p. 4254-4259.

84.

Hayes, S.A., et al., The phase stability of cerium species in aqueous systems - I. EpH diagram for the Ce-HClO4-H2O system. Journal of the Electrochemical
Society, 2002. 149(12): p. C623-C630.

85.

86.

Nobial, M., et al., The nitrate reduction process: A way for increasing interfacial
pH. Journal of Electroanalytical Chemistry, 2007. 600(1): p. 87-94.

87.

Giannuzzi, L.A. and F.A. Stevie, A review of focused ion beam milling techniques
for TEM specimen preparation. Micron, 1999. 30(3): p. 197-204.

88.

Jennings, A.T., M.J. Burek, and J.R. Greer, Microstructure versus Size:
Mechanical Properties of Electroplated Single Crystalline Cu Nanopillars.
Physical Review Letters, 2010. 104(13): p. -.

177

89.

90.

Weber, W.H., K.C. Hass, and J.R. Mcbride, Raman-Study of Ceo2 - 2nd-Order
Scattering, Lattice-Dynamics, and Particle-Size Effects. Physical Review B, 1993.
48(1): p. 178-185.

91.

Kosacki, I., et al., Raman scattering and lattice defects in nanocrystalline CeO2
thin films. Solid State Ionics, 2002. 149(1-2): p. 99-105.

92.

93.

Phok, S. and R.N. Bhattacharya, Effect of samarium doping on electrodeposited
CeO2 thin film. Physica Status Solidi a-Applications and Materials Science, 2006.
203(15): p. 3734-3742.

94.

95.

Chang, H.Y. and H.I. Chen, Morphological evolution for CeO2 nanoparticles
synthesized by precipitation technique. Journal of Crystal Growth, 2005. 283(34): p. 457-468.

96.

97.

M. E. Orazem, B.T., Electrochemical Impedance Spectroscopy. 2008: John Wiley
& Sons.

98.

Edwards, J.D. and F. Keller, Formation of Anodic Coatings on Aluminum.
Transactions of The Electrochemical Society, 1941. 79(1): p. 135-144.

99.

Hunter, M.S. and P. Fowle, Factors Affecting the Formation of Anodic Oxide
Coatings. Journal of the Electrochemical Society, 1954. 101(10): p. 514-519.

178

100.

Houser, J.E. and K.R. Hebert, The role of viscous flow of oxide in the growth of
self-ordered porous anodic alumina films. Nature Materials, 2009. 8(5): p. 415420.

101.

Skeldon, P., et al., A tracer study of porous anodic alumina. Electrochemical and
Solid State Letters, 2006. 9(11): p. B47-B51.

102.

Garcia-Vergara, S.J., et al., Compositional evidence for flow in anodic films on
aluminum under high electric fields. Journal of the Electrochemical Society,
2007. 154(9): p. C540-C545.

103.

104.

Fan, Z.Y., et al., Electrical and photoconductive properties of vertical ZnO
nanowires in high density arrays. Applied Physics Letters, 2006. 89(21): p. -.

105.

Yasui, N., A. Imada, and T. Den, Electrodeposition of (001) oriented CoPt L1(0)
columns into anodic alumina films. Applied Physics Letters, 2003. 83(16): p.
3347-3349.

106.

Fusil, S., et al., Nanolithography based contacting method for electrical
measurements on single template synthesized nanowires. Nanotechnology, 2005.
16(12): p. 2936-2940.

107.

Sander, M.S., et al., Template-assisted fabrication of dense, aligned arrays of
titania nanotubes with well-controlled dimensions on substrates. Advanced
Materials, 2004. 16(22): p. 2052-+.

108.

Yang, Y., et al., Anodic alumina template on Au/Si substrate and preparation of
CdS nanowires. Solid State Communications, 2002. 123(6-7): p. 279-282.

109.

Vlad, A., et al., Nanowire-decorated microscale metallic electrodes. Small, 2008.
4(5): p. 557-560.

110.

Chong, M.A.S., et al., Combinational template-assisted fabrication of
hierarchically ordered nanowire arrays on substrates for device applications.
Applied Physics Letters, 2006. 89(23): p. -.

111.

Colvin, V.L., From opals to optics: Colloidal photonic crystals. Mrs Bulletin,
2001. 26(8): p. 637-641.

179

112.

Yu, X.D., et al., Filling fraction dependent properties of inverse opal metallic
photonic crystals. Advanced Materials, 2007. 19(13): p. 1689-+.

113.

Chung, Y.W., et al., Fabrication of various nickel nanostructures by manipulating
the one-step electrodeposition process. Journal of the Electrochemical Society,
2007. 154(6): p. E77-E83.

114.

Bartlett, P.N., T. Dunford, and M.A. Ghanem, Templated electrochemical
deposition of nanostructured macroporous PbO2. Journal of Materials Chemistry,
2002. 12(10): p. 3130-3135.

115.

Bartlett, P.N., et al., Optical properties of nanostructured metal films. Faraday
Discussions, 2004. 125: p. 117-132.

116.

Yan, H.W., et al., Fabrication of 2D and 3D ordered porous ZnO films using 3D
opal templates by electrodeposition. Electrochemistry Communications, 2005.
7(11): p. 1117-1121.

117.

Zhu, R., et al., Controlling the electrodeposition of mesoporous metals for
nanoplasmonics. Nanoscale, 2009. 1(3): p. 355-359.

118.

119.

Wang, T.W., et al., Preparation of a large Mesoporous CeO2 with crystalline
walls using PMMA colloidal crystal templates. Colloid and Polymer Science,
2006. 285(1): p. 1-9.

120.

Waterhouse, G.I.N., et al., Physical and optical properties of inverse opal CeO2
photonic crystals. Chemistry of Materials, 2008. 20(3): p. 1183-1190.

121.

Zhao, J.P., et al., Fabrication of three-dimensionally ordered macroporous
gadolinia-doped ceria films. New Journal of Chemistry, 2008. 32(6): p. 10141019.

122.

Zhou, W., et al., A novel efficient oxide electrode for electrocatalytic oxygen
reduction at 400-600 degrees C. Chemical Communications, 2008(44): p. 57915793.

123.

Kim, S., et al., Anomalous Electrical Conductivity of Nanosheaves of CeO2.
Chemistry of Materials, 2009. 21(7): p. 1182-1186.

180

124.

Yu, K.L., et al., Convenient synthesis of CeO2 nanotubesl. Materials Science and
Engineering B-Solid State Materials for Advanced Technology, 2007. 139(2-3):
p. 197-200.

125.

126.

Jiang, X.C., T. Herricks, and Y.N. Xia, CuO nanowires can be synthesized by
heating copper substrates in air. Nano Letters, 2002. 2(12): p. 1333-1338.

127.

Kumar, A., et al., The effect of growth parameters on the aspect ratio and number
density of CuO nanorods. Journal of Physics-Condensed Matter, 2004. 16(47): p.
8531-8543.

128.

Zhang, K.L., et al., Synthesis of large-area and aligned copper oxide nanowires
from copper thin film on silicon substrate. Nanotechnology, 2007. 18(27): p. -.

129.

130.