Bowen Michaels 2021

AN ABSTRACT OF THE THESIS OF
Michael S. Bowen for the degree of Master of Science in Mechanical Engineering
presented on June 10, 2021
Title: Evaluation of Materials for Use in Open-Cycle Magnetohydrodynamic Power Generation
Abstract approved:
______________________________________________________________________________
David P. Cann C. Rigel Woodside
Abstract:
Recent interest in oxy-fuel combustion for carbon capture, as well as advancements in
technologies such as magnetics, materials, and computational modeling has sparked renewed
interest in magnetohydrodynamic (MHD) power generation. The increased temperatures of oxy-
fuel combustion versus air-fuel combustion poses a challenge in the selection of materials for
plasma exposed components. Further, the combustion products are seeded with an ionized gas
such as potassium which creates a highly corrosive environment for many materials. A selection
of insulator and electrode materials from the refractory oxides are evaluated for their stability in
extreme environments from plasma exposure. Electrical properties are also considered to gauge
conductor or insulator performance and provide accurate values to enhance generator design
modeling. Of those evaluated, ceria-based materials offer the best properties as electrode materials
with electrical conductivities exceeding 100 S/m at 1300˚C while magnesia exhibits electrical
conductivities well below 1 S/m for temperatures under 1600˚C. Both materials were found to
resist potassium corrosion in controlled experiments suggesting minimal degradation in the
presence of potassium seeded MHD environments.
© Copyright by Michael S. Bowen
June 10, 2021
All Rights Reserved
Evaluation of Materials for Use in Open-Cycle Magnetohydrodynamic Power Generation
by
Michael S. Bowen
A THESIS
submitted to
Oregon State University
in partial fulfillment of
the requirements for the
degree of
Master of Science
Presented June 10, 2021

Commencement June 2022
Master of Science thesis of Michael S. Bowen presented on June 10, 2021.
APPROVED:
________________________________________
Co-Major Professor, representing Mechanical Engineering
________________________________________
Co-Major Professor, representing Mechanical Engineering
________________________________________
Head of the School of Mechanical, Industrial, and Manufacturing Engineering
________________________________________
Dean of the Graduate School
I understand that my thesis will become part of the permanent collection of Oregon State
University libraries. My signature below authorizes release of my thesis to any reader upon
request.
______________________________________________________________________________
Michael S. Bowen, Author
ACKNOWLEDGEMENTS
First and foremost, I would like to thank the Oak Ridge Institute for Science and
Education for funding my participation in research at both Oregon State University and the
National Energy Technology Laboratory. Without the ORISE Professional Internship Program, I
don’t see how my graduate career would be possible.
Thanks are extended to my co-advisors for their imparted wisdom in science and
research. Dr. David P. Cann has offered patience, understanding and an open, welcoming
environment for thought provoking discussions and the freedom to learn from mistakes. Dr. C.
Rigel Woodside has taught me how to develop experiments thoughtfully to thoroughly answer
research questions, how to bridge the gap between science and engineering, and how to navigate
government-funded research.
Finding success in a graduate program is difficult alone, and I would like to thank the
many colleagues and friends who have helped me succeed in my studies, research, and general
wellness. A special mention to Dr. Charles Culbertson, Dr. Ryan McQuade, and Dr. Lee
Aspitarte for passing on their experience knowledge. There are too many others to mention here,
but if you feel you deserved a mention, consider this sentence so.
Of course, I would like to finally thank my family. Their patience, love, and support
throughout my entire educational has been paramount to my success.
CONTRIBUTION OF AUTHORS
The authors who have contributed to the works published which are featured in this thesis
have all provided guidance in executing experimental methods, interpreting results, and revising
and editing.
Michael Johnson and Bryce Wright contributed to the experimental methods performed
in Chapter 2, assisting in developing the fabrication and measurement procedures. Dr. Ryan
McQuade assisted in performing electrochemical impedance spectroscopy measurements and
interpreting results collected in Chapter 2. Richard Chinn (NETL) performed x-ray diffraction
measurements published in Chapter 3. Dr. David P. Cann, Dr. C. Rigel Woodside, Dr. Kyei-Sing
Kwong, and Dr. Peter Y. Hsieh contributed to the conception of ideas, interpretation of results,
analysis of data, writing, revising, and editing of the contents of Chapter 2 and 3.
TABLE OF CONTENTS
Page
1. Introduction ______________________________________________________________ 1
1.1. Magnetohydrodynamics _________________________________________________ 1
1.1.1. History ____________________________________________________________ 1
1.1.2. Application _________________________________________________________ 3
1.2. Materials _____________________________________________________________ 5
1.3.1. Requirements _______________________________________________________ 6
1.3.2. Electrode Materials _________________________________________________ 12
1.3.3. Insulators _________________________________________________________ 16
1.3. Bibliography _________________________________________________________ 20
2. Electrical Properties of Gadolinia-Doped Ceria for Electrodes for Magnetohydrodynamic
Energy Systems ______________________________________________________________ 24
2.1. Introduction _________________________________________________________ 24
2.2. Review ________________________________________________________________ 26
3. High temperature corrosion stability of ceramic materials for magnetohydrodynamic
generators ___________________________________________________________________ 46
3.1. Introduction _________________________________________________________ 46
3.2. Materials ____________________________________________________________ 50
3.3. Experiment __________________________________________________________ 50
3.4. Results and Discussion _________________________________________________ 54
3.4.1. Magnesia _________________________________________________________ 54
3.4.2. Alumina __________________________________________________________ 56
3.4.3. Yttria-stabilized zirconia _____________________________________________ 58
3.4.4. Ceria _____________________________________________________________ 59
3.5. Conclusions _________________________________________________________ 61
3.6. Bibliography _________________________________________________________ 62
4. The Yttria-Hafnia System __________________________________________________ 65
4.1. Fabrication __________________________________________________________ 65
4.2. Electrical Properties ___________________________________________________ 68
4.3. Corrosion Results _____________________________________________________ 71
4.4. Bibliography _________________________________________________________ 74
5. Conclusions _____________________________________________________________ 75
5.1. Electrical properties ___________________________________________________ 75
5.2. Corrosion Resistance __________________________________________________ 76
5.3. Final Remarks _______________________________________________________ 77
Appendix A: Full Documentation of Experimental Procedures _________________________ 79
LIST OF FIGURES
Figure Page
Figure 1.1) Implementation of topping cycles generating power from high temperature gas flows.
A) A turbogenerator converting thermal and kinetic energy into mechanical power through a
rotating turbine and finally electrical power through induction and brushes. B) A MHD generator
converting thermal and kinetic energy into electrical power through the induction of ionized
combustion gases.______________________________________________________________ 3
Figure 1.2) Stable potassium phases based on the minimization of Gibb’s Free Energy of (A)
stoichiometric oxy-methane combustion with an addition of 1% potassium by weight in the form
of potassium carbonate and (B) an argon to potassium ratio of 5:1. Molar concentrations of phases
are calculated at 1 atm pressure and incremental temperature steps. ______________________ 10
Figure 1.1.3) Location of atoms in the cubic fluorite structure of MO2 oxides.28 ____________ 12
Figure 1.4) Phase diagram of MgO and K2O adapted from Yazhenskikh based on equilibrium
phases of potassium oxide and magnesium oxide in a slag environment.41 _________________ 17
Figure 1.5) Equilibrium phase diagram of Al2O3 and K2O adapted from Spear and Allendorf.21 18
Figure 1.6) Electrical conductivity values of various oxide electrode and insulator materials of
interest for MHD use. __________________________________________________________ 19
Figure 2.1) Schematic of a combustion driven open-cycle MHD power generator showing direct
plasma exposure of the channel wall, composed of electrode and insulator segments. ________ 26
Figure 2.2) Comparison of accumulated data from various sources27–37 of Gd0.1Ce0.9O1.95. _ 29
Figure 2.3) Comparison of accumulated data from various sources18,19,27,29,31,33,36,38–42 of

Gd0.2Ce0.8O1.9. _____________________________________________________________ 30
Figure 2.4) Equivalent circuit diagram of a brick layer model for AC circuit modeling44. _____ 32
Figure 2.5) A) above: Average density and standard deviation of 8 samples of each GDC
composition. B) right: XRD patterns of GDC pellets ground into powders. The Miller indices
shown are for fluorite reflections. ________________________________________________ 33
Figure 2.6 A) Impedance spectra of one sample of each composition at 600˚C showing the grain
boundary for each except 50GDC. Second semicircles, where present, represent electrode
contributions. B) Imaginary bode plot normalized to the frequency and impedance at which the
peak value of the grain boundary impedance occurs for a pure ceria disk sample. ___________ 34
LIST OF FIGURES (Continued)
Figure Page
Figure 2.7) Electrical conductivity measured by ac impedance spectroscopy for disk samples
prepared in this work for increasing Gd dopant concentration over the temperature range of 300˚C
to 900˚C. Solid lines show the fit of equation (2.4) for each sample. _____________________ 35
Figure 2.8) Arrhenius plot of electrical conductivity of different measurement techniques for
undoped ceria. Solid lines represent Arrhenius equation fitting for each method. ___________ 37
Figure 2.9) Lattice parameter a0 with increasing Gd content for this work in comparison to data
reported by Artini et al. [46] and Grover, Tyagi [45]. Deviation is expected due to the method of
calculation. __________________________________________________________________ 44
Figure 3.1) Schematic of a direct fired open-cycle MHD generator with a segmented channel.
Schematic shows plasma flowing into a rectangular channel where an electrode-insulator wall is
exposed to potassium seeded combustion products. An orange arrow shows the direction of plasma
flow perpendicular to a green arrow representing the magnetic field. _____________________ 46
Figure 3.2) Predominant potassium phases calculated from minimization of Gibb’s Free Energy
using CEA at varying temperatures and 1 atm pressure.7 Open circles represent phase transitions.
(A) Theoretical phases resulting from an input of stoichiometric oxygen-methane with an addition
of potassium carbonate to produce 1% potassium by weight. (B) Theoretical phases resulting from
an input of argon and potassium carbonate with a 20:1 mass ratio._______________________ 48
Figure 3.3) (A) Crucible, sample, and potassium carbonate assembly. (B) Furnace setup with
representative oxygen and water vapor sensors. Approximate crucible and thermocouple locations
show temperature sampling occurs at position of assembly. ____________________________ 51
Figure 3.4) Mole fractions of potassium vapor and potassium hydroxide vapor in an argon
environment containing potassium carbonate at 1 atm with varying trace amounts of water vapor.
Shaded region shows the range of water vapor content measured to be between 10 ppm and 50
ppm across all tests. ___________________________________________________________ 52
Figure 3.5) Carbon dioxide peak intensity extracted from gas chromatograms at varying
temperature throughout a potassium exposure experiment of YSZ. Peak intensity shows potassium
carbonate breakdown beginning above approximately 900˚C and proceeding until cooling below
800˚C. ______________________________________________________________________ 53
Figure 3.6) Mass of magnesia samples through each step with mass changes reported between each
step. The null and exposure test both indicated less than 0.1 mg mass changes._____________ 55
Figure 3.7) X-ray diffraction pattern of an magnesia sample surface before exposure (orange) and
after exposure (green) in the range of 2θ from (A) 10˚ to 90˚ and (B) the emphasized section of
30˚ to 45˚. ___________________________________________________________________ 55
Figure Page
Figure 3.8) Mass of alumina samples through each step with mass changes reported between each
step. The null test showed <0.1 mg mass change while the K-exposure tests indicate an
approximate 3.0, 4.2, and 4.1 mg increase in mass for each respective test. ________________ 57
Figure 3.9) X-ray diffraction pattern of an alumina sample surface before exposure (orange) and
30˚ to 45˚. ___________________________________________________________________ 57
Figure 3.10) Mass of YSZ samples through each step with mass changes reported between each
step, indicating <0.1 mg mass change for each sample. The exposure step resulted in >9 mg
increase in mass. ______________________________________________________________ 58
Figure 3.11) X-ray diffraction pattern of a YSZ sample surface before exposure (orange) and after
exposure (green) in the range of 2θ from (A) 10˚ to 90˚ and (B) the emphasized section of 25˚ to
40˚. ________________________________________________________________________ 59
Figure 3.12) Mass of ceria samples through each step with mass changes reported between each
step. The null and exposure test both indicated <0.3 mg mass changes. ___________________ 60
Figure 3.13) Diffraction pattern of a ceria sample surface before exposure (orange) and after
35˚. ________________________________________________________________________ 60
Figure 4.1) X-ray diffraction patterns in the 2q range of 26˚ to 33˚ of ground calcined and sintered
powders of hafnia with 20 mol% and 50 mol% yttria additions. _________________________ 66
Figure 4.2) X-ray diffraction pattern of sintered 20 mol% and 30 mol% yttria-doped hafnia with a
single calcination step compared to the 20 mol% sintered pellet subjected to incremental
calcinations in Figure 4.1A. Triangles mark residual monoclinic phase peaks while Miller indices
mark fluorite peaks. ___________________________________________________________ 67
Figure 4.3) Circuit model used to fit impedance spectroscopy data of YDH samples. ________ 69
Figure 4.4) Circuit model fitting (orange boxes) versus measured data (blue dots) of a 20mol%
yttria-doped hafnia sample using electrochemical impedance spectroscopy techniques. ______ 70
Figure 4.5) Arrhenius plot of the electrical conductivity of yttria-doped hafnia samples measured
using AC electrochemical impedance spectroscopy techniques. Reference data from Schieltz
included. ____________________________________________________________________ 71
Figure 4.6) Mass change values of 20 mol% yttria-doped hafnia samples exposed to potassium
vapor in an argon environment. __________________________________________________ 73
Figure Page
Figure 4.7) X-ray diffraction pattern of 20 mol% yttria-doped hafnia samples before and after
exposure to potassium vapor in an argon environment. ________________________________ 73
Figure 5.1) Electrical conductivity of materials of interest for oxy-fuel open-cycle MHD power
generators. Solid lines are fitted curves for measured data in this work and others while dashed
lines are values extrapolated to higher temperatures. _________________________________ 76
Figure A.1) Temperature profile of single ramp up, hold, and ramp down from null test of a MgO
sample. _____________________________________________________________________ 85
Figure A.2) Dimensions and electrode placement of samples for electrical conductivity
measurements. A) Disk sample with electrodes for AC EIS. B) Bar sample for DC resistance
measurements. _______________________________________________________________ 89
LIST OF TABLES
Table Page
Table 2.1) Activation energy and A intercept for each of the Arrhenius equation plotted in Figure
2.7 fit to equation (2.4). ________________________________________________________ 35
Table 2.2) Fitted Arrhenius equation values of the data in Figure 2.8 for the activation energy and
A −intercept defined in equation (2.4) above. _______________________________________ 38
Table 2.3) Table of calculated masses combined to produce the appropriate stoichiometric solid
solution of GdxCe1 − xO2 − δ. Deviation between measured values deviated and calculated
values were less than ±1mg. _____________________________________________________ 43
Table 2.4) Table of lattice parameters from two sources as well as those calculated from the XRD
data provided in the manuscript. _________________________________________________ 43
Table 4.1) Fitting parameters resulting from a least squares fitting of the circuit model in Figure
4.3 to the produce the data represented in Figure 4.3. _________________________________ 69
Table A.1) Equipment used for fabrication and characterization of materials in the Electroceramics
Laboratory at Oregon State University. ____________________________________________ 79
Table A.2) Template table of masses for observing mass change references in the null and
potassium exposure portions of a corrosion evaluation experiment. ______________________ 87
Dedicated to my oldest and dearest friend
William “Billy” F. Butler
1
1. Introduction
In this thesis, materials are evaluated to characterize their critical properties for performance
in magnetohydrodynamic (MHD) power generation systems. The electrical properties of select
oxides with rare-earth dopants are determined through literature review and state of the art
measurements to determine their viability as electrode or insulator components in MHD channels.
Stability in MHD conditions is evaluated by subjecting the materials to a simulated MHD
environment containing highly corrosive potassium species from the volatilization of potassium
salts in a controlled atmosphere tube furnace. The experimental method is adapted from an ASTM
standard test method for refractory bricks. Conclusions are drawn on the materials’ viability and
functionality as important components in MHD power generation.
1.1.Magnetohydrodynamics
1.1.1. History
Magnetohydrodynamics (MHD) is a branch of physics which studies the effects of a

conductive fluid flowing through a magnetic field. It is primarily used in understanding
astrophysical phenomena such as solar flares, but it is scarcely found in applications on Earth. The
first experiment in MHD was conducted by Michael Faraday in 1831, who attempted to measure
the potential difference between two plates suspended from the Waterloo Bridge by copper wire
and submerged in water.1 Theoretically, the ions contained within the flowing water interacting
with the earth’s magnetic field would have produced a voltage based on the Lorenz Force.
Unfortunately, the current produced from his circuit was too small to measure with the
galvanometers of his time.2 In 1851, the experiment was repeated in the English Channel and a
voltage was successfully measured; Dr. William Hyde Wollaston is credited with the experiment,
though the year supersedes his reported death (1828) by more than a decade making this a story of
two incredible feats. It was nearly 100 years later that the first recorded use of the term
magnetohydrodynamics was presented by Hannes Alfven. At the time it was presented, physicists
2
criticized the theory claiming that if these MHD waves that Alfven predicted truly existed,
Maxwell would have discovered them. Overcoming these harsh criticisms, Alfven won the Nobel
Prize in Physics for pioneering the field of magnetohydrodynamics and other works in
electromagnetic phenomena.3
Meanwhile, the earliest patents of MHD power generation date back to 1910, though they
were not given such a name as the term did not yet exist. In 1940, a Hungarian engineer, Bela
Karlovitz, developed an MHD generator with Westinghouse Electric Corporation.4 Their design
was unsuccessful due to a lack in understanding of ionized gases. Furthermore, the generator
design was rather complex and difficult to execute. Research came to a halt, reportedly at the fault
of World War II. Then, researchers in the U.S. at Cornell University (1959) and in the U.K. at
University of Sheffield (1960) published significant progress in the understanding of ionized gases
and plasmas sparking worldwide interest in MHD power generation.5,6 Facilities at AVCO Electric
and Westinghouse Electric were constructed in the U.S. in the 1960s to develop MHD for
commercial implementation into coal fired power plants. Meanwhile, the USSR constructed a
model facility, labeled U-02, followed by a full scale combined MHD – steam cycle power plant
in Moscow, named the U25 facility, fueled by natural gas.7–9 An international collaborative effort
was led by 18 countries in the International Liaison Group on MHD Electrical Power Generation,
including a yearly symposium spanning three decades known as the Symposium on Engineering
Aspects of Magnetohydrodynamics.
From this work, major technical challenges prevented feasible commercialization of MHD
power generation technology. Primarily, the coal-fired generator designs, which relied on slag
formation on the surface of the generator walls to prevent corrosion, eventually failed because of
seed penetration through the slag layer.10,11 Other technologies, like the turbogenerator, developed
faster with less challenges and resulted in a decrease in interest in MHD power generation. Since
this decline, advancements in research and technologies have once again made MHD a competitive
option for commercial power plants. Many of these are laid out in a conference paper by Woodside
and others with emphasis on the economic progress in superconducting magnets, oxygen
production, and simulation and modeling.12 The use of oxy-fuel combustion is advantageous to
simplify carbon capture in commercial power plants but increases operating temperatures above
3
the limits permissible with current turbogenerator designs. A natural alternative is MHD power
generation, which not only can operate at these elevated temperatures, but has the potential to
convert the additional heat into electrical power.
1.1.2. Application
The MHD power generator and the turbogenerator operate in similar manners, converting
thermal and kinetic energy from high temperature combustion into electrical power. They both
were developed to serve as topping cycles for steam power plants which are driven by a high
temperature combustion process. While they both use magnetic induction to drive a current, they
do so through very different processes.
Figure 1.1) Implementation of topping cycles generating power from high temperature gas flows.
A) A turbogenerator converting thermal and kinetic energy into mechanical power through a
rotating turbine and finally electrical power through induction and brushes. B) A MHD generator
converting thermal and kinetic energy into electrical power through the induction of ionized
combustion gases.
In a turbogenerator, a turbine is designed to be spun as the hot, fast-moving gases flow

across the blades converting the thermal energy and kinetic energy into mechanical energy. That
mechanical energy is then converted to electrical power using induction to drive a current through
a rotating coil and brushes to carry the current to an external load (Fig 1.1A). The efficiency of
turbogenerators is largely limited by the maximum operating temperature of the system, typically
4
around 1000˚C for current technologies.13 In particular, the quest for higher temperature materials
for turbogenerator components is one focus for improving efficiency.
On the other hand, MHD power generators remove the need for moving parts and convert
thermal and kinetic energy directly into electrical power. The magnetic field applied to the hot,
fast-moving conductive fluid produces a Lorenz force perpendicular to the flow which drives
electrons towards current collecting electrodes (Fig 1.1B). The omission of moving parts enables
higher operating temperatures and thus higher efficiencies than the turbogenerator. However, the
conditions for MHD power generation requires the material to perform under extreme
environments. It is important to note that there is very little relevant experimental data for existing
materials tested under MHD conditions, so this thesis aims to provide critical materials
performance data that will help guide the materials selection as part of the design of MHD power
generation systems.
To generate power with MHD, the working fluid starts with standard combustion products
which reach ultra-high temperatures, greater than 2500˚C for oxy-fuel combustion. At high
temperatures, thermal ionization of gases results in free electrons, but the products of combustion
have relatively high ionization energies and the concentration of free carriers is relatively low. A
seed material is introduced into the combustion products which produces gaseous species which
have much lower thermal ionization energies and significantly increases the electrical conductivity
of the plasma. Potassium, in the form of potassium carbonate, was a popular choice in early works
as it is benign and abundant. Others also discuss advantages of seed regeneration, for which
potassium reforms potassium carbonate upon cooling.10,14,15 Elevated temperatures, high velocities
from choked flow, and potassium vapor are sources of material degradation and failure in MHD
components.
To extract power, materials must be exposed to these extreme conditions. Some generator
designs attempt to reduce the harsh conditions through water cooling to reduce operating
temperatures and relying on protective coatings from combustion by-products (slag from coal, for
example). The disadvantages to these designs will be discussed in the next section, but focus will
be placed on materials for direct plasma exposure of ultra-high temperature (>1600˚C) materials.
5
1.2. Materials
In Bowen and Rossing’s summary of materials for MHD power generation, there are three
groups of materials for an MHD generator which are categorized by composition and limit the
operating temperature ranges.16 The lowest temperature materials are metals, which are limited by
melting points and oxidation temperatures. Combustion gases reach upwards of 3000˚C, so
aggressive water cooling is required to prevent overheating. In legacy works, metals were used in
conjunction with coal fired systems as ash in the coal produced a slag layer which theoretically
protected the electrodes from potassium corrosion. In practice, potassium diffused through the slag
layer. The cold wall resulted in surface temperatures below the potassium liquidus line and
ultimately shorted the system with a conduction path between the segmented electrodes.10,16 This
critical design flaw motivates increasing the wall temperature above the potassium liquidus line.
In addition, water cooling and reduced wall temperatures result in heat loss through the walls and
lower efficiencies of the cycle.
The next temperature range of materials are boride-, carbide-, and nitride-ceramics. These
materials have extraordinary melting temperatures, typically exceeding 3000˚C. However, at
elevated temperatures (>1500˚C), oxidation of the metal cation or the anion destabilizes the
material at oxygen partial pressures present in typical MHD plasmas.16 Maintaining operating
temperatures below the oxidation point or optimizing the oxidation reactions could improve
material performance, but it’s more desirable to utilize stable materials which will not degrade
from use.
This leads to the next category of materials: high temperature oxide ceramics. With
functional melting temperatures between 1500˚C and 3000˚C, these materials have the potential
to withstand the operating temperatures and oxygen partial pressures necessary for efficient MHD
power generation. Thus, the evaluation of materials can be reduced to a few necessary properties
to consider whether a material will resist degradation in MHD applications.
The use of oxides for MHD applications is well known to be a solution to increasing the
operating temperature of MHD channels. Rohatgi published a summary of materials research
through 1984 encompassing work across various test facilities.17 The electrode materials tested for
MHD applications primarily consisted of zirconia, ceria, ferric aluminates, and lanthanum
6
chromates. As for insulator materials, magnesia and alumina were the most used materials for
clean and coal slag open-cycle systems respectively.
In this work, critical properties of MHD materials are evaluated in preliminary tests to predict
performance in MHD operating conditions. The properties of interest have been narrowed down
to critical properties which define the stability and functionality of a material system. The first is
the electrical properties, which determine whether a material can perform as an insulator or
electrode. Chapter 2 consists of a publication on electrical properties of Gd-doped ceria and
outlines the methods implemented for characterizing this property. The second is the potassium
vapor stability condition which evaluates materials degradation of materials in the presence of
potassium vapor. The results of multiple oxides exposed to potassium vapor are presented in the
publication that is Chapter 3. The properties evaluated are expected to be consistent across dense
polycrystalline materials regardless of fabrication. Upon completion of the evaluation procedure,
materials are vetted and those which show promise as candidate materials can be further researched
into engineering and fabrication of the complex shapes which constitute the MHD channel.
The materials of interest in this work are high temperature oxide ceramics with a focus on
applications with direct plasma exposure in oxy-fuel combustion systems. The following section
outlines the conditions which the material must be able to withstand followed by a literature review
of the relevant properties of the materials studied in this work.
1.3.1. Requirements
Temperature
Before a material can be considered a candidate material for MHD power generation, it
must be evaluated on a few key properties. First, the material must be stable at operating
temperatures exceeding 1600˚C to prevent potassium condensation on the surface, but higher
temperatures are preferred to improve generator efficiency.10,16 The melting points of oxides are
well known and summarized by Schneider, so limiting material selection to those exceeding
2000˚C prevents concerns of material degradation from melting and enables higher operating
temperatures and efficiencies.18
7
Electrical Requirements
The electrode materials serve as conductors to create an electrical circuit between conduits
and the plasma. Others have considered the minimum electrical conductivity required to operate
as an electrode and reported values are 100 S/m by Rohatgi and 1000 S/m by Bowen and
Rossing.16,17 These are general guidelines to assist in evaluating materials, but the success of a
material is better evaluated through modeling performance and efficiency of the generator design.
This work will use the value provided by Rohatgi, 100 S/m, as the minimum requirement to
consider a material a viable electrode material. Metals meet this requirement by default due to the
metallic bonding nature of the materials. Ceramics, on the other hand, are just as often insulating
as they are conducting. Most electrically conducting ceramics remain relatively insulating below
temperatures between 400˚C and 800°C, depending on factors such as the band gap, dopant
species, etc. Other ceramics, such as aluminum oxide, remain insulating through the entire
temperature range, though their conductivities still increase with temperature. In order to properly
act as an insulator, the electrical conductivity must be less than 1 S/m at operating temperatures.11,17
The high temperature behavior of electrical properties of ceramics and metals stems from
the bonding mechanisms of the material. Equation 1.1 represents the conductivity as a function of
temperature dependent quantities. The number of carriers, 𝑛, can quantify electrons, holes, ions,
or any other carrier that may be mobile in the material. The mobility of that specific carrier, 𝜇, is
simply the ability of the carrier to move through the material under an applied field. Finally, the
charge of each carrier, 𝑞, completes the conductivity.
𝜎 = 𝑛𝜇𝑞 (1.1)
By representing the conductivity as a function of mobility and number of carriers, one can
understand the trends in temperature and the behaviors of conduction mechanisms.
For metals, the number of free electrons is exceedingly high due to the nature of metallic
bonds. When bonding, orbitals combine to create bands from valance and directions to adjacent
atoms. These bands are then filled with the delocalized electrons, which are free to move about the
8
lattice. This is the source of the high electrical conductivities observed for metals, on the order of
108 S/m at room temperature. Since the population of free carriers is largely fixed, at elevated
temperatures the rate of electron-phonon collisions rapidly increases and the mobility decreases.
The result is a decreasing electrical conductivity with temperature, though in regard to the
requirements for the electrode in MHD applications, the conductivity remains adequate for power
generation.
On the other hand, the metal-oxides studied in this work are stabilized by ionic bonds. The
ionic bonds are generally stronger than metallic bonds due to coulombic attractions between ions,
which result in more stable materials and higher melting temperatures. However, the population
of free carriers, and thus the electrical conductivity, is significantly lower than metals, <101 S/m
at room temperature. In contrast to metals, as the temperature increases, so does the population of
free carriers, which can be free electrons, holes, or ionic defects such as lattice vacancies in oxide
ceramics. This effect dominates over the temperature changes in mobility, and the trends in
electrical conductivity can be modeled with an activation energy by the Arrhenius equation. The
Arrhenius equation is simply a model for thermodynamic processes which are thermally activated.
In Equation 1.2, 𝑇 is the temperature, 𝐸! is the activation energy, 𝑘" is the Boltzmann constant,
$
𝜎# is a theoretical model parameter representing the intercept conductivity at the limit of % !& → 0,
"
and 𝑠 is a placeholder which is equal to 0 for a simple Arrhenius behavior and equal to 1 for a
process like ionic conductivity.
𝜎# −𝐸!
𝜎= exp < = (1.2)
𝑇' 𝑘" 𝑇
The magnitude of the activation energy is a measure of how rapidly the conductivity increases
with temperature. It is related to the conduction mechanism and the electronic band gap.
While the oxides of interest in this study are insulating at room temperature, these materials
have the potential to function as electrodes in MHD operating conditions. By understanding the
temperature dependent electrical properties, generator designs can be catered to control wall
temperatures and optimize material performance. Furthermore, the electrical conductivity of
9
insulating materials also follow this behavior and must be known to prevent conductivity from
reaching levels which may result in electrical shorting of electrodes.
Stability Requirements
The melting temperature of a material is a stability condition which is easily assessed as
this material property is one that is critical for all applications. In the words of Ken Mills, “At high
temperatures, everything reacts, and it gets increasingly worse as temperature increases.”19
However, there is very little experimental data on the high temperature stability of materials in the
presence of potassium. In MHD applications, potassium carbonate is introduced into the
combustion products resulting in potassium vapor. At these temperatures, the most stable phases
of potassium are in the form of K and KOH. Figure 1.2A shows the potassium phases in methane
combustion seeded with potassium carbonate calculated using Gibb’s free energy minimization
with the commercial Chemical Equilibrium with Applications (CEA) software developed by
McBride and Gordon with NASA.20 These potassium containing vapors are highly corrosive and
lead to destabilization, erosion, and degradation of materials exposed to the plasma. Material
behavior in the presence of vapor potassium is thus a critical property for performance in MHD
applications.
10
Figure 1.2) Stable potassium phases based on the minimization of Gibb’s Free Energy of (A)
stoichiometric oxy-methane combustion with an addition of 1% potassium by weight in the form
of potassium carbonate and (B) an argon to potassium ratio of 5:1. Molar concentrations of
phases are calculated at 1 atm pressure and incremental temperature steps.
While the potassium corrosion problem is not widely studied in this context, there is some
data available which assists in predicting a material’s resistance to potassium corrosion. The
optimal scenario is the availability of a binary oxide phase diagram with the metal oxide of interest
and potassium oxide, like the one presented by Spears and Allendorf.21 The reason for representing
potassium oxide in the phase diagram is that in order for potassium to penetrate the lattice it must
break the oxygen bonds of the cation and form its own. One can consider the oxide material system
as a lattice of oxygen atoms which are linked by metal cations through chemical bonding.
Potassium, which also must act as a cation, will have to bond with oxygen in order to penetrate the
structure and destabilize the initial material phase. So, in an equilibrium phase diagram, one can
conclude whether the potassium reacted phases will be stable at the operating temperatures of
interest.
In general chemical principles, the concept of neutralization by reacting acids and bases is
introduced. The reactions are founded on the formation of H+ in acids and OH- in bases. Similarly,
oxides have been characterized as basic and acidic.22 Generally, elements in Groups I and II of the
11
periodic table form basic oxides while those in Groups XVI and XVII form acidic oxides. Other
elements in between can be basic, acidic, or amphoteric. The lanthanides have also been classified
as basic. It stands to reason, since potassium is a basic oxide, it will likely not react with another
basic oxide. The acid-base characteristic of oxides can be rather complex, especially when
combining multiple oxide systems in solid solution or forming new phases entirely. Thus, looking
towards the basic oxides for potassium corrosion resistance provides only some guidance for
material selection and prediction for potassium corrosion behavior.
While binary oxide phase diagrams with potassium are not available for every oxide, there
are vast databases of information on phases in the form of both x-ray diffraction (XRD) data and
density functional theory (DFT) analysis on crystal structures. Both databases provide information
on possible stable phases containing the oxide of interest and potassium. Together, DFT provides
insight to the theoretical existence of a phase while XRD patterns provide empirical evidence of
the observed phase. Some materials have numerous potassium-metal-oxide phases which may be
stable under varying conditions, which makes for a rigorous task of discovering corrosion
reactions. Therefore, these databases act as tools to determine likelihood of corrosion, and like
acidity, only provides predictions to how a material will behave rather than a robust scientific
theory.
Interest in the potassium corrosion problem extends beyond just MHD applications.
Materials which endure coal combustion products, including potassium in the form of sulfates, are
of wide interest for thermal barrier coatings (TBCs) on turbines in coal-fired power plants.23,24
There are also applications in aviation which rely on TBCs to increase durability of components
exposed to extreme environments. Furthermore, refractory applications such as glass melting
furnaces and slag furnaces for co-gasification create harsh environments which degrade
components.21 While each of these applications have unique reactants and environments which
complicate the thermodynamics of material stability, the evaluation of material properties is
consistent across each. Materials are exposed to environments similar to the application to evaluate
stability, corrosion, and degradation.
12
1.3.2. Electrode Materials
This study focuses on the oxides of elements in the Group IV column of the periodic table
for electrically conducting materials. Namely, zirconium-, cerium-, and hafnium- based oxides
doped with rare earth metals. Yttria-stabilized zirconia, or YSZ, was one of the first material
systems studied for hot-wall MHD power generation in the collaborative U.S.-U.S.S.R.
development program. It was found to be able to perform on the order of 100s of hours, but post-
testing analysis confirmed degradation due to potassium reactions.8 Introduction of ceria into the
YSZ system showed improvements in corrosion, suggesting ceria-based materials may perform
better in the presence of MHD plasmas, but a robust evaluation of ceria-based materials was not
performed.8 Hafnia was studied by Bates and Marchant in later years for coal-fired MHD.25 It is
also a popular refractory material used in reentry vehicles, nuclear applications, and other extreme
environments.26,27 The other Group IV elements are not included either due to radioactivity or an
insufficient melting point.
O
M
Figure 1.1.3) Location of atoms in the cubic fluorite structure of MO2 oxides.28
These materials are capable of forming the cubic fluorite phase which has a face-centered
cubic metal cation structure with an oxygen atom at each of the eight tetrahedral sites (Figure 1.3).
When doped with rare-earth oxides, such as gadolinia or yttria, there is a direct substitution of the
dopant cation on the base cation resulting in an oxygen vacancy compensation as in the defect
reaction in Equation 1.3. The oxygen vacancies that form in the fluorite structure are the primary
source of electrical conductivity and the motivation for studying the Group IV metal oxides. The
13
other advantage to stabilizing the fluorite structure is that it is the highest temperature phase for all
three systems, and therefore prevents destructive phase transitions upon heating and cooling.
Specifics for each system will be described in the sections that follow.
*+#
𝐵( 𝑂) @⎯B 2𝐵*, + 3𝑂+- + 𝑉+⋅⋅
(1.3)
The dopant materials, namely Gd2O3 and Y2O3, are rare-earth lanthanides with a crystal
structure named as such. Specifically, the cubic rare-earth structure is depicted in Figure 1.3B.
Their structure being cubic improves the solubility of the dopants into the fluorite stabilized
structures. However, when the dopant material dominates in concentration, the structure of the
system undergoes a phase transition from fluorite to the cubic (c-type) rare-earth structure for
hafnia- and ceria-based materials. This phase transition occurs somewhere around 50 mol % as
observed for the systems in this study.
Zirconia
Zirconium oxide (ZrO2) is one of the most commonly used electrically conducting
refractory oxides. It is a popular material in applications such as oxygen sensing, solid oxide fuel
cells, and other ionic conductor uses. In early works, it was the first ceramic material considered
for use in MHD applications. In such research, Y2O3 and CeO2 were primary dopants to stabilize
the fluorite phase and improve conductivity.
At room temperature, it forms a monoclinic phase which is non-conducting, even at
elevated temperatures. At 1170˚C, there is a destructive phase transition to tetragonal, followed by
the fluorite phase at 2370˚C.29 However, with a small amount of doping, as low as 8 mol%, the
fluorite phase can be formed, stabilizing the material from room temperature up to its melting
point. In the monoclinic phase, the energy to form oxygen vacancies is high, but the fluorite phase
can easily produce oxygen vacancies increasing the ionic conductivity. Above 1000˚C, YSZ has
been shown to conduct current indicating the material is electronically conducting in addition to
ionically conducting.29
14
The performance of zirconium oxide in MHD applications shows that potassium corrosion
can occur during operation. In particular, bloating and spalling was reported on the cathode side
due to the negative charge attracting the highly corrosive potassium ions. The testing of zirconia-
based materials in the presence of potassium led to the publication of potassium zirconate x-ray
diffraction patterns by Gatehouse, Nesbit, and Lloyd in the 1970s.30–32 In the 1990s, a K-Zr-O
ternary phase diagram at 700˚C was published showing numerous stable potassium-zirconium-
oxide phases.33 These observations of zirconia behavior in the presence of potassium vapor show
that corrosion is unavoidable, and alternatives are desired. However, this work performs the
evaluation on zirconia materials both to act as a reference material with expected results and to
provide a clearer analysis on the behavior of zirconia materials in the presence of potassium vapor
species at high temperatures.
Ceria
Cerium oxide (CeO2) is unique in this study as the fluorite phase is naturally occurring for
this material. Rare-earth dopants are still used in order to increase the electrical conductivity of the
system. The use of gadolinium as a dopant into ceria to enhance the electrical properties has been
extensively studied and a comprehensive collection of the results is presented in Chapter 2.
Mogensen et al. present an elegant literature review of doped ceria, carefully explaining the source
of conduction mechanisms.34 Ceria is referred to as a mixed ionic, electronic conductor (MIEC) as
the electronic and ionic conductivities are on the same order of magnitude. Blumenthal and
Panlener have studied ceria in depth to determine the behavior of ceria in oxidizing and reducing
environments finding that ceria has is capable of maintaining its structure even with a high
concentration of oxygen vacancies.35,36 The conclusions found that ceria could be easily reduced
to increase the electrical conductivity. The introduction of gadolinium activates this high
conductivity level in less reducing environments enabling applications in solid-oxide fuel cells.
In the various applications of ceria, either very few involve potassium environments or
there is no interaction of potassium with cerium oxide. Based on literature reports, there is little
published data on the formation of a potassium-cerium-oxygen phase. In crystallography
databases, the K2CeO3 phase has been calculated using DFT with a crystal structure based on the
15
Na2CeO3 structure, but no experimental phase has been found to the extent of this literature
search37. Even reports on legacy U.S. research, which used CeO2-ZrO2 compounds, published that
observed degradation decreased with increasing CeO2 content.8 Thus no potassium-ceria phases
are predicted to form.
Hafnia
Hafnia (HfO2) is chemically similar to zirconia with phase transitions from monoclinic to
tetragonal at 1750°C and then to fluorite above 2700°C, which occur at much higher temperatures
than zirconia. The addition of rare-earth oxides of the composition R2O3 are found to stabilize the
fluorite phase at room temperatures to permit a single-phase structure up to the melting point of
the material. Again, this fluorite phase increases electrical conductivity significantly, through the
introduction of oxygen vacancies, to permit the material to perform as an electrode in MHD
applications. Specifically, additions of yttrium oxide have been found to stabilize the fluorite phase
at concentrations as low as 8 mol%. Bates and Marchant, who worked on materials for MHD
applications, studied rare-earth doped hafnia, including yttria.25 Furthermore, thesis work
completed by Schieltz at the Iowa State University describes the electrical properties of yttria-
doped hafnia as well as fabrication procedures enabling the evaluation of this material system at
Oregon State University.38
In works by Bates and Marchant, the stability of rare-earth doped hafnia in the presence of
potassium is evaluated using various techniques for coal-fired, open-cycle MHD power
generation.39 Their evaluation procedure primarily consisted of an electrochemical corrosion test
of two electrodes submerged in liquid slag seeded with potassium sulfate to mimic the corrosive
environment produced by a slag layer. The apparatus is heated above 900˚C and a current is run
between the electrodes, through the slag. Their findings report varied reaction rates for different
dopants, with yttrium and terbium showing low reactions rates which met their stability
condition37. These promising results motivate further testing of fluorite hafnia for oxy-fuel and
slag-less MHD channels where the major corrosion species are potassium vapors as opposed to
potassium sulfates and liquid slag.
16
1.3.3. Insulators
Unlike the electrodes, selection of insulator materials is much simpler in terms of chemical
and structural considerations. Thermodynamically favorable phases of oxides have already been
observed and studied and there is little that can be done to improve the material’s stability as
dopants introduce defects which reduce stability and produce sources of electronic and ionic
conduction. These problems are accelerated as temperature increases as well. There are primarily
two insulator materials of interest in MHD applications, alumina and magnesia. Both are naturally
occurring oxides with melting temperatures above 2000˚C. The metal-oxygen bonds are extremely
stable which reduces the formation of oxygen vacancies and free electrons making them ideal for
insulator materials.
In previous research, these two materials were the most commonly used insulators in
generators constructed in the first development stages in the late 60s and 70s.16,17 Many reports
have provided results on the stability of these materials in the presence of MHD conditions and
most have reported the same conclusions. Magnesia is chemically resistant to potassium corrosion
with very little potassium infiltration reported into the structure, though some potassium
compounds were found to penetrate the pores of the low-density insulators.8,9 Alumina has
consistently been found to react with potassium with formations of potassium aluminate phases in
addition to evident erosion of the surface.11
Magnesia has a melting temperature around 2500˚C with the rock salt structure. The
electrical conductivity has been measured up to 1600˚C to be around 8 x 10-4 S/m making it a
suitable insulator for MHD applications.40 With interest in slag and fuel ash corrosion of magnesia,
Yazhenskikh published work containing a K2O-MgO phase diagram showing the potassium-
magnesium-oxide phases are only stable up to approximately 750˚C, below the breakdown
temperature of potassium carbonate.41 Above this temperature, the potassium then remains in the
slag or volatilizes (for their specific system) and does not react with magnesia. This explanation
strongly supports the lack of potassium reactions observed with MgO in MHD systems.
17
Figure 1.4) Phase diagram of MgO and K2O adapted from Yazhenskikh based on equilibrium
phases of potassium oxide and magnesium oxide in a slag environment.41
Alumina has a melting temperature near 2050˚C and has multiple stable and metastable
crystal structures but the most common is corundum (a-Al2O3). The electrical conductivity has
been measured up to 1300˚C to be around 4.4 x 10-4 S/m, which is sufficient for insulators in MHD
applications. Experiments in electrical conductivity have reported the resistance of alumina at high
temperatures exceeds that of air, requiring complex equipment for measuring the electrical
conductivity.42,43 Alumina is used for many refractory applications on its own and in alumino-
silicates like mullite. Spear and Allendorf were interested in the performance of alumina in glass
melting furnaces containing potassium and sodium hydroxides, motivating the development of a
potassium-aluminum oxide phase diagram.21 Their results found stable potassium aluminate
phases at temperatures as low as 500˚C with a melting temperature exceeding 1500˚C, indicating
potassium aluminate phases are likely to form under MHD operating conditions and degrade the
insulator material.21
18
Figure 1.5) Equilibrium phase diagram of Al2O3 and K2O adapted from Spear and Allendorf.21
The electrical properties of single crystal and polycrystalline insulators have been studied to
great extent in prior work and are difficult to replicate using the techniques available. However, a
thorough literature search has provided accurate values for the electrical properties of these
materials. As for the stability in extreme MHD environments, experiments have provided a rather
clear picture on the expected corrosion results for use as a plasma-exposed material. These results
can act as a measure of precision in the corrosion evaluation testing performed in this work. Should
the results of corrosion for these insulators not reflect those observed in previous MHD works, it
will be evident that the evaluation techniques do not accurately reflect true MHD conditions.
19
Figure 1.6) Electrical conductivity values of various oxide electrode and insulator materials of
interest for MHD use.
In the following chapters, the experimental procedures and results are presented on
evaluating the materials selected. Specifically, the electrical properties of GDC in Chapter 2 in a
manuscript publication and YDH in Chapter 4 are measured using techniques described. The
literature values of electrical conductivity of YSZ, magnesia, and alumina are used to evaluate the
electrical properties of those materials. The establishment of experimental methods which
constitute the potassium exposure testing as well as the results for all materials except YDH are
presented in Chapter 3 in the form of a manuscript publication. Results of potassium exposure of
YDH then follow in Chapter 4. The results are then discussed in Chapter 5 with conclusions on
the performance of these materials as MHD generator components.
20
1.3. Bibliography
(1) Faraday, M. Experimental Researches in Electricity, 2nd ed.; Richard and John Edward
Taylor, Printers and Publishers to the University of London, Red Lion Court, Fleet Street:
London, UK, 1849; Vol. 1.
(2) Lipscombe, G.; Penney, J.; Leyser, R.; Allison, H. J. Water under the Bridge. University of
Leicester Journal of Physics: Special Topics 2014, A5 (4), 2.
(3) Fälthammar, C.-G. The Discovery of Magnetohydrodynamic Waves. Journal of
Atmospheric and Solar-Terrestrial Physics 2007, 69 (14), 1604–1608.
https://doi.org/10.1016/j.jastp.2006.08.021.
(4) Karlovitz, B.; Halasz, D. US2210918.
(5) Proceedings of Advances in Magnetohydrodynamics; McGrath, I. A., Siddall, R. G.,
Thring, M. W., Eds.; Pergamon Press: Oxford, UK, 1963.
(6) Lin, S.; Resler, E. L.; Kantrowitz, A. Electrical Conductivity of Highly Ionized Argon
Produced by Shock Waves. Journal of Applied Physics 1955, 26 (1), 95–109.
https://doi.org/10.1063/1.1721870.
(7) Bates, J. L.; Bein, J. D.; Black, D. L.; Cadoff, L. H.; Calvo, R.; Book, L.; Daniel, J. L.;
Farabaugh, E.; Frantti, E. W.; Hosler, W. R.; Kochka, E. L.; Kuszyk, J. A.; McDaniel, C.
L.; Negas, T.; Perloff, A.; Retallick, F. D.; Rosing, B. R.; Sadler, J. W.; Borodina, D. R.;
Brozevsky, V. V.; Burenkov, D. K.; Gordon, V. G.; Ivanov, A. B.; Kirillov, V. V.;
Romanov, A. I.; Strekalov, N. V.; Telegin, G. P.; Vysotsky, D. A.; Zalkind, V. I. Joint
U.S.-U.S.S.R. Test of U.S. MHD Electrode System in the U02 Facility (Phase III);
National Bureau of Standards, 1979.
(8) Rudins, G.; Schneider, S.; Bates, L.; Bowen, K.; Crawford, L.; Frederikse, H.; Rossing,
B.; Marchant, D. D.; Telegin, G. P.; Burenkov, D. K.; Romanov, A. I.; Kirillov, V. V.
Joint U.S.-U.S.S.R. Test of U.S. MHD Electrode Systems in U.S.S.R. U-02 MHD Facility
(Phase I); National Bureau of Standards, 1975; p 273.
Joint U.S.-U.S.S.R. Electrode Material Test System U-02 Westinghouse (Phase II);
(10) Kayukawa, N. Open-Cycle Magnetohydrodynamic Electrical Power Generation: A
Review and Future Perspectives. Progress in Energy and Combustion Science 2004, 30
(1), 33–60. https://doi.org/10.1016/j.pecs.2003.08.003.
(11) Bowen, H. K. Ceramics for Coal-Fired MHD Power Generation. In Materials science in
energy technology; Libowitz, G. G., Whittingham, M. S., Eds.; Materials science and
technology; Academic Press: New York, 1979; pp 181–198.
(12) Woodside, C. R.; Richards, G.; Huckaby, E. D.; Marzouk, O. A.; Haworth, D. C.; Celik, I.
B.; Ochs, T.; Oryshchyn, D.; Strakey, P. A.; Casleton, K. H.; Pepper, J.; Escobar-Vargas,
J.; Zhao, X. DIRECT POWER EXTRACTION WITH OXY-COMBUSTION: AN
OVERVIEW OF MAGNETOHYDRODYNAMIC RESEARCH ACTIVITIES AT THE
NETL-REGIONAL UNIVERSITY ALLIANCE (RUA); Pittsburgh, PA, USA, 2012; p
19.
21
(13) Lagow, B. W. Materials Selection in Gas Turbine Engine Design and the Role of Low
Thermal Expansion Materials. JOM 2016, 68 (11), 2770–2775.
https://doi.org/10.1007/s11837-016-2071-2.
(14) Rosa, R. J. Magnetohydrodynamic Energy Conversion, Rev. print.; Hemisphere Pub.
Corp: Washington, 1987.
(15) Kessler, R. Magnetohydrodynamics. Kirk-Othmer Encyclopedia of Chemical Technology;
John Wiley & Sons, Inc., 2000.
(16) Bowen, H. K.; Rossing, B. R. Materials Problems in Open Cycle Magnetohydrodynamics.
In Critical materials problems in energy production; Stein, C., Ed.; Academic Press: New
York, 1976.
(17) Rohatgi, V. K. High Temperature Materials for Magnetohydrodynamic Channels. Bull.
Mater. Sci. 1984, 6 (1), 71–82. https://doi.org/10.1007/BF02744172.
(18) Schneider, S. J. Compilation of the Melting Points of the Metal Oxides; Monograph 68;
National Bureau of Standards, 1963; p 40.
(19) Mills, K. The Estimation of Slag Properties; South African Institute of Mining and
Metallurgy: Misty Hills, South Africa, 2011.
(20) McBride, B. J.; Gordon, S. Computer Program for Calculation of Complex Chemical
Equilibrium Compositions and Applications. NASA Reference Publication 1996, No.
1311, 178.
(21) Spear, K. E.; Allendorf, M. D. Thermodynamic Analysis of Alumina Refractory Corrosion
by Sodium or Potassium Hydroxide in Glass Melting Furnaces. J. Electrochem. Soc. 2002,
149 (12), B551. https://doi.org/10.1149/1.1516773.
(22) Petrucci, R. H. General Chemistry: Principles and Modern Applications; Pearson Prentice
Hall: Toronto, 2011.
(23) Konter, M.; Bossmann, H.-P. Materials and Coatings Developments for Gas Turbine
Systems and Components. In Modern Gas Turbine Systems; Elsevier, 2013; pp 327–381e.
https://doi.org/10.1533/9780857096067.2.327.
(24) Clarke, D. R.; Phillpot, S. R. Thermal Barrier Coating Materials. Materials Today 2005, 8
(6), 22–29. https://doi.org/10.1016/S1369-7021(05)70934-2.
(25) Bates, J. L.; Marchant, D. D. Development of Materials for Open-Cycle
Magnetohydrodynamics: Ceramic Electrode; Pacific Northwest Laboratory, 1986.
(26) Buckley, J. D.; Wilder, D. R. EFFECTS OF CYCLIC HEATING AND THERMAL
SHOCK ON HAFNIA STABILIZED WITH CALCIA, MAGNESIA, AND YTTRIA.
NASA TECHNICAL NOTE August 1969.
(27) Stacy, D. W.; Wilder, D. R. The Yttria-Hafnia System. J American Ceramic Society 1975,
58 (7–8), 285–288. https://doi.org/10.1111/j.1151-2916.1975.tb11476.x.
(28) Momma, K.; Izumi, F. VESTA 3 for Three-Dimensional Visualization of Crystal,
Volumetric and Morphology Data. J. Appl. Cryst. 2011, 44, 1272–1276.
https://doi.org/10.1107/S0021889811038970.
(29) Zavodinsky, V. G. The Mechanism of Ionic Conductivity in Stabilized Cubic Zirconia.
Phys. Solid State 2004, 46 (3), 453–457. https://doi.org/10.1134/1.1687859.
(30) Gatehouse, B. M.; Lloyd, D. J. The Crystal Structure of Potassium Metazirconate,
K2ZrO3, and Its Tin Analogue, K2SnO3. Journal of Solid State Chemistry 1970, 2 (3),
410–415. https://doi.org/10.1016/0022-4596(70)90099-X.
22
(31) Gatehouse, B. M.; Lloyd, D. J. The Crystal Structure of Beta-Potassium Dizirconate: β-

K2Zr2O5. Journal of Solid State Chemistry 1970, 1, 478–483.
https://doi.org/10.1016/0022-4596(70)90130-1.
(32) Gatehouse, B. M.; Nesbit, M. C. The Crystal Structure of the 2:5 Phase in the K2O-ZrO2
System: K4Zr5O12, a Compound with Octahedral and Trigonal Prismatic Zirconium(IV)
Coordination. Journal of Solid State Chemistry 1980, 31 (1), 53–58.
https://doi.org/10.1016/0022-4596(80)90007-9.
(33) Dash, S.; Sood, D. D.; Prasad, R. Phase Diagram and Thermodynamic Calculations of
Alkali and Alkaline Earth Metal Zirconates. Journal of Nuclear Materials 1996, 228 (1),
83–116. https://doi.org/10.1016/0022-3115(95)00199-9.
(34) Mogensen, M.; Sammes, N. M.; Tompsett, G. A. Physical, Chemical and Electrochemical
Properties of Pure and Doped Ceria. Solid State Ionics 2000, 129 (1–4), 63–94.
https://doi.org/10.1016/S0167-2738(99)00318-5.
(35) Panlener, R. J. A THERMODYNAMIC STUDY OF NONSTOICHIOMETRIC CERIUM
DIOXIDE. J. Phys. Chem. Solids 1975, 36, 1213–1222.
(36) Blumenthal, R. N.; Lee, P. W.; Panlener, R. J. Studies of the Defect Structure of
Nonstoichiometric Cerium Dioxidet. Journal of The Electrochemical Society 1975, 118
(1), 7. https://doi.org/10.1149/1.2407923.
(37) Jain, A.; Ong, S. P.; Hautier, G.; Chen, W.; Richards, W. D.; Dacek, S.; Cholia, S.;
Gunter, D.; Skinner, D.; Ceder, G.; Persson, K. A. Commentary: The Materials Project: A
Materials Genome Approach to Accelerating Materials Innovation. APL Materials 2013, 1
(1), 011002. https://doi.org/10.1063/1.4812323.
(38) Schieltz, J. D. Electrolytic Behavior of Yttria and Yttria Stabilized Hafnia. Retrospective
Theses and Dissertations, Iowa State University, 1970.
(39) Marchant, D. D.; Bates, J. L. RARE-EARTH HAFNIUM OXIDE MATERIALS FOR
MAGNETOHYDRODYNAMIC (MHD) GENERATOR APPLICATION. In The Rare
Earths in Modern Science and Technology: Volume 2; McCarthy, G. J., Rhyne, J. J.,
Silber, H. B., Eds.; Springer US: Boston, MA, 1980; pp 553–558.
https://doi.org/10.1007/978-1-4613-3054-7.
(40) Osburn, C. M.; Vest, R. W. Electrical Properties of Single Crystals, Bicrystals, and
Polycrystals of MgO. Journal of the American Ceramic Society 1971, 54 (9), 428–435.
https://doi.org/10.1111/j.1151-2916.1971.tb12380.x.
(41) Yazhenskikh, E.; Jantzen, T.; Hack, K.; Müller, M. Critical Thermodynamic Evaluation of
Oxide Systems Relevant to Fuel Ashes and Slags: Potassium Oxide–Magnesium Oxide–
Silica. Calphad 2014, 47, 35–49. https://doi.org/10.1016/j.calphad.2014.05.006.
(42) Will, F. G.; deLorenzi, H. G.; Janora, K. H. Effect of Crystal Orientation on Conductivity
and Electron Mobility in Single-Crystal Alumina. J American Ceramic Society 1992, 75
(10), 2790–2794. https://doi.org/10.1111/j.1151-2916.1992.tb05506.x.
(43) Peters, D. W.; Feinstein, L.; Peltzer, C. On the High-Temperature Electrical Conductivity
of Alumina. The Journal of Chemical Physics 1965, 42 (7), 2345–2346.
https://doi.org/10.1063/1.1696298.
23
Electrical Properties of Gadolinia-Doped Ceria for Electrodes for Magnetohydrodynamic

Energy Systems
by
Michael S. Bowen, Michael Johnson, Ryan McQuade, Bryce Wright, Kyei-Sing Kwong, Peter
Y. Hsieh, David P. Cann, and C. Rigel Woodside
SN Applied Sciences
Springer Nature Switzerland AG
Gewerbestr. 11
6330 Cham
Switzerland
Volume 2 Article1529
24
2. Electrical Properties of Gadolinia-Doped Ceria for Electrodes for Magnetohydrodynamic

Energy Systems
2.1. Introduction
Research and development of a direct power extraction concept using

magnetohydrodynamics (MHD) is motivated by the fact that this technology can provide a
significant increase in energy efficiency for chemical-to-electrical energy conversion by enabling
the operation of power plants at higher temperatures1. By utilizing high temperature gases as a
working fluid in a MHD topping unit, a combined-cycle coal-fired power plant could eventually
lead to plant thermal efficiencies above 60%2. While the thermodynamic advantages of a MHD
system have been known for years, legacy US Department of Energy (DOE) research was impeded
by challenges in the supporting technology and unfavorable techno-economics. Fortunately,
technological advancements in magnets, materials, and thermal fluids have improved since
previous DOE research efforts, improving the viability of MHD power generation. More recently,
interest in using oxy-fuel combustion to enable carbon dioxide capture has renewed R&D into
MHD topping cycles especially with the increased use of high temperature oxy-fuel combustion3.
Thus, implementation of MHD technology has the potential to significantly reduce fossil fuel
consumption and reduce greenhouse gas emissions.
A great challenge in development of a functional generator is material selection. The
working fluid of a MHD generator creates an extreme environment with combustion temperatures
as high as 3000°C, gas velocities up to mach-2, and the presence of ionizable species, specifically
potassium vapor. Materials are functionally exposed to less intense conditions at the boundary
layer, with temperatures up to 2400°C3,4. Previous versions of MHD generators used cold
electrodes which were high electrical conductivity metals; such as tungsten or copper, with
extreme water cooling to avoid reaching melting temperatures1,5. With direct exposure to the
seeded plasma, oxidation and potassium corrosion resulted in material failure. Alternative designs
for use with coal combustion relied on a protective slag which would shield the electrodes from
potassium and insulate the plasma to increase temperatures and improve efficiencies. The failure
in this design was potassium diffusion through the slag, at which point the temperature at the
25
surface of the electrodes was below the dewpoint of potassium resulting in a conductive liquid
which shorted the electrode pairs5–7. A review of MHD technologies conducted in 2004 reports a
more in-depth analysis of designs of MHD technologies5.
This motivates a transition to hot electrodes with direct exposure to the plasma, which
limits materials to refractory metals and high temperature ceramics1. There are a number of oxides
which can perform as conductors at MHD operating conditions, but the two of greatest interest
during initial MHD work were zirconia-based and lanthanum chromate-based oxides6–15. Zirconia
is stable in a monoclinic phase, which is generally a poor conductor, but rare earths and group IV
fluorites stabilize the system in the cubic fluorite phase, which shows to be a good electrical
conductor16. Yttria stabilized zirconia (YSZ) and ceria-zirconia were both tested in early research;
but failure occurred likely due to potassium corrosion, specifically reaction with zirconium
oxide6,8,14. This drives a push for zirconia free, high temperature conducting oxides. One particular
system, Gd doped ceria (GDC), is used in place of and in conjunction with YSZ, indicating they
share many similar properties in operating temperature ranges, corrosion resistance, and electrical
properties17–21.
In order to be considered a viable candidate material for MHD power generation, there are
a number of stability conditions and material properties that must be met. Phase stability plays a
huge role, as materials ideally can withstand operating temperatures ranging from 1500˚C to
2100˚C depending on thermal conductivity. Furthermore, materials must resist chemical reactions
with potassium, specifically K vapor which is highly reactive. For optimal power generation, the
conductivity of the material should be at least on the same order of magnitude as the plasma
conductivity, which is theorized to be between 101 to 102 S/m3,4. Finally, a specific consideration
for oxide materials is the pO2 level to determine whether the environment is reducing or oxidizing.
This can vary vastly based on the application, the equivalence ratio (a measure of fuel-oxygen),
seed concentration and temperature of the gases. For oxy-methane combustion, oxygen partial
pressures range between 10-9 and 1 atm. This work is solely focused on the electrical properties,
though other work is being carried out to address other requirements for the material system.
26
a gnet
M
ne l Wall
Chan Electrode
a Flow
Plasm Insulator
Figure 2.1) Schematic of a combustion driven open-cycle MHD power generator showing direct
plasma exposure of the channel wall, composed of electrode and insulator segments.
2.2. Review
This work considers use of cerium oxide doped with gadolinium, 𝐺𝑑- 𝐶𝑒/0- 𝑂(0$ , (GDC)
#
as a high temperature electrode material for MHD applications. The GDC system has broad interest
for oxygen transport applications including solid oxide fuel cells (SOFC) and catalysis17–21, which
are heavily reliant on maximizing oxygen transport. As a result, extensive research has been carried
out on the electrical properties of doped ceria, particularly GDC. This work has accumulated the
results of many others in order to draw general conclusions about properties of GDC and further
conduct an independent investigation to ensure our techniques meet the standards of current state
of the art.
Ceria is stable in the cubic fluorite structure (Fm-3m) with no reported phase changes up
to the melting point around 2750K22. Gadolinium oxide (Gd2O3) is typically stabilized in a
monoclinic phase, but transitions to the cubic c-type rare earth structure with small amounts of Ce
doping. The similarity in structure between the two supports the stable solid solution between CeO2
27
and Gd2O3. A phase diagram simulation from Žguns23 shows a stable fluorite phase for Gd doped
ceria (GDC) with intermediate amounts of doping and a phase transition to c-type rare earth near
50%.
Undoped ceria has relatively low electrical conductivity at atmospheric oxygen levels.
Blumenthal et al.24 show that under reducing conditions, however, the conductivity exhibits a pO2-
1/5
dependence and Panlener25 found the same relation for oxygen vacancies in reduced CeO2. The
promotion of oxygen vacancies derives from multiple valence states of cerium, represented by the
defect reaction below in equation (2.1). There is a similar reaction of hole compensation which
explains the electronic conductivity of ceria in equation (2.2). The pO2 dependence is consistent
with p-type conductivity according to defect equilibrium analysis by Smyth26. The introduction of
Gd3+ doping improves conductivity by the defect reaction in equation (2.3). This reaction shows
the promotion of oxygen vacancies with Gd doping, but there is a similar electronic compensation
to equation (2.2), which is likely to occur in oxidizing environments.
12+#
,
1
2𝐶𝑒𝑂( @⎯B 2𝐶𝑒12 + 3𝑂+- + 𝑉+⋅⋅ + 𝑂( ↑ (2.1)
2
12+#
,
𝐶𝑒𝑂( @⎯B 𝐶𝑒12 + 2𝑂+- + ℎ⋅ (2.2)
12+#
,
𝐺𝑑( 𝑂) @⎯B 2𝐺𝑑12 + 3𝑂+- + 𝑉+⋅⋅ (2.3)
An extensive literature search into measured electrical properties of GDC has resulted in
numerous sources reporting electrical conductivity for various dopant levels18,19,27–42. Most reports
included data on compositions with 10% (10GDC) and 20% (20GDC) dopant levels as
summarized in Figures 2.2 and 2.3 respectively. The conduction mechanisms in this system are
oxygen vacancies and small polarons, concluded in the review by Mogensen22, which are both
/
hopping mechanisms. As a result, an additional &
term is included in the Arrhenius equation to
determine the activation energy of the system, given by22:
𝐴 𝐸!
𝜎= exp N− O (2.4)
𝑇 𝑘" 𝑇
3⋅4
where 𝐸! is the activation energy, and 𝐴 is the intercept conductivity in 56
.
28
The conductivity data shown in Figures 2.2 and 2.3 show quite significant variability in the
magnitude of the conductivity values. Using 500 °C as a reference point, 20 mol% Gd-doped CeO2
exhibited an average conductivity of 0.38 S/m with a standard deviation of 0.19 S/m, and 10
mol% Gd-doped CeO2 exhibited a larger variation with an average conductivity of 0.43 S/m with
a standard deviation of 0.26 S/cm. There are many potential explanations for these large ranges of
conductivity values. The data represented in Figures 2.2 and 2.3 were obtained from different
measurement techniques, including ac impedance spectroscopy vs dc I-V measurements and
four-point vs two-point probe methods, as well as different electrodes which may impart differ-
ences in the contact resistance. In addition, while the compositions are nominally the same
stoichiometry, the purity of the starting materials varied between the different literature reports.
This is likely to affect the overall conductivity values, especially when considering the grain
boundary contribution to the overall conductivity. There may also be differences in oxygen
stoichiometry due to the presence of impurities or variations in processing conditions. Finally, in
considering that all the samples in Figures 2.2 and 2.3 were polycrystalline ceramics, variations in
density and grain size would be expected to influence variability in the measured conductivity
values.
29
Figure 2.2) Comparison of accumulated data from various sources27–37 of 𝐺𝑑#./ 𝐶𝑒#.8 𝑂/.89 .
30
Figure 2.3) Comparison of accumulated data from various sources18,19,27,29,31,33,36,38–42 of

𝐺𝑑#.( 𝐶𝑒#.: 𝑂/.8 .
2.3. Methods
Ceramic pellets were fabricated for this study to confirm the current state of the art in
electrical conductivity of GDC. Starting CeO2 (99.9%, <5𝜇𝑚, Sigma Aldrich) and Gd2O3 (99%,
Sigma Aldrich) powders were mixed in the appropriate stoichiometric ratios as reported in Online
Resource 1. The mixed powders were combined with ethanol and 8% YTZ media, then milled for
8 hours. Initial powders were reported by Sigma Aldrich to have a D50 distribution <5𝜇𝑚 further
confirmed by measurements with a laser diffraction particle size analyzer (Malvern Mastersizer
31
3000). Post-milled powders were measured to be <2.5𝜇𝑚 using the same measurement. The milled
powders were then dried at 115˚C for 8 hours and pressed into green bodies using a 13mm diameter
die under 170 MPa pressure for 30 seconds. The green bodies were finally sintered at 1600˚C for
30 hours in magnesium oxide crucibles. This method differs from standard solid-state techniques
in that no calcination process was used, motivated by stable fluorite phase of the ceria powder and
long sintering times.
Densities of sintered pellets were measured per Archimedes principle with a Mettler
Toledo XS64 balance and the standard density kit. Averages of eight samples of each composition
showed above 95% densities for nearly all samples. Phase stability was verified by performing X-
ray diffraction (XRD) with a Bruker AXS D8 Discover on pellets crushed into powders with a
mortar and pestle using a Bragg-Brentano geometry. Fluorite peak locations were extracted, and a
lattice constant was calculated from each peak using Bragg’s law. The experimental lattice
parameter was approximated as the average lattice spacing of all peaks to calculate theoretical
density.
Electrical conductivity was measured on disk and bar samples using 2-point AC
Electrochemical Impedance Spectroscopy (EIS), outlined by Sinclair and others44, to extract bulk
and grain boundary conductivities. Disk samples were approximately 10.75mm diameter with
varying heights between 1.5mm and 3mm. Platinum electrodes were applied to the circular faces
and AC EIS was performed using a Solartron SI1260A impedance analyzer along with the
Solartron 1296A dielectric interface. Non-linear least squares fittings were performed with the
following equivalent circuit model (Figure 2.4) to extract grain (g), grain boundary (g.b.), and
electrode (el.) resistances.
𝑍 = 𝑅;<'= + (𝑅𝑄)> + (𝑅𝑄)>.?. + (𝑅𝑄)2@. (2.5)

1
𝑅𝑄 = + 𝑇(𝑖𝜔)< (2.6)
𝑅
32
Figure 2.4) Equivalent circuit diagram of a brick layer model for AC circuit modeling44.
The total sample conductivity is a combination of grain and grain boundary resistivities as follows:
1 1
𝜎=A= = = (2.7)
𝜌=A= 𝜌> + 𝜌>.?.
where resistivity is defined by:
𝑅𝐴
𝜌= (2.8)
𝐿
Separate electrical conductivity measurements were performed with a 4-point DC method
on bar sample geometries to verify the total conductivity values extracted from AC EIS. Pellets
pressed into 25.4mm diameter by 3.5mm height were cut into bars with final geometries of
approximately 3.5mm length by 5mm width by 18.5mm height. Surface electrodes were applied
to both end surfaces as well as on the perimeter of a cross section at 5mm from each end. A DC
current was applied at the surfaces using a Keithley 6221 Current Source Meter while measuring
the voltage across the ring electrodes with a Keithley 2182a nanovoltmeter. Total sample
resistance is calculated from the slope of the I-V curve and conductivity is calculated from
equations (2.5-2.8) above, with the length 𝐿 being the spacing between ring electrodes.
All electrical measurements were performed in a NorECS ProboSTAT high temperature
measurement cell heated in a horizontal tube furnace at 100˚C increments from room temperature
to 1100˚C. The temperature inside the cell was tracked with an r-type thermocouple logged with
time to ensure temperature stability.
33
2.4. Results & Discussion
X-Ray diffraction patterns are shown in Figure 2.5B for increasing gadolinium content.
Peak locations verify the cubic fluorite structure for all compositions. As Gd content increases,
peaks shifted to the left indicating an increase in lattice parameter which is expected 45. The average
lattice spacing was calculated from peak heights and reported in Online Resource 1. The respective
d-spacing shows trends similar to those observed by Artini et al.46, verifying the deviation from
Vegard’s Law despite being a solid solution. The majority of pellets fabricated were greater than
95% of theoretical density. Pure CeO2 is found to have lower densities, which indicates Gd doping
improves sinterability. Samples with density below 95% were not used to measure electrical
properties.
Figure 2.5) A) above: Average density and

standard deviation of 8 samples of each GDC
composition. B) right: XRD patterns of GDC
pellets ground into powders. The Miller indices
shown are for fluorite reflections.
At temperature increments of 100˚C, AC EIS measurements were performed on two disk

samples of each composition to resolve grain and grain boundary resistances. Figure 2.6A displays
impedance spectra representative of the specimens in this study, where only two mechanisms are
activated throughout the frequency range of the measurement for any given temperature, aside
from a few samples. As a result, data is typically fitted with only two of the three RQ elements in
34
equation (2.5) and Figure 2.4 above. Whether the two elements correspond to the bulk, grain
boundary, or electrode is determined based on the capacitance of the fitted equivalent circuit
elements43,47. Furthermore, as the temperature increased, the frequency at which the peak
imaginary resistance occurs, known as 𝜔# , increased following the Arrhenius equation. Figure
2.6B depicts the gradual appearance of the lower frequency mechanism as temperature increases,
which corresponds to the electrode impedance contribution. At 300˚C, the peak in imaginary
impedance arising from the bulk conductivity is apparent at around 104 𝜔/𝜔# , as in Figure 2.6B,
but disappears from the measurable frequency range of this system at higher temperatures.
Figure 2.6 A) Impedance spectra of one sample of each composition at 600˚C showing the grain
boundary for each except 50GDC. Second semicircles, where present, represent electrode
contributions. B) Imaginary bode plot normalized to the frequency and impedance at which the
peak value of the grain boundary impedance occurs for a pure ceria disk sample.
35
Figure 2.7) Electrical conductivity measured by ac impedance spectroscopy for disk samples
prepared in this work for increasing Gd dopant concentration over the temperature range of 300˚C
to 900˚C. Solid lines show the fit of equation (2.4) for each sample.
𝐶𝑒𝑂( 𝐺𝑑./ 𝐶𝑒.8 𝑂/.89 𝐺𝑑.( 𝐶𝑒.: 𝑂/.8 𝐺𝑑.) 𝐶𝑒.B 𝑂/.:9 𝐺𝑑.9 𝐶𝑒.9 𝑂/.B9
𝐸! [𝑒𝑉 ] -1.00 -1.01 -0.87 -1.11 -1.41
𝑆
log(𝐴) N 𝐾O 7.1 8.3 8.1 8.8 9.6
𝑚
Table 2.1) Activation energy and 𝐴 intercept for each of the Arrhenius equation plotted in Figure
2.7 fit to equation (2.4).
As described above, total conductivity was calculated for each composition and fitted to
the Arrhenius equation (2.4). The results in Figure 2.7 show trends similar to those observed by
others27,29,31,33,36 who found a maximum conductivity in 20% Gd doping and that at higher
36
temperatures, variation between samples decreases. By looking at Hohnke’s work, which exceeded
the temperature range of others, conductivities converged with a change in activation energy at the
temperature where each composition meets in the Arrhenius plot31. Above this temperature,
samples appear to have little variation in conductivity with dopant level, suggesting doping has
little effect on the overall conductivity at ultra-high temperatures. A more detailed analysis into
the higher temperature conductivity is needed to confirm this hypothesis.
The limiting factor for higher temperature measurements with 2-point AC EIS is the
instrument resistance, which is on the order of magnitude of the samples at temperatures of 900˚C.
Four-point resistance measurements are one resolution to this issue. A further study to observe
discrepancies between the 2-point AC EIS and 4-point DC resistance on bar geometries was done
for undoped ceria. It was found that conductivities from each of the three techniques, AC EIS for
disk geometries, AC EIS for bar geometries, and 4-point DC resistance for bar geometries, all
range within a factor of 4 as observed in the data in Figure 2.8. The change in mechanism at
temperatures above 1000˚C is apparent, though the number of data points is insufficient to reliably
fit an activation energy. The use of bar geometries over disk geometries has little effect on the
conductivity values obtained from AC impedance spectroscopy, as the activation energies are
nearly equal with values of 0.98 eV and 1.0 eV, respectively. However, there is a clear difference
in activation energies between AC and DC measurements performed on the bar sample. This
discrepancy is possibly from error in low temperature measurements with the AC technique.
37
Figure 2.8) Arrhenius plot of electrical conductivity of different measurement techniques for
undoped ceria. Solid lines represent Arrhenius equation fitting for each method.
38
AC Disk AC Bar DC Bar

𝐸! [𝑒𝑉 ] -1.00 -0.98 -0.86
𝑆
log (𝐴) N 𝐾O 7.1 7.2 6.5
𝑚
Table 2.2) Fitted Arrhenius equation values of the data in Figure 2.8 for the activation energy
and 𝐴 −intercept defined in equation (2.4) above.
2.5. Conclusions
In this study, we explored the electrical properties of pure and Gd-doped ceria to determine
its viability as an electrode material for MHD power generation. While thermal engineering
challenges restrict electrical measurements performed at the expected operating temperatures
(~2000°C), it is apparent that the expected conductivity values will reach those required for
efficient power generation with values on the order of 10 S/m at temperatures as low as 1100˚C.
The accumulation of data from many sources leads to a reasonable range of values which helps
understand both the material property and the limits in manufacturing of GDC for MHD
applications. Beyond the electrical properties, there are other important material characteristics
needed to fully evaluate the suitability of GDC for MHD applications. Based on our findings, Gd-
doped ceria has adequate electrical properties for use in hot-electrode MHD generators.
39
2.6. Bibliography
(1) Kessler, R. Magnetohydrodynamics. Kirk-Othmer Encyclopedia of Chemical Technology;
John Wiley & Sons, Inc., 2000.
(2) Rosa, R. J. Magnetohydrodynamic Energy Conversion, Rev. print.; Hemisphere Pub.
Corp: Washington, 1987.
(3) Woodside, C. R. Retrospective and Prospective Aspects of MHD Power Generation, 2014.
(4) Bedick, C. R.; Kolczynski, L.; Woodside, C. R. Combustion Plasma Electrical
Conductivity Model Development for Oxy-Fuel MHD Applications. Combustion and
Flame 2017, 181, 225–238. https://doi.org/10.1016/j.combustflame.2017.04.001.
(5) Kayukawa, N. Open-Cycle Magnetohydrodynamic Electrical Power Generation: A
Review and Future Perspectives. Progress in Energy and Combustion Science 2004, 30
(1), 33–60. https://doi.org/10.1016/j.pecs.2003.08.003.
Joint U.S.-U.S.S.R. Test of U.S. MHD Electrode Systems in U.S.S.R. U-02 MHD Facility
(Phase I); National Bureau of Standards, 1975; p 273.
(7) Rudins, G. US and USSR MHD Electrode Materials Development. Technology
Assessment Office December 1974.
Joint U.S.-U.S.S.R. Electrode Material Test System U-02 Westinghouse (Phase II);
(9) Bates, J. L.; Bein, J. D.; Black, D. L.; Cadoff, L. H.; Calvo, R.; Book, L.; Daniel, J. L.;
Farabaugh, E.; Frantti, E. W.; Hosler, W. R.; Kochka, E. L.; Kuszyk, J. A.; McDaniel, C.
L.; Negas, T.; Perloff, A.; Retallick, F. D.; Rosing, B. R.; Sadler, J. W.; Borodina, D. R.;
Brozevsky, V. V.; Burenkov, D. K.; Gordon, V. G.; Ivanov, A. B.; Kirillov, V. V.;
Romanov, A. I.; Strekalov, N. V.; Telegin, G. P.; Vysotsky, D. A.; Zalkind, V. I. Joint
U.S.-U.S.S.R. Test of U.S. MHD Electrode System in the U02 Facility (Phase III);
(10) Telegin, G. P.; Romanov, A. I.; Rekov, A. I.; Spiridonov, E. G.; Barodina, T. I.; Vysotsky,
D. A. Tests and Studies of U.S.S.R. Materials at the U.S. Coal Burning MHD Facility
UTSI-2. 40.
(11) Bates, J. L.; Marchant, D. D. Development of Materials for Open-Cycle
Magnetohydrodynamics: Ceramic Electrode; Pacific Northwest Laboratory, 1986.
(12) Meadowcroft, D. B. Electronically-Conducting, Refractory Ceramic Electrodes for Open
Cycle MHD Power Generation. Energy Conversion 1968, 8, 185–190.
(13) Sadler, J. W.; Bein, J.; Black, D. L.; Dilmore, J. A.; Driesen, G. E.; Eggers, A. G.;
Kochka, E. L. Development, Testing and Evaluation of MHD Materials and Component
Designs. 205.
(14) U25 MHD Pilot Plant - Part II - MHD Generator Studies. September 1974.
(15) Patil, D. S.; Venkatramani, N.; Rohatgi, V. K. Electrical Conductivity of (ZrO2)_0.85
(CeO2)_0.12 (Y2O3)_0.03. Journal of Materials Science 1988, 23, 3367–3374.
40
(16) Zavodinsky, V. G. The Mechanism of Ionic Conductivity in Stabilized Cubic Zirconia.

Phys. Solid State 2004, 46 (3), 453–457. https://doi.org/10.1134/1.1687859.
(17) Ai, C.; Zhang, Y.; Wang, P.; Wang, W. Catalytic Combustion of Diesel Soot on Ce/Zr
Series Catalysts Prepared by Sol-Gel Method. Catalysts 2019, 9 (8), 646.
https://doi.org/10.3390/catal9080646.
(18) Badwal, S. P. S.; Fini, D.; Ciacchi, F. T.; Munnings, C.; Kimpton, J. A.; Drennan, J.
Structural and Microstructural Stability of Ceria – Gadolinia Electrolyte Exposed to
Reducing Environments of High Temperature Fuel Cells. J. Mater. Chem. A 2013, 1 (36),
10768. https://doi.org/10.1039/c3ta11752a.
(19) Eguchi, K.; Setoguchi, T.; Inoue, T.; Arai, H. Electrical Properties of Ceria-Based Oxides
and Their Application to Solid Oxide Fuel Cells. Solid State Ionics 1992, 52 (1–3), 165–
172. https://doi.org/10.1016/0167-2738(92)90102-U.
(20) Taroca, H. A.; Santos, J. A. F.; Dominguez, R. Z.; Matencio, T. Ceramic Materials for
Solid Oxide Fuel Cells. In Advances in Ceramics - Synthesis and Characterization,
Processing and Specific Applications; Sikalidis, C., Ed.; InTech, 2011.
https://doi.org/10.5772/18297.
(21) Tuller, H. L.; Nowick, A. S. Doped Ceria as a Solid Oxide Electrolyte. J. Electrochem.
Soc. 122 (2), 5.
(22) Mogensen, M.; Sammes, N. M.; Tompsett, G. A. Physical, Chemical and Electrochemical
Properties of Pure and Doped Ceria. Solid State Ionics 2000, 129 (1–4), 63–94.
https://doi.org/10.1016/S0167-2738(99)00318-5.
(23) Žguns, P. A.; Ruban, A. V.; Skorodumova, N. V. Phase Diagram and Oxygen–Vacancy
Ordering in the CeO 2 –Gd 2 O 3 System: A Theoretical Study. Phys. Chem. Chem. Phys.
2018, 20 (17), 11805–11818. https://doi.org/10.1039/C8CP01029C.
(24) Blumenthal, R. N.; Lee, P. W.; Panlener, R. J. Studies of the Defect Structure of
Nonstoichiometric Cerium Dioxidet. Journal of The Electrochemical Society 1975, 118
(1), 7. https://doi.org/10.1149/1.2407923.
(25) Panlener, R. J. A THERMODYNAMIC STUDY OF NONSTOICHIOMETRIC CERIUM
DIOXIDE. J. Phys. Chem. Solids 1975, 36, 1213–1222.
(26) Smyth, D. M. THE DEFECT CHEMISTRY OF METAL OXIDES, 1st ed.; Monographs on
the Physics and Chemistry of Materials; Oxford University Press: New York, 2000.
(27) Acharya, A. S.; Gaikwad, V. M.; Sathe, V.; Kulkarni, S. K. Influence of Gadolinium
Doping on the Structure and Defects of Ceria under Fuel Cell Operating Temperature.
Applied Physics Letters 2014, 104, 113508. https://doi.org/10.1063/1.4869116.
(28) Balazs, G. Ac Impedance Studies of Rare Earth Oxide Doped Ceria. Solid State Ionics
1995, 76 (1–2), 155–162. https://doi.org/10.1016/0167-2738(94)00242-K.
(29) Esposito, V.; Traversa, E. Design of Electroceramics for Solid Oxides Fuel Cell
Applications: Playing with Ceria. Journal of the American Ceramic Society 2008, 91 (4),
1037–1051. https://doi.org/10.1111/j.1551-2916.2008.02347.x.
(30) Grilo, J. P. F.; Macedo, D. A.; Nascimento, R. M.; Marques, F. M. B. Electronic
Conductivity in Gd-Doped Ceria with Salt Additions. Electrochimica Acta 2019, 318,
977–988. https://doi.org/10.1016/j.electacta.2019.06.148.
(31) Hohnke, D. K. Ionic Conduction in Doped Oxides with the Fluorite Structure. Solid State
Ionics 1981, 5, 531–534. https://doi.org/10.1016/0167-2738(81)90309-X.
41
(32) Kabir, A.; Santucci, S.; Van Nong, N.; Varenik, M.; Lubomirsky, I.; Nigon, R.; Muralt, P.;
Esposito, V. Effect of Oxygen Defects Blocking Barriers on Gadolinium Doped Ceria
(GDC) Electro-Chemo-Mechanical Properties. Acta Materialia 2019, 174, 53–60.
https://doi.org/10.1016/j.actamat.2019.05.009.
(33) Kharton, V. V.; Figueiredo, F. M.; Navarro, L.; Naumovich, E. N.; Kovalevsky, A. V.;
Yaremchenko, A. A.; Viskup, A. P.; Carneiro, A.; Marques, F. M. B.; Frade, J. R. Ceria-
Based Materials for Solid Oxide Fuel Cells. J. Mat. Sci. 2001, 36, 1105–1117.
(34) Kudo, T.; Obayashi, H. Mixed Electrical Conduction in the Fluorite-Type Cel_ Gd O. ,. J.
Electrochem. Soc. 1976, 123 (3), 5.
(35) Sındıraç, C.; Büyükaksoy, A.; Akkurt, S. Electrical Properties of Gadolinia Doped Ceria
Electrolytes Fabricated by Infiltration Aided Sintering. Solid State Ionics 2019, 340,
115020. https://doi.org/10.1016/j.ssi.2019.115020.
(36) Tianshu, Z.; Hing, P.; Huang, H.; Kilner, J. Ionic Conductivity in the CeO2–Gd2O3
System (0.05VGd/CeV0.4) Prepared by Oxalate Coprecipitation. Solid State Ionics 2002,
148, 567–573.
(37) Wang, S.; Kobayashi, T.; Dokiya, M.; Hashimoto, T. Electrical and Ionic Conductivity of
Gd-Doped Ceria. Journal of The Electrochemical Society 2000, 147 (10), 3606–3609.
(38) Fagg, D. P.; Kharton, V. V.; Frade, J. R. P-Type Electronic Transport in
Ce0.8Gd0.2O2−δ: The Effect of Transition Metal Oxide Sintering Aids. Journal of
Electroceramics 2003, 9, 199–207.
(39) Inaba, H.; Tagawa, H. Ceria-Based Solid Electrolytes. Solid State Ionics 1996, 83 (1), 1–
16. https://doi.org/10.1016/0167-2738(95)00229-4.
(40) Torrens, R. S.; Sammes, N. M.; Tompsett, G. A. Characterisation of
(CeO2)0.8(GdO1.5)0.2 Synthesised Using Various Techniques. Solid State Ionics 1998,
111 (1–2), 9–15. https://doi.org/10.1016/S0167-2738(98)00172-6.
(41) Van herle, J.; Horita, T.; Kawada, T.; Sakai, N.; Yokokawa, H.; Dokiya, M. Low
Temperature Fabrication of (Y,Gd,Sm)-Doped Ceria Electrolyte. Solid State Ionics 1996,
86–88, 1255–1258. https://doi.org/10.1016/0167-2738(96)00297-4.
(42) Zhang, C.; Sunarso, J.; Zhu, Z.; Wang, S.; Liu, S. Enhanced Oxygen Permeability and
Electronic Conductivity of Ce0.8Gd0.2O2−δ Membrane via the Addition of Sintering
Aids. Solid State Ionics 2017, 310, 121–128. https://doi.org/10.1016/j.ssi.2017.08.020.
(43) Sinclair, D. C. Characterization of Electro-Materials Using Ac Impedance Spectroscopy.
Bol. Soc. Esp. Ceram. Vidrio 1994, 34 (2), 12.
(44) Irvine, J. T. S.; Sinclair, D. C.; West, A. R. Electroceramics: Characterization by
Impedance Spectroscopy. Adv. Mater. 1990, 2 (3), 132–138.
https://doi.org/10.1002/adma.19900020304.
(45) Grover, V.; Tyagi, A. K. Phase Relations, Lattice Thermal Expansion in CeO2–Gd2O3
System, and Stabilization of Cubic Gadolinia. Materials Research Bulletin 2004, 39 (6),
859–866. https://doi.org/10.1016/j.materresbull.2004.01.007.
(46) Artini, C.; Costa, G. A.; Pani, M.; Lausi, A.; Plaisier, J. Structural Characterization of the
CeO2/Gd2O3 Mixed System by Synchrotron X-Ray Diffraction. Journal of Solid State
Chemistry 2012, 190, 24–28. https://doi.org/10.1016/j.jssc.2012.01.056.
42
(47) Guo, X.; Mi, S.; Waser, R. Nonlinear Electrical Properties of Grain Boundaries in Oxygen
Ion Conductors: Acceptor-Doped Ceria. Electrochem. Solid-State Lett. 2005, 8 (1), J1.
https://doi.org/10.1149/1.1830393.
43
2.7. Supplementary Information
Below is supplementary data which has been cited in the text, but excluded for conciseness.
X [mol%] 0% 10% 20% 30% 50%

Mass CeO2 [g] 25 22.38 19.79 17.23 12.18
Mass Gd2O3 [g] 0 2.62 5.21 7.77 12.82
Total [g] 25 25 25 25 25
Table 2.3) Table of calculated masses combined to produce the appropriate stoichiometric solid
solution of 𝐺𝑑- 𝐶𝑒/0- 𝑂(0C . Deviation between measured values deviated and calculated values
were less than ±1mg.
Lattice Parameter
X [mol%] 0% 10% 20% 30% 50%
Grover&Tyagi
[45] 5.411 5.41 5.432 5.437 5.431
Artini et al. [46] 5.4073 5.4209 5.4271 5.4321 5.4342
This Work 5.4116 5.4182 5.4221 5.4271 5.4252
Table 2.4) Table of lattice parameters from two sources as well as those calculated from the
XRD data provided in the manuscript.
44
Figure 2.9) Lattice parameter a0 with increasing Gd content for this work in comparison to data
reported by Artini et al. [46] and Grover, Tyagi [45]. Deviation is expected due to the method of
calculation.
45
High temperature corrosion stability of ceramic materials for magnetohydrodynamic

generators
by
Michael S. Bowen, Kyei-Sing Kwong, Peter Hsieh, David P. Cann, and C. Rigel Woodside
ASTM International Journal on Material Performance and Characterization
ASTM International
100 Barr Harbor Drive
PO Box C700
West Conshohocken, PA
19428-2959 USA
Volume 2, Issue 11
46
3. High temperature corrosion stability of ceramic materials for magnetohydrodynamic

generators
3.1. Introduction
Figure 3.1) Schematic of a direct fired open-cycle MHD generator with a segmented channel.
Schematic shows plasma flowing into a rectangular channel where an electrode-insulator wall is
exposed to potassium seeded combustion products. An orange arrow shows the direction of plasma
flow perpendicular to a green arrow representing the magnetic field.
Magnetohydrodynamic (MHD) power generators achieve direct power extraction through

the Lorentz force acting on ionized gases in a channel where the motion of the working fluid is
perpendicular to an external magnetic field, as depicted in Figure 3.1. The power output is
proportional to the plasma’s electrical conductivity multiplied by the squares of the velocity and
applied magnetic field. Open-cycle MHD can use combustion of different fuels such as methane,
coal, and other hydrocarbon fuels in air or pure oxygen as its energy source. Molecules that are
formed from the combustion process have very high ionization energies, which means that even
at temperatures in excess of 3000 K, the concentration of free electrons in the plasma and the
electrical conductivity are low too for efficient MHD power generation. As explained in
Magnetohydrodynamic Energy Conversion, introduction of a new species, alkali metals
47
specifically, can increase conductivity by several orders of magnitude with very low
concentrations.1 For example, Bedick et al. report a conductivity level over 50 S/m in oxy-methane
combustion products with a 1% mass fraction potassium “seed” added via a water solution
containing dissolved potassium carbonate.2
There are many advantages to using potassium over other elements. First, the alkali metals
have the lowest ionization energies on the periodic table.3 Cesium, which has the lowest ionization
energy of the elements, has been proposed for use in closed-cycle MHD system, but has been
avoided in open-cycle systems due to its higher cost. In comparison, potassium and sodium
containing compounds are relatively abundant, and potassium has the lower ionization energy of
the two. Next, while there are many potassium compounds which could be used to seed the
generator, potassium carbonate is particularly suitable for numerous reasons. The compound is
safe to handle, can be dissolved into water or methanol for injection into the system with the fuel,
can be recovered as K2CO3 from the exhaust product and recycled for continued use, and the
carbonate primarily decomposes into gaseous carbon dioxide at combustion temperatures.4–6
The added potassium carbonate also decomposes into potassium gas and potassium
hydroxide gas which forms via a reaction to the water vapor present. Using the Chemical
Equilibrium and Applications (CEA) software developed by NASA, the stable potassium phases
were predicted in both oxy-methane combustion and argon environments at 1 atm up to 3000 K
seeded with potassium carbonate and displayed in Figure 3.2.7 The increase in electrical
conductivity of the plasma arises from the partial ionization of potassium into ions and free
electrons (Figure 3.2A). The carbon dioxide produced from potassium carbonate injection only
increases the carbon dioxide content by 1% which does not affect material reactions. On the other
hand, the addition of potassium containing vapor is a major concern for materials corrosion.
48
Figure 3.2) Predominant potassium phases calculated from minimization of Gibb’s Free Energy
using CEA at varying temperatures and 1 atm pressure.7 Open circles represent phase transitions.
(A) Theoretical phases resulting from an input of stoichiometric oxygen-methane with an addition
of potassium carbonate to produce 1% potassium by weight. (B) Theoretical phases resulting from
an input of argon and potassium carbonate with a 20:1 mass ratio.
To extract power from the plasma, channel design and material selection play a significant
role in the lifetime and durability of the generator. The channel’s electrically insulating and
conducting materials are exposed to the hot gasses and plasma at temperatures up to 3000 K. Most
highly conductive materials, metals in particular, have significantly lower melting temperatures
than this but manageable channel temperatures can be achieved by water cooling the channel.
However, this lowers the boundary layer temperature to below the liquidus line of potassium
resulting in condensation of potassium on the surface which can electrically short the electrodes
leading to damage and reduced performance. Choosing materials which operate at temperatures
above around 1500 ˚C can circumvent this issue, and this limits the channel material selection to
refractory metals and ceramics. In direct fired channels, the high vapor pressure of some refractory
metal oxides, like tungsten and molybdenum, eliminates these metals as viable candidates. Simple
ceramics like magnesia, alumina, silica, and boron nitride have been previously studied, as well as
more complex ones like yttrium-stabilized zirconia (YSZ) and lanthanum chromate (LaCrO3) for
49
use in MHD power generation systems. Potassium reactions with materials can be predicted by
considering the basicity of the oxide. Potassium oxide, a basic oxide, reacts with acidic (e.g. silicon
oxide) and amphoteric oxides (e.g. aluminum oxide) whereas basic oxides (e.g. MgO) can be
expected to resist corrosion.8
Resistance to corrosion by alkali metal vapor and their compounds at high temperatures is
a challenging problem that is not unique to MHD applications. Fireside corrosion in biomass-fired
boilers, thermal barrier coatings for gas turbines, and silica-based refractory ceramics in glass
furnaces are some examples where high temperature potassium corrosion poses a threat to
materials.9–12 Use of biomass to complement or replace fossil fuels and recycling of waste glass
are growing industries with a high demand for structural materials. Operation of gas turbines in
“dirty environments” result in alkali contamination of water and fuel, resulting in similar operating
conditions to MHD. Each of these respective fields can benefit from methods for rapid screening
of candidate materials capable of sustained operation at high temperatures while exposed to vapor
potassium species.
This study established a baseline experiment to study potassium vapor exposure as a way
to evaluate candidate MHD channel materials for corrosion resistance. This approach uses a simple
case with the oxides: magnesia (MgO), alumina (Al2O3), ceria (CeO2), and yttria-stabilized
zirconia (Y2O3-ZrO2). Magnesia and alumina were two of the most common MHD insulator
materials due to availability and cost.13 Magnesia was observed to resist potassium corrosion in
earlier reports which can be expected as it is strongly basic as reported by Kurushkin and
Kurushkin.14 On the other hand, alumina’s amphoteric nature results in the potassium-alumina
reactions which have been observed at high temperatures. Fluorite-stabilized zirconia was the most
successful ceramic electrode, despite known reactions with potassium.15 Pure zirconium dioxide
(ZrO2) undergoes a destructive phase transformation at around 1300˚C, however the introduction
of yttria above 8 mol% stabilizes the cubic fluorite phase over a wide temperature range.16 Yttrium
oxide is strongly basic while zirconia is amphoteric, so it follows that the potassium containing
phases are strictly zirconia based. It was later discovered that adding ceria to zirconia-based
ceramics resulted in improved corrosion resistance.15 Ceria, undoped and doped with gadolinium,
exhibits good electrical conductivity, reaching values up to 100 S/m at 1600 ˚C and has a melting
50
point around 2600 ˚C making it an attractive base material for MHD electrodes.17,18 Ceria is unique
to other oxides in that it can be stabilized under reduced conditions between fluorite CeO2 at
nominal conditions and rare-earth Ce2O3 in reducing environments.19 The rare-earth structured
ceria is basic and trends towards amphoteric with increasing oxidation. These four materials
systems were screened to reproduce potassium reactions in a more controlled environment and
provide a better understanding of the reaction mechanisms.
3.2. Materials
Discs of 10 mm diameter by 3 mm height were acquired from American Elements for

magnesia, ceria and YSZ, while alumina discs were cut from a 10 mm diameter rod obtained from
a separate commercial vendor (CoorsTek).20–23 Samples were exposed to vapor potassium through
volatilization of potassium carbonate in a tube furnace under a controlled atmosphere. Magnesium
oxide crucibles were used to contain the potassium carbonate, as it has been previously observed
that magnesia resists potassium corrosion.13,15,24–27 This was further validated by the results of this
experiment. The hygroscopic nature of potassium carbonate imposes an additional challenge in
preventing water contamination. As reported in the Merck Index, potassium carbonate readily
absorbs water from the atmosphere up to 16%, which was mitigated by calcining for up to 6 hours
at 800˚C.28 Measuring the mass after calcination provides a more accurate value for the mass of
initial potassium carbonate but risks exposure to moisture.
3.3. Experiment
A modified version of ASTM Standard Test Method for Vapor Attack on Refractories for
Furnace Superstructures was implemented in this study.29 Potassium carbonate is melted and
volatilized inside a covered crucible. The crucible assembly is altered from the prescribed
dimensions laid out in ASTM C987-10 to accommodate sample dimensions commonly used to
characterize engineering ceramics. The assembly consists of a crucible of 30 mm diameter by 30
51
mm height, 2 g of potassium carbonate, a complementing lid with a 5 mm bore hole, and the
sample. As described in ASTM C987-10, the reactant height should not exceed 40% of the crucible
depth, where in this test the potassium carbonate height remained below 25% of the crucible depth
preventing molten reactant from contacting the sample surface. Figure 3.3A provides a diagram of
the assembly and exploded view of the components. The sample is placed over the hole so that
potassium vapor may react with the material of interest as it escapes. The 5 mm bore hole permits
sufficient sample area to be exposed without extending beyond the circular face. The sample,
crucible, and potassium carbonate are placed in a closed-atmosphere furnace as in Figure 3.3B and
heated to 1250 ˚C at a rate of 5 ˚C/min. The furnace is then held at temperature for 24 hours before
cooling to room temperature at 8 ˚C/min. Mass measurements of each component are taken before
and after heating to measure potassium carbonate evaporated and potential changes in sample
mass.
Figure 3.3) (A) Crucible, sample, and potassium carbonate assembly. (B) Furnace setup with
representative oxygen and water vapor sensors. Approximate crucible and thermocouple locations
show temperature sampling occurs at position of assembly.
The environment inside the tube furnace is carefully monitored and controlled to ensure a
simplified reaction model. An inert environment provides a simple breakdown reaction of
potassium carbonate based on minimization of Gibb’s Free Energy with the Chemical Equilibrium
with Applications code developed by McBride and Gordon.7 Under a pure argon environment,
monoatomic potassium gas is the sole potassium species but trace amounts of water vapor permit
52
formation of potassium hydroxide. As depicted in Figure 3.4, by maintaining below 50 ppm water
vapor content, monoatomic potassium vapor dominates over potassium hydroxide by a factor of
10, ensuring this is the primary species in corrosion. While the presence of potassium hydroxide
is important in the thermochemistry of MHD combustion, focusing on exposure to potassium
vapor simplifies the reaction model.
Figure 3.4) Mole fractions of potassium vapor and potassium hydroxide vapor in an argon
environment containing potassium carbonate at 1 atm with varying trace amounts of water vapor.
Shaded region shows the range of water vapor content measured to be between 10 ppm and 50
ppm across all tests.
53
Ultra-high purity argon (99.999 %) was consecutively flown through a trace oxygen
analyzer (Alpha-Omega 3500) followed by a dewpoint meter (Vaisala DMT 3408), as shown in
the diagram in Figure 3.3B, to monitor gas contaminants with a precision of 1-100 ppm. A gas
chromatograph (Agilent Micro 3000) sampled gases at the exit to monitor potassium carbonate
volatilization in experiments. As mentioned above, carbon dioxide is a byproduct of potassium
carbonate decomposition. After identifying the carbon dioxide peak, multiple exposure tests were
measured for carbon dioxide prevalence and showed potassium carbonate volatilization for the
extent that the furnace temperature exceeded 900˚C, which is consistent with results from
thermogravimetric analysis performed by Lehman and others.5,30 From Figure 3.5, it was evident
that potassium breakdown initiated between 800˚C and 1000˚C and continued to volatilize
throughout the duration of the experiment. This demonstrated that the material is concisely being
exposed to potassium containing gas throughout the duration of the experiment. The sampling rate
of the gas chromatograph system was too low in relation to the heating rate of the furnace to resolve
the temperature at which potassium carbonate breakdown initiated.
Figure 3.5) Carbon dioxide peak intensity extracted from gas chromatograms at varying
temperature throughout a potassium exposure experiment of YSZ. Peak intensity shows potassium
54
carbonate breakdown beginning above approximately 900˚C and proceeding until cooling below
800˚C.
After exposure, samples were determined to have reacted based on two main
characterization techniques. Changes in mass are measured with a precision of 0.01 mg using an
analytical balance (Sartorius CP225D). Each sample was subjected to an initial null test under
conditions of the corrosion test excluding potassium. Samples which showed mass changes during
the initial null test were subjected to a second null test which resulted in less than a 0.25 mg
difference across all samples. The mass loss of the initial null test may be a result of either
dehydration of samples or reduction due to the low oxygen environment. After the null test,
samples are subjected to the potassium exposure test and similar mass measurements are made.
Corrosion was suspected if the variation in mass of the sample is greater than 0.5 mg. The second
characterization technique is surface x-ray diffraction (XRD) performed using a Bruker Discover
D8. The scan is completed on one of the faces of the sample before potassium exposure, and again
on both faces after exposure. Any peaks introduced in the scan of the surface exposed would lead
to the conclusion that a reaction has occurred, whether they are confirmed to contain potassium or
not. Using the methods described, three samples each of magnesia, alumina, ceria and YSZ were
tested for potassium corrosion resistance.
3.4. Results and Discussion
3.4.1. Magnesia
The mass changes of magnesia samples from the null and exposure step of the process are
shown in Figure 3.6 for each sample. A large decrease in mass >30 mg was observed in the initial
step, referred to as a bakeout, likely due to removal of absorbed water. The second null test showed
a stable mass to less than ±0.1 mg. Similarly, when exposed to potassium, the change in mass was
within ±0.1 mg, which suggests there was no reaction between potassium and magnesia under
55
these conditions. This was further validated in the x-ray diffraction data as seen in Figure 3.7.
These results are in line with those observed in previous MHD works.13,15,24–27
Figure 3.6) Mass of magnesia samples through each step with mass changes reported between each
step. The null and exposure test both indicated less than 0.1 mg mass changes.
Figure 3.7) X-ray diffraction pattern of an magnesia sample surface before exposure (orange) and
30˚ to 45˚.
56
3.4.2. Alumina
Unlike magnesia, alumina is known to react with potassium in various refractory

applications, including MHD tests.13,15,16,31 Several experiments using alumina components report
potassium reactions. More so, a calculated oxide phase diagram between alumina and potassium
oxide has been developed by Spear and Allendorf indicating a stable potassium-aluminum-oxygen
phase is expected at temperatures as high as 1250˚C.11
The alumina samples showed no mass change during an initial null test indicating a low
absorption of water and high stability in a reduction atmosphere. Thus, the initial null test was
accepted without a second test, as the mass change remained less than 0.1 mg for all three samples.
The mass change data provided in
Figure 3.8 clearly shows a reaction has occurred between alumina samples and potassium with a
consistent increase in mass on the order of 3 to 4 mg. A distinct corundum structure was evident
in the initial XRD pattern in Figure 3.9. Upon removal, the alumina samples had visible changes
in the form of a residue layer, assumed to be a hydrated reaction product. Several new peaks arose
from potassium reactions, some of which were identifiable as a potassium aluminate phase in
Figure 3.9.
57
Figure 3.8) Mass of alumina samples through each step with mass changes reported between each
step. The null test showed <0.1 mg mass change while the K-exposure tests indicate an
approximate 3.0, 4.2, and 4.1 mg increase in mass for each respective test.
Figure 3.9) X-ray diffraction pattern of an alumina sample surface before exposure (orange) and
30˚ to 45˚.
58
3.4.3. Yttria-stabilized zirconia
This study used 8 mol % yttria-stabilized zirconia to evaluate the corrosion resistance of
fluorite phase zirconia to potassium vapor. Like the magnesia samples, YSZ appeared to have a
significant mass loss between 1 mg and 2 mg during an initial bakeout, but a subsequent null test
showed a stable mass to 0.1 mg as in Figure 3.10. Though Sample 3 (fig. 10) was not subjected to
the same sequence of tests as Samples 1 and 2, the mass change resulting from the potassium
exposure is significant when compared the null tests from Samples 1 and 2. After exposure to
vapor potassium, a clear reaction was evident even by simple observation with the naked eye. A
protuberance was visible on the surface exposed to the potassium. Further mass measurements
show an increase in mass indicating reaction products on the surface of the zirconia sample
presumably chemically adhered. An increase in mass of approximately 1 mg was found for each
of the three samples. New peaks in the post-exposure XRD pattern indicated formation of a
potassium containing phase in Figure 3.11. The predominant new phase was identified as K4Zr5O12,
however minor peaks suggest other K-Zr-O phases are present.
Figure 3.10) Mass of YSZ samples through each step with mass changes reported between each
step, indicating <0.1 mg mass change for each sample. The exposure step resulted in >9 mg
increase in mass.
59
Figure 3.11) X-ray diffraction pattern of a YSZ sample surface before exposure (orange) and after
40˚.
3.4.4. Ceria
Prior to this study, ceria had not been studied for potassium corrosion resistance, though it
had been used as a stabilizer for YSZ in MHD experiments to reduce potassium corrosion.15 Phases
containing potassium, cerium and oxygen are not prevalent in the literature, though there exists an
XRD pattern (PDF 00-31-0989) for K2CeO3 which is similar to the Na2CeO3 structure.32 With the
knowledge that alkali corrosion is possible, a reaction between ceria and potassium is conceivable
but not necessarily expected.
Like the magnesia and YSZ samples, a mass change was observed from the initial null test
requiring a second null test resulting in a change in mass of less than 0.15 mg (fig. 12). While this
was greater than results observed in other samples, it was still low enough to accept for the null
test. The potassium exposure test resulted in one sample having a greater mass change of 0.23 mg,
which while greater than the null, was not large enough to suspect a potassium reaction. Further
x-ray (fig. 3.13) analysis showed that, within the detection limits of the XRD instrument, no new
60
phases were present on the surface exposed to potassium. Thus, it can be concluded that ceria
resists potassium vapor attack up to 1250 ˚C.
Figure 3.12) Mass of ceria samples through each step with mass changes reported between each
step. The null and exposure test both indicated <0.3 mg mass changes.
Figure 3.13) Diffraction pattern of a ceria sample surface before exposure (orange) and after
35˚.
61
3.5. Conclusions
This study shows that the methods used for potassium vapor exposure are valid for
evaluating the potassium corrosion resistance of a series of MHD channel materials. For the sample
geometries and quantity of potassium carbonate volatilized, mass changes that are less than 0.5
mg are considered to be inconclusive. Rather, potassium corrosion is considered to be evident if
the change in mass of the sample exceeds 0.5 mg. X-ray diffraction should be used in conjunction
with mass changes to confirm reactions through the detection of newly formed phases. If both
characterization techniques show imperceptible changes to the sample, then one can conclude that
the material resists corrosion under these conditions. For the purposes of this study, XRD was
sufficient to evaluate the corrosive behavior of materials in potassium, but X-ray photoelectron
spectroscopy offers a more precise measure of potassium contaminants for applications which may
be more sensitive to potassium infiltration. It should be noted that higher temperatures and
different oxygen partial pressures may still create thermodynamically favorable conditions for
corrosion reactions to occur.
The experimental data obtained in this study leads to the conclusion that alumina and YSZ
are susceptible to high temperature vapor potassium corrosion, particularly in an inert atmosphere.
In contrast, magnesia and ceria resist potassium vapor attack under these conditions. A more
accurate assessment can be undertaken by using an atmosphere more similar to actual MHD
conditions, including compounds such as carbon dioxide or water vapor. The temperatures of this
experiment were limited by the maximum temperature of the furnace. The test should be repeated
with a carbon dioxide-containing atmosphere to more accurately reflect MHD conditions, as well
as be performed at higher temperatures up to 2000˚C. This study found that ceria does not react
which might imply that GDC will perform well in MHD applications, however it is possible that
gadolinium doping may influence the susceptibility to potassium reactions or stabilize the
formation of a potassium-gadolinium-oxygen phase. Overall, these findings have helped provide
valuable experimental data for critical materials selection issues in the extreme environments in
the MHD channel, which is one of the principal limiting factors in the development of MHD-based
power systems.
62
3.6. Bibliography
(1) R. J. Rosa, Magnetohydrodynamic Energy Conversion, (Washington: Hemisphere

Publishing Corp., 1987)
(2) C. R. Bedick, L. Kolczynski, C. R. Woodside, "Combustion Plasma Electrical
Conductivity Model Development for Oxy-Fuel MHD Applications," Combustion and
Flame 181 (2017): 225–238, https://doi.org/10.1016/j.combustflame.2017.04.001
(3) R. H. Petrucci, General Chemistry: Principles and Modern Applications, (Toronto:
Pearson Prentice Hall, 2011)
(4) A. Y. Platonov, A. N. Evdokimov, A. V. Kurzin, H. D. Maiyorova, "Solubility of
Potassium Carbonate and Potassium Hydrocarbonate in Methanol," J. Chem. Eng. Data
47, no. 5 (2002): 1175–1176, https://doi.org/10.1021/je020012v
(5) R. L. Lehman; J. S. Gentry, N. G. Glumac, "Thermal Stability of Potassium Carbonate
near Its Melting Point," Thermochimica Acta 316, no. 1 (1998): 1–9,
https://doi.org/10.1016/S0040-6031(98)00289-5.
(6) T. R. Brogan, "Recent Progress in the Development of Combustion Fired M.H.D.
Generators," Philosophical Transactions of the Royal Society of London. Series A,
Mathematical and Physical Sciences 261, no. 1123 (1967): 360–367.
https://www.jstor.org/stable/73505
(7) B. J. McBride, S. Gordon, "Computer Program for Calculation of Complex Chemical
Equilibrium Compositions and Applications," NASA Reference Publication No. 1311
(1996), https://ntrs.nasa.gov/citations/19950013764
(8) H. G. Schurecht, "Reactions of Slag with Refractories: I. Surface Reactions," J American
Ceramic Society 22, (December 1939): 116–123, https://doi.org/10.1111/j.1151-
2916.1939.tb19437.x
(9) G. Zhang, M. Reinmöller, M. Klinger, B. Meyer, "Ash Melting Behavior and Slag
Infiltration into Alumina Refractory Simulating Co-Gasification of Coal and Biomass,"
Fuel 139, (January 2015): 457–465. https://doi.org/10.1016/j.fuel.2014.09.029
(10) M. Konter, H. P. Bossmann, "Materials and Coatings Developments for Gas Turbine
Systems and Components." in Modern Gas Turbine Systems, ed. P Jansohn, (Woodhead
Publishing, 2013), 327–381. https://doi.org/10.1533/9780857096067.2.327.
(11) K. E. Spear, M. D. Allendorf, "Thermodynamic Analysis of Alumina Refractory
Corrosion by Sodium or Potassium Hydroxide in Glass Melting Furnaces," J.
Electrochem. Soc. 149, no. 12 (March 2002): B551-B559.
https://doi.org/10.1149/1.1516773
(12) D. R. Clarke, S. R. Phillpot, "Thermal Barrier Coating Materials," Materials Today 8, no.
6, (June 2005): 22–29. https://doi.org/10.1016/S1369-7021(05)70934-2
(13) Materials Science in Energy Technology, ed. G. G. Libowitz, M. S. Whittingham,
Materials science and technology; (New York: Academic Press, 1979)
(14) M. Kurushkin, D. Kurushkin, "Acid–Base Behavior of 100 Element Oxides: Visual and
Mathematical Representations," J. Chem. Educ. 95, no. 4 (February 2018): 678–681.
https://doi.org/10.1021/acs.jchemed.7b00576
63
(15) G. Rudins, S. Schneider, L. Bates, H. K. Bowen, L. Crawford, H. Frederikse, B. Rossing,

D. D. Marchant, G. P. Telegin, D. K. Burenkov, A. I. Romanov, V. V. Kirillov, "Joint
U.S.-U.S.S.R. Test of U.S. MHD Electrode Systems in U.S.S.R. U-02 MHD Facility
(Phase I)," Final Report, United States: N. p. (1975), https://doi.org/10.2172/7237502
(16) D. B. Meadowcroft, "Electronically-Conducting, Refractory Ceramic Electrodes for Open
Cycle MHD Power Generation," Energy Conversion 8 (July 1968): 185–190,
https://doi.org/10.1016/0013-7480(68)90016-8
(17) R. N. Blumenthal, P. W. Lee, R. J. Panlener, "Studies of the Defect Structure of
Nonstoichiometric Cerium Dioxide," Journal of The Electrochemical Society 118 no. 1
(1971): 123-129. https://doi.org/10.1149/1.2407923
(18) M. S. Bowen, M. Johnson, R. McQuade, B. Wright, K.-S. Kwong, P. Y. Hsieh, D. P.
Cann, C. R. Woodside, "Electrical Properties of Gadolinia-Doped Ceria for Electrodes for
Magnetohydrodynamic Energy Systems," SN Appl. Sci. 2, no. 9 (2020)
https://doi.org/10.1007/s42452-020-03280-2
(19) M. Mogensen, N. M. Sammes, G. A. Tompsett, "Physical, Chemical and Electrochemical
Properties of Pure and Doped Ceria," Solid State Ionics 129, no. 1-4 (April 2000): 63–94,
https://doi.org/10.1016/S0167-2738(99)00318-5
(20) Yttria-Stabilized Zirconia 8 Mol % Disc, ZRO-Y08-02M-D, AmericanElements Merelex
Corporation, Los Angeles California.
(21) MgO Disc, MG-OX-02M-D, AmericanElements Merelex Corporation, Los Angeles
California.
(22) CeO2 Disc, CE-OX-02M-D, AmericanElements Merelex Corporation, Los Angeles
California.
(23) Alumina (99.8%) Rod, 1cm Diameter, Alumina AD-998 65956, CoorsTek Inc., Golden,
Colorado.
(24) V. K. Rohatgi, "High Temperature Materials for Magnetohydrodynamic Channels," Bull.
Mater. Sci. 6, no. 1 (1984): 71–82, https://doi.org/10.1007/BF02744172.
(25) G. Rudins, S. Schneider, L. Bates, H. K. Bowen, L. Crawford, H. Frederikse, B. Rossing,
D. D. Marchant, G. P. Telegin, D. K. Burenkov, A. I. Romanov, V. V. Kirillov, "Joint
U.S.-U.S.S.R. Electrode Material Test System U-02 Westinghouse (Phase II)," Final
Report, United States: N. p., (December 1977), https://doi.org/10.2172/6795339
(26) J. L. Bates, J. D. Bein, D. L. Black, L. H. Cadoff, R. Calvo, L. Book, J. L. Daniel, E.
Farabaugh, E. W. Frantti, W. R. Hosler, E. L. Kochka, J. A. Kuszyk, C. L. McDaniel, T.
Negas, A. Perloff, F. D. Retallick, B. R. Rosing, J. W. Sadler, D. R. Borodina, V. V.
Brozevsky, D. K. Burenkov, V. G. Gordon, A. B. Ivanov, V. V. Kirillov, A. I. Romanov,
N. V. Strekalov, G. P. Telegin, D. A. Vysotsky, V. I. Zalkind, "Joint U.S.-U.S.S.R. Test of
U.S. MHD Electrode System in the U02 Facility (Phase III)," Final Report, United States:
N. p. (1978), https://doi.org/10.2172/6263617
(27) J. W. Sadler, J. Bein, D. L. Black, J. A. Dilmore, G. E. Driesen, A. G. Eggers, E. L.
Kochka, "Development, Testing and Evaluation of MHD Materials and Component
Designs," Westinghouse Research and Development Center, Quarterly Report, United
States, N. p. January-March 31, 1978. https://doi.org/10.2172/6175307
(28) M. J. O’Neil, A. Smith, P. E. Heckelman, "Potassium Carbonate," The Merck Index An
Encyclopedia of Chemicals, Drugs, and Biologicals, (John Wiley & Sons, Inc., 2001).
64
(29) ASTM Committee C08 on Refractories. Test Method for Vapor Attack on Refractories for
Furnace Superstructures; ASTM International. https://doi.org/10.1520/C0987-10R19.
(30) M. S. Bowen, P. Y. Hsieh, C. R. Woodside, "Thermal Decomposition of Potassium
Carbonate for Seeding of Magnetohydrodynamic Generators," Poster Series, Am Chem
Soc NORM (June 2019)
(31) P. V. Ananthapadmanabhan, A. V. Bapat, A. K. Das, R. Majumder, R. K. Marwah, D. S.
Patil, V. K. Rohatgi, K. G. Sankaran, R. Sharma, P. B. Shrivastava, M. K. Totlani, N.
Venkat, "An Indo-Soviet Experiment on an MHD Generator Test Section at the Soviet U-
02 Facility," Proceedings of the Indian Academy of Sciences Section C: Engineering
Sciences 5 (1982): 169–195, https://doi.org/10.1007/BF02897682.
(32) S. D. Gates-Rector, T. N. Blanton, "The Powder Diffraction File: A Quality Materials
Characterization Database," International Centre for Diffraction Data 34, no. 4 (2019):
352–360
65
4. The Yttria-Hafnia System
4.1. Fabrication
Samples of yttria doped hafnia (YDH) were fabricated using the standard solid-state
practices presented in Chapter 2 and Appendix A. The diffusion of yttrium into hafnia is crucial to
producing a stable fluorite hafnia phase. According to the phase diagram from Stacy and Wilder,
yttria added into hafnia will stabilize the fluorite structure between 8 mol% and 50 mol%
concentrations when sintered at 1600˚C1. However, their processing technique used
coprecipitation methods along with isostatic pressing and reducing atmospheres to promote
oxygen vacancies, which were not executable in this study. Using standard oxide starting powders,
calcination temperatures were tested incrementally at 1000˚C, 1200˚C, and 1400˚C to determine
the minimum temperature to initiate the fluorite crystallization. The XRD patterns in Figure 4.1
show the results of increasing calcination temperatures indicating that 1400˚C is the minimum
calcination temperature to produce fluorite crystal powder. Two compositions were chosen for this
portion of the study, 20 mol% and 50 mol% Y-doped hafnia. As 50 mol% resides at the phase
boundary of the fluorite and c-type rare earth phases, it is ideal to observe the maximum yttria
diffusion into hafnia is achievable under the conditions presented. The 20 mol% Y-doped hafnia
was chosen to confirm that the fluorite phase could be stabilized.
The same powders which were calcined at incremental temperatures were sintered at
1600˚C for 30 hours to confirm that the sintering process could continue to promote the growth of
fluorite phase. After the final sintering step, XRD was performed yet again to analyze the phases
present. For the 20 mol% doping, there was a primary fluorite phase with some residual tetragonal
hafnia, likely a result of the calcinations at lower temperatures promoting grain growth of the
undoped phase. This could be reduced through longer sintering times, higher sintering
temperatures, and a secondary milling and calcination step. As for the 50 mol% doping, the
primary phase is c-type rare earth indicating that yttria diffusion is rapid at the sintering
temperatures used, but a low intensity fluorite peak still remains. This could be attributed to non-
66
ideal sintering conditions, errors in molar concentrations or the limitations from the
approximations interpreted from the phase diagram provided by Stacy and Wilder1.
Figure 4.1) X-ray diffraction patterns in the 2q range of 26˚ to 33˚ of ground calcined and sintered
powders of hafnia with 20 mol% and 50 mol% yttria additions.
67
After confirming the calcination and sintering temperatures, 20 mol% and 30 mol% Y-
doped hafnia were fabricated by calcining at 1400˚C for 8 hours and sintering at 1600˚C for 30
hours. It was concluded, from the residual tetragonal phase in the higher doped systems, that lower
concentrations of Y-doping would be more difficult to stabilize the fluorite phase. Thus, 20 mol%
and 30 mol% YDH samples were fabricated to evaluate the material system for MHD use. The
XRD patterns of the samples sintered for characterization showed a much lower impurity
tetragonal phase than those used in process development making them of good quality for MHD
material evaluation tests (Fig 4.2). Pellet densities were found to be greater than 85% of the
theoretical density, calculated using Archimedes Principle as described in Appendix A. The
residual monoclinic phase peaks are marked with a triangle in Figure 4.2.
Figure 4.2) X-ray diffraction pattern of sintered 20 mol% and 30 mol% yttria-doped hafnia with a
single calcination step compared to the 20 mol% sintered pellet subjected to incremental
calcinations in Figure 4.1A. Triangles mark residual monoclinic phase peaks while Miller indices
mark fluorite peaks.
68
4.2. Electrical Properties
The electrical properties of YDH were evaluated for 20 and 30 mol% Y-doping. While the
maximum conductivity appears at 8 mol% Y, concerns over fully stabilizing the fluorite phase at
this low concentration motivated the use of a higher content of yttria2. Electrical conductivity was
measured using AC EIS, described in depth in Chapter 2, at 50˚C increments for two samples of
20YDH and one sample of 30YDH. The circuit used to model the mechanisms in YDH was
equivalent to the one in Chapter 2 (Fig. 4.3). A fitting of the data for 20YDH at 544˚C compares
the calculated values of the circuit model to the measured data showing good agreement (Fig 4.4).
The fitting parameters are given in Table 4.1. The Arrhenius equation fitting is plotted as a solid
line against measured values as dots (Fig. 4.5). Error in conductivity measurements is dominated
by the standard deviation of circuit model fittings and uncertainty in temperature measurements.
A least-squares fitting was used to calculate the activation energy from Equation 1.2. The
activation energies for each of 20YDH, 30YDH and Schieltz’s data for 20YDH2 are displayed in
Figure 4.5. Schieltz also observed a decrease in electrical conductivity with increasing dopant
concentration, which is consistent with the results found from 30 mol% compared to 20 mol%, as
well as trends observed for the other rare-earth doped fluorites. The activation energy for 20 mol%
fits very well for the data measured up to 900˚C. Assuming this fitting is valid for temperatures up
to 2000˚C, a minimum temperature of 1800˚C is necessary to reach the required 100 S/m for MHD
use.
For the 20 mol% YDH, two samples were used to measure the electrical conductivity
variation within samples. The activation energies of the material are calculated for each sample
and found to be 1.395 eV ±0.031 eV for sample 1 and 1.360 ±0.019 eV for sample 2. The average
of the two samples falls within the error of both activation energies suggesting these samples are
roughly the same composition. However, the measured data from Schieltz was found to have an
activation energy of 1.324 eV ±0.006 eV which suggests there is some chemical variation between
the samples fabricated in our study versus the one fabricated in Schieltz’s study. There are many
possibilities for the variation, but the most likely candidate is impurities. First, the samples used
in this study were found to have some phase impurity, specifically a residual monoclinic phase.
69
Second, the quality of sintering can influence the electrical properties of the final sample. Third,
the quality and impurities of starting powders can translate into varied electrical properties of final
samples. Other possible sources of variation are the different temperature ranges measured and the
techniques used to measure electrical conductivity.
\Given these factors, the values for the activation energies are considered within reasonable
variation to be considered the same composition.
1 (Inst) 2 (Grain) 3 (G.B.) 4 (El)

Resistance [Ω] 1.649e3 1.943e4 8.329e3 Not used
Capacitance [F] 6.942e-11 1.419e-6 Not used
Phase (1 = Not used
0.957 0.612
capacitor)
Table 4.1) Fitting parameters resulting from a least squares fitting of the circuit model in Figure
4.3 to the produce the data represented in Figure 4.3.
Figure 4.3) Circuit model used to fit impedance spectroscopy data of YDH samples.
70
Figure 4.4) Circuit model fitting (orange boxes) versus measured data (blue dots) of a 20mol%
yttria-doped hafnia sample using electrochemical impedance spectroscopy techniques.
71
Figure 4.5) Arrhenius plot of the electrical conductivity of yttria-doped hafnia samples measured
using AC electrochemical impedance spectroscopy techniques. Reference data from Schieltz
included.
4.3. Corrosion Results
The material system was exposed to potassium vapor under a flowing argon environment
with less than 1 PPM oxygen. Three samples of 20 mol% YDH were tested for null and potassium
vapor reactions using the experiment described in Chapter 3 and Appendix A. The mass change
data (Fig. 4.6) shows a significant increase in mass for all three samples exposed to potassium.
Two of the three samples were found to have approximately equivalent mass changes while the
third sample was considerably less. However, all three have an increase in mass which reflects a
72
significant change in relation to that measured during the null test. XRD scans were performed
before and after testing to record any new phases that may have formed from potassium exposure.
In Figure 4.7, it is evident that the potassium vapor reacts with hafnia to form the K4Hf5O12 phase,
which is the most probable source of the increase in mass. Zhao et al. report that two primary K-
Hf-O phases form in potassium reactions, but based on the intensity of peaks, it is apparent that
K4Hf5O12 dominates over the K2Hf2O5 phase3. Finally, simple observation with the naked eye
clearly shows degradation of the sample in the form of visible bulging. Unlike the YSZ sample,
the protuberances on the surface of the YDH were sporadic where that of the YSZ was a single
smooth bubble. It is unclear whether this is due to the nature of the growth of the potassium
containing phases or the surface roughness of the samples.
These results can be compared to those described in the works of Bates and Marchant. In
their electrochemical corrosion testing, summarized in Chapter 1, they found an insignificant
change in sample before and after exposure to potassium seeded liquid slag4. As the materials
tested were of similar quality, i.e., mostly phase pure yttria-doped hafnia, the explanation of
differing results lies in the methods of potassium exposure. In the experiments performed in this
work, careful control was taken to ensure the exposure of material was to potassium vapor as
opposed to potassium compounds contained in a liquid slag (see Chapter 2). It is likely that the
liquid slag changed the chemistry of potassium reactions, with the potential for potassium
compounds such as sulfates and liquid carbonates. Another possible explanation is their exposure
temperatures appear to be less carefully controlled, only reporting a lower limit of 900˚C across
all tests. At higher temperatures, reactions would be accelerated, and greater mass changes are
expected to be observed.
73
Figure 4.6) Mass change values of 20 mol% yttria-doped hafnia samples exposed to potassium
vapor in an argon environment.
Figure 4.7) X-ray diffraction pattern of 20 mol% yttria-doped hafnia samples before and after
exposure to potassium vapor in an argon environment.
74
4.4. Bibliography
(1) Stacy, D. W.; Wilder, D. R. The Yttria-Hafnia System. J. Am. Ceram. Soc. 1975, 58 (7–8),
285–288. https://doi.org/10.1111/j.1151-2916.1975.tb11476.x.
(2) Schieltz, J. D. Electrolytic Behavior of Yttria and Yttria Stabilized Hafnia. Retrospective
Theses and Dissertations, Iowa State University, 1970.
(3) Zhao, Q.; Bugaris, D. E.; Stackhouse, C. A.; Smith, M. D.; zur Loye, H.-C. Crystal
Growth and Structure Determinations of Potassium Hafnates: K2Hf2O5 and K4Hf5O12.
Mater. Res. Bull. 2011, 46 (2), 166–169.
https://doi.org/10.1016/j.materresbull.2010.11.025.
(4) Marchant, D. D.; Bates, J. L. RARE-EARTH HAFNIUM OXIDE MATERIALS FOR
MAGNETOHYDRODYNAMIC (MHD) GENERATOR APPLICATION. In The Rare
Earths in Modern Science and Technology: Volume 2; McCarthy, G. J., Rhyne, J. J.,
Silber, H. B., Eds.; Springer US: Boston, MA, 1980; pp 553–558.
https://doi.org/10.1007/978-1-4613-3054-7.
75
5. Conclusions
5.1. Electrical properties
A summary of electrical properties is plotted in Figure 5.1. The limits of electrical

conductivity for electrodes and insulators provided by Rohatgi are plotted as horizontal dashed
lines. Materials are ideally well above or well below these limits at operating temperatures. The
curves for 8 mol% YSZ, MgO, and Al2O3 are the most accurate data pulled from literature based
on measurement controls or averages of multiple sources. The curves for GDC and YDH are
measured values from experimental data in this work. It is evident that YSZ is the best conductor
for MHD applications but as explained in other chapters, does not perform as well in terms of
corrosion resistance. Alternatives, such as GDC and YDH have lower electrical conductivities but
may still reach the suitable values at elevated temperatures. Above 1300˚C, GDC exceeds the
requirements for MHD electrodes making it a good alternative to YSZ. On the other hand, 20
mol% YDH has not been measured to exceed the minimum conductivity value but may still be
able to perform at higher temperatures. Extrapolated values of Arrhenius fittings indicate that YDH
can reach 100 S/m near 2000˚C making this the minimum temperature for this electrode to operate.
Also, others have found dopants such as terbia could improve conductivity over yttria to enable
lower operating temperatures. The insulator materials, Al2O3 and MgO, evidently remain
insulating for an extremely wide temperature range. Extrapolating the Arrhenius equation, alumina
approach concerning conductivity levels until temperatures exceeding 2100˚C while MgO does
not reach 1 S/m below its melting temperature, making them excellent candidates for MHD
channels.
The electrical conductivity has been determined for all materials in this study within the
limits of equipment available, but the temperature range of interest exceeds those of current
capabilities. Higher temperature measurements should be attempted to accurately measure the
electrical conductivity up to the operating temperature of MHD electrodes and insulators to
properly gauge material performance. These measurements can then improve modeling
capabilities when simulating the efficiencies of MHD generators constructed of these materials.
76
Of note, the temperature dependence of electrical conductivity can significantly impact loss
through Joule heating in MHD generators, as the plasma exposed surfaces will be much hotter than
those where electrical contact points are made to the external load.
Figure 5.1) Electrical conductivity of materials of interest for oxy-fuel open-cycle MHD power
generators. Solid lines are fitted curves for measured data in this work and others while dashed
lines are values extrapolated to higher temperatures.
5.2. Corrosion Resistance
From the preliminary results presented above, only two candidate MHD materials have
shown suitable corrosion resistance to vapor potassium species. One insulator, MgO, has shown
insignificant mass change in comparison to the null and no new phase formations from XRD for a
24-hour span of potassium vapor exposure. Similarly, the electrode material, ceria, shows no
reactions with potassium vapor. These are the best candidate materials for a direct plasma exposed
MHD channel based on the guidelines outlined in this work. Further work into in-situ testing with
MHD plasmas, structural and thermal properties, and fabrication techniques should be further
examined for channel fabrication.
On the other hand, YSZ, YDH, and alumina appear to lack the corrosion resistant nature
required for oxy-fuel combustion MHD materials. The presence of potassium vapors from seed
injection into combustion has the potential to quickly degrade these materials and reduce
performance. For alumina, the potassium reactions form a hygroscopic powder which is likely to
77
erode in MHD channels and degrade the material. For YSZ and YDH, the reacted compounds are
likely to be bonded with the surface and have the potential to vastly change the surface properties
of the material as well as introduce cracks and compromise the structural stability of the material.
These materials may still be viable for other MHD applications with less corrosive species, as
Bates and Marchant conclude in works on doped hafnia interacting with slag coating.
The corrosion results presented in this work only provide a preliminary evaluation as
replicating the conditions of MHD creates an extremely harsh environment for materials and
creating a containment chamber to simulate the MHD environment is as difficult as creating an
MHD channel itself. Improved evaluation methods are being researched at the National Energy
Technology Laboratory in Albany, OR involving the exposure of materials directly to a potassium
carbonate seeded oxy-kerosene combustion jet with temperatures, velocities, and species which
better represent MHD environments.
5.3. Final Remarks
As stated above, the corrosion evaluation procedure does not guarantee a material will
survive MHD conditions. Measuring the conductivity of a material at realistic operating
temperature is crucial for considering electrode materials, and further, the material will exhibit an
effective conductivity when subjected to the thermal gradients expected in an MHD channel. The
properties have been evaluated for dense, ideal polycrystalline pellets, but fabrication and
engineering of a MHD channel may result in properties that deviate from ideal. While MgO and
GDC appear to be strong candidates for MHD components, there are numerous additional
properties that must be understood in order to construct a successful channel. Rohatgi’s review of
MHD materials discusses them in great detail with large attention placed on the current density
values necessary for electrode materials and the breakdown voltage of insulator materials. Low
thermal conductivities are preferred to reduce heat loss; however, this problem can be solved
through design and careful choice in materials not exposed directly to the plasma. Finally, the
structural properties, namely the stress-strain relations and thermal expansion, of these materials
will play a role in the method of fabrication and channel design. Many of these issues will be
78
incorporated into future work on the development of high-performance materials for MHD
applications.
79
Appendix A: Full Documentation of Experimental Procedures
Equipment Model Number

Vibratory Mill Sweco M16-5 Finishing Mill
X-Ray Diffractometer Bruker D8 Discover
Hydraulic Press Carver 3912
Analytical Balance – Density Mettler-Toledo XS64
Analytical Balance – Corrosion Sartorius CP225D
Hygrometer Vaisala DMT3408
O2 meter Alpha-Omega 3500
Diamond Saw Buehler IsoMet Saw 11-1280-160
Platinum Ink for Electrodes SPI Platinum Conductive Paint
Current Source Meter Keithley 6221
Nano-voltmeter Keithley 2182A
Frequency Analyzer for ACEIS Solartron 1260A
Dielectric Interface for ACEIS Solartron 1296A
Table Appendix A.1) Equipment used for fabrication and characterization of materials in the
Electroceramics Laboratory at Oregon State University.
Sample Fabrication
Samples used in this study were both purchased from material suppliers and fabricated using
facilities in the Electroceramics Laboratory at Oregon State University. Specifically, ceria doped
with gadolinium and hafnia doped with yttrium were the two material systems fabricated in this
work. Samples were fabricated using basic solid-state practices used widely across the field. The
process requires some equipment listed in Table A.1., where model numbers of those found in the
lab are also provided.
The fabrication process begins with oxide powders where specifications of purity and particle
size are provided. The chemical formula is used to calculate the mass required of each powder in
order to produce the desired final stoichiometry.
80
Mass Calculations:
Consider the reaction below of a Group IV metal oxide (𝑀𝑂( ) doped with a rare-earth oxide
(𝑅( 𝑂) ) which form a solid solution. In this equation, 𝑥 is the percentage of the cations which are
the dopant metal.
𝑥
(1 − 𝑥)𝑀𝑂( + 𝑅( 𝑂) → 𝑅- 𝑀/0- 𝑂(0- (A.1)
2 (
The molar weights (in grams per mole) of each substance must be calculated. The desired
final weight of total powder (y) is then input to find the amount of each powder to mix. Typically,
batches of 10, 15, or 25 grams are made. Simply using the molar weight of the product, this mass
is converted to moles. The balanced chemical formula is used to find the ratio of moles of reactant
to moles of product. Finally, the molar weights of the reactants are used to find the required masses
of starting powders. In equations A.2 – A.4, the subscripts 𝐴, 𝐵, 𝐶 represent the reactants and
products in equation A.1 above.
𝑚; −mass of compound 𝑖
𝑀; −molar mass of compound 𝑖
𝑎 𝑦
𝑚* = 𝑀 (A.2)
𝑐 𝑀1 *
𝑏 𝑦
𝑚" = 𝑀 (A.3)
𝑐 𝑀5 "
The final calculations can be checked by the equation
𝑚" + 𝑚" = 𝑦 (A.4)
The correct mass ratios of powder are measured using the analytical balance used for density
measurements in Table A.1 to +/- .1 mg and added to a Nalgene bottle. Twelve yttria-stabilized
zirconia milling media are added for every 10 g of powder. Then, the solvent is added at 85%
volume to 15% powder volume. In this case, ethyl alcohol is used.
81
Milling:
The bottle is then capped and secured in the vibratory mill (see Table A.1) and milled for 8
hours to thoroughly mix the powders and distribute the various components evenly. Often
completed overnight, the removed bottles typically contain a slurry which has settled and excess
ethyl alcohol on the surface. This excess is poured out as waste before transferring the slurry to a
drying dish. The media is removed from the slurry. A small amount of ethyl alcohol is added, and
the bottle is shaken by hand to remove residual powders that may have adhered to the wall. Foil is
placed over the drying dish and punctured with a pin or needle. The foil protects the slurry from
potential candidates in the drying oven while the pin holes allow for release of vaporized solvent.
The slurry is dried for a minimum of 8 hours. Once dried, it is stored in a sample bottle until ready
for calcination.
Calcining:
Calcination is the first step of the process where phase transitions occur. The milled powders
are added to a crucible, here magnesia, capped, and heated to temperature for 8 hours. Calcination
temperatures vary depending on the phase diagram of the system, but 75% of the sintering
temperature is often employed as a starting value.
Calcined powders are then ground with a mortar and pestle and a small amount is placed in
a sample holder for powder x-ray diffraction (XRD) using the instrument in Table A.1. The EVA
XRD-software is used to analyze the diffraction patterns and identify phases. If after calcining, the
desired final phase is not present, the calcination must be repeated at a higher temperature.
Conversely, if the XRD indicated the calcination has stabilized the desired phase, the powder can
be progressed into the sintering stage.
Sintering:
Powder is compressed using the hydraulic press (Table A.1) and dye kit for either ½-inch or
1-inch pellets at 170 MPa for 1 minute. In general, longer pressing times and higher pressures
increase final densities to some extent. The pellets are placed in crucibles, capped and put into
furnaces for sintering. Sintering times and temperatures vary for materials, but the calcination and
82
sintering parameters used for the ceria and hafnia systems studied are presented in Chapters 2 and
4 respectively. Once the final pellets are produced, one pellet is ground into fine powder again to
measure the XRD pattern and ensure the samples are single phase. Measuring phases with XRD
is typically only necessary for determining the fabrication process before batching samples for
evaluation.
Characterization:
In order to characterize the materials, many different techniques are used to verify structure,
measure sample quality, and evaluate material properties. These techniques will be presented in
chronological order; that is, the order which the characterization will take place for a batch of
samples.
X-Ray Diffraction
The first method of characterization has already been mentioned in the sample fabrication
process. The structure of a material is found using X-ray diffraction (Table A.1) which uses copper
K-a x-ray waves of nominal wavelength 1.5414 Å to reflect off samples and measure distances
between atoms in the crystal structure. Typical scans measured at steps of 0.05˚ with a rate of
0.5˚/s. This is the primary tool for determining phases of materials.
Density
The quality of a sample can further be evaluated by its density. Using the Archimedes
principle, the weight of the sample is taken in both air (𝑚!;D ) and liquid (𝑚@;E ), and the difference
is used to calculate the volume of the sample. Masses are measured on the Mettler-Toledo Balance
in Table A.1. This method is more accurate than measuring the dimensions of the sample and
calculating volume geometrically. For some samples, particularly lower density ones, it is found
that the sample absorbs water which is observed as an increase in mass over time when submerged
in the liquid. This can introduce significant error in the final density calculation. A subsequent
83
,
mass measurement after being submerged (𝑚!;D ) will reveal the mass of the water that has been
absorbed by the sample. Then, the density can be calculated from equation A.5. The measurements
performed in this work were all done with water which has a density of approximately 1 g/cm3.
Samples with less than 85% density are considered low quality and properties measured may
deviate significantly from a dense polycrystalline sample.
𝑚!;D
𝜌'A@ = , 𝜌 (A.5)
𝑚!;D − 𝑚@;E @;E
The theoretical density of a sample is determined through either reported literature values or
calculations of lattice parameters using Bragg’s Law with data extracted from XRD. Equation A.6
is a less general form of the Bragg equation specifically used to calculate lattice parameter for the
cubic systems studied here. From the XRD pattern, one can determine the angle at which
constructive interference occurs by the location of the peak intensity. The distance between atoms
is determined by the Miller indices which are determined by comparisons to experimental data in
a database such as ICDD1.
𝑎#
𝜆 = 2 < = sin(𝜃) (A.6)
√ℎ( + 𝑘 ( + 𝑙 (
𝜆 − Wavelength of x-ray (Cu-𝛼)

𝑎# − Unit cell lattice parameter
[ℎ, 𝑘, 𝑙] − Miller index of peak
𝜃 − Angle location of peak
For the cubic fluorite systems, calculating the unit cell volume is trivial. Then, using the
crystal structure of the system, one can determine the average mass of the unit cell from the
chemical formula of the result. Using Figure 1.3 above, the number of oxygen atoms and cations
contained within the unit cell volume is counted. There are eight interstitial oxygen atoms, and a
total of four cation atoms contained within the boundaries. However, the crystal represented in
Figure 1.3 represents an equilibrium MO2 system. The average mass of the doped unit cell is
84
calculated using the chemical formula of 𝑅- 𝑀/0- 𝑂(0$ . Thus, theoretical density is calculated by
#
equation _
2𝑥𝑚G + 2(1 − 𝑥)𝑚H + (4 − 𝑥)𝑚+
𝜌=F =
𝑎#)
Potassium Exposure Testing

Once dense samples of the desired phase(s) are formed, they can be characterized for MHD
applications. In order to determine phase stability and performance in the presence of MHD
plasmas, samples are exposed directly to potassium vapor at elevated temperatures to mimic the
seed. The details of the experiment are laid out clearly in Chapter 3. Careful mass measurements
of each component are made before and after testing. Changes in mass of components will suggest
that the crucible material has reacted with potassium or the sample and enables a mass balance to
determine what mechanisms have caused the change in mass of the sample. Table A.2 is a template
of the required measurements. Each mass is taken three times consecutively using the Sartorius
Analytical Balance in Table A.1 and the average is used. The balance used in corrosion testing
measurements offers five decimals of precision at the gram scale which is more precise than the
balance used for density. The crucibles, lids, and trays are MgO with up to 2% yttrium substitution
purchased from Tateho-Ozark. Dimensions are described in Chapter 3. The lid is modified by
boring a 5 mm hole through the center with a refractory drill at NETL-Albany. Mass changes are
recorded for both a null experiment and a potassium exposure experiment. Structure is
characterized initially, after the null test, and after the potassium exposure.
The atmosphere gas is flown through the gas monitor system and the tube furnace is flushed.
The end cap at the exit is left open to insert the samples. Three samples are massed and recorded
as in the template in Table A.2. The mass of the tray is also recorded. The samples are placed in
the tray in a specific order which is recorded. The samples are inserted into the exit of the tube
furnace to the center of the furnace, where the thermocouple is positioned, and the tube is capped.
The bubbler system is connected, and the gas flow is decreased using a needle valve until the
bubbles cease. Then, flow is increased carefully until a slow bubble initiates. For reducing
85
atmospheres, more specifically an argon environment, the oxygen and water content are monitored
using the PPM level O2 meter and hygrometer listed in Table A.1. When the flow rate is set, the
PPM meters typically take 30 minutes to reach equilibrium, less than 5 PPM O2 and less than 25
PPM H2O. The furnace may then be heated to temperature at a rate of 5˚C/min up to 1300˚C. The
maximum temperature setpoint of the furnace is 1500˚C, but the system is limited by its age and
peaks at 1200˚C internal temperature. The furnace is programmed to hold for 24 hours at
temperature before cooling at 5˚C/min back to room temperature. Figure A.1 shows the internal
measured temperature profile from a run with a MgO sample representative of a typical furnace
run. After the furnace has cooled to room temperature, the samples are removed and the mass for
each component is measured and recorded in the After section of the template in Table A.2. The
samples are then stored until the potassium exposure test is conducted.
Figure Appendix A.1) Temperature profile of single ramp up, hold, and ramp down from null
test of a MgO sample.
A similar procedure is done for the potassium exposure test, though with more components.
The mass of the lid, crucible, and tray are taken separately and recorded for each sample being
86
tested as in Table A.2. If the corrosion test is performed directly following the null test, the mass
of the sample from the previous measurement can be used. Otherwise, the mass of the sample
should be taken again, and any changes be noted. Then, between 2 and 2.5g of potassium carbonate
is added to the crucible and the mass is taken. Since potassium carbonate readily absorbs water
from the atmosphere, the mass will increase during consecutive measurements. Next, the lid is
placed atop the crucible and the mass of this assembly is taken. Finally, the sample is used to cap
the hole in the crucible lid and the total mass is recorded. This can be done for up to three crucibles
assemblies and samples. The three assemblies are arranged on two trays and inserted into the tube
furnace to the position of the thermocouple. The trays should be centered at the thermocouple
carefully, as the samples do not provide enough friction to prevent sliding and uncapping the hole.
The order of the samples is noted to correlate final masses with initial masses. Again, the furnace
is heated per the temperature profile used in the null test (Fig. 1). After program completion the
mass is recorded again in virtually reverse order. The total mass is first recorded. The sample is
removed and weighed. Some samples may adhere to the crucible lid, in which case a spatula can
be used to pry the sample gently from the lid. Next the assembly mass is taken. The lid is removed,
which may also require gently prying with the fingers or a spatula, and the mass is recorded. The
crucible weight is taken and recorded. Finally, the tray mass is measured.
During exposure tests, some simple observations can be made to confirm if the experiment
completed successfully. Most importantly, the supply gas must not have depleted during testing.
This can be checked from the pressure regulator, or in the data log from the water or oxygen
content meters. Any error indicators on the furnace may suggest the temperature cycle did not
successfully complete. Again, this can be confirmed by the data logger by plotting the temperature
with time. The experiment is designed such that approximately 25% or 0.5 grams of potassium
carbonate volatilizes throughout the duration of the exposure. If all potassium carbonate breaks
down, the assumption that the sample has been exposed for 24 hours is no longer valid and the
mass change results cannot be correlated to other samples with the same conditions. Any failed
test should be repeated to provide three consistent experiments with reproducible results.
87
Null Test
Before 1 2 3 Mean After 1 2 3 Mean
Tray Tray
Sample 1 Sample 1
Sample 2 Sample 2
Sample 3 Sample 3
K-Exposure Test
Sample 1
Tray Tray
Crucible Crucible
Lid Lid
Crucible + Crucible +
K2CO3 K2CO3
Assembly Assembly
Sample Sample
Total Total
Sample 2
Tray Tray
Crucible Crucible
Lid Lid
K2CO3 K2CO3
Assembly Assembly
Sample Sample
Total Total
Sample 3
Tray Tray
Crucible Crucible
Lid Lid
K2CO3 K2CO3
Assembly Assembly
Sample Sample
Total Total
Table Appendix A.2) Template table of masses for observing mass change references in the null
and potassium exposure portions of a corrosion evaluation experiment.
88
Electrical Properties
Electrical properties of materials are evaluated using two techniques which will be
presented separately. However, the fundamental material property measured is the same using
either technique and the values provided from each agree as they should.
1 𝐿
𝜎= = (A.7)
𝜌 𝑅𝐴
DC Bar
Firstly, preparing samples for DC Bar measurements requires cutting, applying electrode
contacts, and performing measurements. Ideal samples are pressed and sintered into 1-inch
diameter pellets approximately 5mm thick. The pellet is cut using a diamond saw (Table A.1.) into
bars of 5mm by 5mm by 20mm. Grooves are cut into the corners approximately 5mm from each
end to a shallow depth for inserts to secure contact points for the platinum wire probes using the
diamond saw as well.
Once the sample is cut into the appropriate geometry, platinum electrodes are applied to
the contact surfaces. The smallest area faces are coated with a thin layer of platinum ink (Table
A.1). A thin ring of platinum ink is also applied to the outer surface at the heights of the grooves,
using a paper clip to ensure the inner grooves are coated as well. All in all, 4 contact points are
provided as shown in Figure A.2B. The platinum ink is then heated in a furnace to 325˚C at
1˚C/min then to 1000˚C at 10˚C/min. Curing too fast during the initial heating step results in
delaminating of the platinum, but after initial curing, the platinum diffusion into the sample is
stable beyond the peak curing temperature.
The sample is inserted into the NORECS Probostat fitted with the appropriate bar sample
parts described in the manual. A platinum wire is wrapped around the sample and secured into the
corners at each end where ring electrodes are applied. The sample is then loaded into the Probostat
with a mesh electrode on the bottom and top surfaces. The rod spacer for spring loading is placed
atop and the springs are fixed to the base. The wires are connected to the leads which exit the
Probostat for voltage measurements. Typically, the high voltage and high current are applied to
the top surface and ring electrodes and the low voltage and low current are applied to the bottom
89
surface and ring electrodes as in Figure A.2B. The Probostat is finally sealed with the alumina
protection tube, inserted into the furnace, and heated for conductivity measurements at each
temperature.
The furnace is heated at 50˚C increments from room temperature to maximum furnace
temperature and held for 40 minutes before taking electrical conductivity measurements to ensure
temperature stability. Typically, samples do not conduct at lower temperatures and initial
measurements begin around 400˚C. The resistance of the sample is measured using Keithley
current source and nanovoltmeter units (Table A.1.) controlled by a custom Labview program.
Current is swept at minimum of 21 points between symmetric minimum and maximum current
values. The values are adjusted depending on the order of magnitude of the resistance of the
sample, gauged by the linearity and smoothness of the I-V curve produced by the Labview
program. Voltage is measured using the automated trigger function. The resistance is then
calculated from the slope of a linear fit of voltage versus current data. The resistivity is finally
calculated using Equation A.7 where 𝐴 is the area of the face where the electrode is applied and 𝐿
is the length between the two ring electrodes where voltage is measured.
Ring electrodes
A B
10mm
5mm
2mm
5m
m
5mm 5mm
20mm
Figure Appendix A.2) Dimensions and electrode placement of samples for electrical conductivity
measurements. A) Disk sample with electrodes for AC EIS. B) Bar sample for DC resistance
measurements.
AC Disk
The other method of measuring the electrical conductivity of samples is known as AC
Electrochemical Impedance Spectroscopy (EIS). Platinum ink is applied to the flat surfaces of ½-
90
inch diameter samples and cured in a furnace by heating to 325˚C at 1˚C/min then to 1000˚C at
10˚C/min. Again, initial curing temperature is limited to prevent platinum delamination, but the
cured electrodes can withstand significantly higher temperatures. Samples are then inserted into
the Probostat fitted with the disk sample parts described in the manual. Platinum mesh electrodes
are fixed on the top and bottom of the sample and the spring load system is fixed to add pressure
to the electrodes.
Impedance spectroscopy is measured using the Solartron 1260A with the Solartrion 1296A
dielectric interface (Table A.1). Instruments are controlled using the SMaRT software sweeping
frequency from 106 Hz to 10-1 Hz at 10 steps per decade and real and imaginary impedance are
recorded at each. Equivalent circuit modeling is used to analyze the real and imaginary impedance
values and extract resistance and capacitance values. Figures 2.4 and 4.4 are the circuit model used
where typically only two resistance-capacitance circuit elements are visible within the impedance
spectrum. A python script is used to fit the data using a linear least squares algorithm to fit
modeling parameters. The resistivity is calculated for the bulk and grain boundary resistances
using equation A.7 where 𝐴 is the area of the circular face and 𝐿 is the thickness of the sample.
The total resistivity is found through a combination of the bulk and grain boundary resistivities
using equation A.8 based on the brick layer model2.
𝜎=A= = 𝜎?I@% + 𝜎J.". (A.8)
Bibliography:
(1) Gates-Rector, S. D.; Blanton, T. N. The Powder Diffraction File: A Quality Materials
Characterization Database. International Centre for Diffraction Data 2019, 34 (4), 352–
360.
(2) Irvine, J. T. S.; Sinclair, D. C.; West, A. R. Electroceramics: Characterization by
Impedance Spectroscopy. Adv. Mater. 1990, 2 (3), 132–138.
https://doi.org/10.1002/adma.19900020304.

Bowen Michaels 2021

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Bowen Michaels 2021

Uploaded by

Copyright:

Available Formats

AN ABSTRACT OF THE THESIS OF

Michael S. Bowen for the degree of Master of Science in Mechanical Engineering

presented on June 10, 2021

Title: Evaluation of Materials for Use in Open-Cycle Magnetohydrodynamic Power Generation

Oregon State University

Presented June 10, 2021

Figure 2.2) Comparison of accumulated data from various sources27–37 of Gd0.1Ce0.9O1.95. _ 29

Figure 2.3) Comparison of accumulated data from various sources18,19,27,29,31,33,36,38–42 of

Magnetohydrodynamics (MHD) is a branch of physics which studies the effects of a

In a turbogenerator, a turbine is designed to be spun as the hot, fast-moving gases flow

1.3.2. Electrode Materials

(31) Gatehouse, B. M.; Lloyd, D. J. The Crystal Structure of Beta-Potassium Dizirconate: β-

Electrical Properties of Gadolinia-Doped Ceria for Electrodes for Magnetohydrodynamic

2. Electrical Properties of Gadolinia-Doped Ceria for Electrodes for Magnetohydrodynamic

Research and development of a direct power extraction concept using

Figure 2.3) Comparison of accumulated data from various sources18,19,27,29,31,33,36,38–42 of

𝑍 = 𝑅;<'= + (𝑅𝑄)> + (𝑅𝑄)>.?. + (𝑅𝑄)2@. (2.5)

where resistivity is defined by:

2.4. Results & Discussion

Figure 2.5) A) above: Average density and

At temperature increments of 100˚C, AC EIS measurements were performed on two disk

AC Disk AC Bar DC Bar

(16) Zavodinsky, V. G. The Mechanism of Ionic Conductivity in Stabilized Cubic Zirconia.

2.7. Supplementary Information

X [mol%] 0% 10% 20% 30% 50%

High temperature corrosion stability of ceramic materials for magnetohydrodynamic

ASTM International Journal on Material Performance and Characterization

100 Barr Harbor Drive

3. High temperature corrosion stability of ceramic materials for magnetohydrodynamic

Magnetohydrodynamic (MHD) power generators achieve direct power extraction through

Discs of 10 mm diameter by 3 mm height were acquired from American Elements for

3.4. Results and Discussion

Unlike magnesia, alumina is known to react with potassium in various refractory

3.4.3. Yttria-stabilized zirconia

(1) R. J. Rosa, Magnetohydrodynamic Energy Conversion, (Washington: Hemisphere

(15) G. Rudins, S. Schneider, L. Bates, H. K. Bowen, L. Crawford, H. Frederikse, B. Rossing,

4. The Yttria-Hafnia System

4.2. Electrical Properties

1 (Inst) 2 (Grain) 3 (G.B.) 4 (El)

4.3. Corrosion Results

5.1. Electrical properties

A summary of electrical properties is plotted in Figure 5.1. The limits of electrical

5.2. Corrosion Resistance

5.3. Final Remarks

Appendix A: Full Documentation of Experimental Procedures

Equipment Model Number

𝜆 − Wavelength of x-ray (Cu-𝛼)

Potassium Exposure Testing

You might also like