livrepository.liverpool.ac.uklivrepository.liverpool.ac.uk/19353/4/simofra_aug2014_19353.pdf · 2...
Post on 12-Oct-2020
0 Views
Preview:
TRANSCRIPT
Novel Oxide Materials for Solid Oxide
Fuel Cells Applications Thesis submitted in accordance with the requirements of the
University of Liverpool for the degree of Doctor in
Philosophy by:
František Šimo
Supervised by
Professor M. J. Rosseinsky
Dr. J. B. Claridge
2
Abstract
The work of this thesis focuses on three perovskite-based compounds:
YSr2Cu3−xCoxO7+δ cuprates, Gd2BaCo2O5+δ related phases and Sr2SnO4 Ruddlesden-Popper
structures. Both YSr2Cu3−xCoxO7+δ and Gd2BaCo2O5+δ are cathode material candidates for
solid oxide fuel cells (SOFCs). Doping of Sr2SnO4 aims to enhance the ionic conductivity of
the parent phase and explore the phases as a potential SOFCs electrolyte material.
The cobalt content in the layered perovskite YSr2Cu3−xCoxO7+δ has been increased to a
maximum of x = 1.3. A slight excess of strontium was required for phase purity in these
phases, yielding the composition Y1−ySr2+yCu3−xCoxO7+δ (where y = 0.03 and 0.05). The
potential of Y1−ySr2+yCu3−xCoxO7+δ (where x = 1 to 1.3) as a cathode material for a solid
oxide fuel cell has been explored through optimisation of processing parameters, AC
impedance spectroscopy and DC conductivity measurements. The stability of
Y0.95Sr2.05Cu1.7Co1.3O7+δ with commercial electrolytes has been tested along with the stability
under CO2. This material exhibits a significant improvement in properties compared to the
parent member, Y0.97Sr2.03Cu2CoO7+δ, and is compatible with commercially available doped
ceria electrolytes at 900 °C.
Energetics of Ln2BaCo2O7 (Ln = Gd, Nd, Ce) materials consisting of a layer of
LnBaCo2O5+δ (Ln = Gd, Nd) and a fluorite layer (CeO2 or Ln2O3, Ln = Gd, Nd) have been
studied using DFT calculations. Various reactions including binary oxides and double
perovskites were taken into an account for the formation energy calculations. Phases
favourable in DFT calculations were observed also in PXRD patterns of the materials
prepared by a solid state synthesis.
DFT prediction has been also used in the work with Ruddlesden-Popper phases. The
structures of experimentally prepared Nb- and Ta-doped Sr2SnO4 phases were investigated
using high resolution diffraction methods. The conductivity of single phased materials was
studied by AC impedance spectroscopy. A significant improvement in conductivity was
observed in Sr2Sn1−xTaxO4 compounds with x = 0.03 and 0.04. The origin of the enhancement
has been studied using different techniques such as solid state Sn-NMR, UV-vis and NIR
spectroscopy methods and it tends to be explained by an ionic contribution.
3
4
Acknowledgements
I would like to offer my special thanks to my primary supervisor Prof. Matthew
Rosseinsky for the given opportunity to learn new methods and techniques and for the
guidance throughout my time in the group. I would also like to thank my secondary
supervisor Dr. John Claridge for his valuable advices.
I am grateful for the assistance of past and present members of the MJR research
group I have been closely working with. In particular, Dr. Antoine Demont and Dr. Ruth
Sayers for their help through the first synthetic attempts. I would like to thank Dr. George
Darling and Dr. Matthew Dyer for the patience and support with density functional theory
applications. Thanks also go to Dr. Phil Chater, Dr. Mike Pitcher, Dr. Julia Payne and Dr
Alex Corkett for their great help in structural refinements and valuable suggestions. Advices
given by Dr. Hripsime Gasparyan and Dr. Ming Li have been a great help in electrochemical
data analysis. I wish to acknowledge the help with UV-vis spectroscopy measurements
provided by Dr. Troy Manning and Borbala Kiss. I would like to thank Dr. Frédéric Blanc for
the help with Sn solid state NMR measurement and Natasha Flack for the SEM images. In
addition, thanks go to Dr. Hongjun Niu and Mike Chatterton for their technical support.
I would like to extend my thanks to Prof. Ken Durose, Dr. Laurie Phillips and Dr.
Robert Treharne for their kind assistance with NIR spectroscopy. X-ray and neutron
diffractometers support has been provided by instrument scientists from ISIS and Diamond,
namely Prof. C. Tang, Dr. S. Thomson, Dr. J. Parker (I11), Dr. Aziz Daoud-Aladine (HRPD)
and Dr. Winfried Kockelman (GEM).
Finally, I would like to thank my family, Eftychia and friends for their constant
support and good company outside of the workplace.
Contents List of Abbreviations...............................................................................................................................9
1 Introduction ................................................................................................................................... 11
1.1 Current energy demand ......................................................................................................... 11
1.2 SOFCs ................................................................................................................................... 12
1.2.1 Principle of operation .................................................................................................... 12
1.2.2 Cell efficiency ............................................................................................................... 13
1.2.3 SOFC component requirements .................................................................................... 15
1.3 Mass transport ....................................................................................................................... 16
1.3.1 Vacancy diffusion mechanism ...................................................................................... 17
1.3.2 Interstitial diffusion mechanism .................................................................................... 17
1.4 Charge transport .................................................................................................................... 18
1.5 Ionic and electronic conductivity .......................................................................................... 19
1.5.1 Ionic conductivity ......................................................................................................... 19
1.5.2 Electronic conductivity ................................................................................................. 19
1.6 Structures description ............................................................................................................ 21
1.6.1 Perovskite ...................................................................................................................... 21
1.6.2 Ruddlesden-Popper phases ........................................................................................... 22
1.6.3 Fluorite .......................................................................................................................... 23
1.7 Materials review .................................................................................................................... 24
1.7.1 Cathodes materials ........................................................................................................ 24
1.7.2 Electrolyte materials ..................................................................................................... 26
1.7.3 Anode materials ............................................................................................................ 29
1.8 Aims of the work .................................................................................................................. 31
2 Experimental and theoretical methods .......................................................................................... 33
2.1 Material synthesis ................................................................................................................. 33
2.2 Powder diffraction techniques .............................................................................................. 34
2.2.1 Fundamentals of diffraction .......................................................................................... 34
2.2.2 Diffraction of X-rays ..................................................................................................... 35
2.2.3 Diffraction of neutrons .................................................................................................. 38
2.2.4 Powder diffraction......................................................................................................... 39
2.2.5 The Rietveld Method .................................................................................................... 40
2.2.6 Laboratory X-ray diffraction ......................................................................................... 43
2.2.7 Synchrotron X-ray powder diffraction .......................................................................... 43
2.2.8 Neutron Sources and Time-of-Flight Diffraction ......................................................... 44
6
2.3 Scanning Electron Microscopy ............................................................................................. 47
2.4 Electrical conductivity measurements .................................................................................. 47
2.4.1 Fundamentals ................................................................................................................ 47
2.4.2 The four-probe DC method ........................................................................................... 48
2.4.3 Cold Isostatic Pressing .................................................................................................. 49
2.4.4 Density measurements .................................................................................................. 50
2.5 AC Electrochemical Impedance Spectroscopy (EIS) ........................................................... 50
2.5.1 Fundamentals ................................................................................................................ 50
2.5.2 Data analysis ................................................................................................................. 52
2.6 Ultraviolet-visible and Infrared Spectroscopy ...................................................................... 53
2.6.1 Ultraviolet and visible Spectroscopy ............................................................................ 53
2.6.2 Infrared Spectroscopy ................................................................................................... 54
2.7 Solid state NMR technique ................................................................................................... 55
2.8 Iodometric Titrations ............................................................................................................ 56
2.9 Thermogravimetric Analysis................................................................................................. 57
2.10 Dilatometry ........................................................................................................................... 58
2.11 Density Functional Theory (DFT) ........................................................................................ 59
2.11.1 The energy functional ................................................................................................... 60
2.11.2 Kohn-Sham equations ................................................................................................... 61
2.11.3 Exchange Correlation Functionals ............................................................................... 63
2.11.4 Pseudo potentials........................................................................................................... 65
2.11.5 DFT+U .......................................................................................................................... 66
3 Synthesis and characterization of Y1−ySr2+yCu3−xCoxO7+δ .............................................................. 68
3.1 Introduction ........................................................................................................................... 68
3.2 Synthesis ............................................................................................................................... 70
3.3 Structural characterization .................................................................................................... 71
3.3.1 Laboratory P-XRD data ................................................................................................ 71
3.3.2 Neutron Powder Diffraction data .................................................................................. 75
3.4 DC conductivity measurements ............................................................................................ 79
3.5 Thermal stability ................................................................................................................... 82
3.6 Chemical compatibility of Y0.95Sr2.05Cu1.7Co1.3O7+δ with electrolytes................................... 86
3.7 AC impedance spectroscopy of Y0.95Sr2.05Cu1.7Co1.3O7+δ ...................................................... 89
3.7.1 AC impedance data at 500 ‒ 800 °C ............................................................................. 89
3.7.2 AC impedance data for the dwelling at 650 °C............................................................. 94
3.7.3 SEM study of symmetrical cells ................................................................................... 96
3.8 AC impedance spectroscopy of Y0.97Sr2.03Cu2CoO7+δ ........................................................... 97
7
3.9 CO2 stability tests of Y0.95Sr2.05Cu1.7Co1.3O7+δ ....................................................................... 99
3.10 Thermal expansion studies of Y0.95Sr2.05Cu1.7Co1.3O7+δ ....................................................... 102
3.11 Discussion and conclusions ................................................................................................ 104
4 Prediction and synthesis of GBCO related phases ...................................................................... 107
4.1 Introduction ......................................................................................................................... 107
4.2 Computational methods ..................................................................................................... 109
4.3 Experimental methods......................................................................................................... 113
4.4 Computational results ......................................................................................................... 113
4.4.1 CeO2 ............................................................................................................................ 113
4.4.2 LnBaCo2O5 .................................................................................................................. 115
4.4.3 LnBaCo2O5.5 ................................................................................................................ 116
4.4.4 Double perovskite with fluorite layer ......................................................................... 116
4.4.5 Formation energies ...................................................................................................... 118
4.5 Experimental results ............................................................................................................ 122
4.6 Formation energies - different lanthanides ......................................................................... 126
4.7 Other fluorite layer .............................................................................................................. 128
4.7.1 Formation energies ...................................................................................................... 128
4.7.2 Experimental results .................................................................................................... 130
4.8 Discussion and conclusions ................................................................................................ 133
5 Ruddlesden-Popper phases ‒ stannates ....................................................................................... 136
5.1 Introduction ......................................................................................................................... 136
5.2 Computational methods ...................................................................................................... 138
5.3 Experimental methods......................................................................................................... 139
5.4 Computational results ......................................................................................................... 140
5.5 Structural characterization .................................................................................................. 142
5.5.1 Laboratory P-XRD ...................................................................................................... 142
5.5.2 Synchrotron data ......................................................................................................... 146
5.5.3 High Resolution Powder Diffraction data ................................................................... 151
5.6 AC Electrochemical Impedance Spectroscopy (EIS) ......................................................... 154
5.6.1 AC impedance data at 600 ‒ 900°C ............................................................................ 154
5.6.2 AC impedance data at 300 ‒ 600°C ............................................................................ 158
5.6.3 AC impedance data at different partial oxygen pressure ............................................ 160
5.7 Thermal stability ................................................................................................................. 164
5.8 UV-vis spectroscopy measurements ................................................................................... 166
5.8.1 As made materials ....................................................................................................... 166
5.8.2 Reduced materials ....................................................................................................... 167
8
5.9 Sn Solid-state NMR ............................................................................................................ 168
5.10 IR spectra ............................................................................................................................ 171
5.11 Discussion and conclusions ................................................................................................ 173
6 General Conclusions and Perspectives ....................................................................................... 177
References...........................................................................................................................................178
APPENDIX A: EDX data of Sr2Sn0.96Ta0.04O4....................................................................................187
APPENDIX B: Lattice parameters and cell volume of Sr2Sn1−xNbxO4...............................................188
APPENDIX C: Lattice parameters and cell volume of Sr2Sn1−xTaxO4................................................189
APPENDIX D: Joint I11 and HRPD Rietveld refinement of Sr2Sn0.97Nb0.03O4..................................190
APPENDIX E: Joint I11 and HRPD Rietveld refinement of Sr2Sn0.96Ta0.04O4...................................191
List of Abbreviations
AFM Anti-ferromagnetic
ASR Area specific resistance
BSCF Ba1−xSrxCo1−yFeyO3−δ
CIP Cold isostatic pressing
CPE Constant phase element
DFT Density functional theory
Ea Activation energy
EDX Energy dispersive X-ray
EIS Electrochemical impedance
spectroscopy
eV/FU Electron volts per formula unit
FM Ferromagnetic
GBCO GdBaCo2O5+δ
GDC Gadolinia-doped ceria
GGA Generalised-gradient
approximation
GOF Goodness-of-fit
HK Hohenberg-Kohn
HRPD High resolution neutron
powder diffraction
ICSD Inorganic crystal structure
database
IR Infrared
IT-SOFC Intermediate solid oxide fuel
cell
LAMOX La2Mo2O9
LDA Local density approximation
LNO La2NiO4+δ
LSC LaCoO3
LSCF La1−xSrxCo1−yFeyO3−δ
LSF Lanthanum strontium ferrite
LSGM La1−xSrxGa1−yMgyO3−δ
LSM Lanthanum strontium
manganite
MAC Multianalyzing crystal
MCFC Molten carbonate fuel cell
NIR Near infrared
NPD Neutron powder diffraction
PAFC Phosphoric acid fuel cell
PAW Projector augmented wave
method
PBE Perdew-Burke-Ernzerhof
PEFC Polymer electrolyte fuel cell
PM Pechini sol-gel method
ppm Parts per million
PVA Polyvinyl alcohol
PXRD Powder X-ray diffraction
RE Rare earth
RP Ruddlesden-Popper
RT Room temperature
ScSZ Scandia-stabilised zirconia
SDC Samarium-doped ceria
SEM Scanning electron microscopy
SIE Self interaction error
10
SOFC Solid oxide fuel cell
ss NMR Solid state NMR
TCO Transparent conductors
TEC Thermal expansion coefficient
TGA Thermogravimetric analysis
TOF Time of flight
UV-vis Ultraviolet-visible
VASP Vienna ab initio simulation
package
XC Exchange correlation
YDC Yttria-doped ceria
YSZ Yttria-stabilised zirconia
3ap Triple perovskite
1 Introduction
1.1 Current energy demand
Traditional power generation based on fossil fuels, oil and coal presents a huge threat
for the environment with the worldwide consequences such as the climate change.1,2
Growing
population and thus increasing demand enhances the need to find environmentally friendly
energy sources. This has led to more intense search for alternative energy sources along with
novel technologies in order to reduce carbon dioxide emissions and fossil fuel dependence.
Numerous reviews bring a discussion on energy and technology alternatives.3-5
The target of this thesis includes fuel cell technologies, specifically solid oxide fuel
cells (SOFCs). Fuel cells are electrochemical power generation devices providing higher
efficiencies than conventional power production.6 Their potential application ranges from
providing power for portable devices (mobile phones, laptops) and transport applications to
stationary power applications. Recent development in fuel cell technology has brought
different types of fuel cells, characterized by an electrolyte, such as the polymer electrolyte
fuel cell (PEFC), the phosphoric acid fuel cell (PAFC), the molten carbonate fuel cell
(MCFC) and the SOFC. Main advantages of fuel cells compared to conventional power
generation devices are the higher conversion efficiency and environmental benefit of
producing less CO2 altogether with the lower fuel requirement. The principle of fuel cells was
revealed in the middle of 19th
century, when the first fuel cell was also built.7,8
The first fuel
cell operated at room temperature, using a dilute sulphuric acid electrolyte, a hydrogen anode
and an oxygen cathode.
The main concepts and principles of SOFCs are discussed in Section 1.1 considers the
transport and conductivity mechanisms (Sections 1.3 to 1.5). Sections 1.6 and 1.7 bring an
overview of the structure and material types of each of the individual components of SOFCs.
The objectives of the work presented in this thesis are explained in Section 1.8.
Chapter 1. Introduction
1.2 SOFCs
1.2.1 Principle of operation
SOFCs are complex electrochemical conversion devices (converting fuel directly into
electrical current) composed of three main components: an electrolyte, a cathode and an
anode. Their operating temperature (500‒1000 °C), relatively high compared to other fuel
cells, is required to ensure adequate ionic conductivity in the electrolyte. Moreover, operating
temperatures above 500 °C show additional benefits, such as the possibility of avoiding
expensive platinum metal-based catalysts. High temperature also reduces the need for fuel
purity, i.e. natural gas can be used as the fuel without preliminary reforming. SOFCs can be
divided into two groups according to operating temperatures; high (800‒1000 °C) and
intermediate (IT-SOFC, 500‒700 °C). Lowering the operation temperature of SOFCs towards
IT-SOFC has become one of the main SOFC research goals,9 driven by significant
restrictions for materials at high temperatures such as chemical stability and compatibility.
A schematic picture of an SOFC is shown in Figure 1.1, displaying main three
compartments of the device. An SOFC can also contains interconnect components providing
electrical connection between the cathode of one individual cell and the anode of the
neighbouring cell.10
This component should prevent mixing of cathodic and anodic gases and
thus prevent their mutual reaction.
Figure 1.1: Schematic layout of an SOFC, showing the flow of electrons in order to provide electrical
energy to an external circuit. Reactions taking place on the cathode and the anode are shown on the
right side.
Chapter 1. Introduction
13
Oxygen reduction occurs on the cathode (Equation 1.1) while hydrogen is oxidised on
the anode (Equation 1.2). Transport of O2−
anions is allowed by the electrolyte exhibiting
high oxygen ion conductivity and low electronic conductivity. A wide range of fuels from
hydrogen to hydrocarbons, is available for SOFCs. The chemical energy of the fuel is
converted into electrical power with water and CO2 as a waste product (Equation 1.3).
Cathode reaction: O2 (g) + 4 e− → 2 O
2− (1.1)
Anode reaction: 2 H2 (g) + 2 O2−
→ 2 H2O (g) + 4 e− (1.2)
Overall reaction: O2 (g) + 2 H2 (g) → 2 H2O (g) (1.3)
1.2.2 Cell efficiency
The main advantage of fuel cells compared to traditional power sources is their higher
efficiency which can normally reach up to 65%.11
The main requirements for SOFC include
power loss not exceeding 0.1% during continuous operation for 1000 h and lifetime of at least
5 years. However, SOFCs materials often react with materials from other parts of the device
(cathode ‒ electrolyte interface) causing a rapid degradation. Another drawback is presented
by poisoning of the SOFC surface by chromium compounds. This is due to the doped
lanthanum chromite (La1−xMxCrO3, M = Ca, Sr) often used as the interconnect materials.12
The chromites can be substituted by temperature-resistant steels which inquires lowering the
SOFC temperature to 500 ‒ 700 °C.9
An increased SOFC efficiency can be generally achieved by lowering of the cell
polarisation losses (η, known also as overpotential). The cell overpotential is the difference
between the theoretical and actual cell voltage present in the cell. The difference is due to the
presence of polarisation losses. These losses are expressed by Equation 1.4 and are induced
by the following four types of polarisation: charge transfer (activation) polarisation (ηa,
associated with the electrodes), diffusion (concentration) polarisation (ηm, associated with
mass transport), reaction polarisation (ηr, similar to concentration polarisation) and resistance
or ohmic polarisation (ηΩ, associated with ionic and electronic conduction and contacts
between the cell compartments).
V = E0 − ηa − ηm − ηr − jR (1.4)
Chapter 1. Introduction
14
where V is the actual voltage output, E0 is the theoretical cell voltage and jR is equal to the
ohmic polarisation losses while R is the total cell resistance.12-15
The relationship between the
polarisation losses is displayed in the voltammogram of a working SOFC (Figure 1.2).12
Figure 1.2: Current ‒ voltage curve of a working SOFC with the polarization losses, taken from.12
Ohmic losses are related to Ohm's law and thus these losses are reduced by an
increase of temperature. Concentration losses contribute significantly at high currents when
the rates of electrochemical processes are considerable. The largest contribution of losses is
given by the activation polarisation, due to the complexity of electrochemical processes
taking place on electrodes commonly involving several steps. Lowering the operation
temperature makes a huge impact on activation polarisation. Decreasing temperature slows
the electrochemical reaction on electrodes.12
The resistance of each of the individual SOFC components can be quantified by area
specific resistance (ASR, in Ω cm2), also described in this thesis (Chapter 3). Material
properties for each part of SOFC including total performance need to be estimated.13,16
Power
density of a single cell is required to be of 490 mW cm−2
(in order to deliver 0.7 A at 0.68 V).
Chapter 1. Introduction
15
The total polarisation, made up of ohmic, anodic and cathodic overpotentials, is then ≈ 0.32
V. Neglecting contributions of other polarisation effects gives a single cell resistance of
≈ 0.45 Ω cm2. As a consequence of this power density requirement, a limitation of resistance
of ≈ 0.15 Ω cm2 for each of the SOFC component needs to be applied.
17
1.2.3 SOFC component requirements
The requirements for SOFC components are discussed below. There are some general
restrictions for all of the parts of SOFC such as: material (chemical) compatibility with other
device components, similar thermal expansion behaviour compared to other elements of
SOFC, cost efficiency for material synthesis, environment impact, low operation temperature
(for IT-SOFC).
Electrolytes are required to have high ionic conductivity (oxide ion or proton), high
ionic transport numbers, negligible (or no) electronic conductivity (any contribution causes
ohmic losses in a cell). The materials must be electrochemically and mechanically stable over
a wide range of temperature and oxygen partial pressure and they have to show good
sinterability for their fabrication as gas-tight membranes with very small thickness.6,17,18
Porous cathodes are critical components in SOFCs. They are commonly mixed ionic-
electronic conductors with high electronic conductivity under oxidizing conditions, stability
in the cathodic gas atmosphere and a high electrocatalytic activity towards oxygen
reduction.6,12,17
Materials suitable for anodes are also required to have high electronic conductivity.
They need to be stable in severely reducing environments with electrocatalytic activity
towards hydrogen oxidation.6,17
In addition, the anode must have a porous structure since a
large area for gaseous fuel is required.
Chapter 1. Introduction
16
1.3 Mass transport
SOFC electrolytes exhibit high ionic conductivity which is related to mass transport
processes controlled by diffusion. Electrode materials are mixed ionic-electronic conductors
displaying additional charge transport processes compared to electrolytes.
The SOFC's operation requires a transport of oxygen ions through materials via
diffusion. Diffusion is proportional to the negative of the concentration gradient (dC/dx;
Equation 1.5), i.e. the movement of particles (atoms or molecules) from an area with a high
concentration to an area with a low concentration. A mathematical approach is stated in
Fick's first and second law. Fick's first law in one direction (x) is given by:
dCJ D
dx
(1.5)
where J is the particle flux (equals to number of particles per unit area per unit time) and D is
the diffusion coefficient. When steady conditions cannot be reached, the diffusion is
described by Fick's second law (Equation 1.6).
2
2
C J CD
t x x
(1.6)
where ∂C/∂t is the change in concentration with time.
Diffusion in oxide materials is a complex phenomenon influenced by anion and cation
sublattices of a crystal structure. In most of the studied systems, oxygen ions diffusion is
significantly faster than the diffusion of cations.19
Diffusion in crystalline materials can be
imagined as a migration of atoms away from their equilibrium positions. The way that an
atom moves is described by a diffusion mechanism. Point defects play an important role in
enabling the move.20
Chapter 1. Introduction
17
1.3.1 Vacancy diffusion mechanism
In the vacancy mechanism of diffusion, an atom jumps to a neighbouring vacancy, i.e.
by a hopping mechanism (Figure 1.3). A series of adjacent exchanges with vacancies
provides a mechanism for atoms to move through the crystal. The presence of lattice
vacancies is required and their concentration has an impact on the kinetics.21
The vacancy
mechanism is common for oxygen self-diffusion in fluorite- and perovskite-related systems
and is of great interest in SOFC electrolytes. For instance, the consequence of adding CaCl2
into a NaCl crystal structure, means every Ca2+
ion replaces two Na+ ions and creates
vacancies within the structure. Oxide ion diffusion of a number of widely used cathode
materials for SOFCs with cubic perovskite structure (e.g. La1−xSrxMO3−δ, M = Mn, Fe, Co) is
based on the vacancy migration mechanism.21
Figure 1.3: Schematic of vacancy diffusion mechanism with a vacancy indicated by a square.
1.3.2 Interstitial diffusion mechanism
There are two different interstitial mechanisms: direct or indirect, also known as the
interstitialcy mechanism. The direct mechanism (Figure 1.4a) occurs by ions jumping from
one interstitial site to a neighbouring vacant interstitial site. There is no permanent
displacement of the other ions present after a single jump is completed. This diffusion type is
characteristic for small atoms, such as hydrogen and carbon.
The indirect (interstitialcy) mechanism, shown in Figure 1.4b, is described by two
steps. Firstly, an atom of an occupied lattice site (light grey) moves to an unoccupied
Chapter 1. Introduction
18
interstitial site becoming the new interstitial atom (dark grey) and secondly, an original
interstitial atom moves to an un-occupied lattice site.
An example of a SOFC material where oxygen ion transport is driven by an interstitial
diffusion mechanism is reported for lanthanum nickelate (Section 1.7.1).22
Figure 1.4: Schematic of interstitial diffusion mechanism; a) direct interstitial mechanism and b)
indirect (interstitialcy) mechanism. Dark grey colour represents interstitial sites while light grey lattice
site.
1.4 Charge transport
The measure of charge transport is expressed by the electrical conductivity, which is a
material's ability to transport charged particles under an applied electric field.23
The electrical
conductivity is defined as the charge flux (Ji) per unit of electric field (E) for a particle i with
a charge (Zi e) given by:
i i i i i
i i i i
J Z e C v Z eC Z e
E E (1.7)
where Ci is the particle concentration, vi indicates the particle velocity and μi refers to the
particle electrical mobility per unit field of charge. The conductivity of a material, as it is
shown in Equation 1.7, depends on the particle concentration and mobility. The mobility of
smaller electronic carriers such as electrons and holes is typically 2-3 orders of magnitudes
higher than the mobility of ionic carriers (vacancies). The total electrical conductivity of a
Chapter 1. Introduction
19
material is composed of contributions of each of the individual charged particles. Electrical
conductivity in ceramics can be either electronic, ionic or mixed ionic-electronic.
1.5 Ionic and electronic conductivity
1.5.1 Ionic conductivity
The high diffusion rate of oxygen ions species is required for the SOFC electrolytes.
The mechanism of ionic conductivity can be explained via vacancy or interstitial diffusion
types (Sections 1.3.1 and 1.3.2). Thermally dependent mobility is always a part of ionic
conductivity while the temperature dependence of carrier concentration may vary, depending
on the defect type.23
An increase of the ionic conductivity of a SOFC material is typically
achieved by increasing the defect concentration usually done by introducing electron rich or
poor elements (e.g. Sm doping in ceria).24
Another task lays on finding a minimum of the
activation energy for conductivity isotherms for a relevant doped structure.25
1.5.2 Electronic conductivity26
Electrical properties of solids, electronic conductivity including, were well described
in the middle of the twentieth century when Band Theory was completed and generally
approved. In this theory, conductive electrons are not linked to any atom; they are delocalised
enabling them to move easily through the crystal. The fundamental role is played by the outer
electrons being responsible for both chemical and electronic properties. The outer electrons
occupy bands of allowed energies while the regions between these bands cannot be occupied
and are called band gaps. According to contribution, the bands may be mainly of s, p, d or f
character. Overall, the band structure of a solid is more complex with possible overlaps in a
specific crystallographic direction whilst separated in others.26
In general, materials can be
classified into metals, insulators and semiconductors (Figure 1.5). An insulator normally does
not show electrical conductivity. Previous classifications between metals and semiconductors
were based on the magnitude of the measured electrical conductivity. The difference between
Chapter 1. Introduction
20
the electronic/band structures of materials are shown in Figure 1.5 and should be used to
define the structures. Semiconductors are of great interest in SOFC material research.
Figure 1.5: Band structure of: a) an insulator; b) a semiconductor and c) a metal. The valence band is
indicated by grey areas while the conduction band is represented by white areas.
Semiconductors have similar band picture to insulators except the narrower band gap
(Eg). The separation between the empty (conduction) and filled (valence) bands is small
enough that some electrons have sufficient energy to be transferred from the valence band to
the conduction band at room temperature. As the temperature increases, more electrons will
gain the energy to cross the band gap, which leads to enhancement of the conductivity. Each
time an electron is removed from the valence band to the conduction band, two mobile charge
carriers are created: an electron and a hole.26
The value of band gap can be estimated by the
conductivity measurement of a semiconductor at various temperatures in order to obtain the
thermal band gap. Another, more conventional determination, is based on using the energy of
photons (usually from UV-vis region) to excite an electron across the band gap to estimate
the optical band gap (such as UV-vis spectroscopy, Section 2.6).
The presence of point defects may increase the electronic conductivity. A point defect
on band diagram is represented as an energy level localised in the band gap. The effects of
point defects highly depend on which part of band gap they lie in. Those sitting close to the
edges of Eg (shallow levels) play an important role in electronic conductivity while those
lying in the middle of Eg (deep levels) influence optical properties. Shallow donor levels,
lying close to the conduction band, cause the material to become n-type semiconductor (using
electrons as carriers, e.g. Si doped by P). Shallow acceptor levels, lying close to the valence
band, take up electrons from it, thus creating holes and producing p-type semiconductors (e.g.
Chapter 1. Introduction
21
Si doped by B). The same consideration can be also applied to vacancies. An anion vacancy
may increase shallow donor levels close below the conduction band resulting in an n-type
semiconductor. A cation vacancy will act the opposite way, typical for p-type
semiconductors.
The presence of shallow levels is commonly found in transition-metal oxides (typical
also for SOFC materials) which can exist in several valence states. Their conduction band is
made up mainly from metal d orbitals (possibly mixed with s orbitals), while the valence
band is derived from oxygen 2p orbitals.
1.6 Structure descriptions
1.6.1 Perovskite
Perovskites and perovskite-related structures are common material types for SOFCs,
especially for cathodes. Their general formula is ABO3, with two different cation sites, A and
B respectively. The ideal perovskite structure has cubic symmetry with space group 3Pm m
and its example is given in Figure 1.6. The perovskite structure consists of larger 12-
coordinated A-site cations and smaller 6-fold coordinated B-site cation. BO6 octahedra are
linked forming a BO3 layer with cavities occupied by A-site cation. A-site atoms are typically
presented by alkaline and alkaline earth ions or rare earths, while B-site is typically occupied
by a transition metal.27,28
Figure 1.6: Structure of cubic perovskite SrTiO3 with Sr cations (green) on A-site and Ti cations
(blue) on B-site, oxygen anions are marked with red.
Chapter 1. Introduction
22
Many perovskite structures differ from the ideal cubic symmetry. Distortions within a
structure lower the symmetry (e.g. to hexagonal or orthorhombic). Additionally, a large
amount of oxygen or cation deficiency has been observed in many compounds. Certain types
of distortions are related to the structure's properties.
In order to understand the deviations from the ideal cubic structure, ABO3 oxides are
treated as purely ionic crystals where the relationship between the radii of A, B and O2−
is
given by:
2A O B Or r r r (1.8)
where rA, rB and rO are the ionic radii of A, B and O ions respectively. The deviation from the
ideal structure (and a measure of the geometrical driving force towards distortion) can be
expressed by the value of tolerance factor, t defined as:
2
A O
B O
r rt
r r
(1.9)
The value of t in perovskites varies between approximately 0.80 to 1.10.28
The oxides with
lower t values crystallise in the ilmenite structure.28
The values of t for ideal cubic structures
are mainly close to 1. Deviations of t are represented by systems with lower symmetry such
as orthorhombic and monoclinic.
1.6.2 Ruddlesden-Popper phases
Apart from cubic perovskite, there is a large family of perovskite-related structural
types. A few of the layered perovskite structures are mentioned in Section 1.7. Layered
perovskite structures are of much interest due to their diverse applications based on ionic
conductivity, photocatalytic activity, magnetic and dielectric properties.
Chapter 1. Introduction
23
Ordered B-sites and oxygen defects are presented within Ruddlesden-Popper phases
(RP, general formula An+1BnO3n+1). Their structure is composed of ABO3 perovskite units
intercepted by the AO rock salt unit (Figure 1.7). Based on the intergrowth of different
number of individual layers (n), various structures are accessible (such as RP1, RP2 phases
for n = 1 and 2 respectively). The list of possible compounds is expanded by structures with
two different A cations forming perovskite and rock salt layer (such as LaO∙nSrFeO3) or
phases with two different anions (i.e. SrFeO3∙SrF).28
Figure 1.7: Schematic of the structure of Sr2RuO4 and related compounds of Ruddlesden-Popper
series of Srn+1RunO3n+1, where n indicates the number of repeating RuO2 layers.29
1.6.3 Fluorite
Fluorite structures are common for SOFC electrolytes. It has a general formula given
by AO2 and an example of this structural type, CaF2, is shown in Figure 1.8. The cubic
closest-packed structure of CaF2 consists of eight-coordinated calcium cations with fluoride
ions adopting a tetrahedral arrangement. Oxide ion conductivity in AO2 fluorite oxides can be
enhanced by heterovalent substitution of A-site cation resulting in oxygen deficiency, such as
in yttria-stabilised zirconia (YSZ)30
Chapter 1. Introduction
24
Figure 1.8: Structure of fluorite CaF2 with Ca cations in blue and F anions in green.
1.7 Materials review
1.7.1 Cathodes materials
Several structural types have been examined for potential use as SOFC cathodes. This
review of cathode materials includes the three main structural families: perovskites,
Ruddlesden-Popper phases and layered perovskites.
Perovskite oxides ABO3−δ with 3d-transitional metals such as B = Mn, Fe, Co, Ni and
Cu were found to be most promising for SOFC cathodes.12
The most common material is the
doped lanthanum strontium manganite (LSM) showing high electrical conductivity
(320 S cm−1
at 800 °C for La0.6Sr0.4MnO3−δ composition),31
high electrochemical activity and
no compatibility problems.32
It has become an important cathode choice for high temperature
operations (700 ‒ 900 °C) although it exhibits no significant oxide-ion conductivity.12
Fe-
containing perovskite materials are favoured due to their low cost. Thus, A-site doped mixed
Chapter 1. Introduction
25
conductors of lanthanum strontium ferrite (LSF) have been studied for lower operation
temperatures.33-35
Cobalt-containing perovskites display high electronic and fast ionic conductivity both
with excellent electrocatalytic activity.17
Strontium-doped LaCoO3 (LSC) perovskites exhibit
p-type electronic conductivity due to the change of Co oxidation state and formation of
oxygen vacancies.36,37
Their total electrical conductivity is reported to be 1500 S cm−1
in air
at 600 °C,38
but the cathode application is limited by the high values of thermal coefficients
(TEC), usually 20 × 10−6
K−1
resulting in significant mismatch compared to common
electrolytes.17,39,40
A large amount of iron doping, such as in La0.6Sr0.4Co0.2Fe0.8O3−δ (LSCF),
reduces thermal expansion and provides good candidates for IT-SOFCs with ceria based
electrolytes.41-43
Reduced cathode performance due to strontium diffusion remains their main
limitation for long-term SOFC operation.44,45
Another widely studied perovskite is Ba1−xSrxCo1−yFeyO3−δ (BSCF), especially that of
x = 0.5 and y = 0.2 with a cubic structure.46
BSCF exhibits very low ASR values of
0.055‒0.071 Ω cm2 at 600 °C in Sm0.2Ce0.8O1.9 electrolyte based cells and high power
densities of about 1 W cm−2
at 600 °C.47,48
Incompatibility with the common electrolytes due
to the reactivity and high thermal expansion, typical for cobalt containing compounds, is the
main limitation for BSCF use as a SOCF cathode material.49
In addition, BSCF undergoes
phase decomposition under SOFC operating conditions50,51
and the presence of CO2 results in
the formation of thermodynamically stable BaCO3.52
Improving stability by increasing the Fe
content, such as observed in Ba0.5Sr0.5Co0.2Fe0.8O3−δ, leads to lower electrochemical
performance.53
Ruddlesden-Popper phases with general formula of An+1BnO3n+1 are of interest for
SOFC cathode application due to the high electronic conductivity caused by the mixed
valence of B-site element and the oxygen ion conductivity occurring via an interstitial
mechanism (e.g. lanthanum nickelates).22
Phases with n = 1 are most commonly studied,
although a few n = 2 and 3 phases, such as (La,Sr)n+1(FeCo)nO3n+1, have been examined as
cathodes.54,55
Depending on the nature of the cation, these compounds can crystallise in
different structural types. Nickel- and copper-containing oxides exhibit promising properties
for IT-SOFC.12
La2NiO4+δ (LNO) exhibits total conductivity of approximately 30 S cm−1
at
700 °C. An improvement of conductivity was observed for the A-site strontium-doped
compounds.56
LNO shows thermal expansion behaviour close to those for YSZ, GDC and
Chapter 1. Introduction
26
LSGM electrolytes.17,57
Chemical phase instability of LNO and its derivatives remains the
main limitation for its SOFC application.58,59
Double perovskites with general formula of AA'Co2O5+δ (where A is rare earth
element and A' = Ba, Sr) consist of double layers of square pyramidally coordinated Co
anions. These oxides are good mixed conductors with a rapid oxygen ion transport due to the
A-site ordering.17,60
Double perovskites with Gd and Pr are characterised by high coefficients
of diffusion and surface exchange of oxygen.61,62
The ASR value of GdBaCo2O5+δ (GBCO)
with GDC electrolyte was found to be 0.25 Ω cm2 at 625 °C.
62 Electrode performance of
GBCO, at the temperatures below 700 °C, has been examined also in other studies.63,64
Additionally, GdBaCo2O5+δ, as a most studied compound of the structural family (with
additional properties mentioned in Section 4.1), shows good chemical compatibility with
GDC and LSGM electrolytes and very good stability in CO2 atmospheres at temperatures up
to 700 °C.17
A vast number of AA'Co2O5+δ can be prepared and studied as potential SOFC
cathodes, including examples such as: YBaCo2O5+δ,65
LaBaCuFeO5+δ, LaBaCuCoO5+δ,66
and
SmBa0.5Sr0.5Co2O5+δ.67
Layered cuprates such as YSr2Cu2MO7+δ (M = Fe, Co) represents another group of
compounds studied as SOFC cathodes.68-70
The structural and electrochemical studies of
these materials are discussed in more details in Section 3.1.
1.7.2 Electrolyte materials
SOFC electrolytes are solely oxygen ion conductors. In order to achieve the
movement of the ions, the crystal structure of a material needs to contain unoccupied sites
equivalent to those of the lattice oxygen ions. Materials suitable for SOFC electrolytes belong
to several structural types, such as fluorite based (yttria and ceria derivatives), perovskite and
perovskite-related (LaGaO3 based oxides, brownmillerites), La2Mo2O9 (LAMOX) and
apatites.
The fluorite structure is shown in Figure 1.8. The zirconia-based oxides, are mainly
used as electrolytes for high temperatures (800 ‒ 1000 °C), more details about their properties
are available in selected reviews.71,72
Zirconia-based electrolytes show minimal electronic
contribution to the total conductivity and stable electrochemical performance over a broad
Chapter 1. Introduction
27
p(O2) range. The properties of zirconia-based electrolytes vary depending on the presence of
divalent and trivalent cation dopants.
Yttria-stabilised zirconia (YSZ) has become one of the most common SOFC
electrolytes. Zirconia (ZrO2) is known to exist in three different structures of monoclinic,
tetragonal and cubic phases.72
Yttria (Y2O3) is used to stabilise the high temperature cubic
phase, which is achievable even with a small amount dopant. A 8 mol% of Y2O3 is reported
to be the lowest concentration required to stabilise the cubic phase at room temperature.73
Substitution of Y3+
by Zr4+
also creates vacancies in the oxygen sublattice since the lower
valence of Y3+
compared to Zr4+
. This 8 mol% of Y2O3 in zirconia exhibits the conductivity
of 0.16 S cm−1
at 1000 °C.72
Operation at that temperature would bring significant constraints
to other parts of SOFC device. Scandia-stabilised zirconia (ScSZ) is an alternative to YSZ. It
shows two times higher ionic conductivity than those for any of the known zirconia-stabilised
electrolytes. The higher conductivity is attributed to the smaller ionic radii mismatch between
Zr4+
and Sc3+
. The use of ScSZ is limited by the presence of many phases, by the cost of Sc
and by the difficulty to obtain equilibrium in Sc2O3-ZrO2 system.72,74
Stoichiometric ceria (CeO2) is not a good oxygen ion conductor, but its conductivity
can be dramatically enhanced by low valance doping.24
It has been reported that the
conductivity and activation energy correlate with the ionic radius of the dopants ions, with
the lowest activation energy for the cations whose radius matches most closely to that of
Ce4+
.73
Ionic radii of Gd3+
and Sm3+
match most closely with that of the Ce4+
. Gadolinia-
doped ceria (GDC, CGO) and samarium-doped ceria (SDC) have become one of the leading
electrolytes for IT-SOFCs. High values of conductivity were also obtained for yttria-doped
ceria (YDC). All three ceria based electrolytes exhibit similar conductivities at 750 °C with
values of 6.5 × 10−2
S cm−1
for YDC,75
6.7 × 10−2
S cm−1
for GDC76
and 6.1 × 10−2
S cm−1
for SDC.77
At high temperatures and low oxygen partial pressure, ceria-doped electrolytes
exhibit n-type electronic conduction that is detrimental to SOFC operation. Co-doping has
also been carried out to improve the properties of ceria-based electrolytes. A decrease of the
electronic conductivity has been achieved for example in co-doping with Mg (in
Ce0.85Gd0.1Mg0.05O1.9)78
or with Pr (in Ce1−x−yGdxPrO2−z).79
Perovskite and perovskite-related electrolytes include two main structural types:
LaGaO3-based and Ba2In2O5 brownmillerites. Solid solutions of LaGaO3 oxide, especially
those with Sr and Mg (LSGM), exhibit better ionic conductivity than zirconia stabilised
electrolytes (Figure 1.9).80
The basic LaGaO3 compound can be doped with divalent ions
Chapter 1. Introduction
28
with appropriate sizes. Sr2+
and Mg2+
are favourable due to their low solution energies.73
Solid solution with Sr2+
and Mg2+
dopants was studied in extended range.81-83
The highest
values of conductivity of 0.17 and 0.08 S cm−1
were obtained for La0.8Sr0.2Ga0.83Mg0.17O2.815
at 800 and 700 °C respectively.84
A decrease of performance was observed in SOFCs
composed of LSGM, lanthanum-containing perovskite cathodes and Ni-LSGM anodes,
which was attributed to the anode-electrolyte reaction forming LaNiO3 and La2NiO4.85
Chemical stability of LSGM under reducing atmosphere at high temperatures is questionable
and causes significant changes in the surface morphology of the electrolyte.86
The presence of
several phases due to the vaporisation of Ga was detected by X-ray diffraction.
Figure 1.9: Conductivity data comparison of YSZ87
, CGO88
and LSGM84
; figure taken from17
Brownmillerite structures have been studied since the order-disorder transition in
Ba2In2O5 was observed.89
In order to suppress the phase transition, several different dopants
Chapter 1. Introduction
29
were tried.90,91
The conductivity of 0.12 S cm−1
obtained at 800 °C for (Ba0.3Sr0.2La0.5)InO2.75
was higher than that for YSZ.92,93
A new family of materials to be considered as electrolytes was reported by Lacorre's
group.94-96
These materials, based on La2Mo2O9 compound (also known as LAMOX),
represent alternative to common electrolytes, with the comparable conductivity properties
above 600 °C due to a phase transition from monoclinic α form to the cubic β form,
improving conductivity of the α form by two orders of magnitude. Several different doping
strategies have been applied to suppress the phase transition and improve the stability.97-100
The practical derivatives of La2Mo2O9 are limited due its reactivity towards some electrodes
and its high thermal expansion coefficients.
Apatites, with the general formula M10(XO4)6O2+y (where M is rare-earth element, X
is P, Si or Ge), have been proposed as potential electrolytes.101-103
Their values of
conductivity at higher temperatures are comparable with those for YSZ, but the conductivity
may be anisotropic, due to the structure, as in the case of lanthanum silicate apatites. The
conductivity in apatites involves interstitial oxide-ions, which differentiates them from
fluorite and perovskite based electrolytes.104-107
The conductivity of apatites is encouraging,
while high temperatures required for the synthesis and to material densification remains
unsolved.108
1.7.3 Anode materials
SOFC anodes provide the electrochemical oxidation of fuels (see Section 1.2.1) with
relevant requirements for the materials properties and fabrication (Section 1.2.3). Several
reviews about these anode materials are available,109-113
while here only a summary
highlighting the most common materials is outlined.
Early stage SOFC anode research was focused on the materials based on graphite,
platinum, iron, cobalt and nickel.73
Graphite was found to electrochemically corrode;
platinum did not show sufficient mechanical properties due to the water-vapour evolution.
Iron at a certain partial oxygen pressure oxidises, resulting in formation of a red iron oxide.
Cobalt was more stable but more expensive for fabrication. Nickel showed a big thermal
expansion mismatch with YSZ electrolyte and a tendency to aggregate at higher
Chapter 1. Introduction
30
temperatures. None of the metallic candidates showed ideal match for materials requirements.
Later on, a nickel-zirconia concept was reported114
and has become the leading
material for SOFC anodes. Ni provides the required electronic conductivity and catalytic
activity for both oxidation and for steam reforming. The YSZ provides ionic conductivity and
improves the thermal expansion match with YSZ electrolyte. The typical amount of Ni is at
least 30 vol% to achieve the percolation threshold for electronic conductivity.115
Few
disadvantages are connected with the use of Ni-YSZ. Since the catalytic activity of Ni is for
both steam reforming and hydrogen cracking, carbon deposition occurs when hydrocarbon is
used as the fuel and thus it prevents this material's use for operation with hydrocarbon fuels.
109,110,116,117 Impurities in the fuel stream, especially sulphur, are detrimental to the Ni-YSZ
anode, due to the low tolerance to sulphur.118
The problem of carbon deposition can be solved
by decreasing operating temperature88
or by a deposition of (Y2O3)0.15(CeO2)0.85 porous film
between YSZ and Ni-YSZ anode.119
Cu containing materials represent an option to replace Ni-YSZ cermets. For instance,
an anode composed of Cu‒CeO2‒YSZ/SDC showed improved cell performance, especially
for hydrocarbon fuels.120,121
Electrocatalytic properties of Cu-based materials are not as good
as those with Ni. Moreover, Cu is a relatively low melting point metal, having compatibility
problems with many standard SOFC fabrication procedures. An alternative is represented by
Cu-containing alloys with a second metal providing sufficient catalytic activity, with Ni as a
reasonable option.14
The bimetallic anodes (such as Cu/Ni, Cu/Co in CeO2/YSZ) exhibited
improved performance and in the case of Cu/Co also improved carbon tolerance.122
A tri-
metal anode, FexCo0.5−xNi0.5/Sm0.2Ce0.8O1.9(SDC) cermet, with GDC electrolyte and
Sm0.5Sr0.5CoO3/SDC cathode (with x = 0.25) exhibits much better electrochemical
performance than this observed for Ni/SDC.123
Alternative anode materials may be categorised into several groups depending on the
structural type they posses. Ceramics based on CeO2 having a mixed ionic-electronic
conductivity, are the most common fluorite candidates for SOFC anodes. Addition of Ni, Cu,
Pt, Ru improves their catalytic activity.124-126
The most studied perovskite materials are the
titanates and chromites due to their stability in reducing conditions,109,127
with
La0.75Sr0.25Cr0.5Mn0.5O3−δ as an examples having similar electrochemical performance to that
of Ni/YSZ cermets and good catalytic activity.128,129
Tungsten bronze materials were also
studied as SOFC anodes showing poor oxygen exchange kinetics.130,131
Pyrochlores, based on
Gd2Ti2O7, have been found to have a high mixed conductivity but their potential use as SOFC
Chapter 1. Introduction
31
anodes is reduced by the stability problems causing changes of electrochemical performance
depending on p(O2) region.132,133
1.8 Aims of the work
The work for this thesis is focused on new perovskite-based materials for solid oxide
fuel cells (SOFCs) including the development of both cathode and electrolyte materials which
can be divided into three structural types: YSr2Cu3−xCoxO7+δ cuprates (Chapter 3), GBCO
related phases (Chapter 4) and Ruddlesden-Popper structures (Chapter 5). Both experimental
and theoretical methods have been applied during the research described within this thesis.
Previously reported work on YSr2Cu2MO7+δ (M = Fe, Co) revealed the triple
perovskite materials as a potential SOFCs cathode candidate.69,70
Cobalt plays an important
role in catalysis which is necessary for the SOFC operation.17
By increasing the cobalt
content in Y1−ySr2+yCu3−xCoxO7+δ compounds (Chapter 3), and thus introducing electronic
carriers, improved conductivity properties are expected. Perovskite cuprates are found to
have thermal and compatibility problem,68,70,134-137
which may be detrimental to SOFCs
application. Therefore, the increasing of the electrochemical stability and compatibility of
Y1−ySr2+yCu3−xCoxO7+δ represents the other challenge of the work with triple perovskite
materials.
Expansions in unit cell size driven by the ordering of cations and oxide vacancies (in
order to satisfy local coordination preferences) are known for many layered perovskite
structures. Double perovskites are one of the superstructures of interest for SOFCs studies.
The ordering of Gd and Ba and accompanied oxygen vacancy ordering in GdBaCo2O5+δ
results in high oxide mobility, producing excellent SOFC cathode performance.60,62,138
By
adding a fluorite layer to a mixed double perovskite conductor, an improvement of ionic
conductivity is expected. DFT studies are used as a tool for predicting the chemical stability
of different double perovskite structures with different fluorite layers (Chapter 4). The
structural model is based on Y2SrFeCuO6.5 structure.139,140
The most stable compounds
predicted are then synthesized using a solid state chemistry synthesis.
SOFC research on Ruddlesden-Popper phases is focused mainly on the mixed ionic-
electronic conductors such as A2NiO4+x phases (A = La, Sr, Pr, Nd).56,141-143
The presented
Chapter 1. Introduction
32
work on the RP1 phases aims to find ionic conductors with interstitial oxygen. The target is
to find pure ionic conductor with a potential use as a SOFC electrolyte. The RP1 structures
can support interstitial and vacant oxygen sites (as reported for La1.5+xSr0.5−xCo0.5Ni0.5O4+δ).144
Several A- and B-site doped RP1 structures (with Sr, Ba on A-site and Sn, Zr, Hf on B-site)
are looked at using DFT calculations to investigate interstitial oxygen energetics (Chapter 5).
The theoretical work is followed by the structural and electrochemical characterization of
synthesized phases of doped Sr2SnO4 phases.
Chapter 2. Experimental and theoretical methods
2 Experimental and theoretical methods
2.1 Material synthesis
The majority of the materials discussed within this thesis were prepared via the
standard ceramic methods.145,146
The starting powdered materials (oxides and carbonates)
were treated in a way to remove any moisture absorbed in the powder when stored at room
temperatures. Most of the starting oxides were calcined at ≈ 200 °C except lanthanide oxides,
which were kept at ≈ 900 °C due to their stronger hygroscopic properties. Carbonates were
stored in a desiccator. Precise stoichiometric amounts of starting materials were weighed out
and ground together using a pestle and mortar. Mixing of the starting materials is required in
order to ensure the homogenity of the mixture, to reduce the size of the grains and thus
maximising the surface contact between the reagent grains which enhances the reaction rate.
A few synthetic routes and physical property measurements (e.g. DC conductivity, AC
impedance measurements) require improved mixing with smaller particle sizes, which is
ensured by ball milling.146
A FRITSCH Pulverizette 7 planetary mill was used for ball
milling of the materials in this thesis. The ground reaction mixture is then pressed to form a
pellet, which maximises the surface contact between reagent grains and enhances the reaction
rate. Some sample may require a calcination step at a lower temperature to begin the
chemical reaction and improve the homogeneity of the sample. A calcination step is also
required for the synthesis of volatile starting materials.
Once the mixing and/or calcination step has been completed, the powder or pellet is
sintered to high temperatures (typically over 1000 °C) for several hours or days. Sintering at
high temperatures allows the high kinetic barrier to reaction to be overcome, allowing the
thermodynamically favoured product to form.27
The sintering process needs to be commonly
repeated with re-grinding of the sample in order to produce a single phase sample. Planetary
milling used at the very beginning of the synthesis may decrease the number of sintering
cycles. The vast majority of the syntheses in this thesis were carried out in ambient air.
Sintering at different atmospheres (oxygen, nitrogen) may be used to alter the oxygen content
and to change the final product. It should also be taken into account that syntheses in
reducing atmospheres may lead to the production of materials that are not stable in certain
Chapter 2. Experimental and theoretical methods
34
conditions. Since SOFCs materials are required to be stable at oxidising/reducing conditions,
which depends on the part of the (Section 1.2.3), reducing atmosphere was not used for the
synthesis. The final powder products were then characterised using the methods described in
the following sections.
Several Ruddlesden-Popper phase were synthesised using Pechini sol-gel method
(PM).147
Samples of outstanding homogenity can be prepared by this polymerizable complex
method. PM is based on in situ polymerization of monomers specially introduced into
solution in addition to the required metal cations. Suitable metal salts are introduced into the
ethylene glycol after dissolution of citric acid forming metal-citrate complex. The mixture is
then heated to ≈ 100 °C while stirring to enhance the formation of polyester. After the
formation of a plastic-like gel, the temperature is increased in order to remove the excess of
ethylene glycol. The obtained mixture is then heated at 500 ‒ 600 °C in order to oxidise the
organic compounds and to form precursor powders ready for further annealing at
temperatures depending on the product required.
2.2 Powder diffraction techniques27
2.2.1 Fundamentals of diffraction
The main technique used to describe structures of the solids is diffraction. A crystal
structure consists of an infinite array of constituents units (a unique unit of atoms or
molecules) forming a three-dimensional lattice. The lattice can be described by the unit cell ‒
chosen as the repeating unit. In order to define the unit cell, three translation vectors a, b, c
and the interaxial angles α, β, γ are required. Based on the values of the six lattice parameters,
seven crystal systems are defined. In the case where a ≠ b ≠ c and α ≠ β ≠ γ, the lattice has
intrinsic translation symmetry and may have inversion symmetry, but it has no rotation axes
or mirror planes. Such a lattice needs to be defined by six unique lattice parameters and is
described as a 'triclinic' system. On the other hand, the 'cubic' system has the highest number
of non-translation symmetry elements and is defined by only two unique lattice parameters a
(= b, c) and α (= β = γ = 90°).
There are fourteen independent ways of arranging points in space giving rise to 14
Bravais lattices. If the unit cell contains just a single lattice point, it is described as primitive.
Chapter 2. Experimental and theoretical methods
35
Sometimes it is necessary to define a unit cell with more than one lattice point and such a cell
is described as centred. Based on the studies of the symmetry the crystals may have one or
more than ten basic symmetry elements (e.g. proper rotation axes, inversion or improper
axes, centres of inversion and mirror planes). A set of symmetry elements intersecting at a
common point within a crystal is called the point group. The 10 basic symmetry elements
along with their 22 possible combinations constitute the 32 crystal classes. There are two
additional symmetry elements: screw axis and glide plane, which arise in crystal but have no
counterpart in molecular symmetry. A combination of these elements with the point group
symmetry is called a space group. There are 230 possible space groups in total. The extended
translation symmetry in crystals allows diffraction methods to be used as a powerful tool for
structural characterization and determination, as described in the following sections.
2.2.2 Diffraction of X-rays148,149
The discovery of X-rays opened a vast area of scientific research in many fields. It
plays a crucial role in crystallography. For diffraction to take place, the wavelength of the
incident light has to be of the same order of magnitude as the spacings of the grating (X-ray
wavelength has a magnitude of approximately 1 × 10−10
m or 1 Å). The first theory of the X-
rays diffraction was developed by Laue at the beginning of 20th
century. The crystal was
considered as a three dimensional lattice with rows of regularly spaced atoms, but the theory
can be illustrated by a one dimensional lattice (Figure 2.1). A parallel monochromatic X-ray
beam is incident upon a row of regularly-spaced atoms. Each atom in the row behaves as a
point scatterer ‒ a secondary source of monochromatic X-rays with a spherical wavefront.
Interference then occurs between these scattered beams and maximum constructive
interference will occur when the path lengths differ by an integer (n) of wavelength (λ)
according to:
(cos cos )i i dAB CD X n (2.1)
where AB and CD are path lengths (Figure 2.1). This equation is valid for a one-dimensional
system. A real crystal is a three dimensional system with three Laue equations, corresponding
Chapter 2. Experimental and theoretical methods
36
to each orthogonal x, y, z axis, which must be simultaneously satisfied for a constructive
interference to occur and can be written as follows:
0(cos cos )n xa n (2.2)
0(cos cos )n yb n (2.3)
0(cos cos )n zc n (2.4)
Laue equations indicate that a periodic lattice produces diffraction maxima at specific
angles, which are defined by lattice parameters. The equations give the most general
representation of a three dimensional diffraction pattern and they may be used in the form of
Equations 2.2-2.4 to describe the geometry of a single crystal diffraction. The solution of the
Laue equations involves the determination of up to 12 variables. This complexity limits their
usage in practical crystallography.
Figure 2.1: Schematic diagram illustrating Laue diffraction from a lattice in the x direction. The
labelled distances and points are used to construct the first Laue equation (Equation 2.1).
Chapter 2. Experimental and theoretical methods
37
An alternative formulation to the Laue's model was suggested by W. L. Bragg.
Bragg's equation describes the principle of X-ray diffraction in terms of reflection of X-rays
by sets of lattice planes. Lattice planes are crystallographic planes characterized by the index
triplet hkl known as Miller indices. Parallel planes have the same indices and are equally
spaced, separated by the distance dhkl. Bragg's analysis deals with X-rays like visible light
being reflected by the surface of a mirror (Figure 2.2). In contrast to the visible light, the X-
rays penetrate inside the material. This causes additional reflections at many consecutive
parallel planes. Since all X-rays are reflected in the same direction, superposition of the
scattered rays takes place. For a constructive interference to occur, the path difference
(B'C − BC) has to be equal to an integer number of wavelengths, resulting in the following
condition:
' ; ( ' ) cos 2sin
hkldB C BC B C
(2.5)
therefore for the maximum constructive interference:
cos 2
sin sin
hkl hkld dn
(2.6)
which is simplified to the equation known as Bragg's equation:
2 sinhkld n (2.7)
Other useful equivalent variations of the Bragg's equation are:
0
2sin 1h s s
d
(2.8)
4 sin 2Q
d
(2.9)
The vector Q is the physicist's equivalent of the crystallographer's scattering vector h while s
and s0 are unit vectors along the directions of the diffracted and incident beams respectively.
Chapter 2. Experimental and theoretical methods
38
The physical meaning of Q is the momentum transfer on scattering and differs from h by a
factor of 2π.
Figure 2.2: Schematic diagram of Bragg diffraction from planes of atoms, θ indicates the incident and
diffracted beam angles and d(dhkl) the inner spacing within the crystal structure with Miller indices
hkl.
2.2.3 Diffraction of neutrons
The diffraction of neutrons is described in the same way as X-ray diffraction
(Section 2.2.1). Neutrons can be diffracted by a crystal if they have an appropriate energy.
The scattering mechanism in the case of neutrons is slightly different since neutrons are
scattered by atomic nuclei via the nuclear force. The nuclear force acts at a different range
(10−15
m, compared to a common incident wavelength of 10−10
m). Thus the scattering objects
for the neutron diffraction are extremely small and the atomic scattering factors are
independent of the incident angle (θ) of the beam.
Neutron scattering amplitude is not related to electron count but it depends on the
scattering length of atoms. Every isotope has a unique coherent scattering length (the
effective size of the nucleus), which may take a negative value and is not related to chemical
properties. Typical scattering lengths range from −5 fm to 16 fm. This makes neutron
diffraction an outstanding tool to differentiate light elements (such as oxygen) in the presence
Chapter 2. Experimental and theoretical methods
39
of heavy ones. Sample containers are generally made from vanadium which posses a very
small coherent scattering length (≈ −0.38 fm).
Another important feature of neutrons is their magnetic moments (spin 1/2 particle).
This means that the neutrons can also be diffracted by a lattice of ordered magnetic moments.
A neutron diffraction pattern is thus a superposition of a structural pattern and a magnetic
pattern. Magnetic structure crystallography represents another important application of
neutron diffraction.
2.2.4 Powder diffraction
Diffraction methods mentioned in previous sections were related to single crystals. In
practice, it is difficult to grow single crystals for many materials. Therefore, instead of single
crystals, polycrystalline samples have to be used. A distribution of crystallite sizes of powder
samples should be of a few microns in order to have a distribution from many crystallites.
Ideally, every crystallite of the sample should be randomly orientated to the incident beam, so
that each hkl reflection gives a uniform cone-shaped diffracted beam (Debye-Scherrer cone)
with contributions from many crystallites (Figure 2.3). Although the powder diffraction data
lack the three-dimensionality, the powder diffraction pattern represents a one-dimensional
snapshot of the three-dimensional reciprocal lattice of a crystal.148
The quality of the pattern
depends on the nature and the energy of the applied radiation, the resolution of the
instrument, and the physical and chemical conditions of the sample. In typical powder
diffraction experiments the intensity of the diffracted beam is measured as a function of 2θ.
Since all collected information is condensed into one dimension, hkl peaks with identical or
similar d-spacing overlap to some degree. Peak overlap tends to be more problematic for the
materials with lower crystal symmetry. Thus, powder methods are mostly used for high-
symmetry solids and less for complex molecular solids. Problems with overlapping most
often increase for multi-phased materials.
Phase and overlapping problems make the use of powder diffraction difficult as a tool
for a direct structure solution. After recent developments of instruments and characterization
methods, powder diffraction techniques have become more powerful and useful as a tool for
phase purity identification. High quality data (synchrotron ‒ Section 2.2.7, neutron diffraction
Chapter 2. Experimental and theoretical methods
40
‒ Section 2.2.8) combined with Rietveld refinement149-151
can reveal structural details that in
turn can explain material properties and phenomena.
Figure 2.3: Schematic diagram of a Debye-Scherrer reflection cone from a diffraction of an incident
beam by a polycrystalline powder material.
2.2.5 The Rietveld Method
The Rietveld method was originally developed for powder neutron diffraction by H.
M. Rietveld150,151
to derive the maximum amount of information from a polycrystalline
powder. It is not a structure solution method, instead it relies on the input of a starting model
which generates a theoretical diffraction pattern. Rietveld refinements shown in this thesis
were carried out using the Topas152
software package.
Since a powder diffraction pattern is a set of peaks with overlapping and slightly
varying background, a Rietveld refinement may present a complex curve fitting problem. To
identify the true contribution to Bragg reflections to an observed peak, correct intensity to
each peak needs to be assigned. The Rietveld method does not use the integrated intensity of
each peak, but it considers the intensity of each individual data point as a single observable.
The intensity of a data point yi may contain contributions from several Bragg reflections. The
Fhkl values are generated from the structural model and are used to calculate theoretical
intensity for this point (yci).
Chapter 2. Experimental and theoretical methods
41
The sum of other contributions to an observed intensity is as follows:
2(2 2 )ci bi hkl hkl hkl i hkl
hkl
y y s L P A F (2.10)
where s ‒ scale factor, Lhkl ‒ combined Lorentz polarisation and multiplicity, Phkl ‒ preferred
orientation and A ‒ absorption are four factors depending mainly on experimental set-up;
θ ‒ reflection profile function which describes the peak shapes and ybi ‒ the background
intensity. Fhkl refers to the structure factor for a suggested structural model.
Structural factor (Fhkl) contains information on the atomic scattering contribution at
each reflection from all of the atoms of the unit cell and is defined by:
2
1
( ) ( )j j jn
i hx ky lzj j j
hkl
j
F g t s f s e
(2.11)
where:
n denotes the number of atoms of the unit cell,
gj − fractional occupancy of atoms j, where atom j fully occupies the site g
j,
(s) − the angular dependence,
f j − the atomic scattering factor depending on the type of the diffraction,
exponential term includes the fractional coordinates (x, y, z) for atom j by the
corresponding h, k or l value.
Once a pattern of the material is calculated, corresponding values of yci are obtained
and the difference between the theoretical and observed patterns (yi − yci) is calculated. The
difference is then incorporated into a 'residual function' Sy. Adjusting the structural or set-up
parameters allows the value of Sy to be minimised by a least-squares method:
2( )i ciy
i i
y yS
y
(2.12)
The least squares optimisation is a feedback loop with a certain user-defined parameter of a
starting model and an improved model as an output model of the calculation which is used for
Chapter 2. Experimental and theoretical methods
42
the next loop. When the refinement is finished, an adequate model closely corresponding to
the observed pattern is obtained.
The accuracy of the refinement can be expressed by several statistical outputs, known
as R-factors. The factors describing the quality of the full-pattern fitting are based on directly
observed intensities. These include the R-pattern (Rp) and R-weighted pattern (Rwp) expressed
by:
i ci
ip
i
i
y y
Ry
(2.13)
2
2
( )i i ciyi
wp
i i i i
i i
w y yS
Rw y w y
(2.14)
where wi is the weighting factor. The final important R-factor is the R-expected factor (Rexp).
Rexp can be calculated from a pattern, which can be considered as the best expected fit for a
pattern.149
varexp 2
obs
i i
i
N NR
w y
(2.15)
where Nobs and Nvar are the number of observables and refined parameters, respectively. For a
common powder experiment Nobs ≈ 2000, therefore Nobs ≫ Nvar. The only difference between
the expected R factor and Rwp value if the model was perfect is due to statistical fluctuations.
The end of the refinement is typically quoted by a goodness-of-fit (GOF), χ, related to both
Rwp and Rexp:
var exp
y wp
obs
S R
N N R
(2.16)
For an ideal powder diffraction experiment with sufficient data point number and
counting time (background intensity is less dominated in yci) with a superb structural model,
the value of GOF is expected to be in the range 1.0 ≤ χ ≤ 1.3. Practically the χ are higher,
Chapter 2. Experimental and theoretical methods
43
mostly due to insufficiency of at least one of the set-up parameters. It needs to be noted that
the statistical evaluation should not be the only criterion taken into an account. A graphical
representation is an excellent way to judge the quality of the refinement and it should not be
omitted. A model is considered to be acceptable if it fits the observed data well, with the
reasonable statistical factors and it makes physical and chemical sense.
2.2.6 Laboratory X-ray diffraction
To generate an X-ray in laboratory based instruments, electrons are produced by
heating a metal filament, accelerated using high voltages of 40 kV through a vacuum tube
into an anode material, typically made of copper, cobalt of molybdenum. The incident
electrons have energy sufficient to initiate a number of electronic transitions within the metal.
This leads to the emission of photons with a broad spectrum of X-rays at most voltages (low
intensities) and with a sharp emission (where the exact voltage is needed) at specific
wavelengths typical for each anode metal. There are many electronic transitions available in
metals, therefore it is common to observe multiple characteristic emissions. The core
electrons of Cu atoms include 2p or 3p electrons. Kα photon is given by a 2p → 1s transition,
when Kβ photon is given by a 3p → 1s transition. Due to the spin-coupling, the transitions are
complicated yielding in two different spin states and thus resulting in two characteristic
wavelengths (Kα1 and Kα2). Since a monochromatic X-ray beam is required, one radiation is
typically selected and filtered using an appropriate monochromator. The instruments used for
the laboratory Powder X-ray Diffraction (PXRD) characterization had cobalt (a Panalytical
X'pert Pro diffractometer, λ = 1.789 Å) and copper target (a Bruker D8 Advance
diffractometer, λ (Cu Kα1= 1.541 Å).
2.2.7 Synchrotron X-ray powder diffraction
Synchrotron sources show notable advantages compared to a common laboratory
sources. The more intense beam allows faster data collection with an improved resolution of
the obtained diffraction patterns. Synchrotrons are considered, together with time-of-flight
Chapter 2. Experimental and theoretical methods
44
neutron diffraction, as the highest available resolution tools for powder crystallography.
Another advantage of the synchrotron-based experiments is the variation of available
wavelengths. This can be useful for extreme environment studies such as high-pressure
diffraction.
The synchrotron data presented in this thesis were collected using I11 High-
Resolution Diffractometer at Diamond Light Source, U. K. The beamline is described in
more details by Thompson et al.153
The source, an in-vacuum undulation, is inserted in an
array of permanent magnets, and it produces X-ray ranging from 11 to 20 keV. The optical
hutch, consisting of slits, monochromators and harmonic rejection mirrors, focuses the beam
resulting in a small final beam size. The instrument works in Debye-Scherrer geometry. Si
and Ge are known as a suitable analyzers for synchrotron powder beamlines.154
I11 beamline
detecting system is based on multianalyzing crystal (MAC) assembly. To enhance the data
collection, five identical nine-crystal MAC-arms of total of 45 detectors are mounted at 30 °
intervals. Each individual detector corresponds to Si (111) analyser crystal. A picture of I11
experimental hutch with the closer look at the sample detector area is shown in Figure 2.4.
Samples for I11 beamline are mounted in spinning glass capillaries.
Figure 2.4: a) I11 experimental hutch showing the diffractometer (DIF), 5 arms for MAC-detectors
(MACs), robotic arm (ROB), carousel with 200 specimen positions (CAR), and heavy duty Table
(XYZ); b) sample-detector arrangement with 5 sets of MAC detectors. Both images are taken from.153
2.2.8 Neutron Sources and Time-of-Flight Diffraction
Neutrons for diffraction experiments can be generated using two different ways, by
reactor sources or by spallation sources. Neutron diffraction data presented in this thesis were
Chapter 2. Experimental and theoretical methods
45
collected at the ISIS facility (Rutherford Appleton Laboratory, U. K.). ISIS sources
(schematic of which is shown in Figure 2.5), as other spallation sources, do not produce a
continuous beam. The neutron generation starts with the formation of hydride ions which are
accelerated in a linear accelerator. A beam of protons, generated after the collision of hydride
ions with an aluminium oxide target, are then accelerated in a synchrotron to obtain an energy
of 800 MeV. The high energy proton beam then hit a tungsten target resulting in neutron
production. A pulsed neutron beam causes energy-disperse mode of the neutron diffraction
experiments known as time of flight (TOF) diffraction.
The behaviour of neutrons is described by the de Broglie relationship:
n
h h
p m v (2.17)
where h is Planck constant, p is momentum, mn is neutron mass, and v is velocity.
Considering the fixed path length between moderator and detector L, and the time t for
neutron to reach the detector Equation 2.17 can be modified to:
n
ht
m L (2.18)
which can be substituted into Bragg's equation (Equation 2.7) giving the relationship between
the neutron TOF and d spacing:
2 sinhkl
n
htd
m L (2.19)
The values of dhkl can be determine as the variable in TOF experiments due to the detectors at
fixed 2θ positions and fixed path length L.
The dependence of the intensities of neutron Bragg reflections to 2θ is not strong.
Neutron diffraction is described by the same equations as the X-ray diffraction but the
scattering mechanism is different since neutrons are scattered by atomic nuclei via the
nuclear force. Typical wavelength of the nuclear force is at a range of 10−15
m, compared to
10−10
m incident wavelength. Thus, the scattering objects for neutron diffraction are
extremely small and are independent of the incident angle. In order to prevent any
Chapter 2. Experimental and theoretical methods
46
overlapping of pulses, the neutron beam is not heavily moderated at ISIS. As a consequence
of this, a relatively high population with short wavelengths is observed. Therefore, reflections
with short d-spacing are common and their intensities are stronger than those observed by
synchrotron. The θ position of the detector affects the range of the observed d-spacing, which
can be seen from Equation 2.18. Thus the detector at a low scattering angle will show the
longest d-spacing and vice versa, a detector at high scattering angles the smallest d-spacing.
Having a multiple detector banks in TOF experiments is common. Low angle banks are
typically used for magnetic structure refinements, since magnetic peaks are common for a
long-range structural ordering. High-angle banks provide the highest resolution data of
reactions with small d-spacing. An intermediate bank is used to collect the data with
intermediate counting statistics and resolution for intermediate d-spacing. The High
Resolution Neutron Powder Diffraction (HRPD) instrument, used for the TOF data collection
shown in Section 5.5.3 of this thesis, has three banks of detectors, fixed at 30, 90 and 168 °.
The data can be collected over the d-spacing between 0.3 ‒ 16.5 Å. The NPD data at 740 °C
shown in Section 3.3.2 were collected using GEM instrument. The GEM diffractometer
allows collection of the data over a large range of d-spacing using six banks of detectors in a
relatively short time.155
Figure 2.5: Schematic of the ISIS neutron source, with all of the associated neutron diffraction
instrumentation highlighted, figure taken from the STFC-ISIS webpage.
Chapter 2. Experimental and theoretical methods
47
2.3 Scanning Electron Microscopy
Electron microscopy is often used for the study of structure, morphology, crystallite
size or defects of solids. An electron beam used in Scanning Electron Microscopy (SEM) is
focused on a sample surface.27,156
The electron beam is produced by heating a tungsten
filament and accelerating to energies between 2 and 40 keV. Beam electrons interact with the
surface of the sample producing secondary electrons. These electrons are detected giving a
topographic map of the studied material. Secondary electrons can be observed only from the
near-surface region of a conducting surface. Therefore, gold or carbon thin coating is
required for insulating materials. Images of the scanned material with various range of
magnification can be achieved. The very short wavelength of the electron allows resolution to
0.1 nm. The SEM study of Y0.95Sr2.05Cu1.7Co1.3O7+δ material presented in this thesis
(Section 3.7.3) was performed using a Hitachi S-4800 scanning electron microscope and the
analysis of the sample was carried out by a low 3 kV electron beam.
2.4 Electrical conductivity measurements
2.4.1 Fundamentals
One of the key properties of solids is their electrical resistivity. Different materials
exhibit resistivity varying over 20 orders of magnitude. There is no single method available
to measure all of the materials. Electrical resistivity (ρ) of a material describes how the
material resists the flow of electricity. In a simplified microscopic model electricity is shown
as a simple movement of electrons157
. Electrons may collide with atoms of the material.
Every collision slows down the electron. A material, which produces few collisions is a low-
resistivity material, those producing lots of collisions is referred as a high-resistivity material.
Electrical resistivity is thus geometry dependent. The resistivity can significantly vary with a
change of temperature. The resistivity of metals increases as temperature increases while the
resistivity of semiconductors usually decreases. Electrical conductivity (σ) is defined as the
inverse of electrical resistivity and is given by:
Chapter 2. Experimental and theoretical methods
48
RA
l and
1
(2.20)
where A is the cross-sectional area, l is the length and R is the electrical resistance of a
sample. The electrical resistance is defined by voltage (V) and an electric current (I)
following Ohm's law:
VR
I (2.21)
It should be noted that the resistance R depends on the size and shape of the measured
material while the resistivity ρ is independent of the shape and size of specimen.
2.4.2 The four-probe DC method
All of the DC measurements mentioned in this thesis (Section 3.4) were carried out
using the four-probe measurement technique (using a Keithley 220 Current Source and a
Keithley 2182 Nanovoltameter). The schematic of the measurement arrangement is shown in
Figure 2.6. The measurement is performed on a bar of material with four wires (contacts)
attached to it. The contacts are made of a wire and paste (typically made from gold or silver).
An ammeter measures the current I passing through the specimen. Inner contacts are
connected to voltmeter measuring the voltage V. The four-probe resistivity is then expressed
by:
1
Vwh
ll (2.22)
where w is the width of the sample bar, h ‒ height of the sample bar, l ‒ the distance between
the two outer contacts (where the ammeter is connected) and l1 is the distance between the
two inner contacts (connected with voltmeter). All the values of distance are in metres. The
quality of contacts is essential for a DC measurement. That includes the geometry of the
Chapter 2. Experimental and theoretical methods
49
contacts where l1 distance should be much larger than the thickness of the bar. The contacts
should be thin as possible for the satisfactory accuracy and completely independent. Another
necessary condition is the density of a measured material, which is expected to be over 90%.
Figure 2.6: A four-probe DC technique of a bar of material. The voltmeter measures the voltage
between inner contacts whilst the ammeter is connected to outer contacts.
2.4.3 Cold Isostatic Pressing
Cold IsostaticPressing (CIP) is a method of applying pressure from multiple direction
(using hydrostatic pressure) to a sample in order to obtain great uniformity of compaction
over the entire surface.158
The CIP is commonly used to increase the density of a material for
physical property measurements and it was used during the work presented in this thesis;
before the DC (Section 3.4) or AC impedance measurements (Section 5.6.1). Samples were
first pelletised, then sealed in waterproof bags and lowered into the hydraulic fluid. Dense
pellets were then made by CIP using an Autoclave Engineers Cold Isostatic Press, under a
pressure of 206.85 MPa. The samples were dwelled at the pressure for 3 min, then the
pressure was slowly released and the sample bag was taken out of the autoclave.
Chapter 2. Experimental and theoretical methods
50
2.4.4 Density measurements
The density of ceramic materials before the conductivity measurements was obtained
using an Archimedes' Principle balance. According to Archimedes' principle, the upward
force that is exerted on a body immersed in a fluid (fully or partially submerged) is equal to
the weight of the fluid that the body displaces. Sample densities (ρs) were calculated from the
following formula:
1
3 2
s l
m
m m
(2.23)
where m1 is the mass of dry sample, m2 is the mass of immersed sample, m3 is the mass of
soaked sample and ρl is the density of a liquid. All of the samples were immersed in distilled
water. The relative density (ρrel) of the prepared pellets was calculated as a fraction of the
actual density of the sample (ρs) to the theoretical crystallographic density (ρtheor, calculated
using lattice parameter data from the Inorganic Crystal Structure Database, ICSD) according
to:
(%) 100srel
theor
(2.24)
2.5 AC Electrochemical Impedance Spectroscopy (EIS)
2.5.1 Fundamentals
The use of Ohm’s law is limited to only one circuit element ‒ an ideal resistor. The
real elements (electrolytes or electrodes in our studies) are more complex and need to be
described by impedance (Z) rather than resistance. Like resistance, impedance defines the
ability of a circuit to resist a flow of electrical current. Electrochemical impedance159
is
typically carried out by applying an AC potential to an electrochemical cell and measuring
the current flowing through the cell. The current response, It, is shifted in phase (by θ, which
is related to the radial frequency ω = 2 πf) and has a different amplitude compared to a
sinusoidal signal of applied voltage (Vt) expressed by following equations:
Chapter 2. Experimental and theoretical methods
51
sint mV V t (2.25)
sint mI I t (2.26)
where Vm and Im are magnitudes of voltage and current respectively.
An equation to calculate impedance, analogous to Ohm’s law, is given by:
sin sin
sin sin
mtm
t m
V t tVZ Z
I I t t
(2.27)
with the impedance magnitude Zm = Vm/Im.
It is common to express the impedance as a complex number composed of real (Z’, ReZ) and
imaginary (Z’’, ImZ) terms:
cos sini
m m mZ Z e Z i Z
(2.28)
where Z’ = |Zm| cos(θ) and Z’’ = |Zm| sin(θ). Figure 2.7 shows one of the common impedance
data representations ‒ the Nyquist plot. The impedance is presented as a vector of length |Z|
with θ, the angle between the vector and the x-axis.
Figure 2.7: Nyquist plot obtained from impedance data collection (taken from160
) consisting of a
single semicircle.
Chapter 2. Experimental and theoretical methods
52
2.5.2 Data analysis
Electrochemical impedance plots often contain more than one semicircle. Each of the
individual semicircles is analysed by fitting to an equivalent electrical circuit model. Most of
the models are based on common electrical elements such as resistors, capacitors and
inductors with their impedance defined by following equations:
:resistor Z R (2.29)
:inductor Z j L (2.30)
1:capacitor Z
j C (2.31)
In the cases where the impedance semicircles are depressed, a constant phase element (CPE)
can be used for data analysis. The CPE impedance is expressed by:
( )Z A j (2.32)
where A is the inverse of the capacitance (= 1/C) and α, an exponent which is equal to 1 for a
capacitor. The capacitance of CPE is related to the depression angle by:
1
1( )n nC R Q (2.33)
where R is the element resistance, Q is the pseudo-capacitance and n is a parameter related to
the depression angle. The output of impedance data analysis contains further information
about the physical processes occurring within the measured electrochemical cell. That is
often due to the various frequencies at which the different physical phenomena relax. The
values of capacitance extracted from the equivalent circuit fits depend on different
contributions161
(examples given in Table 2.1).
Chapter 2. Experimental and theoretical methods
53
Table 2.1: Capacitance values and their possible interpretation.161
Capacitance (F m−1
) Phenomenon responsible
10−12
Bulk of electrolyte
10−11 Minor, second phase
10−11
‒10−8
Grain boundary
10−4
Electrochemical reactions
Once all of the equivalent circuits are assigned, the value of bulk and grain boundary
conductivity (σ) can be calculated from the sample dimension: l ‒ length and SA ‒ surface
area (Equation 2.34). The cathode performance of a material is commonly expressed by the
value of area specific resistance (ASR) which is obtained using total resistance for the
cathode surface area divided by 2 (taking the cell symmetry into account, as is shown in
Equation 2.35.
l
R SA
(2.34)
2
totR SAASR
(2.35)
2.6 Ultraviolet-visible and Infrared Spectroscopy
2.6.1 Ultraviolet and visible Spectroscopy
Ultraviolet-visible (UV-vis) spectroscopy uses light in the visible and adjacent (near
UV and near infrared ‒ NIR) ranges. It is a technique for determining if a material is able to
absorb radiation in the UV or visible region. That can be found from the optical absorption of
the material determined from the optical absorption coefficient as a function of the energy of
photon corresponding to wavelengths of light. The optical absorption spectrum of a
semiconductor contains information on its band-gap. UV-vis spectroscopy is commonly
carried out in solutions, since transmission in solid materials is very low. The actual
Chapter 2. Experimental and theoretical methods
54
absorption coefficient cannot be measured directly. The Kubelka Munk (KM) remission
function F(R) is used to represent the absorption coefficient α, and is calculated from the
reflectance of the solid sample, with s ‒ the scattering coefficient according to:
2
1( )
2
RF R
R s
(2.36)
The variation in absorption coefficient α as a function of photon energy can be fitted to a
power law relationship (Equation 2.37), where B is a constant, Eg is the band gap energy and
n takes on a value that depends on the nature of the transition; values of n = 1/2 and n = 2
result in a linear fit for direct and indirect transitions respectively.162
( ) ,n
g gE E B E E E E (2.37)
Equation 2.37 is valid only for photon energies greater or equal to the band gap energy. The
value of the band gap energy is obtained as x axis intercept from a plot of (Eα(E))n vs E.
The UV-vis study of the Sr2SnO4 related materials shown in this thesis (Section 5.8)
deals with the direct band gap determination. The values of the band gaps were determined
from the plots (F(R) × E)1/2
vs E. The linear part of the plot was fitted to straight line using
Origin while the linear equation parameter was calculated. The value of the direct band gap is
equal to the intercept of the x-axis (when y = 0). The indirect band gap values can be obtained
by a similar method by plotting (F(R) × E)2 vs E.
2.6.2 Infrared Spectroscopy
Infrared spectroscopy deals with the radiation with a longer wavelength than visible
light. The IR measurements shown in Section 5.10 used the radiation from the higher energy
near-infrared (NIR) part which is commonly measured in the region of 800 ‒ 2500 nm. IR
spectroscopy uses the fact that molecules absorb specific ‒ resonant frequencies characteristic
for their structure and functional group within the structure. IR spectroscopy of solid state
Chapter 2. Experimental and theoretical methods
55
materials includes mainly identification of OH groups. NIR spectra application is often used
for the opto-electronic characterisation of transparent materials.163
NIR spectra of Sr2SnO4 related phases were carried out using a Shimadzu SolidSpec-
3700 UV-VIS-NIR Spectrophotometer with three detectors: photomultiplier tube detector for
UV-VIS region and InGaAs and PbS detectors for near infrared region. The InGaAs detector
can be switched to PbS detector in the region of 1600 to 1800 nm. Although the use of all of
the three detectors provides smooth spectra over a wide range, a noise represented by a bump
of the spectra is usually accompanied the detector switchover. Thus, it is advised to collect
the spectra in various wavelength ranges in order not to mis-interpret the spectra peaks.
2.7 Solid state NMR technique164,165
Solid state NMR spectroscopy is a technique providing information about the
structure of diamagnetic materials and about the dynamics of processes occurring within the
studied structures. Unlike the X-ray diffraction methods, solid state NMR can be applied for a
study both of disordered (melts and colloid gels) and ordered single crystals materials.
Solution NMR spectra consist of a series of very sharp transitions which are caused by
averaging of anisotropic NMR interactions by rapid random tumbling. Solid state NMR
spectra are on the contrary very broad due to many effects, such as anisotropic or orientation-
dependent interactions. The shape of the spectra provides additional information on
chemistry, structure and dynamics in the solid state. Determination of the relationship
between the experimental spectrum and recurring structural motifs is necessary in NMR. This
can be done by a ‘fingerprinting’ procedure when a database of NMR spectra is established
for related materials and are compared with an unknown structure. Another way of solving
the relationship is represented by ab initio theoretical calculation of the electric field gradient
at a nucleus of an atom in a known crystal structure.
Most of the NMR active nuclei in the periodic table are also available for the solid
state NMR. To produce an NMR spectrum, a nucleus must have a nuclear spin. The most
important for solid state NMR are nuclei with odd mass numbers (e.g. 29
Si, 27
Al). Another
factor, which needs to be considered, is a natural abundance of an NMR nucleus. In order to
minimize anisotropic NMR interactions several methods have been developed, e.g. Magic-
angle spinning, Dilution, Multiple-Pulse Sequences, Cross Polarization. Solid state NMR
Chapter 2. Experimental and theoretical methods
56
technique used in this thesis includes the experiments with Sn. Three NMR isotopes of Sn
exist: 119
Sn (8.5% natural abundance), 117
Sn (7.5%) and 115
Sn (0.35%), all are spin 1/2. For
practical reasons (better sensitivity, higher natural abundance combined with higher
resonance frequency) 119
Sn is used. The Sn NMR data shown in Section 5.9 were collected
and analysed by Dr. Frédéric Blanc.
2.8 Iodometric Titrations
Iodometry is a volumetric chemical analysis, where an addition of a precise amount of
standardised compound to a sample leads to the redox reaction characterised by a colour
change of a solution. Colour change is often enhanced by the addition of a small amount of
indicator. Starch solution is used to indicate the equivalence point in iodometry. Iodometric
titrations are performed to obtain the information about the relative molecular mass or
oxygen content of an oxide material as it was performed for Y0.95Sr2.05Cu1.7Co1.3O7+δ
(Section 3.3.1). Iodometry involves indirect titration of iodine and it can be described by
following reaction:
2,2 2
2 2 3 4 62 2H H O
I S O I S O
(2.38)
An excess of I− (typically in form of KI) is required to reduce the sample and to
generate I2. Prior to iodometric titration, standardisation of thiosulphate (Na2S2O3) has to be
carried out. The most common way for titration of Na2S2O3 is by using potassium iodate
(KIO3). KIO3 is a source of I2, which is then titrated by Na2S2O3 standard in acidic solution.
The titration can be sum up as follows:
2,
3 3
H H OKIO K IO
(2.39)
2,
3 2 25 6 3 3H H O
IO I H I H O
(2.40)
In every studied oxide material, reduced by this method, it is important to identify the
reduced products. The titrations performed on Y0.95Sr2.05Cu1.7Co1.3O7+δ were based on the
Chapter 2. Experimental and theoretical methods
57
reduction of Cu and Co from an unknown oxidation state to +1 and +2 respectively. The
presence of Cu in the titrated system requires an addition for the iodometric titrations. An
excess of KI is easily absorbed by CuI and needs to be released before the end point of the
titration.166
CuI interferes with the sharpness of the end point. To prevent that, KSCN is
added after the starch end point.167
The oxygen content ([O]) was calculated using following formula:
( )
1
w red
red
Os
ex s
MO O
Mm
n m
(2.41)
where [Ored] is the oxygen content of the reduced sample, Mw(red) is its molecular mass, ms is
the mass of the titrated sample, MO is the atomic weight of oxygen, and nex indicates the
number of moles of titrated I2, what is determined from:
2
s sex
c Vn
(2.42)
where Cs and Vs refers to the concentration and volume of Na2S2O3 solution used for titration.
2.9 Thermogravimetric Analysis
Thermogravimetric analysis (TGA) measures the weight of a sample as a function of
time as the temperature is increased at a controlled uniform rate.27
The sample is loaded into
a platinum or alumina pan and enclosed in a furnace. The sample mass is controlled by a
balance relative to the empty pan. The experiment can be performed at ambient air or under
different oxidising or reducing atmosphere. Monitored changes in mass are typically due to
oxidation or adsorption of gas (observed as a weight gain) while dehydratation, reduction and
decomposition cause a weight loss. Materials from this thesis (Y0.95Sr2.05Cu1.7Co1.3O7+δ and
Sr2SnO4 derivatives) were studied by TGA to inspect their stability at a define temperature
range in ambient air using a TA Thermogravimetric Analyzer TGA Q600.
Chapter 2. Experimental and theoretical methods
58
TGA analysis under CO2 or Ar atmosphere were carried out using a TA Thermogravimetric
Analyzer TGA Q500.
2.10 Dilatometry
Dilatometry measurements mentioned in this thesis (Section 3.10) were done in order
to measure the thermal expansion, which is one of the important properties of SOFCs
components. The thermal expansion coefficient α at temperature T is given by:168
/T
dL dT
L (2.43)
where L is the length of the sample at room temperature. The dilatometry measurement was
carried out using a Nietzch 402 C Dilatometer. A schematic of the instrument is shown in
Figure 2.8. A dense pellet of a 6 mm diameter is placed in alumina sample carrier. The
temperature of both the sample and furnace is monitored by the thermocouple. The change in
length whilst the sample is heated is not only due to the expansion of the sample. The
changes in length of the sample support and the pushrod are also monitored and measured.
The sample length change is corrected by the expansion of the sample support as this change
is not included in the sample expansion. The expansion results in a change in voltage,
transformed by an amplifier to DC voltage, which is proportional to the displacement.
Chapter 2. Experimental and theoretical methods
59
Figure 2.8: DIL 402 C measuring unit: 1 - tube, 2 - vacuum flange, 3 - slider knob, 4 - fan, 5 -
furnace, 6 - reversing tube with shut-off valve and clamping nut, 7 - protective tube, 8 - front panel, 9
- support, 10 - retaining nut, 11 - sample carrier with thermocouple and pushrod, 12 - sample, 13 -
stop plate. Adopted from Operating Instructions DIL 402 C.
2.11 Density Functional Theory (DFT)
Development of screening methods for materials, their properties and simulations of
processes in materials has been attracting attention in recent decades. Density Functional
Theory (DFT) based predictions represent a powerful and accurate tool to accelerate
materials discovery process by orienting experimental chemists to computationally predicted
compounds. Its advantage lies on the expression of electron-electron interactions in many
electron systems as an effective one-electron potential, which is a function of the electron
density only.169
The basic problem of describing the studied materials falls on the presence of many
electrons. The Hamiltonian operator corresponding to the total energy of the system can be
expressed:
22 22 2
, , ,
1 1ˆ2 2 2 2
I JIi I
i i I i j I I Ji I I Ji j
Z Z eZ e eH
m r R M R Rr r
(2.44)
Chapter 2. Experimental and theoretical methods
60
An approximation from an early stage of quantum mechanics was suggested by Born-
Oppenheimer, whereby the nuclei were fixed and the kinetic terms of nuclei were neglected.
If we were able solve the Schrödinger equation for many electron (Equation 2.45), we could
predict the behaviour of any electronic system.
1 2 1 2ˆ ( , ,..., ) ( , ,..., )N NH r r r E r r r (2.45)
Many electron wave function Ψ(r1,r2,...,rN) is a function of 3N variables, where N is the
number of electron. The construction of many-electron wave function is very demanding
problem due to the complexity of the system. Three main approaches of many-body problem
can be applied. Independent electron approximation represented by Hartree-Fock, statistical
Quantum Monte Carlo method and DFT – considering the electron density instead of the
wave function. Since the theoretical work of this thesis is based on DFT calculations
(performed using Vienna Ab Initio Simulation Package – VASP) this section is concerned
about the basics of DFT, its modifications and applications.
2.11.1 The energy functional
The basic proof that the density of electron charge is a unique description of a system
and that it corresponds to the energy minimum was outlined by Hohenberg and Kohn.170
When we have a system of N electrons in the ground state of an atom, a molecule or a solid,
the ground state is characterized by the positions of the nuclei, their potential Vext(ri)
including the electrostatic potential and some other scalar components, and the number of
electrons N. The Hamiltonian is then expressed:
0ˆ ˆ ˆ( ) ( )ee ext ext i
i
H T V V H V r (2.46)
Where, the operator 0H contains the kinetic energy operator T and the electron-electron
interaction potential Vee. If the external potential is fixed, then a number of N electrons leads
to a unique wave function Ψ of the system, which results in a unique density of charge ρ(r).
Chapter 2. Experimental and theoretical methods
61
The main statement of the first Hohenberg-Kohn (HK) theorem states the reverse of
this finding, that the potential Vext(r) and the many-electron wave function Ψ are uniquely
determined by the density of charge ρ(r). The proof is based on the opposite statement ‒ the
assumption that two wave functions leads to the same density of charge resulting in a
contradiction. Since the opposite statement is evidently wrong, the original must be correct.
The second theorem proves that the ground state density is not only unique, but also the only
one which minimizes the total energy. This can be proven in two steps. First, the total energy
is written as a functional of density (E), since the density is unique.171
ˆ ˆ ˆee ext
ee ext
E H E T V V d
T V V
r r r r
r r r
(2.47)
The property that the ground state is the state of minimum energy is applied for the second
step. Hence, if we have two different densities, one of which corresponds to the ground state
density ρ(r) when the other ρ(rʹ) does not correspond. The inequality is as follows:
ˆ ˆ' '
' ' ' 'ee ext
E H H
ETE V V
r
r r r r r
(2.48)
This gives a formula to find the ground state density in a numerical procedure: to start with a
trial density, to evaluate the total energy and to change the density until a minimum is
reached.
2.11.2 Kohn-Sham equations
An early attempt to construct the functionals before the HK theorem was the Thomas-
Fermi model.172,173
It does not bring an accurate solution since it misses essential features of a
physical system in condensed matter.174
For better understanding of HK theorem, an extended
system with non-interacting electron is considered. In this system we deal only with the
kinetic energy and the external potential (μ) and the Euler-Lagrange equation171
of the
problem takes the simply form:
Chapter 2. Experimental and theoretical methods
62
00ext
TV
r
r (2.49)
Notice, that the kinetic energy, T0, is the kinetic energy of non-interacting electrons and it is
different from the kinetic energy of interacting electrons T. The many electrons wave
function in a non-interacting electron gas is described exactly. If every local orbital λ is
occupied by two electrons, the kinetic energy is given by a sum over the same orbital171
:
2/2
*
0
1
22
n
T dm
r r r (2.50)
The single electron orbitals are obtained by solving the single particle Schrödinger equation:
2
2V
m
r r r (2.51)
After addressing the problem of the kinetic energy functional and the functional of
electron-electron interactions, the main idea of Kohn and Sham was then to formulate the
problem of interacting electrons in exactly the same way by just changing the potential:
00eff
Tv
r
r (2.52)
The effective potential (υeff) counts all missing parts of the electron interactions in.
0,
e
eff
e TVV
Tv
r r
r r r (2.53)
The charge density is expressed by:
/2
2
1
2n
r r (2.54)
Chapter 2. Experimental and theoretical methods
63
If the effective potential is described, a comprehensive solution for the problem can be found.
Thus the focus of finding the ground state density is changed to finding the universal
effective potential, which can only be approached by various approximations.
2.11.3 Exchange Correlation Functionals
An important feature about the Kohn-Sham theory is that fact that the theory is not
based on the density alone. The density is obtained from the density of single electron states.
The total energy is given by171
:
* *
0, , H xcE T E E V d r r r (2.55)
The exchange correlation energy (Exc[ρ] represents the difference between the energy of
interacting electrons (first bracket on the right side of Equation 2.56) and the energy of
electrons interacting only via their Coulomb interaction (second bracket):
0xc ee HE V E T T (2.56)
Finally, after expressing Hartree energy (EH[ρ]) and the energy due to electron-ion
interactions, the exchange correlation energy is given by:
*
;
' ' '
xc
xc xc
x
xc
x
xc c
c
EE
V
Ed V
V d
V
r r r rr
r r r r r r rr
(2.57)
Thus, the effective potential is expressed as follows:
eff ext H xcv V eV V r r r r (2.58)
Chapter 2. Experimental and theoretical methods
64
where the exchange correlation (XC) potential is the functional derivative:
0
xc
xc ee H
EV V E T T
r
r r (2.59)
This potential depends only on the number of electrons and has to be calculated separately.
That means in every density functional simulation a table with the values of the potential for
a given charge density is looked up. Several different methods have been developed in order
to calculate XC potential.
Local Density Approximation (LDA)175
functional have been mostly used in early
stage of DFT calculations. LDA is based on an assumption that the exchange and correlation
potentials depend only on the value of the charge density at a specific point of the system.
The exchange-correlation energy for LDA can be given by:
3 homLDA
xc xcE n d rn n r r (2.60)
where εxchom
is the XC energy per electron in a homogeneous electron gas with the
corresponding electron density n(r). LDA tends to overestimate the bond strength in solids.
The calculated lattice parameters are too small, cohesive energies are overestimated, and
energy gaps in semiconductors and insulators are vastly underestimated.
The Generalised-Gradient Approximation (GGA)175
takes the value of the density at
each point as well as the magnitude of the gradient of the density.
3 homGGA
xc xc xcE n d r n F s r r (2.61)
with an additional term Fxcs(r) compared to LDA which indicates (with the term s(r)) how
the electron density gradient varies. The GGA makes improvement over LDA for many
cases. It corrects the overbinding tendency in LDA, with a certain overcorrection in few
situations. Other equilibrium properties that are sensitive to lattice constant such as phonon
frequencies, bulk modulus, and magnetic moments are sometimes also over-corrected by
GGA.176
Chapter 2. Experimental and theoretical methods
65
2.11.4 Pseudo potentials177
As we have systems with increasing number of electrons (e.g. transition metals), the
computational cost is bigger. A method commonly used in DFT is to create a pseudo
potential. This involved typically two approximations. Firstly, the core electrons are treated
as a frozen core assuming that the interactions of core electrons with the surrounding
chemical environment are negligible and thus only the valence electrons interact with the
chemical environment. Core-electron wave functions are localised. For the plane wave basis
set, the number of planes would need to be very large (> 106) to describe localised core-
electrons. To make the plane wave more feasible, pseudo potential needs to be applied.
A pseudo potential replaces nucleus and core electrons by a fixed effective potential.
Only valence electrons are taken into account in the calculations. The core state is removed
from the spectrum (see schematic in Figure 2.9). In practise the core electrons wave functions
are generated from calculations on the isolated atoms, libraries of which are provided by DFT
packages. In some cases in order to improve the accuracy even some of the sub-valent
orbitals are treated as if they were valence. Pseudo potential generation can be summed into
three main steps:
1) Calculate exact all electron wave functions for a reference atom.
2) Replace the exact wave function by a node less pseudo-wave function.
3) Invert Schrödinger equation to obtain the pseudo potential.
Pseudo potentials must conserve exactly the scattering properties of the original
atomic configuration. They must be generated with the same functional that will be later used
in DFT calculations. It should also be noted that the choice of pseudo potential is not unique;
there is a lot of freedom to construct them. A pseudo potential is called soft when a few plane
waves are needed. If a pseudo potential can be used in various environment (molecule, solid,
metal) it is called transferable. Creation of a good pseudo potential must meet requirements
for both soft and transferable pseudo potentials. A variety of different methods using pseudo
potential is available.178-181
A related method used within this thesis is the Projector
Augmented Wave method (PAW)178
implemented in VASP.
Chapter 2. Experimental and theoretical methods
66
Figure 2.9: Schematic of pseudo potential concept taken from.182
The pseudo wave function and
potential are labelled in red; the all electron wave function and potential are in blue.
2.11.5 DFT+U
DFT+U represent one of the recent and essential developments in DFT.183
An
inherent problem of the DFT methods comes from the Coulomb term in the Kohn-Sham
equations, in which an electron interacts with the total electron density. That results in the
electron interacting with itself, what is known as the self interaction error (SIE). The SIE
causes localised electron orbitals to be destabilised and so DFT often results in orbitals being
spread out spatially in order to minimise self interaction. The DFT+U includes the orbital
dependence of the self-energy operators which are missing from the Kohn-Sham potential,
neglecting the fine details of the spatial variation of the Coulomb potential. In order to obtain
correct computation of band structures and total energies of systems that should have
localised orbitals (such as in semiconductors with partially filled d and f orbitals), corrections
for SIE are required. In DFT+U for each atom we apply a numerical parameter Ueff to
specified orbitals (d orbitals in this thesis) with Ueff defined as177
:
Chapter 2. Experimental and theoretical methods
67
effU U J (2.62)
Where U and J are parameterised. Ueff is a numerical parameter which requires fitting to a
relevant material, which is commonly done by varying the value U, with J remaining fixed.
The parameter sets used in this theses were taken from published related DFT works (given
in relevant chapter) in order to determine reaction energies to form perovskite related oxides
and oxygen gas where available.
3 Synthesis and characterization of Y1−ySr2+yCu3−xCoxO7+δ
3.1 Introduction
Previous research in the area of perovskite cuprates was focused on the
superconductor YBa2Cu3O7-δ. It has been shown as a possible cathode material for SOFCs
(high electronic conductivity and mixed conductivity at temperatures above 300 °C).68,184-186
However at temperatures above 900 °C and at low current densities materials, react with YSZ
electrolyte and decompose.68
This was indicated by the presence of significant amounts of
Y2BaCuO5 and BaZrO3 phases. Despite the insufficient stability, this material showed quite
good electronic and oxide ion conductivity.187,188
There is also electrochemical
decomposition observed at low current densities (20 mA cm−2
).68
The instability of
YBa2Cu3O7-δ was explained by electrostatic effects, observed as an increase in the Cu(1)-
O(1) distance.189
In order to improve thermal stability, copper can be partially replaced by
other metal cations with different coordination preferences (M = Fe, Co, Ga, Al, etc).134-137
Replacing Ba with Sr and partial substitution of copper with Co increased the stability of the
material.189
Figure 3.1: Structure of YSr2Cu2CoO7+δ adopted from,190
with Y (grey-blue) and Sr (green) on the A-
sites, with the square pyramids of Cu (orange) and the tetrahedral Co ordering (dark blue) on B-site:
a) view along c-axis; b) rotated view around a- and b-axis.
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
69
The orthorhombic structure of YSr2Cu2CoO7+δ (Figure 3.1) was first reported by
Huang et al.190
and later work showed that it exhibits a transformation to a tetragonal
structure at high temperature (above 800 °C).70
Its structure may be described as an ordered
perovskite, with Y, Sr, Co and Cu layers ordered along the a-axis. CuO5 units adopt a corner
sharing, square based pyramidal geometry, whilst corner sharing CoO4 tetrahedra adopt a
disordered arrangement with both oxygen atoms adopting a split site model. Strontium and
yttrium are also ordered, due to their different local coordination preferences. The disordered
CoO4 tetrahedral layers are between the two strontium layers, whilst the CuO layers sit either
side of the yttrium containing layers. This structure has two copper sites: the one (Cu2) is
associated with the superconducting and a second site (Cu1) is connecting Cu2 layers along
the c-axis. Previous structural works based on high-resolution transmission-electron
microscopy demonstrated that all the Co atoms are located in Cu1 sites.191,192
The fact that
Co2+/3+
prefers tetrahedral coordination to square planar leads to cooperative oxygen ordering
and therefore to an expanded unit cell.
Previous research on YSr2Cu3−xCoxO7+δ and related phases was focused on structural
studies.189,192-194
The most detailed neutron diffraction study193,194
confirmed previous
structural works192,195
and compared them with superconductor YBa2Cu3O7−δ material.
Reported studies on Co-containing phases included materials of x = 1. The structure of Co-
1212 phase was found to be best described in Ima2 space group.193
The overall oxygen
content 7.01(2) was determined from refined oxygen occupancies from neutron powder
diffraction (NPD) data.193
Later studies on triple copper perovskites (M = Fe, Co) were
focused on the materials as a potential SOFCs cathode,69,70
showing promising conductivity
values, for Fe σ550°C = 35 S cm−1
and for Co σ900°C = 15 S cm−1
, but also significant
compatibility problems with common electrolytes (reactivity with LSGM and ceria-based
electrolytes).70
The compatibility tests were carried out at the temperatures between 900 and
1000 °C with the reactions in all cases except for the ceria-based electrolytes at 900 °C.
Thermogravimetric analysis showed that no oxygen loss occurred in the 25 – 900 °C
temperature range under both air and nitrogen atmospheres.70
The work reported in this chapter is focused on Y1−ySr2+yCu3−xCoxO7+δ compounds
including synthesis, structural and electrochemical characterization of a series of materials
with different Co doping levels. Increasing the cobalt content improves conductivity
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
70
properties of studied phases by introducing electronic carriers.17
Cobalt also plays important
role in catalysis which is necessary for the operation of SOFC (see Section 1.6.1).17
The
beginning of the work includes structural characterisation of these materials by powder
diffraction, followed by thermal and compatibility studies and comparison with previously
reported works of YSr2Cu2CoO7+δ phase. Since the work is oriented towards SOFC
application, conductivity measurements of the symmetrical cells play an important role and
are presented at the end of this chapter, together with CO2 stability tests and thermal
expansion studies.
3.2 Synthesis
Y1−ySr2−yCu3−xCoxO7+δ samples with nominal values of x = 1.00, 1.10, 1.20, 1.25,
1.30, 1.40, and 1.50 and of y = 0, 0.03, and 0.05 were prepared by solid state synthesis.
Stoichiometric mixtures of pre-dried Y2O3 (99.99%), SrCO3 (99.99%), Co3O4 (99.9%) and
CuO (99.995%) – all purchased from Alfa Aesar – were ground by hand and heated in
alumina crucibles to 1050 °C in air at a rate of 5 °C min−1
and then immediately cooled down
to room temperature at a rate of 5 °C min−1
. The samples were then re-ground and pelletised
before re-firing under the same conditions. In total, the samples were re-fired four times until
phase pure materials were obtained.
The initial syntheses followed the literature synthesis conditions for the parent phase
YSr2Cu2CoO7+δ (1000 °C for 16 h).69
Attempts to synthesise YSr2Cu3−xCoxO7+δ with larger
cobalt contents yielded the yttrium containing impurity phases Y2Cu2O5 and
Y2SrCu0.6Co1.4O6.5. The presence of these impurity phases was eliminated by controlling the
Y:Sr ratio to compensate partially for the increasing positive charge on the B-site, yielding
compositions of type Y1−ySr2+yCu3−xCoxO7+δ. In the range x = 1.00 – 1.25, phase pure samples
were obtained for y = 0.03, and at x = 1.30 a single phase sample was obtained with y = 0.05.
For x > 1.30 single-phase samples could not be obtained. At x = 1.0, samples were
synthesised with y = 0.0 and 0.03.
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
71
3.3 Structural characterization
3.3.1 Laboratory P-XRD data
Phase purity was verified by powder XRD (PXRD) collected at room temperature
using a Panalytical X'pert Pro diffractometer using Co Kα1 radiation in Bragg-Brentano
geometry (measured over a 2θ range of 5 – 120°, step size 0.0167°, time per step 3 s).
Rietveld-quality PXRD data was collected on a Bruker D8 Advance in Debye-Scherrer
geometry using a 0.2 mm glass capillary with Mo Kα1 (λ = 0.7093 Å) radiation (d-spacing
range 0.58 – 20.25 Å). Structural parameters were refined by the Rietveld method150,151
using
the Topas academic software.152
Samples of YSr2Cu3−xCoxO7+δ with x = 1.00 and 1.10 were
prepared as single phases, with all diffraction lines indexed on the basis of a
6ap × ap√2 × ap√2 body-centred unit cell. The observed decrease in a lattice parameter
(a = 22.7691(5) Å for x = 1.00 and a = 22.7563(5) Å for x = 1.10) suggested that
compositions with different cobalt contents are accessible. Rietveld refinement of
Y0.97Sr2.03Cu2CoO7+δ is shown in Figure 3.2. A decrease of lattice parameters of
YSr2Cu2CoO7+δ compared to Y0.97Sr2.03Cu2CoO7+δ was observed (Figure 3.3). The higher Co
content and higher Sr content is required to obtain single phase samples. With increasing Co
content, there is a decrease in cell parameters, in agreement with Vegard’s law (observed for
the constant Y0.97Sr2.03 ratio for x = 1 to 1.25). Figure 3.3 and Table 3.2 show the evolution
of cell parameters with cobalt content of Y1−ySr2−yCu3−xCoxO7+δ samples (x = 1 to 1.5 and y =
0, 0.03 and 0.05). The values of the parameters were obtained from Rietveld refinements of
the PXDR data collected with lanthanum hexaboride, LaB6, used as a standard. The presence
of Y2Cu2O5 and Y2SrCu0.6Co1.4O6.5 impurity phases was the limiting factor in obtaining
single phase Y1−ySr2+yCu3−xCoxO7+δ for x >1.3 (Figure 3.4).
Rietveld refinement of the room temperature structure was carried out using
laboratory PXRD data. The ambient temperature PXRD pattern was indexed to an
orthorhombic cell of dimensions 22.744(1) × 5.4210(3) × 5.4430(3) Å, with systematic
absences consistent with Imcm or Ima2 symmetry. The structure of YSr2Cu2CoO7 published
by Huang et al.190
with orientationally disordered CoO4 tetrahedra in space group Imcm was
selected as a suitable starting point for Rietveld refinement. Cell parameters, peak shape,
zero error and 18 background parameters were refined. The Cu:Co ratio on the Cu site was
fixed at 0.85:0.15, in line with the nominal composition, due to the lack of contrast between
Co and Cu using XRD data alone. The individual Y, Sr, Co and Cu coordinates and isotropic
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
72
temperature parameters were refined. The oxygen coordinates were fixed at the published
values and a common isotropic temperature factor was refined. This model provided a good
fit to the data, as displayed in Figure 3.2. The refined structural parameters are listed in Table
3.1.
Figure 3.2: Rietveld refinement of Y0.95Sr2.05Cu1.7Co1.3O7+δ to laboratory PXRD data at room
temperature based on the published model of YSr2Cu2CoO7 in Imcm, Rwp = 1.51%.
Table 3.1: Structural parameters obtained for Y0.95Sr2.05Cu1.7Co1.3O7 following Rietveld refinement
using laboratory PXRD data. Space group Imcm. a = 22.744(1) Å, b = 5.4210(3) Å, c = 5.4430(3) Å.
Rwp = 1.51 %, χ = 1.64.
Site Wycoff
Site
x y z occ Beq (Å2)
Y1 4a 0 0 0 0.95 0.5(2)
Sr1 4a 0 0 0 0.05 0.5(2)
Sr2 8h 0.3452(2) 0.000(5) 0 1 1.7(1)
Cu1 8h 0.4258(3) 0.494(3) 0 0.85 1.4(1)
Co1 8h 0.4258(3) 0.494(3) 0 0.15 1.4(1)
Co2 4e 0.25 0.526(6) 0 1 2.9(3)
O1a 8i 0.25 0.614 0.386 0.39 1.0(2)
O1b 8i 0.25 0.386 0.386 0.11 1.0(2)
O2 8g 0.4327 0.75 0.25 1 1.0(2)
O3 8g 0.4352 0.25 0.75 1 1.0(2)
O4 8g 0.3255 0.473 0 1 1.0(2)
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
73
Figure 3.3: Lattice parameter evolution of Y1−ySr2+yCu3−xCoxO7+δ (samples with YSr2(○),
Y0.97Sr2.03(∆), and Y0.95Sr2.05(□) ratio) as a function of Co doping level (x): a) lattice parameter a; b)
lattice parameters b and c; c) cell volume; d) (b+c)/2. Errors bars are within the symbol size.
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
74
Figure 3.4: PXRD pattern of Y0.95Sr2.05Cu1.5Co1.5O7+δ, with the main reflections of Y2SrCu0.6Co1.4O6.5
(*) and Y2O3 (+) impurity phases.
Table 3.2: Lattice parameters and cell volume of Y1−ySr2+yCu3−xCoxO7+δ plotted against different Co
doping level x (x = 1 to 1.5, for y = 0, 0.03 and 0.05). The values of the parameters were obtained after
Rietveld refinement and compared with previously reported data on x = 1 material.193
x y a (Å) b (Å) c (Å) V (Å3)
1-reported193
0 22.7987(2) 5.45150(5) 5.40890(5) 672.26(1)
1 0 22.7618(4) 5.4525(1) 5.4088(2) 671.28(3)
1 0.03 22.7691(5) 5.4545(1) 5.4099(1) 671.88(3)
1.1 0.03 22.7563(5) 5.4567(1) 5.4135(1) 672.22(2)
1.2 0.03 22.7405(5) 5.4475(1) 5.4156(1) 670.88(2)
1.25 0.03 22.7338(4) 5.4425(1) 5.4169(1) 670.22(2)
1.3 0.05 22.7294(4) 5.4394(1) 5.4179(1) 669.83(2)
1.4 0.05 22.7362(5) 5.4380(1) 5.4174(1) 669.81(2)
1.5 0.05 22.743(1) 5.4460(3) 5.4190(2) 671.20(5)
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
75
Iodometric titrations were carried out to determine the oxygen content. Samples of
Y0.95Sr2.05Cu1.7Co1.3O7+δ were titrated by standard sodium thiosulfate solution (0.1 M),
standardized by potassium iodate (99.995% purchased from Sigma Aldrich). For the
titrations, an approximate 0.05 g of the Y0.95Sr2.05Cu1.7Co1.3O7+δ sample was weighed out and
dissolved in a mixture of 10 ml distilled water and 10 ml of concentrated HCl. The solution
was kept under argon atmosphere and constantly stirred while 1.0 g of KI was added and the
solution was then titrated by thiosulfate standard solution using starch indicator. A 5 ml
portion of KSCN was added after the starch end point (see Section 2.8). The same procedure
was done four times and an average (mean) titre value was obtained. Details about the sample
mass, volume of used standard solution and calculated oxygen content are shown in Table
3.3. The oxygen content was determined by iodometric titration and calculated to be
Y0.95Sr2.05Cu1.7Co1.3O7.02(3). This is in good agreement with literature values for related
phases, e.g. an oxygen content of YSr2Cu2CoO7.03(4) was determined for the parent non-
substituted phase by Sansom et al.69
using thermogravimetric methods.
Table 3.3: Iodometry determination of Y0.95Sr2.05Cu1.7Co1.3O7+δ sample.
Sample Sample weight (g) volume (Na2S2O3) (ml) O content
1 0.0497 2.2 6.93(3)
2 0.0508 2.5 7.08(3)
3 0.0496 2.4 7.06(3)
4 0.0516 2.4 7.00(3)
average 0.0504 2.4 7.02(3)
3.3.2 Neutron Powder Diffraction data
In order to determine the structure at typical SOFC operating temperatures, time-of-
flight powder neutron diffraction data were collected on the high-intensity, medium-
resolution GEM diffractometer at ISIS at 740 °C. Data from banks 2, 3, 4, 5 and 6 were used
simultaneously in the refinement. The refined room temperature structure from laboratory
PXRD was used as the starting model in the refinement. Cell parameters, atomic coordinates
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
76
and anisotropic temperature parameters, along with 12 background parameters, a scale factor
and peak shape were refined. The composition of the Y1 site was fixed at a nominal value of
Y0.95Sr0.05 due to poor scattering contrast between Y (7.75 fm) and Sr (7.02 fm). However
there is good contrast between Co (2.49 fm) and Cu (7.718 fm) and the occupancies of their
crystallographic sites were refined to allow Cu/Co sharing, whilst constraining to a total
occupancy of 1 at each site. In the final stages of the refinement all thermal displacement
parameters were allowed to refine anisotropically. The partially-occupied O1a and O1b sites
were constrained to a common thermal displacement ellipsoid and their site occupancies
refined with a constraint to maintain the overall composition. The refined Co2 and O4 sites
show elongated thermal displacement ellipsoids consistent with orientational disorder of the
CoO4 tetrahedra. Split-site models for both the Cu1 and O4 sites were tested (as applied to
YSr2Cu2CoO7 by Babu et al.)192
by moving both atoms from the ..m plane (8h) to a general
position (16j) with an occupancy of 0.5. However the coordinates of both atoms were found
to refine to within 1 estimated standard deviation of their original 8h positions, with no
significant change to the fit, so this site splitting was not employed in the final refinement.
No Cu was found to occupy the tetrahedral cobalt site and its partial occupancy was
subsequently fixed to zero in the final stages of the refinement. A significant fraction of Co
on the square-pyramidal Cu sites was found, with a refined Co occupancy of 0.163(4), in
agreement with the nominal cobalt content of 1.30 per formula unit. The partial occupancy of
the Cu site by Co is a key structural difference between this non-stoichiometric material and
the stoichiometric parent phase, which may play an important part in improving the
material’s properties with respect to the parent material. The final refined structural
parameters are presented in Table 3.4-3.5 and corresponding Rietveld fits are shown in
Figure 3.5. The refined structure is illustrated in Figure 3.6. Bond lengths (Table 3.6) and O-
Cu-O angles (Table 3.7) are in good agreement with those published by Huang et al.190
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
77
Figure 3.5: Fits to GEM neutron powder diffraction data at 740 °C using Imcm (a = 22.9740(2) Å,
b = 5.51692(6) Å, c =5.47661(5) Å). Rwp = 2.12%: a) Fit to bank 3; b) Fit to bank 4; c) Fit to bank 5;
d) Fit to bank 6.
Table 3.4: Refined parameters obtained from Rietveld refinement of Y0.95Sr2.05Cu1.7Co1.3O7+δ using
GEM 740 °C neutron powder diffraction data. Space group Imcm (a = 22.9740(2) Å,
b = 5.51692(6) Å, c = 5.47661(6) Å).
Site Wycoff Site x y z Occ
Y1 4a 0 0 0 0.95
Sr1 4a 0 0 0 0.05
Sr2 8h 0.34813(4) 0.0069(4) 0 1
Cu1 8h 0.42638(5) 0.500(5) 0 0.836(3)
Co1 8h 0.42638(5) 0.500(5) 0 0.163(3)
Co2 4e 0.25 0.558(1) 0 1
O1a 4e 0.25 0.6210(7) 0.4006(9) 0.414(3)
O1b 4b 0.25 0.347(4) 0.312(4) 0.086(3)
O2 8g 0.4351(1) 0.75 0.25 1
O3 8g 0.43631(9) 0.25 0.75 1
O4 8h 0.32453(7) 0.4639(5) 0 1
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
78
Table 3.5: Anisotropic temperature factor coefficients obtained from the Rietveld refinement of
Y0.95Sr2.05Cu1.7Co1.3O7 using powder neutron diffraction data collected at 740 °C.
Site U11 U22 U33 U12 U13 U23
Y11 0.0195(9) 0.0095(9) 0.021(1) 0 0 0
Sr1 0.0195(9) 0.0095(9) 0.021(1) 0 0 0
Sr2 0.0237(8) 0.034(1) 0.031(1) 0.006(2) 0.017(3) 0.0(1)
Cu1/Co1 0.0274(8) 0.0118(7) 0.0095(8) -0.002(1) 0.00(1) 0.00 (1)
Co2 0.018(3) 0.027(6) 0.072(6) 0 0 0
O1a 0.0371(2) 0.031(2) 0.012(3) 0 0 0
O1b 0.0371(2) 0.031(2) 0.012(3) 0 0 0
O2 0.057(3) 0.018(1) 0.015(2) 0.00(5) 0.00(9) -0.005(2)
O3 0.027(2) 0.016(1) 0.016(2) 0.0(3) 0.0(2) -0.008(1)
O4 0.0189(9) 0.042(2) 0.065(2) 0.010(2) 0.010(5) 0.043(3)
Table 3.6: Selected bond lengths obtained from Rietveld refinement of Y0.95Sr2.05Cu1.7Co1.3O7+δ.
Y-O (Å) Sr-O (Å) Cu-O (Å) Co-O (Å)
Y-O(2):
2.4488(15)
Sr1-O(2): 2.4488(15) Cu-O(2): 1.9536(3) Co1-O(2): 1.9536(3)
Y-O(3):
2.4325(13)
Sr1-O(3): 2.4325(13) Cu-O(3): 1.9565(3) Co1-O(3): 1.9565(3)
Sr2-O(1a): 2.4177(19) Cu-O(4): 2.350(2) Co1-O(4): 2.350(2)
Sr2-O(1c): 2.576(7) Co2-O(1a): 1.848(8)
Sr2-O(2): 2.807(2) Co2-O(1a): 2.229(4)
Sr-O(4): 2.761(4) Co2-O(1c): 2.080(17)
Sr-O(4): 2.574(3) Co2-O(1c): 2.493(15)
Sr-O(4): 3.048(3) Co2-O(4): 1.795(3)
Table 3.7: Selected bond angles obtained from Rietveld refinement of Y0.95Sr2.05Cu1.7Co1.3O7+δ.
O-Co-O (°) O-Cu-O (°)
O(1a)-Co-O(1a): 106.54(19) O(2)-Cu1-O(2): 88.990(19)
O(1c)-Co-O(1c): 80.74(18) O(2)-Cu1-O(3): 89.741(12), 167.53(12)
O(1a)-Co-O(4): 106.54(19) O(3)-Cu1-O(3): 88.822(18)
O(1c)-Co-O(4): 106.0(4) O(4)-Cu1-O(2): 93.06(8)
O(4)-Co-O(4): 145.4(4) O(4)-Cu1-O(2): 99.39(9)
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
79
Figure 3.6: Refined structure of Y0.95Sr2.05Cu1.7Co1.3O7+δ at 740 °C: a) orthorhombic unit cell with
atoms plotted as 99% displacement ellipsoids; b) the tetrahedral CoO4 layer viewed along the stacking
axis with atoms plotted isotropically to allow representation of oxide occupancies by pie-charts.
Grey = Y3+
, green = Sr2+
, blue = Co2+
, orange = Cu2+
and red = O2-
.
3.4 DC conductivity measurements
DC conductivity data were collected on single phase materials of
Y1−xSr2+xCu3−yCoyO7+δ (1≤ x ≤ 1.3, y = 0.03 and 0.05). In order to obtain dense pellets for the
measurements, single phase samples were ball-milled using a FRITSCH Pulverizette 7
planetary ball mill for 12 h in ethanol using zirconium oxide balls. The resulting fine powder
was mixed with a 2% polyvinyl alcohol (PVA) solution and dried at 80 °C overnight. The
samples were then pressed into pellets using cold isostatic pressing (Section 2.4.3) and
sintered in alumina crucibles under ambient air in a box furnace at 1050 °C with a heating
rate of 5 °C min−1
(with no dwell time at temperature). This resulted in pellets with densities
greater than 93% of the theoretical density, which was checked using Archimedes’ Principle
balance (Section 2.4.4).
The DC conductivity data were collected using the standard four probe technique
(Section 2.4.2) on a bar with approximate dimensions of 2 × 2 × 13 mm3. Gold paste was
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
80
used to bond the gold wires in a four-in-a-line contact geometry. DC conductivity data were
collected (using a Keithley 617 Programmable Electrometer) in air as a function of
temperature (300 – 800 °C) at each 50 ° with a 90 min equilibrium time both on heating and
cooling. Sample purity before and after the measurements was confirmed by using X-ray
diffraction.
The data collection of each of the samples started at 600 °C. Then the samples were
heated to 800 °C cooled down to 300 °C and re-heated to 800 °C. The plots with DC
conductivity data of Y1-ySr2+yCu3-xCoxO7+δ (1 ≤ x ≤ 1.3, y = 0.03 and 0.05) are shown in
Figure 3.7. The materials showed semiconducting behaviour, decreasing the resistivity by
increasing temperature. Increasing the Co doping level increases the measured DC
conductivity. The Arrhenius plot of conductivity for YSr2Cu2CoO7+δ (x = 1 sample) shows an
abrupt increase above 800 °C which has been attributed to an orthorhombic-tetragonal phase
transition.69
A significant increase in conductivity above 700 °C is observed for all of the
measured samples (Figure 3.7b). Y0.95Sr2.05Cu1.7Co1.3O7+δ exhibited the highest conductivity
with a value of 191 S cm−1
at 800 °C. This is one order of magnitude greater than the value
obtained for Y0.97Sr2.03Cu2CoO7+δ (20.7 S cm−1
at 800 °C) and the value reported for the
parent, YSr2Cu2CoO7+δ material in previous work (15 S cm−1
at 800 °C).69
The activation
energies (Table 3.8) were obtained from a linear fit of the Arrhenius plots in the 300 to
600 °C temperature region. The activation energy increases with increasing Co doping level,
from a value of 0.14 eV for x = 1 to 0.26 eV for x = 1.3. This results in a more modest
increase in conductivity with Co doping at lower temperatures. At 650 °C, the DC
conductivity of Y0.95Sr2.05Cu1.7Co1.3O7+δ is four times higher than in Y0.97Sr2.03Cu2CoO7+δ.
The PXRD patterns of Y0.95Sr2.05Cu1.7Co1.3O7+δ collected before and after the DC
measurement are shown in Figure 3.8.
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
81
Figure 3.7: DC conductivity of Y1−ySr2+yCu3−xCoxO7+δ. a) linear temperature scale of DC conductivity
as a function of Co doping level; b) logarithmic scale of DC conductivity as a function of temperature.
Table 3.8: Values of DC conductivity at 800 °C of Y1-ySr2+yCu3-xCoxO7+δ materials with the values of
activation energies (Ea) compared with reported value for YSr2Cu2CoO7+δ.69
x DC conductivity at 800 °C (S cm−1
) Ea (eV)
1 – reported69
15 –
1 20.7 0.14
1.1 26.8 0.18
1.2 29.3 0.25
1.25 36.3 0.24
1.3 191 0.26
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
82
Figure 3.8: PXRD of Y0.95Sr2.05Cu1.7Co1.3O7+δ – before and after DC measurement with the mark of
the most intense reflection of triple perovskite phase and gold.
3.5 Thermal stability
Since thermal instability of the related cuprate materials was a significant
problem,68,70
thermal stability behaviour of the studied ap materials was inspected. Thermal
stability tests of the most conducting material Y0.95Sr2.05Cu1.7Co1.3O7+δ were investigated for
different dwelling times (6 h and 24 h) at various temperatures (900, 950, and 1000 °C). Prior
to heating, 0.1 g of Y0.95Sr2.05Cu1.7Co1.3O7+δ material was weighed out and pressed into 6 mm
diameter pellets. The samples were sintered in alumina crucibles in a box furnace at ambient
air with heating rate 5 °C min−1
. After the tests each of the samples was re-ground manually
using a pestle and a mortar. PXRD patterns collected after the annealing (Figure 3.9 and
Figure 3.10) indicated that no impurity phases were present. For further investigation,
Rietveld refinements in Topas152
were done. The values of lattice parameters of the samples
after the tests are summarised and compared with as made powder in Table 3.9. The changes
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
83
of lattice parameters after the thermal stability tests were negligible compared to the values
obtained from as made Y0.95Sr2.05Cu1.7Co1.3O7+δ (Figure 3.11).
Figure 3.9: PXRD data of Y0.95Sr2.05Cu1.7Co1.3O7+δ with the most intense 3ap reflections labelled. The
data were collected after the 6 h thermal stability tests carried out at various temperatures.
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
84
Figure 3.10: PXRD data of Y0.95Sr2.05Cu1.7Co1.3O7+δ with the most intense 3ap reflections labelled.
The data were collected after the 24 h thermal stability tests carried out at various temperatures.
Table 3.9: Summary of the lattice parameters of Y0.95Sr2.05Cu1.7Co1.3O7+δ obtained by Rietveld
refinement from PXRD data collected after the thermal stability tests at different temperatures
compared with as made powder.
Conditions a (Å) b (Å) c (Å) V (Å3)
As made powder 22.7210(8) 5.43220(5) 5.43140(4) 670.37(2)
900 °C, 6 h 22.7181(8) 5.4298(5) 5.4303(5) 669.8(1)
900 °C, 24 h 22.7261(8) 5.4303(6) 5.4311(5) 670.2(1)
950 °C, 6 h 22.7062(5) 5.4293(5) 5.4286(5) 669.2(1)
950 °C, 24 h 22.7131(7) 5.4305(9) 5.4304(9) 669.7(2)
1000 °C, 6 h 22.7207(3) 5.4306(8) 5.4299(8) 669.9(2)
1000 °C, 24 h 22.7210(7) 5.4300(6) 5.4304(7) 670.0(1)
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
85
Figure 3.11: Lattice parameters of Y0.95Sr2.05Cu1.7Co1.3O7+δ obtained from Rietveld refinements using
laboratory PXRD data a) to c) show the samples annealed for 6 h and d) to f) show the samples
annealed for 24 h compared to as made material Y0.95Sr2.05Cu1.7Co1.3O7+δ. a) and d) show the a cell
parameters; b) and e) show the b and c cell parameters and c) and f) show the cell volume.
Simultaneously with thermal stability tests, a TGA experiment with
Y0.95Sr2.05Cu1.7Co1.3O7+δ material was conducted. The sample was annealed in an alumina pan
with the heating rate 5 °C min−1
in air to 900 °C and cooled down to room temperature with
the same heating rate. Figure 3.12 shows the weight loss changes during the heating and
cooling process. There is no significant weight loss, which could be assigned to any
decomposition process. The small change between the 25 – 150 °C temperatures is mostly
due to the moisture taken by the material before the measurement. The PXRD data collected
after the TGA experiment shows no impurity phases.
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
86
Figure 3.12: TGA measurement of Y0.95Sr2.05Cu1.7Co1.3O7+δ material at ambient air on heating from
RT to 900 °C and cooling back to room temperature.
3.6 Chemical compatibility of Y0.95Sr2.05Cu1.7Co1.3O7+δ with electrolytes
It is important to understand the reactivity of potential cathode materials with an
electrolyte, as any reaction could create an interface containing impurities with properties that
are detrimental to the operation of the SOFC. The stability of the Y0.95Sr2.05Cu1.7Co1.3O7+δ
cathode with SDC20, GDC10, and LSGM electrolytes was tested by annealing a thoroughly
ground mixture of Y0.95Sr2.05Cu1.7Co1.3O7+δ and electrolyte at temperatures of 900, 950 and
1000 °C for one week. Compatibility tests were carried out using SOFC electrolytes of
followed properties:
SDC20, Ce0.80Sm0.20O2−x, TC Grade, surface area: 6.0 m2 g
−1, purchased from
fuellcellmaterial.com
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
87
GDC10, Gd0.10Ce0.90O2-x, TC Grade, surface area: 5.9 m2 g
−1, purchased from
fuellcellmaterial.com
LSGM, La0.9Sr0.1Ga0.8Mg0.2Oxide, 99.9%, surface area: 0.50 m2 g
−1, purchased
from PRAXAIR Surface Technologies
Prior to annealing in ambient air, mixture of cathode and electrolyte (1:1 mass ratio,
0.05 g each) was ground using a pestle and a mortar and pressed into a 6 mm pellet and put in
an alumina crucible in a box furnace. After one week annealing the pellets of the cathode and
electrolyte were cooled down to room temperature (5 °C min−1
rate). The composition and
the lattice parameters of the materials after the tests were investigated from the collected
PXRD data.
Y0.95Sr2.05Cu1.7Co1.3O7+δ reacts with LSGM at all the studied temperatures (Table 3.10)
and is unsuitable for use with this electrolyte. The reaction gave rise to the formation of
perovskite (LaCuO3, La0.6Sr0.4CoO3, LaSrCuGaO5) and Y2Cu2O5 impurities.
After annealing the SDC20 and GDC10 electrolytes with Y0.95Sr2.05Cu1.7Co1.3O7+δ, at
900 °C, PXRD patterns showed that no impurity phases were present, indicating that no
reaction between the electrolyte and Y0.95Sr2.05Cu1.7Co1.3O7+δ had occurred (as shown in
Figure 3.13). Increasing the temperature to 950 °C resulted in observation of
Y1.5Ce0.5Sr2Cu2CoO9+δ impurities in the PXRD data, indicating reaction between the
Y0.95Sr2.05Cu1.7Co1.3O7+δ material and the electrolyte. This is in agreement with previous
electrolyte compatibility studies of YSr2Cu2CoO7+δ with GDC10 which also showed the
presence of Y1.5Ce0.5Sr2Cu2CoO9+δ impurity phases after annealing at 1000 °C for 1 week.69
The compatibility test show a possibility to use Y0.95Sr2.05Cu1.7Co1.3O7+δ material with
ceria based electrolytes at lower temperatures (900 °C and less – depends on adhesion step
required for processing). Table 3.11 shows the comparison of the lattice parameters of as
made powder with the lattice parameters obtained from the Rietveld refinements of the data
collected after the compatibility tests with ceria based electrolytes. There are small changes in
volume and lattice parameters but since there were no additional phases observed after one
week annealing, these electrolytes (SDC20, GDC10) were used for further work including
AC impedance spectroscopy on symmetrical cells (see Section 3.7).
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
88
Figure 3.13: PXRD data of Y0.95Sr2.05Cu1.7Co1.3O7+δ after one week of annealing with GDC10 at 900
and 1000 °C. The reflections of the different phases are identified as follows: + for GDC10, * for
Y1.5Ce0.5Sr2Cu2CoO9+δ, † for Co3O4, and ‡ for Gd0.18Y1.82O3.
Table 3.10: Summary of the reactivity and additional phases observed after compatibility tests of
Y0.95Sr2.05Cu1.7Co1.3O7+δ with SDC20, GDC10, and LSGM respectively at various temperatures for
one week.
Temp. (°C) SDC20 GDC10 LSGM
900 No reaction No reaction SrLaGa3O7
950 Y1.5Ce0.5Sr2Cu2CoO9+δ Y1.5Ce0.5Sr2Cu2CoO9+δ SrLaGa3O7
1000 Y1.5Ce0.5Sr2Cu2CoO9+δ Y1.5Ce0.5Sr2Cu2CoO9+δ Multi phased
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
89
Table 3.11: Comparison of the lattice parameters of Y0.95Sr2.05Cu1.7Co1.3O7+δ after compatibility test
with SDC20 and GDC10 respectively with as made powder and reported YSr2Cu2CoO7+δ phase.193
Material a (Å) b (Å) c (Å) V (Å3)
+SDC20 at 900°C 22.746(7) 5.4524(7) 5.4114(6) 671.11(7)
+GDC10 at 900°C 22.653(4) 5.4500(3) 5.4169(2) 668.76(5)
Y0.95Sr2.05Cu1.7Co1.3O7+δ 22.7294(4) 5.4394(1) 5.4179(8) 669.83(2)
YSr2Cu2CoO7+δ 193
22.7987(2) 5.45150(5) 5.40890(5) 672.26(1)
3.7 AC impedance spectroscopy of Y0.95Sr2.05Cu1.7Co1.3O7+δ
3.7.1 AC impedance data at 500 ‒ 800 °C
To further Y0.95Sr2.05Cu1.7Co1.3O7+δ investigate as a cathode material for SOFCs, AC
impedance spectroscopy measurements were performed (see Section 2.5 for more details).
Since the compatibility tests showed no reaction between the potential cathode material and
ceria based electrolytes (GDC10, SDC20) the symmetrical cells were fabricated of ceria
based electrolytes and of Y0.95Sr2.05Cu1.7Co1.3O7+δ or Y0.97Sr2.03Cu2CoO7+δ cathode. Pellets of
SDC20 and GDC10 (with densities over 95%) were prepared by annealing in air at 1400 °C
for 5 h. The surface of the electrolytes was polished with SiC paper using a polishing
machine (Struers Tegramin-30). Cathode inks were produced from powders of
Y1−ySr2+yCu3−xCoxO7+δ ball milled with ethanol for 12 h. Following evaporation of the
solvent, the dried powder was mixed with an organic binder (Paste Vehicle purchased from
Fuel Cell Materials) in a powder : binder mass ratio of 0.67 : 0.33, before milling the mixture
for a further 12 h. The resulting ink was used for screen printing to both sides of the
electrolyte and the adherence of the ink to the surface of the electrolyte was achieved after
calcinations at 800 or 900 °C for 1 h. Various processing conditions (drying temperature after
screen printing, number of screen printed cathode layers, different level of polishing of the
electrolyte surface) were applied to improve the electrochemical performance of the cells.
Gold mesh with gold paste was used as current collector for the AC impedance
spectroscopy measurements. AC impedance spectroscopy was carried out using a Solatron
1255B Frequency Response Analyzer and Solatron SI 1287 Electrochemical Interface. The
data were collected over a frequency range of 1 MHz – 0.1 Hz with a modulation potential of
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
90
10 mV, over the temperature of 500 – 800 °C in static air with a 90 min step every 50 °C.
Impedance measurements and corresponding equivalent circuit modelling were performed
using the ZPlot and ZView software196
. The ASR of the cathode was calculated by
normalising the measured resistance for the electrode area and dividing by two to take into
account the symmetry of the cell (Equation 2.34, Section 2.5.2).
Figure 3.14 shows the AC impedance response of the cathode at different
temperatures, from which ASR values were obtained. The ohmic impedance associated with
the GDC10 and SDC20 electrolytes has been subtracted from the spectra and the AC
impedance arcs normalised to zero on the real (x) axis for easy comparison.
Figure 3.14: Nyquist plot of measured AC impedance arcs of the cell made of
Y0.95Sr2.05Cu1.7Co1.3O7+δ (dried at 800 °C) and unpolished GDC10 at temperatures 600 ‒ 700 °C.
It is generally accepted that the cathode processing conditions can have a dramatic
effect on the physical properties (e.g. porosity, surface area, interface, adherence) of cathodes
and therefore altering the processing conditions (such as drying temperature after screen
printing, polishing of the electrolyte surface and the number of printed layers) may cause
different ASR values which are difficult to predict.47,62
To investigate this, various processing
conditions for symmetrical cells of Y0.95Sr2.05Cu1.7Co1.3O7+δ were applied, as outlined in
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
91
Table 3.12. Throughout this chapter, we will use a labelling system (e.g. GDC Cell 1) for the
symmetrical cells as outlined in Table 3.12. Here we will use GDC or SDC to denote the
electrolyte used, whilst cell followed by a number (1-4) will indicate the processing
conditions used (e.g. GDC Cell 1 describes a symmetrical cell made from GDC10, with a
polished electrolyte surface and a cathode drying temperature of 800 °C). Eight symmetrical
cells (4 with GDC10, 4 with SDC20 electrolyte) were prepared and measured on first heating
(600 – 800 °C), cooling (800 – 600 °C) and second heating (600 – 800 °C). Figure 3.15 and
Table 3.13 show the ASR values for all cells collected during the first heating. The lowest
ASR values of 0.08 Ω cm2 at 700 °C, were given by GDC Cell 3 (dried at 800 °C and
polished electrolyte surface; Table 3.13, Figure 3.15a and Figure 3.16).
Table 3.12: Processing conditions for symmetrical cells containing one layer of
Y0.95Sr2.05Cu1.7Co1.3O7+δ, as applied to two ceria based electrolytes (GDC10 and SDC20). The cells
will be referred to as e.g. “GDC Cell 1”, which would refer to a cell with GDC10 electrolyte
processed under condition (1).
Symmetrical cell Drying temperature
(°C)
Electrolyte surface Printed layers
(1) 800 Unpolished 1
(2) 900 Unpolished 1
(3) 800 Polished 1
(4) 900 Polished 1
Figure 3.15: AC impedance spectroscopy data of symmetrical cells made of Y0.95Sr2.05Cu1.7Co1.3O7+δ
as a cathode and using different processing conditions: (1) to (4) ‒ see Table 3.12, with a) GDC10 and
b) SDC20 electrolyte.
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
92
Table 3.13: Values of ASR and Ea calculated from the AC impedance data collected on the
symmetrical cells made by altering processing conditions (Table 3.12).
Processing
condition
ASR (Ω cm2) Ea
(eV)
(650 °C) (700 °C) (750 °C) (800 °C)
GDC Cell 1 0.287 0.136 0.122 0.090 0.35
GDC Cell 2 0.236 0.184 0.092 0.070 0.25
GDC Cell 3 0.207 0.081 0.065 0.057 0.27
GDC Cell 4 0.235 0.155 0.083 0.073 0.26
SDC Cell 1 0.530 0.239 0.172 0.115 0.61
SDC Cell 2 0.311 0.174 0.152 0.094 0.36
SDC Cell 3 0.243 0.117 0.083 0.063 0.32
SDC Cell 4 0.206 0.172 0.156 0.114 0.20
Although the Y0.95Sr2.05Cu1.7Co1.3O7+δ material shows promising ASR values for
symmetrical cells with both GDC10 and SDC20 (Figure 3.15, Table 3.13), there is an
increase of ASR values of one order of magnitude (in the 600 – 700 °C region) between the
first and second heating cycles as shown in Figure 3.16. The same trend was observed for all
of the symmetrical cells shown in Figure 3.15. This is presumably the reason for the
deviation observed in Arrhenius-type plots (Figure 3.15 and Figure 3.16). PXRD data of the
symmetrical cells after the AC impedance (Figure 3.17) confirmed the presence of other
phases (Y2Cu2O5, Co3O4, Sr2CeO4), and this has been attributed to in-situ electrochemical
decomposition. In-situ electrochemical decomposition has also been observed in
YBa2Cu3O7−δ at low current densities (≈ 20 mA cm−2
).68
This is different compared to the
thermal stability tests of Y0.95Sr2.05Cu1.7Co1.3O7+δ (Section 3.5), where no impurity phases
were observed. In order to improve the stability of Y0.95Sr2.05Cu1.7Co1.3O7+δ, AC impedance
measurements were carried out with a reduced maximum temperature of 650 °C.
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
93
Figure 3.16: The comparison of ASR values between the 1st and 2
nd heating of the cell made of
polished GDC, dried at 800 °C (GDC Cell 3).
Figure 3.17: XRD data of Y0.95Sr2.05Cu1.7Co1.3O7+δ before and after the AC impedance spectroscopy
(GDC Cell 3) with the marked reflections of GDC10 electrolyte, gold, the main reflections of
additional phases presented after the measurement: +Y2Cu2O5, †Co3O4, *Sr2CeO4 and missing (200)
reflection after the AC impedance spectroscopy.
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
94
3.7.2 AC impedance data for the dwelling at 650 °C
As mentioned in previous section, Y0.95Sr2.05Cu1.7Co1.3O7+δ material exhibited
promising ASR values for symmetrical cells with ceria based electrolytes, but also in-situ
electrochemical decomposition at higher temperatures (800 °C). Current research for SOFCs
cathodes is focused for the materials working in intermediate temperature range
(600 − 800 °C). Materials with a good cathode performance (processing, stability,
conductivity, etc.) at the temperature close to 600 °C are of significant interest for both
research and industry.9,17
Hence, the AC impedance of Y0.95Sr2.05Cu1.7Co1.3O7+δ material was
carried out at temperatures below 800 °C. From the previous results on symmetrical
cells (Table 3.13), the triple perovskite showed promising ASR values at 650 °C. The other
processing factor, which has not been considered is the thickness of the cathode layer of a
symmetrical cell. The thickness can be changed by the number of screen printed layers of the
cathode material.
The symmetrical cells were prepared following the procedure mentioned in
Section 4.7.1. The processing conditions (polishing, drying temperature and type of
electrolyte) were selected from previously measured cell – from the one with the best ASR
values. Thus symmetrical cells of the Y0.95Sr2.05Cu1.7Co1.3O7+δ material and polished GDC10
dried at 800 °C (GDC Cell 3) were constructed. The cells were constructed with 1 or 6 layers
(GDC Cell 3 L1 and GDC Cell 3 L6 respectively) of screen printed Y0.95Sr2.05Cu1.7Co1.3O7+δ
in order to see if there was any influence of cathode thickness on cathode performance and
electrochemical stability.
Figure 3.18 shows the ASR values obtained whilst dwelling at 650 °C for 12 h. For
both 1 and 6 layered cells, there is no significant increase of ASR observed (Table 3.14),
indicating that the number of screen printed cathode layers does not affect the
electrochemical stability. The total increase of ASR for both the 1 and 6 layered cells after
12 h dwelling was within 5% of the starting values. The Y0.95Sr2.05Cu1.7Co1.3O7+δ symmetrical
cell with 6 layers shows lower (improved) ASR values compared to the cell with 1 screen
printed cathode layer. The starting ASR value of 0.17 Ω cm2 for the 6 layered cell of the
x = 1.3 sample increased after 12 h annealing to 0.18 Ω cm2. After the measurement, PXRD
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
95
confirmed that the sample was phase pure (Figure 3.19), indicating that in-situ
electrochemical decomposition had been avoided.
Figure 3.18: ASR values obtained when annealing GDC Cell 3 L1 (1 cathode layer) and GDC Cell 3
L6 (6 cathode layers) at 650 °C for 12 h.
Table 3.14: Comparison of the ASR values obtained after 1 h and 12 h dwelling at 650 °C of
symmetrical cells of Y0.95Sr2.05Cu1.7Co1.3O7+δ and GDC10 (GDC Cell 3 L1 and GDC Cell 3 L6).
Symmetrical cell ASR (Ω cm2) (1 h) ASR (Ω cm
2) (12 h)
GDC Cell 3 L1 (1 layer) 0.39 0.41
GDC Cell 3 L6 (6 layers) 0.17 0.18
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
96
Figure 3.19: PXRD data of Y0.95Sr2.05Cu1.7Co1.3O7+δ after AC impedance at 650 °C for 12 h. The main
reflections of GDC10 and Au are indicated by + and * respectively. All other reflections are indexed
to Y0.95Sr2.05Cu1.7Co1.3O7+δ.
3.7.3 SEM study of symmetrical cells
The morphology of symmetrical cell of Y0.95Sr2.05Cu1.7Co1.3O7+δ cathode with GDC
Cell 3 L6 (6 layered) was evaluated by SEM (instruments details available in Section 2.3).
The impedance data of the cells are shown in Section 3.7.2 and suggests that alternating the
number of screen-printed layers of the cathode material can improve the ASR values.
Studying the cell by SEM provides a better insight into the real thickness of the screen-
printed cathode layer. Any other additional reaction on boundaries between the layers of the
symmetrical cells can be also observed.
Figure 3.20 shows the SEM data with different magnifications collected on the
symmetrical cell made of 6 layers of Y0.95Sr2.05Cu1.7Co1.3O7+δ (with GDC Cell 3 L6) after the
AC impedance. We can clearly see all three parts of the symmetrical cell (Au current
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
97
collector, cathode, and electrolyte layer). No additional interfacial layers were observed. The
total cathode thickness (for a total of 6 screen printed layers) is 40 μm.
Figure 3.20: Cross-sectional image of the symmetrical cell that exhibited best performance
measurements (with ASR650 °C = 0.18 Ω cm2) after AC impedance spectroscopy data collection. The
cell was constructed of 6 layer of screen printed Y0.95Sr2.05Cu1.7Co1.3O7+δ cathode and GDC Cell 3 L6
is shown with different magnifications: a) 50 μm; b) 30 μm.
3.8 AC impedance spectroscopy of Y0.97Sr2.03Cu2CoO7+δ
The previous section (Section 3.7) was focused on the impedance measurement of the
most conductive Y0.95Sr2.05Cu1.7Co1.3O7+δ material (Section 3.4). AC impedance data from
Y0.97Sr2.03Cu2CoO7+δ phase were collected to compare with the data obtained on x = 1.3
material. GDC Cell 1 with one screen printed cathode layer was selected for this comparative
study. The ASR values and activation energies obtained from the AC impedance data of
Y0.97Sr2.03Cu2CoO7+δ phase are shown and compared with Y0.95Sr2.05Cu1.7Co1.3O7+δ in Figure
3.21 and Table 3.15. At lower temperatures (600 and 650 °C), the ASR values of
Y0.95Sr2.05Cu1.7Co1.3O7+δ are one order of magnitude lower than those obtained for
Y0.97Sr2.03Cu2CoO7+δ. The improvement of the electrochemical performance of
Y0.95Sr2.05Cu1.7Co1.3O7+δ compared to x = 1 sample was achieved for the whole temperature
range. The activation energy value for the x = 1.3 material is almost of one half of the
activation energy obtained from the data collected on the parental x = 1 phase.
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
98
Figure 3.21: ASR values comparison of symmetrical cells of Y0.95Sr2.05Cu1.7Co1.3O7+δ and
Y0.97Sr2.03Cu2CoO7+δ, with GDC Cell 1 (800 °C drying temperature and unpolished electrolyte
surface). The activation energy shown was calculated from the values in the 650 to 800 °C region.
Table 3.15: Values of ASR and activation energies (calculated for 650 – 800 °C) for symmetrical
cells of Y0.95Sr2.05Cu1.7Co1.3O7+δ and Y0.97Sr2.03Cu2CoO7+δ, with one screen printed cathode layers using
GDC Cell 1 (800 °C drying temperature and unpolished electrolyte surface).
Co doping level ASR (Ω cm2) Ea (eV)
(650 °C) (700 °C) (750 °C) (800 °C)
x = 1 2.82 1.14 0.25 0.13 0.62(9)
x = 1.3 (1st heating) 0.29 0.14 0.12 0.09 0.35(8)
x = 1.3 (2nd
heating) 2.26 0.91 0.28 0.12 0.57(3)
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
99
3.9 CO2 stability tests of Y0.95Sr2.05Cu1.7Co1.3O7+δ
The presence of CO2 in the air feedstock gas used at the cathode means that potential
cathode materials must be tested for their stability in CO2 containing atmospheres (typical
CO2 levels in air are in the region of 400 ppm by volume). Reaction of the cathode material
with CO2 in air at SOFC operating temperatures can lead to performance loss due to the
formation of carbonate containing impurities.
The stability of Y0.95Sr2.05Cu1.7Co1.3O7+δ was tested at low and high CO2
concentrations. Approx. 0.3 g of the material was weighed out, pressed into an 8 mm pellet
and put into an alumina crucible. A mixture of 1% CO2 in Ar was passed over
Y0.95Sr2.05Cu1.7Co1.3O7+δ for 12 h at temperatures of 600, 650, and 700 °C. The resultant
PXRD data are shown in Figure 3.22. No impurity phases were observed at 600 °C,
indicating that at this temperature Y0.95Sr2.05Cu1.7Co1.3O7+δ is stable under low CO2. At the
higher temperatures of 650 and 700 °C, SrCO3 and CuO impurities were observed in the
diffraction patterns. Lattice parameters and volume of the unit cell obtained from Rietveld
refinements (Table 3.16-3.18) show expansion of the unit cell which may be due to the
reduction of Co and Cu. To investigate this annealing in pure Ar atmospheres were
performed using the same procedure and amounts of material as in CO2 stability tests. There
were no other phases presented. Table 3.16-3.18 show the same trend of expanded lattice
parameters and volumes both in the case of annealing in pure Ar and in mixture Ar + 1%
CO2, meaning that these changes are due to the reduction and are not related to the presence
of CO2.
In order to test the behaviour of Y0.95Sr2.05Cu1.7Co1.3O7+δ in a pure CO2 atmosphere,
TGA was performed on the sample whilst annealing at 650 °C for 12 h under pure CO2. TGA
showed an increase in mass over time (Figure 3.23), indicating a reaction with CO2,
compared to no change in mass when the TGA was carried out in air (Figure 3.12). This was
confirmed by PXRD data (Figure 3.22 and Figure 3.24) collected after the TGA experiment,
which showed the presence of SrCO3 and CuO in the diffraction pattern, indicating
decomposition of Y0.95Sr2.05Cu1.7Co1.3O7+δ.
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
100
Figure 3.22: PXRD data of Y0.95Sr2.05Cu1.7Co1.3O7+δ after exposure to 1% CO2 in Ar (at temperatures
of 600, 650 and 700 °C) and to pure CO2 at 650 °C, * denotes the CuO reflections, + indicates SrCO3
phase.
Table 3.16: Comparison of lattice parameters of the Y0.95Sr2.05Cu1.7Co1.3O7+δ sample after the 12 h
annealing at 600 °C in Ar + 1% CO2 and pure Ar atmosphere with as made powder, obtained from
Rietveld refinements using Topas.152
600°C a(Å) b(Å) c(Å) V(Å3)
As made 22.7210(8) 5.43220(5) 5.43140(4) 670.37(2)
Ar/1% CO2 22.7470(4) 5.45489(8) 5.42001(9) 672.53(2)
Ar 22.7451(3) 5.45134(7) 5.41735(7) 671.70(1)
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
101
Table 3.17: Comparison of lattice parameters of the Y0.95Sr2.05Cu1.7Co1.3O7+δ sample after the 12 h
annealing at 650 °C in Ar + 1% CO2 and pure Ar atmosphere with as made powder, obtained from
Rietveld refinements using Topas.152
650°C a(Å) b(Å) c(Å) V(Å3)
As made 22.7210(8) 5.43220(5) 5.43140(4) 670.37(2)
Ar/1% CO2 22.7420(4) 5.4594(1) 5.4204(1) 672.99(2)
Ar 22.7320(3) 5.46414(7) 5.42150(8) 673.41(2)
Table 3.18: Comparison of lattice parameters of the Y0.95Sr2.05Cu1.7Co1.3O7+δ sample after the 12 h
annealing at 700 °C in Ar + 1% CO2 and pure Ar atmosphere with as made powder, obtained from
Rietveld refinements using Topas.152
700°C a(Å) b(Å) c(Å) V(Å3)
As made 22.7210(8) 5.43220(5) 5.43140(4) 670.37(2)
Ar/1% CO2 22.738(1) 5.4620(2) 5.4204(2) 673.22(5)
Ar 22.7227(3) 5.46804(6) 5.42137(6) 673.60(1)
Figure 3.23: TGA experiment of Y0.95Sr2.05Cu1.7Co1.3O7+δ material, after exposure to pure CO2 at
650 °C for 12 h.
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
102
Figure 3.24: Rietveld refinement of Y0.95Sr2.05Cu1.7Co1.3O7+δ sample collected after the TGA
experiment carried out in pure CO2 at 650 °C for 12 h.
3.10 Thermal expansion studies of Y0.95Sr2.05Cu1.7Co1.3O7+δ
Parent YSr2Cu2CoO7+δ material showed an orthorhombic-tetragonal phase transition
at above 800 °C.69
The unit cell contraction at the phase transition was also observed using
dilatometry.70
Any phase transition of a cathode material may be detrimental for a SOFC
device causing mechanical stress of the cell. The thermal expansion studies were carried out
on the most conductive Y0.95Sr2.05Cu1.7Co1.3O7+δ material using a dilatometry measurement.
Prior to the dilatometry measurement, Y0.95Sr2.05Cu1.7Co1.3O7+δ was weighed out,
pressed into a 6 mm pellet and sintered in ambient air at 1050 °C (following the synthesis
conditions in Section 3.2). Linear thermal expansion of the sample was measured in the range
20 – 850 °C with the heating rate 10 °C min−1
. The sample was then cooled down to 20 °C
and heated again in order to see any changes in behaviour between two heating cycles. The
data obtained during the first heating cycle are shown in Figure 3.25. There is a slight
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
103
decrease of the thermal expansion recorded during the second heating compared to the first
one (Table 3.19). In both cycles, a linear change of expansion is observed without any sharp
change in the values especially above 800 °C, where the phase transition was reported.70
To
ensure there is no phase transition, first derivation (dL/dt) of the linear thermal expansion can
be plotted (Figure 3.25). There is no peak presented. That indicates no phase transition (apart
slow phase transitions) in Y0.95Sr2.05Cu1.7Co1.3O7+δ in the temperature range 20 – 850 °C.
Table 3.19 shows the value of the linear thermal parameter – α obtained from the
linear thermal expansion data (see Chapter 2.10) for the first and second heating.
The α values for selected temperatures (8.4-8.8 × 10−6
K−1
for the temperatures between 600
and 850 °C) are all within narrow range and they increase with the increasing temperature.
Reported dilatometry measurement of YSr2Cu2CoO7+δ material70
clearly showed phase
transition with the values of thermal coefficient (TEC): 13.0 × 10−6
K−1
above and
12.3 × 10−6
K−1
below phase transition. Measured values for Y0.95Sr2.05Cu1.7Co1.3O7+δ material
are lower than the ones for the starting phase. The difference can be assigned to different
Y/Sr ratio and higher Co-amount.
Figure 3.25: Linear thermal expansion of Y0.95Sr2.05Cu1.7Co1.3O7+δ measured in 20 ‒ 850 °C range.
Black data points indicate the obtained values for the 1st heating whilst the red points show the first
derivation of the collected data (dL/dt).
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
104
Table 3.19: Linear thermal expansion coefficients (α, 100 °C – selected temperature) of the
Y0.95Sr2.05Cu1.7Co1.3O7+δ material calculated from the data collected for both heating cycles.
Temperature (°C) α (K−1
) – 1st cycle α (K
−1) – 2
nd cycle
600 8.45 × 10−6
8.43 × 10−6
650 8.54 × 10−6
8.52 × 10−6
700 8.63 × 10−6
8.61 × 10−6
750 8.70 × 10−6
8.69 × 10−6
800 8.76 × 10−6
8.76 × 10−6
850 8.78 × 10-6
8.81 × 10−6
3.11 Discussion and conclusions
An increase of the cobalt content in the layered YSr2Cu2CoO7+δ perovskite was
achieved by altering the Y:Sr cationic ratio, resulting in materials of the form
Y1−ySr2+yCu3−xCoxO7+δ (1≤ x ≤ 1.3, y = 0, 0.03 and 0.05). The higher the Co content, the
higher the Sr content is required to obtain single phase samples. This most probably is a
result of charge balance and is consistent with cobalt commonly averaging higher formal
oxidation state than copper in perovskite oxides, or more generally in transition metal oxides.
An oxygen content of YSr2Cu2CoO7.03(4) was observed for the parent non substituted phase
by Slater et al.69
using thermogravimetric methods. The oxygen content is independent of the
level of Co substitution which suggest that a good stability of the peculiar oxygen vacancy
ordering pattern, associated here with the O7 stoichiometry, is responsible for the variable Y
: Sr ratios needed to obtain phase pure samples as the Co content varies. Indeed, at a fixed
oxygen content, incorporation of extra Co needs more Sr to reduce the total charge on the A-
site, in order to compensate the increased charge on the B-site. Assuming a copper oxidation
state of (+2), the Co oxidation state in Y0.95Sr2.05Cu1.7Co1.3O7+δ is lower (+2.8) than the Co3+
found in the parent YSr2Cu2CoO7+δ phase and Co substitution therefore allows the formation
of mixed valent compounds.
The retention of the O7 stoichiometry throughout the series of Co enriched materials
allows the production of mixed valent compounds with a higher content of charge carriers,
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
105
which may imply easier electron delocalisation. This is reflected in the DC conductivity data,
which show that we were able to produce an enhancement of the conductivity by one order of
magnitude when comparing Y0.95Sr2.05Cu1.7Co1.3O7+δ to Y0.97Sr2.03Cu2CoO7+δ. The obtained
values for Y0.95Sr2.05Cu1.7Co1.3O7+δ are lower than those for the pure Co containing
La0.7Sr0.3CoO3−δ, which has a total conductivity of 1650 S cm−1
at 800 °C,197
but they are
comparable with dominant cathode materials such as La0.6Sr0.4Fe0.8Co0.2O3−δ (LSCF) which
has a total conductivity of 280 S cm−1
at 800 °C,40
and is better than the total conductivity of
22.7 S cm−1
(at 800 °C) reported for Ba0.5Sr0.5Co0.8Fe0.2O3−δ.198
Parent YSr2Cu2CoO7+δ showed an orthorhombic-tetragonal phase transition at above
800 °C.69
The dilatometry study of Y0.95Sr2.05Cu1.7Co1.3O7+δ revealed no phase transition
between 20 – 850 °C. The obtained linear thermal expansion coefficient (8.7 × 10−6
K−1
) is
smaller than the TEC of LSCF ((La0.6Sr0.4)1−xCo0.2Fe0.8O3−δ, for x = 0.00-0.15) series,
showing a TEC of 13.8-14.2 × 10−6
K−1
at the temperature range 25 – 700 °C.199,200
TEC
values of measured triple perovskite are comparable with LSM series with TEC of 10.8 ×
10−6
K−1
for La0.79Sr0.2MnO3−x for the temperature range 30 – 800 °C.201
The thermal expansion of Y0.95Sr2.05Cu1.7Co1.3O7+δ material matches well with YSZ
electrolyte (10.5 × 10−6
K−1
for Zr0.85Y0.15O1.93 between 30 – 800 °C)202
and LSGM
electrolytes (10.4 × 10−6
K−1
for La0.8Sr0.2Ga0.9Mg0.1O3−x between 30 ‒ 800 °C).203
Higher
mismatch in TEC values is observed in comparison with Ce-based electrolytes
(12.5 × 10−6
K−1
for Ce0.8Gd0.2O1.90, 30 – 800 °C).202
The suitability of a material as a cathode can be assessed from the ASR values
obtained during cell testing; small ASRs are desirable as the total resistance of the cell is
usually dominated by the contribution from the cathode. The small increase in Co content in
Y1−ySr2+yCu3−xCoxO7+δ results in an improved (decreased) ASR compared with the parent
material. Although the material suffers electrochemical decomposition at the higher
temperatures studied here, we have also shown that as low as 15% Co substitution in the
square based pyramidal site of the initial structure enhances the electrochemical properties of
the 3ap material, again by an order of magnitude at 650 °C, where the material is stable. The
ASR value for Y0.95Sr2.05Cu1.7Co1.3O7+δ at 700 °C is 0.08 Ω cm2 and compares favourably to
the ASR reported for some of the best reported materials such as the pure cobalt containing
double perovskite – SmBa0.5Sr0.5Co2O5+δ which exhibits an ASR value of 0.092 Ω cm2
at 700 °C. 67
Y0.95Sr2.05Cu1.7Co1.3O7+δ shows similar or improved properties compared to other
Co compounds which are considered as intermediate temperature SOFC cathodes, such as the
double perovskite GdBaCo2O5+δ (ASR700 °C = 0.1 Ω cm2),
62 and the tetrahedrally coordinated
Chapter 3. Synthesis and characterisation of Y1−ySr2+yCu3−xCoxO7+δ
106
Co compound YBaCo4O7 (ASR600 °C = 0.40 Ω cm2).
204,205 The ASR values of
Y0.95Sr2.05Cu1.7Co1.3O7+δ at 700 °C are lower than those reported for LSCF
(ASR700 °C = 0.18 Ω cm2).
206 Y0.95Sr2.05Cu1.7Co1.3O7+δ shows a higher ASR than that obtained
for the perovskite Ba0.5Sr0.5Co0.8Fe0.2O3−δ (which has an ASR value of 0.01 Ω cm2 at
700 °C),47
which displays one of the lowest ASR values to date, although has significant
stability problems. The good electrochemical performance we observe here may be related to
some key structural features of the Y1−ySr2+yCu3−xCoxO7+δ. The cobalt enriched compound
retains a crystal structure that should be highly favourable for oxide ion mobility with several
functional sites able to enhance it, such as a fully oxygen vacant layer at the level of Y site or
a half filled layer with ordered vacancies at the level of the Co site.207
At the same time, the
incorporation of Co in the square based pyramid containing layers may provide a catalytic
boost for the oxygen reduction reaction, as this element is generally believed to provide good
catalytic activity17
while square based pyramids are favourable environments for O2 molecule
dissociation.208
Therefore the Y1−ySr2+yCu3−xCoxO7+δ series offers a combination of crystal
chemical features which may provide a high level of electrochemical performance, while
achieving the required values of total conductivity for SOFC cathode materials, owing to the
charge carriers generated by the association of Co doping and Y : Sr ratio changes.
In conclusion, although the Co substitution has substantially improved the
electrochemical properties, further work is needed to enhance the chemical and
electrochemical stability of this class of complex oxides for SOFC cathode applications ‒ this
is not uncommon for the most active electrode systems. This class of candidate cathode
materials is based on a rare example of high spin tetrahedral Co3+
and further investigation is
needed to understand how to optimise materials of this type for fuel cell applications, as this
study tends to suggest that a high level of electrochemical performances could be achievable
within this type of 3ap structured materials.
4 Prediction and synthesis of GBCO related phases
4.1 Introduction
Ordered double perovskites have attracted attention in SOFCs and oxygen storage
research in recent years. Their composition is represented by the general formula AA'Co2O5+δ
(A = RE, Y and A' = Ba, Sr).60
RE-O and A'-O layers alternate along the c-axis (Figure 4.1)
with the A' atoms located in the Co double layers. An interesting feature of double perovskite
is the variety of possible oxygen contents within the range of 0 ≤ δ ≤ 1.0. Generation of
electron-hole pairs, or Co2+
-Co4+
states determines the transport behaviour of AA'Co2O5+δ.209-
214 It has been observed that the oxide-anion vacancies are localized in the RE-O layers due to
the preference of lower coordination number of smaller rare earth element anions compared
to larger Ba2+
or Sr2+
anions. That results in the formation of CoO5 square pyramids and
CoO6 octahedra.60,215
The magnetic and transport properties of AA'Co2O5+δ are highly
dependent on the oxygen content. The oxygen content, δ, controls the mixed valence state of
Co ions216
YBaCo2O5 is an example of the oxygen deficient compound where all Co ions
have the same environment while in RBaCo2O5.5 (R = Tb, Gd) there are two different Co
sites. 217
Figure 4.1: a) Reported structure for GdBaCo2O5.5 210
b)√2×√2×1 cell used for DFT calculation
expanded in a, b lattice parameters. Atoms coloured as follows: gadolinium (purple), barium (green),
cobalt (blue), and oxygen (red).
Chapter 4. Prediction and synthesis of GBCO related phases
108
An example of a material with the double perovskite structure is GdBaCo2O5+δ
(Figure 4.1). Its good mixed conductor properties were first demonstrated by Taskin et
al.218,219
when high rates of oxygen uptake were measured. Clear indication of the importance
of vacancy layers to this process was also demonstrated. GdBaCo2O5+δ (GBCO) as a potential
SOFC cathode materials was introduced by Chang et al.63
A further study showed an
impedance spectroscopy data of GdBaCo2O5+δ and BSCF cathodes with LSGM electrolyte.138
Reported ASR values of symmetrical cells with GBCO at the temperatures above 700 °C are
lower than 0.15 Ω cm2.138
Another electrochemically studied double perovskites were
LnBaCo2O6−δ (Ln = Pr, Nd, Sm, and Gd) phases.220
An orthorhombic to tetragonal phase
transition of LnBaCo2O5+δ (Ln = Pr, Nd and Sm) was evidenced by high temperature X-ray
diffraction data.221
The chemical stability of the LnBaCo2O5+δ materials has been tested against the
common electrolyte materials such as YSZ, LSGM, and GDC. The Ln = La, Nd, and Sm are
stable against LSGM electrolyte, while the Ln = Gd and Y structures show significant side
reaction.222
Similarly to that, the Ln = La and Nd materials are compatible with GDC, but the
Ln = Gd, Y do react with GDC electrolyte.223
Thermal stability and thermal expansion is one
of the problems of Co-containing perovskite cathodes. High values of linear TEC
(αL = 16‒14 × 10−6
K−1
) were also reported for LnBaCo2O5+δ materials.223
However, materials
with medium-size cations (Gd3+
and Y3+
) show TEC values (≈ 16 × 10−6
K−1
) comparable to
some of commonly used electrolytes, especially ceria-based ones.
The work presented in this chapter is focused on double perovskite related materials.
Studied materials, Ln2BaCo2O7 (Ln = Gd, Nd, Ce), consist of a layer of LnBaCo2O5+δ
(Ln = Gd, Nd) and a fluorite layer (CeO2 or Ln2O3, Ln = Gd, Nd). Work with these new
materials is aimed towards potential SOFC cathode materials. By adding a fluorite layer to a
mixed double perovskite conductor an improvement of an ionic conductivity is expected. A
part of the work involves the study of Sr analogues. The materials are studied both by
experimental and theoretical methods. The first part is focused on the prediction of the
stability of the material based on DFT calculations. The structures for theoretical studies are
shown in Figure 4.2. The structural model for the materials with double perovskite and
fluorite layer was based on the reported structure of Y2SrCuFeO6.5 (Figure 4.3)24,25
and it is
discussed in more details in Section 4.2. Relaxations of the individual parts (fluorite, double
Chapter 4. Prediction and synthesis of GBCO related phases
109
perovskite layers) were obtained beforehand and compared with the experimental data. The
synthesis of all the materials was attempted using solid state methods. The study was also to
target materials with the different fluorite layer composed of Ln2O3 oxide (Ln = Gd, Nd).
4.2 Computational methods
All the calculations were carried out using the plane wave DFT package, Vienna Ab-
initio Simulation Package (VASP) version 4.6.26178,224
with Perdew, Burke and Ernzerhof
(PBE)225
exchange correlation functional used mainly in this work. Different exchange
correlation functional: LDA226,227
and PW91228
were used at the beginning of the work
(Section 4.4.1). For Co, the first sub-valence p orbital was treated as valence, while for Ba it
was s orbital. For lanthanides there is a special GGA potential available in which f electrons
are kept frozen in the core. The number of f electrons in the core is equal to the total number
of valence electrons minus the formal valency. These potentials were used for Gd and Nd,
unlike for Ce where the standard potential with f states treated as valence states was used.
The automatic Monkhorst-pack was used for a k-point grid. The number of k-points varied
depending on the ratio of lattice parameters for a calculated structure. Thus 3 × 3 × 3 k-point
mesh was used for CeO2 and LnBaCo2O5 (Ln = Gd, Nd) calculations, 4 × 3 × 3 for the rest of
the structures. The unit cell size and atomic co-ordinates were relaxed until forces on atoms
were less than 0.01 eV/Å. The relaxation process was stopped when the energy difference
between two steps was smaller than 10−5
eV. The cut-off energy for plane wave was set to
450 eV. The computational setup was chosen after the preliminary study on CeO2 study using
different settings and the comparison of the obtained data with the experiments.
The initial atomic coordinates of LnBaCo2O5 and LnBaCo2O5.5 structures were used
from the reported data (Table 4.2 and Table 4.3). The reported unit cell was rotated in 45°,
and lattice parameters a and b were expanded by √2 (Figure 4.1). The unit cell was expanded
in order to include the structures with both ferromagnetic (FM) and anti-ferromagnetic
ordering (AFM) due to the Co mixed valence, although G-type antiferromagnetism for
LnBaCo2O5.5 (Ln = Tb, Gd) phases has been reported.217
Blocks (CeO2, GdBaCo2O5,
GdBaCo2O5.5) were relaxed separately and their relaxed atomic positions were used for the
starting structure of the double perovskites with the fluorite CeO2 layer. The unit cell of
LnBaCo2O5(5.5) with CeO2 layer was then doubled in c parameter (Figure 4.2c) in order to
Chapter 4. Prediction and synthesis of GBCO related phases
110
accommodate the fluorite layer. The atomic positions of the additional cell were shifted in a
direction for the value of a/2 compared to the ones obtained by the previous DFT relaxations.
This structure (with the shifted two blocks of a double perovskite and a fluorite block) was
reported for the Y2SrCuFeO6.5139,140
and it served as a template for the DFT relaxations with
the LnBaCo2O5(5.5) + CeO2 structures.
Figure 4.2: Structures used for DFT calculations: a) GdBaCo2O5, b) CeO2, c) GdCeBaCo2O7, with
GdBaCo2O5 and fluorite CeO2 layer between double perovskite blocks. Atoms coloured as follows:
gadolinium (purple), cerium (yellow), barium (green), cobalt (blue), and oxygen (red).
The values of total energies of the relaxed structures were used to calculate the
formation energies of the structures with fluorite CeO2 layer. These values were obtained
from various equations (Equations 4.1-4.4, where Ln = Gd, Nd, and AE = Ba, Sr) depending
on the reaction enthalpies of final products from double perovskites, binary oxides or
AECeO3 (calculated because of the well known high stability of cerates). Since LnBaCo2O5
is not very stable (thus, the formation energies are higher) it is more precise to calculate the
formation energies in comparison to the more stable compound ‒ LnBaCo2O5.5. The
formation energies were calculated in electron Volts per Formula Unit, eV/FU. The more
negative values were obtained, the more stable were the structures on the right side of the
Chapter 4. Prediction and synthesis of GBCO related phases
111
equations. As it was mentioned in Section 2.11.5 in order to calculate more accurate
formation energies for d electron-containing perovskites, U must be applied to the d orbitals
of the transition metals.229
For the calculations in this chapter the value of U = 4.3 eV was
used for the Co atom.230
Corrected value −8.5 eV of O2 gas was taken into account for the
formation energy calculations.230
LnAECo2O5 + CeO2 → LnCeAECo2O7 (4.1)
LnAECo2O5.5 + CeO2 → LnCeAECo2O7 +
O2 (4.2)
AEO +
Ln2O3 +
Co3O4 + CeO2 → LnCeAECo2O7 +
O2 (4.3)
AECeO3 +
Ln2O3 +
Co3O4 → LnCeAECo2O7 +
O2 (4.4)
For the Gd-containing materials with the fluorite layer built of two layers of CeO2
Equations 4.1-4.4 were adjusted (Equations 4.5-4.8).
GdBaCo2O5 + 2 CeO2 → GdCe2BaCo2O9 (4.5)
GdBaCo2O5.5 + 2 CeO2 → GdCe2BaCo2O9 +
O2 (4.6)
BaO +
Gd2O3 +
Co3O4 + 2CeO2 → GdCe2BaCo2O9 +
O2 (4.7)
BaCeO3 +
Gd2O3 +
Co3O4 → GdCe2BaCo2O9 + BaO +
O2 (4.8)
Similar to previous equations, various equations (Equations 4.9-4.11) were used to calculate
the formation energies of the materials with different fluorite layers (Ln2O3, Ln = Gd, Nd). A
Ruddlesden-Popper phase (RP2) was taken into account as another possible structure because
of the same nominal composition. The atomic coordinates of Gd2SrCo2O7 231
(Figure 4.3)
were used as a template for relaxation of Ln2BaCo2O7 (Ln = Gd, Nd) in the RP2 structural
form. The final formation energies were calculated for both structural models and compared
each other.
LnBaCo2O5 +
Ln2O3 → Ln2BaCo2O7 −
O2 (4.9)
LnBaCo2O5.5 +
Ln2O3 → Ln2BaCo2O7 (4.10)
BaO + Ln2O3 +
Co3O4 → Ln2BaCo2O7 −
O2 (4.11)
Chapter 4. Prediction and synthesis of GBCO related phases
112
Figure 4.3: Reported structure for Gd2SrCo2O7 material231
used as a template for Ln2BaCo2O7
(Ln = Gd, Nd) DFT relaxation in RP2 structural model. Atoms coloured as follows: gadolinium
(purple), strontium (green), cobalt (blue), and oxygen (red).
Different atoms of lanthanides were used in relaxations of the LnBaCo2O5 structure (Ln = Ce,
Dy, Er, Eu, Gd, Ho, La, Nd, Pr, Tb, and Tm). PBE-GGA exchange correlation functional was
applied with the same configuration for the rest of the set-up, mentioned at the beginning of
this chapter. The formation energies values of the LnBaCo2O5 double perovskites were
calculated compared to binary oxide according to Equation 4.12. Since the most common
oxidation state of Ce in an oxide is (+4) (thus CeO2 is more stable than Ce2O3 whilst in non-
oxides Ce3+
is more common), Equation 4.13 was used to calculate the formation energy of
CeBaCo2O5.
Ln2O3 + BaO +
Co3O4 → LnBaCo2O5 +
O2 (4.12)
CeO2 + BaO +
Co3O4 → CeBaCo2O5 +
O2 (4.13)
Chapter 4. Prediction and synthesis of GBCO related phases
113
4.3 Experimental methods
GdBaCo2O5+δ related materials with fluorite layer were prepared by solid state
synthesis. The synthesis conditions were based upon the reported ones for GdBaCo2O5+δ
materials62,138,220
were the sintering temperatures varied from 1000 to 1100 °C with the
different dwelling times (from 5 h to 20 h). The starting oxides for the synthesis with a listed
purity: Gd2O3 and Nd2O3 – 99.9%, BaCO3 – 99.99%, Co3O4 – 99.7%, and CeO2 – 99.99% all
purchased from Alfa Aesar, were measured out in the required stoichiometric amounts. The
inhomogeneous powder mixture of the starting oxides and carbonate was ground using a
pestle and a mortar. At the beginning of the synthetic work, LnCeBaCo2O7 (Ln = Gd, Nd)
materials were heated to 1000 °C for 12 h. Samples were then re-ground and pressed into
pellets and re-fired at 1000 °C for 12 h again. The samples were fired three times. The same
process was repeated for different synthesis temperatures (1050 and 1100 °C, see
Section 4.5). The same synthesis procedure was followed for the synthesis of Sr-analogues
(LnCeSrCo2O7) and materials with different fluorite layer (Ln2O3, Ln = Gd, Nd).
Phase compositions of all of the prepared samples were verified by PXRD collected at
room temperature using a Panalytical X'pert diffractometer using Co Kα1 radiation in Bragg-
Brentano geometry (Section 2.2.5). The PXRD data were measured over a 2θ range of 5-90 °
with step size 0.0334 ° and time per step 2 s. X'pert Highscore Plus software232
was used for
phase identification using the pdf-2 database.233
Pawley fits and Rietveld refinements
(Section 2.2.4) were performed by Topas academic program.152
4.4 Computational results
4.4.1 CeO2
At the first stage of the computational work each of the blocks (CeO2 and double
perovskites) were treated individually. Atomic coordinates of relaxed structures of the blocks
were then used for the main LnAECo2O7 phase. The unit cell of CeO2 used in DFT
relaxations is shown in Figure 4.2b. Different potentials (see Table 4.1 and Figure 4.4) were
applied. All the cells of CeO2 were relaxed using various energy cut-offs and different
Chapter 4. Prediction and synthesis of GBCO related phases
114
k‒points. The values of the lattice parameter a of the relaxed CeO2 cells are shown in Figure
4.4 and compared with the reported value.234
There is a significant difference ‒0.1 Å in the
case of the DFT relaxations using LDA potential. Better agreement with the reported data
was obtained from the calculations using PBE and PW91 potentials. The difference against
the experimentally obtained value is for both of the potentials in the third decimal place. The
calculated values are within the 1 percent range from the reported lattice parameter of CeO2.
Increasing the energy cut-off value from 400 to 440 eV improves the accuracy of the relaxed
lattice parameter. Further increase of the cut-off energy does not enhance the accuracy.
Energy cut-off 450 eV was used as a 'medium' setting value in INCAR file for the later DFT
relaxations.
Table 4.1: Lattice parameters of relaxed CeO2 obtained by DFT calculations using different potentials
compared with experimental data.234
Lit.234
PBE LDA PW91
a lattice parameter (Å) 5.47441(1) 5.475 5.371 5.476
Figure 4.4: Comparison of the lattice parameters obtained by DFT relaxation methods with the
experimentally obtained data using different potentials and different cut-off energy values.
Chapter 4. Prediction and synthesis of GBCO related phases
115
4.4.2 LnBaCo2O5
The values of lattice parameters obtained after the DFT relaxations of the double
perovskites LnBaCo2O5 (Ln = Gd, Ba) (Figure 4.2c) for both ferromagnetic and anti-
ferromagnetic structures are shown in Table 4.2. The values are compared with the reported
values for GdBaCo2O5 and NdBaCo2O5 respectively.235,236
Since the unit cell used in DFT
calculations was expanded by √2 (Section 4.2), the reported values of a and b lattice
parameters are multiplied by the same value of √2. The values of the lattice parameters of
GdBaCo2O5 obtained from DFT calculations are smaller than the reported (experimental)
ones.235
The difference for the ferromagnetic structure varies between 1-1.5% compared to
the reported structure. The values of the calculated lattice parameters of the anti-
ferromagnetic structure differ more with the 2.9% difference from the reported value235
for
the lattice parameter b. The lattice parameters obtained for NdBaCo2O5 structures are closer
to the reported values and all of them lie within sufficient 1% range. The lattice parameters a
and b are smaller than the reported ones,236
as in the case of GdBaCo2O5. The relaxed c
lattice parameter is on the contrary bigger than that obtained experimentally. The ratio
between the calculated and experimental value of a and b gives a number of 0.99, whilst for
lattice parameter c is the value of 1.01.
Table 4.2: Lattice parameters obtained after the DFT relaxation of LnBaCo2O5 materials, compared
with experimentally obtained data.235,236
Material a (Å) b (Å) c (Å)
GdBaCo2O5 − FM 5.504 5.504 7.474
GdBaCo2O5 − AFM 5.432 5.432 7.499
NdBaCo2O5 − FM 5.543 5.543 7.575
NdBaCo2O5 − AFM 5.557 5.557 7.607
GdBaCo2O5 − lit.235
3.955 (5.593) 3.934 (5.564) 7.540
NdBaCo2O5 − lit.236
3.961 (5.602) 3.930 (5.558) 7.533
Chapter 4. Prediction and synthesis of GBCO related phases
116
4.4.3 LnBaCo2O5.5
The values of the relaxed lattice parameters of LnBaCo2O5.5 for both ferromagnetic
and anti-ferromagnetic structures are shown in Table 4.3. Similarly to LnBaCo2O5
relaxations, the calculated values are compared to reported ones.210,237
Lattice parameter b is
multiplied by √2 due to the expanded unit cell used for the DFT calculations (Section 4.2).
The calculated lattice parameters b, c for both Gd and Nd containing phases are smaller than
the reported values. The values of the lattice parameter a obtained after DFT relaxations are
on the contrary bigger than the reported ones. The lattice parameters of the relaxed unit cell
of GdBaCo2O5.5 are in ± 1% from the experimental values. The lattice parameters of
NdBaCo2O5.5 after the DFT relaxations show less accuracy with experimentally obtained data
(± 2%).
Table 4.3: Lattice parameters obtained after the DFT relaxations of LnBaCo2O5.5 materials, compared
with experimental data.210,237
Material a (Å) b (Å) c (Å)
GdBaCo2O5.5 − FM 7.954 7.645 7.543
GdBaCo2O5.5 − AFM 7.938 7.660 7.569
NdBaCo2O5.5 − FM 8.037 7.666 7.577
NdBaCo2O5.5 − AFM 7.955 7.746 7.577
GdBaCo2O5.5 − lit.210
7.867 3.862 (7.724) 7.571
NdBaCo2O5.5 − lit.237
7.802 3.901 (7.802) 7.615
4.4.4 Double perovskite with fluorite layer
Previous sections showed the lattice parameters of the relaxed structures of the double
perovskites and the fluorite structures. The total energies of the individual blocks obtained
before are necessary for the calculation of the formation energies (Section 4.4.5). The model
for the further calculations with a structure combined of the perovskite double layer and the
fluorite CeO2 layer is mentioned and showed previously (Section 4.2, Figure 4.2c). The
lattice parameters obtained after the DFT relaxations are summarized in Table 4.4. All of the
Chapter 4. Prediction and synthesis of GBCO related phases
117
calculations were also done on Sr-analogues, where the Ba atom on A-site was completely
substituted by Sr (Table 4.5).
Table 4.4: Lattice parameters of calculated unit cells of LnCeBaCo2O7 materials obtained after the
DFT relaxations.
Material a (Å) b (Å) c (Å)
1-GdCeBaCo2O7 ‒ Gd-FM 5.523 5.523 20.688
2-GdCeBaCo2O7 ‒ Gd-AFM 5.448 5.448 20.478
3-NdCeBaCo2O7 ‒ Nd-FM 5.560 5.547 20.943
4-NdCeBaCo2O7 ‒ Nd-AFM 5.583 5.584 20.679
Table 4.5: Lattice parameters of calculated unit cells of LnCeSrCo2O7 materials obtained after the
DFT relaxations.
Material a (Å) b (Å) c (Å)
1-GdCeSrCo2O7 ‒ Gd-FM 5.468 5.458 20.649
2-GdCeSrCo2O7 ‒ Gd-AFM 5.492 5.494 20.475
3-NdCeSrCo2O7 ‒ Nd-FM 5.513 5.515 20.787
4-NdCeSrCo2O7 ‒ Nd-AFM 5.528 5.537 20.534
The electronic structure of GdCeBaCo2O7 material was studied by the density of
states (DOS) from GGA+U potential based on the optimized lattice structure. Figure 4.5
shows the spin-up and spin-down DOS of GdCeBaCo2O7 for both FM and AFM magnetic
ordering. A previous study238
on the electronic and magnetic structure of GdBaCo2O5.5 was
focused on several possible spin states and their relation to magnetic, electronic, and
structural properties. Our study was targeted on the role of CeO2 between the perovskite
blocks and its influence on the electronic structure of the material. Looking at the calculated
DOS (Figure 4.5a) of the FM ordered structure there is no band gap calculated. The expected
band gap region is overlapped by the Co d-orbitals. In contrast, anti-ferromagnetic
GdCeBaCo2O7 structure (Figure 4.5b) shows a band gap of 0.7 eV, with the valence band of
Co d orbitals. In general, LDA and GGA functional underestimate band gaps for
semiconductor and sometimes incorrectly predict a metal.239,240
This needs to be also
considered in the study of the density of state of GdCeBaCo2O7. It is worth mentioning very
Chapter 4. Prediction and synthesis of GBCO related phases
118
low lying Ce 4f states (Figure 4.5). This could be corrected by including a +U parameter for
Ce.
Figure 4.5: Density of states of GdCeBaCo2O7 material obtained from the relaxed structure with a)
ferromagnetic; b) anti-ferromagnetic ordering.
4.4.5 Formation energies
Equations for the calculation of the formation energies (Equations 4.1-4.4) were
selected to consider the various ways of forming a double perovskite with CeO2 fluorite
layer. Equations 4.1 and 4.2 give comparison with the starting building blocks: CeO2 and
double perovskite. Comparing the stability of LnAECo2O5 with LnAECo2O5.5 phase, the
phase with oxygen amount O5 is less stable than the one with O5.5 (for Ln = Gd, Nd). Thus
the values of the formation energies obtained from Equations 4.1 are expected to be more
negative that means LnCeAECo2O7 is more stable with respect to the O5 phase than the O5.5
phase. Equation 4.3 represents the calculation of the formation energy of LnCeAECo2O7
from binary oxides since they are starting materials for the solid state synthesis (Section 4.3).
Many reported studies with Ba and Ce containing materials showed chemical and thermal
stability of BaCeO3 phase and various cerates as well,241,242
although BaCeO3 exhibits an
instability in water containing atmospheres.243,244
Therefore, Equation 4.4 takes the stability
of cerates (AECeO3) into account.
Chapter 4. Prediction and synthesis of GBCO related phases
119
The values of the formation energies of LnCeBaCo2O7 materials are shown in Figure
4.6. The lowest formation energy values were calculated according to Equations 4.1 and 4.3
with the values of −2.4 and −2.8 eV/FU for anti-ferromagnetic Gd and Nd-containing
material (Equation 4.3) and −2.2 eV/FU for both ferromagnetic and anti-ferromagnetic
GdCeBaCo2O7 material (Equation 4.1). Formation energies compared with the LnBaCo2O5.5
(Equation 4.2) are more than 1 eV/FU higher than the ones compared with the LnBaCo2O5
(Equation 4.1) and are between −1 to −0.6 eV/FU for all of the calculated materials. The
highest values of formation energies were calculated using Equation 4.4 where the stability of
cerates was considered. All of the values are above 0 eV/FU and varies from +0.6 to +0.8
eV/FU. That predicts that the LnCeAECo2O7 compounds are unstable versus cerates
(AECeO3, AE = Ba, Sr), which is confirmed in Section 4.5.
Figure 4.7 displays the value of the formation energies for LnCeSrCo2O7 materials.
The formation energies calculated from Equation 4.1 were found to have the most negative
values with the most negative value −2.3 eV/FU for ferromagnetic NdCeSrCo2O7.
Comparison with the binary oxides (Equation 4.3) gives the formation energies 1 eV/FU
higher than the data obtained from Equation 4.1. Similar to the Ba-containing materials,
reaction enthalpies with LnSrCo2O5.5 are higher (approximately 1.5 eV/FU) than the ones
with the LnSrCo2O5 with the values between −0.5 and −0.6 eV/FU. The only positive values
are those obtained from the comparison with cerates (Equation 4.4) with the highest
formation energy +0.5 eV/FU for the ferromagnetic GdCeSrCo2O7.
Chapter 4. Prediction and synthesis of GBCO related phases
120
Figure 4.6: Formation energies of (Gd,Nd)CeBaCo2O7 materials, (1-4) see Table 4.4 for phase setting
calculated from the total energies obtained after the DFT relaxations by the Equations 4.1-4.4 in
Section 4.2.
Chapter 4. Prediction and synthesis of GBCO related phases
121
Figure 4.7: Formation energies of (Gd,Nd)CeSrCo2O7 materials, (1-4) see Table 4.5 for phase setting
calculated from the total energies obtained after the DFT the relaxations by the Equations 4.1-4.4 in
Section 4.2.
Previous formation energies included materials with one layer of the CeO2. A part of
the study of the GdBaCo2O5+δ related materials was also focused on the materials with more
layers of the fluorite CeO2 block. Table 4.6 shows the formation energies of the GdBaCo2O5
materials (with both ferromagnetic and anti-ferromagnetic ordering) with one and two layers
of the CeO2. Equations 4.5-4.8 (Section 4.2) were used to calculate the formation energies for
the materials with two layered fluorite block. The formation energies of the materials with
two layers of CeO2 are lower compared to one layered CeO2 structures for both magnetic
orderings and for most of the equations (Table 4.6). The improvement of the formation
energies varies from 1.3-1.5 eV/FU for both of the structures using Equations 4.5-4.7 (except
the GdBaCo2O5 material with AFM ordering for Equation 4.7 where the improvement was
noticed to be 0.4 eV/FU). The formation energies calculated from Equation 4.8 were the only
ones with an increase of the formation energies, +1.4 and +1.3 eV/FU for FM and AFM
ordering respectively. This is due to the stability of the BaCeO3, relevant also for the
materials with one layer of CeO2.
Summarizing all of the formation energies, all cases show that the cerate phase is
more stable than the phase with the CeO2 block within the structure. Even though the
Chapter 4. Prediction and synthesis of GBCO related phases
122
structures with the fluorite layer seem to be more stable than the binary oxides or the double
perovskite + fluorite, we might not be able to prepare these materials, at least not using a
synthetic method relying on thermodynamic stability, which would expect to form the
cerates.
Table 4.6: Comparison of the formation energies (in eV/FU) of the GdBaCo2O5 double perovskites
with one and two layers of CeO2. The energy values were calculated using corresponding equations
from the data obtained after the DFT relaxations of each of the materials.
Material Equation + 1 CeO2 layer + 2 CeO2 layers
GdBaCo2O5 FM 4.1/4.5 −2.204 −3.719
GdBaCo2O5 FM 4.2/4.6 −0.998 −2.513
GdBaCo2O5 FM 4.3/4.7 −1.248 −2.762
GdBaCo2O5 FM 4.4/4.8 +0.828 +1.390
GdBaCo2O5 AFM 4.1/4.5 −2.180 −3.498
GdBaCo2O5 AFM 4.2/4.6 −0.883 −2.282
GdBaCo2O5 AFM 4.3/4.7 −2.410 −2.844
GdBaCo2O5 AFM 4.4/4.8 +0.748 +1.308
4.5 Experimental results
Figure 4.8 shows the comparison of the PXRD of the GdCeBaCo2O7 materials
sintered to various temperatures (1000, 1050, and 1100 °C) with the main peaks of the
presented phases. In order to obtain the weight percentage (w%) of the phases, Rietveld
refinements were done (Table 4.7). For the material sintered to 1000 °C almost all of the
reflection of the PXRD pattern are indexed to the GdBaCo2O5.5 and the CeO2 phase with a
small amount (1-2 w%) of BaCoO3). For the material synthesized at higher temperatures, an
additional phase BaCeO3 phase (4-5 w%) is presented. There are no other peaks presented,
only those which could be assigned to a double perovskite + fluorite structure. The presence
of the double perovskite and the CeO2 phase indicates that the CeO2 was not introduced
between the blocks of the GdBaCo2O5.5. There are no changes in the phase compositions
comparing various sintering temperatures. The values of the main phases vary in the range
± 3 %.
Chapter 4. Prediction and synthesis of GBCO related phases
123
Figure 4.8: Comparison of the PXRD data collected on the GdCeBaCo2O7 materials synthesized at
various temperatures for 12 h with the main GdBaCo2O5.5 double perovskite phase. † indicates the
main reflections of the CeO2, ‡ represent the BaCeO3 reflection, and * denotes the main reflection of
the BaCoO3 phase.
Table 4.7: Phase fraction (w%) of the GdCeBaCo2O7 materials sintered three times at various
temperatures for 12 h. Phase percentages were obtained from the Rietveld refinements in Topas
against the PXRD data.
Sint. temp GdBaCo2O5.5 CeO2 BaCeO3 BaCoO3
1000 °C 71 27 0 2
1050 °C 68 28 5 < 1
1100 °C 71 24 4 1
Similar to the GdCeBaCo2O7 materials, their Nd-analogues were prepared at various
temperatures (1000, 1050 and 1100 °C) and characterized by the PXRD (Figure 4.9). The
percentages of the presented phases are shown in Table 4.8. NdBaCo2O5.7 and CeO2 are the
main phases for all of the synthesis temperatures with 5-8% of the BaCeO3 phase. There is no
extra peak which could be indexed to a perovskite + fluorite structure. NdCoO3 is the other
phase evident, especially abundant after the synthesis at 1050 °C. The percentage of the CeO2
Chapter 4. Prediction and synthesis of GBCO related phases
124
phase is constant for all of the sintering temperature, the amount of NdBaCo2O5.7 varies
depending on the presence of the NdCoO3 phase. The presence of the BaCeO3 is in
agreement with the calculated formation energies for these materials (Figure 4.7). The
formation energies calculated compared to the formation of cerates (Equation 4.4 in
Section 4.2) are positive ‒ predicting the formation of cerates which is also observed on the
PXRD data collected on the LnCeBaCo2O7 materials, Ln = Gd, Nd.
Figure 4.9: Comparison of the PXRD data collected on the NdCeBaCo2O7 materials synthesized at
various temperatures for 12 h with the main the NdBaCo2O5.7 double perovskite phase. † indicates the
main reflections of the CeO2, ‡ represent the BaCeO3 reflection, and * denotes the main reflections of
the NdCoO3 phase.
Chapter 4. Prediction and synthesis of GBCO related phases
125
Table 4.8: Phase fraction (w%) of the NdCeBaCo2O7 materials sintered three times at various
temperatures for 12 h. Phase percentages were obtained from the Rietveld refinements in Topas
against the PXRD data.
Sint. temp NdBaCo2O5.7 CeO2 BaCeO3 NdCoO3
1000 °C 61 25 8 6
1050 °C 58 22 6 14
1100 °C 71 24 5 0
DFT prediction of the LnCeSrCo2O7 materials showed more stability of these
compounds compared to cerates formation (Equation 4.4 in Section 4.2) than in the case of
the LnCeBaCo2O7 materials. PXRD data of the NdCeSrCo2O7 material with the presented
phases is shown in Figure 4.10. The percentages of the phases obtained after the Rietveld
refinements are displayed in Table 4.9. There are two main Co-containing phases presented:
perovskite NdCoO3 and Ruddlesden-Popper NdSrCoO4 phase with an additional cobalt
oxide. The amount of all of the Co-containing phases stays constant for all of the synthesis
temperatures. As in the previous synthesis with the Ba-analogues, CeO2 is not incorporated
into the perovskite structure and it is observed as a CeO2 phase with similar w% (23-27%) for
all of the sintering temperatures. According to the DFT calculations a formation of cerates
was expected to be a problem for the synthesis. Although the presence of the SrCeO3 was not
observed on the PXRD data after the synthesis, double perovskite material with the CeO2
fluorite layer was not synthesized due to the presence of the other Co-containing perovskite
or perovskite based phases.
Chapter 4. Prediction and synthesis of GBCO related phases
126
Figure 4.10: Comparison of the PXRD data collected on the NdCeSrCo2O7 materials synthesized at
various temperatures for 12 h with the main NdCoO3 phase. † displays the main reflection of the
CeO2, ‡ indicates the main reflections of the NdSrCoO4 phase, and * denotes the Co3O4 main
reflections.
Table 4.9: Phase fraction of the NdCeSrCo2O7 materials sintered three times at various temperatures
for 12 h. Phase percentages were obtained from the Rietveld refinements in Topas against the PXRD
data.
Sint. temp NdCoO3 NdSrCoO4 CeO2 Co3O4
1000 °C 67 2 27 4
1050 °C 52 18 24 6
1100 °C 59 13 23 5
4.6 Formation energies - different lanthanides
Previously mentioned work (see Sections 4.4.5 and 4.5) was focused on the double
perovskite materials with the CeO2 fluorite layer both from the theoretical and experimental
Chapter 4. Prediction and synthesis of GBCO related phases
127
point of view. Examples of other double perovskites with some of the different lanthanides
(Pr, Sm,220
La222
) were mentioned in the introduction of this chapter. The presented work
involved materials with Gd and Nd. For further study, other lanthanides were used for the
DFT relaxations of the LnBaCo2O5 structures to look at the other potential double perovskite
as a building block for materials with the fluorite layer in between. 'Medium' setting in VASP
described in Section 4.2 were used for the DFT calculations with the double perovskites.
The values of the reaction enthalpies (means also formation energies per formula unit,
in eV) of Equations 4.12 and 4.13 are displayed in Figure 4.11. As in previous plots with
formation energies, the lower value of the energy the more stable phase on the right side of
the equation. Comparing the formation energies of the selected lanthanide double
perovskites, the most stable ones are Nd and Eu materials followed by Gd, La, Pr, and Dy.
The presented values show only comparison between the stability and the presence of the
various lanthanides in LnBaCo2O5 structure. Variety of the oxygen content is typical for
double perovskite systems thus a material can be more stable in LnBaCo2O5+ δ with δ close
to 1.
Since the lowest energy LnBaCo2O5 structure is the Nd-containing material ‒ already
used for the study, another fluorite block is going to be used for the further work.
Figure 4.11: Formation energies (calculated according the Equations 4.12, 4.13 in Section 4.2) of
LnBaCo2O5 structures with the different lanthanides.
Chapter 4. Prediction and synthesis of GBCO related phases
128
4.7 Other fluorite layer
4.7.1 Formation energies
Both theoretical and experimental study of the LnCeAECo2O7 materials (Ln = Gd,
Nd, and AE = Ba, Sr) showed the problem with the synthesis of these materials involving the
stability of cerates. In order to avoid this different fluorite blocks (Gd2O3 and Nd2O3
respectively) were used. The theoretical part of the study followed a similar procedure to that
applied to the double perovskites with the CeO2 fluorite layer.
Formation energies (Equations 4.9-4.11) were selected to compare the stability of the
Ln2BaCo2O7 materials versus LnBaCo2O5, LnBaCo2O5.5 and binary oxides. Table 4.10 shows
the values of the lattice parameters after the DFT relaxations. The values are compared with
the reported values of RP2 Gd2SrCo2O7 phase231
since the presence of a RP2 phase was
observed on the PXRD data collected after the synthesis of the Nd2SrCo2O7 sample
(Section 4.7.2). The values of lattice parameters of calculated structures are higher than the
reported which is due to the higher ionic radius of Ba2+
(1.61 Å) compared to Sr2+
(1.44 Å).245
The values of formation energies are shown in Figure 4.12. All of the values are
negative, that means materials on the right side of the Equations 4.9-4.11 are more
thermodynamically stable than the materials on the left side. The lowest values are calculated
for the formation energies compared to the LnBaCo2O5 double perovskites. As it was
mentioned before, LnBaCo2O5 phases are less stable than the LnBaCo2O5.5 and thus the
formation energies with the LnBaCo2O5 included are expected to be lower. Formation energy
values compared to binary oxides were approximately 0.5 eV/FU higher than these obtained
after the comparison to LnBaCo2O5 phases. The higher values of formation energies
(between −2.4 and −3 eV/FU) were obtained from the comparison of Ln2BaCo2O7 versus
LnBaCo2O5.5 (Equation 4.10).
The stability of the Ln2BaCo2O7 materials with several possible phases was
compared. Both ferromagnetic and anti-ferromagnetic ordering together with RP2 were taken
into account as a potential structural form. The formation energy values favour the formation
of the Ln2BaCo2O7 phases. The next chapter is focused on the experimental results obtained
after the solid state synthesis of the Ln2BaCo2O7 materials.
Chapter 4. Prediction and synthesis of GBCO related phases
129
Table 4.10: Lattice parameters obtained after the DFT relaxations of the Ln2BaCo2O7 materials.
Material a (Å) b (Å) c (Å)
1- Gd2BaCo2O7 − FM 5.473 5.473 20.512
2 - Gd2BaCo2O7 − AFM 5.483 5.483 20.631
3 - Gd2BaCo2O7 − RP2 5.435 5.435 19.612
4- Nd2BaCo2O7 − FM 5.651 5.675 21.035
5 - Nd2BaCo2O7 − AFM 5.654 5.649 21.037
6 - Nd2BaCo2O7 − RP2 5.438 5.438 20.095
Gd2SrCo2O7 – lit.231
5.37506(2) 5.37506(2) 19.35807(6)
Figure 4.12: Formation energies of the Ln2BaCo2O7 materials, Ln = Gd, Nd, (1-6) see Table 4.8 for
the phase composition, calculated from the total energies obtained after the DFT relaxations from
Equations 4.9-4.11 in Section 4.2.
Chapter 4. Prediction and synthesis of GBCO related phases
130
4.7.2 Experimental results
PXRD data of the Gd2BaCo2O7 material collected after the synthesis at various
temperatures is shown in Figure 4.13. There is no difference in the collected patterns. Table
4.11 shows the percentage of presented GdBaCo2O5.5 and Gd2O3 phases obtained after the
Rietveld refinements (Figure 4.14). There are no additional reflections in the PXRD data.
The phase fraction 70 : 30 of double perovskite to Gd2O3 phase remains constant for all of the
temperatures. This fact is in contradiction to the prediction of the stability based on DFT
calculations mentioned in Section 4.7.1. The values of the formation energies compared with
the GdBaCo2O5.5 were calculated to be close to −2.6 eV/FU and thus a formation of a
Gd2BaCo2O7 phase was expected.
Figure 4.13: PXRD data collected on the Gd2BaCo2O7 material sintered three times at various
temperatures for 12 h with the main double perovskite GdBaCo2O5.5 phase. The main reflections of
the Gd2O3 are labelled with †.
Chapter 4. Prediction and synthesis of GBCO related phases
131
Table 4.11: Phase fraction (w%) of the Gd2BaCo2O7 material sintered three times at various
temperatures for 12 h. Phase percentages were obtained from the Rietveld refinements of the PXRD
data in Topas.
Sint. temp GdBaCo2O5.5 Gd2O3
1000 °C 70 30
1050 °C 70 30
1100 °C 71 29
Figure 4.14: Rietveld refinement of the Gd2BaCo2O7 material sintered three times at 1000 °C. First
line of reflections corresponds to the GdBaCo2O5.5 phase and the second one to the Gd2O3,
GOF = 1.445.
Nd2BaCo2O7 materials were prepared at the same time and using the same synthesis
conditions as for the Gd2BaCo2O7 synthesis. The materials were characterized by Rietveld
refinement (Figure 4.16) against the PXRD data (Figure 4.15) collected after the solid state
synthesis with different sintering temperature. The percentages of the presented phases are
displayed in Table 4.12. The Ruddlesden-Popper phase (RP2) Nd2BaCo2O7 is the main phase
for all of the sintering temperature with the abundance of 85-92 w%. No crystallographic data
for the Nd2BaCo2O7 RP2 phase have been reported. All of the reflections of the main phase
were indexed using the symmetry and the data of the reported Gd2SrCo2O7 phase.231
This
phase was also used as a starting model for the Rietveld refinements. The rest of the
Chapter 4. Prediction and synthesis of GBCO related phases
132
reflections are indexed to NdBaCo2O5.7 double perovskite and the Nd2O3 phase. According to
the calculated values of formation energy (Figure 4.12) a formation of the Nd2BaCo2O7
phase is expected. The values of formation energy of ferromagnetic, anti-ferromagnetic
ordered and RP2 structures are within a close range.
Figure 4.15: PXRD data collected on the Nd2BaCo2O7 material sintered three times at various
temperatures for 12 h with the Ruddlesden-Popper (RP2) main phase. The main reflection of the
double perovskite NdBaCo2O5.7 are marked with †, the Nd2O3 reflections are labelled with ‡.
Table 4.12: Phase fraction (w%) of the Nd2BaCo2O7 material sintered three times at various
temperatures for 12 h. Phase percentages were obtained from the Rietveld refinements of the PXRD
data in Topas.
Sint. temp NdBaCo2O5.7 Nd2O3 RP2 phase
1000 °C 10 2 88
1050 °C 12 3 85
1100 °C 5 3 92
Chapter 4. Prediction and synthesis of GBCO related phases
133
Figure 4.16: Rietveld refinement of the Nd2BaCo2O7 material sintered 3 times at 1000 °C with the
presented phases as follow: RP2-phase ‒ 1st line of the reflections, NdBaCo2O5.7 ‒ 2
nd line, Nd2O3 ‒
3rd
line, GOF = 1.677.
4.8 Discussion and conclusions
In summary, it has been shown that DFT methods can be used to calculate the
formation energies of perovskite based materials from binary and ternary oxides. The method
was used as a prediction to guide for the synthetic isolation of double perovskite
GdBaCo2O5+δ related structures with a fluorite layer. Final structures were built from the
individual relaxed blocks. The data of the starting structures obtained from DFT calculations
were compared with the reported data.210,235-237
Both FM and AFM magnetic ordering was
used for the calculations. Several possible structures were taken into account. The synthesis
of the most stable materials (according calculated formation energies) was attempted. The
selection based on DFT calculation was predictive and time saving tool for phase screening
of selected structural types.
Calculated formation energies were based on various reaction including binary oxides
and double perovskites. For the prediction of the stability of materials is important to take
Chapter 4. Prediction and synthesis of GBCO related phases
134
into account all the stable oxides/materials. An unstable material (e.g. GdBaCo2O5 in our
prediction) can result in very negative values of formation energies. Since the work was
followed by the experimental work, few changes for the formation energy calculations were
done after the first steps of the solid state synthesis. All of the collected PXRD patterns
confirmed the presence of the stable cerates. Additional calculated formation energies
showed higher positive values, which was in an agreement with the experimentally obtained
data.
PXRD data obtained after the solid state synthesis of LnCeAECo2O7 (Ln = Gd, Nd;
AE = Ba, Sr) showed the presence of the double perovskite and the CeO2, with an additional
cerate phase. Hence, no CeO2 was introduced between the double perovskite layers. A charge
imbalance between the CeO2 fluorite layer and the double perovskite layer represents maybe
another problem for the synthesis of these materials.
As a minor part of the DFT study, electronic structure of the structure relaxed
materials with the fluorite layer was studied to see the influence of the CeO2 layer for the
conduction or the valence band. Since other factors (magnetic ordering, spin states) play an
important role in the electronic structure, the effect of the additional CeO2 layer cannot be
explained individually. Structures with the two fluorite layers were built and their formation
energies were calculated. The values follow the same trend as it was observed in the materials
with the one layer of CeO2.
Since the materials with the CeO2 fluorite layer could not be synthesized, another
fluorite Ln2O3 (Ln = Gd, Nd) layer was used for both DFT prediction and solid state
synthesis to avoid the formation of cerates. Calculated formation energies for Ln2BaCo2O7
show that the existence of the structures is favourable. The comparison between included
Ln2BaCo2O7 structural types (RP2 phase and double perovskite + fluorite block) showed
similar values of formation energies. There was no Gd2BaCo2O7 phase synthesized
(GdBaCo2O5.5 and Gd2O3 phases were only presented), while collected PXRD data for Nd-
containing material indicates the presence of a RP2 phase. Gd2BaCo2O7 appearance did not
follow the energetic prediction based on DFT calculations, whilst Nd2BaCo2O7 agreed with
the favourability of RP2 phase compared to double perovskite structure.
In conclusion, DFT calculation methods were used for material synthesis prediction
based on energetics comparison of selected possible phases in order to prepare GdBaCo2O5+δ
related structures with a fluorite layer. A charge misbalance between the CeO2 fluorite layer
and the double perovskite layer was overcome in materials with fluorite layers based on
Chapter 4. Prediction and synthesis of GBCO related phases
135
Ln2O3 (Ln = Gd, Nd). The data of experimentally prepared phases indicate the presence of a
RP2 phase. The formation of a RP2 phase does not meet our main goal – to synthesize a
double perovskite structure with a fluorite inter-layer, but the Nd2BaCo2O7 phase could be a
candidate for further crystallographic and material properties study.
5 Ruddlesden-Popper phases ‒ stannates
5.1 Introduction
Ruddlesden-Popper phases (general formula An+1BnO3n+1, structure represented in
Figure 5.1) represent promising structures for a SOFC cathode application due to their ability
to host oxide vacancies in perovskite layer (such as observed in Sm2BaCo2O7−δ)246
and oxide
interstitial in the rock salt layer (with an example of La1.5+xSr0.5−xCo0.5Ni0.5O4+δ).144
Recent
studies involved mainly the A2NiO4+δ (A = La, Sr) phase56,141-143
and are described in details
in Section 1.7.1. This work is focused on the n = 1 (RP1) stannate (Sr2SnO4) phases doped on
A- or B-site, where the A-site Sr, Ba atoms are substituted by La and in the case of B-site
doping, Sn atom is doped by M5+
(Nb, Ta) and M6+
(Mo, W) elements, in order to introduce
mobile interstitial oxide ions into a non-metallic system. The target of this work is to find a a
pure ionic conductor with a potential use as an electrolyte for SOFCs, thus any open-shell
transitional metals are not suitable for the consideration. Those transitional metals would
increase electronic conductivity, which is detrimental to any SOFC electrolyte application.
Figure 5.1: Schematic of Ruddlesden Popper Srn+1TinO3n+1 phases with different number of
perovskite layer within the structure: a) n = 1; b) n = 2; c) n = 3. Sr atoms are in green, Ti in blue and
O atoms are in red.
Chapter 5. Ruddlesden-Popper phases − stannates
137
The structure of Sr2SnO4 (Figure 5.2) was described as a K2NiF4-type.247,248
Later,
high resolution neutron powder diffraction studies249,250
clarified the relationship between the
SnO6 octahedral tilts and the structural description of the material. The Rietveld refinements
showed that the structure Sr2SnO4 at room temperature is well described using Pccn space
group, which is a subgroup of the both Bmab and P42/ncm. The structure can be derived from
the K2NiF4 structural type by tilting of the SnO6 octahedra along the a- and b-axis and with
the different values of tilting angles: α ≠ β ≠ 0.249
According to a variable temperature
neutron diffraction study there are two phase transition observed.250
At temperatures above
423 K, the structure changes to another orthorhombic structure, Bmab, retaining both tilts but
with equal angles α = β ≠ 0. At temperatures higher than 573 K there are no tilts and the
structure is described in the tetragonal I4/mmm space group.
Figure 5.2: Structure of Sr2SnO4 described in Pccn space group,249
with Sr (green) on A-site and Sn
(grey) in octahedral ordering on B-site, oxygen ions are labelled with red: a) view along c-axis; b)
view along a-axis.
The work presented in this chapter contains both experimentally and computationally
obtained data of A- and B-site doped Sr2SnO4 phases. The computational part deals with
Chapter 5. Ruddlesden-Popper phases − stannates
138
formation energies of doped RP1 phases e.g. stannates, hafnates, and zirconates, to identify
favourable doping strategies. The experimental part is focused on Nb- and Ta-doped Sr2SnO4
materials. Prepared materials are structurally characterized by PXRD techniques. The
conductivity of single phase materials are measured by AC impedance spectroscopy and
discussed later on together with the electronic structure studies by spectroscopic methods
(UV-vis, IR, and solid state Sn-NMR).
5.2 Computational methods
All of the calculations were performed using the plane wave DFT package, Vienna
Ab-initio Simulation Package (VASP) version 4.6.26178,224
with Perdew, Burke and
Ernzerhof (PBE) exchange correlation functional.225
For the A-site atoms (strontium, barium)
and hafnium, the first sub-valent s orbital was treated as valence, as were d orbital for tin and
p orbital for the rest of the atoms. A k-point 3 × 3 × 3 grid was used for the DFT calculations.
The unit cell size and atomic co-ordinates were relaxed until forces on atoms were less than
0.01 eV/Å. The relaxation process was stopped when the energy difference between two
steps was smaller than 10−5
eV. The cut-off energy for plane wave was set to 450 eV.
The coordinates of initial un-doped phases were used from the reported data (for
stannates,250
for hafnates and zirconates251
). The lattice parameters a and b of the unit cells
used for the DFT relaxations were doubled. This was necessary for a lower doping level (1/8
and 1/16 respectively) on both A- and B-site. In the case of M6+
doping (Mo, W) one atom of
tin was replaced by M5+
element (doping level 1/16). Increased positive charge was
compensated by introducing one interstitial O atom (Oint) placed in the SrO rock salt layer.
The coordinates of Oint used where those reported for the mixed conductor
La1.5+xSr0.5−xCo0.5Ni0.5O4+δ.144
When doping of M5+
was carried out, two atoms of tin were
replaced by M5+
(doping level 1/8). Structures with various positions of the doping M5+
elements were taken into account. The relaxed structure with the lowest total energy was used
for the formation energy calculation. Similarly to M6+
doping, doping level on A-site was 1/8.
Two atoms of tin were replaced by La. Several possible configurations of La atoms were used
for DFT relaxations.
The values of the total energies of the doped Ruddlesden-Popper phases were used to
calculate the formation energies from the starting un-doped phases. Equations 5.1-5.3 were
Chapter 5. Ruddlesden-Popper phases − stannates
139
used to calculate the formation energies of either A- and B-site doped materials. The
formation energies were calculated in electron Volts per Formula Unit, eV/FU. The more
negative the values obtained, the more stable were the structures on the right side of the
equations.
Sr2BO4 +
M2O5 −
SnO2 → Sr2B7/8M1/8O4+1/16 (5.1)
Sr2BO4 +
MO3 −
SnO2 → Sr2Sn15/16M1/16O4+1/16 (5.2)
Sr2BO4 +
La2O3 −
SrO → Sr2−1/8La1/8BO4+1/16 (5.3)
B = Sn, Hf, Zr
Previous equations show the formation energies of the structures with Oint within the doped
structure. Doped structures of the stannates without the Oint atom (with the tin mixed valence
oxidation state of Sn2+
/Sn4+
) were also relaxed. The formation energies of these phases were
calculated using Equation 5.4-5.6 and compared with the ones with Oint included.
Sr2SnO4 +
M2O5 −
SnO2 → Sr2Sn7/8M1/8O4 +
O2 (5.4)
Sr2SnO4 +
MO3 −
SnO2 → Sr2SnM1/16O4 +
O2 (5.5)
Sr2SnO4 +
La2O3 −
SrO → Sr2La1/8SnO4 +
O2 (5.6)
5.3 Experimental methods
Parent phase of Sr2SnO4 was synthesized by a solid state method.249,250
Sr2SnO4 was
prepared from powders of SrCO3 and SnO2. The mixture was ground and heated in an
alumina crucible in air at 1177 °C. The undoped phase was heated for one week with
repeated regrinding with a pestle and a mortar every day following the literature synthesis
protocol.249,250
Synthesis conditions of Sr2SnO4 were also followed for the doped materials.
Starting doping levels x = 0.05 and 0.10 were selected for both A- (La) and B-site (Nb, Ta,
Mo, W) doped Sr2SnO4. The starting oxides for the synthesis with a listed purity: SnO2,
SrCO3, La2O3 ‒ 99.99 %; MoO3 ‒ 99.95 %; Nb2O5, Ta2O5, ZrO2 ‒ 99.9 % all purchased from
Alfa Aesar and WO3 ‒ 99.995% purchased from Sigma-Aldrich, were measured out in the
Chapter 5. Ruddlesden-Popper phases − stannates
140
required stoichiometric amounts. The inhomogeneous powder mixture of the starting oxides
and carbonate was ground using a pestle and a mortar. In the case of La-, Nb-, and Ta-doped
materials the initial synthesis produced RP1 phases with significant amounts of binary oxides
(La2O3, Nb2O5, and Ta2O5) as confirmed by the PXRD data. Synthesis conditions were
improved by experimenting with temperature, mixing technique (using a planetary ball mill
and reaction atmospheres as shown in Section 5.5.1). Final synthesis conditions for the single
phased B-site doped Sr2SnO4 were as follows: heating at 1250 °C in O2 atmosphere for Nb-
doping and heating in ambient air at 1300 °C for the Ta-doped materials. La-, Mo-, and W-
doped materials could not be prepared as single phases (even after modification of the
synthesis conditions) and thus the work was focused on the Nb- and Ta-doped phases.
The phase composition of all of the prepared samples was verified by powder PXRD
collected at room temperature using a Phillips X'pert Panalytical diffractometer using Co Kα1
radiation in Bragg-Brentano geometry. X'pert Highscore Plus software232
was used for phase
identification using the pdf-2 database.233
Pawley fits and Rietveld refinements were
performed by Topas academic program.152
5.4 Computational results
All of the studied Ruddlesden-Popper phases were relaxed using the DFT methods.
Total energy values obtained after the DFT relaxations were used to calculate the formation
energies using Equations 5.1-5.3. Figure 5.3 shows the values of the formation energies for
the A- and B-site doped stannates (Sr2SnO4), zirconates (Sr2ZrO4) and hafnates (Sr2HfO4).
The same trend of binding energy values is observed for all of the studied phases. The lowest
values of the formation energies are those for the La-doped materials with the values
−1.2 eV/FU. The more negative the value calculated, the more stable is the doped material
expected to be. All of the formation energies for M5+
doped materials (except that for the
Sr2Hf7/8Nb1/8O4+1/16) were found to have negative values. The highest values of the formation
energies (and thus expected to be least stable) were calculated for the M6+
doped phases, with
the positive values from +0.02 to +0.3 eV/FU. Comparing the values of all doped structures
for all three phases, the most stable are doped zirconates phases, followed by stannates and
hafnates.
Chapter 5. Ruddlesden-Popper phases − stannates
141
Figure 5.3: Formation energies of the doped Sr2(Sn,Zr,Hf)O4 materials calculated from the total
energies obtained after the DFT relaxations by the Equations 5.1-5.3 in Section 5.2.
Most of the work presented in this chapter is focused on the stannate structures.
Figure 5.4 shows a comparison of the formation energies of the Sr2SnO4 phases compared to
the Ba2SnO4. Calculated formation energies of Ba2SnO4 derivatives (except for Nb-doping)
are higher than the ones of Sr2SnO4 phases for both A- and B-site doped materials. As it was
mentioned in Section 5.2, the Sr2SnO4 phases were also relaxed without the Oint, with the
formation energies shown in Figure 5.4. The formation energy values of the phases without
Oint are almost identical to the ones with Oint for M5+
and La-doping. That means that the
stability of the phases with and without Oint with respect to the parent undoped RP phase is
very similar. In the case of M6+
doping, the formation energies of the structures without Oint
are lower (by 0.05 for W and 0.1 eV/FU for Mo doping) than those obtained for the materials
with interstitial oxygen.
Chapter 5. Ruddlesden-Popper phases − stannates
142
Figure 5.4: Formation energies of the doped Sr2SnO4 materials calculated from the total energies
obtained after DFT relaxations by the Equations 5.1-5.3 in Section 5.2 compared with the doped
materials without Oint (Equations 5.4-5.6, Section 5.2) and with the Ba2SnO4 materials.
5.5 Structural characterization
5.5.1 Laboratory P-XRD
A- and B-site doped Sr2SnO4 materials were prepared by solid state synthesis
described in Section 5.3. All of the synthesized stannates were refined using the orthorhombic
Pccn space group with the reported refined atomic positions.249
Figure 5.5 shows the
Rietveld refinement of the Sr2Sn0.95Nb0.05O4 material. Changes of the intensity of the
Sr4Nb2O9 impurity main peak were found to vary depending on the synthesis conditions.
Since the materials with starting doping levels x = 0.05 and 0.1 were synthesized with an
impurity phase, materials with lower doping levels were also prepared. For Nb-doped
Sr2SnO4 materials, the highest doping level of phase pure sample was obtained for the
x = 0.03 in Sr2Sn0.97Nb0.03O4. Maximum Ta doping for single phased stannates was achieved
for x = 0.04 in Sr2Sn0.96Ta0.04O4. PXRD data of the Ta-doped materials with x > 0.4 showed a
Chapter 5. Ruddlesden-Popper phases − stannates
143
presence of Sr-Ta-O oxide (found in ICSD database with structural formula of
Sr1.907Ta0.593O2.73) secondary phase.
La-doped materials were not prepared as single phases, as a presence of La2O3 impurity was
confirmed even on PXRD data for low La doping (Sr1.97La0.03SnO4 material, Figure 5.6).
Neither Mo- or W-doped Sr2SnO4 materials were prepared as single phases due to the
presence of the Sr3(Mo,W)O6 impurity phases; these compounds were not considered in the
theoretical DFT study.
Figure 5.5: Rietveld refinement of the Sr2Sn0.95Nb0.05O4 material, with a presence of the Sr4Nb2O9
impurity phase.
Chapter 5. Ruddlesden-Popper phases − stannates
144
Figure 5.6: Rietveld refinement of the PXRD data obtained after the synthesis of the Sr1.97La0.03SnO4
material at 1300 °C with the La2O3 impurity phase.
Lattice parameters of the Nb-doped Sr2SnO4 materials obtained after the Rietveld
refinement against the laboratory PXRD data are shown in Figure 5.7 and in Appendix B. The
data were collected using a powder of the synthesized material mixed in 1:1 ratio with KCl as
an internal standard. There is a step-like increase of all lattice parameters observed for the
samples with x ≥ 0.03. This increment, more significant for the c lattice parameter, could be
explained due to the presence of the Sr4Nb2O9 impurity phase (Figure 5.5) for the materials
with higher Nb doping level or due to a change in charge compensation mechanism after the
doping with Nb5+
(presence of Oint or Sn2+
/Sn4+
valence state). A significant decrease of a/c
ratio is observed, showing two regions for x < 0.03 and x > 0.03 with the different values
(Figure 5.7d). The same trend of the lattice parameter values is observed for the Ta-doped
Sr2SnO4 materials (Figure 5.8 and Appendix C). The composition of the Sr2Sn0.96Ta0.04O4
material was studied by the Energy dispersive X-ray (EDX) diffraction. The incorporation of
the Ta atom within the RP1 structure was confirmed by multiple measurements. The average
ratio between the A- and B-site atoms: Sr1.934(4)Sn1.029(4)Ta0.04(2) is in good agreement with the
expected stoichiometry (Table of EDX analysis of Sr2Sn0.96Ta0.04O4 is shown in Appendix A) .
Chapter 5. Ruddlesden-Popper phases − stannates
145
Figure 5.7: Lattice parameter evolution of the Sr2Sn1−xNbxO4 materials; values were obtained after
Rietveld refinements of laboratory PXRD data with KCl as an internal standard: a) lattice parameters
a, b; b) lattice parameter c; c) cell volume; d) a/c ratio.
Chapter 5. Ruddlesden-Popper phases − stannates
146
Figure 5.8: Lattice parameter evolution of the Sr2Sn1−xTaxO4 materials; values were obtained after
Rietveld refinements of laboratory PXRD data with KCl as an internal standard: a) lattice parameters
a, b; b) lattice parameter c; c) cell volume; d) a/c ratio.
5.5.2 Synchrotron data
The previous section showed the laboratory PXRD data. Sr2Sn0.97Nb0.03O4 and
Sr2Sn0.96Ta0.04O4 materials were studied using high resolution synchrotron X-ray techniques
and high resolution neutron diffraction (HRPD, Section 5.5.3). Synchrotron experiments were
carried out using the I11 instrument at the Diamond Light Source (UK). The data were
collected at the room temperature with incident wavelength λ = 0.827157 Å. Small amounts
of the Sr2Sn0.97Nb0.03O4 and Sr2Sn0.96Ta0.04O4 materials were loaded into 0.2 mm diameter
borrosilicate capillary and sealed with a gas/oxygen torch. Measurements were taken from
spinning capillaries to improve the powder average. Lattice parameters and atomic positions
obtained after the Rietveld refinement of the I11 data were used as starting data for the joint
refinements (Section 5.5.3).
Chapter 5. Ruddlesden-Popper phases − stannates
147
Several models for the Rietveld refinement of the doped Sr2SnO4 materials were
applied. The parent Sr2SnO4 phase is described in two orthorhombic (Pccn, Bmab) and
tetragonal (P42/ncm) space groups due to the different tilting angles around the a- and b-
axes.250
At the room temperature, the phase is described in Pccn space group, which was used
for the starting model. The refinement was also performed in orthorhombic supergroup
(Bmab). A better refinement was achieved using Pccn space group (Figure 5.9) with all peaks
of Ruddlesden-Popper Sr2SnO4 phase indexed. However, the starting model could not fit the
data correctly since the peak anisotropy was observed.
Several peak shape functions for Sr2SnO4 were applied, the best results are given by
orthorhombic Stephens Anisotropic Peak Broadening. A closer look at the refinement shows
additional peaks (inset of Figure 5.9) and broader reflections of the main phase. The
broadening of the peaks can be explained by stacking faults (observed in several Ruddlesden-
Popper phases)252-255
and by a presence of other compound of Ruddlesden-Popper series. The
position of low angle peaks was crucial for impurity phases determination.
Figure 5.9: Rietveld refinement of the Sr2Sn0.96Ta0.04O4 material in Pccn space group, Rwp = 3.554%.
The inset shows magnified area of additional reflections found to be indexed for SrSnO3 perovskite
(P) and Sr3Sn2O7 Ruddlesden-Popper n = 2 phase (RP2).
Chapter 5. Ruddlesden-Popper phases − stannates
148
The previous refinement (Figure 5.9) was improved after including of SrSnO3
perovskite and Sr3Sn2O7 RP2 phases (Figure 5.10). All of the extra reflections are indexed
using these two phases and since the main peaks of these phases overlay with the main
Sr2SnO4 reflections, anisotropic peak broadening (due to stacking faults) was fitted. SrSnO3
phase is known to exist in several space groups a Pbnm space group was used in Pawley fit.
The final Rietveld refinement of Sr2Sn0.96Ta0.04O4 with SrSnO3 and Sr3Sn2O7 as secondary
phases, is shown in Figure 5.10. The refined parameters of the main RP1 phase are displayed
in Table 5.1 whilst the refined lattice parameters are compared with the parent Sr2SnO4 phase
in Table 5.3. The occupancy of Ta was fixed to nominal value of 4%.
Figure 5.10: Final refinement of I11 data of Sr2Sn0.96Ta0.04O4 in Pccn space group with secondary
phases of SrSnO3 and Sr3Sn2O7; Rwp = 2.292%.
Chapter 5. Ruddlesden-Popper phases − stannates
149
Table 5.1: Refined parameters of the Sr2Sn0.96Ta0.04O4 sample from the data collected from I11
beamline.
Atom Site x y z Beq (Å2) Occ
Sn1 4a 0 0 0 0.392(9) 0.96
Ta1 4a 0 0 0 0.392(9) 0.04
Sr1 8e −0.0050(4) 0.0014(7) 0.35184(2) 0.297(9) 1
O1 4c 0.250 0.250 0.016(1) 0.74(8) 1
O2 4d 0.750 0.250 0.006(1) 0.43(7) 1
O3 8e 0.027(2) −0.025(2) 0.1656(2) 0.71(8) 1
The I11 data refinement of Sr2Sn0.97Nb0.03O4 followed the same procedure as it was
described for Sr2Sn0.96Ta0.04O4, with the refinement of the main Ruddlesden-Popper phase.
Similar to previous refinements, the occupancy of Nb was fixed to nominal value of 3 %. The
best refinement of the main phase was obtained using Pccn space group. Similarly to
Sr2Sn0.96Ta0.04O4, a broadening of the main reflection was observed in Nb-doped material,
while the extra peaks were not observed. The refinement with SrSnO3 and Sr3Sn2O7 phases
(Figure 5.11) significantly decreased the Rwp values (1.812% compared to 2.473%) and
improved the fitting of the main RP peaks. Table 5.2 shows the refined parameters of
Sr2Sn0.97Nb0.03O4.
Chapter 5. Ruddlesden-Popper phases − stannates
150
Figure 5.11: Rietveld refinement of the Sr2Sn0.97Nb0.03O4 sample in Pccn space group with secondary
phases of SrSnO3 and Sr3Sn2O7; Rwp = 1.812%.
Table 5.2: Refined parameters of the Sr2Sn0.97Nb0.03O4 sample from the data collected from I11
beamline.
Atom Site x y z Beq (Å2) Occ
Sn1 4a 0 0 0 0.77(1) 0.97
Nb1 4a 0 0 0 0.77(1) 0.03
Sr1 8e −0.0021(7) 0.0024(1) 0.35213(2) 0.148(9) 1
O1 4c 0.250 0.250 0.0187(4) 0.73(8) 1
O2 4d 0.750 0.250 −0.003(1) 0.76(7) 1
O3 8e 0.0011(2) −0.0359(8) 0.1656(2) 0.87(6) 1
Chapter 5. Ruddlesden-Popper phases − stannates
151
Table 5.3: Refined lattice parameters of the Sr2Sn0.97Nb0.03O4 and Sr2Sn0.96Ta0.04O4 material from I11
data after the Rietveld refinement in Topas and their comparison with the undoped Sr2SnO4 phase.
Material a (Å) b (Å) c (Å) V (Å)
Sr2SnO4 250
5.72898(5) 5.73524(5) 12.58110(6) 413.378
Sr2Sn0.97Nb0.03O4 5.73504(6) 5.74146(6) 12.65833(7) 416.807(7)
Sr2Sn0.96Ta0.04O4 5.73495(13) 5.74169(13) 12.66376(24) 416.99(1)
5.5.3 High Resolution Powder Diffraction data
The solid state synthesis of Sr2Sn0.97Nb0.03O4 and Sr2Sn0.96Ta0.04O4 materials was
scaled up in order to synthesise larger amount of the materials for the HRPD analysis.
Approximately 5 g of the Sr2Sn0.97Nb0.03O4 and Sr2Sn0.96Ta0.04O4 materials were loaded into a
vanadium 8 mm diameter can and measured at ambient temperature by time-of flight neutron
diffraction for the time 3 h. Data were collected using all three detector banks (35 °, 90 °, and
145 °). Magnetism in these compounds is irrelevant, thus the data of first two banks only
(35 °, 90 °) were used in a joint Rietveld refinement with the I11 data. Twelve background
parameters, scale factors and sample-dependent peak shapes were refined per bank.
A representative refinement of the Sr2Sn0.97Nb0.03O4 and Sr2Sn0.96Ta0.04O4 sample is
displayed in Figure 5.11 and Figure 5.13 respectively. Starting model obtained from I11 data
analysis (orthorhombic Pccn space group of Sr2SnO4 phase) was applied for both of the
refinements. All of the reflections of RP1 phase were observed and fitted, although some
additional reflections and broadening of several peaks of the main phase were noticed, similar
to I11 data analysis. Refinements with refined coordinates of RP1 phase only, obtained from
I11 fits, gave imperfect with Rwp = 2.871% for Sr2Sn0.97Nb0.03O4 and Rwp = 4.163% for
Sr2Sn0.96Ta0.04O4. Initially, models with one phase only were applied to improve the joint
refinements. Inspection of Fourier maps showed regions of negative nuclear density where
interstitial oxygen could be placed. Several positions of Oint were tried all together with the
Oint position evidenced in La1.5+xSr0.5−x Co0.5Ni0.5O4+δ,144
but no one improved the refinements
significantly. It also did not solve the problem with additional and broad reflections. Many
RP phases are known to contain stacking faults in their crystal structures.252-255
A model
including a stacking fault (with doubled oxygen sites placed in positions obtained from
Chapter 5. Ruddlesden-Popper phases − stannates
152
Fourier maps) improved the joint refinement by modelling the peak anisotropy, though did
not index the additional peaks.
After trying several models of one RP phase, the model with secondary phases of
SrSnO3 and Sr3Sn2O7 was applied for the joint refinement of I11 and HRPD data. The
refinements were improved (with Rwp values of 2.083% for Sr2Sn0.97Nb0.03O4 and 2.724% for
Sr2Sn0.96Ta0.04O4 respectively) and the additional peaks were all indexed. Including of Oint
(placed in rock salt layer with coordinates: 0, 0.5, 0.25) did not affect the refinements and
showed similar fits to previous ones, with Rwp value of 2.077% for Nb-doped and 2.714% for
Ta-doped sample. Refined thermal parameters of Oint were negative, which could be related
to the small concentration of Oint within the structure, the wrong position of Oint or the
absence of Oint. Several other Oint positions, all obtained from Fourier maps, were used; all of
them provided negative values of thermal parameter for Oint. To maintain the charge balance
of the phase, occupancy of Nb or Ta were included in the refinements after addition of Oint.
Due to the lack of contrast between Sn and Nb or Ta in their neutron scattering lengths (6.225
fm for Sn, 7.054 fm for Nb and 6.91 fm for Ta), the occupancy of Nb and Ta was fixed to
nominal values of 3% and 4% respectively. The final joint refinement of Sr2Sn0.97Nb0.03O4
and Sr2Sn0.96Ta0.04O4 are shown in Figure 5.12 and Figure 5.13; refined structures with the
refined parameters of both compounds are available in Appendices D and E. Refined lattice
parameters (Table 5.4) compared to parental Sr2SnO4 phase show the same trend as observed
in the values obtained only from I11 data analysis: an increase in volume and lattice
parameter c after the doping in comparison with Sr2SnO4. The presence of Oint within the
Sr2SnO4 structure after M5+
doping (Nb, Ta) remained questionable as the dopant levels are
relatively small and the adding of Oint did not improve the refinements significantly. Other
techniques (including spectroscopic methods) were used to explore the doping chemistry of
Sr2Sn1−xMxO4 (M = Nb, Ta).
Chapter 5. Ruddlesden-Popper phases − stannates
153
Figure 5.12: HRPD data refinement of Sr2Sn0.97Nb0.03O4 using Pccn space group with the presence of
secondary phases of SrSnO3 and Sr3Sn2O7, a) bank 1; b) bank 2; Rwp = 2.083%.
Figure 5.13: HRPD data refinement of the Sr2Sn0.96Ta0.04O4 material using Pccn space group with the
presence of secondary phases of SrSnO3 and Sr3Sn2O7, a) bank 1; b) bank 2; Rwp = 2.724%.
Table 5.4: Refined lattice parameters of the Sr2Sn0.97Nb0.03O4 and Sr2Sn0.96Ta0.04O4 material from
HRPD data after the Rietveld refinement in Topas and their comparison with the undoped Sr2SnO4
phase.
Material a (Å) b (Å) c (Å) V (Å)
Sr2SnO4 250
5.72898(5) 5.73524(5) 12.58110(6) 413.378(6)
Sr2Sn0.97Nb0.03O4 5.73644(6) 5.74219(6) 12.65926(6) 416.992(6)
Sr2Sn0.96Ta0.04O4 5.73478(3) 5.74230(3) 12.66306(6) 417.005(4)
Chapter 5. Ruddlesden-Popper phases − stannates
154
5.6 AC Electrochemical Impedance Spectroscopy (EIS)
5.6.1 AC impedance data at 600 ‒ 900°C
The conductivity of single phased (confirmed by laboratory PXRD data analysis) Nb-
and Ta-doped Sr2SnO4 phases was measured using AC impedance spectroscopy and
compared with the values obtained for the undoped phase. A powder of materials was ball
milled in EtOH for overnight. Ball milled and dried powder was pressed in a 10 mm pellet
using a uni-axial press at a pressure of 3.5 tons. Approximately 0.6 g of the material was
weighed out in order to obtain 2 mm thick electrolyte. Cold isostatic pressing (CIP,
Section 2.4.3) was used to increase the density of the pellets applying a pressure
of ≈ 200 MPa. Samples were then heated for 24 h following the synthesis conditions for each
of the doped materials (Section 5.3). The density of all of the measured pellets was calculated
from the obtained values of mass and volume and it was over 97% of crystallographic density
for each of the pellets. Since synthesised materials showed hygroscopic properties,
Archimedes' Principle balance could not be used for density determination. The surface of the
pellets was manually polished after the sintering. Gold wire, mesh, and paste were used as
current collectors.
AC impedance data were collected in ambient air over the temperature range
500 ‒ 800 °C and at the frequencies 1 MHz to 0.01 Hz. The samples were dwelled for 90 min
each 50 °C to reach the thermal equilibrium. Each of the experiments started with 90 min
dwelling at 600 °C. The data were collected using Smart software and equivalent circuit
modelling were performed using the ZView software.196
More details about the data analysis
are described in Section 2.5.2. The spectra showed one high-frequency arc (bulk contribution)
and one intermediate frequency arc (grain boundary contribution). The high-frequency
intercepts of the impedance arcs with the Z' axis was used for the total conductivity
determination, which was calculated from the total resistance, Rtot (bulk + grain boundary).
An example of the AC impedance data collected on Sr2SnO4 material is shown in Figure
5.14. The values of total conductivities are shown in Table 5.5-5.7.
Chapter 5. Ruddlesden-Popper phases − stannates
155
Figure 5.14: AC impedance spectroscopy data of the Sr2SnO4 collected at temperatures of 700, 750
and 800 °C.
Figure 5.15 shows the total conductivities for the temperatures 600 ‒ 800 °C of the
Nb-doped materials compared with the undoped Sr2SnO4 phase (with the values in Table 5.5).
All of the measured materials show an increase of conductivity with the increasing
temperature. The increase of the conductivity follows linear trend in logarithmic scale of σtot.
Sr2Sn0.97Nb0.03O4 material tends to show two linear regions (Figure 5.15). The shape of the
trend line suggests a phase transition might have occurred but more data points at the
temperatures between 650 ‒ 700 °C are needed to confirm that. Activation energy of
Sr2Sn0.97Nb0.03O4 was calculated using the data obtained at the temperatures 700 ‒ 800 °C
while for the rest of the doped Sr2SnO4 and the undoped phase data from 600 ‒ 800 °C region
were used. At the temperature of 600 °C the value of the total conductivity of the
Sr2Sn0.98Nb0.02O4 material is one order of magnitude higher than the ones collected on the
undoped phase. There is no significant improvement of conductivities of both Nb-doped
materials compared to parent Sr2SnO4 phase at the higher temperature of 700 ‒ 800 °C.
Chapter 5. Ruddlesden-Popper phases − stannates
156
Figure 5.15: The total conductivity and activation energy values of the Sr2Sn1−xNbxO4 materials
(x = 0, 0.02, 0.03) collected using AC impedance spectroscopy at the temperatures 600 ‒ 800 °C.
Table 5.5: The values of the total conductivity of the Sr2Sn1−xNbxO4 materials (x = 0, 0.02, 0.03)
collected using AC impedance spectroscopy at the temperatures 600 ‒ 800 °C.
x σtot (S cm−1
)
600 °C 650 °C 700 °C 750 °C 800 °C
0 4.76 × 10−6
8.56 × 10−6
1.85 × 10−5
2.12 × 10−5
9.20 × 10−5
0.02 1.29 × 10−5
2.61 × 10−5
3.28 × 10−5
1.13 × 10−4
1.18 × 10−4
0.03 1.86 × 10−6
4.59 × 10−6
1.28 × 10−5
3.60 × 10−5
1.10 × 10−4
Total conductivity and activation energy values for the data obtained at the
temperatures 600 ‒ 800 °C of the Ta-doped Sr2SnO4 material with nominal values of x = 0,
0.2, 0.3, and 0.4 are shown in Figure 5.16 and Table 5.6. The values of the total conductivity
of Ta-doped materials are higher compared to the undoped Sr2SnO4 phase. The improvement
Chapter 5. Ruddlesden-Popper phases − stannates
157
is more significant for the x = 0.03 and 0.04 samples were the σtot values are for several
temperatures over one order of magnitude higher than those collected for the undoped
material. The conductivity of Sr2Sn0.98Ta0.02O4 material remains close to the Sr2SnO4 for all
the values in 600 ‒ 800 °C. The values of activation energies are significantly increased
(0.5 eV and more compared to Ea value of Sr2SnO4) for all of the Ta-doped samples.
Figure 5.16: The total conductivity and activation energy values of the Sr2Sn1−xTaxO4 materials
(x = 0, 0.02, 0.03, 0.04) collected using AC impedance spectroscopy at the temperatures
600 ‒ 800 °C.
Table 5.6: The values of the total conductivity of the Sr2Sn1−xTaxO4 materials (x = 0, 0.02, 0.03, 0.04)
collected using AC impedance spectroscopy at the temperatures 600 ‒ 800 °C.
x σtot (S cm−1
)
600 °C 650 °C 700 °C 750 °C 800 °C
0 4.76 × 10−6
8.56 × 10−6
1.85 × 10−5
2.12 × 10−5
9.20 × 10−5
0.02 3.99 × 10−6
1.34 × 10−5
3.38 × 10−5
7.87 × 10−5
2.83 × 10−4
0.03 2.45 × 10−5
4.06 × 10−5
1.07 × 10−4
2.63 × 10−4
2.30 × 10−3
0.04 5.38 × 10−5
1.76 × 10−4
5.32 × 10−4
9.91 × 10−4
2.61 × 10−3
Chapter 5. Ruddlesden-Popper phases − stannates
158
Previous impedance data show the values of the total conductivity at the temperatures
600 ‒ 800 °C. The conductivity of samples with the highest Nb and Ta doping levels
(x = 0.03 and 0.04 respectively) was also measured for the higher temperatures until 900 °C.
Table 5.7 shows the comparison of this data with parental Sr2SnO4 phase. The value of total
conductivity for Sr2Sn0.97Nb0.03O4 at 900 °C is two times lower than that for the Sr2SnO4
phase. Sr2Sn0.96Ta0.04O4 sample, similarly to lower temperatures range shows an improvement
of the total conductivity values. The highest value collected at 900 °C: 1.02 × 10−2
S cm−1
is
more than one order of magnitude higher than for the Sr2SnO4 phase. This value is
comparable (one order of magnitude lower) to YSZ256
and comparable to the reported values
of ceria based electrolytes, GDC and SDC.257,258
Table 5.7: The values of total conductivities of the Sr2SnO4 and Nb0.03 and Ta0.04 doped materials
collected using AC impedance spectroscopy at the high temperatures 800 ‒ 900 °C.
Material σtot (S cm−1
)
800 °C 850 °C 900 °C
Sr2SnO4 9.20 × 10−5
4.51 × 10−4
9.18 × 10−4
Sr2Sn0.97Nb0.03O4 1.10 × 10−4
2.87 × 10−4
4.52 × 10−4
Sr2Sn0.96Ta0.04O4 2.61 × 10−3
4.76 × 10−3
1.02 × 10−2
5.6.2 AC impedance data at 300 ‒ 600°C
Impedance data of Sr2SnO4 derivatives mentioned in previous section showed an
improvement of the conductivities for the Ta-doped materials and almost no change in the
case of Nb doping. It is important to find out the origin of the increased conductivity in
Sr2Sn1−xTaxO4 materials (i.e. due to electronic or ionic contribution). This can be due to the
few different compensation effects after the Ta doping: reduction of the Sn4+
to Sn2+
,
presence of free electron carriers (electrons) or the presence of Oint. The structural study of
the HRPD data did not give the clear answer about the presence of Oint. The presence of
interstitial oxygen would significantly increase the ionic conductivity which can be observed
at the temperature higher than 600 °C. Gradual increase of the conductivity would point
towards the electronic contribution (implying mixed-valence Sn4+
/Sn2+
).
Chapter 5. Ruddlesden-Popper phases − stannates
159
The values of the impedance data at lower temperatures 300 ‒ 600 °C (collected for
90 min dwelling at every 100 °C step) are shown in Figure 5.17 and Table 5.8. The data were
collected on the undoped material and single phased materials with highest doping level of
Nb and Ta (Sr2Sn0.97Nb0.03O4 and Sr2Sn0.96Ta0.04O4 respectively). For the Nb-doped sample
the values of total conductivity are lower than the ones collected for the undoped Sr2SnO4
material. The comparison of the Sr2Sn0.96Ta0.04O4 sample with undoped material shows
higher values of the conductivity of the Ta-doped material. The difference between the
materials is decreasing with the increase of the temperature. Gradual increase of the
conductivity could be explained by the increased electronic conductivity. The trend of the
conductivity values in lower temperature region for the data collected on the undoped and
Sr2Sn0.96Ta0.04O4 material can be used as evidence tool for the conductivity contribution in
Ta-doped materials but not as a main proof. Further study of the electronic structure and its
changes after the doping is needed, including UV-vis spectroscopy and Sn solid state NMR.
These methods can provide more precise results and better evidence about the valence state
of Sn in Sr2Sn0.96Ta0.04O4.
Figure 5.17: The total conductivity values and activation energies of the Sr2SnO4 and Nb0.03 and Ta0.04
doped materials collected after the AC impedance at the temperatures 300 ‒ 600 °C.
Chapter 5. Ruddlesden-Popper phases − stannates
160
Table 5.8: Total conductivity values of the Sr2SnO4 and Nb0.03 and Ta0.04 doped materials collected
after the AC impedance at the temperatures 300 ‒ 600 °C.
Material σtot (S cm−1
)
300 °C 400 °C 500 °C 600 °C
Sr2SnO4 1.75 × 10−7
2.67 × 10−7
4.12 × 10−6
1.68 × 10−5
Sr2Sn0.97Nb0.03O4 2.86 × 10−8
2.99 × 10−7
1.14 × 10−6
4.31 × 10−6
Sr2Sn0.96Ta0.04O4 6.29 × 10−7
3.31 × 10−6
9.99 × 10−6
3.52 × 10−5
5.6.3 AC impedance data at different partial oxygen pressure
One of the requirements for electrolytes for SOFCs is the stability and constant
electrochemical properties for a wide range of oxidising/reducing conditions. More details
about the electrochemical processes can be obtained from the p(O2) vs σt dependence. Thus
AC impedance data of Ta-doped materials were collected under various p(O2) pressure.
Pellets were prepared following the procedure mentioned at the beginning of Section 5.6.1.
Measurement with the Sr2Sn0.97Ta0.03O4 material was carried out in a large oxygen partial
pressure region of p(O2) from 1 to 10−4
atm for the various temperatures increasing from 600
and 700 to 800 °C. Every step at different p(O2) was dwelled for 60 min while the collected
AC impedance data were stabilised. Measurements at each individual temperature were
performed during a day. The sample was kept at 600 °C and atmospheric O2 pressure
overnight between the measurements. AC impedance data of the cell were collected at these
conditions before heating to higher temperatures. The obtained data were compared with
those collected before in order to see any changes of impedance arc, which would indicate a
presence of unwanted processes such as reduction occurring during the measurement. Pellet
of the Sr2Sn0.96Ta0.04O4 material was tested for wider p(O2) range: 1 to 10−10
atm at the
temperature of 600 °C.
Figure 5.18 shows the AC impedance data at various p(O2) collected on the
Sr2Sn0.97Ta0.03O4 material. The shape of each of the individual arcs (Figure 5.18a) remains
the same for all the measured p(O2). The total conductivity is a sum of contributions of the
bulk and grain boundary. The changes of the conductivity in studied p(O2) range were
Chapter 5. Ruddlesden-Popper phases − stannates
161
negligible (Figure 5.18b) at all of the measured temperatures. The calculated slope of the data
showed in logarithmic scale is for all of the measured temperatures close to 0. That means
that the studied Sr2Sn0.97Ta0.03O4 material behaves in large oxygen partial pressure region of
p(O2) from 1 to 10−4
atm as pure ionic conductor.
Figure 5.18: EIS data of the Sr2Sn0.97Ta0.03O4 material collected at various temperature (600, 700, and
800 °C) and at different p(O2) partial pressure; a) Nyquist plot of the impedance data collected at
various p(O2) at 800 °C; b) comparison of the log σtot values against log p(O2) collected at various
temperatures.
The EIS data of Sr2Sn0.96Ta0.04O4 material collected at various p(O2) partial pressure
are shown in Figure 5.19 and Table 5.9. For the higher p(O2) pressures we can see the same
behaviour as was observed for the measurement of the Sr2Sn0.97Ta0.03O4 sample. The values
of the total conductivity are all within close range (from 7.7 to 5.8 × 10−5
S cm−1
). The main
contribution of the bulk and grain boundary remains the same for the x = 0.04 sample as for
the x = 0.03. For the values of p(O2) lower than 10−5
atm there is an abrupt increase of the
conductivity with the total conductivity 6.7 × 10−3
S cm−1
at 10−7
atm which is two orders of
magnitude higher than it is observed at p(O2) = 10−5
atm. With further decrease of p(O2)
pressure the conductivity increases more.
Chapter 5. Ruddlesden-Popper phases − stannates
162
Figure 5.19: EIS data of the Sr2Sn0.96Ta0.04O4 material collected at 600 °C. Inset shows the shape of
the AC impedance arcs at various p(O2) pressure.
Table 5.9: Values of the total conductivity of the Sr2Sn0.96Ta0.04O4 material collected at 600 °C
depending on various p(O2) pressure.
p(O2) (atm) σtot (S cm−1
)
2.1 × 10−1
7.77 × 10−5
1.0 × 10−2
6.43 × 10−5
1.1 × 10−3
5.82 × 10−5
1.2 × 10−5
6.21 × 10−5
1.8 × 10−7
6.66 × 10−3
1.2 × 10−10
7.93 × 10−2
The analysis of the PXRD data of Sr2Sn0.96Ta0.04O4 after the p(O2) measurement (Figure 5.20)
confirms the presence of the SrCO3 phase. The values of the refined lattice parameters are
shown in Table 5.10. There is a small change in lattice parameter a and b (on third decimal
place). Comparison of the c lattice parameters shows a significant decrease of the parameter
Chapter 5. Ruddlesden-Popper phases − stannates
163
obtained from the data after the p(O2) measurement compared to as made Sr2Sn0.96Ta0.04O4.
The decrease of the c parameter is followed by the decrease in the volume. The change of the
conductivity can be explained by the mixed Sn4+
/Sn2+
valence state. The presence of Sr-
containing impurity phase might also indicate changes on A-site perovskite layer of
Ruddlesden-Popper phase, which could be partially occupied by Sn2+
. All these changes can
lead to decrease of lattice parameter c and cell volume. More important information related to
transport properties is included in the p(O2) vs conductivity dependence. There is no
significant change of conductivity observed in large oxygen partial pressure region of p(O2)
from 1 to 10−5
atm, which means Sr2Sn0.96Ta0.04O4 behaves as a pure ionic conductor. The
same trend was also observed for the Sr2Sn0.97Ta0.03O4 material. The increase in conductivity
at p(O2) < 10−5
atm suggests a change of behaviour − transition from pure ionic conductor to
mixed ionic-electronic conductor.
Figure 5.20: Pawley fit of the data of Sr2Sn0.96Ta0.04O4 material collected after the p(O2) measurement.
Chapter 5. Ruddlesden-Popper phases − stannates
164
Table 5.10: Refined lattice parameters obtained from the Rietveld refinements of the data collected
on the undoped Sr2SnO4 material with as made Sr2Sn0.96Ta0.04O4 compared with the Ta0.04 doped
sample after the p(O2) measurement.
Material a (Å) b (Å) c (Å) V (Å3)
Sr2SnO4 5.72962(8) 5.73509(8) 12.5862(1) 413.58(1)
Sr2Sn0.96Ta0.04O4 as made 5.7273(2) 5.7389(3) 12.6722(4) 416.51(3)
Sr2Sn0.96Ta0.04O4 after p(O2) 5.7215(2) 5.7326(2) 12.5895(3) 412.92(2)
5.7 Thermal stability
Thermal stability of synthesized materials was monitored by a TGA experiment
(Section 2.9). Figure 5.21 shows the TGA measurement of the Sr2Sn0.97Nb0.03O4 material.
The experiment was carried out in alumina pan. Small amount of the sample (28.1720 mg)
was annealed in air from RT to 900 °C with a dwelling for 2 h and it was cooled down back
to RT with the rate 5 °C min−1
. The small change (0.4 weight %) observed at the beginning of
the heating is due to the moisture taken by the material before the measurement. Further
heating does not decrease the mass significantly. There are no other phases observed on the
PXRD data collected after the TGA experiment. Table 5.11 shows the comparison of the
lattice parameters of as made Nb0.03 doped and the Sr2Sn0.97Nb0.03O4 material after the TGA
analysis. The differences in the values of lattice parameters obtained after the Rietveld
refinement are minimal in all cases. The thermal behaviour of Sr2SnO4 and Sr2Sn0.96Ta0.04O4
was also investigated using a TGA method. The materials exhibit the same behaviour over
the studied temperature range (RT ‒ 900 °C) as it was observed in the TGA measurement of
Sr2Sn0.97Nb0.03O4.
Chapter 5. Ruddlesden-Popper phases − stannates
165
Figure 5.21: TGA of the Sr2Sn0.97Nb0.03O4 material at ambient air sintered from RT to 900 °C and
cooled back to RT.
Table 5.11: Comparison of the lattice parameters and cell volume of Sr2SnO4 and Sr2Sn0.97Nb0.03O4 as
made material and Sr2Sn0.97Nb0.03O4 after the TGA experiment. All of the parameters were obtained
using a Rietveld refinement.
Material a (Å) b (Å) c (Å) V (Å3)
Sr2SnO4 5.72962(8) 5.73509(8) 12.5862(1) 413.58(1)
Sr2Sn0.97Nb0.03O4 as made 5.74330(7) 5.73340(6) 12.6564(1) 416.759(7)
Sr2Sn0.97Nb0.03O4 after TGA 5.7410(6) 5.7334(6) 12.4654(5) 416.46(5)
Chapter 5. Ruddlesden-Popper phases − stannates
166
5.8 UV-vis spectroscopy measurements
5.8.1 As made materials
Sr2SnO4 based materials have attracted attention due to various optical properties,
such as photoluminescence or machanoluminescence.259,260
Previous works on Sn containing
materials (doped SnO2, SrSnO3, Sr2SnO4) showed direct band transitions occurring.163,259,261
Direct band gaps were also calculated for Sr2SnO4 related materials of our study. Linear fits
for the direct band gap determination of corresponding material are shown in Figure 5.22c-e.
A-site Sr cation in Sr2SnO4 has a minor contribution to the electronic structure near the Fermi
level.262
The band gap depends mainly on the bonding of the octahedral network of SnO6.
Conduction band is expected to come from O 2p states whilst the valence band is
predominantly composed of Sn 5s states.262,263
UV-vis spectroscopy was used to look at the
electronic structure of parent Sr2SnO4 and the doped phases in order to find any evidence of
Sn2+
within the structure of doped materials. The diffuse reflectance of solid materials of
Sr2SnO4, Sr2Sn0.97Nb0.03O4, and Sr2Sn0.96Ta0.04O4 was measured on a Shimadzu UV-2550
UV-vis spectrometer. The powder of the measured sample was put in a solid sample holder
with a quartz glass window and placed in the UV-vis spectrometer. The reflectance was
recorded over the spectral range of 200 ‒ 800 nm (6.2 ‒ 1.55 eV). The reflectance at each
wavelength was converted to F(R) using Equation 2.35 (Section 2.6). The data obtained for
all of the samples was converted for both values of n = 1/2 for direct transition.
Parental Sr2SnO4 phase exhibited a band gap of 4.00 eV, which is smaller than the
reported 4.43 eV.259
That could be due to the different crystallinity of the material or different
synthetic route. The purpose of this study was the comparison of the band gap values and
electronic structures of the parental Sr2SnO4 phase with its derivatives. The change of the
electronic structure could be related to the presence of dopant such as in the reported works
of the doped SnO2,163,264-267
or due to the mixed Sn4+
/Sn2+
valence state. The comparison
between the parent and the doped phases is shown in Figure 5.22a, b. For both Nb- and Ta-
doped material, a blue shift towards the higher energy levels (equal to lower wavelengths)
have been observed. This trend is observed also on the values of band gap, which are higher
than that for Sr2SnO4 phase; 4.24 eV for Sr2Sn0.97Nb0.03O4 and 4.40 eV for Sr2Sn0.96Ta0.04O4.
The increase of band gap value was observed in several Sn-containing materials upon M5+
doping.267,268
Chapter 5. Ruddlesden-Popper phases − stannates
167
Figure 5.22: a) Diffuse reflectance spectra of Sr2SnO4, Sr2Sn0.97Nb0.03O4, and Sr2Sn0.96Ta0.04O4
collected over the range of 200 ‒ 800 nm; b) spectra representation using F(R) function; the
extrapolation of the linear section of the diffuse reflectance of c) Sr2SnO4 d) Sr2Sn0.97Nb0.03O4, and e)
Sr2Sn0.96Ta0.04O4, plotted to determine the direct band gap, with the intercept of the x axis (direct band
gap) at c) 4.00 eV, d) 4.24 eV, and e) 4.40 eV.
5.8.2 Reduced materials
The change of the electronic structure of Sr2SnO4 related materials after the doping
and the variation (in the case of Ta-doped Sr2SnO4 phases increase, Section 5.6.1) of
conductivity could be explained due to the presence of Sn in mixed Sn4+
/Sn2+
oxidation state.
Section 5.8.1 showed the diffuse spectra of the Sr2SnO4 related materials as made. These
samples were then reduced in under N2 at 600 °C (following an established literature
procedure for indium-tin-oxide reduction).269
Direct band gaps were calculated from the
spectra obtained after the reduction of the materials (linear fits are shown in Figure 5.23c-e).
All three of the samples show a decrease of the band gap values compared to those
obtained from as made compounds and a shift in their spectra towards lower wavelengths
(Figure 5.23a, b). Reduced Sr2SnO4 sample exhibits the band gap value of 3.57 eV and a red
shift compared to the as made compound with band gap of 4.00 eV. The shift can be
explained by the presence of tin in a mixed oxidation state of Sn4+
and Sn2+
. Band gap values
Chapter 5. Ruddlesden-Popper phases − stannates
168
of reduced phases of Sr2Sn0.97Nb0.03O4 and Sr2Sn0.96Ta0.04O4 were of 3.91 eV and 4.01 eV
respectively. Band gap values of all of the reduced samples are lower compared to as made
materials. One of the possible compensating effects after M5+
doping of Sr2SnO4 is the
presence of Sn4+
/Sn2+
mixed oxidation state. Reduced Sr2SnO4 (with mixed Sn4+
/Sn2+
valence
state) exhibits narrower band gap compared to as made RP1 phase. If there was a Sn4+
/Sn2+
mixed oxidation state in Sr2Sn0.97Nb0.03O4 and Sr2Sn0.96Ta0.04O4, the band gaps for as made
materials would be expected to be smaller compared to the parent Sr2SnO4 phase. The values
are on the contrary higher, which suggests that Sn is present only in Sn4+
oxidation state. This
can be confirmed by solid state 119
Sn NMR (Section 5.9).
Figure 5.23: a) Diffuse reflectance spectra of the reduced samples of Sr2SnO4, Sr2Sn0.97Nb0.03O4, and
Sr2Sn0.96Ta0.04O4 collected over the range of 200‒800 nm; b) spectra representation using F(R)
function; the extrapolation of the linear section of the diffuse reflectance of c) Sr2SnO4 d)
Sr2Sn0.97Nb0.03O4, and e) Sr2Sn0.96Ta0.04O4, plotted to determine the direct band gap, with the intercept
of the x axis (indirect band gap) at c) 3.57 eV, d) 3.91 eV, and e) 4.01 eV.
5.9 Sn Solid-state NMR
Previous work on the doped Sr2SnO4 materials showed an increase in conductivity
Chapter 5. Ruddlesden-Popper phases − stannates
169
after the doping with Ta atom. Neutron diffraction did not confirm the presence of the Oint of
the doped samples. Sn solid-state NMR was used to investigate the oxidation state of the Sn
atom in the Ta-doped Sr2SnO4 materials. Solid-state NMR experiments were performed on a
9.4 T Bruker DSX NMR spectrometer equipped with a 4 mm HXY triple-resonance MAS
probe (in double resonance mode) tuned to 119
Sn at ν0(119
Sn) = 149.1 MHz. 119
Sn single pulse
experiments were performed at room temperature under magic angle spinning (MAS) at
frequency rate of νr = 12.5 kHz, using a π/2 pulse width of 3 μs (i.e. at a radio-frequency (rf)
field amplitude of ν1(119
Sn) = 83 kHz) and a recycle delay of 60 s allowing full relaxation of
the 119
Sn spins. 1024 – 4096 scans were averaged with (experimental time ranging from 17 h
to 3.5 days). 119
Sn chemical shifts were externally referenced to SnO2 at iso = −604.3 ppm.
Figure 5.24 shows the obtained 119
Sn MAS NMR spectra of Sr2Sn1−xTaxO4 (x = 0,
0.02, 0.04). The peak at −564 ppm corresponds to Sn4+
in Sr2SnO4. The peak has a multiplet
lineshape coming from spin-spin coupling between 119
Sn and 117
Sn (doublet as a spin 1/2 ‒
approximately 7 % of the signal) and between 119
Sn and 115
Sn (doublet too). This coupling
arises from crystallographically equivalent but magnetically inequivalent 119
Sn, 117
Sn and
115Sn nuclei). All three samples show residual SnO2 at −604 ppm (approximately 1% of SnO2
in Sr2SnO4 and Sr2Sn0.98Ta0.02O4 and 2% in Sr2Sn0.96Ta0.04O4). SnO2 may relax quicker than
Sr2SnO4, thus its signal will be more intense. High resolution diffraction data collected using
the I11 instrument at the Diamond Light Source (UK) did not show any presence of SnO2,
which may indicate that the amount of SnO2 is very small (< 1%). There is no Sn2+
peak
presented (expected to be at −200 ppm vs −600 for Sn4+
). The shift between Sn2+
and Sn4+
is
quite clear. The fact that there is no Sn2+
presented is important in understanding the defect
chemistry in Ta-doped Sr2SnO4. Enlarged view (Figure 5.25) of the Sn4+
region shows two
peaks at −553 and −580 ppm appearing upon Ta doping. The intensities of these peaks seem
to increase with Ta doping level. The additional peaks might be related to the secondary
phases of SrSnO3 and Sr3Sn2O7 which were used in both I11 and HRPD refinements and they
fitted low-intensity and low-angle peaks (Sections 5.5.2 and 5.5.3). Solid state NMR spectra
of the Sr2Sn1−xTaxO4 compounds show no evidence of Sn2+
presented after the Ta-doping,
implying no mixed valence of Sn4+
/Sn2+
, which was also suggested by the UV-vis
measurements. Ta-doping is compensated for another way − by the presence of Oint or free
electrons. The evidence of free electrons can be confirmed by IR spectra (Chapter 5.10).
Chapter 5. Ruddlesden-Popper phases − stannates
170
Figure 5.24: 119
Sn MAS NMR spectra of Sr2Sn1−xTaxO4 obtained at 9.4 T. Asterisks denote spinning
side bands.
Chapter 5. Ruddlesden-Popper phases − stannates
171
Figure 5.25: Magnified view of the 119
Sn MAS NMR spectra of Sr2Sn1−xTaxO4 obtained at 9.4 T.
5.10 IR spectra
The spectra of 10 mm pellets of Sr2SnO4, Sr2Sn0.97Nb0.03O4, and Sr2Sn0.96Ta0.04O4
materials were measured using a Shimadzu SolidSpec-3700 UV-VIS-NIR Spectrophotometer
(see Section 2.6.2 for more details).
Before the analysis of the collected spectra of Sr2SnO4 related phases, a short
description of transparent conductors must be given. Previous chapters of doped stannates
Chapter 5. Ruddlesden-Popper phases − stannates
172
showed an improvement of conductivity properties after doping with Ta (Section 5.6.1). It is
important to reveal the reason of the increase in conductivity. HRPD data analysis showed no
Oint (within the experimental error limits) within the doped structure (Section 5.5.3) whilst
solid state NMR demonstrated the presence of Sn4+
only. The enhancement of conductivity
can be due to the presence of free electrons which is well known and described in transparent
conductors (TCO).163
The opto-electronic properties of conventional TCO materials can be
represented by a typical TCO representative ZnO (Figure 5.26). The gradual decrease in the
transmission starts at ≈ 1000 nm while the increase in the reflection is observed at ≈ 1500 nm.
The increase of reflectivity (R) at this region is typical for TCO. If the number of electrons in
the conduction band is increased, the wavelength shifts to shorter wavelengths, at the very
high electron concentration even in the visible region.163
This characteristic feature of TCO
was considered in the analysis of the spectra of Sr2SnO4 related materials.
Figure 5.26: Optical spectra of ZnO transparent conductor, taken from163
The spectra of Sr2SnO4, Sr2Sn0.97Nb0.03O4, and Sr2Sn0.96Ta0.04O4 materials collected
over the range of 1600 ‒ 2500 nm are shown in Figure 5.27. No increase in reflection, typical
for TCO is observed for either of the studied materials. A decrease of reflectivity at the
region of ≈ 1600 nm is found for all of the three samples. This bump lies at the lowest region
where PbS detector is used and thus a noisy spectra are commonly collected at this region.
The inset of Figure 5.27 shows the spectra of the same materials collected at wavelengths
from 1000 nm to 1700 nm using InGaAs detector especially focused on the decrease of the
Chapter 5. Ruddlesden-Popper phases − stannates
173
reflectivity at ≈ 1600 nm which is found to be less than at previous spectra (1600 ‒ 2500 nm,
using PbS detector). Even at the wavelengths above 2000 nm no increase of reflectivity is
observed, this tends to indicate that no increased concentration of electrons in the conduction
band is found. It is worth noting that the presented method is mainly used on TCO materials
prepared as thin films while our spectra were collected on solid pellets. It is expected that the
presence of grain boundaries within the solid sample might reduce the sensibility of the
method and thus might not be a strong tool for pellet samples.
Figure 5.27: NIR spectra of Sr2SnO4, Sr2Sn0.97Nb0.03O4, and Sr2Sn0.96Ta0.04O4 material measured at the
wavelengths between 1600 and 2500 nm. The inset shows the spectra of the same materials collected
in 1000 ‒ 1750 nm region.
5.11 Discussion and conclusions
According to the theoretical study of A- and B-site doped Ruddlesden-Popper phases
e.g. A2SnO4, A2HfO4, A2ZrO4, La3+
and M5+
doping is favourable which was shown by
negative values of formation energies obtained in energetic comparison to parental phases
based on the data obtained from DFT calculations. Experimental work was focused on the
Chapter 5. Ruddlesden-Popper phases − stannates
174
Sr2SnO4 related phases. Nb- and Ta- doped Sr2SnO4 were prepared by a solid state synthetic
route. Laboratory obtained PXRD data showed single phase materials with the maximum
doping level of 3% for Nb and 4% Ta. The incorporation of Ta in the RP1 structure was
confirmed by EDX and the average cation ratio was in a good agreement with the expected
stoichiometry. The lattice parameter evolution of both Nb- and Ta-doped materials shows the
same step-like increase of cell volume and c lattice parameter at the maximum doping levels
of single phased samples. There is no further increase observed for higher doping levels
(higher than 3% for Nb and 4% Ta-doped materials). The observed increase can be explained
by charge compensation effect (presence of Oint or Sn4+
/Sn2+
mixed oxidation state). La-
doped materials, the most stable (with most negative formation energies) ‒ as obtained from
DFT calculation, were not synthesised as single phases due to the presence and chemical
stability of La2O3 even after attempts to improve the synthesis conditions.
Ta-doped samples exhibit a significant increase of the conductivity obtained by AC
impedance spectroscopy whilst the Nb-doped materials show similar values as those for the
parental Sr2SnO4 phase. It is of much interest to find out the origin of the enhanced
conductivity of Sr2Sn1−xTaxO4, especially for x = 0.04 where the total conductivity in
temperature range of 700 ‒ 800 °C was increased by more than one order of magnitude
compared to Sr2SnO4. The highest total conductivity 1 × 10−2
S cm−1
of Sr2Sn0.96Ta0.04O4
compound was measured at 900 °C, which is one order of magnitude lower than that obtained
for YSZ.72
Other common electrolytes such as ceria based, exhibit similar conductivity
properties but at the lower temperatures (750 °C): 6.7 × 10−2
S cm−1
for GDC76
and
6.1 × 10−2
S cm−1
for SDC.77
La0.8Sr0.2Ga0.83Mg0.17O2.815 electrolyte shows conductivity of
0.08 S cm−1
at 700 °C.84
The conductivity of Sr2Sn0.97Ta0.03O4 and Sr2Sn0.96Ta0.04O4 were also
studied under various partial oxygen pressures. In high oxygen partial pressure region
(p(O2) ˃ 10−5
atm) both of the studied materials showed no change in conductivity, which is
common for pure ionic conductors. A significant increase of conductivity (of two orders of
magnitude) was observed at the p(O2) < 10−5
atm at 600 °C for Sr2Sn0.96Ta0.04O4. That
indicates mixed ionic-electronic conductivity due to the reduction of Sn4+
to Sn2+
. Potential
SOFC electrolyte needs to exhibit constant oxygen ion conductivity over wide range of p(O2).
Any electronic contribution to total conductivity is detrimental for any practical use as an
electrolyte for SOFCs.
Three possible mechanisms explaining the defect chemistry and improvement of
conductivity in Sr2Sn1−xTaxO4 were taken into an account: increasing of ionic conductivity
due to the presence of interstitial oxygen atoms, electronic contribution as a result of partial
Chapter 5. Ruddlesden-Popper phases − stannates
175
reduction of tin and therefore mixed valence state of Sn4+
and Sn2+
and electronic
contribution due to the presence of free electrons typical for transparent conductors. The
presence of Oint was investigated using the high resolution data from synchrotron (I11) and
neutron (HRPD) sources. The analysis of the data indicates stacking defects in crystal
structure. The doping level of studied phases was rather small (3 and 4% respectively) and
even using high resolution data gives no certain answer. The interstitial oxygen
concentration, if there is any, is very low and is most likely lower than the detection limits of
HRPD. The presence of stacking faults which could be explained by the changes of synthesis
route (using a planetary ball mill to decrease the total synthesis time from seven to five days)
makes the Oint determination even more difficult.
Other methods such as Sn solid state NMR, UV-vis and infrared spectroscopy, have
been used to reveal the chemistry of M5+
doping on B-site of Sr2SnO4 materials. UV-vis
spectra of Sr2SnO4, Nb- and Ta-doped compounds show a blue shift of the doped material
compared to the parent phase expressed by the band gap values. More interesting is the
spectra comparison of the reduced samples. Reduced Sr2SnO4, with expected mixed valence
of Sn4+
and Sn2+
, shows a lower band gap value of 3.57 eV than the one of as made Sr2SnO4.
Comparison of the band gap values of the doped materials tends to show that Sn atoms are
not presented in mixed oxidation state of Sn4+
and Sn2+
. Further work based on solid state Sn-
NMR clearly showed no Sn2+
presented in either the parent or the Sr2Sn1−xTaxO4 (for x = 0.02
and 0.04) phases.
An increase of reflectivity, typical for transparent conductors with free electrons, was
not observed on the IR spectra of parental and doped materials (Sr2Sn0.97Nb0.03O4 and
Sr2Sn0.96Ta0.04O4). The increase is observed at various wavelengths, which depends on the
electrons concentration. This absence could be a result of no or very low concentration of free
electrons in the doped Sr2SnO4 phases. The method conditions (samples is in polycrystalline
pellet form instead of thin films), which reduce the sensibility, must be taken into an account
as well.
In conclusion, Ta substitution in Ruddlesden-Popper Sr2SnO4 phases has improved
the total conductivity properties. Electrochemical performance may be improved more by
altering the processing conditions. Following study for a potential use of this material as a
SOFC electrolyte, revealed severe drawbacks for a practical use such as inconstant
electrochemical behaviour under lower p(O2) atmospheres. Several different techniques have
Chapter 5. Ruddlesden-Popper phases − stannates
176
been applied in order to determine the defect chemistry in Sr2Sn1−xTaxO4. The results tend to
show an ionic contribution to the enhanced conductivity.
Chapter 6. General Conclusions and Perspectives
6 General Conclusions and Perspectives
The aim of this work was to find new materials and improve the properties of already
known materials for solid oxide fuel cells. The material investigations were based on both
experimental and theoretical methods.
The first results chapter (Chapter 3) includes an examination of the cathode properties
of the triple perovskite Y1−ySr2+yCu3−xCoxO7+δ. The promising electrochemical properties
reported in the literature for YSr2Cu2CoO7+δ69,70
are devalued by significant problems with
compatibility and electrochemical stability. Further substitution of Cu for Co presented in this
work, has considerably improved the cathode performance compared to the material with the
parent material. The values of ASR are in agreement with general SOFC cathode
requirements. The limits of the nonstoichiometry on A-site, needed for the synthesis of single
phase compounds with higher dopant amount, was not studied in full and thus higher Co
doping levels might be achievable. It is worth noting that the Co enhancement retains a
crystal structure favourable for high oxide ion mobility. The crystal structure of
Y0.95Sr2.05Cu1.7Co1.3O7+δ compound was studied in more details using the neutron diffraction
data. Sufficient electrochemical stability is a common issue for state of the art cathode
material research and more stability research is, also required for 3ap structured materials.
Most of the known Co-containing cathode materials exhibit thermal expansion higher than
that of common electrolyte while the thermal expansion of Y0.95Sr2.05Cu1.7Co1.3O7+δ material
matches well with YSZ and LSGM electrolytes. The structural family of 3ap materials
represents interesting cathode candidates with further study on material optimisations
principally needed to improve the electrochemical stability.
The aim of a theoretical approach based on DFT calculations of GBCO related
structures (Chapter 4) was to find candidates with double perovskite and fluorite layer
suitable for further experimental work. Energetics of these materials were studied including
several phases as potential by-products to predict thermodynamically favourable phases,
which were then observed by PXRD data after the solid state synthesis. Synthesis of the
materials with a CeO2 fluorite layer showed the presence of the starting building blocks of
double perovskite and fluorite layer resulting in no CeO2 implementation within the double
perovskite layer. The charge misbalance between the layers was suppressed by changing the
Chapter 6. General Conclusions and Perspectives
178
fluorite layer by Gd2O3 and Nd2O3 respectively. Other structures, such as Ruddlesden-Popper
phases are thermodynamically accessible, which was also taken into an account in theoretical
study. Primary solid state synthesis showed Nd2BaCo2O7 RP2 phase as a possible candidate
for further investigation including synthesis optimisation and material properties study
although it does not possess the required structure.
Selected A- and B-site doped Ruddlesden-Popper phases and their formation energies
compared to parental phases were part of the study presented in Chapter 5. The DFT study
showed preferential B-site doping in the case of Nb and Ta, which was confirmed by the
single phased Nb- and Ta-doped Sr2SnO4 materials obtained after conventional solid state
synthesis. The conductivity properties of Sr2Sn1−xTaxO4 with x = 0.03 and 0.04 were
increased by more than one order of magnitude compared to those obtained from the parent
Sr2SnO4 phase. The following study was focused on identifying the chemical mechanism
responsible for the enhancement of the conductivity in Sr2Sn1−xTaxO4. Crystal structure
investigation based on synchrotron and neutron diffraction data revealed stacking faults and
was unable to show definitely whether oxide Oint had formed within the RP. Solid state Sn-
NMR data analysis showed no Sn2+
present, which does not support the mechanism of
increasing of the conductivity in Sr2Sn1−xTaxO4 by an electronic contribution due to the mixed
valence Sn2+
/Sn4+
state. The presence of free electrons after the Ta-doping, which was
assumed as another potential mechanism of the defect chemistry, was not confirmed by the
IR spectra collected on parent and doped materials. None of the used technique gave clear
answer about the mechanism explaining the improvement of the conductivity in
Sr2Sn1−xTaxO4, although the AC impedance measurement at various p(O2) showed
Sr2Sn1−xTaxO4 materials (x = 0.03 and 0.04) as ionic conductors in high oxygen partial
pressure region (p(O2) ˃ 10−5
atm). An increase of the dopant concentration could be helpful
to understand doping chemistry in this Ruddlesden-Popper phase. This should include an
optimisation of solid state synthesis or use of different synthetic routes, e.g. sol-gel methods.
References
References
(1) Kordesch, K. V.; Simader, G. R. Chem Rev 1995, 95, 191. (2) Dresselhaus, M. S.; Thomas, I. L. Nature 2001, 414, 332. (3) MacKay, D. Sustainable Energy - without the hot air; UIT Cambridge, 2009. (4) Craddock, D. Renewable Energy Made Easy; Atlantic Publishing Group: Florida, 2008. (5) Wengenmayr, R. a. B., T. Renewable Energy: Sustainable energy concepts for the future; Wiley-VCH: Weinheim, 2008. (6) Orera, A.; Slater, P. R. Chem Mater 2010, 22, 675. (7) Schoenbeinf, C. F. Philos Mag 1838, 14, 3. (8) Grove, W. R. Philos Mag 1839, 14, 4. (9) Steele, B. C. H.; Heinzel, A. Nature 2001, 414, 345. (10) Aguadero, A.; Alonso, J. A.; Escudero, M. J.; Daza, L. Solid State Ionics 2008, 179, 393. (11) Yamamoto, O. Electrochim Acta 2000, 45, 2423. (12) Istomin, S. Y.; Antipov, E. V. Russ Chem Rev+ 2013, 82, 686. (13) Steele, B. C. H. Solid State Ionics 2000, 134, 3. (14) Lashtabeg, A.; Skinner, S. J. J Mater Chem 2006, 16, 3161. (15) Brandon, N. P.; Skinner, S.; Steele, B. C. H. Annu Rev Mater Res 2003, 33, 183. (16) Steele, B. C. H. Solid State Ionics 1996, 86-8, 1223. (17) Jacobson, A. J. Chem Mater 2010, 22, 660. (18) Goodenough, J. B. Annu Rev Mater Res 2003, 33, 91. (19) Miyoshi, S.; Martin, M. Phys Chem Chem Phys 2009, 11, 3063. (20) Mehrer, H. Diffusion in Solids; Springer: Berlin Heidelberg, 2007. (21) Chroneos, A.; Yildiz, B.; Tarancon, A.; Parfitt, D.; Kilner, J. A. Energ Environ Sci 2011, 4, 2774. (22) Chroneos, A.; Parfitt, D.; Kilner, J. A.; Grimes, R. W. J Mater Chem 2010, 20, 266. (23) Chiang, Y.-M., Birnie, D. I. and Kingery, W. D. Physical Ceramics; John Wiley & Sons Inc, 1997. (24) Inaba, H.; Tagawa, H. Solid State Ionics 1996, 83, 1. (25) Kharton, V. V.; Wiley-VCH: Weinheim, 2009. (26) Tilley, R. J. D. Defects in Solids; John Wiley & Sons, Inc., 2008. (27) West, A. R. Solid State Chemistry and its Applications John Wiley & Sons, 1987. (28) Ishihara, T.; Springer: 2009. (29) Liu, Y. New J Phys 2010, 12. (30) Malavasi, L.; Fisher, C. A. J.; Islam, M. S. Chem Soc Rev 2010, 39, 4370. (31) Jiang, S. P. Solid State Ionics 2002, 146, 1. (32) Jiang, S. P. J Mater Sci 2008, 43, 6799. (33) Plonczak, P.; Gazda, M.; Kusz, B.; Jasinski, P. J Power Sources 2008, 181, 1. (34) Mai, A.; Haanappel, V. A. C.; Uhlenbruck, S.; Tietz, F.; Stover, D. Solid State Ionics 2005, 176, 1341. (35) Simner, S. P.; Anderson, M. D.; Pederson, L. R.; Stevenson, J. W. J Electrochem Soc 2005, 152, A1851. (36) Bak, T.; Nowotny, J.; Rekas, M.; Ringer, S.; Sorrell, C. C. Ionics 2001, 7, 380. (37) Figueiredo, F. M.; Marques, F. M. B.; Frade, J. R. Solid State Ionics 1998, 111, 273. (38) Mizusaki, J.; Tabuchi, J.; Matsuura, T.; Yamauchi, S.; Fueki, K. J Electrochem Soc 1989, 136, 2082. (39) Huang, K.; Lee, H. Y.; Goodenough, J. B. J Electrochem Soc 1998, 145, 3220.
References
180
(40) Tai, L. W.; Nasrallah, M. M.; Anderson, H. U.; Sparlin, D. M.; Sehlin, S. R. Solid State Ionics 1995, 76, 273. (41) Bae, J. M.; Steele, B. C. H. Solid State Ionics 1998, 106, 247. (42) Oishi, N.; Atkinson, A.; Brandon, N. P.; Kilner, J. A.; Steele, B. C. H. J Am Ceram Soc 2005, 88, 1394. (43) Dusastre, V.; Kilner, J. A. Solid State Ionics 1999, 126, 163. (44) Tietz, F.; Mai, A.; Stover, D. Solid State Ionics 2008, 179, 1509. (45) Tietz, F.; Fu, Q.; Haanappel, V. A. C.; Mai, A.; Menzler, N. H.; Uhlenbruck, S. Int J Appl Ceram Tec 2007, 4, 436. (46) Magnone, E. J Fuel Cell Sci Tech 2010, 7. (47) Shao, Z. P.; Haile, S. M. Nature 2004, 431, 170. (48) Shao, Z. P.; Haile, S. M.; Ahn, J.; Ronney, P. D.; Zhan, Z. L.; Barnett, S. A. Nature 2005, 435, 795. (49) McIntosh, S.; Vente, J. F.; Haije, W. G.; Blank, D. H. A.; Bouwmeester, H. J. M. Chem Mater 2006, 18, 2187. (50) Svarcova, S.; Wiik, K.; Tolchard, J.; Bouwmeester, H. J. M.; Grande, T. Solid State Ionics 2008, 178, 1787. (51) Mueller, D. N.; De Souza, R. A.; Weirich, T. E.; Roehrens, D.; Mayer, J.; Martin, M. Phys Chem Chem Phys 2010, 12, 10320. (52) Yan, A.; Cheng, M.; Dong, Y. L.; Yang, W. S.; Maragou, V.; Song, S. Q.; Tsiakaras, P. Appl Catal B-Environ 2006, 66, 64. (53) Chen, Z. H.; Ran, R.; Zhou, W.; Shao, Z. P.; Liu, S. M. Electrochim Acta 2007, 52, 7343. (54) Lee, K. T.; Bierschenk, D. M.; Manthiram, A. J Electrochem Soc 2006, 153, A1255. (55) Lee, K. T.; Manthiram, A. Chem Mater 2006, 18, 1621. (56) Aguadero, A.; Escudero, M. J.; Perez, M.; Alonso, J. A.; Pomjakushin, V.; Daza, L. Dalton T 2006, 4377. (57) Sun, C. W.; Hui, R.; Roller, J. J Solid State Electr 2010, 14, 1125. (58) Zinkevich, M.; Aldinger, F. J Alloy Compd 2004, 375, 147. (59) Amow, G.; Skinner, S. J. J Solid State Electr 2006, 10, 538. (60) Maignan, A.; Martin, C.; Pelloquin, D.; Nguyen, N.; Raveau, B. J Solid State Chem 1999, 142, 247. (61) Frison, R.; Portier, S.; Martin, M.; Conder, K. Nucl Instrum Meth B 2012, 273, 142. (62) Tarancon, A.; Skinner, S. J.; Chater, R. J.; Hernandez-Ramirez, F.; Kilner, J. A. J Mater Chem 2007, 17, 3175. (63) Chang, A. M.; Skinner, S. J.; Kilner, J. A. Solid State Ionics 2006, 177, 2009. (64) Tarancon, A.; Pena-Martinez, J.; Marrero-Lopez, D.; Morata, A.; Ruiz-Morales, J. C.; Nunez, P. Solid State Ionics 2008, 179, 2372. (65) Liu, Y. J Alloy Compd 2009, 477, 860. (66) Zhou, Q. J.; He, T. M.; He, Q.; Ji, Y. Electrochem Commun 2009, 11, 80. (67) Kim, J. H.; Cassidy, M.; Irvine, J. T. S.; Bae, J. J Electrochem Soc 2009, 156, B682. (68) Fletcher, J. G.; Irvine, J. T. S.; West, A. R.; Labrincha, J. A.; Frade, J. R.; Marques, F. M. B. Mater Res Bull 1994, 29, 1175. (69) Sansom, J. E. H.; Rudge-Pickard, H. A.; Smith, G.; Slater, P. R.; Islam, M. S. Solid State Ionics 2004, 175, 99. (70) Sansom, J. E. H.; Kendrick, E.; Rudge-Pickard, H. A.; Islam, M. S.; Wright, A. J.; Slater, P. R. J Mater Chem 2005, 15, 2321. (71) Etsell, T. H.; Flengas, S. N. Chem Rev 1970, 70, 339. (72) Badwal, S. P. S. Solid State Ionics 1992, 52, 23. (73) Fergus, J. W., Hui, R., Li, X., Wilkinson, D. P., Zhang, J. Solid Oxide Fuel Cells - Material Properties and Performance; CRC Press: Boca Raton, 2009. (74) Duwez, P.; Brown, F. H.; Odell, F. J Electrochem Soc 1951, 98, 356.
References
181
(75) Vanherle, J.; Horita, T.; Kawada, T.; Sakai, N.; Yokokawa, H.; Dokiya, M. J Eur Ceram Soc 1996, 16, 961. (76) Kudo, T.; Obayashi, H. J Electrochem Soc 1975, 122, 142. (77) Eguchi, K.; Setoguchi, T.; Inoue, T.; Arai, H. Solid State Ionics 1992, 52, 165. (78) Wang, F. Y.; Chen, S. Y.; Qin, W.; Yu, S. X.; Cheng, S. F. Catal Today 2004, 97, 189. (79) Lubke, S.; Wiemhofer, H. D. Solid State Ionics 1999, 117, 229. (80) Ishihara, T.; Matsuda, H.; Takita, Y. J Am Chem Soc 1994, 116, 3801. (81) Feng, M.; Goodenough, J. B. Eur J Sol State Inor 1994, 31, 663. (82) Huang, K. Q.; Tichy, R. S.; Goodenough, J. B. J Am Ceram Soc 1998, 81, 2565. (83) Huang, K. Q.; Tichy, R. S.; Goodenough, J. B. J Am Ceram Soc 1998, 81, 2576. (84) Huang, P. N.; Petric, A. J Electrochem Soc 1996, 143, 1644. (85) Huang, K. Q.; Feng, M.; Goodenough, J. B.; Schmerling, M. J Electrochem Soc 1996, 143, 3630. (86) Yamaji, K.; Horita, T.; Ishikawa, M.; Sakai, N.; Yokokawa, H. Solid State Ionics 1999, 121, 217. (87) Minh, N. Q., Takahashi, T. Science and Technology of Ceramic Fuel Cells; Elsevier: Amsterdam, 1995. (88) Steele, B. C. H. Solid State Ionics 2000, 129, 95. (89) Goodenough, J. B.; Ruizdiaz, J. E.; Zhen, Y. S. Solid State Ionics 1990, 44, 21. (90) Goodenough, J. B.; Manthiram, A.; Paranthaman, P.; Zhen, Y. S. Solid State Ionics 1992, 52, 105. (91) Manthiram, A.; Kuo, J. F.; Goodenough, J. B. Solid State Ionics 1993, 62, 225. (92) Kakinuma, K.; Yamamura, H.; Haneda, H.; Atake, T. Solid State Ionics 2002, 154, 571. (93) Kakinuma, K.; Arisaka, T.; Yamamura, H.; Atake, T. Solid State Ionics 2004, 175, 139. (94) Lacorre, P.; Goutenoire, F.; Bohnke, O.; Retoux, R.; Laligant, Y. Nature 2000, 404, 856. (95) Goutenoire, F.; Isnard, O.; Retoux, R.; Lacorre, P. Chem Mater 2000, 12, 2575. (96) Goutenoire, F.; Isnard, O.; Suard, E.; Bohnke, O.; Laligant, Y.; Retoux, R.; Lacorre, P. J Mater Chem 2001, 11, 119. (97) Georges, S.; Goutenoire, F.; Altorfer, F.; Sheptyakov, D.; Fauth, F.; Suard, E.; Lacorre, P. Solid State Ionics 2003, 161, 231. (98) Georges, S.; Goutenoire, F.; Laligant, Y.; Lacorre, P. J Mater Chem 2003, 13, 2317. (99) Corbel, G.; Laligant, Y.; Goutenoire, F.; Suard, E.; Lacorre, P. Chem Mater 2005, 17, 4678. (100) Georges, S.; Goutenoire, F.; Bohnke, O.; Steil, M. C.; Skinner, S. J.; Wiemhofer, H. D.; Lacorre, P. J New Mat Electr Sys 2004, 7, 51. (101) Nakayama, S.; Aono, H.; Sadaoka, Y. Chem Lett 1995, 431. (102) Nakayama, S.; Kageyama, T.; Aono, H.; Sadaoka, Y. J Mater Chem 1995, 5, 1801. (103) Nakayama, S.; Sakamoto, M.; Higuchi, M.; Kodaira, K.; Sato, M.; Kakita, S.; Suzuki, T.; Itoh, K. J Eur Ceram Soc 1999, 19, 507. (104) Islam, M. S.; Tolchard, J. R.; Slater, P. R. Chem Commun 2003, 1486. (105) Bechade, E.; Masson, O.; Iwata, T.; Julien, I.; Fukuda, K.; Thomas, P.; Champion, E. Chem Mater 2009, 21, 2508. (106) Tolchard, J. R.; Islam, M. S.; Slater, P. R. J Mater Chem 2003, 13, 1956. (107) Jones, A.; Slater, P. R.; Islam, M. S. Chem Mater 2008, 20, 5055. (108) Shaula, A. L.; Kharton, V. V.; Marques, F. M. B. Solid State Ionics 2006, 177, 1725. (109) Fergus, J. W. Solid State Ionics 2006, 177, 1529. (110) Atkinson, A.; Barnett, S.; Gorte, R. J.; Irvine, J. T. S.; Mcevoy, A. J.; Mogensen, M.; Singhal, S. C.; Vohs, J. Nat Mater 2004, 3, 17. (111) Gross, M. D.; Vohs, J. M.; Gorte, R. J. J Mater Chem 2007, 17, 3071. (112) Goodenough, J. B.; Huang, Y. H. J Power Sources 2007, 173, 1.
References
182
(113) Sun, C. W.; Stimming, U. J Power Sources 2007, 171, 247. (114) Spacil, H. S.; Patent, U., Ed. 1970; Vol. 3,558,360. (115) Dees, D. W.; Claar, T. D.; Easler, T. E.; Fee, D. C.; Mrazek, F. C. J Electrochem Soc 1987, 134, 2141. (116) Zhu, W. Z.; Deevi, S. C. Mat Sci Eng a-Struct 2003, 362, 228. (117) McIntosh, S.; Gorte, R. J. Chem Rev 2004, 104, 4845. (118) Matsuzaki, Y.; Yasuda, I. Solid State Ionics 2000, 132, 261. (119) Murray, E. P.; Tsai, T.; Barnett, S. A. Nature 1999, 400, 649. (120) Park, S. D.; Vohs, J. M.; Gorte, R. J. Nature 2000, 404, 265. (121) Lu, C.; Worrell, W. L.; Gorte, R. J.; Vohs, J. M. J Electrochem Soc 2003, 150, A354. (122) Lee, S. I.; Vohs, J. M.; Gorte, R. J. J Electrochem Soc 2004, 151, A1319. (123) Xie, Z.; Zhu, W.; Zhu, B. C.; Xia, C. R. Electrochim Acta 2006, 51, 3052. (124) Saeki, M. J.; Uchida, H.; Watanabe, M. Catal Lett 1994, 26, 149. (125) Hibino, T.; Hashimoto, A.; Yano, M.; Suzuki, M.; Sano, M. Electrochim Acta 2003, 48, 2531. (126) Hibino, T.; Hashimoto, A.; Asano, K.; Yano, M.; Suzuki, M.; Sano, M. Electrochem Solid St 2002, 5, A242. (127) Tao, S. W.; Irvine, J. T. S. Chem Rec 2004, 4, 83. (128) Tao, S. W.; Irvine, J. T. S. Nat Mater 2003, 2, 320. (129) Ruiz-Morales, J. C.; Canales-Vazquez, J.; Pena-Martinez, J.; Marrero-Lopez, D.; Nunez, P. Electrochim Acta 2006, 52, 278. (130) Slater, P. R.; Irvine, J. T. S. Solid State Ionics 1999, 124, 61. (131) Slater, P. R.; Irvine, J. T. S. Solid State Ionics 1999, 120, 125. (132) Kramer, S.; Spears, M.; Tuller, H. L. Solid State Ionics 1994, 72, 59. (133) Porat, O.; Heremans, C.; Tuller, H. L. Solid State Ionics 1997, 94, 75. (134) Awana, V. P. S.; Menon, L.; Malik, S. K. Physica C 1996, 262, 266. (135) Roth, G.; Adelmann, P.; Heger, G.; Knitter, R.; Wolf, T. J Phys I 1991, 1, 721. (136) Huang, Q.; Sunshine, S. A.; Cava, R. J.; Santoro, A. J Solid State Chem 1993, 102, 534. (137) Isobe, M.; Matsui, Y.; Takayamamuromachi, E. Physica C 1994, 222, 310. (138) Pena-Martinez, J.; Tarancon, A.; Marrero-Lopez, D.; Ruiz-Morales, J. C.; Nunez, P. Fuel Cells 2008, 8, 351. (139) Li, R. K.; Greaves, C. Phys Rev B 2000, 62, 14149. (140) Kim, J. S.; Lee, J. Y.; Swinnea, J. S.; Steinfink, H.; Reiff, W. M.; Lightfoot, P.; Pei, S.; Jorgensen, J. D. J Solid State Chem 1991, 90, 331. (141) Wan, J.; Goodenough, J. B.; Zhu, J. H. Solid State Ionics 2007, 178, 281. (142) Kharton, V. V.; Yaremchenko, A. A.; Shaula, A. L.; Patrakeev, M. V.; Naumovich, E. N.; Loginovich, D. I.; Frade, J. R.; Marques, F. M. B. J Solid State Chem 2004, 177, 26. (143) Kim, G. T.; Jacobson, A. J. Mater Res Soc Symp P 2007, 972, 175. (144) Tonus, F.; Greaves, C.; El Shinawi, H.; Hansen, T.; Hernandez, O.; Battle, P. D.; Bahout, M. J Mater Chem 2011, 21, 7111. (145) Segal, D. J Mater Chem 1997, 7, 1297. (146) Moulson, A. J. a. H., J. M. Electroceramics: materials, properties, applications; Chapman & Hall: London, 1990. (147) Pechini, M. P. 1967; Vol. U. S. Pat. No. 3 330 697. (148) Pecharsky, V. K., Zavalij, P. Y.; Springer: NY, 2009. (149) Dinnebier, R. E., Billinge, S. J. L.; RSC Publishing: Cambridge, 2008. (150) Rietveld, H. M. Acta Crystallogr 1967, 22, 151. (151) Rietveld, H. M. J Appl Crystallogr 1969, 2, 65. (152) Coelho, A. A. J Appl Crystallogr 2003, 36, 86. (153) Thompson, S. P.; Parker, J. E.; Potter, J.; Hill, T. P.; Birt, A.; Cobb, T. M.; Yuan, F.; Tang, C. C. Rev Sci Instrum 2009, 80.
References
183
(154) Hastings, J. B.; Thomlinson, W.; Cox, D. E. J Appl Crystallogr 1984, 17, 85. (155) Hannon, A. C. Nucl Instrum Meth A 2005, 551, 88. (156) Smart, E. L., Moore, E. A. Solid State Chemistry; Taylor & Francis: Boca Raton, 2005. (157) Heaney, M. B.; CRC Press LLC: 1999. (158) Groover, M. P. Fundamentals of Modern Manufacturing: Materials, Processes, and Systems; John Wiley & Sons, 2010. (159) MacDonald, J. R. Impedance Spectroscopy: Emphasising solid materials and systems; John Wiley & Sons, 1987. (160) Ulgut, B. Electrochemical Impedance Spectroscopy and Its Applications to Battery Analysis, 2014. (161) Irvine, J. T. S., Siclair, D. C., and West, A. R. Adv Mater 1990, 2, 7. (162) Smith, R. A.; Second ed. ed.; Cambridge University Press: Cambridge, 1978. (163) Ginley, D., Hosono, H., Paine, D. C. Handbook of Transparent Conductors; Sprinfer, 2011. (164) MacKenzie, K. J. D., Smith, M. E. Multinuclear Solid-State NMR of Inorganic Materials, 2002. (165) Buannic, L.; Blanc, F.; Middlemiss, D. S.; Grey, C. P. J Am Chem Soc 2012, 134, 14483. (166) Chen, W. M.; Wu, X. S.; Geng, J. F.; Chen, J.; Chen, D. B.; Jin, X.; Jiang, S. S. J Supercond 1997, 10, 41. (167) Laitinen, H. A., Harris, W. E. Chemical Analysis; McGraw-Hill: New York, 1975. (168) Hayashi, H.; Saitou, T.; Maruyama, N.; Inaba, H.; Kawamura, K.; Mori, M. Solid State Ionics 2005, 176, 613. (169) Kohn, W.; Sham, L. J. Phys Rev 1965, 140, 1133. (170) Hohenberg, P.; Kohn, W. Phys Rev B 1964, 136, B864. (171) Hanke, F., Persson, M., Hofer, W.; Liverpool, T. U. o., Ed. 2010. (172) Thomas, L. H. P Camb Philos Soc 1927, 23, 542. (173) Fermi, E. Z Phys 1928, 48, 73. (174) Teller, E. Rev Mod Phys 1962, 34, 627. (175) Burke, K.; California, U. o., Ed. Irvine, 2007. (176) Wu, Z. G.; Cohen, R. E. Phys Rev B 2006, 73. (177) Sholl, D. S. a. S., J. A.; Wiley: Hoboken, N.J, 2009. (178) Kresse, G.; Joubert, D. Phys Rev B 1999, 59, 1758. (179) Blochl, P. E. Phys Rev B 1994, 50, 17953. (180) Vanderbilt, D. Phys Rev B 1990, 41, 7892. (181) Kresse, G.; Hafner, J. J Phys-Condens Mat 1994, 6, 8245. (182) Apirath, P. DFT: basics and practice, 2007. (183) Petukhov, A. G.; Mazin, I. I.; Chioncel, L.; Lichtenstein, A. I. Phys Rev B 2003, 67. (184) Carrillocabrera, W.; Wiemhofer, H. D.; Gopel, W. Solid State Ionics 1989, 32-3, 1172. (185) Macmanus, J. L.; Fray, D. J.; Evetts, J. E. Physica C 1992, 190, 511. (186) Vischjager, D. J.; Vanzomeren, A. A.; Schoonman, J.; Kontoulis, I.; Steele, B. C. H. Solid State Ionics 1990, 40-1, 810. (187) Nowotny, J.; Rekas, M. J Am Ceram Soc 1990, 73, 1054. (188) Nowotny, J.; Rekas, M.; Weppner, W. J Am Ceram Soc 1990, 73, 1040. (189) Slater, P. R.; Greaves, C. Physica C 1991, 180, 299. (190) Huang, Q.; Cava, R. J.; Santoro, A.; Krajewski, J. J.; Peck, W. F. Physica C 1992, 193, 196. (191) Den, T.; Kobayashi, T. Physica C 1992, 196, 141. (192) Babu, T. G. N.; Kilgour, J. D.; Slater, P. R.; Greaves, C. J Solid State Chem 1993, 103, 472. (193) Awana, V. P. S.; Malik, S. K.; Yelon, W. B.; Karppinen, A.; Yamauchi, H. Physica C- Superconductivity and Its Applications 2002, 378, 155.
References
184
(194) Awana, V. P. S.; Takayama-Muromachi, E.; Malik, S. K.; Yelon, W. B.; Karppinen, M.; Yamauchi, H.; Krishnamurthy, V. V. J Appl Phys 2003, 93, 8221. (195) Sunshine, S. A.; Siegrist, T.; Schneemeyer, L. F.; Marsh, P. Mater Res Soc Symp P 1989, 156, 113. (196) Johnson, D. Copyright 1990-2005 Schribner Asociates, Inc. (197) Kakinuma, K.; Arisaka, T.; Yamamura, H. J Ceram Soc Jpn 2004, 112, 342. (198) Li, S. Y.; Lu, Z.; Huang, X. Q.; Su, W. H. Solid State Ionics 2008, 178, 1853. (199) Kostogloudis, G. C.; Ftikos, C. Solid State Ionics 1999, 126, 143. (200) Hansen, K. K.; Hansen, K. V. Solid State Ionics 2007, 178, 1379. (201) Hammouche, A.; Siebert, E.; Hammou, A. Mater Res Bull 1989, 24, 367. (202) Tietz, F.; Stochniol, G.; Naoumidis, A. Euromat 97 - Proceedings of the 5th European Conference on Advanced Materials and Processes and Applications: Materials, Functionality & Design, Vol 2 1997, 271. (203) Tietz, F. Ionics 1999, 5, 129. (204) Kim, J. H.; Kim, Y. N.; Bi, Z. H.; Manthiram, A.; Paranthaman, M. P.; Huq, A. Electrochim Acta 2011, 56, 5740. (205) Vert, V. B.; Serra, J. M.; Jorda, J. L. Electrochem Commun 2010, 12, 278. (206) Steele, B. C. H.; Bae, J. M. Solid State Ionics 1998, 106, 255. (207) Islam, M. S.; Davies, R. A.; Fisher, C. A. J.; Chadwick, A. Solid State Ionics 2001, 145, 333. (208) Huang, K., Goodenough, J. B. Solid Oxide Fuel Cell Technology; Woodhead Publishing: Cambridge, U. K., 2010. (209) Maignan, A.; Caignaert, V.; Raveau, B.; Khomskii, D.; Sawatzky, G. Phys Rev Lett 2004, 93. (210) Frontera, C.; Garcia-Munoz, J. L.; Llobet, A.; Aranda, M. A. G. Phys Rev B 2002, 65. (211) Taskin, A. A.; Lavrov, A. N.; Ando, Y. Phys Rev B 2005, 71. (212) Akahoshi, D.; Ueda, Y. J Solid State Chem 2001, 156, 355. (213) Fauth, F.; Suard, E.; Caignaert, V.; Mirebeau, I. Phys Rev B 2002, 66. (214) Martin, C.; Maignan, A.; Pelloquin, D.; Nguyen, N.; Raveau, B. Appl Phys Lett 1997, 71, 1421. (215) Moritomo, Y.; Takeo, M.; Liu, X. J.; Akimoto, T.; Nakamura, A. Phys Rev B 1998, 58, 13334. (216) Respaud, M.; Frontera, C.; Garcia-Munoz, J. L.; Aranda, M. A. G.; Raquet, B.; Broto, J. M.; Rakoto, H.; Goiran, M.; Llobet, A.; Rodriguez-Carvajal, J. Phys Rev B 2001, 64, art. no. (217) Zhang, Q. F.; Zhang, W. Y. Phys Rev B 2003, 67. (218) Taskin, A. A.; Lavrov, A. N.; Ando, Y. Appl Phys Lett 2005, 86. (219) Taskin, A. A.; Lavrov, A. N.; Ando, Y. Prog Solid State Ch 2007, 35, 481. (220) Chavez, E.; Mueller, M.; Mogni, L.; Caneiro, A. Xix Latin American Symposium on Solid State Physics (Slafes) 2009, 167. (221) Kim, J. H.; Mogni, L.; Prado, F.; Caneiro, A.; Alonso, J. A.; Manthiram, A. J Electrochem Soc 2009, 156, B1376. (222) Manthiram, A.; Prado, F.; Armstrong, T. Solid State Ionics 2002, 152, 647. (223) Kim, J. H.; Prado, F.; Manthiram, A. J Electrochem Soc 2008, 155, B1023. (224) Kresse, G.; Furthmuller, J. Phys Rev B 1996, 54, 11169. (225) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys Rev Lett 1996, 77, 3865. (226) Vosko, S. H.; Wilk, L.; Nusair, M. Can J Phys 1980, 58, 1200. (227) Perdew, J. P.; Zunger, A. Phys Rev B 1981, 23, 5048. (228) Perdew, J. P.; Wang, Y. Phys Rev B 1992, 45, 13244. (229) Wang, L.; Maxisch, T.; Ceder, G. Phys Rev B 2006, 73. (230) Lee, Y. L.; Kleis, J.; Rossmeisl, J.; Morgan, D. Phys Rev B 2009, 80.
References
185
(231) Sanchez-Andujar, M.; Senaris-Rodriguez, M. A. Z Anorg Allg Chem 2007, 633, 1890. (232) PANalytical, B. V.; edn., E., Ed. 2006. (233) International Center for Diffraction Data; edn., D., Ed. 2007. (234) Hull, S.; Norberg, S. T.; Ahmed, I.; Eriksson, S. G.; Marrocchelli, D.; Madden, P. A. J Solid State Chem 2009, 182, 2815. (235) Mevs, H.; Mullerbuschbaum, H. Z Anorg Allg Chem 1989, 573, 128. (236) Burley, J. C.; Mitchell, J. F.; Short, S.; Miller, D.; Tang, Y. J Solid State Chem 2003, 170, 339. (237) Anderson, P. S.; Kirk, C. A.; Knudsen, J.; Reaney, I. M.; West, A. R. Solid State Sci 2005, 7, 1149. (238) Pardo, V.; Baldomir, D. Phys Rev B 2006, 73. (239) Wang, C. S.; Pickett, W. E. Phys Rev Lett 1983, 51, 597. (240) Chan, M. K. Y.; Ceder, G. Phys Rev Lett 2010, 105. (241) Matskevich, N. I.; Wolf, T.; Matskevich, M. Y.; Chupakhina, T. I. Eur J Inorg Chem 2009, 1477. (242) Matsumoto, H.; Kawasaki, Y.; Ito, N.; Enoki, M.; Ishihara, T. Electrochem Solid St 2007, 10, B77. (243) Kreuer, K. D.; Paddison, S. J.; Spohr, E.; Schuster, M. Chem Rev 2004, 104, 4637. (244) Matsui, T. Thermochim Acta 1995, 253, 155. (245) Shannon, R. D. Acta Crystallogr A 1976, 32, 751. (246) Gillie, L. J.; Hadermann, J.; Hervieu, M.; Maignan, A.; Martin, C. Chem Mater 2008, 20, 6231. (247) Weiss, R.; Faivre, R. Cr Hebd Acad Sci 1959, 248, 106. (248) Kennedy, B. J. Aust J Chem 1997, 50, 917. (249) Fu, W. T.; Visser, D.; IJdo, D. J. W. J Solid State Chem 2002, 169, 208. (250) Fu, W. T.; Visser, D.; Knight, K. S.; IJdo, D. J. W. J Solid State Chem 2004, 177, 4081. (251) Scholder, R.; Rade, D.; Schwarz, H. Z Anorg Allg Chem 1968, 362, 149. (252) Detemple, E.; Ramasse, Q. M.; Sigle, W.; Cristiani, G.; Habermeier, H. U.; Keimer, B.; van Aken, P. A. J Appl Phys 2012, 112. (253) Hungria, T.; MacLaren, I.; Fuess, H.; Galy, J.; Castro, A. Mater Lett 2008, 62, 3095. (254) Suzuki, T.; Nishi, Y.; Fujimoto, M. J Am Ceram Soc 2000, 83, 3185. (255) Rodgers, J. A.; Battle, P. D.; Dupre, N.; Grey, C. P.; Sloan, J. Chem Mater 2004, 16, 4257. (256) Arachi, Y.; Sakai, H.; Yamamoto, O.; Takeda, Y.; Imanishai, N. Solid State Ionics 1999, 121, 133. (257) Hui, S. Q.; Roller, J.; Yick, S.; Zhang, X.; Deces-Petit, C.; Xie, Y. S.; Maric, R.; Ghosh, D. J Power Sources 2007, 172, 493. (258) Horita, T.; Sakai, N.; Yokokawa, H.; Dokiya, M.; Kawada, T.; Van Herle, J.; Sasaki, K. J Electroceram 1997, 1, 155. (259) Kamimura, S.; Yamada, H.; Xu, C. N. Appl Phys Lett 2012, 101. (260) Kamimura, S.; Yamada, H.; Xu, C. N. Appl Phys Lett 2013, 102. (261) Van den Eeckhout, K.; Poelman, D.; Smet, P. F. Materials 2013, 6, 2789. (262) Mizoguchi, H.; Eng, H. W.; Woodward, P. M. Inorg Chem 2004, 43, 1667. (263) Liu, Q. Z.; Li, B.; Liu, J. J.; Li, H.; Liu, Z. L.; Dai, K.; Zhu, G. P.; Zhang, P.; Chen, F.; Dai, J. M. Epl-Europhys Lett 2012, 98. (264) Zainullina, V. M. Physica B-Condensed Matter 2007, 391, 280. (265) Shanthi, E.; Dutta, V.; Banerjee, A.; Chopra, K. L. J Appl Phys 1980, 51, 6243. (266) Lee, S. W.; Kim, Y. W.; Chen, H. D. Appl Phys Lett 2001, 78, 350. (267) Brankovic, G.; Brankovic, Z.; Santos, L. P. S.; Longo, E.; Davolos, M. R.; Varela, J. A. Mater Sci Forum 2003, 416-4, 651. (268) Liu, Q. Z.; Wang, H. F.; Chen, F.; Wua, W. B. J Appl Phys 2008, 103.
References
186
(269) Shigeno, E.; Shimizu, K.; Seki, S.; Ogawa, M.; Shida, A.; Ide, A.; Sawada, Y. Thin Solid Films 2002, 411, 56.
APPENDIX A: EDX data of Sr2Sn0.96Ta0.04O4
Table A: EDX data of Sr2Sn0.96Ta0.04O4 collected on 12 different crystal areas. The data
show the percentage of each individual element and normalised (norm.) amount of Sr, Sn and
Ta in respect to nominal values for Sr2Sn0.96Ta0.04O4. The obtained average values are
compared to those expected.
Measurement
number
Sr (%) Sn (%) Ta (%) O (%) Sr
(norm.)
Sn
(norm.)
Ta
(norm.)
1 28.6 14.6 0.6 56.3 1.9589 1 0.0410
2 27.1 14.4 0.8 57.6 1.9219 1.0212 0.0567
3 28.5 14.5 0.3 56.7 1.9745 1.0046 0.0207
4 28 14.7 0.8 56.5 1.9310 1.0137 0.0551
5 27.7 14.9 0.5 57 1.9280 1.0371 0.0348
6 28.5 14.8 0.2 56.5 1.9655 1.0206 0.0137
7 27.5 14.9 0 57.6 1.9457 1.0542 0
8 26.2 15.6 0.2 58 1.8714 1.1142 0.0142
9 27.2 15.1 0.4 57.3 1.9110 1.0608 0.0281
10 27.3 14.6 0.9 57.2 1.9135 1.0233 0.0630
11 27.6 14.1 0.8 57.4 1.9482 0.9952 0.0564
12 27.6 14.4 0.6 57.3 1.9436 1.0140 0.0422
Expected 28.5714 13.7142 0.57142 57.1428 2 0.96 0.04
Average 27.65 14.7166 0.50833 57.1166 1.9344 1.0299 0.0355
188
APPENDIX B: Lattice parameters an cell volume of Sr2Sn1−xNbxO4
Table B: Lattice parameters and cell volume of the Sr2Sn1−xNbxO4 phases obtained after
Rietveld refinement in Topas (KCl used as an internal standard) and compared with
previously reported data on Sr2SnO4 material.250
Nb doping level a (Å) b (Å) c (Å) V (Å3)
0 – lit.250
5.72898(5) 5.73524(5) 12.58110(6) 413.378
0 - as made 5.72962(8) 5.73509(8) 12.5862(1) 413.58(1)
0.01 5.7283(1) 5.7352(1) 12.5874(2) 413.53(1)
0.02 5.72808(9) 5.7353(1) 12.5874(2) 413.53(1)
0.025 5.7255(3) 5.7334(3) 12.5925(3) 413.38(3)
0.03 5.7363(3) 5.7411(3) 12.6509(3) 416.62(4)
0.05 5.7352(3) 5.7418(2) 12.6493(3) 416.54(3)
0.075 5.7336(2) 5.7390(3) 12.6493(3) 416.23(3)
0.1 5.7340(2) 5.7406(2) 12.6449(4) 416.23(3)
189
APPENDIX C: Lattice parameters and cell volume of Sr2Sn1−xTaxO4
Table C: Lattice parameters and cell volume of the Sr2Sn1−xTaxO4 phases obtained after
Rietveld refinement in Topas (KCl used as an internal standard) and compared with
previously reported data on Sr2SnO4 material.250
Ta doping level a (Å) b (Å) c (Å) V (Å3)
0 – lit.250
5.72898(5) 5.73524(5) 12.58110(6) 413.378
0 - as made 5.72962(8) 5.72962(8) 12.5862(1) 413.58(1)
0.01 5.7279(1) 5.7353(1) 12.5898(2) 413.60(1)
0.02 5.7277(1) 5.7356(1) 12.5912(3) 413.65(1)
0.025 5.7264(2) 5.7356(2) 12.5910(3) 413.54(2)
0.03 5.7343(1) 5.7401(1) 12.6555(2) 416.56(1)
0.04 5.7273(2) 5.7389(3) 12.6722(4) 416.51(3)
0.05 5.7346(2) 5.7399(2) 12.6532(3) 416.50(2)
0.075 5.7339(3) 5.7390(3) 12.6507(3) 416.29(3)
0.1 5.7346(3) 5.7401(4) 12.6459(4) 416.27(4)
190
APPENDIX D: Joint I11 and HRPD Rietveld refinement of
Sr2Sn0.97Nb0.03O4
Table D: Refined parameters of the Sr2Sn0.97Nb0.03O4 sample from the joint Rietveld
refinement of the I11 and HRPD data.
Atom Site x y z Beq (Å2)
Sn1 4a 0 0 0 0.56(1)
Nb1 4a 0 0 0 0.56(1)
Sr1 8e 0.0007(1) −0.0039(5) 0.35184(2) 0.27(1)
O1 4c 0.250 0.250 0.004(1) 0.75(8)
O2 4d 0.750 0.250 −0.0158(5) 0.71(8)
O3 8e −0.0127(2) 0.026(1) 0.1638(1) 0.35(4)
Figure D: Structure of Sr2Sn0.97Nb0.03O4 obtained from joint Rietveld refinement of I11 and
HRPD data: a) view along c axis; b) view along a axis. Sn atoms are grey, Sr is green and O
atoms are red.
191
APPENDIX E: Joint I11 and HRPD Rietveld refinement of
Sr2Sn0.96Ta0.04O4
Table E: Refined parameters of the Sr2Sn0.96Ta0.04O4 material obtained from the joint
Rietveld refinement of the I11 and HRPD data.
Atom Site x y z B (Å2)
Sn1 4a 0 0 0 0.518(7)
Ta1 4a 0 0 0 0.518(7)
Sr1 8e 0.0010(6) −0.0031(4) 0.35187(2) 0.285(9)
O1 4c 0.250 0.250 0.005(1) 0.52(9)
O2 4d 0.750 0.250 −0.0153(5) 0.15(8)
O3 8e −0.011(1) 0.0335(6) 0.1629(1) 0.36(4)
Figure E: Structure of Sr2Sn0.96Ta0.04O4 obtained from joint Rietveld refinement of I11 and
HRPD data: a) view along a axis; a) view along c axis. Sn atoms are grey, Sr is green and O
atoms are red.
top related