SYNTHESIS AND CHARACTERIZATION OF NOVEL COMPLEX IRON OXIDES
WITH LAYERED AND TUNNEL STRUCTURES
By
Shaun R. Bruno

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Chemistry—Doctor of Philosophy
2014

 
 

ABSTRACT
SYNTHESIS AND CHARACTERIZATION OF NOVEL COMPLEX IRON OXIDES
WITH LAYERED AND TUNNEL STRUCTURES
By
Shaun R. Bruno
Novel cathode materials for lithium ion batteries were synthesized via lithium for sodium
ion exchange from the known compounds, ß-NaFeO2 and NaFeTiO4. The resulting
lithium analogs of these known sodium compounds, T-LiFeO2 and LiFeTiO4, contain
tunnel-like structures that were characterized using Rietveld refinement of Powder X-ray
diffraction, electrochemical measurements, Mössbauer spectroscopy, thermogravimetric
analysis, and inductively coupled plasma spectroscopy. Similarly characterized, α- and ßNaFe2O3, with a double layered rock salt structure, were synthesized for the first time as
a bulk powder using an oxygen pressure regulation method that provided the appropriate
conditions for the two polymorphs to form.
Further, investigation into T-LiFeO2 and NaFe2O3 by doping other transition metals into
the iron position, to control specific properties of the two materials was performed with
success. T-LiFeO2 and the parent phase, ß-NaFeO2, were doped with up to 0.1 and 0.15
parts of cobalt per formula unit respectively. NaFe2O3 was successfully doped with cobalt
up to 0.5 moles with pure phases of both the α-NaFe1.5Co0.5O3 and ß-NaFe1.5Co0.5O3
forming. Manganese doping into NaFe2O3 also showed the formation of the α- phase.

 
 

Probing the Fe3+/4+ redox potential of both LiFeO2 and LiFeTiO4 resulted in the
decomposition of each. The cobalt doped LiFeO2 though did show a greater possibility of
cycling at Fe3+/4+ redox potential, but also resulted in a reaction with the organic
electrolyte. Chemical deintercalation of LiFeO2 and LiFeTiO4 were performed with
resulting in the decomposition of LiFeTiO4. LiFeO2, indicated successful lithium
deintercalation with preliminary Mössbauer results illustrating Fe4+ formation. Both
LiFeTiO4 and LiFeO2 successfully cycled electrochemically at the Fe2+/3+ redox potential,
with the new calcium ferrite structure polymorph LiFeTiO4 cycling 17 % higher capacity
than the previously reported spinel and rock salt structure compounds.

 
 

Copyright by
SHAUN R. BRUNO
2014

 
 

This thesis is dedicated to my family; Ashley and Ryan Bruno, Lisa and David Heagle,
Barbara Skodak, Dave and Mary Sebesta, John and Betty Sebesta, Robert and Gloria
Carpenter, Shane, Dane, Lindsey and Kailey.

v 
 

ACKNOWLEDGEMENTS

I would like to thank my advisor; Professor Viktor Poltavets for his advice and guidance
during my Ph.D. research, my colleague and lab mate Dr. Colin Blakely for the
conversations and teamwork we had throughout our tenure at MSU, Dr. Richard Staples
for his help with X-ray diffraction and instrumentation, and my Ph.D. committee
Professor Hamann, Professor Weliky, and Professor Mahanti for the help and insight they
provided when I needed it.

vi 
 

TABLE OF CONTENTS

LIST OF TABLES

.....................................................................................................ix

LIST OF FIGURES

......................................................................................................x

Chapter 1.

Introduction......................................................................................1

Chapter 2.

Instrumentation.................................................................................4
Powder X-ray Diffraction.................................................................4
Inductively Coupled Atomic Emission Spectroscopy......................9
Mössbauer Spectroscopy................................................................10
Electrochemistry.............................................................................15
Thermogravimetric Analysis and Differential Scanning
Calorimetry.....................................................................................16
Scanning Electron Microscopy......................................................16
Transmission Electron Microscopy................................................19
REFERENCES...............................................................................25

2.1
2.2
2.3
2.4
2.5
2.6
2.7
Chapter 3.
3.1
3.2
3.3
3.4
3.5
3.6
Chapter 4.
4.1
4.2
4.3
4.4
4.5
4.6
4.7

Synthetic Methods..........................................................................27
Pechini (Sol-Gel)............................................................................27
Solid State (Ceramic).....................................................................28
Reduction........................................................................................28
Partial Oxygen Pressure Regulation...............................................30
Moisture and Oxygen Sensitive Techniques..................................32
Chemical oxidation and reduction .................................................34
REFERENCES...............................................................................36
Synthesis of NaFe2O3 Using Partial Oxygen Pressure Regulation
Method............................................................................................38
Introduction....................................................................................38
Synthesis of NaFe2O3 via controlled oxygen pressure...................41
Oxidation via sodium deintercalation.............................................45
Transition Metal Doping................................................................46
Results and Discussion...................................................................47
Conclusion......................................................................................72
Future Direction..............................................................................72
REFERENCES...............................................................................76

vii 
 

Chapter 5.
5.1
5.2
5.3
5.4
5.5
5.6
5.7
Chapter 6.
6.1
6.2
6.3
6.4
6.5
6.6
6.7
Chapter 7.
7.1
7.2

Synthesis of Tunnel Polymorph LiFeO2 and Characterization of the
Electrochemical Performance of Cobalt Doping............................80
Introduction....................................................................................80
Synthesis of ß-NaFeO2 and NaFe1-xCoxO2.....................................86
Ion exchange of NaFe1-xCoxO2 to LiFe1-xCoxO2 (0≤x≤0.15)..........86
Lithium intercalation and deintercalation of LiFeO2 via chemical
reduction and oxidation..................................................................87
Results and Discussion...................................................................87
Conclusion....................................................................................136
Future Direction............................................................................136
REFERENCES.............................................................................140
Synthesis of the Novel Polymorph, LiFeTiO4, with Calcium Ferrite
Structure.......................................................................................143
Introduction..................................................................................143
Synthesis of NaFeTiO4.................................................................145
Ion Exchange of NaFeTiO4 to LiFeTiO4......................................146
Chemical intercalation and deintercalation..................................147
Results and Discussion.................................................................148
Conclusion....................................................................................175
Future Direction............................................................................175
REFERENCES.............................................................................182
Mössbauer Data Analysis.............................................................185
Introduction..................................................................................185
Data Fitting...................................................................................185
REFERENCES.............................................................................232

viii 
 

LIST OF TABLES

Table 4.1

Crystallographic Data for the Rietveld Refinement of ß-NaFe2O3 and ßNa0.78Fe2O3 from SXPD data. (Goodness of Fit Parameters: 2 = 5.87,
Rwp = 10.6%).........................................................................................55

Table 4.2

Phase composition of nominal “NaFe2O3” stoichiometry at different
partial O2 pressures at 850 ºC.................................................................56

Table 4.3

NaxFe2O3 Cell parameters a, c and volume determined from Le Bail fit
from PXD................................................................................................67

Table 5.1

Common transition metal oxide cathode materials.................................82

Table 5.2

Polymorphs of LiFeO2, structure, preparation, preparation procedure,
and best electrochemical performance....................................................83

Table 5.3

Cell parameters and volume of cobalt doped ß-NaFeO2 determined from
Le Bail fit................................................................................................93

Table 5.4

Cell parameters and volume of cobalt doped T-LiFeO2 determined from
Le Bail fit..............................................................................................101

Table 5.5

Mössbauer parameters of the chemically deintercalated Li0.42FeO2 fit
with three sites......................................................................................131

Table 5.6

Mössbauer parameters of the as prepared LiFeO2 fit with one
site.........................................................................................................133

Table 5.7

Mössbauer parameters of the chemically intercalated Li1.57FeO2 fit with
three sites..............................................................................................135

Table 6.1

Crystallographic data for the Rietveld refinement of LiFeTiO4 (space
group Pnma), a=8.9206(5) Å, b=2.9595(3) Å, c=10.7103(7) Å...........163

Table 6.2

Crystallographic data for the Rietveld refinement of Li2FeTiO4 (space
group Pnma), a=9.0525(3) Å, b=2.9566(2) Å, c=10.7575(5) Å...........165

Table 6.3

Mössbauer spectroscopy of LiFeTiO4 and lithium deintercalated
Li0.33FeTiO4...........................................................................................174

ix 
 

LIST OF FIGURES

Figure 2.1

Mössbauer schematic illustrating the decomposition of 57Co into 57Fe
resulting in the emission of energy into an absorber, in this case an iron
source, at which point the absorber emits non-resonant, resonant,
conversion, Auger electrons, and x-rays with other various energies.....12

Figure 2.2

Schematic of Mössbauer isomer shift, quadrupole splitting and hyperfine
splitting illustrating the effect each has upon the Mössbauer spectrum..14

Figure 2.3

The relationship between kI, kD, K, θ, and λ. The initial
wavefront(green), the top of the triangular shape is the point at which the
wavefront is diffracted by the plane of atoms, resulting in the diffracted
wavefront (blue), and the change in the vector due to diffraction (black)
is K..........................................................................................................20

Figure 4.1

Structure models of - and -NaFe2O3. The FeO6 octahedra are
highlighted. For - NaFe2O3 atoms in the shared Na/Fe position are
shown as large blue spheres and small red balls represent O ions. For NaFe2O3 large green spheres correspond to Na ions..............................39

Figure 4.2

Experimental setup for the synthesis of α- and ß-NaFe2O3 under
controlled partial O2 pressure..................................................................45

Figure 4.3

PXD of the resulting product of γ-NaFeO2 (red ticks) and Fe3O4 (green
ticks) after annealing the mixture of α-NaFeO2, Fe powder and Fe2O3
under flowing house nitrogen at 850 °C.................................................48

Figure 4.4

PXD of the resulting product of ß-NaFe2O3 (red ticks), γ-NaFeO2 (green
ticks), and Fe3O4 (blue ticks) after annealing the mixture of α-NaFeO2,
Fe powder and Fe2O3 at a pO2 below ∼1.0×10−10 atm at 850 °C............49

Figure 4.5

PXD of the resulting product of ß-NaFe2O3 (red ticks) after annealing
the mixture of α-NaFeO2, Fe powder, and Fe2O3 between the a pO2 of
∼1.0×10−10 and ∼1.0×10−19 atm at 850 °C............................................50

Figure 4.6

PXD of the resulting product of α-NaFe2O3 (red ticks) after annealing
the mixture of α-NaFeO2, Fe powder, and Fe2O3 below pO2 ∼1.0×10−19
atm at 850 °C...........................................................................................51

x 
 

Figure 4.7

Schematic section of phase composition vs. partial O2 pressure diagram
at 850 °C for Na/Fe=1/2..........................................................................52

Figure 4.8

Observed (crosses), calculated (solid line), difference (bottom solid line)
profiles and Bragg reflections (tick marks) for the final Rietveld
refinement of ß-NaFe2O3 and ß-NaxFe2O3, from SXPD data. The tick
marks correspond to the positions (from top down) of ß-NaxFe2O3, ßNaFe2O3, and Fe3O4. The section at higher angles (>16° 2θ) is enlarged
by a factor of 20 for clarity.....................................................................54

Figure 4.9

Difference Fourier map for α-NaFe2O3 indicating large residual electron
density (green) around the Na/Fe and O3 positions. Atomic positions are
denoted for this layered structure and Fe–O octahedra are highlighted.
Blue and red spheres represent atoms in the Na/Fe and O positions,
respectively.............................................................................................59

Figure 4.10 PXD of unwashed sodium deintercalated ß-NaxFe2O3 (red ticks) with the
byproduct NaBr (green ticks)..................................................................63
Figure 4.11 Cell parameter a as a function of moles of sodium in NaxFe2O3 Le Bail
refined from PXD....................................................................................64
Figure 4.12 Cell parameter c as a function of moles of sodium in NaxFe2O3 Le Bail
refined from PXD....................................................................................65
Figure 4.13 Volume as a function of moles of sodium in NaxFe2O3 Le Bail refined
from PXD................................................................................................66
Figure 4.14 NaFe1.5Co0.5O3 synthesized at 850 °C at a pO2 of about 10-23 atm
resulting in the formation of the α-NaFe2O3 structure type (red ticks)...68
Figure 4.15 NaFe1.75Co0.25O3 synthesized at 850 °C at a pO2 of about 10-23 atm
resulting in the formation of the ß-NaFe2O3 structure type (red) and γNaFeO2 (green). .....................................................................................69
Figure 4.16 PXD of NaFe1.85Co0.15O3 synthesized at 850 °C at a pO2 of about 10-23
atm resulting in the formation of the ß-NaFe2O3 structure type (red), γNaFeO2 (green), and Fe3O4 (blue)..........................................................70
Figure 4.17 NaFe1.5Mn0.5O3 synthesized at 850 °C at a pO2 of about 10-23 atm
resulting in the formation of the α-NaFe2O3 (red), structure type γNaFeO2 (blue), and Fe3O4 (green)..........................................................71

xi 
 

Figure 4.18 Magnetic response of NaFe2O3 as a function of temperature. The
measurement was preformed from room temperature to 4 K at 5000
Oe............................................................................................................73
Figure 4.19 Thermopower (green) and electrical conductivity (black) measurements
of ß-NaFe2O3 with electrical conductivity showing transitions occurring
at about 250 K and at about 100 K (black).............................................75
Figure 5.1

PXD of pure ß-NaFeO2 synthesized at 1000 °C in air, red ticks represent
expected reflections of pure phase..........................................................88

Figure 5.2

Cell parameter a of ß-NaFeO2 as a function of cobalt
doping......................................................................................................89

Figure 5.3

Cell parameter b of ß-NaFeO2 as a function of cobalt
doping......................................................................................................90

Figure 5.4

Cell parameter c of ß-NaFeO2 as a function of cobalt
doping......................................................................................................91

Figure 5.5

Cell volume of ß-NaFeO2 as a function of cobalt
doping......................................................................................................92

Figure 5.6

Thermogravimetric analysis (black) and differential scanning
calorimetry (blue) of the lithium for sodium ion exchange. The
endothermic peak beginning at about 190 °C and exothermic peak at
about 212 °C, (exothermic is positive µV/mg), are indicative of melting
and ionic diffusion respectively..............................................................94

Figure 5.7

PXD of the unwashed lithium for sodium ion exchange of ß-NaFeO2 and
LiNO3 to produce LiFeO2 (red tick marks) and NaNO3 (green tick
marks)......................................................................................................96

Figure 5.8

Cell parameter a of T-LiFeO2 as a function of cobalt
doping......................................................................................................97

Figure 5.9

Cell parameter b of T-LiFeO2 as a function of cobalt
doping......................................................................................................98

Figure 5.10 Cell parameter c of T-LiFeO2 as a function of cobalt
doping......................................................................................................99
volume
of
T-LiFeO2
as
a
function
of
cobalt
Figure 5.11 Cell
doping....................................................................................................100

xii 
 

Figure 5.12 PXD and Rietveld refinement of LiFeO2 Rwp=1.49. Observed (black
crosses), calculated (solid red line), difference (solid blue line) profiles,
and Bragg reflections (tick marks) for the final....................................102
Figure 5.13 Rietveld Refinement of PXD Li0.57FeO2 Rwp=3.90. Observed (black
crosses), calculated (solid red line), difference (solid blue line) profiles,
and Bragg reflections (tick marks) for the final Rietveld refinement...103
Figure 5.14 Synchrotron diffraction of Li1.57FeO2 Rwp=5.73 and χ2=1.969. Observed
(black crosses), calculated (solid red line), difference (solid blue line)
profiles and Bragg reflections (tick marks) for the final Rietveld
refinement.............................................................................................106
Figure 5.15 Thermogravimetric analysis of T-LiFeO2 in oxygen up to a final
temperature of 1000 °C.........................................................................107
Figure 5.16 PXD of the resulting powder from the oxygen TGA of T-LiFeO2, red
ticks are indicative α-LiFeO2 with the rock-salt structure....................108
Figure 5.17 Thermogravimetric analysis of Li0.42FeO2 in oxygen up to a final
temperature of 725 °C...........................................................................109
Figure 5.18 PXD of the resulting powder from of Li0.42FeO2 after TGA in oxygen
red tick are indicative of the compound Fe0.873(Li0.365Fe1.2915)O4.........110
Figure 5.19 Thermogravimetric analysis of T-LiFeO2 in nitrogen up to a final
temperature of 1000 °C with a final mass loss of about 2 %................111
Figure 5.20 PXD of the resulting powder from the nitrogen TGA of T-LiFeO2, red
ticks are indicative Fe3O4, but is likely to be lithium doped Fe3O4......112
Figure 5.21 Thermogravimetric analysis of ß-NaFeO2 in nitrogen to a final
temperature of 250 °C for analysis to determine structural stability for
high temperature electrochemical cycling............................................113
Figure 5.22 PXD of the resulting powder from the nitrogen TGA of ß-NaFeO2, red
ticks are indicative ß-NaFeO2...............................................................114
Figure 5.23 Thermogravimetric analysis of ß-NaFeO2 in oxygen to a final
temperature of 350 °C for analysis to determine structural stability for
high temperature electrochemical cycling............................................115
Figure 5.24 PXD of the resulting powder from the oxygen TGA of ß-NaFeO2, red
ticks are indicative of ß-NaFeO2...........................................................116

xiii 
 

Figure 5.25 Thermogravimetric analysis of the sodium deficient ß-Na0.79FeO2 in
nitrogen to a final temperature of 975 °C..............................................117
Figure 5.26 PXD of the resulting powder from the nitrogen TGA of ß-Na0.79FeO2, ßNaFeO2 (red) and Na3Fe5O9 (green) were the resulting phases............118
Figure 5.27 Thermogravimetric analysis of the sodium deficient ß-Na0.79FeO2 in
oxygen to a final temperature of 725 °C...............................................119
Figure 5.28 PXD of the resulting powder from the oxygen TGA of ß-Na0.79FeO2, ßNaFeO2 (red) and Fe2O3 (green) were the resulting phases..................120
Figure 5.29 Thermogravimetric analysis of ß-NaFe0.9Co0.1O2 in nitrogen to a final
temperature of 1000 °C.........................................................................121
Figure 5.30 PXD of the resulting powder from the nitrogen TGA of ßNaFe0.9Co0.1O2, ß-NaFeO2 (red) and ß-Na1-xFeO2 (green) were the
resulting phases.....................................................................................122
Figure 5.31 Thermogravimetric analysis of the ß-NaFe0.9Co0.1O2 in oxygen to a final
temperature of 1000 °C.........................................................................123
Figure 5.32 PXD of the resulting powder from the nitrogen TGA of ßNaFe0.9Co0.1O2, ß-NaFeO2 (red) and ß-Na1-xFeO2 (green) were the
resulting phases.....................................................................................124
Figure 5.33 Cyclic voltammogram of T-LiFeO2 between 4.0 and 4.6 V probing the
Fe3+/4+ redox potential showing the reaction of T-LiFeO2 with the
electrolyte..............................................................................................125
Figure 5.34 Integrated cyclic voltammogram of T-LiFeO2 illustrating what would
normally be the deintercalation of lithium, but this is the reaction of TLiFeO2 with the electrolyte, this is also seen by absence of lithium
intercalation...........................................................................................126
Figure 5.35 High voltage cycling of LiFe0.9Co0.1O2 between 3.5 and 4.6 V probing
the Fe3+/4+ redox couple, the lithium deintercalation peak is present at
about 4.4 V and where a distinct intercalation peak does not exist due to
the compound’s reaction with the electrolyte.......................................127
Figure 5.36 Low voltage cyclic voltammogram of LiFe0.9Co0.1O2 probing the Fe2+/3+
redox couple between voltages of 1.5 and 3.0 V, with lithium
intercalation occurring at about 1.8 V and deintercalation occurring at
about 2.0 V............................................................................................128

xiv 
 

Figure 5.37 Mössbauer spectrum of Li0.42FeO2 after chemical lithium intercalation fit
with 3 sites the total fit (black) and the three sites correlating to isomer
shifts are 0.129 (red), -0.09 (blue), and 0.53 (green)............................130
Figure 5.38 Mössbauer spectrum of the as prepared LiFeO2 with one site..............132
Figure 5.39 Mössbauer spectrum of Li1.57FeO2 after chemical lithium intercalation fit
with 3 sites, the total fit (black) and the three sites correlating to isomer
shifts are 0.161 (green), 0.438 (red), and 0.494 (blue)..........................134
Figure 5.40 Proposed structure of tunnel NaCoO2 resulting from the ion exchange
using NaNO3 with the known compound KCoO2.................................138
Figure 5.41 Proposed structure of tunnel LiCoO2 resulting from the ion exchange
using LiNO3 with the known compound KCoO2..................................139
Figure 6.1

Single crystal diffraction of Anosovite, Ti3O5, formed during initial
attempt to synthesize NaFeTiO4 from literature preparation. Space group
Cmcm, a=3.7097(2)Å, b=9.7308(4)Å, c=9.9716(5)Å, and volume
357.96(3)ų
from
a
molybdenum
source
single
crystal
diffractometer........................................................................................148

Figure 6.2

The mixture containing NaFeTiO4 (green ticks) and Na0.79Fe0.8Ti1.2O4
(red ticks)..............................................................................................149

Figure 6.3

The mixture containing NaFeTiO4 (green ticks) and Na0.75Fe0.75Ti0.25O2
with the α-NaFeO2 structure type (red ticks)........................................150

Figure 6.4

The PXD pattern of NaFeTiO4 (green ticks) was determined to be a pure
phase sample.........................................................................................151

Figure 6.5

Thermogravimetric analysis of calcium ferrite polymorph LiFeTiO4 in
nitrogen to a final temperature of 1050 °C............................................153

Figure 6.6

PXD of LiFeTiO4 after TGA in nitrogen to 1050 °C. Red ticks illustrate
peak reflection locations of the spinel structure LiFeTiO4...................154

Figure 6.7

Thermogravimetric analysis of calcium ferrite polymorph LiFeTiO4 in
oxygen to a final temperature of 1050 °C.............................................155

Figure 6.8

PXD of LiFeTiO4 after TGA in oxygen to 1050 °C. Red ticks illustrate
peak reflection locations of the spinel structure LiFeTiO4...................156

xv 
 

Figure 6.9

SEM images of as prepared LiFeTiO4 illustrating the particle size of the
compound in the micrometer regime, scale bar represents 5 µm..........157

Figure 6.10 SEM images of as prepared LiFeTiO4 illustrating the particle size of the
compound in the micrometer regime, scale bar represents 5 µm..........158
Figure 6.11 SEM images of as prepared LiFeTiO4 illustrating the particle size of the
compound in the micrometer regime, scale bar represents 500 nm......159
Figure 6.12 Observed (black crosses), calculated (red line), difference (blue) profiles,
and Bragg reflections (tick marks) for the final Rietveld refinement of
LiFeTiO4 from PXD data, Rwp=4.42 and χ2=0.260...............................162
Figure 6.13 Observed (black crosses), calculated (red line), difference (blue) profiles,
and Bragg reflections (tick marks) for the final Rietveld refinement of
Li2FeTiO4 from PXD data, Rwp=5.50 and χ2=0.183.............................164
Figure 6.14 Cyclic voltammogram of LiFeTiO4 between 2.8 and 3.7 V decomposing
after the initial cycles as also illustrated by the low current.................166
Figure 6.15 Integrated cyclic voltammogram illustrating the voltage as a function of
capacity of LiFeTiO4 in the attempt to probe the Fe3+/4+ oxidation states
resulting in the decomposition of the tunnel structure..........................167
Figure 6.16 Capacity as a function of the cycle number showing the instability of the
compound with no indication of a cycling trend...................................168
Figure 6.17 Cyclic voltammogram of LiFeTiO4 between 1.4 and 3.0 V showing
relative stable cycling intercalation and deintercalation of the Fe2+/3+
oxidation states......................................................................................169
Figure 6.18 Integrated cyclic voltammogram of the voltage as a function of capacity
of LiFeTiO4 probing the Fe2+/3+ oxidation state at the two cycling rates
C/5 and C/2.5........................................................................................170
Figure 6.19 Capacity as a function of cycle number of the two cycling rates of C/5
(red) and C/2.5 (black), illustrating the cycling capacity and the capacity
fade........................................................................................................171
Figure 6.20 Rietveld refined structure of LiFeTiO4 and of Li2FeTiO4 after
electrochemical/chemical intercalation of a lithium ion.......................172
Figure 6.21 Mössbauer spectroscopy of LiFeTiO4 fit with one Fe3+ site.................173

xvi 
 

Figure 6.22 Mössbauer spectroscopy of lithium deintercalated Li0.33FeTiO4 with the
total of the fit two sites (red), LiFeTiO4 (green) and the amorphous iron
oxide (blue)...........................................................................................174
Figure 6.23 The known structure NaV2O4 with the calcium ferrite structure as a
potential starting point for probing new cathode materials...................177
Figure 6.24 Proposed structure of LiV2O4 resulting from the successful lithium for
sodium ion exchange of the known compound NaV2O4.......................179
Figure 6.25 Proposed structure of Li2V2O4 after successful chemical lithium ion
intercalation or electrochemical lithium intercalation during cyclic
voltammetry..........................................................................................180
Figure 6.26 Proposed structure of V2O4 after successful chemical lithium ion
deintercalation or electrochemical lithium deintercalation during cyclic
voltammetry..........................................................................................181
Figure 7.1

Two-column space separated raw data saved with the ".exp"
extension...............................................................................................185

Figure 7.2

Folder containing the required program files, raw data, and converted
experimental file....................................................................................187

Figure 7.3

The in_kctl file with highlighted input files and parameters required for
CONUSS analysis.................................................................................188

Figure 7.4

The in_kfit file with highlighted parameters for Mössbauer
refinement.......................................................................................190

Figure 7.5

The in_kmco file with highlighted parameter files and input files.......191

Figure 7.6

The in_kref with highlighted material data input to be changed in each
experiment.......................................................................................193

Figure 7.7

Thickness unit / micron parameter highlighted also in in_kref that will
need to be initially refined to optimize data analysis............................194

Figure 7.8

The NAME_in file illustrating the known constant values of iron for
Mössbauer spectroscopy.......................................................................196

Figure 7.9

The NAME_in file with highlighted number of sites of the Mössbauer
atom.................................................................................................197

xvii 
 

Figure 7.10 The NAME_in file with highlighted parameters that will be analyzed in
Mössbauer refinement.........................................................................198
Figure 7.11 The NAME_in file illustrating electric field gradient as the target for
distribution input file........................................................................199
Figure 7.12 The NAME_in file illustrating isomer shift as the target for distribution
input file...........................................................................................200
Figure 7.13 The NAME_in file illustrating magnetic hyperfine field as the target for
distribution input file.........................................................................201
Figure 7.14 The in_kfit illustrating the input of the % symbol, this opens the
parameter for spectrum fitting............................................................204
Figure 715

Terminal screen shot illustrating input of the kctl command................205

Figure 7.16 Terminal screen shot illustrating the reults of the kctl command after
fitting background and thickness.........................................................205
Figure 7.17 The main folder with out_kctl file highlighted.....................................206
Figure 7.18 The out_kctl file with the resulting values providing the best χ2 from the
kctl command....................................................................................207
Figure 7.19 The in_kref with thickness unit / micron highlighted to be opened for
background refinement........................................................................208
Figure 7.20 The resulting spectrum from the background refinement with unchanged
parameters........................................................................................209
Figure 7.21 The NAME_in file illustraing the isomer shift, magnetic hyperfine field,
and quadrupole splitting parameters open with the "%" for fitting.......210
Figure 7.22 Terminal illustrating kctl command with no convergence after 40
iterations.........................................................................................211
Figure 7.23 The out_kctl with the resulting values providing the best χ2 of the nonconverged data.................................................................................212
Figure 7.24 The NAME_in file illustrating the values from the kctl_out file input
into the NAME_in file for further fitting..............................................214

xviii 
 

Figure 7.25 The resulting spectrum of a non-converged kctl command with the
isomer shift, magnetic hyperfine field, and quadrupole splitting
parameters open.................................................................................215
Figure 7.26 The in_kmco with highlighted number of levels and samples per
level.................................................................................................216
Figure 7.27 The kmco set up in the NAME_in file................................................218
Figure 7.28 The terminal illustration of output from the kmco command...............220
Figure 7.29 The highlighted output file as result from the kmco command............221
Figure 7.30 The best sampling as a result of the kmco command highlighted in the
out_kmco file......................................................................................222
Figure 7.31 The best sampling values input into the NAME_in file for further
refinement...........................................................................................223
Figure 7.32 Number of sites of the MB atom highlighted in the NAME_in file with
1 site..............................................................................................224
Figure 7.33 Number of sites of the MB atom highlighted in the NAME_in file with
the changed number of sites from 1 to 2.............................................225
Figure 7.34 The @ &dist2 parameter highlighted after it was copied from the
@&dist1 parameter and the 1 changed to 2..........................................226
Figure 7.35 The NAME_in file with defining MB site #2 highlighted after the 1 was
changed to 2 and the entire section of defining MB site #1 was copied
and pasted beneath the 0 0 0 at the bottom of defining MB site #1......227
Figure 7.36 The final paramater to be renamed, Make Gaussian 80 @&dist2, is
highlighted after all of the parameters have been changed in the defining
MB site #2.............................................................................................228
Figure 7.37 The weight of sublattice parameter for site 1 within the NAME_in file is
highlighted illustrating its change from 1 to 0.5 due to the addition of the
second site.............................................................................................229
Figure 7.38 The weight of sublattice parameter for site 2 within the NAME_in file is
highlighted illustrating its change from 1 to 0.5 due to the addition of the
second site.............................................................................................230

xix 
 

Chapter 1
Introduction
Synthesis and characterization of new materials have, to this point in time, paved the
proverbial road to many new technologies that are used with little, if any thought, by the
majority of people every day. Continued research in the synthesis of novel materials is
essential to the continued growth of society. New materials have provided the world with
mobile technologies that provide the use of cell phones, lap top computers, and tablets; in
addition to rechargeable batteries. .he hardware found within each of these devices are
examples of advancements in materials research. With every advancement that occurs,
new problems also arise. An example of this is storage devices; computer flash drives
continue to increase in memory but are essentially the same size they were when they
came out in 2000. Eventually, there will be a limit to the memory these flash drives can
hold. However, continued research on semiconductors provides the ability to continue
this trend of increasing the memory while maintaining small size. This is just a simple
example of a material that is used to make the lives of many easier. The topics that will
be discussed within will focus on the use of relatively unconventional synthetic methods
to synthesize materials in bulk and new materials to be used in rechargeable lithium ion
batteries.
Chapter 2 will discuss the instrumentation that was used in the characterization of the
compounds within. It will include discussion on the background along with terminology
that will be used throughout for each method as well as important features each method

1 
 

provides to materials chemistry. Important theory will also be discussed to further
understand some of the characterization methods.
Chapter 3 will describe all of the synthetic methods used within. It will include
background, a detailed description, the purpose of using the selected method over
alternative techniques, and how each method was performed experimentally.
Chapter 4 will explain the oxygen pressure regulation method used to prepare the
divalent polymorphs α- and ß-NaFe2O3 that were previously only prepared as single
crystals. The phase diagram of their synthesis will be described in detail along with the
structural and physical properties of the two polymorphs. Chemical oxidation via sodium
deintercalation to synthesize Na1-xFe2O3 to alter the average oxidation state of iron in the
material along with metal doping into the iron position to synthesize NaFe2-xMxO3 will be
discussed.
Chapter 5 will discuss the synthesis and characterization of new tunnel polymorph
LiFe1-xCoxO2 through the initial synthesis of ß-NaFe1-xCoxO2. Attempts to probe the
Fe4+/3+ redox couple through electrochemical characterization of the undoped sample
along with the electrochemical characterization of the Co doped sample will be
described. Electrochemical characterization of the Fe3+/2+ redox couple will also be
shown with discussion of the potential for two-electron transfer.
Chapter 6 will describe the synthesis of the new tunnel polymorph LiFeTiO4 with the
calcium ferrite structure for the purpose of a two-electron transfer rechargeable lithium
ion battery. Synthesis of the precursor phase NaFeTiO4, ion exchange to form the new
compound LiFeTiO4, structural and electrochemical characterization of LiFeTiO4, and

2 
 

attempts to probe the Fe4+/3+ redox couple will also be discussed. Further characterization
of LiFeTiO4 by chemical reduction of Fe3+ to Fe2+ using n-butyllithium was determined
structurally to be Li2FeTiO4. 

3 
 

Chapter 2
Instrumentation
2.1 Powder X-ray Diffraction
Powder X-ray Diffraction (PXD) uses X-rays to characterize powder materials for phase
identification and structural characterization. In order to acquire an accurate PXD, all
possible orientations of the crystalline solid must be represented equally. This is
generally achieved by grinding the sample in question in a mortar and pestle or ball mill
to reduce the powder size and therefore providing an increased number of representative
crystals of all orientations. The detected orientational averaging of the bulk composition
projects a single dimension spectrum that, in contrast to single crystal diffraction, is a
three dimensional reciprocal space projection. This will be discussed later in section 2.7.
Due to the resulting spectrum being the average of the orientation of the bulk
composition, it is imperative that the sample represent all possible orientations equally. If
it is not representative of all orientations equally, the PXD spectrum can be altered due to
what is called preferred orientation. Preferred orientation is when the distribution of the
powder is non-random. This can result in PXD patterns missing peaks that would
otherwise have been present, therefore leading to difficulties in the data analysis such as
peak indexing. Another way to prevent preferred orientation on top of homogenous
grinding is sample rotation. This provides multiple different particles in different
orientations while the scan is occurring.
A powder X-ray diffractometer is composed of three basic parts: an X-ray tube, a sample
holder, and an X-ray detector. X-rays are generated in a cathode ray tube by heating the

4 
 

filament to produce electrons. These electrons are then accelerated toward the sample by
applying a voltage. The resulting electrons that are diffracted by the material at different
angles are detected producing a spectrum that corresponds to the characteristics of the
sample material. X-rays produced are of a specific wavelength that is characteristic to the
X-ray tube source element (Cu, Mo, Fe, Co). Herein all reported PXD patterns are the
result of Cu radiation with a wavelength of 1.5418Å. X-rays with wavelengths of such
magnitudes are required in PXD because they are comparable to the spacing between
planes in the crystal lattice. Peaks in a PXD spectra result from constructive interference
from waves, with the angle (θ) of incident radiation emerging from planes and with
spacing (d) from within the crystal lattice. This is true for constructive interference when
the distance is integral multiples of the wavelength λ as described by Braggs' law in
Equation 2.1.
nλ=2dsinθ

Equation 2.1

Determining parameters such as d-spacing, cell parameters, or Miller indices of simple
cubic structures can be relatively easy with an adequate PXD spectrum, but can become
much more difficult with an increase in complexity of space groups. For instance,
determining the d-spacing requires solving Braggs law Equation 2.1. Further calculations
to determine a compound's cell parameters (a, b, and c) or Miller indices (h, k, and l)
would require Equation 2.2.
1
d2hkl

=

h2
a2

+

k2
b

2

+

l2
c2

Equation 2.2

For compounds that contain a simple cubic cell space group where a=b=c, Equation 2.2
can be simplified into Equation 2.3.

5 
 

1
d2hkl

=

h2 +k2 +l2
a2

Equation 2.3

The solution of Equations 2.2 or 2.3 can only be determined if the unknown variables are
decreased to one. This is accomplished with the knowledge of destructive interference
leading to systematic absences of reflections unique to the compound's space group.
Continuing with the example of a simple cubic cell, the Miller indices, M, can be
determined from Equations 2.4 and 2.5.
M2 =h2 +k2 +l2

Equation 2.4

M= h2 +k2 +l2

Equation 2.5

The resulting integer value of M2 can then be compared to an appropriate table of known
reflections of the compound's space group, which in turn provides three values for the
sum of the h, k, and l. These indices, i.e. [220], only describe the mathematical result of
M. More probable than not, this is the incorrect order of which the indexed values are for
a specific peak i.e. [202] or [022]. This does though, correlate to the multiplicity, mhkl,
which is indicative to the number of planes that have the same d-spacing. Due to the dspacing being inversely proportional to h2+k2+l2, as seen in Equation 2.3, all
combinations of h, k, and l with the same values contribute to the observed peak intensity
(Equation 2.6).
Ihkl α	mhkl F2hkl

1+cos2 θ
sin2 θcosθ hkl

Equation 2.6

Continuing this example with the hkl value of 220, because M2 is a result of the square
root of h2, k2, and l2 (Equation 2.4), there are 12 possible ways to represent this value
([220], [202], [022], [-220], [-2-20]...), therefore 12 is the multiplicity. Though each

6 
 

Miller indices contributes to each peak's intensity of which it is a subset of its
multiplicity, each Miller indice can be its own peak. This becomes more complicated
when the resulting M2 value is 9, therefore the resulting h, k, and l values could be 300 or
221, again most likely not in that specific order. As seen in Equation 2.6, the peak
intensity, Ihkl, is proportional to the structure factor, Fhkl, which is dependent upon how
the material scatters incident radiation, calculated using Equation 2.8. In order to do so,
the scattering factor, f, must first be determined using the electron density of an atom,
ρ(r), using Equation 2.7.
f=4π

∞
sin ((4π/λ)sinθ)r 2
ρ (r)
r
0
(4π/λ)sinθ)r

Fhkl = ∑j fj e

dr

2πi(hxj +kyj +lzj )

Equation 2.7
Equation 2.8

For simple cubic cells containing two, or perhaps three elements, all angles of the unit
cell are 90° and all cell parameters are equal. These calculations can be easily performed
using common software spreadsheets. Increasing the complexity of the space group along
with increased variables, such as multiple atoms, different cell parameters, or different
unit cell angles all add to the overall complexity of the calculations; therefore, requiring
the use of specialized software such as TOPAS or GSAS for the compound's cell
parameters, indexing, and crystal structure Rietveld refinement.
Further analysis of PXDs and the use of specialized programs allow for determination of
the crystal structure using Rietveld refinements (TOPAS or GSAS), particle size
determination can be performed using a program such as PM2K, or long hand calculation
using the Scherrer Equation, Equation 2.9.

7 
 

τ=

Κλ
βcosθ

Equation 2.9

This equation is used, initially solving for Κ, the shape factor, by using a standard of
which τ, the mean size of the ordered crystals is known. β, the line broadening FWHM, θ,
the Bragg angle, and λ, the X-ray wavelength are known or determined values from the
PXD spectra. LaB6 powder or Al2O3, corundum, with known crystallite size are common
standards. The Κ value is then input into the Sherrer formula with the spectrum of which
crystallite size is of question and τ can be solved. This formula is limited to crystallites in
size up to approximately 0.1µm, which are generally much smaller than compounds
synthesized at high temperatures (greater than 500 °C).
Rietveld refinement method uses a least squares approach of refinement to derive a
theoretical line profile calculated based on a known crystal structure model. This method
is performed until the calculated line overlaps the measured reflections in a PXD
spectrum. This method uses height, width, and peak positions in a spectrum to determine
the structural characteristics in powder materials.

Powder samples are hand ground until homogenous and fine, poured onto a zero
background silicon plate, and flattened with a glass slide. The sample is placed in the
powder diffractometer and run from 10 °-65 ° 2θ with a step size of a 0.02 °. The
resulting diffraction pattern is matched to corresponding known diffraction patterns from
the International Centre for Diffraction Data database. If the resulting pattern is
determined to be unknown, further analysis using the Rietveld refinement programs
TOPAS or General Structure Analysis System (GSAS) is performed [1].

8 
 

2.2 Inductively Coupled Plasma Atomic Emission Spectroscopy
Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES) furthermore
referred to as ICP, is an emission spectroscopy that uses a plasma torch created by argon
gas and a radio frequency output coil to determine metal concentrations within a sample.
The emission spectrum is the result of the excited atoms or ions emitting electromagnetic
radiation with characteristic wavelengths indicative of its corresponding element. The
intensity of the emission peak in the spectra corresponds to the concentration of that
element within the sample. ICP can be used for the analysis of up to 70 elements at a time
while being accurate to elemental analysis in the parts per billion regimes. Limitations to
ICP can include inter-elemental interfaces resulting in inaccurate data, and due to the
nature of the analysis, loss of sample.

The final sample must be dissolved in a 2-5 % acidic solution, preferably a HNO3 (Nitric
Acid) solution. Therefore, depending on the sample, direct addition into concentrated
HNO3 will result in the dissolution of the metal compound. Due to the stability of some
metal oxide compounds, concentrated HNO3 will not dissolve them. Therefore, a
concentrated solution of aqua regia is required. This is a mixture of concentrated HCl
(hydrochloric acid) and concentrated HNO3 (all within solutions of aqua regia are
composed of a 1:1 mixture of concentrated HNO3:HCl). The resulting solution of
dissolved metal oxides is then diluted to the 2-5 % acidic solution for ICP analysis.

9 
 

Sample calibration must be performed on every element to be analyzed. This is achieved
by the appropriated dilution into 2 % HNO3 of a purchased 1000 parts per million (ppm)
standards, using a micropipette to construct a calibration curve of which the analyzed
sample concentrations will fall within. Qualitative ICP is performed by the analysis of the
molar ratio of multiple metals within the sample, while quantitative ICP requires the
accurate mass of the analyzed sample and accurate volume the dissolved sample is
diluted to. Qualitative ICP is useful for analysis when determining only the metal-tometal ratios within a sample. Quantitative ICP provides increased information due to the
careful preparation of the metal oxide compound. This allows for the assumption that the
remaining unobservable concentration in ppm, via ICP, are the non-metals within the
compound; therefore, providing oxygen content of the material without the use of
titration.

2.3 Mössbauer Spectroscopy
Mössbauer spectroscopy is used to probe the oxidation states and local environments of
metals. It is most commonly used for iron in mineralogy or geophysics. It is significant
in solid state chemistry for investigating the oxidation state of iron (Fe0, Fe2+, Fe3+ or
Fe4+) and exploring the local environment of iron's coordination polyhedron (trigonal,
tetrahedral,

or

octahedral).

Mössbauer

spectroscopy

is

an

extremely

useful

characterization tool for explaining observed properties of different Mössbauer active
compounds.

10 
 

Mössbauer spectroscopy uses 57Fe, the decay product of 57Co, and an unstable isotope of
iron, which in turn emits different types of energy including γ-rays. These energies can
result in a number of different possible processes when they collide with the sample (the
absorber), such as the emission of core shell electrons, which are not of interest in
Mössbauer spectroscopy. If the emitted γ-rays were to collide with the nucleus of the
absorber iron, and have the correct energy (14.4 keV), it would move the nucleus of iron.
The nucleus must then recoil with equal and opposite energy, resulting in an emitted γray with the as mentioned 14.4 keV recoil energy. This can be seen in Figure 2.1.

11 
 

Figure 2.1 Mössbauer schematic illustrating the decomposition of

57

Co into

57

Fe

resulting in the emission of energy into an absorber, in this case an iron source, at which
point the absorber emits non-resonant, resonant, conversion, Auger electrons, and x-rays
with other various energies.

The units of energy for Mössbauer spectroscopy are in mm/sec due to the set up of the
spectrometer. The source for conventional spectrometers is mounted on a drive that
oscillates the source toward and away from the absorber, therefore increasing the energy
of the emitted photons. For sources such as a synchrotron, the absorber is oscillated with
respect to the source.

12 
 

In order to understand the resulting spectrum of Mössbauer spectroscopy, three important
parameters must be discussed: the isomer shift, quadrupole splitting, and the magnetic
hyperfine splitting. The combination of these three parameters is used to determine the
oxidation state along with the local iron polyhedron environment. Figure 2.2 illustrates
the observed results of isomer shift, quadrupole splitting, and magnetic hyperfine
splitting in a Mössbauer spectrum.

Isomer shift is the comparison of two different nuclear isomeric states in two different
physical or chemical environments due to the combined effect of the recoil-free transition
between the two states. This difference is observed on the Mössbauer spectrum by the
shift from the 0 mm/sec point. The general trend when considering two iron atoms within
the same local environment is: as the isomer shift increases in positive value the
oxidation state of iron is less.

Quadrupole splitting is the result of the nuclei having a non-radially-symmetric shape and
is in the presences of an external electric field gradient. This results in the splitting of
what would normally be a singlet peak in the Mössbauer spectrum into a doublet. The
separation between the doublet peaks can be measured to provide further information on
the chemical environment of the nuclei.

Magnetic hyperfine splitting (also referred to as electronic field gradient or EFG), is the
result of the energy of the nuclear magnetic dipole moment of the magnetic field

13 
 

generated by the electrons. This is observed in a Mössbauer spectrum by the further
splitting of a singlet or doublet into a sextet from the resulting magnetic field.

Figure 2.2 Schematic of Mössbauer isomer shift, quadrupole splitting, and hyperfine
splitting illustrating the effect each has upon the Mössbauer spectrum [2,3].

The following procedure takes place inside an inert, dry nitrogen glove box to prevent
any reaction that may occur between the sample and oxygen or water from air. The
prepared powder sample of interest is spread in a homogenous fashion on the sticky side
of a piece of Kapton tape, with a concentration of at least 10 mg of iron per square
centimeter of tape. A second piece of Kapton tape is then placed sticky side to sample to
"sandwich" the sample between the two pieces of tape. The now sandwiched tape sample

14 
 

is cut into a square with sides of approximate length of one centimeter. Next, all sides of
the square are taped to seal the edges from ambient conditions when placed in the
Mössbauer spectrometer. The sample is placed inside a small vial and parafilm wrapped,
then placed in a larger vial and parafilm wrapped. The large vial is finally placed inside a
zip lock bag for transportation to the measuring facility.

2.4 Electrochemistry
Electrochemical characterization of potential cathode materials were performed in an
argon filled glove box due to the reactivity of the anode (lithium metal) with nitrogen to
form Li3N. The cells were assembled in a split test cell. The cells used a lithium disc as
an anode and counter electrode. A 1 M solution of LiPF6 in ethylene carbonate (EC),
dimethyl carbonate (DMC), and diethyl carbonate (DEC) (EC/DMC/DEC = 1:1:1 by
volume) was used as the electrolyte. The anode/cathode separator was a 25 pm
Microporous Monolayer Membrane. The cathode was a 19 mm disc of slurry prepared by
ball milling 75% by weight active material, 10 % by weight PTFE, and 15 % by weight
Super C65 carbon in 1-methyl-2-pyrrolidone. A 30 micrometers thick slurry layer was
deposited on aluminum foil using a doctor-blade method. The layer was then vacuum
dried and pressed for increased particle-particle connectivity. Cyclic voltammetry were
initially run from 1.5 V to 5.0 V to determine the oxidation and reduction potentials of
the active material. Further analysis of the active material was then performed with a
narrower window and slower scan rate.

15 
 

2.5 Thermogravimetric Analysis and Differential Scanning Calorimetry
Thermogravimetric analysis (TGA) is the analysis of materials in which the physical
properties are measures as a function of increasing temperature at a constant rate in
different gases. Differential scanning calorimetry (DSC) is the analysis for which the
difference for heat required to increase the temperature of a sample function of increasing
temperature at a constant rate. When coupled, TGA-DSC provides information in
different gasses about mass loss or gain in oxidizing, reducing, or inert atmospheres. This
information can explain exothermic or endothermic reactions, the temperature at which
point a phase transition occurs, or basic information such as the temperature required to
dehydrate a sample without decomposition.

2.6 Scanning Electron Microscopy
Scanning electron microscopy (SEM) uses high-speed electrons to study surface
morphology and particle size in the sub micrometer regime as low as about 10 nm into
the micro-regime. Morphology of the micro sized particle can provide insight into
characteristics such as crystal growth along a particular axis. This may provide insight
into possible observations such as preferred orientation within the PXD or intercalation or
deintercalation of the electrochemical results.

16 
 

Elemental analysis within both SEM and TEM is performed using energy measurements
of X-rays produced by elements in the sample from the electron beam of the instrument
called energy-dispersive spectroscopy (EDS). This is accomplished by evaluating the
critical excitation energy difference between different elements. The differences in the
critical excitation energy within the analytical spatial resolution of the whole sample are
measured by the loss of energy over the distance the X-rays travel into the sample.
Comparing two elements of different critical excitation energies, the greater distance the
incident electron travels, the more energy is lost due to scattering. Therefore, an element
with a low excitation energy would produce X-rays from a greater distance within the
sample as compared to an element with a greater excitation energy. The X-rays from
within the sample lead to the production of characteristic X-rays of the specific elements
the sample is composed of. Each element has a fingerprint per say, a unique energy
created by the filling of an inner orbital shell electron by an electron from a shell of
higher energy. This specific energy difference is emitted as detectable X-rays for most
elements. An element often produces many types of X-rays because multiple beam
electrons are striking the sample. Each electron may differ in the interaction with the
sample leading to the emission of different X-ray wavelengths produced from the
relaxation of different sub-shell electrons. These different electron relaxations are named
based on which shell the vacancy was created in (K, L, M, N), and the number of shells
by which the electron jumped to fill the vacancy (α, ß, γ ...). A downfall of EDS is the
inability to detect lighter elements with lower critical excitation energies, due to the
beryllium window absorbing the excitation energies of about 1 keV of energy. Peak

17 
 

overlap in the spectrum is another disadvantage of EDS; therefore, analysis of
compounds with multiple elements may cause peak overlap resulting in difficulties for
quantitative elemental analysis. EDS has several advantages over other elemental
determining methods for SEM, such as wavelength-dispersive spectroscopy (WDS).
WDS does not exhibit peak overlap as does EDS, but EDS is generally the preferred
elemental analysis spectroscopy in SEM and TEM due to its ability to analyze more than
two elements at a time unlike WDS. Analysis software can provide relatively accurate
quantitative analysis even in the instance of peak overlap in an EDS spectrum.
Backscattering electron imagining (BSE) uses a different source of excitation radiation.
This radiation is created by an elastic interaction of beam electrons with the nuclei of the
atom. Heavy atoms have a larger atomic number, therefore, scatter stronger than lighter
elements with smaller atomic numbers. Due to this fact, backscattering images can
provide an accurate elemental composition and an image illustrating the distribution of
elements within the sample. This is increasingly important when determining the metal
ratio and composition of new compounds or determining phase composition of impurities
or element deficient compounds [4–7].

Powder samples are spread on a sticky carbon tab on an aluminum stub with the excess
powder removed using compressed air. This is performed to prevent any damage or
contamination by loose particles ejecting from the stub into the detectors, vacuum
system, or any other parts of the SEM. Depending on the electrical conductivity of the
sample, it can then be left alone or coated with carbon, gold, or osmium with appropriate

18 
 

sputter coaters to increase the sample's electrical conductivity for better resolution. The
sample is then inserted into the instrument for analysis.

2.7 Transmission Electron Microscopy
Transmission electron microscopy (TEM) uses high-speed electrons to study particle
sizes or layer thickness of nano-materials or thin films. TEM was employed on prepared
nano-crystalline samples to determine morphology, particle size, and potential core shell
structures. TEM was used rather than SEM due to the magnification ability of over
500,000 times providing resolution of particles with sizes as low as 0.2 nm.

TEM is useful to determine particle size and morphology of nano-crystals, however, high
resolution TEM (HRTEM) provides the ability to probe further into the crystalline nanoparticles. HRTEM permits an increased accuracy in particle size measurements and
greater detailed images of the sample. HRTEM also grants the ability to observe the
lattice fringes of the nano-crystalline structure. The lattice fringe is the observed plane
within the nano-crystal that correlates to the Miller indice observed in the PXD.
Therefore measuring the distance between two lines of the observed lattice fringe
provides the ability to determine which plane the nano-crystal is orientated. Furthermore,
single crystal diffraction of nano-crystals is performed using Fourier transforms of
HRTEM resulting in electron diffraction spots.

19 
 

In order to discuss single crystal diffraction and better understand some aspects of PXD
(as mentioned in 2.1), the topic of reciprocal lattices must be discussed. First, a few
mathematical definitions must be defined and worked through. Initially considering
planewave front vectors in terms of k, where the initial wavefront, kI, at a specific
wavelength, λ, is scattered by a plane of atoms resulting in a diffracted wavefront, kD,
that scatters at an angle of 2θ. With this, we can determine the difference in kI and kD as
seen in Equation 2.9, where K is the change in the vector due to diffraction as illustrated
in Figure 2.3.

Figure 2.3 The relationship between kI, kD, K, θ, and λ. The initial wavefront (green),
the top of the triangular shape is the point at which the wavefront is diffracted by the
plane of atoms, resulting in the diffracted wavefront (blue), and the change in the vector
due to diffraction (black) is K.

K=kD -kI

20 
 

Equation 2.10

If the energy of the electron is unchanged during the diffraction, therefore is elastic so we
can relate these wavefront vectors as seen in Equation 2.11.
1

|kI |=|kD |= =|K|
λ

Equation 2.11

Furthermore using trigonometry to write an expression in terms of θ,
sin θ=

|K|/2

Equation 2.12

|kI |

then algebraically rearranging, Equation 2.13 is observed.
|K|=

2 sin θ

Equation 2.13

λ

In relating Equation 2.13 to Braggs' law, Equation 2.1, these conditions must be met: the
angle θ must equal the Bragg angle and the integer value of n is one. The equations can
be combined into Equation 2.14, where the vector K at the Bragg angle is KB.
|KB |=

1
d

Equation 2.14

Under the conditions where vector K is at the Bragg angle, the magnitude of vector K has
a special value, therefore is defined by g. Equation 2.15 is also very important because it
represents the Laue conditions for constructive interference. The Laue diffraction will
later be discussed.
KB =g

Equation 2.15

Now that it has been established that K is reciprocally related to d, and K at the Bragg
angle is an important value, of which further discussion will come later, it is important
that real space lattice and reciprocal space lattice vectors be discussed. Here a lattice in
real space, rn, will be defined as follows;
rn =n1 a+n2 b+n3 c

21 
 

Equation 2.16

where the vectors a, b, and c are unit-cell translations in real space and n1, n2, and n3 are
integers. Similarly the reciprocal-lattice vector, r*, can be defined as:
rn =m1 a* +m2 b* +m3 c*

Equation 2.17

where in this case the vectors a*, b*, and c* are unit-cell translations in reciprocal space
and m1, m2, and m3, are integers. The direction of the vectors can be seen using the dot
product of the real space vectors with the reciprocal space vectors as seen in Equation
2.16, where a* is normal to both vectors b and c.
a* ·b=a* ·c=b* ·c=b* ·a=c* ·a=c* ·b=0

Equation 2.18

The length of these vectors can be related as follows:
a* ·a=1;b* ·b=1;c* ·c=1

Equation 2.19

Here it is seen that the dot product of the reciprocal unit-cell length on the real space unitcell length is defined to be one, consequently, providing the scale or dimension of the
reciprocal lattice. Furthermore, this relationship provides insight into the dimensionality
of the reciprocal lattice, in that if a, b, and c are all large the corresponding reciprocal
lattice vectors will be small.
Continuing discussion on the concept of g, the K vector at the Bragg angle, it can be
established that any vector in reciprocal space can be defined as seen in Equation 2.20.
K=ξa* +ηb* +ζc*
Here , , and

Equation 2.20

are three numbers, which do not have to be integers. The ghkl vector,

similarly written to the K vector, is an important vector in reciprocal lattice vectors
because it provides the definition of h, k, and l, which are integer values that together
define a plane (hkl). Due to h, k, and l being in reciprocal space, the definition of the

22 
 

plane (hkl) is that it cuts the a, b, and c axes at 1/h, 1/k, and 1/l respectively. With this
notation the vector BC, can also be written as c/l-b/k.
ghkl =ha* +kb* +lc*

Equation 2.21

Acquiring the dot product of g and rn by using Equations 2.16 and 2.20, an integer value,
N, is always achieved, therefore Equation 2.21 describes similarly, that in order to have
Laue (or Bragg) diffraction, certain conditions must be satisfied. In other words, from
Equation 2.20, , , and

must be integer values and the integer values are h, k, and l in

order for Laue diffraction to occur.
K·rn =N

Equation 2.22

Setting rn to three unit vectors, establishes the following relationships, which are the core
principle of Laue diffraction:
K·a=h

Equation 2.23

K·b=k

Equation 2.24

K·c=l

Equation 2.25

Knowledge of reciprocal space is required to understand Ewald's sphere, which is
observed in single crystal diffraction in TEM. When the Bragg (or Laue) diffraction
condition is satisfied, the scattering vector is equal to the reciprocal lattice vector, as
previously described. Therefore, diffraction will occur only for the reciprocal lattice
points that lie on the surface of the Ewald sphere. If the diffraction pattern is aligned
perfectly, at which the Bragg condition is satisfied, where no spots will be observed,
tilting of the sample will result in the appearance of diffraction spots. Continued tilting of

23 
 

the sample down different planes of the sample will result in the appearance and
disappearance of diffraction spots that cut through different Laue zones. [4–7]

Powder samples are dispersed in a solvent of choice in a test tube. The solvents must be
completely evaporated prior to analysis in the TEM, therefore methanol or ethanol is
chosen due to their volatilities. The dispersed sample is then sonicated for 15 minutes to
reduce agglomerations. A drop of the resulting solution is then placed on a TEM grid and
allowed to dry and be inserted into the instrument for analysis.

24 
 

REFERENCES

25 
 

REFERENCES

[1]

V.K. Pecharsky, P.Y. Zavalij, Fundamentals of Powder Diffraction and Structural
Characterization of Materials, Second Edi, Springer US, Boston, MA, 2009.

[2]

K. J. Kuhn, K. Th. uller, N.J. A. Ruckenstein, U. F. Steiner, G. J. Tumper, K. P.
Wölfle, Nuclear Condensed Matter Physics with Synchrotron Radiation Basic
Principles, Methodology and Application, Spring Berlin Heidelberg, Berlin,
Heidelberg, 2004.

[3]

P. Gütlich, E. Bill, A.X. Trautwein, Mössbauer Spectroscopy and Transition Metal
Chemistry, Springer Berlin Heidelberg, Berlin, Heidelberg, 2011.

[4]

D.B. Williams, C.B. Carter, Transmission Electron Microscopy A Textbook for
Materials Carter, Springer US, New York, n.d.

[5]

S. Amelinckx, D. Van Dyck, J. Van Landuyt, G. Van Tendeloo, Electron
Microscopy Principles and Fundamentals, VCH Verlagsgesellschaft mbH,
Weinheim, 2008.

[6]

P. Echlin, Handbook of Sample Preparation for Scanning Electron Microscopy and
X-Ray Microanalysis, Springer US, Boston, MA, 2009.

[7]

L. Reimer, H. Kohl, Transmission Electron Microscopy, Fifth Edit, Springer US,
New York, 2008.

26 
 

Chapter 3
Synthetic Methods
3.1 Pechini (Sol-Gel)
The Pechini method, also referred to as the polymerisable complex method liquid mix
technique, or sol-gel, was first proposed in the late 1960's as a technique to deposit
different elements in the production of capacitors. It was later determined useful as a
synthetic technique for the synthesis of multicomponent metal containing materials. The
method relies on the mixing of ions in solution then controlled temperature increase to
transform the solution into a gelatin-like polymer. Increasing the temperature furthermore
removes the matrix resulting in homogenously mixed metal oxide precursor.
The sol-gel process involves the addition of stoichiometric ratios of metal salts or metal
oxides with citric acid, ethylene glycol, nitric acid, and water. The citric acid is added to
form citric complexes that upon heating to above 100 °C form the polymer citrate gel
with the ethylene glycol. The addition of nitric acid is to increase the dissolution of the
metal oxides within the solution. Continued heating above 400 °C removes the matrix
leading to the formation of a uniform metal oxide and/or metal carbonate mixture to be
further annealed under appropriate conditions to acquire the desired material.
This method has been widely used for the synthesis of materials with many different
properties such as superconductivity, dielectrics, catalyst, etc. Nanocrystalline powders
have also been synthesized using the Pechini method for different purposes such as dyesensitized solar cells (DSSCs) and water splitting reactions for hydrogen production. [1–
6]

27 
 

3.2 Solid-state (Ceramic)
Solid-state reaction, also referred to as ceramic method, is generally the standard
technique of synthesis in material sciences. Limitations are due to slow diffusion in solids
as compared to liquid/liquid chemistry. High temperatures are required to increase the
rate of diffusion within the reaction. The ceramic method is generally the easiest way to
synthesize stable, solid-state materials directly from the metals, metal oxides, or metal
oxoanion precursors. To increase the rate of diffusion, it is preferable the starting
materials are homogenously mixed prior to sintering. This provides a decreased path
length of atomic diffusion for nucleation sites of the desired material to form. This type
of method can result in incomplete reactions if the starting reagents are inhomogeneously
mixed. After initial sintering, the material is generally reground and reannealed. This
allows the nucleation sites to react with unreacted starting materials resulting in increased
purity of the final product. This method has been given the nickname "the heat and beat
method" because of the firing and grinding repetition that is sometimes needed to acquire
a pure product. High temperature synthesis can also lead to the loss of volatile metals
such as lithium or sodium; therefore, stoichiometric compensation is taken into
consideration in the initial reactant addition. [7]

3.3 Reduction
Reduction of complex metal oxides, if carefully performed, can result in the formation of
meta-stable complex metal oxides that may contain properties otherwise not seen in the

28 
 

non-reduced phase. There are multiple reduction types in the material science synthesis
like solvothermal reduction, which involves the use of metal hydrides such as sodium
hydride (NaH) and a solvent inside a high-pressure autoclave with the temperature of the
reaction elevated to a desired temperature. An approach resembling a solid-state synthesis
reaction involves the mixing and grinding of metal hydrides with the complex metal
oxide. The mixture is then heated to evolve hydrogen gas that can be involved in the
reduction, or in the formation of the hydrogen gas. The resulting metal acts as a reducing
agent toward the complex metal oxide. A third reduction type includes the flow of
hydrogen gas at an elevated temperature to form water and the reduced complex metal
oxide. All reduction of complex metal oxides discussed within will include only the third
reduction process.
Reduction using 100% hydrogen gas was performed with extensive care due to the high
flammability of hydrogen and the fact that hydrogen gas is the only gas that, upon
expansion, is exothermic. For these reasons, all reactions with hydrogen gas were
performed in a tube furnace using a fused silica tube rather than the typical alumina tube,
due to the porosity of the alumna tube and the elevated temperature at which the reaction
takes place. All reactions using hydrogen gas also took place inside a fume hood rather
than out in the open for extra safety. The hydrogen gas was flowed at a rate of
approximately 1 ml per second through the reaction tube into an oil bubbler. Final
compounds of reduction are moisture sensitive; therefore, the hydrogen gas was also
flowed over glass wool coated with P2O5 to ensure the gas was dry prior to entering the
reaction tube. P2O5 was also placed at the end of the furnace to collect water that was

29 
 

evolved from the reaction to ensure diffusion of water back into the reaction tube did not
occur. The sample was placed in the tube as a non-pelleted powder and spread in an
alumina boat in a thin layer to increase surface exposure to the reactive gas. Intermittent
grindings inside a glove box were performed for homogenous sample exposure to the
reactive gas for optimal purity of the reduced phase.

3.4 Partial Oxygen Pressure Regulation
The ceramic method provides the ability to control the atmosphere within reaction by
using a reactive gas that provides different conditions to acquire desired products.
Oxidizing atmospheres such as oxygen, inert atmospheres such as nitrogen or argon gas,
or reducing atmospheres such as hydrogen gas, are all common gasses used in solid-state
reactions providing versatility in the synthesis process of new compounds. In the
synthesis of compounds, generally only one valence electronic state is available within
the transition metal of the complex compound. Therefore, synthesis of bulk compounds
with multiple oxidation states for the transition metals can only be achieved by increasing
the temperature of the reaction; thus, decreasing the oxygen reactivity and forming
oxygen deficient phases. In these circumstances, though common, the transition metal is
only partially divalent due to the oxygen deficiencies within the compound only resulting
in a small percentage while still maintaining the overall structure. Single crystals on the
other hand, can be synthesized regularly with multiple oxidation states of the transition
metal. This usually occurs within vacuum-sealed ampoules composed of different
transition metals such as iron, copper, or platinum, or the ampoule can be non-metallic

30 
 

such as glass or quartz. These different types of ampoules, though under vacuum, can
provide different types of oxidizing environments, like the metal's reduction potentials in
electrochemistry. If under the correct heating and cooling conditions, this can result in
single crystal formation of compounds that can have metals with multiple oxidation
states. As mentioned, synthesizing bulk materials with transition metals in multiple
oxidation states of significant magnitude cannot be performed using normal reactive
gasses such as oxygen or hydrogen. Therefore, the use of an oxygen buffer system to
control the oxygen partial pressure within a reaction vessel to synthesize complex
compounds, to this point, is the only known reaction system to be successful.
Controlled oxygen partial pressure reactions were performed using a two zone furnace;
one zone to control the temperature of a metal/metal oxide buffer such as Cu/Cu2O
(copper/copper(I)oxide) or Ni/NiO (nickel/nickel(II) oxide) and the other zone to control
the temperature of the sample. Nitrogen gas was flowed over glass wool covered with
P2O5 then flowed through the reaction tube with P2O5 at both ends to collect any water
that may have been released due to the increase in temperature during the reaction and
absorbed by the powders. The temperature of the sample for all reactions is kept constant,
while the temperature of the metal/metal oxide buffer is changed to appropriate
temperatures for the desired partial oxygen pressure characteristic of the specific buffer.
The reaction vessel is purged for at least one hour with nitrogen gas, the flow is then
stopped, and valve 1 as seen in Figure 4.2, is closed. The two zones of the furnace are
then initiated simultaneously with an identical ramp time, therefore, each zone reaches
the desired temperatures at the same time. At the time the temperatures reach their

31 
 

desired values, valve 2 is closed (Figure 4.2). The reaction is allowed to proceed for a
minimum of 48 hours, but can go longer due to the equilibrium state at which the system
is in. The reaction vessel is quenched to room temperature to prevent the buffer from
decreasing the partial oxygen pressure in the reaction vessel for a prolonged period at an
increased temperature therefore altering the final state the product. The sample is then
placed in a vial and taken into an inert glove box for further analysis.

3.5 Moisture and Oxygen Sensitive Techniques
Synthesis of metal-stable, oxygen sensitive, or moisture sensitive compounds requires an
increased measure of care to prevent decomposition, uncontrolled, or unknown reactions
from occurring that alter the properties of the final product. Properties of moisture
sensitive samples can be altered. This can be seen in the example of Na0.35CoO2·yH2O,
where in the case of y=1.3, the product is superconducting whereas the anhydrous
product or the further water-intercalated products do not exhibit such properties. The
discovery of the superconducting material was somewhat serendipitous due to the high
relative humidity in the area where the magnetic properties were measured, which
resulted in the intercalation of water molecules into the material. The original material
did not contain the intercalated water; thus, the exposure to the humid atmosphere
resulted in the final product's properties. Further structure refinement provided insight on
the water intercalation. Without the further analysis that was performed on the material,
incorrect data could have easily been presented, if the researcher assumed that, in fact,
the material they sent out for measurements maintained its original composition. To

32 
 

prevent any unwanted water intercalation, oxidation, or decomposition, samples can be
prepared in inert nitrogen or argon glove boxes.
All samples that are prepared are initially assumed to be moisture sensitive until further
analysis determines otherwise. This assumption is made for all sodium and lithium
containing compounds due to the general hygroscopic nature of most compounds that
contain these elements. This assumption is also made for meta-stable compounds
synthesized through solid-state reduction. To determine whether the sample is in fact
moisture sensitive, the sample can be washed in water then analyzed using PXD and
compared with previous PXD of the non-water washed phase. Precautions for sample
exposure to ambient conditions are minimized as much as possible. In the process of
solid-state reduction of Perovskite structure to infinite-layer structure, the fused silica
tube was made to fit within the antechamber of the nitrogen glove box, therefore, the
entire reaction vessel can be inserted into the glove box without any exposure. In the case
of α- and ß- NaFe2O3, due to the increased size of the reaction vessel to fit in a two-zone
furnace, the transfer of the entire vessel cannot occur. The sample is removed from the
reaction vessel, placed in a vial, immediately placed inside the glove box antechamber,
and evacuated. Removing the sample from the reaction vessel is performed under a
nitrogen gas flow creating an inert "blanket" over the sample when possible.
Materials that are synthesized from moisture sensitive precursors, such as NaFe2O3 with
the moisture sensitive precursor α-NaFeO2, or the preparation of α-NaFeO2 which is
synthesized from the hygroscopic compound NaOH and non-hygroscopic compound

33 
 

Fe2O3, are all prepared in a glove box. The compounds are ground, mixed, and pelleted,
if required, inside the glove box then transferred to an appropriate reaction vessel.

3.6 Chemical oxidation and reduction
Studying the properties of new materials can sometimes lead to further questions, such as
the effects of lithium intercalation and deintercalation on the overall structural integrity of
the new material. To answer such a question, a few different processes can occur; either
in situ PXD and electrochemical characterization, or oxidation via chemical lithium
intercalation and reduction via chemical lithium deintercalation. In situ PXD and
electrochemical characterization provides insight on the structural properties directly as
the cyclic voltammogram is produced. The downfall is the expensive set up required to
do such measurements. Generally, these measurements are performed at national
laboratories. Oxidation and reduction via chemical lithium intercalation and
deintercalation can provide the compound with deficient or inserted lithium from or into
the structure respectively; therefore allowing PXD to be performed on the samples
providing information on the structure and the effect of the inserted or removed lithium.

Chemical lithium intercalation was performed in a nitrogen glove box using 1 M nbutyllithium in hexanes. The ground powder sample was dispersed by stirring in pentane
and the n-butyllithium solution was added drop wise until the correct volume for the
required molar ratio was achieved. The mixture was stirred for at least 48 hours then

34 
 

washed with pentane and filtered. The resulting powder was then stored in an inert glove
box for further analysis.
Chemical lithium deintercalation was performed in a nitrogen glove bag due to the
oxidizing nature of the compounds used and the metal surroundings inside a glove box.
Bromine (Br2), iodine (I2), nitrosyl tetrafluorborate (NOBF4), or nitrosonium
tetrafluoroborate (NO2BF4) are all oxidizing reagents with different reduction potentials
providing the ability to deintercalate different amounts of lithium from a compound. The
sample is dispersed and stirred in acetonitrile (CH3CN) if the reaction was with NOBF4
or NO2BF4, or pentane if the reaction was with bromine or iodine. The slow addition of
the dilute oxidizing agent was then performed until the required volume to reach the
desired molar ratio was achieved. The mixture was then stirred for at least 48 hours or
until the reaction had come to completion by visual indication. Oxidation using bromine
and iodine, upon completion, would result in the color change from purple to clear
indicating the bromine or iodine had reacted to produce the bromide or iodide salt. The
color change provided information that the reaction has finished, which became useful
for some reactions that only require as little as 10 hours to become complete. NOBF4 and
NO2BF4 are both colorless; therefore the reaction cannot be visibly determined to be
complete, so controlled reaction times were performed. [8–12]

35 
 

REFERENCES

36 
 

REFERENCES

[1]

M.P. Pechini, US Patent, 3,3306,97, 1967.

[2]

L.-W. Tai, P. a. Lessing, J. Mater. Res. 7 (1992) 502.

[3]

L.-W. Tai, P. a. Lessing, J. Mater. Res. 7 (1992) 511.

[4]

J.-Y. Bae, D. Lim, H.-G. Yun, M. Kim, J. Jin, B.-S. Bae, RSC Adv. 2 (2012) 5524.

[5]

M. Kakihana, M. Yoshimura, Chem. Soc. Japan 72 (1999) 1427.

[6]

T. Sreethawong, Y. Suzuki, S. Yoshikawa, J. Solid-state Chem. 178 (2005) 329.

[7]

B.D. Fahlman, Materials Chemistry, Springer Netherlands, Dordrecht, 2011.

[8]

A.R. Wizansky, P.E. Rauch, F.J. DiSalvo, J. Solid-state Chem. 81 (1989) 203.

[9]

R. Tripathi, T.N. Ramesh, B.L. Ellis, L.F. Nazar, Angew. Chem. Int. Ed. Engl. 49
(2010) 8738.

[10]

B.L. Ellis, W.R.M. Makahnouk, Y. Makimura, K. Toghill, L.F. Nazar, (2007) 749.

[11]

L.A. de Picciotto, M.M. Thackeray, Mater. Res. Bull. 20 (1985) 1409.

[12]

K. Takada, H. Sakurai, E. Takayama-muromachi, Nature 422 (2003) 53.

 

37 
 

 

Chapter 4
Synthesis of NaFe2O3 Using Partial Oxygen Pressure Regulation Method
4.1 Introduction
Crystal structures of a large class of transition metal oxides AMO2 (A = alkali metal, M =
3d metal) can be described as an ordered rock-salt lattice with cubic close-packed oxygen
layers where A and M occupy the octahedral sites of alternating layers. These structures
contain MO2 layers composed of face-shared MO6 octahedra (CdI2-type layers) separated
by A+ ions. -NaFeO2 is the prototype compound, which lends its name to the structural
type.
Compounds with the -NaFeO2 structure are important from both fundamental and
technological points of view. NaMO2 phases demonstrate a large variety of physical
properties: frustrated magnetism in NaTiO2 [1], two successive orbital ordering
transitions in NaVO2 [2], broad fluctuating cross-over regimes in NaCrO2 [3, 4], strongly
coupled antiferromagnetic chains in NaMnO2 [5], two antiferromagnetic structures in NaFeO2 [6], and a nonmagnetic insulator state in NaCoO2 [7, 8]. From the technological
point of view, LiCoO2 is an important cathode material for Li-ion batteries [9, 10] and
NaxCoO2 is a candidate for thermoelectric applications [11-13].
Knowledge on AM2O3 rock-salt related compounds with layered A and M ordering is
very limited. The only known example is ß-NaFe2O3; its crystal structure consists of
double rock-salt layers of octahedrally coordinated iron (Fe2O3 blocks) separated by Na
ions (Figure 4.1) [14-17]. Generally speaking, 3d metal compounds with ß-NaFe2O3
structures are expected to demonstrate different properties than their -NaFeO2 type

38 
 

 

counterparts due to the difference in M oxidation states as well as in the dimensionality
of the 3d metal - oxygen blocks. At the same time, synthesis of the mixed valent
AM 22.5 O3 compounds is expected to require delicate control of the preparation

conditions. In the case of ß-NaFe2O3, the situation is complicated by the existence of an
-NaFe2O3 polymorph. Similarly to the ß phase, the crystal structure of -NaFe2O3
contains Fe2O3 blocks; however, these blocks are separated by two layers of octahedrally
coordinated cations with (1/3 Fe + 2/3 Na) in this shared crystallographic position (Figure
4.1).

Figure 4.1 Structure models of - and ß-NaFe2O3. The FeO6 octahedra are highlighted.
For - NaFe2O3, atoms in the shared Na/Fe position are shown as large blue spheres and
small red balls represent O ions. For ß-NaFe2O3, large green spheres correspond to Na
ions.

39 
 

 

Synthesis methods and crystal structures of - and ß-NaFe2O3 were reported in the 1970's
by El Balkhi et al. [14, 15]. For synthesis of the NaFe2O3 polymorphs, mixtures of
NaFeO2, Fe2O3, and Fe were sealed in ampoules and annealed at 1000 ºC for 48 hours.
The  and ß phases were prepared in sealed steel and Cu ampoules respectively. Fe
powder was also present in a separate tube inside the ampoule for the preparation of NaFe2O3.
It needs to be emphasized that the experimental technique used relies on kinetic factors
rather than on equilibrium conditions. At high temperatures, metal oxides generate some
oxygen pressure. Materials of the ampoules as well as Fe powder reacted with O2,
decreasing the partial O2 pressure (pO2), thus stabilizing NaFe2O3. At the same time, Fe
metal from the ampoule walls cannot be in equilibrium with a Fe3+ containing phase, i.e.
with NaFe2O3; therefore, reactions for extended times would result in further ampoule
wall oxidation to FeO as well as NaFe2O3 decomposition to Na doped wüstite (FeO).
This consideration is in agreement with a study of the equilibrium oxygen pressure in a
broad composition range in the Fe–Na–O system by measurements in electrochemical
cells with an oxygen-ion-conducting solid electrolyte [18].
A synthetic approach relying on equilibrium conditions is needed for the systematic
search for AM2O3 compounds with different 3d metals.

40 
 

 

4.2 Synthesis of NaFe2O3 via controlled oxygen pressure.
The syntheses of α- and ß-NaFe2O3 were carried out in closed silica tubes under
controlled partial oxygen pressure. Stoichiometric mixtures of α-NaFeO2, α-Fe2O3
(99.5%, Alfa Aesar) and Fe metal (≥99.9%, Sigma Aldrich) were mixed, pressed into
pellets, placed in alumina boats, and closed in quartz tubes together with a second boat
containing an oxygen buffer (oxygen getter) mixture. A split-open two-zone furnace was
used to control the temperature of both the sample and getter mixtures independently. All
samples were annealed at 850 °C for 48–72 hours. The temperature of the getter mixtures
was varied between 350 °C and 900 °C in 29 different experiments, allowing for the
precise control of pO2 in the system. After annealing, all quartz tubes were quenched to
ambient temperature in air. The syntheses of all samples and subsequent X-ray diffraction
measurements were successfully reproduced. It was reported that NaFe2O3 slowly
hydrolyzes in air [14]; therefore, all operations with initial mixtures and final compounds
were performed under nitrogen.
α-NaFeO2 was synthesized by the reaction of α-Fe2O3 (99.5%, Alfa Aesar) with a 20%
stoichiometric excess of NaOH (98.8%, Mallinckrodt Chemicals) under O2 flow at
500 °C for 48 hours. Excess NaOH was removed by washing with anhydrous methanol in
a nitrogen-filled glove box.
The samples were characterized by Powder X-ray diffraction (PXD) on a Bruker D8
Advanced with DAVINCI design diffractometer using Cu Kα radiation. The patterns

41 
 

 

were recorded at room temperature with a step size of 0.02° (2θ) on a zero background
sample holder.
High resolution synchrotron powder X-ray diffraction data were collected using beamline
11-BM at the Advanced Photon Source, Argonne National Laboratory with an average
wavelength λ=0.37382 Å. Discrete detectors covering an angular range from −6 to 16° 2θ
were scanned over a 34° 2θ range, with data points collected every 0.001° 2θ and a scan
speed of 0.01°/s. A three-axis translation stage holds the sample mounting and allows it
to be spun, typically at ∼5400 RPM. Rietveld refinement [19] of SPXD data was
performed with the GSAS [20] program and EXPGUI [21] interface. The profile function
4, which contains pseudo-Voigt, asymmetry, and microstrain broadening components,
was employed for all phases. The DRAWxtl program was employed to prepare the
difference Fourier map, (Figure 4.9) [22].
Elemental composition was determined by inductively coupled plasma (ICP)
spectroscopy using a Vista-MPX CCD Simultaneous ICP-OES instrument (Varian Inc.).
Oxygen stoichiometry was determined using triplicate iodometric titrations. KI (ACS
Grade, Jade Scientific), K2S2O3 (≥95%, Sigma Aldrich), KIO3 (ACS Grade, Columbus
Chemical), and Soluble Starch (ACS Grade, Columbus Chemical) were used in these
titrations [23].
The initial attempt at the synthesis of NaFe2O3 at 850 °C under N2 flow resulted in a
mixture of NaFeO2 and Fe3O4. In both compounds, the average iron oxidation state was
above that in AM22.5+O3. It appears that the presence of about 2 ppm O2 in the N2 used

42 
 

 

was sufficient to stabilize the Fe3+ compound NaFeO2, leading to the conclusion that the
pO2 of the system was too high. Synthesis in a sealed quartz ampoule resulted in the
mixture of α- and ß-NaFe2O3 polymorphs. At very low pO2, Na2O and Na doped wüstite
(FeO) can be expected as equilibrium phases [18]. From this data, it becomes obvious
that the synthesis of NaFe2O3 requires very careful control over oxygen stoichiometry.
Since both the α- and ß-NaFe2O3 polymorphs were prepared earlier at distinct, but
unknown pO2 [14], the dependence of phase composition vs. pO2 was studied in order to
prepare the pure compounds.
To control the pO2 during experiments, mixtures of Cu/Cu2O or Ni/NiO were employed
as oxygen buffers. Under equilibrium conditions for mixtures of metal and metal oxides,
the pO2 depends on both the chemical nature of the metal as well as temperature.
Equilibrium pO2 temperature dependences were determined experimentally for many
systems [24] and [25]. In the case of Ni/NiO, the oxygen fugacity follows the Equation:

logfO2 atm =9.36-

24930
T

519 K<T<1319 K

Equation 4.1

where T is temperature in Kelvin. For a Cu/Cu2O mixture, the temperature dependence of
the oxygen fugacity can be approximated as:
logfO2 atm =-

18162.2
T

+12.855-0.6747×ln(T) 700 K<T<1338 K Equation 4.2

The fugacity of a real gas is an effective pressure, which replaces the true mechanical
pressure in accurate chemical equilibrium calculations. For low partial pressure

43 
 

 

situations, as in our experiments, the fugacity is approximately equal to the partial vapor
pressure; therefore, partial oxygen pressure is used throughout the text instead of the
more thermodynamically strict fugacity term. This terminology is commonly accepted for
solid-state synthesis [26].
It was a formidable challenge to seal quartz ampoules with two boats inside while
maintaining the boat's positions without spilling the contents; therefore, a more
convenient approach was developed. A simple setup consisting of a quartz tube with
24/40 standard taper joints, 2 glass stopcocks and a bubbler with mineral oil was used
(Figure 4.2). Krytox high-temperature fluorinated grease was used with taper joints and
all joints were held with Keck clamps. After placing the boats in the quartz tube and
attaching adapters to the ends, the tube was purged with dry N2. Afterwards, valve 1
(Figure 4.2) was closed and the furnace was heated up to desired temperatures. The
increasing N2 pressure was relieved through the bubbler. After the temperatures were
reached, valve 2 was also closed. After annealing, all quartz tubes were quenched to
ambient temperature in air.

44 
 

 

Figure 4.2 Experimental setup for the synthesis of α- and ß-NaFe2O3 under controlled
partial O2 pressure.

4.3 Oxidation via sodium deintercalation
Oxidation via sodium deintercalation was performed using bromine as an oxidizing
agent, resulting in the formation of NaBr and the sodium deficient NaxFe2O3. Bromine
was added to acetonitrile and then added with a micropipette, in the amount of the desired
stoichiometric ratio, into a vial of acetonitrile with ß-NaFe2O3. This was all performed in
a nitrogen glove bag. The vial was sealed then stirred until the light to dark purple
colored solutions became clear. This color change is indicative that all of the Br2 reacted
to form NaBr. The resulting mixture of NaBr and NaxFe2O3 was then washed with
anhydrous ethanol to remove NaBr.

45 
 

 

4.4 Transition Metal Doping
Due to the interesting properties that many different compounds with the rock-salt
layered α-NaFeO2 structure type contain [1-13], further investigation into the doublelayered NaFe2O3 by doping into the iron site with similar transition metals was
performed.

Equation 4.3

Equation 4.4

Equation 4.5
When

mixing

of

a

non-oxidized

transition

metal

with

its

oxidized

form

stoichiometrically, such as iron (Fe) and iron (III) oxide, as seen in Equation 4.3, it can be
considered the addition of FeO. This results in the exact stoichiometry for all metals and
oxygen in the product. Due to this, similar compounds were reacted when attempting to
substitute iron with other transition metals, as seen in Equation 4.4. Manganese has an
oxide that exists with a stable 2+ oxidation state therefore direct substation can be
performed as seen in Equation 4.5.

46 
 

 

4.5 Results and Discussion
A series of tests were performed to ensure that low O2 pressures can be reached and
reliably maintained in a tube with standard taper joints. For example, Ni powder was
annealed in a tube at 850 °C for 20 days. The small amount of NiO formed can be related
to an O2 admixture in the N2 purge gas. During experiments on complex bismuth oxides
in the presence of oxygen buffers in an identical setup, the formation of metallic Bi from
Bi2O3 at low pO2 was observed [27]. This serves as additional proof of the method
validity. It appears that even if a small amount of O2 diffuses into the closed quartz tube,
the equilibrium pO2 is efficiently maintained by the buffer mixtures. The temperature of
the buffer mixture was never lower than 350 °C for all experiments in order to provide
high kinetic activity. PXD measurements were used after all experiments to confirm the
presence of both metal and metal oxides in the buffer mixtures.
Annealing of the nominal NaFe2O3 composition at ∼2×10−6 atm pO2 resulted in the
mixture of γ-NaFeO2 and Fe3O4 (Figure 4.3). The γ-NaFeO2 polymorph was reported to
be stable in air above 1010 °C and to transform into ß form at lower temperatures [28]. It
appears that at lower pO2, the γ-NaFeO2 region of stability extends to lower temperatures.
Synthesis at ∼2.7×10−10 atm pO2 led to the formation of ß-NaFe2O3 with γ-NaFeO2 and
Fe3O4 as secondary phases (Figure 4.4). Even lower pO2 was required to prepare pure ßNaFe2O3 (Figure 4.5) while α-NaFe2O3 forms at pO2 below ∼3.0×10−19 atm (Figure 4.6).
The experimental results of NaFe2O3 synthesis under controlled pO2 are summarized as a
schematic section of phase composition vs. pO2 diagram at 850 °C (Figure 4.7). Buffer

47 
 

 

mixtures used and corresponding pO2, as well as phase compositions are given in Table
4.1.

Figure 4.3 PXD of the resulting product of γ-NaFeO2 (red ticks) and Fe3O4 (green ticks)
after annealing the mixture of α-NaFeO2, Fe powder and Fe2O3 under flowing house
nitrogen at 850 °C.

48 
 

 

Figure 4.4 PXD of the resulting product of ß-NaFe2O3 (red ticks), γ-NaFeO2 (green
ticks), and Fe3O4 (blue ticks) after annealing the mixture of α-NaFeO2, Fe powder and
Fe2O3 at a pO2 below ∼1.0×10−10 atm at 850 °C.

49 
 

 

Figure 4.5 PXD of the resulting product of ß-NaFe2O3 (red ticks) after annealing the
mixture of α-NaFeO2, Fe powder, and Fe2O3 between the pO2 of ∼1.0×10−10 and
~1.0×10−19 atm at 850 °C.

50 
 

 

Figure 4.6 PXD of the resulting product of α-NaFe2O3 (red ticks) after annealing the
mixture of α-NaFeO2, Fe powder, and Fe2O3 below pO2 ∼1.0×10−19 atm at 850 °C.

51 
 

 

Figure 4.7 Schematic section of phase composition vs. partial O2 pressure diagram at
850 °C for Na/Fe=1/2.

52 
 

 

The most important result of this study is that the bulk preparation conditions for pure αand ß-NaFe2O3 were determined. The synthetic procedure is highly reproducible; it does
not require any special equipment and can be easily implemented in virtually any
chemistry laboratory.
Mixtures of α- and ß-NaFe2O3 polymorphs were obtained when furnace cooling was used
in initial attempts instead of sample quenching. At the same time, there are no α- and ßNaFe2O3 two-phase regions on the diagram at 850 °C. On cooling, the equilibrium pO2
controlled by the buffer mixtures decreased drastically, allowing for rapid ß-NaFe2O3 to
α-NaFe2O3 transformation. This observation shows that phase relationships are very
sensitive to both pO2 and temperature, further emphasizing importance of controlled pO2
method for the preparation of pure α- and ß-NaFe2O3.
Importantly, the 3-phase refinement, (Figure 4.8), led to insight into the chemistry of the
system. Smaller cell parameters of the ß-NaxFe2O3 phases can be related to a higher Fe
oxidation state accompanied by Na deficiency. The high volatility of sodium oxide at the
experimental temperature (850 °C) is the most probable reason for Na deficiency in the
sample. The devitrification (crystallization) of fused silica tubes was apparent near the
sample boats. The quartz devitrification process was accelerated in the presence of the
reaction product of Na2O with SiO2, which acts as crystallization seeds.

53 
 

 

Figure 4.8 Observed (crosses), calculated (solid line), difference (bottom solid line)
profiles and Bragg reflections (tick marks) for the final Rietveld refinement of ß-NaFe2O3
and ß-NaxFe2O3, from SXPD data. The tick marks correspond to the positions (from top
down) of ß-NaxFe2O3, ß-NaFe2O3, and Fe3O4. The section at higher angles (>16° 2θ) is
enlarged by a factor of 20 for clarity.

54 
 

 

Table 4.1 Crystallographic Data for the Rietveld Refinement of ß-NaFe2O3 and ßNa0.78Fe2O3 from SXPD data. (Goodness of Fit Parameters: 2 = 5.87, Rwp = 10.6%).
Crystallographic Data for ß-Na1Fe2O3.a

Atom

Wyckoff
position

x

y

z

102 Uiso (Å2)

Occ.

Na

1b

0

0

0.5

1.06(3)

1

Fe

2d

1/3

2/3

0.1675(1)

0.64(1)

1

O1

1a

0

0

0

2.26(7)

1

O2

2d

1/3

2/3

0.7040(2)

1.03(3)

1



a

Space group P 3m1 (No. 164). a = 3.05795(1) Å, c = 7.79807(2) Å; weight fraction

73%.
Crystallographic Data for ß-Na0.78Fe2O3.b

Atom

b

Wyckoff
position

x

y

z

102 Uiso (Å2)

Occ.

Na

1b

0

0

0.5

1.06(2)

0.78(1)

Fe

2d

1/3

2/3

0.1705(2)

0.77(2)

1

O1

1a

0

0

0

2.28(9)

1

O2

2d

1/3

2/3

0.6416(6)

2.28(9)

1



Space group P 3m1 (No. 164). a = 3.05106(4) Å, c = 7.7898(2) Å; weight fraction

21%.

55 
 

 

Table 4.2 Phase composition of nominal “NaFe2O3” stoichiometry at different partial O2 pressures at 850 ºC.

Phase composition

Equilibrium partial O2
pressure, atm [19, 20]

Oxygen buffer
(getter) mixture

Temperature of the oxygen
buffer mixture, ºC

-NaFeO2, Fe3O4

~2  10-6

- (N2 flow)

-

ß-NaFe2O3, NaFeO2, Fe3O4

2.7  10-10

Cu/Cu2O

750

3.5  10-11

Cu/Cu2O

700

2+
3+
Na 0.88(2) Fe 0.93
Fe1.07
O 2.98(2)

1.8  10-14

Cu/Cu2O

550

2+
3+
Na 0.80(2) Fe 0.89
Fe1.11
O 2.96(2)

7.5  10-16

Cu/Cu2O

500

2+
3+
Na 0.88(2) Fe 0.9
Fe1.1
O 2.99(2)

5.5  10-17

Ni/NiO

700

2+
3+
Na 0.82(2) Fe 0.89
Fe1.11
O 2.97(2)

3.0  10-19

Cu/Cu2O

400

2+
3+
Na 0.76(2) Fe1.00
Fe1.00
O 2.88(2)

6.4  10-20

Ni/NiO

600

1.2  10-21

Ni/NiO

550

2+
Na 0.92(2) Fe1.09
Fe3+
0.91O 2.91(2)

1.3  10-23

Ni/NiO

500

2+
3+
Na 0.74(2) Fe 0.98
Fe1.02
O 2.88(2)

2.3  10-31

Ni/NiO

350

ß-NaFe2O3

-NaFe2O3

56 
 

Chemical analysis results

 

ICP analysis confirmed Na loss via volatilization, giving an average composition of ßNa0.88(2)Fe2.00O3 for the sample. Knowing the sample composition and fraction of each
phase present, one can calculate Na content (x) for ß-NaxFe2O3. ß-NaFe2O3, ß-NaxFe2O3,
and Fe3O4 weight fractions were calculated during the Rietveld refinement and were
equal to 0.73, 0.21, and 0.06, respectively, giving a derived composition of ßNa0.76Fe2O3. The Na content in ß-NaxFe2O3 refined from SPXD data was equal to 0.78, in
good agreement with the chemical analysis results.
According to the SPXD data, α-NaFe2O3 was essentially a pure phase. According to ICP
analysis, a compositional formula of Na0.98(2)Fe2.00O3 was obtained. Peaks related to the
ß-NaxFe2O3 polymorph were absent in the SPXD pattern while the strongest peak of the
Fe3O4 admixture was at ∼0.1 % of relative intensity level. Attempts to perform a Rietveld
refinement using the literature structural model [14] were unsuccessful (χ2≈22). While
peak positions and peak profile could be fitted adequately, there were substantial
differences in peak intensities, which did not change systematically with peak indices,
indicating a problem with the structural model rather than preferred orientation. Attempts
to include preferred orientation in the refinement were unsuccessful. Difference Fourier
maps were calculated to determine the cause of said discrepancies. The difference Fourier
map indicated regions of significant electron density around the shared Na/Fe and
adjacent O3 positions (Figure 4.9). The extra electron density has a spherical shape near
the O3 position, suggesting splitting of the position. The donut-like shape of the extra
electron density around the shared Na/Fe position can be related to a Na/Fe shift from the
average position. Attempts to split O3 and the shared Na/Fe positions to account for the

57 
 

 

observed differences did not result in an improvement to the refinement; therefore, the
crystal structure published by El Balkhi et al. [14] describes the average structure well,
while the real structure is more complex with substantial disorder in the shared Na/Fe and
O3 positions. Pair distribution function (PDF) analysis is needed to study the structure of
α-NaFe2O3 in further detail.

58 
 

 

Figure 4.9 Difference Fourier map for α-NaFe2O3 indicating large residual electron
density (green) around the Na/Fe and O3 positions. Atomic positions are denoted for this
layered structure and Fe–O octahedra are highlighted. Blue and red spheres represent
atoms in the Na/Fe and O positions, respectively.

59 
 

 

A nominal NaFe2O3 composition of both α- and ß-NaFe2O3 polymorphs implies the same
formal Fe oxidation state in both compounds. It is not clear why α-NaFe2O3 begins to
form at a pO2 eight orders of magnitude lower than ß-NaFe2O3. To clarify the origin of
such a drastic difference in the stability conditions, stoichiometry of representative
samples were determined by a combination of ICP analysis and iodometric titration. In
spite of slight differences in Na content, oxygen stoichiometry of ß-NaFe2O3 samples
prepared in a wide range of pO2 (3.5×10−11−5.5×10−17) were very close to ideal with a
negligible amount of oxygen vacancies (Table 4.1). The Fe2+/Fe3+ ratio calculated from
the sample's stoichiometry and charge balance requirements resulted in a 0.9:1.1 ratio
(within error) for all ß-NaFe2O3 samples. Contrary to our expectations, oxygen
nonstoichiometry in α-NaFe2O3 was minimal, with 0.1 oxygen vacancies per formula unit
(Table 4.1). The Fe2+/Fe3+ ratio was between 1.0/1.0 and 1.1/0.9 for α-NaFe2O3 samples
prepared under different experimental conditions.
It has been shown that order in layered structures is favored by a large difference in the
ionic radii (R) of A and M. An ionic radii ratio RM/RA<0.86 is typical for ordered phases
[30], which coincides well with the empirical rule requiring less than a 15% ionic radii
difference between two elements for substitution to occur. Theoretical investigation
identified that the elastic energy associated with ions of different size stabilizes the
ordering of M and A into alternate layers [31]. In high spin configurations with
coordination number 6, the Shannon ionic radii values for Fe2+ and Fe3+ ions are equal to
0.920 and 0.785 Å, respectively [32]. The formal Fe ionic radii for α- and ß-NaFe2O3 are
very close: 0.859 and 0.846 Å, respectively. Taking the large ionic radius of Na+ (1.16 Å)

60 
 

 

into consideration, one can see that RFe/RNa≈0.74, which is well below the 0.86 value
required for ordering. According to previous reports in the literature, structural strain has
been shown to facilitate structural transformations [33]. The global instability index (GII)
is often utilized as a metric of structural stress due to deviation of the bond valence sums
of ions from ideal values [34]. The GII calculated for ß-Na1Fe2O3 using structural data
from Table 4.2 was 0.14. While the GII indicates presence of structural strain, values
above 0.2 generally correspond to unstable structures; thus, the simple radii ratio picture
cannot unveil a reason for the ß- to α-NaFe2O3 transformation at low pO2.
A correlation between Na content and α- or ß-NaFe2O3 polymorph formation (Table 4.2)
was not established. Differences in Na content seem to be related to minute details of
individual sample preparation, which indicates that further optimization of the synthetic
conditions, is needed. Rietveld refinement of SPXD data for ß-NaFe2O3 points to the
presence of two phases with different Na compositions. It is probable that multiple α- or
ß-NaFe2O3 type line phases with well defined Na stoichiometry exist, as shown
previously in NaxCoO2 [35]. Importantly, the existence of ß-NaxFe2O3 indicates that
partial deintercalation of Na should be possible by soft chemistry methods at room
temperature, allowing for the fine tuning of the properties of the compounds by
controlling the oxidation state of Fe. In the case of NaxCoO2, such fine tuning results in a
complex Na-content vs. physical properties diagram [36]. It would be intriguing to
explore the properties of ß-NaxFe2O3 with different Na compositions.

61 
 

 

Oxygen stoichiometry of α- and ß-NaFe2O3 differs by ∼0.1 per formula unit. Values very
close to full oxygen stoichiometry were consistently observed for ß-NaFe2O3 while the αNaFe2O3 phase compositions were systematically close to the 2.9 oxygen stoichiometry
(Table 4.1). Based on the above consideration, it seems reasonable to assume that a
higher stability of α-NaFe2O3 at low pO2 is related to its ability to accommodate a greater
amount of oxygen vacancies than that for ß-NaFe2O3.
The sodium volatility during the reaction led to further investigation of controlled sodium
deintercalation using bromine. The bromine oxidation potential to deintercalate sodium
against iron in the 3+ oxidation states is minimal; therefore, the oxidation of Fe2+ to Fe3+
in NaFe2O3 will occur and form the stable salt NaBr, as seen in Figure 4.10.

62 
 

 

Figure 4.10 PXD of unwashed sodium deintercalated ß-NaxFe2O3 (red ticks) with the
byproduct NaBr (green ticks).
Similar to the results seen in Table 4.1, the result of the controlled removal of sodium
from the initial compound will lead to the formation of an increased amount of Fe3+.
Chemical sodium deintercalation was performed in increasing increments of five mole
percent of bromine starting at 0% and ending at 50%. This would result in a phase
diagram of the sodium deficient NaxFe2O3 from the initial sodium deficient NaFe2O3
from the initial synthetic reaction, to approximately 0.5 parts per formula unit of sodium
removed, or approximately Na0.5Fe2O3. The samples were prepared starting with the
63 
 

 

initial compound ß-Na0.96Fe2O3, determined from ICP, and reacted as previously
described. The resulting powders were washed and analyzed for purity using a powder xray diffractometer. After the samples were deemed pure, further analysis using ICP to
determine the actual stoichiometry of the sodium deficient phase was performed. Le Bail
fit, to acquire cell parameters, was then performed on the pure phases and the cell
parameters were plotted against the sodium content as seen in Figures 4.11-4.13.

Figure 4.11 Cell parameter a as a function of parts per formula unit of sodium in
NaxFe2O3 Le Bail fit from PXD.

64 
 

 

According to Vegard's law, the cell parameters are expected to continue linearly as a
function of sodium content in a crystal structure. This trend is seen in both the a and c
parameters along with the volume, which is expected due the relationship of volume to
the a and c parameters of a trigonal space group.

Figure 4.12 Cell parameter c as a function of parts per formula unit of sodium in
NaxFe2O3 Le Bail fit from PXD.

65 
 

 

As it can be seen in the Figures 4.11-4.13, there is an absence of the stable phases with
the approximate formulas of Na0.75Fe2O3 and Na0.65Fe2O3. Multiple attempts were made
to synthesize these two compounds to no avail. Therefore, it can be determined that these
two sodium deficient phases are not kinetically stable. Coupled with the Rietveld refined
compound in Table 4.1, Na0.78Fe2O3, it can be then said that the stability of the structure
within this range only varies by a few mole percent of sodium.

Figure 4.13 Volume as a function of parts per formula unit of sodium in NaxFe2O3 Le
Bail fit from PXD.

66 
 

 

Figures 4.11-4.13 are summarized in table 4.3 with the determined Le Bail fit values
from laboratory PXD.
Table 4.3 NaxFe2O3 Cell parameters a, c and volume determined from Le Bail fit from
PXD.
(x) in NaxFe2O3

Cell Parameter a

Cell Parameter c

Cell Volume

0.96

3.0792

7.7765

63.865

0.91

3.0746

7.7746

63.646

0.87

3.0693

7.7705

63.394

0.81

3.0640

7.7658

63.140

0.76

3.0576

7.7625

62.848

0.67

3.0461

7.7576

62.336

0.56

3.0356

7.7513

61.858

0.51

3.0317

7.7476

61.669

0.46

3.0232

7.7448

61.303

Synthesis of transition metal doped NaFe2-XMxO3 was performed using manganese,
cobalt, nickel, and copper. These transition metals were chosen due to their ability to
form stable 2+ oxidation states along with having relatively similar ionic radii to iron.
These two requirements are essential for the doping to be successful. Initial attempts were
performed at a pO2 of about 10-23 atm. This pO2 is in the stability zone of α-NaFe2O3, but
due to the substituted transition metals' oxidation potentials different from that of iron, it
can be expected that the reactivity of the doped NaFe2-xMxO3 will differ from the

67 
 

 

undoped NaFe2O3. With that said, the α-NaFe2O3 zone of stability was chosen to be on
the cautious side. In the case that the reactivity of the doped transition metals required
less oxidizing conditions to form, the formation of the ß- structure would, at the very
least, appear in a mixture of phases in a PXD. This was observed with the cobalt-doped
compounds NaFe1.75Co0.25O3 and NaFe1.85Co0.15O3. As seen in Figure 4.14 at a pO2 of
about 10-23 atm, with NaFe0.5Co0.5O3, the α-NaFe2O3 structure formed.

Figure 4.14 NaFe1.5Co0.5O3 synthesized at 850 °C at a pO2 of about 10-23 atm resulting in
the formation of the α-NaFe2O3 structure type (red ticks).

68 
 

 

Similarly, at a pO2 of about 10-23 atm, NaFe1.75Co0.25O3 (Figure 4.15) and NaFe1.85Co0.15
(Figure 4.16) were synthesized resulting in mixtures of ß-NaFe2O3 structure and NaFeO2
for NaFe1.75Co0.25O3, and ß-NaFe2O3 mixed with NaFeO2 and Fe3O4 for NaFe1.85Co0.15.
This illustrates nicely the effect of the different doping amounts on reactivity at the same
pO2.

Figure 4.15 NaFe1.75Co0.25O3 synthesized at 850 °C at a pO2 of about 10-23 atm resulting
in the formation of the ß-NaFe2O3 structure type (red) and γ-NaFeO2 (green).

69 
 

 

Attempts to synthesize NaFe1.25Co0.75O3 were deemed unsuccessful with the appearance
of cobalt oxide. With this information, it can be said that doping of much more than 0.5
parts per formula unit of a transition metal will lead to the appearance of the respective
transmission metal oxide.

Figure 4.16 PXD of NaFe1.85Co0.15O3 synthesized at 850 °C at a pO2 of about 10-23 atm
resulting in the formation of the ß-NaFe2O3 structure type (red), γ-NaFeO2 (green), and
Fe3O4 (blue).

70 
 

 

Doping NaFe2O3 with manganese was performed and the initial attempt was deemed
successful. The attempt to synthesize NaFe1.5Mn0.5O3 was performed at 850 °C at a pO2
of about 10-23 atm. The resulting PXD, as seen in Figure 4.17, shows the formation of the
α-NaFe2O3 structure, γ-NaFeO2, and Fe3O4. These results with both cobalt and
manganese doping were deemed successful due to the formation of either the α- or ßphase, indicating that further attempts with different buffer conditions to form the pure
phases of both α- and ß-NaFe2-xMxO3 will form the pure phases. 

Figure 4.17 NaFe1.5Mn0.5O3 synthesized at 850 °C at a pO2 of about 10-23 atm resulting in
the formation of the α-NaFe2O3 (red), structure type γ-NaFeO2 (blue), and Fe3O4 (green).

71 
 

 

Attempts to dope copper and nickel into NaFe2O3 were deemed unsuccessful. The
resulting PXD of both reactions still possessed CuO and NiO indicating no reaction
occurred. This result is not surprising due to the relatively high stability of the two
oxides.

4.6 Conclusion
A phase composition vs. partial O2 pressure diagram at 850 °C for Na/Fe=1/2 was
determined which allowed for the reproducible synthesis of the divalent sodium iron
oxides, α- and ß-NaFe2O3 polymorphs. Sodium extraction from ß-NaFe2O3 at room
temperature was successfully investigated thus modifying the average Fe oxidation state.
Doping cobalt and manganese into the iron position was also successfully performed and
illustrates the ability to form dopant series of the compound NaFe2-xMxO3 (0<x<0.5).
These results lead toward the unexplored areas of possibly rich chemistry and physics of
the rock-salt related homolog series AMnOn+1 with layered A and M ordering.

4.7 Future Direction
First, the sodium deficient compounds that have been synthesized require magnetic
measurements from room temperature to 4 K at 1000 Oe and 5000 Oe respectively. This
can present an interesting phase diagram of the materials due to the already measured
magnetic data as seen in Figure 4.18 of ß-NaFe2O3. The magnetic response of the
divalent sodium iron oxide indicates possible interesting properties that are possessed by

72 
 

 

the material. Other divalent iron materials, such as Fe3O4, have the spinel structure that
has iron in both the tetrahedral and octahedral coordination. This is not the case for iron
in NaFe2O3. Different electronic ordering is expected of this material by decreasing the
sodium content and altering the average oxidation state of iron.

Figure 4.18 Magnetic response of NaFe2O3 as a function of temperature. The
measurement was performed from room temperature to 4 K at 5000 Oe.

Similarly, because it was established that transition metal doping into the iron position
can be attained, determining the conditions to successfully synthesize both α- and ßNaFe2-xMxO3 (M= Mn and Co) and (0<x<0.5) should be continued. Furthermore, after
the previously mentioned materials have been synthesized, chemical oxidation via
sodium deintercalation, as previously described with ß-NaFe2O3, could be performed and

73 
 

 

the magnetic response of each material investigated. This would present a series of
compounds with the general formula NayFe2-xMxO3, where (0<y<0.5), (M= Mn and Co),
and (0<x<0.5).
Continuing with just the transition metal doping into NaFe2O3, investigation of NaFe2xMxO3

as thermoelectric materials could also present possible interesting results. As seen

in Figure 4.19, ß-NaFe2O3 underwent transitions from being a semiconductor to insulator
at about 250 K then back to electrically conducting at about 100 K. Each of the
transitions correlated at the same temperatures observed the electron ordering or
disordering in the magnetic response in Figure 4.18. The materials illustrated the
semiconducting properties between 700 K and 250 K, at which point the material
undergoes a Verway like transition into an insulator until about 100 K where the
electrical conductivity begins to sharply increase.

74 
 

 

Figure 4.19 Thermopower (green) and electrical conductivity (black) measurements of ßNaFe2O3 with electrical conductivity showing transitions occurring at about 250 K and at
about 100 K (black).

75 
 

 

REFERENCES

76 
 

 

REFERENCES

[1]A.P. Ramirez Annu. Rev. Mater. Sci., 24 (1994), pp. 453–480
[2]T.M. McQueen, P.W. Stephens, Q. Huang, T. Klimczuk, F. Ronning, R.J. Cava
Phys. Rev. Lett., 101 (2008), p. 166402
[3]D. Hsieh, D. Qian, R.F. Berger, R.J. Cava, J.W. Lynn, Q. Huang, M.Z. Hasan Phys. B
(Amsterdam, Neth.), 403 (2008), pp. 1341–1343
[4]D. Hsieh, D. Qian, R.F. Berger, R.J. Cava, J.W. Lynn, Q. Huang, M.Z. Hasan J. Phys.
Chem. Solids, 69 (2008), pp. 3174–3175
[5]C. Stock, L.C. Chapon, O. Adamopoulos, A. Lappas, M. Giot, J.W. Taylor, M.A.
Green, C.M. Brown, P.G. Radaelli Phys. Rev. Lett., 103 (2009), p. 077202
[6]T. McQueen, Q. Huang, J.W. Lynn, R.F. Berger, T. Klimczuk, B.G. Ueland, P.
Schiffer, R.J. Cava Phys. Rev. B, 76 (2007), p. 024420
[7]G. Lang, J. Bobroff, H. Alloul, P. Mendels, N. Blanchard, G. Collin Phys. Rev. B, 72
(2005), p. 094404
[8]C. de Vaulx, M.H. Julien, C. Berthier, M. Horvatic, P. Bordet, V. Simonet, D.P. Chen,
C.T. Lin Phys. Rev. Lett., 95 (2005), p. 186405
[9]K. Mizushima, P.C. Jones, P.J. Wiseman, J.B. Goodenough Mater. Res. Bull., 15
(1980), pp. 783–789
[10]J.B. Goodenough, K. Mizushima, T. Takeda Jpn. J. Appl. Phys., 19 (1980), pp. 305–
313
[11]I. Terasaki, Y. Sasago, K. Uchinokura Phys. Rev. B, 56 (1997), pp. 12685–12687
[12]M. Lee, L. Viciu, L. Li, Y.Y. Wang, M.L. Foo, S. Watauchi, R.A. Pascal, R.J. Cava,
N.P. Ong Nat. Mater., 5 (2006), pp. 537–540
[13]M. Lee, L. Viciu, L. Li, Y. Wang, M.L. Foo, S. Watauchi, R.A. Pascal, R.J. Cava,
N.P. Ong Physica B: Condens. Matter (Amsterdam, Neth.), 403 (2008), pp. 1564–1568

77 
 

 

[14]A.M. El Balkhi, M. Zanne, C. Gleitzer, A. Courtois J. Solid State Chem., 18 (1976),
pp. 293–297
[15]A.M. El Balkhi, A. Courtois, M. Zanne, C. Gleitzer C. R. Acad. Sci. C. Chim., 285
(1977), pp. 129–131
[16]A.M. El Balkhi, F. Jeannot, M. Zanne, C. Gleitzer Rev. Chim. Miner., 14 (1977), pp.
491–497
[17]A.M. El Balkhi, F. Jeannot, M. Zanne, C. Gleitzer Rev. Chim. Miner., 14 (1977), pp.
557–562
[18]A.A. Lykasov, M.S. Pavlovskaya Inorg. Mater., 39 (2003), pp. 1088–1091
[19]H.M. Rietveld J. Appl. Crystallogr., 2 (1969), pp. 65–71
[20]A.C. Larson, R.B. Von Dreele, Los Alamos National Laboratory Report LAUR 86
(2000) 748.
[21]B.H. Toby J. Appl. Crystallogr., 34 (2001), pp. 210–213
[22]L.W. Finger, M. Kroeker, B.H. Toby J. Appl. Crystallogr., 40 (2007), pp. 188–192
[23]A. Manthiram, J.S. Swinnea, Z.T. Sui, H. Steinfink, J.B. Goodenough J. Amer.
Chem. Soc., 109 (1987), pp. 6667–6669
[24]J.S. Huebner, M. Sato Am. Mineral., 55 (1970), pp. 934–952
[25]H.S. Oneill, M.I. Pownceby Contrib. Mineral. Petrol., 114 (1993), pp. 296–314
[26]V.A. Alyoshin, D.A. Mikhailova, E.V. Antipov Physica C (Amsterdam, Neth.), 271
(1996), pp. 197–204
[27] V. Poltavets, P. Kazin, M. Jansen Solid State Sci., 8 (2006), pp. 1152–1159
[28]A.R. West Nature (London, U. K.), 249 (1974), pp. 245–246
[29]W.C. Hamilton Acta Crystallogr., 18 (1965), pp. 502–510
[30]S. Schougaard, J. Bréger, M. Jiang, C. Grey, J. Goodenough Adv. Mater., 18 (2006),
pp. 905–909
[31]E.J. Wu, P.D. Tepesch, G. Ceder Philos. Mag. B, 77 (1998), pp. 1039–1047

78 
 

 

[32]R.D. Shannon Acta Crystallogr., Sect. A, 32 (1976), pp. 751–767
[33]V.V. Poltavets, K.A. Lokshin, M. Croft, T.K. Mandal, T. Egami, M. Greenblatt
Inorg. Chem., 46 (2007), pp. 10887–10891
[34]I.D. Brown Chem. Rev., 109 (2009), pp. 6858–6919
[35]R. Berthelot, D. Carlier, C. Delmas Nat. Mater., 10 (2011), pp. 74–80
[36]N.P. Ong, R.J. Cava Science, 305 (2004), pp. 52–53

79 
 

 

Chapter 5
Synthesis of Tunnel Polymorph LiFeO2 and Characterization of the Electrochemical
Performance of Cobalt Doping
5.1 Introduction
Ideal cathode materials need to be both light in mass and have a high cycling voltage;
therefore, the ideal lithium ion battery would be LiF. This would contain both a low mass
and a high cycling voltage due to the high electrode potentials of both lithium and
fluorine. However, due to safety issues, this "ideal" battery is not practical. Transition
metal oxides have provided many different compounds that have high cycling voltages
and relatively low masses, the properties that are ideal for good cathode materials.
LiCoO2 illustrates many properties that are promising for a cathode material. It only
contains one early transition metal and only two oxygen atoms making it among the
lightest of cathode materials. The cycling voltage of LiCoO2 is at about 3.95 V, which is
approaching the upper boundary of commonly used electrolytes. These two properties
alone do not represent everything a good cathode material needs to contain. LiCoO2 also
has a two-dimensional diffusion pathway of the lithium between the CoO6 octahedra and
high electrical conductivity, thus requiring no special nano-particle preparation or carbon
nanocoating [1–3].
Since the patent of the intercalation cathode material LiCoO2 with the α-NaFeO2
structure type by Goodenough in 1980, and the intercalation patent on graphite in an
organic solvent in 1981, Sony began to commercially produce lithium ion batteries in the
early 1990's [3]. Because of environmental concerns of the toxicity of cobalt, developing

80 
 

 

lithium ion batteries that reduce or eliminate the amount of cobalt used in the material has
been an ever-increasing field in cathode research. This eventually led to other transition
metal layered cathodes such as LiNiO2 [4], LiNi1-xCoxO2 [5,6], and LiCo1/3Mn1/3Ni1/3O2
[7,8]. Spinel structure material such as LiMn2O4, though lower in energy density, is
commonly used due to its low cost and high abundance [9,10]. The olivine structure
LiFePO4 proposed by Goodenough in 1997, [11] increased momentum in cathode
research eventually leading to an almost perfect cycling material for thousands of cycles.
Originally, LiFePO4 demonstrated poor cycle ability due to its greatest flaw of low
electrical conductivity. This was eventually resolved by carbon nanocoating, leading to
almost 100 % cycling capacity for over 3000 cycles as previously mentioned [12]. The
structure type of LiFePO4, unlike the layered α-NaFeO2 structure type, has tunnel-like
channels that the lithium ion travels out of and into during deintercalation and
intercalation without any structural transformation. LiCoO2, in comparison, undergoes an
irreversible transformation from rhombohedral crystal structure to a monoclinic crystal
structure at about 50% deintercalation of lithium [13,14]. Table 5.1 shows the common
transition metal oxides that are used as cathode materials in secondary lithium ion
batteries.
Although LiFePO4 does not undergo any crystal structure transformation and cycles
thousands of times better than LiCoO2, LiCoO2 is still the preferred material because of
lower production cost, due to not having to be prepared as nano-particles or carbon
nanocoated.

81 
 

 

Table 5.1 Common transition metal oxide cathode materials.
Compound

Structure
Type

Theoretical
Capacity
(mAh/g)

Experimental
Capacity
(mAh/g)

Cycle
Cycle
Voltage
Number
(V)

LiCoO2[1]

Layered
α-NaFeO2

274

160

3.95

~300

LiFePO4[11,15,16]

Olivine

170

160

3.5

>3,000

LiMn2O4[9]

Spinel

144

110

3.2

>1000

Due to the success of LiCoO2 with the layered α-NaFeO2 structure, as previously stated
similar layered compounds were investigated, including LiFeO2.
LiFeO2 has nine known polymorphs including four ordered and disordered cubic rocksalt (NaCl) structure polymorphs. They differ depending on the order or disorder of the
cations and are referred to as α-, ß-, ß'-, and γ-LiFeO2 [17], the layered α-NaFeO2
structure [18], the corrugated-layer [18], Goethite-type [2], Hollandite-type [19], and the
tetrahedral LiFeO2 (T-LiFeO2) [20]. This compound is of particular interest due to iron’s
cost, abundance, and benign toxicity. The nine polymorphs can be synthesized though a
variety of different methods including solid state reaction, low-temperature molten salt
synthesis, hydrothermal, solvothermal, ion-exchange, and heat treatment. Table 5.2
illustrates a few synthetic procedures, the structure, and the best electrochemical
performance for the different LiFeO2 polymorphs.

82 
 

 

Table 5.2 Polymorphs of LiFeO2, structure, preparation, preparation procedure, and best electrochemical performance.
Type

Structure

Preparation

Preparation Procedure

Best Electrochemical
Performance (mAh/g)

α-LiFeO2[17,21]

Fm3m (cubic)
Cations disordered

Solid State Reaction

Li2CO3 and Fe2O3 annealed
above 700 °C

150

-LiFeO2[17,21]

Fm3m (cubic)
Cations disordered

Solid State Reaction

Li2CO3 and Fe2O3 annealed
at 400 °C

*

Heat Treatment

α-LiFeO2 heated at 400 °C

*

'-LiFeO2[17,21]

Fm3m (cubic)
Cations disordered
Fm3m (cubic)
Cations ordered
R-3m
(rhombohedral)
Pbnm
(orthorhombic)

Li2CO3 and Fe2O3 annealed
above 600 °C
α-NaFeO2 and LiNO3
Layered-LiFeO2[18,21]
Ion-exchange
eutectic melt
α-FeOOH and C2H5OLi in
Geothite- LiFeO2[2,21]
Ion-exchange
ethanol at 170 °C
ß-FeOOH and LiOH
HollanditeI4/3 (tetragonal)
Ion-exchange
eutectic melt
LiFeO2[19,21]
Corrugated layerPmnm
γ-FeOOH and C2H5OLi in
Ion-exchange
LiFeO2[2,21,22]
(orthorhombic)
ethanol at 170 °C
Pna21
ß-NaFeO2 and LiNO3
T-LiFeO2[20]
Ion-exchange
eutectic melt
(orthorhombic)
* Do not have electrochemical performance due to the blocked diffusion pathway by the ordering of the cations.
γ-LiFeO2[17,21]

Solid State Reaction

83 
 

*
115
70
150
150
120

 

As seen in Table 5.2, the layered α-NaFeO2 structure type LiFeO2 cannot be prepared by
the conventional high-temperature solid-state route, rather the precursor α-NaFeO2 must
first be prepared followed by ion exchange with a lithium salt. The direct solid-state route
leads to the synthesis of different crystallographic forms of LiFeO2 with the rock-salt
structure. These ordered and disordered rock-salt structures do not show favorable
intercalation and deintercalation due to their structure type, with the exception of αLiFeO2, of which the capacity eventually fades to less than 100 mAh/g. A nanoporous αLiFeO2 carbon composite was shown to have an initial capacity of 255 mAh/g, about 90
% its theoretical capacity of 282 mAh/g. This further illustrates the advancements nanoparticle preparation and carbon coating have on materials [23].
Particle size of LiFeO2 with the layered α-NaFeO2 structure was investigated and it was
similarly determined that smaller particle size increased the discharge capacity. It was
established that samples with an average particle size of 400 nm exhibited a capacity of
80 mAh/g while samples with an average particle size of 40 nm showed a higher capacity
of 115 mAh/g. Each study was performed with a cycling voltage between 2.0 and 4.5 V
[18].
T-LiFeO2 is of particular interest due to its tunnel-like structure similar to that seen in
LiFePO4 [20]. This structure is also the only LiFeO2 polymorph that contains iron-oxygen
in tetrahedral coordination, whereas the other polymorphs are octahedral. T-LiFeO2 is
synthesized through ion exchange from the precursor ß-NaFeO2, which has a similar
structure. The precursor was ball milled prior to ion exchange to decrease the particle size
of the final compound. Electrochemical characterization of the T-LiFeO2 polymorph

84 
 

 

eventually stabilized with a capacity of about 100 mAh/g, but further analysis of the
compounds showed a reaction with the electrolyte resulting in the formation of hydrogen,
carbon monoxide, and carbon dioxide evolution. This reaction also led to the
decomposition of T-LiFeO2 to the spinel structure LiFe5O8.
The major flaw of LiFeO2 in any of its polymorphs is its relatively low electrical
conductivity; therefore, carbon coating or nano-particles, as previously stated, is a route
commonly taken. Doping other transition metals into a structure is not only a route taken
for reducing the amount of some elements as seen in LiCo1/3Mn1/3Ni1/3O2, but can also be
performed to increase desired properties such as electrical conductivity. An example of
this is cobalt doping into the α-NaFeO2 layered structure LiFeO2 to synthesize
LiFe0.9Co0.1O2 [24]. LiFe0.9Co0.1O2 was synthesized simultaneously with an undoped
LiFeO2, both samples with the α-NaFeO2 layered structure. It was reported that LiFeO2
had an initial capacity of 205 mAh/g that eventually faded and stabilized at 140 mAh/g.
Similarly, the LiFe0.9Co0.1O2 had an initial capacity of 205 mAh/g, which faded and
stabilized at 190 mAh/g. This finding illustrates that doping cobalt into layered LiFeO2
produces a higher stabilized capacity than that of layered LiCoO2 or LiFeO2.
Further investigation into the T-LiFeO2 polymorph with a tunnel-like structure will be
discussed herein. The effects of chemical lithium intercalation and chemical lithium
deintercalation on the structure were analyzed along with Mössbauer spectroscopy.
Doping T-LiFeO2 with cobalt was also investigated to determine if the electrochemical
performance and structural stability could be enhanced and maintained respectively
without transformation into LiFe5O8 and decomposition of the electrolyte.

85 
 

 

5.2 Synthesis of ß-NaFeO2 and NaFe1-xCoxO2
ß-NaFeO2 was synthesized through the directed ceramic method with stoichiometric
amounts of Fe2O3 (99.5%, Alfa Aesar) and 5 % excess Na2CO3 (99%, Mallinckrodt
Chemicals), to compensate for sodium volatility. ß-NaFe1-xCoxO2 (0<x<0.15) is
synthesized similarly with the stoichiometric addition of Co3O4 (99.7%, Alfa Aesar) for
the appropriate dopant amount. The samples were prepared by mixing the reagents and
ball milling them until homogenous, at which point the samples are pressed into a 3/4
inch pellet under 6 tons of pressure. The resulting pellet was placed in a furnace
programmed as follows; heat up from room temperature to 850 °C over a time of 6 hours,
hold 850 °C for 12 hours, heat up to 1000 °C over 2 hours, hold at 1000 °C for 48 hours,
then returned to room temperature at a natural cooling rate. The sample must be held at
850 °C to prevent the Na2CO3 from melting (M.P. = 851 °C), the sample was then hand
ground and analyzed using PXD.

5.3 Ion exchange of NaFe1-xCoxO2 to LiFe1-xCoxO2 (0<x<0.15)
After ß-NaFe1-xCoxO2 was successfully synthesized, lithium for sodium ion exchange
was performed using the lithium nitrate/sodium nitrate eutectic melt. Approximately two
molar equivalents of lithium nitrate (LiNO3 99%, Alfa Aesar) were intimately mixed with
ß-NaFe1-xCoxO2 in a nitrogen glove box until the mixture was homogenous. The mixture
was placed in a furnace under oxygen flow and ramped to 225 °C for 48 hours. The
resulting melt was washed with methanol (anhydrous 99.9%) to remove the excess
LiNO3 and NaNO3 produced from the ion exchange. The resulting powders were

86 
 

 

analyzed using PXD and ICP to confirm phase purity and elemental composition
respectively.

5.4 Lithium intercalation and deintercalation of LiFeO2 via chemical reduction and
oxidation
Chemical lithium intercalation was performed using one molar equivalent of the as
prepared T-LiFeO2 with one, five, and ten times excess n-butyllithium (2.5M in hexanes,
Sigma-Aldrich). The mixtures were stirred for 96 hours at room temperature, filtered,
washed with n-pentane (anhydrous 99.8+%), then vacuum dried and stored under
nitrogen in a glove box.
Chemical lithium deintercalaiton was performed in a nitrogen glove bag, using one molar
equivalent of the as prepared T-LiFeO2 mixed with two molar equivalents of bromine
(99.5%, Jade Scientific) or iodine (99.8%, Jade Scientific) inside a round bottom flask
with 100 mL of acetonitrile (99.5%, Mallinckrodt Chemicals). The solutions were stirred
for 96 hours at room temperature. The compound was also mixed with 0.25, 0.5, 0.75, 1,
and 2 molar equivalents of NOBF4 (95%, Sigma-Aldrich) in acetonitrile and stirred for
48 hours. All samples were washed with acetonitrile, vacuum dried, and stored under
nitrogen in a glove box.

5.5 Results and Discussion
Synthesis of ß-NaFe1-xCoxO2 was straightforward using Na2CO3 and Fe2O3. The
synthesis was successful at 770 °C in air, but required increased sintering times on the

87 
 

 

cobalt doped samples. Increasing the temperature resulted in the formation of more
crystalline PXD along with pure cobalt samples upon the initial firing as seen in Figure
5.1. Attempts to dope 0.20 parts per formula unit of cobalt into ß-NaFeO2 resulted in the
formation of the NaCoO2 with α-NaFeO2 layered structure and an impurity peak in the
PXD. This established the maximum amount of cobalt doping to be between 0.15 and
0.20 parts per formula unit. No further doping reactions were performed past 0.15 parts
per formula unit of cobalt.

Figure 5.1 PXD of pure ß-NaFeO2 synthesized at 1000 °C in air, red ticks represent
expected reflections of pure phase.

88 
 

 

The resulting powders were analyzed using ICP to ensure the metal ratios were as
expected in the intial stoichiometry and that there was no loss of sodium due to the high
temperature sinterning. After ICP confirmation, Le Bail fit using GSAS was performed to
determine cell parameters of the cobalt doped ß-NaFeO2 compounds as seen in Figures
5.2-5.5 and Table 5.3.

Figure 5.2 Cell parameter a of ß-NaFeO2 as a function of cobalt doping.

89 
 

 

Figure 5.3 Cell parameter b of ß-NaFeO2 as a function of cobalt doping.

90 
 

 

Figure 5.4 Cell parameter c of ß-NaFeO2 as a function of cobalt doping.

91 
 

 

Figure 5.5 Cell volume of ß-NaFeO2 as a function of cobalt doping.

92 
 

 

Table 5.3 Cell parameters and volume of cobalt doped ß-NaFeO2 determined from Le Bail fit.

(x) ß-NaFe1-xCoxO2

Cell parameter a (Å)

Cell Parameter b (Å)

Cell Parameter c (Å)

Cell Volume (ų)

0.00

5.6739

7.1511

5.3897

218.68

0.05

5.6704

7.1398

5.3832

217.94

0.10

5.6665

7.1256

5.3772

217.11

0.15

5.6630

7.1120

5.3716

216.34

93 
 

 

The resulting Le Bail fits of ß-NaFe1-xCoxO2 established a linear relationship between the
cell parameters and the amount of cobalt doped into the ß-NaFeO2 structure. After it was
determined the ß-NaFe1-xCoxO2 samples were pure, lithium for sodium ion exchange was
performed. The melting points of LiNO3 and NaNO3 are 255 °C and 308 °C respectively,
therefore increasing to a temperature greater than 308 °C would minimize the time at
which the ion exchange would occur. Because the resulting phase is not the most
thermodynamically stable form of LiFeO2, determining the lowest temperature at which
the ion exchange will occur was established using TGA-DSC as seen in Figure 5.6.

Figure 5.6 Thermogravimetric analysis (black) and differential scanning calorimetry
(blue) of the lithium for sodium ion exchange. The endothermic peak begins at about 90
°C and exothermic peak at about 212 °C, (exothermic is positive µV/mg), are indicative
of melting and ionic diffusion respectively.

94 
 

 

The TGA-DSC indicated no change in mass between room temperature and 240 °C
(Figure 5.6), but the DSC did show an endothermic peak at about 190 °C followed by an
exothermic peak at about 212 °C. These peaks represent the eutectic melt (endothermic
peak at about 190 °C) of LiNO3/NaNO3 and the lithium for sodium ion exchange
occurring (exothermic peak at 212 °C). With the temperature of the ion exchange
established to be about 212 °C, all of the ion exchanges performed on ß-NaFe1-xCoxO2
were performed at 225 °C.
As previously mentioned, after the ion exchange, the resulting melt was washed
repeatedly with methanol to remove the excess LiNO3 and produced NaNO3 (Equation
5.1).
Equation 5.1
Figure 5.7 illustrates the formation of NaNO3 from the ion exchange. The sample was
briefly washed to remove the excess LiNO3, which readily dissolved and washed away,
while the NaNO3 required multiple washings to completely remove.

95 
 

 

Figure 5.7 PXD of the unwashed lithium for sodium ion exchange of ß-NaFeO2 and
LiNO3 to produce LiFeO2 (red tick marks) and NaNO3 (green tick marks).

After the produced NaNO3 and excess LiNO3 were completely removed, PXD and ICP
were performed to establish phase purity and elemental analysis respectively. It was
immediately apparent that the resulting compound T-LiFe0.85Co0.15O2 was impure by
PXD due to the appearance of peak reflections from LiCoO2 with the α-NaFeO2 layered
structure type. Le Bail fits were then performed to determine the cell parameters of the
cobalt doped T-LiFeO2 compound as seen in Figure 5.8-5.11 and Table 5.4.

96 
 

 

Figure 5.8 Cell parameter a of T-LiFeO2 as a function of cobalt doping.

97 
 

 

Figure 5.9 Cell parameter b of T-LiFeO2 as a function of cobalt doping.

98 
 

 

Figure 5.10 Cell parameter c of T-LiFeO2 as a function of cobalt doping.

99 
 

 

Figure 5.11 Cell volume of T-LiFeO2 as a function of cobalt doping.

100 
 

 

Table 5.4 Cell parameters and volume of cobalt doped T-LiFeO2 determined from Le Bail fit.

(x) T-Li1--xCoxO2

Cell parameter a (Å)

Cell Parameter b (Å)

Cell Parameter c (Å)

Cell Volume (ų)

0.00

5.4916

6.4078

5.0505

177.72

0.05

5.4833

6.3944

5.0424

176.80

0.10

5.4719

6.3791

5.0341

175.72

0.15

5.4713

6.3743

5.0332

175.54

101 
 

 

The resulting Le Bail fits of T-LiFe1-xCoxO2 established a linear relationship between the
cell parameters and the amount of cobalt doped into the T-LiFeO2 structure. This linear
trend was true through 0.10 parts per formula unit of cobalt, at which point the trend
subsided. This is due to the maximum cobalt concentration in the solid solution;
therefore, leading to an impurity, which as previously stated, was determined by PXD to
be LiCoO2 with the α-NaFeO2 layered structure type.

Figure 5.12 PXD and Rietveld refinement of LiFeO2 Rwp=1.49. Observed (black
crosses), calculated (solid red line), difference (solid blue line) profiles, and Bragg
reflections (tick marks) for the final.

102 
 

 

T-LiFeO2 was previously reported [20] through back-to-back ion exchanges using ten
equivalents of LiCl with ß-NaFeO2 at 160 °C. The resulting structure was determined
from synchrotron data. Ion exchange performed within, though at a higher temperature
than previously reported, only required one exchange.

Figure 5.13 Rietveld Refinement of PXD Li0.57FeO2 Rwp=3.90. Observed (black
crosses), calculated (solid red line), difference (solid blue line) profiles, and Bragg
reflections (tick marks) for the final Rietveld refinement.

103 
 

 

The resulting powder of T-LiFeO2 was Rietveld refined from laboratory PXD spectra
(Figure 5.12). The established cell parameters from the Rietveld refinement of T-LiFeO2
were determined to be a=5.4916Å, b=6.4078Å, and c=5.0505Å, which was in good
agreement with the previously reported data of a=5.5160 Å, b=6.4139 Å, and
c=5.0789 Å.
After T-LiFeO2 was deemed pure by PXD and ICP, chemical lithium intercalation and
deintercalation were performed to determine if the structural integrity of the tunnel
compound would be maintained through the insertion and removal of a lithium ion. Initial
attempts to remove lithium using bromine and iodine in pentane were performed with the
resulting powders analyzed by PXD and ICP. ICP indicated only 0.12 and 0.03 parts per
formula unit of lithium were removed using bromine and iodine respectively. These
results were not surprising because of the relatively low oxidation potentials of bromine
and iodine of 4.1 V and 3.5 V respectively. Due to the relatively high oxidation potential
of Fe3+/4+ with respect to a lithium anode, the oxidizing agent NOBF4, which has an
oxidative potential of 4.7 V with respect to a lithium anode [25], was used. The as
prepared sample of T-LiFeO2 was stoichiometrically added to one-half, one, two, five,
and ten times molar equivalents of NOBF4, dissolved in anhydrous acetonitrile, and
stirred for 48 hours at room temperature. The resulting powders were analyzed using
PXD and ICP to determine phase purity and elemental analysis. The maximum amount of
lithium that was removed from T-LiFeO2 was determined to be 0.58 parts per formula
unit, resulting in the compound Li0.42FeO2. Rietveld refinement of a sample that was
determined to have had 0.43 parts per formula unit removed was performed (Figure

104 
 

 

5.13). The established cell parameters from the Rietveld refinement of T-Li0.57FeO2 were
determined to be a=5.4721Å, b=6.4073Å, and c=5.0405Å.
Similarly, sodium deintercalation was performed on ß-NaFeO2 using NOBF4 to evaluate
this tunnel structure for use as a sodium ion battery. PXD, ICP, and TGA were used to
determine the structural integrity of ß-NaFeO2 and ß-Na1-xFeO2.
Chemical lithium intercalation was performed with concentrated n-butyllithium as
previously described. Surprisingly, the color of the solution of n-butyllithium mixed with
T-LiFeO2 did not instantly change as seen with LiFeTiO4 [26,27], another iron 3+
compound that instantly changed from a salmon color to black. Nevertheless, the nbutyllithium and T-LiFeO2 solutions were stirred at room temperature for four days.
Daily observations of the mixtures were taken. The mixtures slowly changed from the
initial salmon color to black after about 60 hours. The solutions were allowed to continue
to stir for another 36 hours to ensure completion.
The resulting samples were analyzed using PXD and ICP to determine phase purity and
elemental analysis. There were no observable peaks of impurities and it was determined
by ICP that each sample, the one, five, and ten molar equivalents n-butyllithium,
contained the same stoichiometric ratio of lithium to iron of 1.57:1. Synchrotron data was
Rietveld refined (Figure 5.14) with the established cell parameters determined to be
a=5.4991Å, b=6.4090Å, and c=5.0540Å. The slight decrease in cell parameters when
lithium was removed and increase in cell parameters when lithium was inserted correlate
well with the expected structure changes as a function of lithium content.

105 
 

 

Figure 5.14 Synchrotron diffraction of Li1.57FeO2 Rwp=5.73 and χ2=1.969. Observed
(black crosses), calculated (solid red line), difference (solid blue line) profiles, and Bragg
reflections (tick marks) for the final Rietveld refinement.

After the ß-NaFe1-xCoxO2 and T-LixFeO2 compounds were determined to be monophasic
further analysis using TGA was performed.

106 
 

 

Figure 5.15 Thermogravimetric analysis of T-LiFeO2 in oxygen up to a final temperature
of 1000 °C.

Thermogravimetric analysis of T-LiFeO2 was performed up to a final temperature of
1000 °C in oxygen. The TGA exhibited a steady mass loss until about 600 °C, where a
drop of about 0.75% was observed (Figure 5.15). The mass then continued to decrease
slowly until it began to plateau at about 950 °C and the final mass loss was 2 % of the
initial mass. The sudden mass loss at 600 °C is due to lithium volatility and oxygen loss

107 
 

 

during the reorganization of the elements because of the transformation of T-LiFeO2 to
the thermodynamically stable α-LiFeO2.

Figure 5.16 PXD of the resulting powder from the oxygen TGA of T-LiFeO2 red ticks,
are indicative α-LiFeO2 with the rock-salt structure.

The resulting PXD from the oxygen TGA to 1000 °C, (Figure 5.16), illustrates the
product α-LiFeO2 transformation from the kinetically stable T-LiFeO2.

108 
 

 

Figure 5.17 Thermogravimetric analysis of Li0.42FeO2 in oxygen up to a final
temperature of 725 °C.

The thermogravimetric analysis of Li0.42FeO2 in oxgyen resulted in stability up to about
75 °C (Figure 5.17), at which point the compound began to decompose until about 350
°C where the mass loss essentially plateaued after 400 °C. The powder was than analyzed
by PXD (Figure 5.18), and determined to have transformed into Fe0.873(Li0.365Fe1.2915)O4.
Fe0.873(Li0.365Fe1.2915)O4 is a spinel structure compound similar to LiMn2O4. These results
illustrate nicely that the lithium deficient compound that resulted from the oxygen TGA

109 
 

 

is different from the non-lithium deficient starting phase T-LiFeO2, (Figures 5.16 and
5.18), further confirming the success of the lithium removal using NOBF4.

Figure 5.18 PXD of the resulting powder from of Li0.42FeO2 after TGA in oxygen, red
ticks are indicative of the compound Fe0.873(Li0.365Fe1.2915)O4.

Nitrogen TGA of T-LiFeO2 up to 1000 °C (Figure 5.19), showed results similar to the
oxygen TGA with a 2 % mass loss. The resulting PXD (Figure 5.20) was unlike the PXD
that resulted from the oxygen TGA experiment. The resulting PXD from the nitrogen
TGA was indicative of a structure that is likely to be lithium doped Fe3O4.

110 
 

 

Figure 5.19 Thermogravimetric analysis of T-LiFeO2 in nitrogen up to a final
temperature of 1000 °C with a final mass loss of about 2 %.

111 
 

 

Figure 5.20 PXD of the resulting powder from the nitrogen TGA of T-LiFeO2, red ticks
are indicative Fe3O4, but is likely to be lithium doped Fe3O4.

Evaluating ß-NaFeO2 for the purpose of determining whether the compound might
decompose at elevated temperatures during cycling was then performed by ICP and PXD.
ß-NaFeO2 was synthesized at 1000 °C as previously mentioned, therefore its
thermodynamic and structural stability are known to be stable. The issue at hand was
whether or not ß-Na1-xFeO2 was also stable at elevated temperatures. Electrochemical
cycling at elevated temperatures generally means 50 °C and 100 °C, therefore analysis of

112 
 

 

the resulting compounds after heating were evaluated. Figure 5.21 is the
thermogravimetric analysis of ß-NaFeO2 under nitrogen flow up to 250 °C illustrating the
relatively stable mass.

Figure 5.21 Thermogravimetric analysis of ß-NaFeO2 in nitrogen to a final temperature
of 250 °C for analysis to determine structural stability for high temperature
electrochemical cycling.

113 
 

 

Figure 5.22 PXD of the resulting powder from the nitrogen TGA of ß-NaFeO2, red ticks
are indicative ß-NaFeO2.

The PXD resulting from the nitrogen TGA to 250 °C, (Figure 5.22), shows conservation
of ß-NaFeO2. This indicates the initial structure will not decompose at elevated
temperatures, which was initially hypothesized based on the synthetic temperatures
required for pure phase.
Thermogravimetric analysis was similarly performed on ß-NaFeO2 under oxygen flow
(Figure 5.23), with the same expected results as seen in the nitrogen TGA. Similar to the

114 
 

 

nitrogen TGA, there was a slight mass loss in the initial 75 °C. This can be attributed to
water absorption, which is common for lithium and sodium containing compounds
because they are very hygroscopic. The powder sample, when loaded into the crucible for
thermogravimetric analysis, was exposed to ambient conditions briefly, therefore
providing the conditions for water absorption.

Figure 5.23 Thermogravimetric analysis of ß-NaFeO2 in oxygen to a final temperature of
350 °C for analysis to determine structural stability for high temperature electrochemical
cycling.

115 
 

 

Figure 5.24 PXD of the resulting powder from the oxygen TGA of ß-NaFeO2, red ticks
are indicative of ß-NaFeO2.

As previously stated, the expected results for the thermogravimetric analysis up to 350 °C
in oxygen should maintain the pure compound ß-NaFeO2. This was seen in Figure 5.24.
If the temperature were elevated, it would be expected that some sodium would be lost
from the topmost portion of the powder sample due to sodium's volatility, resulting in a
sodium deficient phase.

116 
 

 

Figure 5.25 Thermogravimetric analysis of the sodium deficient ß-Na0.79FeO2 in nitrogen
to a final temperature of 975 °C.

As previously stated, the decomposition of ß-NaFeO2 was not expected at elevated
temperatures; rather the sodium deficient phase would be more likely to decompose.
Therefore, thermogravimetric analysis of the sodium deficient phase, ß-Na0.79FeO2,
which was synthesized by sodium deintercalation from ß-NaFeO2 using NOBF4, was
performed. This compound was more stable to a slightly higher temperature under
oxygen flow, (Figure 5.25), than was the lithium deficient T-LiFeO2. ß-Na0.79FeO2 did

117 
 

 

not show decomposition until almost 100 °C, which is the upper limit of elevated
temperature cycling.

Figure 5.26 PXD of the resulting powder from the nitrogen TGA of ß-Na0.79FeO2, ßNaFeO2 (red) and Na3Fe5O9 (green) were the resulting phases.

The PXD of ß-Na0.79FeO2, run in the TGA to 1000 °C, under nitrogen flow, resulted in
the formation of Na3Fe5O9 and ß-NaFeO2, (Figure 5.26). This result is not surprising due
to the relative stability of ß-NaFeO2 at high temperatures, because of the instability of the

118 
 

 

sodium deficient ß-NaFeO2. The formation of ß-NaFeO2 resulted in even less available
sodium to form another phase which in turn resulted in the formation of Na3Fe5O9
(NaFe0.6O3), and the loss of some unreacted sodium during the formation at these
elevated temperatures.

Figure 5.27 Thermogravimetric analysis of the sodium deficient ß-Na0.79FeO2 in oxygen
to a final temperature of 725 °C.

Similar to the results of the nitrogen TGA, ß-Na0.79FeO2 run under oxygen also resulted
in the decomposition at about 100 °C (Figure 5.27).

119 
 

 

Figure 5.28 PXD of the resulting powder from the oxygen TGA of ß-Na0.79FeO2, ßNaFeO2 (red) and Fe2O3 (green) were the resulting phases.

The PXD of ß-Na0.79FeO2 run in the TGA to 700 °C under oxygen flow resulted in the
formation of ß-NaFeO2 and Fe2O3 as seen in Figure 5.28. This result, unlike that from ßNa0.79FeO2, which was run in the TGA under nitrogen, contained Fe2O3. The Fe2O3
similar to the Fe3O4 previously discussed is most likely doped with some sodium, but in
order to determine this for sure, further analysis would have to be performed and would
provide no beneficial insight into the results.

120 
 

 

Figure 5.29 Thermogravimetric analysis of ß-NaFe0.9Co0.1O2 in nitrogen to a final
temperature of 1000 °C.

Thermogravimetric analysis of the cobalt-doped ß-NaFeO2 was run in nitrogen and
oxygen to a temperature of 1000 °C. As previously mentioned, the results were expected
to be sodium deficient ß-NaFeO2 materials. Figure 5.29 illustrates the TGA of ßNaFe0.9Co0.1O2 under nitrogen flow to 1000 °C with no substantial loss in mass.

121 
 

 

Figure 5.30 PXD of the resulting powder from the nitrogen TGA of ß-NaFe0.9Co0.1O2, ßNaFeO2 (red) and ß-Na1-xFeO2 (green) were the resulting phases.

The resulting PXD of ß-NaFe0.9Co0.1O2 run in the TGA under nitrogen flow (Figure 5.30)
indicated two phases. The main phase was a compound with the same general peak
reflections as ß-NaFeO2. The second phase, which was much less evident, presented
peaks close in 2θ values as seen in the ß-NaFeO2 structures. Therefore, it can be said that
the second phase is the initially hypothesized sodium deficient phase. Due to the increase

122 
 

 

in relative surface area of the powder run in the TGA, sodium and lithium volatility is
/much more probable.

Figure 5.31 Thermogravimetric analysis of the ß-NaFe0.9Co0.1O2 in oxygen to a final
temperature of 1000 °C.

Thermogravimetric analysis was also performed on ß-NaFe0.9Co0.1O2 (Figure 5.31),
under oxygen. The resulting mass loss was relatively low, but again, this is expected due
to the stability of the initial phase and any mass loss is due to sodium loss.

123 
 

 

Figure 5.32 PXD of the resulting powder from the nitrogen TGA of ß-NaFe0.9Co0.1O2, ßNaFeO2 (red) and ß-Na1-xFeO2 (green) were the resulting phases.

Similar to the ß-NaFe0.9Co0.1O2 sample run under nitrogen flow (Figure 5.32), the
resulting PXD from ß-NaFe0.9Co0.1O2 run in the TGA under oxygen presented a mixture
of ß-NaFeO2 and a sodium deficient phase.
Now that the stability of the compounds has been established, electrochemical cycling of
T-LiFeO2 will be discussed. This was done by attempting to probe the Fe3+/4+ redox
couple by cycling the materials at voltages above 4 V. Figure 5.33 illustrates the initial

124 
 

 

cycling of undoped T-LiFeO2. As it can be seen, there is no distinct deintercalation peak
that would typically be seen in the oxidation of a cathode material.

Figure 5.33 Cyclic voltammogram of T-LiFeO2 between 4.0 and 4.6 V probing the
Fe3+/4+ redox potential showing the reaction of T-LiFeO2 with the electrolyte.

This material was previously reported to cycle at about 4.5 V with a capacity of 120
mAh/g, but during this cycling the compound decomposed into LiFe5O8 due to a reaction
that occurred between the cathode material and the electrolyte at increased voltages. The
initial cycling of the as prepared T-LiFeO2 was also integrated to determine what the
capacity was per cycle as seen in Figure 5.34. The initial cycle had a deintercalation

125 
 

 

capacity of 40 mAh/g but quickly faded to 15 mAh/g then 10 mAh/g, and upon
subsequent cycles after that, the capacity continued to fade.

Figure 5.34 Integrated cyclic voltammogram of T-LiFeO2 illustrating what would
normally be the deintercalation of lithium, but this is the reaction of T-LiFeO2 with the
electrolyte. This is also seen by the absence of lithium intercalation.

As seen in Figure 5.34, there was no lithium intercalation into the structure, therefore
resulting in an intercalation capacity of zero. This can be seen by the lack of an

126 
 

 

intercalation curve in Figure 5.34. This response was determined to be caused by the poor
electrical conductivity of the as prepared LiFeO2.

Figure 5.35 High voltage cycling of LiFe0.9Co0.1O2 between 3.5 and 4.6 V probing the
Fe3+/4+ redox couple, the lithium deintercalation peak is present at about 4.4 V and where
a distinct intercalation peak does not exist due to the compound’s reaction with the
electrolyte.

The LiFe1-xCoxO2 was hypothesized to reduce the challenge presented by the undoped TLiFeO2 by increasing the electrical conductivity, which in turn should provide successful

127 
 

 

cycling. As seen in Figure 5.35, a distinct oxidation curve was established at about 4.4 V.
This did indicate that the doping was contributing to the deintercalation of lithium from
LiFe0.9Co0.1O2. However, as seen after the oxidation peak in Figure 5.35, the oxidation
curve continued increasing, this again is due to the cathode material, in this case,
LiFe0.9Co0.1O2, reacting with the electrolyte, therefore decomposing.

Figure 5.36 Low voltage cyclic voltammogram of LiFe0.9Co0.1O2 probing the Fe2+/3+
redox couple between voltages of 1.5 and 3.0 V, with lithium intercalation occurring at
about 1.8 V and deintercalation occurring at about 2.0 V.

128 
 

 

It was established that cobalt doping is essential in the successful cycling of the tunnel
structure, but the reaction with the electrolyte was stopping progress in establishing
successful Fe3+/4+ cycling materials. Therefore, focus was shifted to determine if this
material could accept an additional lithium ion, as it was previously determined
chemically that it could receive about 0.6 lithium ions (Figure 5.14). Electrochemical
characterization probing the Fe2+/3+ redox couple was then performed by initially
intercalating lithium into the material as seen in Figure 5.36. The cyclic voltammogram
started at 3.0 V and ended at 1.5 V with a distinct reduction peak at about 1.8 V.
Successful cycling probing this redox couple was further confirmed by the distinct
oxidation peak at 1.95 V. Though it is obvious there is a dramatic capacity fade, the
ability to intercalate a lithium ion was further established electrochemically.
The as prepared LiFeO2 Mössbauer spectrum, as seen in Figure 5.38, was fit with one site
that consisted of an isomer shift of 0.141 mm/sec, (Table 5.6) which is expected in an
iron in 3+ oxidation state in the tetrahedral coordination.

129 
 

 

Figure 5.37 Mössbauer spectrum of Li0.42FeO2 after chemical lithium intercalation fit with 3 sites the total fit (black) and the
three sites correlating to isomer shifts are 0.129 (red), -0.09 (blue), and 0.53 (green).

130 
 

 

Table 5.5 Mössbauer parameters of the chemically deintercalated Li0.42FeO2 fit with
three sites.
Li0.42FeO2

Weight of
Sublattice

Weight %

Isomer Shift

Site A
Site B
Site C

0.541
0.752
0.827

25.5
35.5
39.0

0.129
-0.09
0.53

Magnetic
Hyperfine
Field
50.96
3.63
0.64

Quadrupole
splitting
-0.37
1.07
1.17

Mössbauer spectroscopy was performed, at room temperature, on the as prepared LiFeO2,
Li1.57FeO2, and Li0.42FeO2 to determine if the chemical oxidation and reduction could
further illustrate the different oxidation states of the three compounds. Figure 5.37 and
Table 5.5 depict the results of Li0.42FeO2 with three sites. The spectrum was fit and
isomer shifts of the three sites were fit and established to be 0.129, -0.09, 0.53 mm/sec.
Of these three values, 0.129 and 0.53 are indicative of iron present with a 3+ oxidation
state, while the -0.09 indicates a 4+ oxidation state. The 0.53 mm/sec value is unusually
high for a tetrahedrally coordinated iron, which is generally expected to be between 0.1
and 0.3 mm/sec. This could indicate that even though the overall fit matches relatively
well with the observed data, there may be other factors that cannot be seen in this data
that was collected at room temperature.

131 
 

 

Figure 5.38 Mössbauer spectrum of the as prepared LFeO2 with one site.

132 
 

 

Table 5.6 Mössbauer parameters of the as prepared LiFeO2 fit with one site.
LiFeO2

Weight of
Sublattice

Weight %

Isomer Shift

Site A

1

100

0.141

Magnetic
Hyperfine
Field
46.46

Quadrupole
splitting
-0.319

The as prepared LiFeO2 Mössbauer spectrum as seen in Figure 5.38 was fit with one site
that consisted of an isomer shift of 0.141 mm/sec, (Table 5.6) which is expected in an
iron 3+ oxidation state in the tetrahedral coordination. T-LiFeO2 is the first seen
polymorph with magnetic ordering observed at room temperature.
The chemically intercalated Li1.57FeO2, (Figure 5.39 and Table 5.7), was also fit with
three sites with isomer shifts of 0.161, 0.438, and 0.494 mm/sec. The values of 0.438 and
0.494 mm/sec are within the expected range of a tetrahedrally coordinated iron with a 2+
oxidation state, which is what was expected from the low voltage electrochemical cycling
that illustrated chemical intercalation, (Figure 5.36).

133 
 

 

Figure 5.39 Mössbauer spectrum of Li1.57FeO2 after chemical lithium intercalation fit with 3 sites, the total fit (black) and the
three sites correlating to isomer shifts are 0.161 (green), 0.438 (red), and 0.494 (blue).

134 
 

 

Table 5.7 Mössbauer parameters of the chemically intercalated Li1.57FeO2 fit with three
sites.
Li1.57FeO2

Weight of
Sublattice

Weight %

Isomer Shift

Site A
Site B
Site C

1.04
0.324
0.105

70.8
22.1
7.1

0.161
0.438
0.494

Magnetic
Hyperfine
Field
47.68
0
0

Quadrupole
splitting
-0.293
-1.038
-0.291

As it can be seen with the somewhat similar isomer shifts values between the Li0.42FeO2
(0.53 mm/sec) and LiFe1.57FeO2 (0.438 and 0.494), it is apparent that room temperature
measurements are not enough to strongly confirm Fe4+ due to the unexpectedly high
value of the other Fe3+ containing site. What can be said though, is that it is highly
unlikely that this compound would contain both iron 2+ and iron 4+ sites. Mössbauer
measurements at liquid helium temperature would exhibit magnetic ordering within each
of the sites within the structure, therefore providing separation of the peaks and more
accurate fitting.

135 
 

 

5.6 Conclusion
The T-LiFeO2 polymorph was successfully doped with up to the 0.10 parts per formula
unit of cobalt and was prepared from ion exchange with the parent doped compound, ßNaFe1-xCoxO2. The cobalt doped T-LiFeO2 did show improved cycling over the undoped
T-LiFeO2 with a distinct reduction peak observed in the Fe3+/4+ redox couple. This was
not observed in the undoped T-LiFeO2. Similar to undoped T-LiFeO2, the cobalt doped
reacted with the electrolyte, therefore decomposing. T-LiFeO2 was investigated to
determine whether chemical lithium intercalation and deintercalation would result in
structural decomposition, and was analyzed using PXD, ICP, electrochemistry, and
Mössbauer spectroscopy. The resulting Mössbauer spectra were matched, with the results
indicating Fe2+/3+, Fe3+, and Fe3+/4+ in their respective spectra. Due to inadequate peak
separation, low temperature Mössbauer spectroscopy should be employed to achieve the
appropriate peak separation to be able to make an accurate conclusion on the final
oxidation states of the T-LixFeO2 compounds.

5.7 Future Direction
The future direction of LiFeO2 can be separated into two-sections: continuing research of
the tetrahedral iron oxide and investigation into similar tunnel-like tetrahedral
compounds.
Focusing on the tetrahedral iron oxides, a collaboration to perform low temperature
Mössbauer spectroscopy on the T-Li0.42FeO2 and T-Li1.57FeO2 to determine the oxidation
states of the respective compounds should be conducted. After confirmation of the

136 
 

 

oxidation states, magnetic response of these divalent materials could also be very
interesting. During synthesis of the lithium deficient compound, it was observed that the
compound was attracted toward a magnet. This was not seen in the initial compound.
Cycling the cobalt-doped T-LiFeO2 against a solid electrolyte or ionic liquids should be
investigated. This could provide the required conditions for two electron cycling without
a reaction with the organic electrolyte, due to the formation of Fe4+, upon lithium
deintercalation.
ß-NaFeO2 and ß-Na1-xCoxFeO2 should be investigated as potential cathode materials and
solid electrolytes. Magnetic response and thermoelectric measurements into these
compounds should also be investigated.
Materials similar to the tunnel-like compound would involve the synthesis and ion
exchange of KCoO2 [28]. This compound contains tetrahedrally coordinated cobalt
similar to T-LiFeO2. The challenge with this compound will be determining conditions to
synthesize KCoO2 in bulk, because the only reported synthesis was a single crystal.
Derivatives from this compound could exhibit very interesting properties, i.e. NaCoO2
from ion exchange with KCoO2, (Figure 5.40), would be a polymorph to the known and
highly characterized NaCoO2 with the α-NaFeO2 layered structure type. Sodium deficient
layered NaxCoO2 have also been reported as thermoelectric materials and
superconductors. Therefore, it would be logical to attempt similar studies with this tunnel
structure.

137 
 

 

Figure 5.40 Proposed structure of tunnel NaCoO2 resulting from the ion exchange using
NaNO3 with the known compound KCoO2.

Successful ion exchange of lithium for potassium in KCoO2 would result in tunnel
LiCOO2 (Figure 5.41), which would be a polymorph to the most commonly used cathode
material in lithium ion batteries.

138 
 

 

Figure 5.41 Proposed structure of tunnel LiCoO2 resulting from the ion exchange using
LiNO3 with the known compound KCoO2.

Tunnel LiCoO2, if successfully synthesized, could result in one full electron transfer,
unlike the layered structure, which can only cycle 50 % of theoretical capacity. Multiple
electron transfer could be investigated with this material based on the information
achieved from the similar structure T-LiFeO2.

139 
 

 

REFERENCES

140 
 

 

REFERENCES
[1]

M. Okubo, E. Hosono, J. Kim, M. Enomoto, N. Kojima, T. Kudo, H. Zhou, I.
Honma, J. Am. Chem. Soc. 129 (2007) 7444.

[2]

Y. Sakurai, H. Arai, S. Okada, J. Yamaki, J. Power Sources 68 (1997) 711.

[3]

H. Awano, R.J. Brodd, J. Cho, N. Dimov, K. Gotoh, T. Horiba, K. Kikuya, M.
Kono, A. Kozawa, M. Maeda, H. Mori, A. Nagai, Y. Nishi, T. Nishida, H.
Noguchi, K. Numata, T. Ogino, Z. Ogumi, M. Okada, S. Okada, T. Osawa, M.
Otsuki, B. Park, P. Ramadass, K. Shimizu, M.E. Spahr, Y. Sun, K. Tagawa, M.
Ue, M. Ueda, H. Wang, J. Yamaki, H. Yamamoto, M. Yoshio, H. Yoshitake, Z.
(John) Zhang, Lithium-Ion Batteries, Springer US, New York, 2009.

[4]

H. Arai, S. Okada, Y. Sakurai, J. Yamaki, Solid State Ionics 95 (1997) 275.

[5]

H. Rim, H. Ryoung Park, M. Youp Song, Ceram. Int. 39 (2013) 7297.

[6]

M.Y. Song, H. Rim, H.R. Park, Ceram. Int. 39 (2013) 6937.

[7]

O. Toprakci, H. a. K. Toprakci, Y. Li, L. Ji, L. Xue, H. Lee, S. Zhang, X. Zhang, J.
Power Sources 241 (2013) 522.

[8]

D. Bhuvaneswari, N. Kalaiselvi, J. Solid State Electrochem. 17 (2012) 9.

[9]

Y. Yang, C. Xie, R. Ruffo, H. Peng, D.K. Kim, Y. Cui, Nano Lett. 9 (2009) 4109.

[10] M.M. Thackeray, P.J. Johnson, L.A. de Picciotto, P.G. Bruce, J.B. Goodenough,
Mater. Res. Bull. 19 (1984) 179.
[11]

A.K. Padhi, K.S. Nanjundaswamy, J.B. Goodenough, J. Electrochem. Soc. 144
(1997) 1188.

[12]

Y. Chiang, Nat. Publ. Gr. 1 (2002) 123.

[13] N. Yabuuchi, Y. Kawamoto, R. Hara, T. Ishigaki, A. Hoshikawa, M. Yonemura,
T. Kamiyama, S. Komaba, Inorg. Chem. 52 (2013) 9131.
[14]

H. Gabrisch, M. Kombolias, D. Mohanty, Solid State Ionics 181 (2010) 71.

[15]

a. Yamada, S.C. Chung, K. Hinokuma, J. Electrochem. Soc. 148 (2001) A224.

[16]

A.K. Padhi, K.S. Nanjundaswamy, C. Masquelier, S. Okada, J.B. Goodenough, J.
Electrochem. Soc. 144 (1997) 1609.

141 
 

 

[17]

M. Barré, M. Catti, J. Solid State Chem. 182 (2009) 2549.

[18]

M. Hirayama, H. Tomita, K. Kubota, H. Ido, R. Kanno, Mater. Res. Bull. 47
(2012) 79.

[19]

T. Matsumura, R. Kanno, Y. Inaba, Y. Kawamoto, M. Takano, J. Electrochem.
Soc. 149 (2002) A1509.

[20]

A.R. Armstrong, D.W. Tee, F. La Mantia, P. Novák, P.G. Bruce, J. Am. Chem.
Soc. 130 (2008) 3554.

[21]

J. Li, J. Li, J. Luo, L. Wang, X. He, Recent Advances in the LiFeO 2 -Based
Materials for Li-Ion Batteries, 2011.

[22]

R. Kanno, T. Shirane, Y. Kawamoto, Y. Takeda, M. Takano, M. Ohashi, Y.
Yamaguchi, J. Electrochem. Soc. 143 (1996) 2435.

[23]

M.M. Rahman, J.-Z. Wang, M.F. Hassan, S. Chou, Z. Chen, H.K. Liu, Energy
Environ. Sci. 4 (2011) 952.

[24] P. Suresh, a. K. Shukla, N. Munichandraiah, J. Power Sources 159 (2006) 1395.
[25]

A.R. Wizansky, P.E. Rauch, F.J. DiSalvo, J. Solid State Chem. 81 (1989) 203.

[26]

S.R. Bruno, C.K. Blakely, C.M. Tenbusch, V. V. Poltavets, Electrochem. Soc.
Trans. 45 (2013) 23.

[27] S.R. Bruno, C.K. Blakely, V. V. Poltavets, Electrochem. Soc. Trans. 29 (2012) 29.
[28]

C. Delmas, C. Fouassier, P. Hagenmuller, J. Solid State Chem. 13 (1975) 165.

142 
 

Chapter 6
Synthesis of the Novel Polymorph, LiFeTiO4, with Calcium Ferrite Structure
6.1 Introduction
Synthesis of novel materials to be used in rechargeable lithium ion batteries (LIBs) has
played a pivotal role in the advancement of modern day technologies because it offers
lighter weight and longer life cycle in portable electronics and hybrid electric vehicles. In
1997, Goodenough's group suggested the cathode material LiFePO4 [1] which in turn
resulted in an array of research into polyoxoanions as potential cathode materials.
Following the synthesis and characterization of LiFePO4, many new compounds were
achieved [2-5]. Further investigation into many of these compounds has resulted in new
synthetic techniques that alter particle morphology and/or size to better engineer the
compounds, therefore enhancing their electrochemical performance [6,7].
Polymorphs of known phases have also shown promise with different physical properties
due to their different structure types [2-4].

Investigation into sodium analogs of

successful LIBs has also shown promise for potential sodium ion batteries along with
mixed stoichiometry LIBs [7]. Polymorph investigation into sodium ion batteries has also
stemmed

interest.

Recently,

Na2MnP2O7,

a

Mn2+/3+

redox

potential,

was

electrochemically probed resulting in the highest capacity for Na-ion batteries known
[8]. Lithium analogs to well characterized oxides such as NaCoO2 have resulted in the αNaFeO2 structure compound LiCoO2, currently the most commonly used cathode
material for rechargeable LIBs. Due to an increased concern over toxicity and cost of the

143 
 

transition metal used, many researchers have focused on inexpensive and less toxic
transition metals such as Mn, Ni, and Fe [9–12].
The lithium analog to α-NaFeO2 was synthesized through ion exchange, then found to
undergo a phase transition by attempting to probe the Fe3+/Fe4+ redox potential [13,14].
Investigation into another lithium analog of a sodium iron oxide was performed through
the ion exchange of lithium for sodium into the ß-NaFeO2 polymorph resulting in the new
compound T-LiFeO2 [15]. It was reported that probing the Fe3+/Fe4+ redox potential of
this tetrahedrally coordinated iron oxide resulted in the transformation of the tunnel-like
structure into LiFe5O8 (Li0.2FeO1.6) through a reaction with the electrolyte leading to
oxygen loss.
Investigating the synthesis of new materials via ion exchange of lithium for sodium, with
known sodium metal oxides allows for the formation of compounds that would normally
not form due to more thermodynamically stable polymorphs that form in one-step or onepot synthetic reactions.

Previous investigations of LiFeTiO4 with different

concentrations of transition metals in the spinel structure have been characterized as an
intercalation anode to form compounds with the general formula Li1+xFe1-3xTi1+2xO4 and
cycling voltages below 2 V [16]. Li2FeTiO4 with the disordered rock salt structure was
synthesized using the carbon precursor approach and heat-treated under CO/CO2
atmosphere at 700 °C. The resulting carbon coated material exhibited reversible cycling
between about 2.3 and 2.7 V, intercalation and deintercalation respectively, with a

144 
 

capacity of about 83% of its theoretical capacity, but only when particle size was between
10 nm and 25 nm [17].
Synthesis of the polymorph LiFeTiO4 with the calcium ferrite structure was performed
[18]. It was initially synthesized to probe the Fe2+/Fe3+ and the Fe3+/Fe4+ redox potentials
due to its tunnel-like structural similarity to LiFePO4. The initial hypothesis was that the
rigid structure would provide stability to intercalate and deintercalate lithium, thus
leading to the first 2-electron transfer iron based cathode material. Previously, the
removal of lithium was observed through chemical deintercalation of the calcium ferrite
structure LiFeTiO4 in the presence of Br2 and I2 [19], therefore providing preliminary
evidence of Fe4+. Rietveld refinement of diffraction data of LiFeTiO4 and Li2FeTiO4 is
presented. Electrochemical characterization of the as prepared LiFeTiO4 showed
reversible cycling between 2 and 2.3 V against lithium with an average particle size of 10
µm determined by SEM. Mössbauer spectroscopy was performed on the as prepared and
lithium deintercalated Li0.33FeTiO4 and determined to only contain Fe3+.
6.2 Synthesis of NaFeTiO4
All reagents used were of ACS grade purity or higher and were used without further
purification. Fe2O3 (99.5%), TiO2 (rutile, 99.5%), and methanol, (anhydrous 99.9%) were
obtained from Alfa Aesar. Na2CO3 (99%) was purchased from Mallinckrodt Chemicals.
Initial attempts to synthesize NaFeTiO4 were performed as described in [20]

with

stoichiometric amounts of Na2CO3, Fe2O3, and TiO2. The starting materials were mixed
in an agate mortar then fired in air at 1200 °C repeatedly with intermittent grindings.

145 
 

Following the procedure as described, the initial batch of NaFeTiO4 was pelleted, placed
into an alumina boat, inserted into a furnace with a heating rate of 100 °C/hour, held at
1200 °C for 48 hours, then allowed to cool to room temperature naturally. The resulting
product had formed single crystals due to the melting of the reactants at such high
temperatures. Synthesis of NaFeTiO4 was then altered, ratios of Na2CO3, Fe2O3, and
TiO2, were intimately mixed stoichiometrically with 10% excess Na2CO3 using a mortar
and pestle. Excess sodium is required due to volatility of sodium at high temperatures.
Using the resulting powder, the mixture was pelleted and fired at 850 °C for 450 hours
with intermittent grindings until pure. The synthetic procedure was optimized with ratios
of Na2CO3, Fe2O3, and TiO2 that were intimately mixed stoichiometrically with 5%
excess Na2CO3 using a planetary ball mill. The powders were mixed for six cycles of five
minutes in steel ball mill jars at 500 RPM using four tungsten carbide balls per jar. The
mixture was then pressed in a 3/4" pellet at 5 tons, placed in an alumina boat, and fired at
850 °C for 12 hours in air. Further annealing at 975 °C was performed for 48 hours.

6.3 Ion Exchange of NaFeTiO4 to LiFeTiO4
LiNO3 (99%) and methanol (anhydrous 99.9%) were obtained from Alfa Aesar.
LiFeTiO4 was synthesized via an ion exchange of Li+ for Na+ in NaFeTiO4. Two times
excess LiNO3 was intimately mixed with one equivalent of NaFeTiO4 then heated at 325
°C for 48 hours under O2 flow. The reaction product was washed with anhydrous
methanol, dried under vacuum, and then stored under N2 in a glove box.

146 
 

6.4 Chemical intercalation and deintercalation
Bromine (99.5%) and iodine (99.8%) were obtained from Jade Scientific and acetonitrile
(99.5%) was purchased from Mallinckrodt Chemicals. Chemical lithium deintercalation
of LixFeTiO4 (0 ≤ x < 1) was performed using a nitrogen glove bag, with two times molar
excess bromine or iodine to LiFeTiO4, in a round bottom flask topped with 100 mL of
acetonitrile. The reaction mixtures were then stirred at room temperature for 48 hours.
Next the mixture was filtered, washed with acetonitrile, vacuum dried and stored under
N2 in a glove box. The resulting compounds were analyzed using PXD and ICP [21].
n-pentane (anhydrous 99.8+ %) was obtained from Alfa Aesar and n-butyllithium (2.5M
in hexanes) was purchased from Sigma-Aldrich. Lithium intercalation into LiFeTiO4 was
performed in a nitrogen glove box using an equivalent of the prepared LiFeTiO4 and one
equivalent of n-butyl lithium. The mixture was stirred at room temperature for 48 hours,
filtered, washed with pentane, vacuum dried, and stored under N2 in a glove box. The
resulting compounds were then analyzed using PXD and ICP [21].

147 
 

6.5 Results and Discussion

 

Figure 6.1 Single crystal diffraction of Anosovite, Ti3O5, formed during initial attempt to
synthesize NaFeTiO4 from literature preparation. Space group Cmcm, a=3.7097(2) Å,
b=9.7308(4) Å, c=9.9716(5) Å, and volume 357.96(3) ų from a molybdenum source
single crystal diffractometer.

The literature preparation of NaFeTiO4, as previously mentioned, was prepared through
the solid-state reaction of the mixed reagents Na2CO3, Fe2O3, and TiO2. This mixture was
pelleted then fired at 1200 °C for 48 hours in air. The sample was allowed to cool to
room temperature at the natural cooling rate of the turned off furnace. The pelleted
melted at temperatures elevated above 851 °C (M.P. of Na2CO3) therefore, acting as a
flux and forming single crystals (Figure 6.1). This was determined to be single crystals of

148 
 

Anosovite, Ti3O5 originally discovered in 1951 and more recently reinvestigated in 1998
[22].
Further attempts to synthesize NaFeTiO4 with altered furnace programming were then
carried out as previously described using the initial reagents Na2CO3, TiO2, and Fe2O3
stoichiometrically mixed with 10% excess Na2CO3, to compensate for sodium volatility
until homogenous. The resulting product was a mixture of NaFeTiO4 and
Na0.79Fe0.8Ti1.2O4 as seen in Figure 6.2.

Figure 6.2 The mixture containing NaFeTiO4 (green ticks) and Na0.79Fe0.8Ti1.2O4 (red
ticks).

149 
 

Further annealing at 975 °C of this sample with intermittent grindings and pelleting
eventually

led

to

the

mixture

of

NaFeTiO4

and

the

α-NaFeO2

structure

Na0.75Fe0.75Ti0.25O4 (Figure 6.3).

Figure 6.3 The mixture containing NaFeTiO4 (green ticks) and Na0.75Fe0.75Ti0.25O2 with
the α-NaFeO2 structure type (red ticks).

Continued intermittent grinding, pelleting, and firing at 975 °C for durations longer than
350 hours led to the formation of pure NaFeTiO4 as seen in Figure 6.4. It was determined
that the initial 10% excess Na2CO3 was too much, therefore, decreasing this excess

150 
 

amount to 5 %, reduced the synthetic preparation of NaFeTiO4 to 48 hours at 975 °C with
no requirement for intermittent grinding and pelleting.

Figure 6.4 The PXD pattern of NaFeTiO4 (green ticks) was determined to be a pure
phase sample.

Initial attempts to synthesize LiFeTiO4 through the ion exchange of Li+ for Na+ from
NaFeTiO4 at 250 °C under O2 flow were unsuccessful. Upon increasing the temperature
to 350 °C, successful ion exchange was observed after 48 hours. This was determined

151 
 

though ICP analysis giving a Li:Fe ratio of 1:1 with no sodium detection. PXD of the
exchange product at 350 °C was impure with peaks of the known LiFeTiO4 spinel
structure appearing. Further attempts at the ion exchange at 300 °C for 250 hours showed
successful exchange with less intense peaks of the spinel structure. Decreasing the
exchange time to 24 hours at 300 °C resulted in partial exchange, while increasing the
time to 72 hours resulted in the formation of the spinel structure. Holding the exchange
time to 48 hours at 300 °C allowed the ion exchange to occur with no sodium precursor
phase, or spinel structure in the PXD. Longer exchange times and higher temperatures
clearly led to the transformation of the calcium ferrite type LiFeTiO4 to the spinel
polymorph. The tunnel structured LiFeTiO4 is only kinetically stable and is unstable
toward the transformation to the thermodynamically preferred spinel polymorph. This
observation was further confirmed by TGA experiments in both nitrogen and oxygen and
their respective PXDs as seen in Figures 6.5-6.8.
Thermogravimetric analysis (TGA) of the calcium ferrite structured LiFeTiO4 in both
oxygen and nitrogen to a temperature of 1000 °C was performed. The resulting sample
was removed and analyzed by PXD, whereby it was determined that the LiFeTiO4
calcium ferrite structure transforms to the more thermodynamically stable spinel
structure, ICSD 050387, under both O2 and N2 atmospheres. The transformation of the
calcium ferrite structure to the spinel structure is thermodynamically driven. This is also
seen in increased annealing times and temperatures as earlier described with the
formation of the spinel phase peaks at 350 °C or exchange times of 250 hours at 300 °C.

152 
 

Figure 6.5 Thermogravimetric analysis of calcium ferrite polymorph LiFeTiO4 in
nitrogen to a final temperature of 1050 °C.

153 
 

Figure 6.6 PXD of LiFeTiO4 after TGA in nitrogen to 1050 °C. Red ticks illustrate peak
reflection locations of the spinel structure LiFeTiO4.

154 
 

Figure 6.7 Thermogravimetric analysis of calcium ferrite polymorph LiFeTiO4 in oxygen
to a final temperature of 1050 °C.

155 
 

Figure 6.8 PXD of LiFeTiO4 after TGA in oxygen to 1050 °C. Red ticks illustrate peak
reflection locations of the spinel structure LiFeTiO4.

Scanning electron microscopy was employed to determine the morphology and particle
size of the as prepared LiFeTiO4. The particles, as seen in Figures 6.9-6.11, were
observed to have a size in the micrometer regime as expected from the synthetic
procedure used. The morphology of the particles was generally rod-like. Smaller particles
were observed on the surface of the LiFeTiO4, but due to exposure to ambient conditions

156 
 

in the mounting process and the hygroscopic nature of the material, these smaller
particles can be attributed to carbonate formation or the slight hydrolysis of LiFeTiO4.

Figure 6.9 SEM images of as prepared LiFeTiO4 illustrating the particle size of the
compound in the micrometer regime, scale bar represents 5 µm.

157 
 

Figure 6.10 SEM images of as prepared LiFeTiO4 illustrating the particle size of the
compound in the micrometer regime, scale bar represents 5 µm.

158 
 

Figure 6.11 SEM images of as prepared LiFeTiO4 illustrating the particle size of the
compound in the micrometer regime, scale bar represents 500 nm.
Oxidation via chemical lithium deintercalation was attempted by the addition of bromine
and iodine as described [19,21]. ICP ratios resulted in surprisingly low values of lithium
to iron Li0.33FeTiO4 and Li0.55FeTiO4 for bromine and iodine respectively [19]. These
values were deemed low due to the relatively low oxidizing power of bromine and iodine
against iron in the 3+ oxidation state. Electrochemical characterization did indicate that
oxidation of the compound, the deintercalation of lithium from LiFeTiO4, during the
initial cycling below 4 V, corresponds to the chemical potential of iodine and bromine

159 
 

[23]. Further investigation using PXD, electrochemistry and Mössbauer spectroscopy,
determined that deintercalation resulted in the decomposition of the compound.
Reduction via chemical lithium intercalation was initially attempted by the addition of
concentrated solutions of one, five, and ten times molar excess of n-butyl lithium to one
equivalent of LiFeTiO4. This resulted in an instant color change for all three reactions,
from the salmon colored precursor LiFeTiO4 to a black powder. PXD of the five and ten
times excess products indicated the decomposition of LiFeTiO4 with no observable peaks
and only an amorphous background [19]. The resulting PXD from the addition of an
equimolar ratio of concentrated n-butyl lithium and LiFeTiO4 had observable peaks of
LiFeTiO4 that were just over the amorphous and unusually high background. They were
similar to those seen in the completely decomposed and amorphous five and ten times
excess n-butyl lithium reactions. ICP of this product indicated a metal ratio of
lithium:iron to be 3.5:1. These results can be explained by the large particle size of the
precursor LiFeTiO4. The high reactivity of the n-butyl lithium over reduced the outside of
the particles resulting in the decomposition and observed amorphous background while
the center of the large particles were unreacted. This explains the observed peaks with
decreased intensities of LiFeTiO4. Further attempts to intercalate lithium into the
structure were successful upon the slow (1-hour) drop wise addition of diluted (0.1
molar) n-butyl lithium into a round bottom flask of 100 ml pentane and LiFeTiO4 in a
nitrogen glove box. Due to the pyrophoric nature of the resulting compound, the PXD
measurements were taken on a Bruker D2 Phaser in a nitrogen filled glove box.

160 
 

LiFeTiO4 was Rietveld Refined using GSAS and determined to have an Rwp=4.42. The
crystal structure of LiFeTiO4 was refined using the NaFeTiO4 structure as an initial
model (Figure 6.12). The space group Pnma was maintained from the precursor and the
cell parameters were determined to be a=8.9206(5) Å, b=2.9595(3) Å, and c=10.7103(7)
Å. LiFeTiO4 has a tunnel structure built up by (Fe/Ti)O6 octahedra that form double
edge-sharing chains. The edge-sharing Fe2O10 dioctahedra also vertex share with other
Fe2O10 dioctahedra. This connectivity gives rise to the "double rutile" chain; the lithium
ions are located in a single tunnel surrounded by four chains. Crystallographic data for
the Rietveld refined LiFeTiO4 can be seen in Table 6.1.

161 
 

Figure 6.12 Observed (black crosses), calculated (red line), difference (blue) profiles, and Bragg reflections (tick marks) for
the final Rietveld refinement of LiFeTiO4 from PXD data, Rwp=4.42 and χ2=0.260.

162 
 

Table 6.1 Crystallographic data for the Rietveld refinement of LiFeTiO4 (space group
Pnma), a=8.9206(5) Å, b=2.9595(3) Å, c=10.7103(7) Å.

Atom

Site

x

y

z

Uiso(102Å2)

Occupancy

Li

4c

0.3409(5)

1/4

0.3498(5)

1.09(4)

1.

Fe1

4c

0.0774(3)

1/4

0.6344(5)

1.00(3)

0.5

Fe2

4c

0.0725(5)

1/4

0.1313(5)

0.99(6)

0.5

Ti1

4c

0.0774(3)

1/4

0.6344(5)

1.03(3)

0.5

Ti2

4c

0.0727(4)

1/4

0.1313(5)

0.99(5)

0.5

O1

4c

0.8945(6)

1/4

0.5201(1)

1.03(3)

1.

O2

4c

0.9996(3)

1/4

0.2875(3)

1.02(8)

1.

O3

4c

0.3079(2)

1/4

0.6299(8)

1.06(7)

1.

O4

4c

0.0676(1)

1/4

0.9434(2)

1.12(5)

1.

Following the synthesis of LiFeTiO4, chemical lithium intercalation was performed as
previously described. PXD and ICP were used to confirm single-phase synthesis and
proper metal stoichiometry. Li2FeTiO4 was Rietveld refined using GSAS and determined
to have an Rwp=5.50 (Figure 6.13). The crystal structure for Li2FeTiO4 was refined using
LiFeTiO4 as an initial model. The space group Pnma was maintained from the precursor
and the cell parameters were determined to be a=9.0525(3) Å, b=2.9566(2) Å, and
c=10.7575(5) Å (Table 6.2).

163 
 

Figure 6.13 Observed (black crosses), calculated (red line), difference (blue) profiles, and Bragg reflections (tick marks) for
the final Rietveld refinement of Li2FeTiO4 from PXD data, Rwp=5.50 and χ2=0.183.

164 
 

Table 6.2 Crystallographic data for the Rietveld refinement of Li2FeTiO4 (space group
Pnma), a=9.0525(3) Å, b=2.9566(2) Å, c=10.7575(5) Å.
Atom
Li
Li

Site
4c
4c
4c

x
0.3733(7)
0.1146(9)
0.0869(3)

y
1/4
1/4
1/4

z
0.3540(6)
0.3124(1)
0.6095(3)

Uiso(102Å2)
0.96(2)
1.15(2)
1.07(3)

Occupancy
1.
1.
0.5

Fe1
Fe2

4c

0.0684(3)

1/4

0.1161(3)

1.00(5)

0.5

Ti1

4c

0.0869(3)

1/4

0.6095(3)

1.09(2)

0.5

Ti2

4c

0.0684(3)

1/4

0.1161(3)

1.00(6)

0.5

O1

4c

0.8982(1)

1/4

0.5221(1)

0.98(5)

1.

O2

4c

0.9528(1)

1/4

0.2728(1)

1.13(7)

1.

O3

4c

0.3216(9)

1/4

0.6590(1)

1.00(4)

1.

O4

4c

0.0830(1)

1/4

0.9215(1)

1.11(7)

1.

Electrochemical characterization was performed using LiFeTiO4 as a cathode vs. a
lithium anode. Initial cycling between 3.0 and 4.0 V results indicated the electrochemical
intercalation was greater than the deintercalation on the same cycle. This trend continued
for 8 cycles with no major fade in the cycling potential, indicating the structure's ability
to harbor more lithium ions in the cavities as seen in Figure 6.20 [18]. Continued cycling
of LiFeTiO4 between these voltages eventually resulted in the decomposition of the
compound as seen in Figures 6.14, 6.15, and 6.16. This explains the surprising ICP
results observed from the bromine and iodine oxidation indicating that the LiFeTiO4
decomposed under the mild oxidizing conditions of bromine and iodine. The PXDs

165 
 

observed previously [19] can be explained due to the particle size with the outermost
parts decomposing and the center remaining unreacted.

Figure 6.14 Cyclic voltammogram of LiFeTiO4 between 2.8 and 3.7 V decomposing
after the initial cycles as also illustrated by the low current.

166 
 

Figure 6.15 Integrated cyclic voltammogram illustrating the voltage as a function of
capacity of LiFeTiO4 in the attempt to probe the Fe3+/4+ oxidation states resulting in the
decomposition of the tunnel structure.

167 
 

Figure 6.16 Capacity as a function of the cycle number showing the instability of the
compound with no indication of a cycling trend.

Cycling the material at voltages between 1.5 and 3.0 V at a rate of C/5 showed initial
intercalation at about 2.3 V (Figure 6.17), with a capacity of about 148 mAh/g (Figures
6.18 and 6.19). This correlates to the insertion of one lithium into the structure (Figure
6.20) confirming the initial hypothesis that the tunnels of this structure could harbor more
lithium. Continued cycling at this rate resulted in a capacity fade of 8% over ten cycles.
Increasing the rate to C/2.5 for 50 cycles had a similar fade trend with a decreased

168 
 

starting capacity of about 121 mAh/g and a final capacity of 99 mAh/g and a capacity
fade of 18% over the 50 cycles as seen in Figures 6.18 and 6.19. In the cycling rate of
C/2.5, the initial capacity fade is about 1% of each subsequent cycle and eventually
decreases to less than 0.5%.

Figure 6.17 Cyclic voltammogram of LiFeTiO4 between 1.4 and 3.0 V showing relative
stable cycling intercalation and deintercalation of the Fe2+/3+ oxidation states.

169 
 

Figure 6.18 Integrated cyclic voltammogram of the voltage as a function of capacity of
LiFeTiO4 probing the Fe2+/3+ oxidation state at the two cycling rates C/5 and C/2.5.

170 
 

Figure 6.19 Capacity as a function of cycle number of the two cycling rates of C/5 (red)
and C/2.5 (black), illustrating the cycling capacity and the capacity fade.

171 
 

Figure 6.20 Rietveld refined structure of LiFeTiO4 and of Li2FeTiO4 after
electrochemical/chemical intercalation of a lithium ion.

Mössbauer spectroscopy was performed on the as prepared LiFeTiO4 and lithium
deintercalated Li0.33FeTiO4. The as prepared sample could be fit with one site as seen in
Figure 6.21, indicating Fe3+ in an octahedral coordination as also determined by PXD.
The chemically deintercalated Li0.33FeTiO4 was fit with two sites, (Figure 6.22), both
indicating Fe3+(Table 6.3), further illustrating the decomposition upon oxidation, as also
seen in the cyclic voltammogram (Figure 6.14), of LiFeTiO4 between 2.8 V and 3.7 V.
The PXD of the chemically delithiated compound [19] did not show a crystalline Fe2O3,
but did show a dramatic increase in background. This can be attributed to an amorphous
Fe2O3 resulting from the decomposition through the oxidation reaction.

172 
 

Figure 6.21 Mössbauer spectroscopy of LiFeTiO4 fit with one Fe3+ site.

The Mössbauer spectroscopy of LiFeTiO4 shows one site with a quadrupole splitting of
0.509 mm/sec and isomer shift of 0.403 mm/sec, which is indicative of Fe3+ in octahedral
coordination. This correlates to the Rietveld refined structure with Fe/TiO6 octahedra.

173 
 

Figure 6.22 Mössbauer spectroscopy of lithium deintercalated Li0.33FeTiO4 with the total
of the fit two sites (red), LiFeTiO4 (green) and the amorphous iron oxide (blue).

Table 6.3 Mössbauer spectroscopy of LiFeTiO4 and lithium deintercalated Li0.33FeTiO4.
Compound
LiFeTiO4

Isomer Shift
(mm/sec)

Hyperfine
Splitting (T)

Quadrupole Splitting
(mm/sec)

Area
%

0.403

0

0.5094

100

0.431

0

0.4795

83.4

0.300

0

0.5581

16.6

Li0.33FeTiO4

174 
 

6.6 Conclusion
New polymorphs LiFeTiO4 and Li2FeTiO4 both with the calcium ferrite structure were
successfully synthesized, structures determined, and electrochemically characterized.
Known polymorphs of both LiFeTiO4 and Li2FeTiO4 have previously been reported with
the spinel crystal structure and disordered rock salt structure respectively. Mössbauer
spectroscopy of the chemically deintercalated LiFeTiO4 was determined to only have iron
with a 3+ oxidation state therefore containing amorphous iron oxide. Electrochemical
characterization at room temperature showed intercalation of one lithium ion and an
initial reversible capacity of about 148 mAh/g between 2 and 2.35 V with a capacity fade
of 8% over 10 cycles. These results show an increased capacity of 17% over the
previously reported Li2FeTiO4 rock-salt-type polymorph, which required nano
preparation and special carbon coating, not required for this polymorph. LiFeTiO4 was
determined to be only kinetically stable, as it transformed into the known
thermodynamically stable spinel polymorph at increased temperatures under both oxygen
and nitrogen.

6.7 Future direction
LiFeTiO4 was explored in-depth as a cathode material. The only further measurements
that may be of interest would be the ionic conductivity of the as prepared sample to
investigate the potential of LiFeTiO4 as a solid electrolyte or performing magnetic
measurements of the lithium-intercalated compound, Li1+xFeTiO4. My hypothesis would
be that the as prepared LiFeTiO4 would not function well as a solid electrolyte due to its

175 
 

instability upon deintercalation of lithium along with the relatively low percentage of
lithium within the structure. My hypothesis for the magnetic measurements of the lithium
intercalated Li1+xFeTiO4 is that it could present some interesting properties within the
tunnel structure. Different lithium insertion amounts would alter the overall oxidation
state of iron between 2+ and 3+, which generally occurs in spinel, or reverse spinel
structures containing iron in both the octahedral and tetrahedral positions.
Future direction of this project would mostly focus on the use of calcium ferrite structure
due to tunnels that would grant structural stability upon oxidation or reduction via lithium
deintercalation or lithium intercalation respectively. An example of this is the compound
NaV2O4, (Figure 6.23), which contains vanadium in the 3+ and 4+ oxidation states [24].

176 
 

Figure 6.23 The known structure NaV2O4 with the calcium ferrite structure as a potential
starting point for probing new cathode materials.

NaV2O4 with the calcium ferrite structure would be a potential starting point for lithium
for sodium ion exchange. Following successful exchange, structural characterization can
be carried out through chemical lithium intercalation and deintercalation. Most
importantly, cyclic voltammetry probing both sodium ion and lithium ion batteries as a
cathode material should be carried out. This structure is nearly identical to NaFeTiO4 and

177 
 

LiFeTiO4, with the benefit of vanadium, which is known for multiple stable oxidation
states when cycled as a cathode material. This would give a theoretical capacity of 155
mAh/g for LiV2O4 (Figure 6.24) and 298 mAh/g for Li2V2O4 (Figure 6.25).
Similar to LiFeTiO4 representing a polymorph of the spinel structure LiFeTiO4, upon
successful lithium for sodium ion exchange, LiV2O4 would also be a polymorph to a
known spinel structure [25]. Therefore, the calcium ferrite structure would most probably
be only kinetically stable as is the calcium ferrite structure LiFeTiO4. If the synthesis of
the calcium ferrite structure LiV2O4 were to be successful, further investigation into this
compound through transition metal doping would be the next appropriate step, as it has
been reported by replacing a vanadium with a nickel in the spinel structure cycling
voltages were observed at 4.8 V [26].

178 
 

Figure 6.24 Proposed structure of LiV2O4 resulting from the successful lithium for
sodium ion exchange of the known compound NaV2O4.

After lithium for sodium ion exchange, a series of experiments could be performed on the
new material including, as previously stated, chemical lithium intercalation as seen in
Figure 6.25, chemical lithium deintercalation as seen in Figure 6.26, and cyclic
voltammetry.

179 
 

Figure 6.25 Proposed structure of Li2V2O4 after successful chemical lithium ion
intercalation or electrochemical lithium intercalation during cyclic voltammetry.

The disadvantage to this material, though it has the potential of nearly 300 mAh/g, is the
lower cycling voltage which could range between 2 and 3 V [27,28], therefore decreasing
the energy density of the compound when cycled as a cathode.
The challenge with this material would be similar to that seen in the synthesis of α- and
ß-NaFe2O3. The transition metal, in this case vanadium, is divalent. This would require

180 
 

the careful synthesis, most likely oxygen pressure regulation, of the starting material
NaV2O4.

Figure 6.26 Proposed structure of V2O4 after successful chemical lithium ion
deintercalation or electrochemical lithium deintercalation during cyclic voltammetry.

181 
 

REFERENCES

182 
 

REFERENCES
[1]

A.K. Padhi, K.S. Nanjundaswamy, C. Masquelier, S. Okada, J.B. Goodenough, J.
Electrochem. Soc. 144 (1997) 1609.

[2]

M. Alvarez-Vega, J.M. Gallardo-Amores, F. García-Alvarado, U. Amador, Solid
State Sci. 8 (2006) 952.

[3]

A.K. Padhi, K.S. Nanjundaswamy, J.B. Goodenough, J. Electrochem. Soc. 144
(1997) 1188.

[4]

H.-K. Song, K.T. Lee, M.G. Kim, L.F. Nazar, J. Cho, Adv. Funct. Mater. 20
(2010) 3818.

[5]

Z. Bi, X. Zhang, W. He, D. Min, W. Zhang, RSC Adv. 3 (2013) 19744.

[6]

M. Minakshi, P. Singh, N. Sharma, M. Blackford, M. Ionescu, Ind. Eng. Chem.
Res. 50 (2011) 1899.

[7]

R. Tripathi, T.N. Ramesh, B.L. Ellis, L.F. Nazar, Angew. Chem. Int. Ed. Engl. 49
(2010) 8738.

[8]

P. Barpanda, T. Ye, M. Avdeev, S.-C. Chung, A. Yamada, J. Mater. Chem. A 1
(2013) 4194.

[9]

C. Nithya, V.S.S. Kumari, S. Gopukumar, Phys. Chem. Chem. Phys. 13 (2011)
6125.

[10] H. Kim, D.-H. Seo, H. Kim, I. Park, J. Hong, K.-Y. Park, K. Kang, Chem. Mater.
24 (2012) 720.
[11] C.V. Rao, A. Leela, M. Reddy, Y. Ishikawa, P.M. Ajayan, Appl. Mater. Interfaces
3 (2011) 2966.
[12] J.B. Goodenough, Y. Kim, Chem. Mater. 22 (2010) 587.
[13]

M. Hirayama, H. Tomita, K. Kubota, R. Kanno, J. Power Sources 196 (2011)
6809.

[14]

M. Hirayama, H. Tomita, K. Kubota, H. Ido, R. Kanno, Mater. Res. Bull. 47
(2012) 79.

[15] A.R. Armstrong, D.W. Tee, F. La Mantia, P. Novák, P.G. Bruce, J. Am. Chem.
Soc. 130 (2008) 3554.

183 
 

[16]

A.D. Robertson, H. Tukamoto, J.T.S. Irvine, J. Electrochem. Soc. 146 (1999)
3958.

[17]

M. Küzma, R. Dominko, A. Meden, D. Makovec, M. Bele, J. Jamnik, M.
Gaberšček, J. Power Sources 189 (2009) 81.

[18] S.R. Bruno, C.K. Blakely, V. V. Poltavets, Electrochem. Soc. Trans. 29 (2012) 29.
[19]

S.R. Bruno, C.K. Blakely, C.M. Tenbusch, V. V. Poltavets, Electrochem. Soc.
Trans. 45 (2013) 23.

[20] S. Zouari, L. Ranno, a. Cheikh-Rouhou, O. Isnard, P. Wolfers, P. Bordet, P.
Strobel, J. Alloys Compd. 452 (2008) 234.
[21] C.K. Blakely, S.R. Bruno, J.D. Davis, V. V. Poltavets, Electrochem. Soc. Trans.
45 (2013) 3.
[22]

M. Onoda, J. Solid State Chem. 73 (1998) 67.

[23]

A.R. Wizansky, P.E. Rauch, F.J. DiSalvo, J. Solid State Chem. 81 (1989) 203.

[24]

K. Yamaura, M. Arai, a. Sato, a. Karki, D. Young, R. Movshovich, S. Okamoto,
D. Mandrus, E. Takayama-Muromachi, Phys. Rev. Lett. 99 (2007) 196601.

[25]

L.A. de Picciotto, M.M. Thackeray, Mater. Res. Bull. 20 (1985) 1409.

[26]

G.T. Fey, Act. Passiv. Electron. Components 18 (1995) 11.

[27]

L.A. de Picciotto, M.M. Thackeray, Solid State Ionics 30 (1988) 1364.

[28]

L.A. de Picciotto, M.M. Thackeray, W.I.F. David, P.G. Bruce, J.B. Goodenough,
Mater. Res. Bull. 19 (1984) 1497.

184 
 

Chapter 7
Mössbauer Data Analysis
7.1 Introduction
COherent NUclear Scattering from Single crystals (CONUSS)[1] was developed in the
mid 1980's after the discovery of nuclear resonant scattering of synchrotron radiation. It
has been continuously updated and improved by one of its primary creators, Wolfgang
Sturhahn, who has an active relationship with Argonne National Laboratory Geophysics
beamline 3-ID-B, Ercan Alp, and the Geological and Planetary Science division at
California Institute of Technology. In November 2012, Argonne National Laboratory
held a workshop on data evaluation using CONUSS and PHOENIX; two programs
written by W. Struhahn for the analysis of elastic and inelastic nuclear resonant
scattering. The workshop was instructed by W. Sturhahn on the two programs' uses.
Because no manual exists for the programs, the only way to learn how to appropriately
use them requires mandatory attendance at this workshop. Therefore, the following will
include a generic manual for the operation of CONUSS for future members use of the
program to evaluate Mössbauer data.
7.2 Data Fitting
The received raw data with the ".cxp" extension must be converted into a two-column
space separated file with the extension ".exp", as seen in A1, and saved in a folder with
the program files for CONUSS analysis. The folder should have 8 files in it, the provided
data_graph.cxp, the converted .exp, and 6 files needed in every analysis: the NAME_in,
in_kctl, in_kfit, in_kmco, in_kmix, and the in_kref as seen in A2.

185 
 

Figure 7.1 Two-column space separated raw data saved with the ".exp" extension.

186 
 

Figure 7.2 Folder containing the required program files, raw data, and converted
experimental file.
Each file has a purpose in the operation of CONUSS; therefore, the lack of presence of
one can result in CONUSS not working. For every experiment data set, the 6 files
(NAME_in, in_kctl, in_kfit, in_kmco, in_kmix, and the in_kref) should just be copied
from the original CONUSS folder, named accordingly to the current experiment, and
each important parameter should be adjusted appropriately as described within. From
here, a brief explanation and major functions of each file will be provided.

187 
 

Figure 7.3 The in_kctl file with highlighted input files and parameters required for
CONUSS analysis.

188 
 

The in_kctl file is a program file that directs the CONUSS program which input files to
use in the analysis as seen in A3. The input files in_kref and in_kfit will be looked at
individually. The in_kmix is used for synchrotron data, which requires enriched iron.
This is beyond the scope of research in this laboratory. A3 also shows the iteration
control. This provides the user with the power to dictate the maximum number of steps
and the minimum iterations that will stop the program from continuing on unnecessarily
after the analysis has converged.
The in_kfit file, as seen in A4, has four parameters of interest (2), (10), (11), and (15)
which are highlighted. The experimental data file (2), instructs CONUSS which file to
use and what format the file is in. This is the two-column ".exp" file that was initially
prepared from the raw data provided. The background parameter (10) must be refined at
the beginning of every analysis along with the scaling factor (10) and thickness (15). The
background parameter (10) will also be maintained open to refinement throughout the
entire process.

189 
 

Figure 7.4 The in_kfit file with highlighted parameters for Mössbauer refinement.

190 
 

Figure 7.5 The in_kmco file with highlighted parameter files and input files.
The in_kmco file, as seen in A5, is the file responsible for the Monte Carlo-like method
that repeatedly takes random samplings in order to eventually approach the best possible

191 
 

fit. The two parameters of interest here are (4) and (5) which represent the number of
levels and the random samples per level. The operator can change each of these
parameters to optimize the analysis.

192 
 

Figure 7.6 The in_kref with highlighted material data input to be changed in each
experiment.

193 
 

Figure 7.7 Thickness unit / micron parameter highlighted also in in_kref that will need
to be initially refined to optimize data analysis.

194 
 

The NAME_in file will be the file that is most often used in the Mössbauer refinement.
This file, as seen in A8-A13, contains the main parameters in Mössbauer. A8 illustrates
the constant values that are to remain unchanged by the user. A9 highlights the number of
sites of the Mössbauer atom. This parameter is to be changed if the sample has multiple
Mössbauer atom sites with different coordination, oxidation states, or local environments.
A10 continues to show the NAME_in file with the @ &dist1, weight of the sublattice,
isomer shift, magnetic hyperfine field, and quadrupole splitting. The @ &dist1 is directly
linked to A11-A13 which allows for peak fluctuations, shifts, or widening depending
upon whether isomer shift, efg, or magnetic hyperfine splitting has been entered as seen
in A11-A13. It should also be noted that entering magnetic hyperfine splitting as the
target for a sample that does not have hyperfine splitting will not provide any benefit. The
same concept goes for entering efg for a sample that does not exhibit quadrupole
splitting. This is generally an issue when the sample contains multiple Mössbauer sites.
The weight of the sublattice is the percentage that this particular Mössbauer site had
within in the sample. Therefore, a sample with only one site would represent 100% of the
data, thus it would have a value of 1. A sample that has multiple sites would have
different representative values for each site. The isomer shift, magnetic hyperfine
splitting, and quadrupole splitting are the common parameters in Mössbauer
spectroscopy, which will be the primary focus throughout the refinement and fitting
process.

195 
 

Figure 7.8 The NAME_in file illustrating the known constant values of iron for
Mössbauer spectroscopy.

196 
 

Figure 7.9 The NAME_in file with highlighted number of sites of the Mössbauer atom.

197 
 

Figure 7.10 The NAME_in file with highlighted parameters that will be analyzed in
Mössbauer refinement.

198 
 

Figure 7.11 The NAME_in file illustrating electric field gradient as the target for
distribution input file.

199 
 

Figure 7.12 The NAME_in file illustrating isomer shift as the target for distribution
input file.

200 
 

Figure 7.13 The NAME_in file illustrating magnetic hyperfine field as the target for
distribution input file.

201 
 

Prior to running CONUSS, a description of the used command must first be undertaken.
The three commands that can be used with CONUSS are krmf, kctl, and kmco. Each one
of the commands can eventually get the operator to the final result with the major
difference of the three commands being the path taken by CONUSS or the user.
The command krmf is much more of a hands on approach by the user. It evaluates the
individual input parameter without any further optimization by CONUSS, therefore;
CONUSS will only evaluate the specific parameter input by the user.
The command kctl takes the open input parameter by the user, then optimizes it by
making minor changes to the parameter while evaluating the χ2 to determine if the change
was better than the initial value. If the value meets the iterations set by the user in the
in_kctl file (A3), the command will continue to run until it converges or meets the
maximum iteration steps, set in in_kctl.
The command kmco uses the Monte Carlo approach by taking the open input parameters
by the user, then optimizing the parameters through a randomized search with restrictions
placed on the command by the operator. This command is generally used as a starting
command for complex spectra and should be followed by the kctl command.
In order to run the CONUSS program to fit the acquired Mössbauer data, the background
and thickness in both the in_kfit and the thickness in in_kref must be refined. This is
performed by inserting a "%" symbol at the beginning of the line of the parameter of
interest. In this case, as seen in A14, the background and thickness both have a "%"
highlighted to illustrate what it should look like. Note there should be a space between

202 
 

the "%" and the next character. This is not as applicable in the other parameters, but is
essential when opening the @ &dist function, therefore, it is suggested to maintain the
space when opening any parameter. It should also be noted in the case of non-Gaussian
shaped peaks, the peak shapes can be changed in the in_kfit file as seen at the top of A14.

203 
 

Figure 7.14 The in_kfit illustrating the input of the % symbol, this opens the parameter
for spectrum fitting.

204 
 

Figure 7.15 Terminal screen shot illustrating input of the kctl command.

Figure 7.16 Terminal screen shot illustrating the reults of the kctl command after fitting
background and thickness.

205 
 

Figure 7.17 The main folder with out_kctl file highlighted.
After the specified parameters have been opened, enter the terminal, open the folder of
interest, and enter the kctl command as illustrated in A15. The terminal will run the
command and finish when the χ2 converges as seen in A16. This will result in an output
file that will be found in the main folder. This file will be labeled as out_kctl, as
illustrated in A17. The results from the kctl command will be in this file. The final data
that resulted in the best χ2 will be at the bottom of the file as seen in A18. By inputting
these values into the appropriate location, in the kref, NAME_in,, kfit files etc. will

206 
 

provide better guesses for CONUSS when fitting the data in the future commands.
Continued background fitting needs to be preformed next by refining the thickness in the
in_kref file as seen in A19. After CONUSS has run the kctl command, at any point the
operator can view the resulting spectra with the following terminal input, "xmgrace *.cxp
*.fit". This will open the program grace and plot the original data file (.cxp) along with
the resulting fit curve (.fit) from the kctl command.

Figure 7.18 The out_kctl file with the resulting values providing the best χ2 from the kctl
command.

207 
 

Figure 7.19 The in_kref with thickness unit / micron highlighted to be opened for
background refinement.

208 
 

Figure 7.20 The resulting spectrum from the background refinement with unchanged
parameters.
As seen in A20, the background has been established by refining the background and
thickness parameters in kfit and the thickness in kref. While the remaining spectra does
not match well, this is due to the inital parameters set in the NAME_in file which were
0.25, 50, and 1 for the isomer shift, magnetic hyperfine field, and quadrupole splitting
respectively.

209 
 

Now that the background has been established, the remaining focus can be switched into
the NAME_in file. From here on, the operator will use the krmf, kctl, and kmco
commands to optimize the Mössbauer fit.

Figure 7.21 The NAME_in file illustraing the isomer shift, magnetic hyperfine field, and
quadrupole splitting parameters open with the "%" for fitting.
In the NAME_in file the isomer shift, magnetic hyperfine field, and quadrupole splitting
parameters need to be opened as seen in A21. Depending upon the complexity of the
210 
 

spectrum, where in the instance there are multiple sites, opening and fitting one
parameter at a time is suggested. For this example, the sample only contains one site,
therefore opening all three parameters will not cause unreasonable results.

Figure 7.22 Terminal illustrating kctl command with no convergence after 40 iterations.
After the desired parameters have been opened and the NAME_in file has been saved, the
kctl command should be run in terminal as seen in A22. If the command does not result
in a convergence, the terminal will note this. In this case, this is due to the iteration steps
set in the in_kctl file only being set to 40.

211 
 

Figure 7.23 The out_kctl with the resulting values providing the best χ2 of the nonconverged data.
Even though the out_kctl parameters did not result in a convergence, the resulting values
can be found in the out_kctl and manually input into the NAME_in file as seen in A23
and A24. At this point, the kctl command can be re-run with the new parameters until a
convergence is achieved. This process should be repeated until all of the parameters no
longer change when the kctl command is run. Throughout this process continual

212 
 

monitoring of the fit pattern as previously described and seen in A25. This should also be
preformed to ensure that no unwanted peaks are appearing. This becomes more
problematic when refining multiple sites. It should also be noted when refining multiple
sites, to not open the same parameter on two different sites. In other words, do not open
the isomer shift for MB site #1 and MB site #2. This can result in CONUSS providing
results that fit the data but do not provide numbers that make scientific sense. Adding
Mössbauer sites will be discussed later.

213 
 

Figure 7.24 The NAME_in file illustrating the values from the kctl_out file input into
the NAME_in file for further fitting.

214 
 

Figure 7.25 The resulting spectrum of a non-converged kctl command with the isomer
shift, magnetic hyperfine field, and quadrupole splitting parameters open.
The kmco command, which uses a Monte Carlo-like method, can potentially be one of
the most useful starting commands when performing a Mössbauer fit. The starting point
again begins with refining the background as previously discussed. When the background
has been established, similar to the procedure of the kctl command, the user will be
performing most of the work in the NAME_in file, but first, the sampling levels and
number of samples per level must be adjusted to the operator's specifications. The more

215 
 

samples per level and the more levels that are specified the longer the operation will take,
therefore, taking the time to decide on reasonable criteria for this can save time in the
end. As seen in A26, the number of levels is 10 and samples per level is 50, this will
provide information that will allow the operator to narrow the parameters.

Figure 7.26 The in_kmco with highlighted number of levels and samples per level.
In the NAME_in file, as seen in A27, the isomer shift, magnetic hyperfine field, and
quadrupole splitting are open to refinement with the "%" symbol. The important aspect at
216 
 

this point is following the initial guess of the specific parameter another number is
inserted. For example, the isomer shift initial guess in A27 is 0.25 which is followed by a
space then the number 0.5. This is directing CONUSS, under the kmco command, to
select a random number within an equal range totaling to 0.5 of the guess. This number,
0.5, covers the entire region around 0.25, therefore the sample region is from 0.0-0.75.
Another example to clarify, is with the quadrupole splitting, the initial guess is 0 and the
range is 3, therefore CONUSS, under the kmco command, will randomly select numbers
between -1.5 and +1.5.

217 
 

Figure 7.27 The kmco set up in the NAME_in file.
After the initial guesses and ranges have been appropriately established, the kmco
command can be input into the terminal for execution. At this point, the terminal will
randomly select points around the initial guess within the region established by the user
and calculate the respective χ2 for these values. If the point provides a lower χ2 value over
the previously established, the terminal will report it. When the random samples per level
is reached, (in this case it is fifty), the next sample level will be entered. This level's

218 
 

starting point will be the previous level best sampling, at which point, again random
points are selected similar to the first level. This process continues until the number of
levels and random samples per level previously input by the user has finished. This is
unlike the kctl command which stops upon convergence. If it were to happen that the best
point was selected immediately by the kmco command, the process would continue until
finished. Upon completion of the kmco command, similar to the kctl command, an output
file will be produced as seen in A29. The best sampling as a result of the kmco command
are toward the bottom of this file as seen in A30. These values should be inserted into the
NAME_in respective parameters as seen in A31. At this point kmco can be ran again
until or if the operator is satisfied with the parameters provided, the kctl command should
be performed as previously described until the fit is ideal.

219 
 

Figure 7.28 The terminal illustration of output from the kmco command.

220 
 

Figure 7.29 The highlighted output file as result from the kmco command.

221 
 

Figure 7.30 The best sampling as a result of the kmco command highlighted in the
out_kmco file.

222 
 

Figure 7.31 The best sampling values input into the NAME_in file for further
refinement.
The final important aspect of Mössbauer fitting using CONUSS is refining data with
multiple sites. It is common for materials or compounds to have multiple sites and this is
performed by the addition of another site into the NAME_in file. This can easily be
performed by changing a few parameters. In the NAME_in, file the first step should be
changing the number of sites of the MB from 1 to 2, or 2 to 3, etc, as seen in A32 and

223 
 

A33. From there, simple copying and pasting of the @ &dist1 then changing the 1 to 2
should be performed. The next step is copying all of the information from defining MB
site #1 to 0 0 0 toward the bottom of the file, as seen at the bottom of A31 and pasting it
below the 0 0 0. At this point, all of the numbers that were previously 1 such as defining
MB site #1, the @ &dist1, or (26.1.4) need to be changed to 2 as in defining MB site #2,
@ &dist2, and 26.2.4 respectively as seen in A33-A36.

Figure 7.32 Number of sites of the MB atom highlighted in the NAME_in file with 1
site.

224 
 

Figure 7.33 Number of sites of the MB atom highlighted in the NAME_in file with the
changed number of sites from 1 to 2.

225 
 

Figure 7.34 The @ &dist2 parameter highlighted after it was copied from the @ &dist1
parameter and the 1 changed to 2.

226 
 

Figure 7.35 The NAME_in file with defining MB site #2 highlighted after the 1 was
changed to 2 and the entire section of defining MB site #1 was copied and pasted
beneath the 0 0 0 at the bottom of defining MB site #1.

227 
 

Figure 7.36 The final paramater to be renamed, Make Gaussian 80 @&dist2, is
highlighted after all of the parameters have been changed in the defining MB site #2.
Now that all of the names of the parameters have been changed accordingly from 1 to 2
or 2 to 3 depending on how many sites have been added, the final parameter to be aware
of is the weight of the sublattice parameter. Previously, this parameter was not necessary
due to only 1 site being responsible for the entirety of the sample. With multiple sites,
each site will have a weighted effect on the peak intensities, widths, splitting etc.

228 
 

Therefore, changing the weight of the sublattice value will increase the effect those
parameters have on the fitting of the Mössbauer spectra. As illustrated by A37 and A38,
the values were both decreased to 0.5. This parameter should also be opened for
refinement.

Figure 7.37 The weight of sublattice parameter for site 1 within the NAME_in file is
highlighted illustrating its change from 1 to 0.5 due to the addition of the second site.

229 
 

When performing a fit refinement with multiple sites, it is strongly suggested to initially
start with one site and refine the site as far as possible before adding a second site. After
the addition of the second site, refine only the second site until the fit no longer improves.
At this point, opening different parameters with both sites may improve the fit. When it
has been determined the addition of a third site is required, again, only focus on the new
site before opening parameters of the previously refined sites.

Figure 7.38 The weight of sublattice parameter for site 2 within the NAME_in file is
highlighted illustrating its change from 1 to 0.5 due to the addition of the second site.

230 
 

At this point, a general understanding of how to operate CONUSS should have been
achieved. Parameters that were not discussed within were not used in laboratory
Mössbauer spectroscopy but were used in synchrotron Mössbauer spectroscopy.

231 
 

REFERENCES

232 
 

REFERENCES

[1]

W. Sturhahn, Hyperfine Interact. 125 (2000) 149.

233