ONE-STEP PROCESS FOR SOLID OXIDE FUEL CELL FABRICATION

1
BOSTON UNIVERSITY
COLLEGE OF ENGINEERING
Thesis
ONE-STEP PROCESS FOR
SOLID OXIDE FUEL CELL FABRICATION
by
CHRISTIAN ROBERT SCHUMACHER
B.S., Boston University, 2000
Submitted in partial fulfillment of the
requirements for the degree of Master of Science
2003

2
Approved by
First Reader
Vinod Sarin, Sc.D.
Professor of Manufacturing Engineering
Second Reader
Uday B. Pal, Ph.D.
Professor of Manufacturing Engineering
Third Reader
Srikanth Gopalan, Ph.D.
Assistant Professor of Manufacturing Engineering

4
ACKNOWLEDGEMENTS
First I would like to thank DOE for financial support for this project. I would like
to thank Prof. Sarin for giving me an entrance to the world of research. I would also like
to thank Prof. Pal and Prof.Gopalan for their input and patience.
I am incredibly grateful to Earl Geary, Helmut Lingertatt, and Bob Sjostrom for
their mentoring, willingness to impart their knowledge and sense of humor. Thanks
Guys.
Dr. Chris Manning was always there to lend a helping hand, for advise, for
listening and for laughing. He would always take time out of his busy day to help a
student, he would analyze any problem and help us work though it. He pushed us, he
nudged us, he taught us. This thesis could not have been completed without him.
Many thanks to the graduate students with whom I have spent much time. Ryan,
Ed N., Scott, Tim, Ajay, Amimesh, Sudeep, Sandeep, Ed C., Simly, Saurabh, Jim and
Chris M. I had a great time and I could not have done it without each of you!

5
ONE-STEP PROCESS FOR
SOLID OXIDE FUEL CELL FABRICATION
CHRISTIAN ROBERT SCHUMACHER
ABSTRACT
Solid oxide fuel cells (SOFCs) are electrochemical devices that convert the
chemical energy of a fuel to electrical energy. They offer clean, efficient, and reliable
power and they can be operated using a variety of fuels. There are two major obstacles
for commercializing the SOFC technology: reducing processing cost and increasing
operating lifetime. The cost reduction necessary for commercialization of SOFC
technology will not be met by optimization alone. Mass production techniques must be
developed for SOFCs to be competitive with current power generating devices. The
current cost goal of $400 / kW can only be met by improved processing techniques. Hot
pressing as was investigated at Boston University as a novel processing technique to
fabricate the planar SOFC in a single step. By removing multiple batch processing steps
and simplifying the manufacturing process, considerable cost reduction can be achieved
over current manufacturing processes. Additionally, the flexibility of hot pressing can
improve interfacial contact and functionally grade interfaces to reduce polarization

6
losses. Finally, by optimizing the process, the processing time and cost can be greatly
reduced.
Fundamental knowledge related to ceramic processing, sintering, and hot pressing
to successfully hot press a single operational SOFC in one step has been developed.
Ceramic powder processing for each of the components of an SOFC has been tailored
towards this goal. Processing parameters for the electrolyte and cathode were
investigated and optimized to attain convergence. Several anode fabrication techniques
were investigated and a novel anode structure was developed and refined. Based on these
results single SOFC cells were fabricated in one step.

7
TABLE OF CONTENTS
List of Tables xi
List of Figures xii
1. INTRODUCTION 1
1.1. Fuel Cell Use World Wide 2
1.1.1. Transportation 2
1.1.2. Stationary 4
1.1.3. Remote 5
1.2. Depletion of the World’s Nonrenewable Fuel Reserves 6
1.3. Fuel Cells Overview 7
1.3.1. Fuel Cell Types 8
1.3.2. Development of SOFC as a Superior Design 9
1.3.3. Current Status of SOFCs 11
1.4. Obstacles to Commercialization 11
1.5. Hot Pressing 12
1.5.1. Novel SOFC Processing Technique 14
2. OVERVIEW OF SOFCS 17
2.1. Operational Theory of SOFCs 17
2.1.1. Fuel Cell Performance 19
2.1.2. Activation Polarization 22
2.1.3. Ohmic Polarization 24
2.1.4. Concentration Polarization 24

8
2.2. SOFC Material Component Selection 25
2.3. SOFC Cell Stacks 28
2.3.1. Evolution of SOFC Stack Designs 29
2.3.2. Tubular Designs 30
2.4. Thin Film Fabrication Processes 32
2.5. Layered SOFC Designs 34
2.5.1. Planar SOFC 34
2.5.2. Monolithic SOFC 35
3. THEORETICAL ASPECTS 38
3.1. Ceramic Powder Processing 39
3.1.1. Powder Particle Size 39
3.1.2. Powder Particle Size Distribution 41
3.2. Ceramic Powder Compaction 42
3.2.1. Pressure Distribution in a Powder Compact 42
3.2.2. Derivation of Pressure Variation in a Powder Compact 43
3.3. Sintering Theory 45
3.3.1. Driving Force for Densification 45
3.3.2. Evolution of Sintering 49
3.3.3. Grain Growth 51
3.3.4. Sintering Transport Mechanisms 52
3.4. Hot Pressing Theory 54
3.4.1. Analytical Treatment of Hot Pressing 55

9
3.4.2. Initial Stage Densification Rate 57
3.4.3. Intermediate and Final Stage Densification Rate 60
3.5. Necessity of Experimental Studies 62
4. EXPERIMENTAL DETAILS 63
4.1. Characterization Techniques 63
4.1.1. Particle Size and Distribution Measurement 63
4.1.2. Density Measurement 63
4.1.3. Hardness Measurement 64
4.1.4. Optical Microscopy 65
4.1.5. SEM Analysis 65
4.1.6. Electron Microprobe Analysis 65
4.2. Sample Fabrication 65
4.2.1. Ball Milling 66
4.2.2. Green Samples 66
4.2.3. Sintering Experiments 67
4.2.4. Typical Hot Pressing Cycle 68
4.2.5. Hot Pressing Experiments: Reducing Environment 68
4.2.6. Hot Pressing Experiments: Oxidizing Environment 70
4.3. Electrochemical Characterization 70
5. RESULTS AND DISCUSSION 72
5.1. Experimental Regime 73
5.1.1. Equipment Limits 74

10
5.1.2. Chemical Interactions 75
5.1.3. Established Sintering Cycles 76
5.1.4. Particle Size Considerations 78
5.2. Experimental Approach Summary 80
5.3. Electrolyte Development 80
5.4. Cathode Development 83
5.5. Anode Development 84
5.5.1. Elimination of Ni Reaction Layer 87
5.5.2. Oxidizing Environment 89
5.6. Anode 2: Mo and YSZ 90
5.7. Anode 3: Mo/ 3% Ni and YSZ 92
5.8. Single Cell 94
5.9. Electrochemical Characterization 94
6. CONCLUSIONS 96
7. Appendix A 97
8. Appendix B 100
9. Bibliography 102
10. Vita 106

11
LIST OF TABLES
Table 2.5 State of the Art SOFC Components 26
Table 2.8 Tubular SOFC Fabrication Processes 31
Table 3.4 Consolidation Stages of a Sintered Powder Compact 49
Table 5.5 Percentage of Driving Force Due to Applied Pressure 79
Table 5.7 Relative Density Estimations as a Function of Hot Pressing
Temperatures for Hot Pressed YSZ Powder
81

12
LIST OF FIGURES
Figure 1.1 Comparison of Developing Fuel Cells 8
Figure 1.2 Selected Energy Generating Devices 10
Figure 1.3 Cost Comparison of Selected Power Generating Systems 12
Figure 1.4 Schematic Representation of SOFC Hot Pressing Technique 14
Figure 1.5 Schematic Representation of Novel Hot Pressing Technique 15
Figure 1.6 Evolution of Hot Press Technique into High Yield
Manufacturing Process
16
Figure 2.1 Conceptual Schematic of Fuel Cell 18
Figure 2.2 H2/ O2 Fuel Cell Ideal Potential as a function of Temperature 20
Figure 2.3 Ideal and Actual Fuel Cell Voltage/ Current Characteristic 21
Figure 2.4 Example of a typical Tafel plot 23
Figure 2.6 SEM Micrographs of a Convention SOFC 27
Figure 2.7 Siemens Westinghouse Tubular SOFC 31
Figure 2.9 Tape Casting 32
Figure 2.10 Tape Calendaring 33
Figure 2.11 Planar SOFC in Cross-flow Configuration 34
Figure 2.12 Monolithic SOFC in Co-flow and Cross-flow Configurations 36
Figure 2.13 Firing Shrinkage as a Function of Temperature for Monolithic
Powders
37
Figure 3.1 Pressure distribution in a Powder Compact Disc 44
Figure 3.2 Chemical Potential as a Function of Surface Curvature 47
Figure 3.3 Equivalent Stress as a Function of Particle Radius 48
Figure 3.5 Schematic of Consolidation Stages of a Sintered Powder
Compact
50

13
Figure 3.6 Sintering Transport Mechanisms 53
Figure 3.7 Particle Size and Pressure Regime in Hot Pressing 56
Figure 3.8 Various Hot Pressed YSZ Powders 56
Figure 3.9 Initial Stage Densification of Sintering Compacts 58
Figure 4.1 Typical Hot Pressing Cycle 69
Figure 5.1 Experimental Approach to Hot Pressing a Single Cell 73
Figure 5.2 Experimental Operating Regime 74
Figure 5.3 Practical SOFC Hot Pressing Experimental Regime 75
Figure 5.4 Practical SOFC Hot Pressing Experimental Regime 77
Figure 5.6 Summary of Experiments 81
Figure 5.8 Optical Microscopy of Electrolyte Cross Section of Hot Pressed
YSZ Powder
82
Figure 5.9 SEM Micrograph of Hot Pressed Cathode Powder 83
Figure 5.10 Equilibrium Oxygen Partial Pressure of Ni and NiO as a
Function of Temperature
87
Figure 5.11 Various Hot Pressed Anode Powders 88
Figure 5.12 NiO Powder Hot Pressed in an Oxidizing Environment 89
Figure 5.13 Hot Pressed Mo/ YSZ Anode Powder and YSZ Electrolyte 91
Figure 5.14 Mo – Ni Binary Phase Diagram 93
Figure 5.15 Mo-Based Anode and Electrolyte Half Cells 93
Figure 5.16 Mo-Based Anode, Single Cell, in a Reducing Environment 95

14
Chapter 1
INTRODUCTION
Dependable, plentiful, and economical energy has been the driving force for
financial, industrial, and political growth in the United States since the mid 19th
century.
For a country whose progress is so deeply rooted in abundant energy and whose current
political agenda involves stabilizing world fossil fuel prices, the development of a
reliable, efficient and environmentally friendly power generating source seems
compulsory. The maturing of high technology fuel cells may be the panacea the country
will find indispensable to free itself from foreign dependence. Fuel cells offer an
efficient, combustion-less, virtually pollution-free power source, capable of being sited in
downtown urban areas or in remote regions. Fuel cells have few moving parts and run
almost silently.
Fuel cells are electrochemical devices that convert the chemical energy of a fuel
directly to electrical energy. Unlike batteries, which store a finite amount of energy, fuel
cells will generate electricity continuously, as long as fuel and oxidant are available to the
electrodes. Additionally, fuel cells offer clean, efficient, and reliable power and they can
be operated using a variety of fuels. Hence, the fuel cell is an extremely promising
technology.

15
Fuel Cell Use World Wide
Fuel cells were developed for the space program and have been used since the late
1950s to provide electricity and drinking water for astronauts. Terrestrial applications
can be classified into categories of transportation, stationary or remote generation uses.
Transportation
Automobile emissions remain one of the largest contributors to urban air
pollution. Worldwide, over one billion people living in urban areas are suffering from
poor air quality, leading to over 700, 000 deaths annually [1]. Estimates from the EPA
[2] indicate that motor vehicles in the US still account for
 78% of all Carbon Monoxide emission
 45% of Nitrogen Oxide emissions
 37% of Volatile Organic Compounds
Offsetting the impact of modern lower emission gasoline vehicles engines are the
growth in the number and size of vehicles on the road, as well as an increase in the
number of miles each vehicle travels. If recent trends continue, Americans will be
driving twice as many miles in 2015 as today [3]. Low gasoline prices also do not

16
encourage fuel efficiency or conservation. Rather, the number of sport utility vehicles
market share is continuing to rise therefore increasing the number of less efficient
vehicles on the road.
Bills such as the 1990 Clean Air Act and the National Energy Policy Act of 1992
paved the way for less polluting gasoline vehicles and the introduction of alternative fuel
vehicles on the roads in the US. Also, in 1990, the California Air Resources Board
recognized that even the cleanest gasoline powered vehicles wouldn’t satisfy the state’s
goals for higher air quality. Meeting state and federal air standards in seriously polluted
area such as Los Angeles would require either restrictions on driving or a large scale
switch to vehicles that don’t pollute. California adopted the Low Emission Vehicle Act
that mandated the seven largest auto makers begin immediately to reduce all tailpipe
emissions and to introduce zero emission vehicles (ZEVs) starting in 1998. Beginning in
2003, 10% of new vehicles sold in CA will be required to be ZEVs or nearly ZEVs - also
known as “equivalent ZEVs” [4]. The California Air Resources Board believes, “the
remarkable developments of fuel-cell engines will help California in its war on smog as
well as provide new consumer choices for transportation.”
Because of the legislative initiative taken by California and subsequent similar
regulations imposed by a number of states in the Northeast, every major automobile
manufacturer has made significant progress toward the development of ultra-low and
zero emission vehicles. This legislation has promoted and encouraged the development
of fuel cells as an energy source. Fuel cells operate at 2-3 times the efficiency of modern
internal combustion engines. An improvement of 1-2 miles per gallon (MPG) is

17
considered substantial in the automotive industry, if fuel cells could directly replace the
car engine, it would mean a 25 to 50 MPG increase in fuel efficiency.
Stationary
The utility sector is expected to be the first areas where fuel cells will be widely
commercialized. Today, only about one-third of the electrical energy produced reaches
the user because of the low energy conversion efficiencies of power plants. Using fuel
cells for utility applications will improve energy efficiencies by as much as 69% while
reducing emissions [5].
Fuel cell commercialization opportunities in the U.S. market are focused in
several large-scale areas: re-powering, central power plants, industrial generators, and
commercial/ residential generators.
The Department of Energy (DOE) predicts that 403 GW of new generating
capacity will be required by 2020 in the United States to replace existing units and meet
growing demand [6]. Both restructuring and rising capacity demands pose new
opportunities for fuel cell adoption in applications as diverse as merchant power plants,
industrial cogeneration plants, data center backup power, and even residential “garage
power” uses.
SOFCs are considered the only fuel cell technology with a wide span of possible
market applications ranging from 2-kilowatt residential systems to wholesale distributed
generation systems of 10-25 megawatts.

18
Remote
Remote or “distributed power” is a new approach utility companies are beginning
to implement – small power generators located at or near the site of consumption provide
advantages for traditional utilities that own both generation and delivery systems. By
placing the power source near the load, the generator avoids energy losses—up to 5
percent of energy transported—via heat dissipation along the transmission and
distribution lines.
Because fuel cells are modular in design and highly efficient, these small units
can be placed on-site. Installation is less of a financial risk for utility planners and
modules can be added as demand increases. Utility systems are currently being designed
to use regenerative fuel cell technology and renewable sources of electricity. Fuel cells
are becoming an alternative choice for rural energy needs in places where there are no
existing power grids or where the power supply is often unreliable.
In a study released in July 2001, the Electric Power Research Institute (EPRI)
estimates that power interruptions and disturbances are responsible for as much as $188
billion in losses every year [7]. This figure can only increase, as more and more sensitive
equipment becomes an integral part of the overall economy. It is no longer possible to
ignore the implications of the reliability and overall quality of our electrical power
supply. Interruptions and disturbances in electric supply can shut down, damage, or even
destroy equipment. A $10 billion global industry has arisen to provide energy consumers
with the solutions they need to protect their operations and property [8].

19
Depletion of the World’s Nonrenewable Fuel Reserves
The world’s production of oil reached a record level of 65 million barrels a day in
1997 and global demand is rising more than 2% per year. Americans spend roughly
$100, 000 per minute to purchase foreign oil [9], and the US transportation sector uses
over 10% of the world’s oil [9]. Consumption of oil by passenger vehicles, which
include automobiles and light duty trucks, exceeds all of the United States’ domestic
production of oil. Reserves of fossil fuels are large but finite, and there is growing
evidence to suggest that world production of crude oil will peak in the early 21st
century
[10]. Since 1985 energy use worldwide has increased 40% in both Latin America and
Africa and 50% in Asia [10].
The Energy Information Agency forecasts that worldwide demand for oil will
increase 60% by 2020[11]. By 2010, Middle East OPEC states, considered unpredictable
and often unstable, will have over 50% of the world oil business, and the switch from
growth to decline in oil production could cause economic and political tension [12]. As
excess oil production capacity begins to decline over the coming decades, oil prices can
be expected to rise. The transportation sector is likely to be most heavily affected by
these fluctuations. World wide, transportation relies almost exclusively on oil, and there
are few viable short-term options.
The introduction of fuel cells will increase fuel efficiency, decrease foreign oil
dependency, and become an important strategy/ technology to mitigate climate change.
In the future the combination of high efficiency fuel cells and fuels from renewable

20
energy sources would nearly eliminate greenhouse gas emissions. The early transition to
hydrogen fuel will lead to cleaner air and stronger national energy security.
Fuel Cells Overview
A commercially useful fuel cell consists of many single cells. These cells
individually generate small currents and voltages, however when configured in series and
parallel, large voltages and currents are possible.
The single cell consists of an electrolyte sandwiched between two porous
electrodes. Dense, electrically conductive, bi-polar plates (or interconnects) allow
individual cells to be connected without shorting anode to cathode.
Fuel cells separate the direct combustion of oxidant and fuel into two separate
electrochemical reactions. By separating these reactions, electrons participating in the
half-cell reactions can be routed through an external circuit providing electricity for a
myriad of purposes. In conventional energy generating devices, direct combustion
produces heat, which is converted into mechanical energy, and finally to electricity. Fuel
cells having direct energy conversion offer an inherently more efficient power production
method by eliminating mechanical energy conversion losses typical of convention power
generating devices.
Fuel cells offer a clean, ultra-low pollution technique to electrochemically
generate electricity at high efficiencies. Fuel cells provide many advantages over
traditional energy-converting systems, including high efficiency, reliability, modularity,

21
fuel adaptability, and very low levels of NOx and SOx emissions. The quiet, vibration-
free operation of fuel cells also eliminates the noise usually associated with conventional
power generating systems.
Fuel Cell Types
There are several types of fuel cells, typically classified by the nature of the
electrolyte; these include Polymer Electrolyte Membrane (PEM), Alkaline (AFC),
Phosphoric Acid (PAFC), Molten Carbonate (MCFC) and Solid Oxide (SOFC). A
summary of the different fuel cells can be seen in Table 1.1.
Figure 1.1: Comparison of Developing Fuel Cells [13].

22
Among these fuel cells types, the PEM is being developed mainly for space and
transportation applications, and the AFC is an important power source on space flights.
The PAFC is presently at the initial stage of commercialization for electric utility and
cogeneration uses. The MCFC is the next most likely candidate for commercialization,
whereas the SOFC is considered third-generation, or the youngest fuel cell technology in
development.
Development of SOFC as a Superior Design
The current commercial leader in electrical energy generating technology is the
gas turbine power plant, due to the low capital cost associated with the device. Fuel cells
are compared against the gas turbine and other large scale energy generating devices in
Table 1.2. Although the gas turbine is presently the least capital intensive, the long term
operating cost is much higher both economically and environmentally [14].
SOFCs show the greatest potential for efficient power generation. High operating
temperatures promote rapid reaction kinetics, allow reforming of hydrocarbon fuels
within the fuel cell, and produce high quality byproduct heat suitable for use in
cogeneration or bottoming cycles. A bottoming cycle is used to generate additional
electricity using fuel cell byproduct heat in a gas turbine or steam engine. Cogeneration
uses the fuel cell byproduct heat for space heating, to supply hot water, or to generate
steam for industrial purposes. Configured in the co-generation format, efficiencies
approaching 70% are projected [13].

23
Figure 1.2: Selected Energy Generating Devices [13]
Also, SOFCs exhibit high power-to-weight ratio since they are made of light-
weight-thin-film ceramic materials. Lower manufacturing times are obtained for SOFCs
since the units are modular in nature and can be assembled on site; and the solid-state
structure can be easily transported as compared to alternative fuel cells.
SOFC additionally have several advantages over alternative fuel cells mostly
arising from the use of a solid electrolyte. The use of a solid electrolyte in ceramic fuel
cells eliminates material corrosion and electrolyte management problems and permits
unique cell designs with performance improvements over other fuel cells. The
conductivity requirement for ceramic electrolytes necessitates high operating
temperatures (600° to 1000°C). Because of the high operating temperature of SOFCs,
natural-gas and other hydrocarbon fuels can be reformed within the cell stack eliminating
the need for expensive external reformer system.

24
Current Status of SOFCs
To date, although still evolving, SOFCs have made excellent technical progress.
Multi-kilowatt fuel cells incorporating various features of a practical power generation
system have been operated for thousands of hours and have shown excellent
performance. Recently, ceramic fuel cell research and development has received much
attention, reflecting widening interest in this technology.
Obstacles to Commercialization
The capital cost associated with SOFCs is still prohibitively high. Fig. 1.3 offers
a comparison of the capital costs of several power generating systems. The most widely
marketed fuel cell, an AFC, by International Fuel Cell, is only used where money is not
an issue, i.e., for continuous power demands and in space applications, it is not a price
level where in can reach the mass market.
There are two major obstacles for commercializing the Solid Oxide Fuel Cell
technology: reducing processing cost and increasing operating lifetime. Cost problems
arise from expensive and time intensive batch processing, while operating lifetime issues
result from material incompatibilities (thermal expansion mismatch) causing thermal
cycling problems such as delamination, thermal shock, and distortion or cracking.
The cost reduction necessary for commercialization of SOFC technology will not
be met by optimization alone. Mass production techniques must be developed for SOFCs

25
to be competitive with current power generating devices. There is a limit to the premium
that customers will be willing to pay for environmentally friendly power or even higher
efficiencies. The current cost goal of $400 per kilowatt can only be met by improved
processing techniques.
Figure 1.3: Cost Comparison of Selected Power Generating Systems [13].
Some of the most energy intensive and expensive fabrication processes, for any
SOFC stack design, are the high temperature sintering steps. Batch processes, in the
tubular and planar designs, use repeated thermal cycling steps to sinter successive layers
upon previous ones. The monolithic design improves upon this approach by co-firing all
layers in a cell in one step, thereby eliminating the batch processing. However, the
monolithic design incorporates extensive post process non-destructive testing and
evaluation due to the complex geometries and multiple shrinkage rates encountered
during processing. This extraneous post testing substantially increases the overall
fabrication cost.

26
Therefore an improved fabrication process is desired, which would reduce or
remove the number of separate batch processes required to produce a SOFC without
adding extreme post process testing.
This is the approach currently being undertaken at Boston University (BU).
Researchers at BU have shown preliminary results using an alternative ceramic
processing technique to eliminate batch, without the added geometrical complexity of the
monolithic design.
Hot Pressing
Hot pressing is the simultaneous application of elevated temperature and
compressive stress to consolidate fine green pressed powders into partially or fully
sintered components. The technique, shown in Fig. 1.4, was developed for the powder
metallurgy industry and has been successfully applied to ceramic components over the
last several decades [14, 15]. Pressure increases the driving force for densification, in
effect, reducing the processing temperature required for a sintering process. Also, hot
pressing results in smaller overall grain size, more precise control over the microstructure
and the flexibility of functionally grading the ceramic layers.

27
Figure 1.4: Schematic Representation of SOFC Hot Pressing Technique.
Novel SOFC Processing Technique
Hot pressing as was investigated at Boston University a novel processing
technique to fabricate the planar SOFC in a single step. The densification parameters of
each individual component were individually determined. Then, by optimizing the
densification parameters for the different layers, the entire fuel cell was hot pressed in a
single step. By removing multiple batch processing steps and simplifying the
manufacturing process, considerable cost reduction can be achieved over current
manufacturing processes. Additionally, the flexibility of hot pressing can improve
interfacial contact and allows for functional grading of interfaces to reduce polarization
losses. Finally, by optimizing the process, the processing time and cost can be greatly
reduced. Hot pressing has never before been applied to the fabrication of SOFCs. A
schematic representation of how a single cell can be processed is shown in Fig. 1.5.

28
Figure 1.5: Schematic Representation of Novel SOFC Hot Pressing Technique
Although reducing the processing to one step, a commercial hot press is a very
capital intensive investment. Hot pressing also requires long heat-up times (±5°C/min)
leading to a relatively slow cycle time. To overcome these limitations it is proposed to
fabricate many cells at once, this is shown schematically in Fig. 1.6. In this schematic
many single cells, consisting of anode, cathode, and electrolyte, will be pressed between
inert spacers. Since each cell can be individually tested, one bad cell will not ruin an
entire stack, as in the monolithic design allowing for a reduction in scrap rate.
With a typical commercially available hot press, one can easily press a single cell
area of 25cm x 25cm (625 cm2
). Furthermore, since the thickness of each cell is not
expected to exceed 4 to 5 mm, at least 20 cells can be fabricated at one time in a single
hot-pressing operation. Operated continuously one hot press one could effectively
produce a total cell area of 37,500 cm2
/ day. It is therefore reasonable to project that with

29
3 to 4 hot presses one could obtain 12-15 m2
of cell area per day. Thus hot pressing can
be a high-volume one step manufacturing process that will result in lower cost, lower
processing time, improved interfacial contact (lower polarization losses) and lower
thermal stresses at the interfaces as a result of compositionally grading the interfaces.
The purpose of this research is to develop the fundamental knowledge related to
ceramic processing, sintering, and hot pressing which must be known to successfully hot
press a single operational SOFC in one step.
Figure 1.6: Evolution of Hot Press Technique into High Yield Manufacturing Process.

30
Chapter 2
OVERVIEW OF SOFCS
Operational Theory of SOFCs
Solid Oxide Fuel Cells convert hydrogen gas, carbon monoxide or hydrocarbon
fuel into water vapor and/ or carbon dioxide. The basic physical structure or building
block of a fuel cell system consists of an electrolyte layer in contact with a porous anode
and cathode on either side. A schematic of a SOFC, showing relevant half reactions and
circuit flow, is shown in Fig. 2.1. There are four main components to a SOFC: two
electrodes (anode and cathode), electrolyte and interconnect, also called a bipolar plate
(not shown).
Oxygen ions are transported from a high oxygen partial pressure region to a lower
one. A low oxygen partial pressure is established at the anode by mixture of hydrogen
and 3-5% water vapor. At the cathode, generally atmospheric molecular oxygen is
reduced to ionic oxygen. Oxygen ions are then able to diffuse through a thin electrolyte
to the anode, along the oxygen electrochemical potential gradient. The oxygen ions are
oxidized at the anode/ electrolyte interface, forming water vapor and free electrons.
These electrons are transferred back to the cathode in an external circuit, providing useful
electrical power.

31
It is critical that the electrolyte is strongly ionically conductive and not
significantly electronically conductive. A fully dense electrolyte also assures that the
oxygen gas side and fuel gas side will not mix directly. Mixing of fuel and oxygen would
reduce the gradient in oxygen potential across the cell, reducing the cell voltage.
A three-phase boundary is established among the reactants, electrolyte, and a
catalytically active current collector in the region of the porous electrode. The nature of
this boundary plays a critical role in the electrochemical performance of a fuel cell. Two
sources of performance loss in the fuel cell are gas phase mass transport in the porous
electrode and resistive losses in the electrolyte. These losses lower the limiting current
density of the cell.
Figure 2.1: Conceptual Schematic of Fuel Cell

32
Thus, a delicate balance must be maintained among the electrode, electrolyte and
gases phase in the porous electrode structure. Recently considerable effort in the
development of fuel cell technology has been devoted to reducing the thickness of cell
components. Refinements and improvements to the electrode structure and the
electrolyte phase have been made to obtain higher and more stable electrochemical
performance.
Fuel Cell Performance
The impact of variables, such as temperature, pressure, and gas composition, on
fuel cell performance needs to be assessed to predict how the cells will respond to
changes and optimizations in processing.
The ideal voltage of a fuel cell is defined by its Nernst potential, represented as a
cell voltage, which can be calculated from the Nernst equation. The Nernst equation
provides a relationship between the ideal standard potential, EO, for the cell reaction and
the ideal equilibrium potential, ER, at other temperatures and partial pressures of
reactants, Preact, and products, PProd. These potentials are related by the gas constant, R,
temperature, T, Faraday’s Constant, F, and the natural log of the oxygen partial pressures
at the anode and cathode. These are related as shown in eq. 2.1.
)(2
)(2
ln
4 a
c
O
O
oR
P
P
F
RT
EE  (2.1)

33
The ideal standard potential of an H2/O2 fuel cell, EO, is 1.229 volts with liquid
water product and 1.18 volts with water as a gaseous product. This value is known as the
oxidation potential of H2 [17]. The potential across the electrodes is directly related to
the change in Gibbs free energy for the reaction of hydrogen and oxygen.
The relation of EO to cell temperature is shown in Fig. 2.2. As the operating
temperature increases the reversible potential decreases, therefore operation at lower
temperatures gives a theoretical advantage in output voltage.
Figure 2.2: H2/ O2 Fuel Cell Ideal Potential as a function of Temperature [after 18].
Useful work (electrical energy) is obtained from a fuel cell only when reasonable
current is drawn. However the actual cell potential is decreased from its equilibrium
potential because of irreversible losses as this current is drawn, this effect is shown in
Fig. 2.3. Several sources contribute to irreversible losses in fuel cells. These losses,
called polarization, overpotential, or overvoltage, , originate primarily from three
sources: activation polarization, act, ohmic polarization, ohm, and concentration

34
polarization, con. These losses result in a cell voltage, V, for a fuel cell that is less than
its ideal potential, E, as shown in eq. 2.2.
TotalEV  (2.2)
The activation polarization loss is dominant at low current density. Charge
transfer barriers have to be overcome prior to current and ion flow. Activation losses
show some increase as current increases.
Figure 2.3: Ideal and Actual Fuel Cell Voltage/ Current Characteristic [after 18].
A gas transport loss, or concentration polarization, occurs assuming that gas
transport is slower than the rate of O2-
ion consumption. Concentration polarization
occurs over the entire range of current densities, but these losses become prominent at
high limiting currents where it becomes difficult to provide reactants to and from the cell
Limiting
Current
Density

35
reaction sites. Ohmic polarization varies directly with current, increasing over the whole
range of current because the area specific resistance of the cell remains essentially
constant. The definition of the ohmic polarization is simply the product of cell current
density and cell area specific resistance. These losses will be further elaborated on
below.
Activation Polarization
Activation polarization is present when the rate of an electrochemical reaction at
an electrode surface is controlled by slow charge-transfer kinetics at the electrodes.
Activation polarization is directly related to the rate of electrochemical kinetics.
If the electrochemical reaction is well stirred or currents are kept so low that the
surface concentrations do not differ appreciably from the bulk values, the current, i, can
be described by the general Butler-Volmer Equation (eq.2.3). Where  is the electron
transfer coefficient of the reaction at the electrode being addressed, io is the exchange
current density, n is number of moles, while F, R, and T remain the same as in eq. 2.1.
 





 







RT
F
i
RT
F
ii Act
o
Act
o
 1
expexp (2.3)

36
For large positive or negative values of , one of the exponential terms in eq. 2.3
becomes negligible. Then eq. 2.3 can be simplified to the Tafel equation (eq. 2.4, or 2.5).
For example, at large negative overpotentials, exp [-nf] >> exp [(1-) nf] yields,
 nfii o  exp (2.4)
or ibai
nf
RT
i
nf
RT
o loglnln 

 (2.5)
Tafel plots (Fig. 2.4) provide a visual understanding of the activation polarization
of a fuel cell. They are used to measure the exchange current density and the electron
transfer coefficient. The exchange current density is given by the extrapolated intercept
at Act = 0 which is a measure of the maximum current that can be extracted at negligible
polarization. The transfer coefficient is simply equal to the slope of the plot.
Figure 2.4: Example of a typical Tafel plot [after 18].

37
The Tafel equation describes the slow kinetics that can be associated with
electrochemical processes. Processes that contribute to slow kinetics include adsorption
of reactant species, transfer of electrons across the double layer, desorption of product
species, and the catalytic activity of the electrode surface can all contribute to activation
polarization.
Ohmic Polarization
Ohmic losses occur because of resistance to the flow of ions in the electrolyte and
resistance to flow of electrons through the electrode materials. The dominant ohmic
losses, through the electrolyte, are reduced by decreasing the electrolyte thickness,
enhancing the ionic conductivity of the electrolyte, and increasing the operating
temperature. Because both the electrolyte and fuel cell electrodes obey Ohm's law, the
ohmic losses can be expressed by eq. 2.6, where i is the current flowing through the cell,
and R is the total cell resistance, which includes electronic, ionic, and contact resistance.
iROhmic  (2.6)
Concentration Polarization
As a reactant/ product is consumed/ produced at the electrode by electrochemical
reaction, a gradient in electrochemical potential occurs due to the inability of the
surrounding fluid to maintain the initial concentration of reactant/ product. That is, gas

38
phase mass transfer leads to the formation of a concentration gradient. Concentration
polarization, conc, is represented (as a voltage) mathematically in eq. 2.7, where iL is the
limiting current, as shown in Fig. 2.4 above.







l
conc
i
i
nF
RT
1ln

 (2.7)
The net result of current flow in a fuel cell is to increase the anode oxygen
chemical potential and to decrease the cathode chemical potential, thereby reducing cell
voltage. For further information on electrochemical testing please see Bard and Faulkner
[19].
SOFC Material Component Selection
The material property requirements for high-temperature (600-1000°C) Solid
Oxide Fuel Cells are quite stringent. The electrolyte must have high oxygen ion
conductivity (0.1-0.01 S/cm), negligible electronic conductivity, be stable in both
oxidizing and reducing conditions and remain dense and impervious to gas. The porous
and gas-permeable cathode and anode must have high electronic conductivity and charge
transfer/surface exchange kinetics (>10-7
cm/s). The cathode must be stable in oxidizing
conditions, while the anode must be stable in reducing conditions and both must be
chemically, mechanically, and structurally compatible with the electrolyte and

39
interconnect materials. The interconnect material must be an electronic conductor,
remain dense and impervious to separate the anodic and the cathodic regions, be stable in
both reducing and oxidizing conditions, and be chemically, mechanically and structurally
compatible with the anode and the cathode materials.
SOFC materials have been studied extensively in the development of these
devices [20]. Current state of the art SOFCs components are shown in Table 2.5. A
convention SOFC electrolyte is 8 molar percent (mol %) yttria stabilized zirconia (YSZ),
a known oxygen ion conductor with negligible electronic conductivity at SOFC operating
temperatures.
The anode is fabricated from a nickel (Ni)/ YSZ cermet. For proper operation the
anode must contain minimum 30 volume percent (vol %) Ni and 20-40% porosity for gas
permeability.
Table 2.5: State of the Art SOFC Components.

40
The cathode is made from A-site, i.e., La site, doped lanthanum manganite
(LaMnO3). Typically strontium is used as the dopant, in which case it is abbreviated
LSM. The cathode also requires 20-40% porosity for gas phase mass transfer. Large
internal electrode surface area, i.e., controlled, fine, uniformly distributed porosity, is also
desired to enhance reaction kinetics.
Finally, the interconnect is made of A/B site doped lanthanum chromate
(LaCrO3). LaCrO3 is chemically stable in reducing and oxidizing atmospheres.
(a) (b)
Figure 2.6: SEM Micrograph of a Conventional SOFC [21].
Cross sections of the state of the art Siemens-Westinghouse SOFC are shown in
Fig. 2.6a and Fig. 2.6b. In these secondary electron (Fig. 2.6a) and backscattered
electron (Fig. 2.6b) images, the cathode, electrolyte and anode are labeled AE (air
electrode), E (electrolyte), and FE (fuel electrode), respectively. The fully dense YSZ
electrolyte is approximately 60-80 m in thickness. Also note the irregular boundaries

41
between layers, which increase reaction surface area and help to promote faster reaction
kinetics. It is evident from these micrographs that pores sizes of approximately 5 to 25
m are typical or acceptable. This cell is cathode supported. However, anode supported
cell are under development. This is represented in the large thickness ranges shown for
the electrodes, in Table 2.5.
SOFC Cell Stacks
As with batteries, individual fuel cells must be combined to produce appreciable
voltage levels and so are joined by interconnects. A single SOFC cell produces
approximately 1 Volt. To obtain practical voltages and currents the cells must be
connected in series and parallel configurations to build voltage and current. These
combinations of blocks are called stacks. Each of these stacks must be contained and
regulated so that the correct type and amount of fuel, oxidant, and heat are maintained.
Additionally, a SOFC stack must be designed to achieve the desired electrical and
electrochemical performance, mechanical integrity, and manifolding requirements.
Design goals for a SOFC stack include:
1. acceptable electrical and electrochemical performance
a. minimize ohmic and polarization losses in cell
2. adequate mechanical integrity
a. strength for assembly and movement
b. maintain electrical contact and gas seals

42
c. handle thermal shock associated with heat up, cool down, and
unexpected failures
3. allow for gas manifolding requirements
a. supply reactant gases
b. remove product and unused gases
4. designed for manufacturability
5. maintain a realistic cost frame
Before improvements to cells or cell processing can be started, the cell stack
design must be decided upon. There are four SOFC stack designs that are in use today,
segmented in series, tubular, planar and monolithic designs. A brief history of the
evolution of SOFC stack design will be presented in the next section.
Evolution of SOFC Stack Designs
Solid Oxide Fuel Cells have undergone numerous design and processes iterations
since their initial development work in the 1960’s. The original designs were planar
discs or a segmented in series design. This was a bell and spigot configuration in which
short tubular segments of electrolyte were joined by conducting seals. These early
designs were supported by the electrolytes. Electrolytes thick enough to support a cell
lead to high internal resistances.
In the 1970’s a banded design, similar to the segmented in series design, was
proposed. This design featured much thinner electrolytes. Development of the banded
design is ongoing even today. In 1980, the sealless tubular design was proposed. This

43
design, a modification of the segmented in series design, incorporated a cylindrical thin
wall electrolyte. The cylindrical configuration allows for the sealless nature of the
design. In 1982, the monolithic design was proposed and developed by Argonne
National Labs. In this design the cells are arranged in a honeycomb like structure, where
the cells act as a baffle and manifold in addition to their normal duties [22].
Currently, the sealless tubular design, having surpassed the segmented in series
design, is the most advanced. Numerous improvements to the tubular design have been
made in the last 5 -10 years. Also, there has been renewed interest in the flat plate, or
planar design, arising from new advances in ceramic processing and forming. The
monolithic design, although having extraordinarily high power density has seems to be
reaching fewer and fewer developmental milestones [23].
Only the tubular, planar and monolithic designs are relevant to the current
research, these designs will be elaborated on in the following sections.
Tubular SOFCs
The most advanced construction for SOFCs is the Siemens -Westinghouse tubular
design (Fig. 2.7). Multi kilowatt stacks have been fabricated and operated for thousands
of hours. The tubular construction can be assembled into large units without seals. The
sealless nature and its robust design are this configurations biggest engineering
advantages.

44
Figure 2.7: Siemens Westinghouse Tubular SOFC [24]
Expensive batch fabrication techniques like Electrochemical Vapor Deposition
(EVD) for the electrolyte, as shown in Table 2.8, result in a fairly high cost for this type
of cell. The use of such exotic processes necessitates the need for control over a number
of parameters, resulting in complicated and expensive stack manufacturing. In addition
to the financial burden imposed, these manufacturing techniques raise the difficulty of
adapting such a system on a commercial scale. The tubular geometry of these fuel cells
limits the specific power density, both on weight and volume basis, to low values while
the electron conduction paths are long and lead to high energy losses due to internal
resistance heating. For these reasons, other configurations are actively being pursued at
the present time.
COMPONENTS MATERIAL FABRICATION PROCESS
Air Electrode
Electrolyte
Interconnection
Fuel Electrode
Sr doped LaMnO3
YSZ
Mg Doped LaCrO3
NI-YSZ Cermet
Extrusion –Sintering
Electrochemical Vapor Deposition (EVD)
Plasma Spraying
Dip Coating followed by Sintering
Table 2.8: Tubular SOFC Fabrication Processes.

45
Before the planar and monolithic designs can be explained fully, current thin
ceramic film fabrication processing used in their construction must be briefly presented.
Thin Film Fabrication Processes
A common ceramic thin film fabrication process is tape casting, shown in Fig.
2.9. A ceramic slip, or slurry is made by combining the desired processed powder, with a
solvent, typically an alcohol, and organic surfactants, dispersants, and stabilizers. This
solution is then spread over a tempered glass bed which is coated with a carrier film. The
thickness of the slurry coating is controlled by two or more doctoring blades which
spread and scrape excess slurry from the bed. This green layer is then carefully heated, to
remove organics, and fired in a controlled sintering cycle. These layers are typically
applied individually, demanding a batch processing approach to layer fabrication.
Figure 2.9: Tape Casting [20].

46
A second ceramic thin film fabrication process is tape calendaring, shown in Fig.
2.10. In this process similar ingredients as tape casting are kneaded into a much thicker
semi-fluid state. This dough like slurry is then fed between rollers yielding a thin 2D
structure. A second rolling operation can mechanically bond individual layers into a
single multilayer structure. This structure can be cut, laminated, corrugated, or formed
prior to firing. Compression molding is used to make the corrugated structure. For a
reliable, robust layer, a carefully controlled combination of temperature, pressing
pressure, pressing time, material consistency, and tape thickness must be used.
Figure 1.10: Tape Calendaring [20].
The advent of these and similar processes have allowed rapid improvement in the
manufacturability and quality of thin ceramic layers.

47
Layered SOFC Designs
Layered designs are some of the most simple, yet they can be the most cost
efficient and most manufacturable. They are fabricated using standard thin film ceramic
processes such as tape casting, for planar cells, and tape calendaring, for monolithic cells.
Planar SOFC
The most common alternative construction to the tubular design is the planar
design, which resembles a cross-flow heat exchanger, shown in Fig. 2.11. The planar
cross flow fuel cell is built from alternating flat single cell membranes, which are tri-
layer anode/ electrolyte/ cathode structures, and interconnection plates, which conduct
current from cell to cell and provide channels for gas flow.
Figure 2.11: Planar SOFC in Cross-flow Configuration [20].

48
The planar SOFC configuration requires high-temperature gas seals at the edges
or inside the gas ports of the plates. The single cell membrane and the interconnect of
every cell must be hermetically sealed at each manifold face to the stack, to prevent cross
leakage. Fuel and oxidant gas cross leakage compromises fuel cell efficiency by
reducing the oxygen chemical potential gradient across the cell (see eq. 2.1).
Compressive seals, cement seals, and glass seals are typically used. The latter is a
sinterable paste, which contains glass particles and whose composition is adjusted so that
it wets the material of the cell components well. Sealant materials with thermal
expansion coefficient matching those of the stack and with a satisfactory lifetime at the
operating temperature are under development at the present time.
Sealing issues represent a serious technical shortcoming for planar SOFCs.
Ignoring the associated sealant problems, planar SOFCs are a much more desirable
configuration than tubular SOFCs. Planar SOFCs offer high current densities, lower
electronic loses than tubular, increased manufacturability and the potential for less
expensive cell fabrication processes.
Monolithic SOFC
The monolithic design consists of many cells combined in a corrugated structure
as shown in Fig. 2.12. The contoured monolithic structure is approximately 200-300 m
thick. The all encompassing structure directs reactant and product gases, eliminating

49
inactive manifolding found in the planar design resulting in an extremely light, compact,
and high power density design.
(a) (b)
Figure 2.12: Monolithic SOFC in Co-Flow (a)and Cross-flow (b) Configurations [20].
The advantages of the monolithic design are small cell size and high power
density. The thin components allow for short current pathways, leading to much smaller
resistive losses. This allows the monolithic design to be operated at higher current
densities, for a given voltage, than competing designs.
The major disadvantage of the monolithic stack is fabricating the complex design
consistently and reliably. Current processing techniques involve tailoring thermal
mismatches and shrinkage rates so the corrugated structure can be co-fired. This design
requires extensive testing prior to operating, to ensure cell integrity.
The fabrication of the monolithic co-fired cell is a complex endeavor. It entails
controlling, adjusting and monitoring many sometimes inversely related parameters at

50
once. For example, reducing particle size, to increase sinterability, may result in more
shrinkage, and possibility cracking due to the increase in specific surface area due to the
smaller particles. The amount of binders, volume loading, and particle surface area must
be stringently controlled. Green state parameters must be controlled to minimize thermal
expansion mismatch during cycling. Initial and final process parameters for the
monolithic SOFC firing cycle can be seen in Fig. 2.13.
Figure 2.13: Firing Shrinkage as a Function of Temperature for Monolithic Powders
[20].

51
Chapter 3
THEORETICAL ASPECTS
Nearly all engineering ceramics and some metal products are fabricated from
powders. Typically, these powders are mixed with a few percent of various organic and
inorganic additives, compacted to form a porous “green” body, and heated (or “fired”) to
below the melting temperature of the material. During the heating, the porosity in the
sample is reduced or eliminated. This phenomenon is called sintering.
The properties of the sintered ceramic body depend to a large extent on the
chemical and microstructural imperfections that were not eliminated by or that were
produced as a result of sintering. Many imperfections can be traced back to the structure
of the green compact or to the nature of the starting powder. Therefore, the details of
these components and aspects will be explained in this chapter. Hot pressing is sintering
technique that increases driving force for densification. Therefore the sintering process
will be used to lay the groundwork of the more complicated hot pressing process. A
theoretical derivation of densification rate for hot pressing will also be presented along
with its practical limitations.
The goal of this chapter is to provide the reader with a clear understanding of how
the microstructure develops in a sintered material as it pertains to the novel hot pressing
technique. The contents of this chapter will be subdivided into powders, powder
compaction, sintering theory and hot pressing theory.

52
Ceramic Powder Processing
The final sintered microstructure of a powder compact is chiefly controlled by the
pore size and pore distribution in the compact. However these parameters are difficult to
measure, therefore the complementary variables, the particle size and distribution, are
measured. The initial packing, or green density, of the compact has a dramatic effect on
the final microstructure of the sintered sample. The green density may be the main
parameter measured in a sintered compact, due to inherently complex nature of pore size
and distribution evolution during sintering. The primary factors controlling green density
will be summarized in this subsection including powder particle size and distribution,
particle packing, and pressure distribution in a powder compact.
Powder Particle Size
Before compaction and sintering operations are performed, it is necessary to
define or regulate the starting powder characteristics. This is typically accomplished by
measuring a particle size. The concept of a “size” parameter is often obscure and
measured in an ad hoc fashion. Size may be defined as: the average of several lengths (as
in microscopic determination), the opening which will allow passage of a particle (as in
sieving), the diameter of a sphere with the same settling rate (as in sedimentation sizing),
or in other ways. It should be noted that although these definitions give somewhat

53
different results, none is basically more “correct” than the others. Several methods of
characterization can be found in the literature [25, 26, 27].
In the same way that the “size” of a particle requires definition, the “average size”
of a series of particles may be subject to several interpretations unless carefully defined.
The average calculated particle size, daverage, depends on the weight given to the factors
of: (a) number, (b) length, (c) surface, and (d) weight or volume of the particles of
different sizes. The average may be defined by eq. 3.1 where y is the percentage of the
total number, length, surface, or volume represented by the particles of diameter dn.
100
* n
average
dy
d  (3.1)
There are several established methods for measuring particle size including,
sieving, microscopy, sedimentation, sensing zone techniques, x-ray line broadening, and
light scattering. Information on these and other techniques can be found in the literature
[26, 27]. Optical microscopic methods may be used for particles size down to 0.5
micron. The microscopic method of determining of an entire range of particle size
distribution suffers from the difficulty that reasonably precise microscopic counts are
laborious. Also, it can be difficult to determine how much material is present that is
smaller than the limit of the resolution of the microscope. Microscopic determination of
separated fractions is particularly important as a check on the screening or sedimentation
calibration and efficiency, as well as shape and other particle characteristics. Electron
microscopy is similarly useful down to much smaller particle sizes, and is unique in its

54
ability to determine particle size and shape for particles below about 0.1 micron.
Broadening X-ray diffraction lines also may be used to determine the size of crystalline
particles finer than 0.1 micron. An advantage (and disadvantage) of this method is that it
determines crystal size rather than particle size. As a result polycrystalline agglomerates
are counted as separate single crystals rather than a single particle. One problem with
particle size measurement methodology is that different techniques cannot usually be
corresponded well with each other due to differences in accounting for particle shapes.
The varied and convoluted nature of particle shape can lead to variations in final products
that are very difficult to account for.
The most common method of representing particle size data is the cumulative
weight percent plot. The abscissa are a logarithm of size, which give equal prominence
to all parts of the size range and is consistent with the practice of geometric series of
openings in screens for sizing (sieves). In the cumulative plot, the total weight finer than
any size is plotted. The weight-frequency plot is the slope of the cumulative plot, or the
weight per cent retained for each geometric interval. The maximum is the mode of the
size distribution on a volume or weight basis.
Powder Particle Size Distribution
The determined particle size distribution depends greatly on the method of
producing the powder. Grinding normally results in a fairly broad distribution of
particles. As the grinding efficiency is increased, as by wet grinding, the distribution of

55
sizes is reduced. In general, a narrower range of particles can be obtained by
precipitation or calcination. Higher calcination temperatures give larger particle sizes
and also a broader particle size distribution [28].
Ceramic Powder Compaction
The firing behavior of ceramic compact depends on the proximity of ceramic
particles relative to each other. It is convenient to conceptualize the compact as porosity
size and distribution.
Pressure Distribution in a Powder Compact
Pressure affects the firing behavior of ceramic compacts. The pressure
distribution will decay as a function of depth from the location of the applied stress. This
effects may be due to: (a) decrease in pore size and better particle contact, (b) strain
energy added due to plastic flow, (c) strain energy added due to particle interlocking, or
(d) fracture of particles at contact points. The first and last are the only effects of
importance for most ceramic composites [28].
The main deleterious effect of pressure variation within a cold pressed compact is
the corresponding differences in green bulk density. These variations will cause non-
uniform shrinking resulting in distortion or warping during firing as high densities are
obtained [29]. Final porosity variations are due to pressure variations throughout the

56
piece. For a simple cylindrical sample, pressure variations in general increase as the
length-diameter ratio increases [30]. Length-to-diameter ratios greater than unity begin
to cause serious problems [31].
Resistance to compaction arises from collapsing bridges, or neck formations and
wall friction. After bridges collapse, further increases in densification may occur by
means of plastic deformation or crushing particles at the contact points. Kamm,
Steinberg, and Wulff, have shown that the major and almost entire cause of pressure
variation is the wall friction. By spraying two layers of a stearic acid lubricant on the
mold walls, complete elimination of pressure variations was obtained. However no
improvement of properties was obtained by adding lubricant to the powder [32].
Derivation of Pressure Variation in a Powder Compact
A first approximation of the mathematical relationships governing pressure
variations in a powder compact can be derived relatively easily.
Assume a cylindrical powder compact of length, l, and diameter, D, is uniaxially
compressed by double action plungers, each applying pressure, Po, as shown
schematically in Fig. 3.1. If we take an arbitrarily thin disc, of thickness dx, at a distance
x from the top plunger, then a force balance on the disc will yield eq. 3.2a and eq. 3.2b.
(3.2a)
    0)(
44
22












 dxDdPP
D
P
D
F XXXX 


57
(3.2b)
Defining the stress at a distance x as x, which is related to the applied pressure,
Px, using eq. 3.3, where k is the interparticle friction, that ranges from 0 to 1. When k
equals unity, the pressure at a depth x equals the applied pressure, the system reduces to
the hydrostatic case.
(3.3)
Figure 3.1: Pressure distribution in a Powder Compact Disc.
Combining (3.2b) and (3.3) yields:
0
4

D
dxkP
dP x
x

(3.4a)
or, (3.4b)
04  dxDdP XX 
XX kP
D
kdx
P
dP
X
X 4


58
Integrating eq. 3.4b using the boundary condition that Px = Po when x = 0 yields
eq. 3.5. Therefore pressure decays exponentially within a compact and is a function of
the coefficient of friction, interparticle friction, and length to diameter ratio.
(3.5)
Sintering Theory
Sintering is the process in which fine particles, in close contact, form single
continuous bulk agglomerations. This agglomeration is typically accompanied by an
increase in relative density due to closure of pores and gaps between the original
particles. The application of temperature inputs energy for atomic rearrangement and
diffusion allowing the surface area to decrease, reducing the surface energy and reducing
the total free energy of the system.
Driving Force for Densification
The driving force for densification has been shown to be the reduction in surface
area associated with the elimination of porosity. This driving force can be better
quantified by examining the thermodynamic energies associated with surface curvature.







D
x
kPP oX 4exp

59
Assume an infinite surface is assigned a chemical potential,  , value of zero. If
material is removed forming a small sphere of radius, R, the volume change, dV, is the
volume of this sphere or the molar volume, , times the change in moles, n, as described
in eq. 3.6.
(3.6)
Therefore the chemical potential change associated with forming this sphere is the
surface energy of the material, ,, times the change in surface area, as shown in eq. 3.7.
Combining eq. 3.6 and eq. 3.7, by eliminating the infinitesimal change in moles, dn,
yields eq. 3.8. Where the surface radius of curvature, , is defined as in eq. 3.9. R1 and
R2 are the principal radii of curvature of the surface.
(3.7)
(3.8)
(3.9)
Equation 3.8 implies that a positive or negative radius of curvature yields a
corresponding positive or negative change in chemical potential change. Therefore a
2
4 RdV  dndR 
 
dV
RdR
dn
aSurfaceAred 
  8
)(


 

 s
s
R
2
21
11
RR


60
concave surface has a lower chemical potential then a convex surface. This is shown
schematically in Fig. 3.2.
Material on a convex surface would tend to move and fill in a concave surface
leaving a flat surface in equilibrium. Examining the sintering of fine particles (see Fig.
3.4 or Fig. 3.5 for a schematic) the convex surface of the spherical particles will tend to
fill in the concave contact, or neck area, between slightly sintered particles. Mass
transport of material will occur following the chemical potential gradient.
Figure 3.2: Chemical Potential as a Function of Surface Curvature.
A mechanical concept of “sintering stress” has been developed to better facilitate
understanding of these abstract ideas. Equating chemical potential energy to mechanical
work, as in eq. 3.10, sintering stress can be defined as in eq. 3.11. Pore size can be
included by combining eq. 3.8 and eq. 3.11
(3.10)
(3.11)
(3.12)
 > 0
ConvexFlat
 = 0
Concave
 < 0 > 0
Convex
 > 0
ConvexConvexFlat
 = 0
Flat
 = 0
Concave
 < 0
ConcaveConcave
 < 0


dV
dndV 





R
s
equivalent


2


61
The equivalent stress in a sintering body is inversely proportional to the pore
radius. Stated another way, the smaller the pore diameter, the greater the driving force is
for densification. Therefore smaller pores should shrink more rapidly than larger pores.
This is shown in Fig. 3.3.
Figure 3.3: Equivalent Stress as a Function of Particle Radius.
Due to the simplicity of the assumptions used, eq. 3.12 represents the driving
force for densification, or sintering stress, for a non crystalline, grain boundary-less,
compact with uniform pore size and uniformly distributed porosity. Although the direct
applicability of this equation is limited, the concepts behind its derivation are important
to keep in mind. In general, the sintering and firing characteristics are determined by the
pore size rather than the particle size.
 Equivalent
Stress
 Equivalent
Stress
R Particle Radius

62
Evolution of Sintering
Sintering requires the surface area of a compact to decrease. This is dependent on
the contact area between particles increasing. The effect of temperature is to control the
rate at which this initial area of contact is increased to a substantial amount. The
geometrical changes by which this decrease in surface area is achieved in actual
compacts are probably incapable of being precisely defined since they depend on the
shape, size, size distribution and modes of packing of the particles.
Microstructural changes in sintering take place continuously. However it is
helpful to define three distinct stages of sintering. Approximate relative densities of these
stages these are shown in Table. 3.4.
Sintering Stages Distinguishing Features Relative
Density
Pre Point Contact Established during green pressing 50%+
I Initial Formation of a skeletal network 50-75%
II Intermediate Pore shrinkage and consumption 75-90%
III Final Isolated porosity, slowing kinetics 90%+
Table 3.4: Consolidation Stages of a Sintered Powder Compact
The stages can be seen schematically in Fig. 3.5. Point contact is established
when the loose powder is green pressed. Green pressing gives a relative density of
greater than 50%, corresponding to loose packing. When added to the powder, binders
and lubricants can result in much higher relative densities. These additives reduce
friction allowing easier particle rearrangement, hence creating higher green densities.

63
Figure 3.5: Consolidation Stages of a Sintered Powder Compact [27].
During the initial stages of sintering, the predominant feature is an increase in the
interparticle contact areas with time, accompanied by a rounding-off of the sharp re-
entrant angles formed at the points of contact, allowing neck growth to begin.
As the growing necks merge, the intermediate stage of sintering begins. During
this stage the original particulate structure disappears and is replaced by that of a poly
crystalline body of intergranular porosity, although the pores still remain interconnected.
During this stage, grain growth usually occurs as the pores shrink and may take the form
of discontinuous growth in which certain grains grow rapidly at the expense of neighbors.
Ultimately, during the final stage of sintering, the pore network is broken into
isolated pores. Further densification results from the shrinking of these pores.
Throughout the sintering process pores have remained in their original positions. Over
time some pores may become isolated from the grain boundary, increasing the vacancy
diffusion distance. Since pores far removed from the grain boundary are only eliminated
through much slower lattice diffusion the rate of sintering greatly diminishes.

64
It is therefore important to understand the factors that control the mobility and
position of grain boundaries in a polycrystalline, porous material.
Grain Growth
The driving force for grain growth is the elimination of grain boundary area,
reducing surface energy in the bulk and lowering the overall Gibbs free energy of the
compact. Grain growth, or grain boundary movement, in a sintering compact is difficult
to characterize due to the complexity of the local pore network which changes
continuously with time. However some general observations can be enumerated.
Grain size increases with increasing temperature. This implies grain growth is an
activated process and is therefore limited by diffusional motion of atoms across the grain
boundaries. The narrower the grain size distribution, the slower the grain growth rate.
Also, larger grains grow at the expense of smaller grains. Therefore to achieve higher
sintered densities, it is important to keep grain size as small as possible and limit the
temperature used during sintering.
The presence of impurities, or second phase particles can slow grain growth if
they remain on the grain boundary and the inclusion diffuses slowly. For a grain
boundary to move past an inclusion surface area must be created. The cross sectional
area of the inclusion must be reformed in the boundary on the far side of the inclusion.
Inclusions that do not remain on the boundary are not as effective grain size growth
inhibitors.

65
Pores will act similar to inclusion, hindering the grain boundary movement and
slowing grain growth. During the initial stage of sintering the volume fraction of pores is
very large and almost no grain growth takes place. However, as sintering proceeds, many
small pores are consumed and grain growth can occur more readily. If a grain boundary
successfully moves past a pore, the pore becomes entrapped. Once entrapped, the pore is
no longer near to the high diffusion rate region, adjacent to the grain boundary. Then the
rate at which material can fill the pore greatly decrease. Pore entrapment is the main
reason why theoretical densities are almost never attained in a sintered sample.
The microstructural development in a compact is governed by the reciprocal
action between densification and grain growth. This is dependent upon whether the
controlling sintering transport mechanism is densifying or non-densifying.
Sintering Transport Mechanisms
There are six main mechanisms for materials transport in sintering: grain
boundary diffusion, volume or bulk diffusion, evaporation/ condensation, surface
diffusion, coarsening (grain growth), and dislocation creep. The main transport
mechanisms are shown schematically in Fig. 3.6. It is convenient to divide the
mechanism into two groups, densifying and non-densifying.

66
Figure 3.6: Sintering Transport mechanisms [after 27].
Mechanisms involving bulk transport, such as grain boundary diffusion, volume
diffusion and dislocation creep, move material by decreasing the distance between
particle centers. Mechanisms involving surface transport, evaporation/condensation,
volume coarsening, and surface diffusion, move material without bringing the particle
centers closer.
Each of these type of mechanisms are important depending on the end goal. If the
goal is fully dense material, the quickest route is to sinter in a regime in which only
densifying mechanisms are prevalent. To control amount and distribution of porosity,
non densifying mechanism regimes are important.

67
Hot Pressing Theory
The driving force for sintering is the excess free energy of the powder over that of
a solid. In contrast to increasing the temperature, hot pressing increases the driving force
for densification rather than the driving force for diffusion. By applying pressure rather
than temperature, hot pressing can limit particle coarsening during sintering, resulting in
more efficient densifcation. Estimates of this increase in driving force can reach 1 to 2
orders of magnitude higher than sintering at the same temperature [29].
Diffusion occurs whenever a powder compact is brought to an adequate
temperature to activate the process. Hot pressing does not change the basic sintering
mechanisms; however it allows those mechanisms to be activated at lower temperature.
For example, in refractory carbides, sintering is generally performed at about 1350°C,
while hot pressing is accomplished at 1200°C [35].
Applied pressure can cause particle rearrangement (grain boundary sliding),
particle fracture, or plastic flow in a ceramic compact. These processes occur on a time
scale of seconds to tens of seconds, while the sintering process depends on a time scale of
minutes to hours, at least an order of magnitude larger time scale. Therefore the
governing mechanism in hot pressing is the same as in sintering: diffusion. Although
plastic flow and creep can be important in certain regimes, their effects are most
prevalent only at high applied pressures (greater than 5000 psi).

68
Analytical Treatment of Hot Pressing
Falling a treatment by Coble [33], models for initial-, intermediate, and final-stage
densification under pressure explicitly including both surface energy and applied pressure
as driving forces will be presented. These densification mechanisms have been shown to
be important when small particles (1 m) and moderate pressures (1-5000 psi) are used
[34].
It is important to consider both applied pressure and surface tension driving forces
when the ratio of the two is approximately unity. The change in concentration, C, due
to applied pressure, Pa, divided by surface energy, , is given by eq. 3.13, where, R is the
particle radius size, and , represents pie. Assuming a moderate surface energy value of
103
ergs/ cm2
(1 J/ m2
), this ratio is plotted as a function of particle size and applied
pressure in Fig. 3.7. As this ratio increases the driving force for densification is
dominated by the applied pressure, as it decreases the driving force is dominated by
reduction of surface energy.

RP
C
C a
rgySurfaceEne
essureApplied


 Pr
(3.13)
Four different particle size YSZ powders were hot pressed under the same
temperature and pressure conditions to demonstrate the relative effects of pressure vs.
particle size. The smallest particle size powder (0.2 m) densified to greater than 90%
relative density, while the largest powder (75 m), resulted in 70% relative density.
Micrographs of the densified YSZ powders are shown in Fig. 3.8.

69
Figure 3.7: Particle Size and Pressure Regime in Hot Pressing.
(a) (b)
(c) (d)
Figure 3.8: Various Hot Pressed YSZ Powders.
YSZ particle sizes (a) 0.2 m (b) 3.5 m, (c) 25 m, and (d) 75 m.
1
10
100
1000
10000
100000
1000000
10000000
0.01 0.1 1 10 100 1000
Particle Radius [um]
Pressure[psi]
90% of Densification due to Applied Pressure 90% of Densification due to Particle Size
Pressure Dominated
Densification
Particle Size Dominated
Densification
Unity Line
50 m
___
50 m
___
50 m
___
50 m
___
50 m

70
Initial Stage Densification Rate
The following treatment will attempt to decipher how applied pressure affects the
driving force for densification in a hot pressed powder compact. A simple two particle
model will be used to explain the applied pressure effects. As two spherical particles,
each of radius R begin to coalesce, there will be material overlap, y, which can be
approximately as a disc of thickness, dy, and disc radius, X. This is shown schematically
in Fig. 3.9.
The radius of curvature, , is equal to the neck radius y. The neck radius can be
written in terms of the particle size and overlap disc radius as described by eq. 3.14. The
volume change can be described in terms of the boundary area and change in overlap
distance, as in eq. 3.15. Differentiating eq. 3.14, yields eq. 3.16, the infinitesimal change
in neck radius as a function of overlap disc radius. The amount of densification is
proportional to the change in volume of this disc. Therefore densification rate is equal to
the rate of volume change, eq. 3.17.
R
x
y
4
2
 (3.14)
dyxdv 2
 (3.15)
R
xdx
dy
2
 (3.16)
Rdt
dxx
dt
dv
2
3

 (3.17)

71
Figure 3.9: Initial Stage Densification of Sintering Compact [after 33].
The rate of volume change must be formulated from geometrical terms into
material property terms. Diffusive vacancy flux from neck surface to the grain boundary
can be expressed as eq. 3.18, where J/l is the flow per unit length of a cylinder, Dv is the
vacancy diffusion coefficient, C is the difference in vacancy concentration between the
surface of the control cylinder and the central axis. The length, l, is equal to the neck
radius for lattice diffusion or one half the grain boundary width, W/2, for boundary
diffusion.
CD
l
J
V  4 (3.18)
Multiplying by the vacancy volume, , and the specific neck length,  or y, yields
eq. 3.19, for lattice diffusion, or eq. 3.20 for grain boundary diffusion. The densification
rate is now the integral of either of these equations depending on the controlling diffusive

72
mechanism. However, before integration, the vacancy-concentration difference must be
evaluated.
    yCDy
l
J
dt
dV
vb 





 4 (3.19)
    


















2
4
2
W
CD
W
l
J
dt
dV
vb (3.20)
The normal stress on the boundary decreases the equilibrium vacancy
concentration; according to the Gibbs-Thompson equation, eq. 3.21, where Co is the
equilibrium concentration at temperature T under a stress-free planar surface,  is the
stress normal to the surface, k is Boltzman’s constant,  is the volume contributed by the
vacancy, and Co is the equilibrium concentration for stresses imposed.
kT
C
CC o
o



 (3.21)
The hot pressing sintering model is obtained by substituting eq. 3.14, for , and
eq. 3.21, for C into eq. 3.19 and then combining with eq. 3.15. Finally by integration
and division of both sides by the radius cubed, R3
,, the final neck growth model is
obtained (eq. 3.22). This can be compared to the analogous result for pressureless
sintering (eq. 3.23). The same procedure as above for grain boundary diffusion yields eq.
3.24 and eq. 3.25, for hot pressed and pressureless sintering neck growth, respectively.

73
t
RP
kTR
D
R
x AL

















 







3
4
32
(3.22)
 t
kTR
D
R
x L





 






3
4
32
(3.23)
t
RP
kTR
D
R
x AB

















 







4
6
96
(3.24)
 t
kTR
D
R
x B





 






4
6
96
(3.25)
By comparing eq. 3.22 with eq. 3.23 or eq. 3.24 with eq. 3.25, it can be seen that
the applied pressure can be added to the surface energy as a driving force for initial stage
densification when multiplied by a dimensional constant (R/) involving the particle
radius. Thus, the explanation for hot pressing increasing the driving force for
densification but not diffusion can be seen from this mathematical standpoint.
Intermediate and Final Stage Densification Rate
The Nabarro-Herring [36] diffusion creep model has been modified for hot
pressing by a number of authors [34, 37, 38, 39]. Uncertainties arise in the calculation of
the effective stress as a function of porosity as the relationship is not precisely known.
There is a factor of 2 uncertainty in both effective stress and surface area/ path length,
yielding less than a factor of 4 uncertainty in densification rate. This equation cannot be
rigorously applied to predict time dependence of densification. However the modified

74
diffusion creep model can be applied to approximate the instantaneous rate of
densification.
The Nabarro-Herring Creep equation (3.26) gives the uniaxial strain rate (d/dt)
which will be caused by lattice diffusion, Dl, under an applied stress, ; where G is the
grain size. For grain boundary diffusion, Db, the strain rate is given by eq. 3.27 [40],
where W is the grain boundary width. Both equations are derived for materials of
theoretical density.
kTG
D
dt
d L
2
3
40  
 (2.26)
kTG
WD
dt
d b
3
5.47  
 (2.27)
Either eq. 3.26 or eq. 3.27 could be used to approximate densification using the
effective stress in a porous mass, and the conversion of the relative strain rate to a
densification rate. However these equations would only be approximate for
instantaneous densification rate as mentioned above. Surface energy effects can be
included in eq. 3.26 and eq. 3.27 by added the pressure difference across the curved
interface to the effective stress. These equations can be approximately used for final
stage densification also [see 33].

75
Necessity of Experimental Studies
Hot pressing theory has been briefly summarized in the preceding section.
Although the conceptual basis of this theory is established, it is difficult to implement it
in practice. Numerous assumptions have been incorporated into the theory. The inability
to precisely monitor the microstructure as a function of time severely limits are ability to
apply the theory. However, the theory is important to understand generally how process
variations should affect the final product. Therefore it is important to have an empirical
understanding of the changes in microstructure so that the one-step fabrication process
can be developed.

76
Chapter 4
EXPERIMENTAL DETAILS
Characterization Techniques
Several techniques were used to characterize the precursor powders and the post
sintered sample. These techniques will be briefly explained below; further information
can be found in the literature [27].
Particle Size and Distribution Measurement
Initial powder characteristics were, in some cases, provided by the manufacturers.
All sample particle sizes and distributions however, were measured using a Horibia
Particle Size Analyzer. This analyzer uses a laser light scattering algorithm to determine
particle size and distribution. Typically this process is repeated several times until results
converge to a reproducible level as recommended by the manufacturer.
Density Measurements

77
For this work, density was quantitatively calculated using two methods. The first
approximate method for measuring density was to cut the sample into a definable
volume, V (e. g., cube), measure the sample mass, m, and calculate the corresponding
density, , using eq. 4.1.
V
m
 (4.1)
The second approach is known as the Archimedes Immersion method. In this
technique the sample is weighed in air, and while being immersed in water. The sample
can be coated with a negligibly thin (assumption) layer of wax to seal any connected
porosity. The dry weight of the sample, WDry, is noted. The sample is supported by a
thin wire, of weight WWire, and placed into a bath of distilled water. The weight of the
sample immersed in the water, its wet weight, WWet, is then measured. Using the known
density of the water, Distilled, the wet and dry weights of the sample can be used to
determine the buoyant force acting on the sample. This can be used to accurately
determine the volume of the sample, which can in turn be used to calculate the sample
density. The density, Sample, can be calculated by as shown in eq. 4.2.
)(
*
WireWetDry
DistilledDry
Sample
WWW
W



 (4.2)
Hardness Measurement

78
Hardness values are also helpful when evaluating the extent of sintering in a
sample. Hardness was measured using standard vertical hardness and microhardness
indenters.
Optical Microscopy
Samples were prepared using standard metallographical techniques. Cross
sectional pictures were taken using an Olympus inverted digital microscope.
SEM Analysis
All SEM characterization was performed on a JOEL scanning electron
microscope, with secondary electron and backscatter electron image capabilities.
Electron Microprobe Analysis
All EMPA work was done at MIT’s Earth Science Department using their
Electron Microprobe.
Sample Fabrication

79
The initial powder size and distribution were tailored to achieve the desired level
of densification in the different material layers. Two methods were used to reduce the
mean particle size, ball milling or attrition milling. Although both methods were used in
early stages of this work, ball milling became the predominant method due to simplicity
and greater repeatability.
Ball Milling
Balling milling was typically performed in 500 or 1000 ml Nalegene containers.
Although used in earlier stages dry milling was discontinued. Alcohol, typically
methanol or isopropyl, was almost exclusively used as suspension for wet milling.
Partially stabilized zirconia was used as the milling media. Zirconia was chosen due to
material compatibility already established by SOFC research. Although many sizes were
employed at early stages, most milling was conducted with 20 mm YSZ media. Rotation
speeds were controlled by an electric motor, ranging from 50-65 rpm. Milling times
ranged from 15 minute to 24 hours.
Green Samples
Prior to sintering, all samples underwent cold-pressing or green pressing in a
Carver mechanical pump under a 2000 lb. load. This process gave the ceramic powder
sufficient green strength for handling, i.e., placing in furnace, weighing, stacking, etc.

80
Green pressing can be accomplished using binders, waxes and lubricants. Boron
Nitride (BN) was used exclusively as a die/ sleeve wall lubricant in some experiments.
The BN lubricants helped in particle repositioning and rearrangement during the pressing.
Although BN is a stable compound and frequently used [27], possible reactions between
BN and the component layers are not known at this time. Future work should investigate
possible interactions or more stable lubricants as substitutes.
Sintering Experiments
Sintering experiments were conducted in a Lindberg Blue three zone tube furnace.
A mullite or alumina tube was sealed on both ends by a pressure clamping fixture. This
seal allowed flexibility for thermal expansion while maintaining a hermetic seal. The
tube was evacuated with a Welch roughing pump down to approximately 100 mTorr.
Then the tube was backfilled with forming gas (Ar-5% H2) to atmospheric pressure.
During the experiment, forming gas was fed into the tube at a slight positive pressure.
Gas pressure and flow rate were controlled by an MKS 1141 mass flow controller. Any
reacted species and excess gas was forced through the vacuum pump to the exhaust.
In experiments where the oxygen particle pressure needed to be controlled, the
forming gas was bubbled through distilled water at room temperature (298 K). At this
temperature the water vapor concentration was set at 3%, and the H2O (l) to H2O (g)
equilibrium ensured an accurate and controlled oxygen partial pressure.

81
Typical Hot Pressing Cycle
The final hot pressed microstructure is controlled by three primary parameters:
temperature, pressure, and time. As shown in the theoretical section, temperature,
pressure, and time have decreasing amounts of effectiveness in changing the initial
microstructure.
An example of a pressured assisted sintering cycle for pore closure in cemented
carbide ceramics is plotted in Fig. 4.1. A key attribute is the simultaneous application of
maximum temperature and pressure. This cycle has been shown to give the smallest
open porosity [27]. This approach has been taken in this research to ensure a fully dense
electrolyte at the lowest temperature possible. Secondary parameters such as particle size
distribution, green density and added binders are ignored in this representation.
Hot Pressing Experiments: Reducing Environment
The experiments were conducted in a Centorr hot-press with graphite heating
elements resulting in a reducing environment. The chamber temperature was measured
with a ‘W’ type thermocouple and was also cross-checked with an optical pyrometer.
The pressure was calculated based on the applied load by the hydraulic pump on the ram
and the cross-sectional area of the die. The vertical movement of the ram was monitored
by a micrometer. The chamber was connected to a mechanical pump to obtain the

82
desired chamber pressure. Argon was backfilled to create an approximately inert
atmosphere in the chamber when required.
Figure 4.1: Typical Hot Pressing Cycle
The green compact, either individual or multi-component layers, were placed in
the hot press and the chamber was evacuated and purged with argon to remove
contaminants. The powder was then slowly heated at 3C per minute, while maintaining
a maximum pressure of 100 mTorr. The hot press was connected to a Honeywell digital
control programmer. The desired temperature-pressure-time cycles were programmed
into the controller which were then executed and plotted automatically. The automatic
program was activated, at approximately 400 C, and allowed to bring the hot press
through its heating and pressure cycles.
Temperature Pressure
Time
Stage 1:
Ramping
Stage 2:
Hold
Stage 3:
Release
Temperature Pressure
Time
Stage 1:
Ramping
Stage 2:
Hold
Stage 3:
Release

83
Hot Pressing Experiments: Oxidizing Environment
A custom resistance furnace was attached to an Instron mechanical testing
machine in order to hot press samples in a high oxygen partial pressure environment. A
cylindrical convection furnace was purchased from Hi-Temp Products Corp., of Danbury,
Connecticut. An aluminum stand was fabricated to support this furnace on the Instron
frame. 303 Stainless Steal rods with Inconel 600 tips were used as plungers. Several die
materials were used including Inconel and alumina. Temperature was manually
controlled by a 240 V variac connected to super kanthal heating element leads. Pressure
was controlled by the Instron machine. The die assembly was fabricated from Alumina
to ensure high temperature stability in an oxidizing environment. Machinable alumina
rod, of 75% relative density and 7.5 cm diameter was used for the die sleeve. Fully dense
Al2O3 rods, of 0.8 cm diameter, were also used for the plungers.
Electrochemical Characterization
Electrochemical characterization was conducted on a series of SOFC single cells
and electrode/ electrolyte half cells in order to determine concentration polarization under
a variety of experimental conditions. These characterizations included open circuit
voltage (OCV), alternating current (AC) impedance, and constant current experiments.
The characterization was carried out on composite discs, either individual or
multi-layer, with electrical connections through platinum screen current collectors. The

84
current collectors were attached to the cell by platinum paste in a two step firing process.
The first step involved firing a thin layer of platinum paste on the electrode. The
platinum screen was then attached to the electrode with more platinum paste and refired
at high temperatures. This technique produces a platinum layer with a relatively large
particle size, which resides on the surface of the anode and cathode and does not hinder
mass transport in the electrodes. Attaching the current collectors in this manner
uniformly distributes the current across the electrode surface.
The current of the fuel cells was controlled by a Perkin Elmer 263A Potentiostat/
Galvanostat (1 amp maximum current), with a Kepco Bipolar Operational Power Supply/
Amplifier (10 amp maximum current) as needed. The AC impedance studies were
conducted with a Solatron frequency response analyzer SI 1250.
In order to test single cells, a reducing atmosphere had to be established on the
anode side, and an oxidizing atmosphere had to be established on the cathode side. This
was established and maintained by gold O-rings between the electrode and the end of a
fully stabilized zirconia tube. After the fuel cell was sealed to the zirconia tube, its
effective electrochemical active area was generally 2.85 cm2
.
The OCV was monitored during heating to the target experimental temperature of
1000C. The theoretical OCV for the H2- 3% H2O mixture vs, air was calculated to be
1.038 V, according to Nernst (eq. 2.1).

85
Chapter 5
RESULTS AND DISCUSSION
To produce high quality reliable cells, the hot pressing SOFC fabrication process
must maintain: a fully dense electrolyte, porous electrodes, adequate interfacial bonding,
and minimal material interactions during fabrication. Ensuring these requirements
necessitate careful and controlled examination of individual components before the
components can be combined into a single cell.
In order to determine the ideal processing conditions for the different material
component layers, the three components of focus in this study, anode, cathode and
electrolyte, were initially studied independently. Only after establishing individual
densification characteristics, were the components combined and studied in tandem.
Although further details are given below, the experimental approach is outlined
presently and is summarized in Fig. 5.1. Initially a moderate pressure was selected. A
moderate pressure was chosen so that in the future the pressure could be increased or
decreased as needed. This pressure was used for all components to establish a baseline
for comparison. Secondly, a starting temperature was chosen. After this, a hold time was
verified. The initial time was considered adequate if double the time did not give any
gross microstructural changes. Next a temperature range was iterated upon to establish
the minimum possible process temperature. Finally, secondary parameters such as

86
particle size, particle packing, particle distribution or pressure modifications were used to
tailor the microstructure to the design goals.
Experimental Regime
It was important to understand the extent of the experimental regime before trials
were commenced. This regime was limited by equipment, chemical interactions and
previously established sintering cycles.
Figure 5.1: Experimental Approach to Hot Pressing a Single Cell.
Cathode
Anode
Electrolyte
Tailor Anode to
Match Electrolyte
Processing
Temperature
Tailor Cathode to
Match Electrolyte
Processing
Temperature
Single Cell
Establish Minimum
Densification
Temperature for
Electrolyte

87
Equipment Limits
The maximum operating temperature of the graphite heating elements used in the
hot press was 2500C. Therefore 2500C was the absolute upper limit for the fabrication
temperature. All hot pressing experiments were initially conducted in a graphite die and
sleeve fixture. The reported maximum compressive stress of for high density, fine-
grained structural graphite is 15, 000 to 18, 000 psi, from room temperature to 2500C
[41]. Assuming a safety factor of 2/3, a maximum pressure limit of 10, 000 psi was
assigned. Therefore a pressure range of 0 to 10, 000 psi was used for these experiments.
This region was further divided into moderate pressure, 0- 5000 psi, and high pressure
5000-10, 000 psi. This is schematically summarized in Fig. 5.2.
Figure 5.2: Experimental Operating Regime.
0
500
1000
1500
2000
2500
3000
0 2000 4000 6000 8000 10000 12000
Pressure [psi]
Temperature[C]
High Pressure RegimeModerate Pressure Regime
Maximum HP Temperature
Maximum
Graphite
Compressive
Stress

88
Chemical Interactions
Chemical interactions between the components become important at higher
temperature. Therefore the maximum temperature before interactions begin was
established.
Thermodynamically, ZrO2 reacts with LaMnO3 to form insulating phases such as
La2Zr2O7 between the cathode and electrolyte, at temperatures above 1100C [42, 43, 44,
45]. Although the reaction kinetics of this reaction are slow, these insulating phases are
undesirable in SOFCs and must be minimized because they hinder electronic conduction
in the cathode causing cell performance to degrade significantly.
Figure 5.3: Practical SOFC Hot Pressing Experimental Regime
MP: 1880 C
500
1000
1500
2000
2500
3000
Temperature[C]
]
ElectrolyteCathode Anode Interconnect
La2Zr2O7 Formation
Mn +2 Diffusion NiCrO 4 Formation
MP: 2660 C
MP: 1453 C
MP: 2510 C

89
Manganese is known to be a mobile species at high temperatures and can easily
diffuse into the electrolyte, changing the electrical characteristics or the structure of both
the cathode and the electrolyte [46]. Fabrication temperature was limited to below
1400C to minimize this migration. Also, above 1400C, Ni or NiO may react with the
LaCrO3 interconnect material to form poorly conducting phases such NiCrO4 [47].
Finally, elemental Nickel melts at 1453C. These facts suggested a maximum processing
temperature of 1400C, lower if possible. These facts, as well as the melting points of
the SOFC component materials are summarized in Fig. 5.3
Established Sintering Cycles
The monolithic fuel cell fabrication temperature using tape calendaring and
pressure-less sintering was reported to lie between 1300 and 1400C [27, 48].
Zirconia powder, of 0.1 m mean diameter, was sintered to 98% theoretical
density at 1300C in air [50]. YSZ powders of sub-micrometer size were formed into a
green body (about 50% green density) and fired to 95% theoretical density in air at
1125C [51]. These data suggest a starting point for hot pressing zirconia to full density
of 1100C to 1300C.
Nickel powder, of 0.5 m particle size, was sintered at 800C to 98% theoretical
density [49]. This suggests that Ni powder will not yield the 30-40% porosity required
for a SOFC anode. However, NiO powder of 0.3 m mean particle size was fired to 57%
theoretical density at 900C [49]. This suggests that NiO powder will yield the 30-40%

90
porosity required for a SOFC anode. This is consistent with the established processing of
the anode using NiO powder as described in the introductory section.
The LSM cathode was air sintered at 1250C, to the required porosity level [18].
Since an additional driving force for densification, arising from the applied pressure, was
available in hot pressing, this ensured densification to the required porosity at or below
1250C.
Based on the above information, the limitation to the experimental regime
imposed by previously established sintering cycles is summarized in Fig. 5.4.
Figure 5.4: Practical SOFC Hot Pressing Experimental Regime
500
700
900
1100
1300
1500
Temperature[C]
ElectrolyteCathode Anode
YSZ Air Sintered
Monolithic Fabrication Temperature
Interconnect
Nano YSZ Air Sintered
Ni 98% Density
NiO 57% Density
LSM 80% Density
500
700
900
1100
1300
1500
Temperature[C]
YSZ Air Sintered
Interconnect
Ni 98% Density
NiO 57% Density
LSM 80% Density
500
700
900
1100
1300
1500
Temperature[C]
YSZ Air Sintered
Interconnect
Ni 98% Density
NiO 57% Density
LSM 80% Density

ONE-STEP PROCESS FOR SOLID OXIDE FUEL CELL FABRICATION

ONE-STEP PROCESS FOR SOLID OXIDE FUEL CELL FABRICATION

Recommended

Recommended

More Related Content

What's hot

What's hot (11)

Similar to ONE-STEP PROCESS FOR SOLID OXIDE FUEL CELL FABRICATION

Similar to ONE-STEP PROCESS FOR SOLID OXIDE FUEL CELL FABRICATION (20)

More from chrisrobschu

More from chrisrobschu (20)

Recently uploaded

Recently uploaded (20)

ONE-STEP PROCESS FOR SOLID OXIDE FUEL CELL FABRICATION