SYNTHESIS AND CHARACTERIZATION OF CORRELATED ELECTRON
MATERIALS BY NOVEL SOFT-CHEMISTRY TOPOTACTIC TECHNIQUES
By
Colin K. Blakely

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 CORRELATED ELECTRON
MATERIALS BY NOVEL SOFT-CHEMISTRY TOPOTACTIC TECHNIQUES
By
Colin K. Blakely
A series of soft-chemistry, topotactic, multistep synthesis methods were
developed to synthesize a variety of metastable functional materials. Throughout
the thesis, a focus is placed on how crystal structure, bonding, and atomic
interactions play a role on determining function or properties of the material.
Materials synthesized were fully characterized, including Rietveld refinement (with
high-resolution diffraction data were available), stoichiometry characterization by
ICP and titration, thermal stability, and decomposition products. Properties
investigation was completed where applicable, including electrochemical cycling
of Li-ion cathode material candidates and magnetic properties characterization for
frustrated magnetic systems with competing coupling pathways.
The first method presented – solvothermal reduction – was used for the
exploratory synthesis of materials with NiO2 square planar units in an extended
lattice. As phases with square planar CuO2 units garnered much attention in the
1990s as high-temperature superconductors, NiO2 analogs may exhibit similar
properties. Synthesis of the infinite-layer compound La4Ni3O8 was used as the test
case for the solvothermal reduction method. A full discussion on reaction variables
as well as reactions routes is presented.

A family of fluorinated materials is also presented, prepared using multistep
fluorination methods. The oxyfluoride SrFeO2F was prepared via the infinite-layer
intermediate SrFeO2, the first phase with FeO2 square planar units in an extended
lattice. An investigation of the effects of experimental conditions on local
coordination is presented for this material. The multistep fluorination of La 4Ni3O8
is also presented, where two different fluorinated phases were prepared –
La4Ni3O8F1 and La4Ni3O8F2. The complex and anisotropic structure of these
phases is discussed, including the interesting T to T’ transformation. Lastly, a
solvothermal fluorination method is developed and demonstrated in the synthesis
of Sr2Co2O5F. The effect of the low-temperature synthesis on product purity is
discussed as well as the importance of experimental conditions.
The third technique described in this thesis is the multistep reductive exchange
(MURE) method for the synthesis of novel layered materials with spin frustrated
Cd-I2 type layers. A series of materials with different transition metals is presented,
focusing on the topotactic nature of the reactions. The synthesis of a group of
Co2+/3+ phases allowed for the further exploration of the low temperature magnetic
phase diagram for Co2.5+/Co4+ in layered phases with Cd-I2 type layers. Thermal
stability, electrochemistry, and reaction kinetics were also investigated.
Lastly, the Whole Powder Pattern Modeling algorithms are described and
shown to be a powerful tool for nanoparticle analysis. Particle size and strain
values were extracted from powder patterns of nanoparticles with high accuracy.

Copyright by
COLIN K. BLAKELY
2014

This dissertation is dedicated to my family and friends who helped make this
happen.

v

ACKNOWLEDGEMENTS

I would like to expressly acknowledge my advisor, Prof. Viktor Poltavets, for his
advice and attention to details as well as my coworker Dr. Shaun Bruno for the
stimulating conversation during my graduate tenure. I would also like to thank my
committee members for their guidance, including Prof. Thomas J. Pinnavaia,
Prof. Tom Hamann, Prof. Tim Hogan, and Prof. Greg Baker.

v

TABLE OF CONTENTS

LIST OF TABLES.................................................................................................. x
LIST OF FIGURES .............................................................................................xiv
Chapter 1.

Introduction to soft chemistry and experimental methods ............. 1
1.1 Introduction to solid state and soft chemistry methods............ 1
1.2 Contents of dissertation .......................................................... 3
1.3 General experimental methods and characterization ............ 11
1.3a Powder diffraction theory and instrumentation ............ 11
1.3b In-house powder diffraction ......................................... 14
1.3c Synchrotron diffraction at Argonne National Lab ......... 15
1.3d Diffraction data workup software ................................. 16
1.3e Thermogravimetric analysis ........................................ 16
1.3f ICP optical emission spectroscopy............................... 18
1.3g Electrochemical characterization................................. 19
REFERENCES ........................................................................... 23

Chapter 2.

Low-temperature solvothermal reduction method ....................... 30
2.1 Introduction ........................................................................... 30
2.2 Description of the novel solvothermal method ...................... 33
2.2a Experimental methods for the test case of La4Ni3O10 .. 33
2.2b Results of the reduction of La4Ni3O10 .......................... 36
2.2c Discussion on the role of pressure in the reaction ....... 45
2.3 LaNiO2 – square planar Ni1+.................................................. 47
2.3a Literature background of LaNiO2 ................................. 47
2.3b Reaction pathways for the formation of LaNiO2 .......... 50
2.3c Synthesis and characterization of LaNiO2 ................... 52
2.3d Reinvestigating the magnetic properties of LaNiO2 ..... 60
2.3e Comparison of solid-solid and solid-gas mechanisms . 66
2.3f DFT study on LaNiO2 and CaCuO2 similarities ............ 71
2.4 Reduction of EuNiO3 ............................................................. 74
2.4a Solvothermal synthesis of EuNiO3............................... 74

v

2.4b Discussion on the reduction of EuNiO3 ....................... 78
2.5 Chapter conclusions.............................................................. 84
2.6 Future work on this topic ....................................................... 85
REFERENCES ........................................................................... 86
Chapter 3.

Soft chemistry synthesis of extended oxyfluoride materials ........ 92
3.1 Survey of synthesis methods of inorganic oxyfluorides ......... 92
3.2 Novel multistep oxyfluoride synthesis procedure .................. 94
3.2a Multistep synthetic route and experimental methods .. 94
3.2b Synthesis and characterization of SrFeO2F ................ 98
3.2c Full magnetic properties investigation ....................... 105
3.2d Local coordination control using 57Fe Mössbauer ..... 108
3.2e DFT structure optimization of ordered SrFeO2F ........ 113
3.2f Concluding remarks for SrFeO2F ............................... 116
3.3 Synthesis of complex oxyfluorides La4Ni3O8Fx (x = 1,2) ..... 117
3.3a Solid state synthesis using XeF2 ............................... 118
3.3b Structural characterization of La4Ni3O8Fx (x = 1,2) .... 121
3.3c Discussion on the T to T’ transformation ................... 129
3.3d Concluding remarks for La4Ni3O8Fx (x = 1,2) ............ 130
3.4 Novel solvothermal synthesis of complex oxyfluorides ....... 131
3.4a Description of the solvothermal fluorination method .. 132
3.4b Proof-of-principle: synthesis of Sr2Co2O5F ................ 134
3.4c Exploring the importance of experimental conditions 137
3.5 Chapter conclusions............................................................ 147
3.6 Future work on this topic ..................................................... 148
REFERENCES ......................................................................... 149

Chapter 4.

MuRE – A multistep topotactic synthetic method ...................... 158
4.1 Literature background on α-NaFeO2 type phases ............... 158
4.1a Review of NaFeO2 type and similar structures .......... 163
4.1b Dependence of oxidation state on properties ............ 174
4.2 The multistep reductive exchange (MuRE) method ............ 186
4.3 Fe2.5+ phases prepared with the MuRE method .................. 192
4.3a Ca0.50FeO2 characterization and intercalation ........... 194
4.3b Magnetic measurements of Fe2.5+ phases................. 212
4.3c Electrochemical cycling of Li0.50Ca0.50FeO2 ............... 216

vi

4.4 Co2+/3+ phases prepared with the MuRE method ................ 221
4.4a Structure characterization of AxCayCoO2 phases ...... 221
4.4b Thermal analysis and stability investigation .............. 233
4.4c Magnetic properties - completing the phase diagram 238
4.4d Electrochemical cycling of layered cobaltates ........... 243
4.4e Intercalation kinetics exploration ............................... 248
4.5 Possibility of lower oxidation states with extended MuRE ... 251
4.6 Chapter conclusions............................................................ 253
4.7 Future work on this topic ..................................................... 254
REFERENCES ......................................................................... 256
Chapter 5.

Particle size analysis using Whole Powder Pattern Modeling ... 266
5.1 Introduction to particle size analysis .................................... 266
5.2 Effects of milling and annealing on stannate nanoparticles . 269
5.2a Experimental methods and WPPM input ................... 271
5.2b Effect of impact milling on particle size and strain ..... 275
5.2c Direct control over size and strain with annealing ..... 279
5.2d Concluding remarks and observations ...................... 281
5.3 Using WPPM to characterize synthesized nanoparticles .... 284
5.3a Direct synthesis of stannate nanoparticles ................ 284
5.3b Microscopy investigation of particle size ................... 287
5.3c Particle size distributions of stannate nanoparticles .. 289
5.4 Efficacy of WPPM on lower symmetry oxides ..................... 291
5.4a Introduction to delafossite type CuAlO2 ..................... 291
5.4b Synthesis of nanoparticle CuAlO2 ............................. 294
5.4c Particle size distribution of CuAlO2 nanoparticles ...... 294
5.5 Chapter conclusions............................................................ 296
REFERENCES ......................................................................... 297

vii

LIST OF TABLES

Table 2-1.

Vapor pressure, enthalpy of vaporization, and calculated
pressures at given temperatures for solvents used in solvothermal
reduction reactions. Chemical data obtained from the 2013 CRC
handbook online. ...................................................................... 36

Table 2-2.

Products of La4Ni3O10 reduction by different techniques.
Autogenous pressures were calculated using either the ClausiusClapeyron equation or the van der Waals equation depending on
the critical point temperature of the solvent. ............................. 40

Table 2-3.

Products of the solvothermal reaction of LaNiO3 with NaH at 150
°C. Calculated autogenous internal pressure was approximately
10 atm....................................................................................... 56

Table 2-4.

Products of the solvothermal reaction of LaNiO3 with NaH at 200
°C. Calculated autogenous internal pressure was approximately
60 atm....................................................................................... 56

Table 2-5.

Cell parameter comparison for LaNiO2. Three literature reported
cell parameter values as well as the cell parameters refined in this
thesis are presented ................................................................. 60

Table 2-6.

Bragg reflections of LaNiO2 and selected reflections of La2Ni2O5
that appear at similar d spacing, thereby complicating the
multiphase refinement .............................................................. 63

Table 3-1.

Unit cell parameters for SrFeO3-x, SrFeO2, and SrFeO2F.
Previously reported literature cell parameters for SrFeO2F:
Pm3̅ m, a = 3.956 Å (6, 7). ...................................................... 103

Table 3-2.

Atomic positions obtained from the Reitveld refinement of
synchrotron diffraction data for SrFeO2F with refined
occupancies............................................................................ 105

Table 3-3.

SrFeO2F Mössbauer spectrum fit parameters obtained using the
Fit:o) software package (35). .................................................. 110

v

Table 3-4.

Atomic parameters of La4Ni3O8F1 from Rietveld refinement of
synchrotron powder diffraction data in the space group I4/mmm
with a = 3.931(8) Å and c = 27.66(1) Å ................................... 123

Table 3-5.

Refined cell parameters of the La4Ni3OxFy family of compounds
prepared. Cell parameters obtained for La4Ni3O8F2 via LeBail
refinement of synchrotron data ............................................... 128

Table 3-6.

Structure parameters of Sr2Co2O5F based on Rietveld refinement
of powder diffraction data in the space group Pm3̅ m with a =
3.8549(1) Å ............................................................................. 136

Table 3-7.

Percent formation of Sr2Co2O5F via the fluorination of Sr2Co2O5
over time using four different methods. The total percent of
unreacted Sr2Co2O5 is the difference between the listed value
below and 100 ........................................................................ 140

Table 4-1.

Summary of structure types, space groups, cation coordination,
and the Co oxidation state in the three layer and two layer families
of the NaxCoO2 system ........................................................... 177

Table 4-2.

Unit cell parameters of sodium parent phases and calcium
exchange phases discussed in this chapter (53, 57). Notice the
topotactic nature of the aliovalent exchange preserves the
structural type between sodium and calcium phases. ............ 183

Table 4-3.

Cell parameters of Ca0.50FeO2 obtained from the LeBail
refinement of powder x-ray and neutron diffraction data at
different temperatures (61). Full Rietveld refinement was
undertaken as well. ................................................................. 196

Table 4-4.

Refined atomic positions for Ca0.50FeO2 obtained by Rietveld
refinement of NPD data. ......................................................... 198

Table 4-5.

Observed and calculated Bragg reflections for Ca0.50FeO2 in the
C2/m space group as proof of correct indexing. Indexing of the
PXD data was done using the TOPAS software package. ..... 199

Table 4-6.

Values for profile fitting parameters used in the Rietveld
refinement of Ca0.50FeO2 NPD data collected at both 75 K and 12
K ............................................................................................. 203

v

Table 4-7.

A comparison of the nuclear and incommensurate magnetic cell
parameters of Ca0.50FeO2 obtained from refinement of NPD
data ........................................................................................ 205

Table 4-8.

Atomic positions for Na0.50Ca0.50FeO2 from Rietveld refinement of
PXD data ................................................................................ 209

Table 4-9.

Atomic positions for Li0.50Ca0.50FeO2 from Rietveld refinement of
PXD data ................................................................................ 209

Table 4-10.

Unit cell parameters for Ca0.50FeO2 and the intercalated
compounds demonstrating the symmetry preservation throughout
the MuRE reactions ................................................................ 211

Table 4-11.

FeO6 interplanar distances comparison for the Fe series of
compounds synthesized with the MuRE method .................... 211

Table 4-12.

Rietveld refined atomic positions for Li0.75Ca0.25CoO2 from PXD
data ........................................................................................ 223

Table 4-13.

Cell parameters of the materials with transition metals in the 2+/3+
oxidation state couple prepared with the MuRE method next to the
cation coordination type and Co oxidation state ..................... 227

Table 4-14.

CoO6 interplanar spacings for all Co-containing layer phases
prepared with the MuRE method ............................................ 229

Table 4-15.

Particle size distributions obtained using the WPPM algorithms
implemented in the PM2K software suite. PM2K results gave the
shortest length (height of platelet) for platelet type particles ... 231

Table 4-16.

Electrochemical properties of common Li-ion cathode materials
compared to the novel cobaltates presented in this thesis ..... 244

Table 4-17.

Compound stoichiometry over intercalation time for the
preparation of the reduced phases presented in the manuscript.
Bolded numbers are stopping points for the preparation each
compound ............................................................................... 250

Table 5-1.

Particle sizes and crystallite strain values calculated using the
PM2K software for BaSnO3 and SrSnO3 samples milled for
different time periods .............................................................. 278

vi

Table 5-2.

Particle sizes and crystallite strain values calculated using the
PM2K software for BaSnO3 and SrSnO3 samples thermally
treated for 24 hours at varied temperatures............................ 282

vii

LIST OF FIGURES

Figure 1-1.

Structural transformation from the perovskite SrFeO3 to the
infinite layer SrFeO2 via reduction with CaH2 (50). The colored
balls represent the following atoms: blue - Fe, red - O, grey Sr............................................................................................... 4

Figure 1-2.

Results of topotactic and non-topotactic chemistry performed on
a pristine Jenga game, illustrating the power and tunability of
topotactic techniques................................................................. 5

Figure 1-3.

α-NaFeO2 type crystal structure, the parent of the phases
presented in Chapter 4 and synthesized using the MuRE
method. The structure consists of alternating layers of edge
shared FeO6 and NaO6 octahedra (teal and blue polyhedra,
respectively). ............................................................................. 9

Figure 1-4.

Actual coin cell setup using for all electrochemical
measurements performed and described in this thesis. These
split cells were purchased from MTI-XTL ................................ 22

Figure 2-1.

Structure models of La4Ni3O10 (left) and La4Ni3O8 (right) with
denoted layers and structural blocks: P, perovskite; RS, rock salt;
IL, infinite layer; F, fluorite. Yellow, cyan, and red balls represent
La, Ni, and O atoms, respectively ........................................... 38

Figure 2-2.

Phase composition versus reaction time during the reduction of
La4Ni3O10 with NaH at 200 °C: (a) solid state reaction, (b)
solvothermal conditions. La4Ni3O10, black squares; La4Ni3O9, red
circles; La4Ni3O8, blue triangles. Relative phase content is
normalized to 100% total ......................................................... 41

Figure 2-3.

Representative PXD patterns for the reduction products
prepared by different experimental methods. La 4Ni3O9, red
circles; La4Ni3O8, blue triangles ............................................... 44

Figure 2-4.

Refinement of PXD data for La4Ni3O8 using the TOPAS software
suite. The sample was prepared via the solvothermal reduction
route. ....................................................................................... 45

Figure 2-5.

d orbital manifold diagram of the crystal field splitting for a
transition metal in square planar configuration ........................ 48

v

Figure 2-6.

Powder X-ray diffraction patterns of products obtained by the
reaction between LaNiO3 and CaH2 at various reaction
temperatures using Cu Kα1 radiation. Reproduced with
permission from (31). .............................................................. 51

Figure 2-7.

Crystal structures of a) LaNiO3 and b) LaNiO2. Grey spheres
represent La3+, teal spheres represent Ni3+ (LaNiO3) or Ni1+
(LaNiO2), and red spheres are O2- .......................................... 53

Figure 2-8.

PXD pattern of the perovskite LaNiO3 after Pechini synthesis and
thermal treatment for 240 hours at 850 °C under an O 2
atmosphere. Diffraction pattern was taken on a D2 Phaser. ... 54

Figure 2-9.

PXD pattern of the perovskite LaNiO3 and the reduced infinitelayer type phase LaNiO2 synthesized via the solvothermal
reduction method at 200 °C for 10 days. ................................. 58

Figure 2-10.

Rietveld refinement of LaNiO2 obtained from the solvothermal
reduction of LaNiO3 at 200 °C for 10 days using NaH. Phase
analysis lead to a LaNiO2:La2Ni2O5:Ni (metal) product ratio of
500:12.35:1 ............................................................................. 61

Figure 2-11.

Susceptibility (closed circles) and inverse susceptibility (open
circles) data collected on a sample of LaNiO2 under a 1000 Oe
field using a SQUID magnetometer. ........................................ 64

Figure 2-12.

Effective magnetic moment of LaNiO2 across the temperature
range studied. A horizontal line was drawn in blue at 1.73 µB –
the effective magnetic moment of an S = ½ ion (Ni1+). ............ 65

Figure 2-13.

Literature fitted molar paramagnetic susceptibility of LaNiO2 from
the Hayward group. Adapted with permission from (9). Copyright
2013 American Chemical Society............................................ 66

Figure 2-14.

TGA/DSC data for the reduction of LaNiO3 under a 40:20 H2:N2
atmosphere from 25 – 1000 °C. The sample holder was loaded
under N2. ................................................................................. 69

Figure 2-15.

TGA/DSC data for the calorimetry experiment of LaNiO3 under a
N2 atmosphere from 25 – 1000 °C. The sample holder was
loaded under N2. ..................................................................... 70

Figure 2-16.

Comparison of calculated LDA projected paramagnetic DOS
LaNiO2 (upper panel) and CaCuO2 (lower panel) using DFT.

v

Reproduced with permission from (37). Copyright 2012 by The
American Physical Society. ..................................................... 74
Figure 2-17.

Crystal structure of the perovskite EuNiO3 (Pbnm, a = 5.2984(3)
Å, b = 5.4572(2) Å, c = 7.5365(3) Å) obtained via solvothermal
synthesis. ................................................................................ 77

Figure 2-18.

TGA data for EuNiO3 recorded under different atmospheres. . 80

Figure 2-19.

PXD spectra after decomposition of EuNiO3 at 950 °C in O2, N2,
and H2. Tickmarks refer to cubic Eu2O3 (wine, top), Ni metal
(olive, middle), and Ni (navy, bottom). All other reflections were
indexed as monoclinic Eu2O3. ................................................. 82

Figure 3-1.

Multistep synthetic route for the preparation of the SrFeO 2F
oxyfluoride. The previously known literature method using PVDF
at 400 °C is also highlighted (6). ............................................. 95

Figure 3-2.

Comparison of PXD patterns of the SrFeO3-x perovskite (bottom
pattern), the SrFeO2 infinite-layer compound (middle), and the
SrFeO2F oxyfluoride (top). ...................................................... 97

Figure 3-3.

Rietveld refinement of the synchrotron powder x-ray diffraction
data of SrFeO2F showing the observed pattern (black crosses),
fitted pattern (red trace), difference curve (gray trace, below), and
Bragg peak positions (magenta tickmarks, below) ................ 101

Figure 3-4.

Thermogravimetric analysis of the fluorinated product SrFeO2F
under an O2 and N2 atmospheres. Positive slope in the DSC
curve represents endothermic signals, while negative slope
represents exothermic signals ............................................... 106

Figure 3-5.

DC magnetization at H = 1000 Oe (a), DC field dependence (b),
AC magnetic susceptibility (c, d) and specific heat measurement
(a) results for SrFeO2F collected on a Quantum Design PPMS
system ................................................................................... 109

Figure 3-6.

Fit (black line) of room temperature 57Fe Mössbauer spectrum of
SrFeO2F with denoted contributions from Fe sites: Fe1 (blue
line), Fe2 (red line), Fe3 (green line) ..................................... 110

Figure 3-7.

Left: Potential energy surface of varied a/c structures. The
minimum position of the surface is highlighted by the white circle.
Right: Contour plot zoomed in around the minimum energy

vi

extracted from the potential surface shown on the left. The dark
blue area represents the minimum energy ............................ 114
Figure 3-8.

Topotactic transformation of La4Ni3O8 to La4Ni3O8F2. The
structure shown for La4Ni3O8F2 depicts O/F ordering. Colored
balls represent the following atoms: yellow – La, red – O, teal –
Ni, green – F ......................................................................... 119

Figure 3-9.

Rietveld refinement profile for the synchrotron powder X-ray
diffraction data of La4Ni3O8F1. Observed intensities (crosses),
calculated pattern (solid line), and difference curve (bottom solid
line). Tickmarks represent the following phases: La4Ni3O8F1 –
magenta (top), La4Ni3O8F2 – green (middle), La4Ni3O8 – blue
(bottom) ................................................................................. 122

Figure 3-10.

Electron diffraction patterns of La4Ni3O8F2 (left column) and
La4Ni3O8F1 (right column) ...................................................... 125

Figure 3-11.

Rietveld refinement of the NPD data of La4Ni3O8F1. Observed
intensities (crosses), calculated pattern (solid line), and
difference curve (bottom solid line). Tickmarks represent the
following phases: La4Ni3O8F1 – magenta (top), La4Ni3O8F2 –
green (middle), La4Ni3O8 – blue (bottom) .............................. 126

Figure 3-12.

X-ray absorption spectra for La4Ni3O10, La4Ni3O8, La4Ni3O8F1,
La4Ni3O8F2 as well as applicable standards .......................... 129

Figure 3-13.

Comparison of the two known methods for the preparation of the
oxyfluoride Sr2Co2O5F. Colored balls represent the following
atoms: blue – Co, red – O, white – Sr, green – F, and purple –
O/F disorder .......................................................................... 133

Figure 3-14.

Observed (green trace), calculated (crosses), and difference
(grey trace, below) XRD profiles obtained from Rietveld
refinement of PXD data for Sr2Co2O5F made at 100 °C. Magenta
tickmarks denote Bragg reflections. ...................................... 135

Figure 3-15.

DC magnetic susceptibility measurements of Sr2Co2O5F. Insets
show the field and temperature dependence on magnetization in
terms of magnetic moment per Co. Data collected on a Quantum
Design PPMS system............................................................ 137

vii

Figure 3-16.

Fluorination rates over time for the fluorination of Sr2Co2O5 using
XeF2 as a fluorine source under different experimental
conditions. The blue lines show the % of Sr2Co2O5F at any given
time ....................................................................................... 139

Figure 3-17.

Possible bonding interaction between XeF2 and two adjacent
CoO4 tetrahedra in the brownmillerite structure Sr2Co2O5. A full
description is presented in the text. Bond lengths were obtained
from the literature (20, 28, 66, 74, 75) ................................... 142

Figure 4-1.

Different structure variations found in the α-NaFeO2 system
based on the coordination of the alkali metal – O for octahedral
and P for trigonal prismatic. A further description of the naming
convention is given in the text ............................................... 158

Figure 4-2.

Rouxel type plot of bond ionicities vs. alkali metal/alkaline earth
metal. Dotted line separates P type (left side) from O type
phases. Bond ionicities were calculated using the Pauling
method (Equation 4-1) (4) .................................................... 162

Figure 4-3.

Graph of the figure-of-merit Z values for the in NaxCoO2 series of
layered cobaltates. Used with permission from (52). ............. 176

Figure 4-4.

Magnetic susceptibility (x) of NaxCoO2∙yH2O. Filled circles, zerofield cooling; open circles, field cooling. The susceptibility was
measured in a magnetic field of 20 Oe. The inset shows x
measured under various magnetic fields by the zero-field cooling
process. Reused without adaptation from (56). ..................... 181

Figure 4-5.

Schematic of the general four step MuRE method. In the
chemical formulae, M represents any 3d transition metal.
Colored spheres represent the following atoms: grey – Na1+, teal
– Mx+, blue – Ca2+, yellow – Na1+/Li1+. MO6 octahedra are shown
in teal for clarity. .................................................................... 185

Figure 4-6.

Rietveld refinement PXD data for α-NaFeO2 performed using the
GSAS software package. Peak anisotropy from preferred
orientation is shown in the difference plot (grey line). ........... 193

Figure 4-7.

PXD spectra of the Fe based family of materials prepared using
the MuRE method presented in this chapter. The background
(broad hump, peak near 16 °2θ) in Na0.50Ca0.50FeO2 was due to
the dome-type sample holder used during collection. ........... 195

viii

Figure 4-8.

Rietveld refinement of the PXD data of Ca0.50FeO2 showing the
observed pattern (black crosses), refined fit (red trace),
difference curve (grey below), and Bragg diffraction positions
(magenta tickmarks, below) .................................................. 197

Figure 4-9.

Rietveld refinement of NPD of Ca0.50FeO2 collected at 75 K at
Oak Ridge National Laboratory SPS showing the observed
pattern (black crosses), fit pattern (red trace), difference curve
(grey line, below), and Bragg peak positions (magenta tickmarks,
below). The refined unit cell is present in the top right corner
showing the topotactically preserved layered structure ......... 202

Figure 4-10.

Rietveld refinement of NPD of Ca0.50FeO2 collected at 12 K at
Oak Ridge National Laboratory SPS showing the observed
pattern (black crosses), fit pattern (red trace), difference curve
(grey line, below), and Bragg peak positions (magenta tickmarks,
below). The second line of magenta tickmarks represents the
magnetic cell observed. The first few reflections of the magnetic
cell are labeled ...................................................................... 204

Figure 4-11.

Refinement of PXD of Na0.50Ca0.50FeO2. Contribution from the
sample holder is the reason for the unindexed peaks/amorphous
background ........................................................................... 207

Figure 4-12.

DC magnetic susceptibility data collected for NaFeO2,
Ca0.50FeO2, and Na0.50Ca0.50FeO2 collected under a 1000 Oe
field........................................................................................ 212

Figure 4-13.

Effective magnetic moment for NaFeO2, Ca0.50FeO2, and
Na0.50Ca0.50FeO2 calculated from magnetic susceptibility
data ....................................................................................... 213

Figure 4-14.

Dependence of magnetization on applied field of
Na0.50Ca0.50FeO2 collected at 300 K and 5 K ......................... 215

Figure 4-15.

Partially optimized charge-discharge curve of Li0.50Ca0.50FeO2
cycled from 2.2 V to 3.8 V vs. a Li-metal anode .................... 218

Figure 4-16.

Secondary electron SEM of Ca0.50FeO2 prepared using the
MuRE method ....................................................................... 220

Figure 4-17.

PXD spectra of the Na0.50CoO2 based family of materials
prepared using the MuRE method presented in this chapter 222

ix

Figure 4-18.

PXD spectra of the NaCoO2 based family of materials prepared
using the MuRE method presented in this chapter ................ 224

Figure 4-19.

PXD spectra of the Na0.37CoO2 family of materials prepared
using the MuRE method presented in this chapter ................ 216

Figure 4-20.

SEM image of Na0.59Ca0.37CoO2 using back scattered electrons
(top) and secondary electrons (bottom) showing the large
anisotropy in crystal dimensions resulting in platelets. Clearly,
the ridges on the larger particles are actually many different
platelets ................................................................................. 232

Figure 4-21.

TGA and DSC curves of Li0.75Ca0.25CoO2 under O2 and N2
atmospheres. Negative signals in the DSC are exothermic .. 233

Figure 4-22.

DSC curves of the Na1+ intercalated products of CaxCoO2 (x =
0.50, 0.37, 0.25) under an N2 atmospheres. Negative signals in
the DSC are exothermic ........................................................ 234

Figure 4-23.

Magnified differential scanning calorimetry trace of
Na0.31Ca0.50CoO2 with the first derivative of the trace ............ 236

Figure 4-24.

Molar magnetic susceptibility of the Na1+ intercalated cobaltates
synthesized with the MuRE method obtained under an applied
field of 0.1 T .......................................................................... 238

Figure 4-25.

Magnetization dependence on field traces of the Na 1+
intercalated cobaltates synthesized with the MuRE method
obtained at 5 K ...................................................................... 240

Figure 4-26.

Low-temperature magnetic phase diagram of layered oxide
phases with Co in different oxidation states .......................... 242

Figure 4-27.

Charge-discharge curves of Li0.47Ca0.50CoO2 showing two
separate discharge events at 2.9 V and 3.2 V vs Li metal..... 245

Figure 4-28.

Per-cycle capacity of Li0.75Ca0.25CoO2 measured at a C/5 rate
against a Li metal anode ....................................................... 247

Figure 4-29.

A schematic diagram demonstrating the theoretical extendibility
of the MuRE method to produce phases with transition metal
oxidation states between 2+ and 2.5+ ................................... 251

x

Figure 5-1.

Powder x-ray diffraction patterns of a) BaSnO3 and b) SrSnO3 as
well as the corresponding calculated particle size distributions in
c) and d), respectively ........................................................... 276

Figure 5-2.

Williamson-Hall plots of a) BaSnO3 and b) SrSnO3 obtained
using the WPPM methods. The particle strain can be extracted
from the slope of the line using equation (5-5) ...................... 279

Figure 5-3.

Williamson-Hall plots of BaSnO3 and SrSnO3 particles after
thermal annealing for 24 hrs at various temperatures. The
particle strain can be extracted from the slope of the line using
equation (5-5) ........................................................................ 280

Figure 5-4.

The left 6 TEM micrograms represent BaSnO3, while the right 6
vertical panels show SrSnO3 particles. The labels a) - f) refer to
milling times: a) none, b) 30 min, c) 1 hr, d) 2 hr, e) 4hr, and f) 8
hr ........................................................................................... 283

Figure 5-5.

TEM (top, left) and HRTEM (top, right) micrographs of BaSnO3
nanoparticles synthesized using procedure I and TEM (bottom,
left) and HRTEM (bottom, right) micrographs of BaSnO 3
nanoparticles synthesized using procedure II. Both materials
were fired at 650 °C for 60 hrs .............................................. 286

Figure 5-6.

(a) & (b) TEM and HRTEM micrographs of SrSnO3 nanoparticles
synthesized at 650 °C for 72 hr (obtained from procedure I), (c)
& (b) TEM and HRTEM micrographs of SrSnO3
nanoparticles ......................................................................... 288

Figure 5-7.

Particle size distributions of the prepared BaSnO3 and SrSnO3
nanoparticles obtained using the WPPM method .................. 290

Figure 5-8.

Transmission electron microscopy pictures of CuAlO2
nanoparticles prepared at 775 °C under a controlled partial
oxygen pressure atmosphere (a,b), 800 °C under ultrapure N 2
(c,d), and 950 °C under ultrapure N2 (e,f).............................. 293

Figure 5-9.

Particle size distribution of CuAlO2 nanoparticles prepared at 775
°C, 800 °C, and 950 °C. Distribution plots calculated using the
PM2K software package. ...................................................... 295

xi

Chapter 1: Introduction to soft chemistry and experimental methods

1.1 Introduction to solid state and soft chemistry methods
For over a hundred years, the solid state chemist has mixed simple metal or
binary reagents and heated the mixture to a high enough temperature in hopes of
preparing a thermodynamically stable phase. Unfortunately, many of the most
interesting functional materials are kinetically stable rather than thermodynamically
stable; therefore, they can only be prepared using much softer synthetic
techniques. The soft-chemistry ideals borrow heavily from organic chemists, truly
the masters of efficient atomic substitution and movement in a specific and planned
fashion. Many very complex natural products have been published in the literature
in just the past year (1–12). These materials were produced starting from simple,
common materials, via a planned series of steps executed in a specific order,
demonstrating the true dominance of organic chemistry techniques over atomic
framework. Obviously, organic chemists have one very important advantage over
the solid-state community - in general, organic molecules are easily accessed
when suspended or dissolved in some type of solution. Solid state chemists are
required to fight both the slow kinetics of solid-solid reactions as well as the inability
to access specific atoms in a crystal lattice due to the extended nature of any
crystal.
In an effort to prepare functional materials, solid state chemists began to
assimilate synthetic techniques from other disciplines into the main stream solid
state toolbox about two decades ago (13). Multiple very nice literature reviews

1

have covered the most common topotactic reactions: intercalation, ion-exchange,
substitution, pillaring, layer construction, exfoliation, deintercalation, and grafting
(13, 14). Annual reports of soft-chemistry and topotactic chemistry in solid-state
materials synthesis have skyrocketed in the past few years with many new groups
joining the field led by the Kageyama, Cava, Wiley, Schaak, and Hayward research
labs (15–37). High temperature analogs of many of the soft-chemistry techniques
are known, especially ion-exchange and substitution; however, the lowtemperature alternatives allow for much greater control over specific structural
components (38, 39).
The ideals of the soft-chemistry mantra run deep in this thesis - a majority of
the chemistry presented employs topotactic methods to specifically modify
structures in a planned fashion to synthesize phases with desired properties.

2

1.2 Contents of dissertation
This thesis is divided into five main sections: the first gives an introduction to
general experimental methods used, the following three introduce and discuss
novel synthetic techniques developed, and a fifth focuses on methods to extract
particle information from powder x-ray diffraction patterns.
Chapter 2 presents a novel solvothermal reduction method to produce phases
with transition metals in infinite-layer coordination environments. The infinite layer
structure is represented in Figure 1-1, in which the primitive cell (P4/mmm) of ABX2
is shown with A in cubical coordination and BX2 shown as infinite square planar
units. This structure was proposed by Buckman as essentially a thought
experiment in a paper published in 2006, dubbing the structure "Kentuckia" (40);
however, the high-temperature superconducting cuprates initially synthesized in
1988 all have infinite-layer CuO2 sheets (38). Considering the superconducting
nature of CaCuO2 (Tc = 40 - 89 K), many different variations and dopings of this
parent phase were investigated in attempts to synthesize a room temperature
superconductor (41–49).

3

Figure 1-1. Structural transformation from the perovskite SrFeO3 to the infinite
layer SrFeO2 via reduction with CaH2 (50). The colored balls represent the
following atoms: blue - Fe, red - O, grey - Sr.

More recently, infinite-layer compounds with transition metals other than Cu
have been synthesized using low-temperature methods. The synthesis of LaNiO2,
with Ni1+ in square-planar geometry, has been attempted with mixed success
multiple times in the literature over the past two decades (29, 51–53). Due to the
metastability of LaNiO2 - and many other infinite layer materials where the
transition metal is not Cu - topotactic methods are generally employed to directly
control the final product. Topotactic reactions - reactions in which small
modifications are made to the crystal lattice while the general structure is
preserved - have grown in use in the recent solid state literature, showing a
movement away from standard "heat-and-beat" approaches (15, 18, 21, 23 –25,

4

60, 61). The concept of topotactic chemistry is depicted in Figure 1-2, using an
analogy to the common Jenga game.

Figure 1-2. Results of topotactic and non-topotactic chemistry performed on a
pristine Jenga game, illustrating the power and tunability of topotactic techniques.

The novel solvothermal method presented in Chapter 2 is investigated using
the literature reduction of La4Ni3O10 as a test case. The roles of the different
experimental variables including: solvent, reductant, reductant concentration,
temperature, and autogeneous pressure are established. After establishing the
method, the reduction of LaNiO3 is attempted to produce the pure phase nickelate
without the Ni and La2Ni2O5 impurities seen in previous literature reports. A full

5

reaction pathway discussion on the formation of LaNiO2 and an investigation into
why certain impurities plague the synthesis is described. A magnetic properties
comparison to previous data is also presented, confirming the higher purity sample
prepared using the solvothermal reduction method presented. Lastly, a
comparison of physical and chemical properties between LaNiO 2 and the
superconducting cuprates is discussed.
The last section of Chapter 2 explores the synthesis of EuNiO 2, a theoretical
compound with cuboidal Eu2+ and Ni2+ in square planar configuration. The difficult
synthesis of EuNiO3 is discussed, as are results from reduction attempts using the
solvothermal method. Future directions of this project are presented as well, based
on recent literature publications on isolated square planar units.
In Chapter 3, the common theme of topotactic, soft-chemistry reactions is
applied to synthesizing novel oxyfluoride materials using simple, less hazardous
reaction conditions. First, the synthesis of SrFeO2F via the infinite layer SrFeO2 is
presented using a multistep synthetic method. Full structural and magnetic
investigation of SrFeO2F was performed with high-intensity radiation sources, AC
and DC magnetic measurements, DFT calculations, and Mössbauer spectroscopy.
A comparison between the properties of the previously synthesized SrFeO2F and
the SrFeO2F synthesized herein is made with a clear conclusion that lowtemperature, topotactic reactions give much more control over phase structure and
configuration than the analogous higher temperature methods.
The multistep synthesis for the preparation of La 4Ni3O8F1 and La4Ni3O8F2 is
also presented in Chapter 3. Structural characterization using synchrotron and

6

neutron diffraction patterns were performed. An interesting discussion on structure
and the possible structural transformations in these materials is included.
The last third of Chapter 3 describes and demonstrates the fluorinating power
of the novel solvothermal fluorination method developed. The details behind how
the method was established and why it allows for the preparation of complex
oxyfluorides without byproducts are discussed. Furthermore, the relationship
between fluorine-source concentration and experimental volume is thoroughly
investigated to pinpoint the general reaction mechanism in the fluorination of the
brownmillerite phase Sr2Co2O5.
In Chapter 4, the third topotactic method developed is presented: multistep
reductive exchange (MuRE). This method, compiled of four different steps that
have been individually well-studied in the literature, is an extremely powerful tool
in the preparation of layered phases with transition metals in lower oxidation states
than previously reported. Because of the complexity of the system, an in-depth
literature background is presented to familiarize the reader with both the rich
literature, as well as the importance of this method as a synthetic tool in creating
new, and, more importantly interesting phases. The parent structure of all of the
phases synthesized is presented in Figure 1-3.
Following the introduction into the many nuances of the structure and synthetic
procedures, experimental details of the MuRE synthetic method are presented.
The procedure is then used to synthesize an array of phases with different
transition metals in AMO2 (A = Na, Li, Ca; M = Co, Fe) type layered structures,
where M is in the 2+/3+ oxidation state couple. The topotactic nature of the

7

reactions is corroborated via multiple techniques including titration, inductively
coupled plasma emission spectroscopy, powder x-ray diffraction, and neutron
diffraction. Intercalation rates and possible intermediate phases are also
presented. Thermal analysis was performed in efforts to determine if the phases
were truly metastable, or if they could be directly prepared and were actually just
low temperature (kinetic) phases.

8

Figure 1-3. α-NaFeO2 type crystal structure, the parent of the phases presented
in Chapter 4 and synthesized using the MuRE method. The structure consists of
alternating layers of edge shared FeO6 and NaO6 octahedra (green and blue
polyhedra, respectively).

As the common theme of this thesis is investigating the relationship between
structure and function, a range of properties were investigated for these materials,
including electrochemical, magnetic, and thermoelectric. Phases containing Li1+

9

were considered as candidates for Li-ion cycling cathode materials, while Na1+
phases underwent magnetic properties characterization and a small investigation
into thermal conductivity. The chapter concludes with a discussion on the future
direction of the MuRE project, as well as a proposal for reaching oxidation states
below 2.5+ - the current lower bound of the 4 step MuRE technique.
In the fifth and final chapter, a technical powder diffraction modeling method
called Whole Powder Pattern Modeling (WPPM) is described (56). The method
and implementation of WPPM in the PM2K software suite is briefly described, then
validated using alkaline earth stannates and delafossite CuAlO2 as case studies
(57, 58). Crystallite parameters including average particle size, strain, or even
defect density can be calculated using the PM2K software, with multiple examples
presented in this chapter.

10

1.3 General experimental methods and characterization
The following experimental methods are general guidelines; the experimental
synthesis and characterization methods used for each compound in this thesis will
be explicitly discussed.

1.3a Powder diffraction theory and instrumentation
In a powder poly-crystalline sample, the assumption is that, due to the large
number of crystals in random orientation relative to the number of possible
diffraction planes of the sample, the reciprocal space lattice can be condensed
from three-dimensional to a single-dimension projection. Given a reciprocal lattice
coordination system using spherical coordinates (r*, θ*, φ*), the reciprocal lattice
of a real space unit cell defined by the primitive vectors (a 1, a2, a3) can be
generated by determining the reciprocal space primitive vectors (b1, b2, b3) via the
following:
b1 =

a2 × a3
a1 ∙(a2 × a3 )

(1-1)

b2 =

a1 × a3
a2 ∙(a1 × a3 )

(1-2)

b3 =

a1 × a2
a3 ∙(a1 × a2 )

(1-3)

Due to the random orientation of the polycrystallites in the sample being
measured, the resultant diffraction pattern taken using a plate detector is a series
of concentric rings with a spacing directly related to the angle between the beam

11

axis and the ring, known as the 2θ angle. Each diffraction ring can be represented
by a reciprocal lattice vector as described previously. The relationship between the
location of the diffraction ring in real space to the reciprocal space lattice vector
can be extracted from the general reciprocal lattice vector equation, the Laue
equation, and the theory of elastic scattering, as follows.
G=hA+kB+lC

(1-4)

G∙(a+b+c)=2π(h+k+l)

(1-5)

The general reciprocal lattice vector G for a crystal lattice with primitive vectors
a, b, c, and reciprocal lattice indices (hkl) can be defined by both equations 1-4
and 1-5 by exploiting the Laue equations for integer value reflections (consistent
with Bragg’s Law) shown in equation 1-6.
a∙∆k=2πhb∙∆k=2πkc∙∆k=2πl

(1-6)

Rearrangement of the three Laue conditions to form a three dimensional
reciprocal lattice vector results in equation 1-7 and furthermore 1-8, which
essentially dictates that constructive interference of diffracted radiation (and a
resulting diffractogram) can only occur at integer (hkl) values.
∆k∙(a+b+c)=2π(h+k+l)

(1-7)

∆k=G

(1-8)

Realizing that ∆k is a measure of the difference between the incident radiation
wave vector ki and the diffracted wave vector ko (1-9) allows for rearrangement of
1-8 to arrive at the general diffraction condition 1-10.
∆k= ko -ki

12

(1-9)

ko -ki =G

(1-10)

Squaring both sides of the equation (1-11) and applying the simplification that
(ko)2 = (ki)2 for elastic scattering leads to equation 1-12, which is directly related to
Bragg’s Law (1-13) where k = 2sin(θ)/λ under the Bragg condition and G = 2π/d
(conversion from reciprocal to real space).
k20 - 2k0 ki + k2i = G
2ki ∙G=G

2

2

(1-12)

2dsin(θ)=nλ
When

multiple diffractograms taken

(1-11)

(1-13)
with

different incident radiation

wavelengths (for example, synchrotron and lab powder diffraction) are to be
compared, the Q space definition derived from 1-12 where Q = G should be
employed to allow for plotting diffractograms as a function of the scattering vector
Q (1-14).
Q=

4πsinθ
λ

(1-14)

The ability to convert between real space and reciprocal space is paramount to
the theory of powder diffraction because in reciprocal space, or momentum space,
the momentum difference between incoming and diffracted radiation is defined by
a reciprocal lattice vector; therefore, the atomic positions in a unit cell are
determinable by combining chemical compositional knowledge and a diffraction
pattern with known incident radiation wavelength.
1.3b In-house powder diffraction

13

The in-house diffractometer that the majority of the PXD scans presented in
this thesis were run on was a Bruker D2 Phaser with LYNXEYE detector. This
diffractometer is optimized for rapid acquisition of diffractograms with electronic
fluorescence control to increase the signal-to-noise ratio. Sample preparation for
measurement was slightly different depending on the air-sensitivity of the
compound being measured.
For an air-stable material, 100 mg – 150 mg were weighed out and placed in
the middle of an etched silicon wafer sample holder. The powder was lightly
tamped with a glass slide to prevent anomalous peak intensity errors due to
sample displacement. The majority of the diffractograms were obtained with a 2θ
range of 10 ° 2θ to 70 ° 2θ for both air-stable and air-sensitive samples. The
presence of straight-through beam intensity below 10 °2θ and lack of diffraction
resolution above 70 °2θ were instrumental parameters that dictated the
experimentally chosen diffraction range. When diffractograms of air-sensitive
materials were needed, the D2 Phaser was placed inside of a sealed N 2 filled
glovebox. Samples were prepared in a similar manner but with a smaller sample
holder to accommodate the smaller amount of mass present (generally 30 mg –
50 mg). Diffractogram scans for both air-stable and air-sensitive samples were run
with a step size of 0.02 °2θ per second. Publication quality results were obtained
with a step size 0.02 °2θ per 10 seconds. For exceptionally air-sensitive samples,
a dome-shaped plastic sample holder was loaded under N2 and sealed.

14

The glovebox environment created to house the D2 Phaser was a custom job
completed in collaboration with Plas-Labs from Lansing. Plas-Labs fabricated the
stainless steel shell with hatch door and placed four KF45 fittings in the back based
on blueprints drafted. After taking delivery of the shell, KF connection feedthroughs were made to pass power, Ethernet, and a chilling system into the
glovebox. Due to the minimal airflow and high heat output from the D2 Phaser, a
multi-faceted chilling setup was devised. The glovebox was cooled by a radiator
type system with 5 gal/min constant flow that was passed through 50 ft. of
aluminum pipe outside of the glovebox, as well as a radiator housed in a bucket of
circulating chilled water. The D2 Phaser was also cooled by an external chiller with
constant recirculation set at 19 °C. These two methods combined held the box
temperature at 30 °C during long measurements.

1.3c Synchrotron diffraction at Argonne National Lab
The Synchrotron Powder x-ray Diffraction (SPXD) pattern was collected on the
Argonne National Lab Synchrotron BM-11 beam line via the sample mail-in
program using a 0.8 mm kapton capillary with an approximately 100 mg sample.
The sample was loaded into the sample holder under an inert atmosphere. The
synchrotron source wavelength was 0.373822 Å with a sample measurement time
of 30 minutes. Data was worked up using the GSAS software package.

15

1.3d Diffraction data workup software
Two software packages, Bruker TOPAS and GSAS/EXPGUI, were employed
to extract sample information from the recorded diffractograms. TOPAS was used
exclusively to index diffractograms due to its indexing and cell parameter interface,
while GSAS was used for all Rietveld refinements.

1.3e Thermogravimetric analysis
Thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC)
measurements were performed on a Netsch Jupiter STA 499 F3 system located
in a custom-fitted purge glovebox. All gas and cooling flow to and from the
instrument was passed through air-tight feedthroughs in the back of the glovebox.
Air or moisture sensitive samples were loaded into the sealed glovebox using
standard air-sensitive techniques.
The Netsch Jupiter STA 449 F3 is a vacuum tight, top-down simultaneous
TGA/DSC instrument with a balance sensitive from the g to µg level. 99.999%
ultrapure N2 or Ar was used as the protective gas in all experiments, safely
protecting the balance from exposure to decomposition products and allowing for
a baseline calibration curve to remove gas buoyancy effects. Calibration curves
were created for sample heating rates of 5 °C/min and 1 °C/min with a maximum
sample temperature of 1300 °C.
TGA is a simple yet powerful technique in which the sample mass is
continuously recorded as the sample temperature changes – allowing for the
calculation of decomposition products by mass changes. DSC analysis

16

instruments constantly record the difference in heat required to raise the
temperature of the unknown sample compared to a known reference sample with
a well-defined heat capacity. From the DSC results, information such as melting,
phase transition temperatures, or enthalpies of transitions can be calculated (∆H
= KA; K = calorimetric constant, A = area under DSC curve).
The typical sample introduction and measurement procedure for an airsensitive sample is described herein. 100 mg of the sample loaded under N 2 in a
paraffin wrapped vial is placed in the antechamber of the purging Jupiter STA 449
F3 glovebox at a purge rate of approximately 2 L/min of ultrapure N2 generated by
a Cryo Industries XG350 N2 generator. Concurrently, an alumina sample holder is
heated to 300 °C for 10 min to remove any moisture adsorbed on the surface,
placed in a vial, and loaded into the antechamber. After purging for 2 hrs, both the
sample and the sample holder are brought in to the purge box. Approximately 25
– 40 mg of sample is loaded into the sample holder, and the sample holder is
loaded onto the TGA balance. After sealing the system with the sample on the
balance and the protective N2 gas flowing, the active gas is introduced to the
sample at a defined flow rate based on the buoyancy calibration curve (generally
40 mL/min protective / 20 mL/min active). After completion of the heating program,
the sample is furnace cooled to room temperature, loaded into a paraffin-wrapped
vial, and moved into the glovebox containing the Bruker D2 diffractometer for postTGA phase characterization in order to further understand the decomposition
products.

17

All TGA and DSC data was worked up using the Netzsch Proteus software
package, which provided the necessary tools to extract pertinent data.

1.3f ICP optical emission spectroscopy
Inductively Coupled Plasma - Optical Emission Spectroscopy (ICP-OES) is a
technique used to determine the concentrations of different elements in a sample.
ICP-OES can be used to find exact concentrations of elements (quantitative ICP)
or relative molar ratios (qualitative ICP), depending on the requirements of the
experiment.
An ICP instrument works on the principle of optical emission - when atoms are
ionized they emit radiation specific to each element. The intensity of the emitted
radiation can then be used to calculate the concentration of each element in the
sample. An Ar plasma is generated in an ICP by passing Ar gas through a coil with
a large electromagnetic field created by induction using high power RF. The initial
plasma is ignited with a discharge from a Tesla coil. As the sample of interest is
passed through the plasma, the atoms are ionized and the resultant emission is
channeled to a detector via plasma separation or optical fibers. Photomultiplier
tubes or other detectors can be used to quantify the intensity of the emission signal.
Sample preparation for ICP analysis is simple for the majority of ceramic
oxides. A 10 mg sample of the oxide was weighed out in the glovebox and
transferred to a 20 mL glass vial. The vial was then filled with spectrophotometric
grade concentrated nitric acid. After stirring and slight heating, 90% of all samples
dissolve. Samples including Cu or main group elements required the addition of

18

HCl to generate aqua regia. The resulting solution has a concentration of
approximately 500 ppm, which is 10x too high to inject into the ICP instrument.
Following complete dissolution of the sample, 1 mL of this mixture was transferred
to a 15 mL test tube filled with spectrophotometric grade water. The sample
concentration of the test tube at this point is approximately 35 ppm in 5% nitric
acid.
The sample is introduced into the ICP instrument with a 5% nitric acid carrier
solution via a peristaltic pump.

1.3g Electrochemical characterization
The optimized procedure for preparing a sample for electrochemical
characterization is summarized in this section. As each sample is different at both
the macroscopic and microscopic scales, small sample-based changes to this
method may be required for optimal results. However, if followed verbatim, the
below procedure will result in initial measurements that assist in identifying the
ideal measurement voltage range or scanning speed.
Initial sample preparation is key to acquiring good electrochemical
characterization data using Li-ion coin cell setups. The biggest disadvantage of
many of the materials presented in this thesis, with respect to electrochemical
performance as Li-ion cathode materials, is the very low electronic conductivity
intrinsic with semiconducting/insulating phases. The easiest way to approach this
deficiency is by intimately mixing the material with a conductive powder (for
example, carbon) in the form of acetylene black or Timcal SuperC 65.

19

The first step in preparing the sample for measurement is intimate mixing with
the carbon additive, a binding additive, and a solvent. Example binding additives
include polytetrafluoroethylene (PTFE), polyvinylidene fluoride (PVDF), and
poly(3-methylthiophene) (59). PVDF was used as the binder in the following guide
for two reasons: lower capacity fade rate upon subsequent charging, and high
solubility in polar, aprotic solvents (60). From both literature and experience, a
good starting ratio for active material:carbon additive:binder additive is 80:12:8
(61). Depending on the conductivity of the sample, the carbon content can be
reduced down to as low as 2% by mass (62, 63). This powder mixture is then
placed in a 20 mL vial with a stir bar in the glovebox. To this vial, approximately 2
mL of N-methyl-2-pyrrolidone (NMP) is added, then the slurry is stirred for 12 hrs.
NMP is a polar, aprotic solvent that PVDF is readily soluble in, as opposed to the
other binder additives. MP also has relatively low surface tension – an important
characteristic during the film phase of sample preparation.
After the paste has stirred for 12 hours, the film is made using a doctor blade
tool. A piece of battery grade Al foil (99.99% MTI-XTL) is cut into an 8”x4” rectangle
and placed longwise on a piece of tempered glass. Approximately 3 mL of paste
is applied drop wise 3" from the left edge of the Al foil and 2" from the bottom,
forming a hemisphere shape with 1" diameter. The doctor blade tool is then set to
between 200 and 250 microns and slowly moved across the paste. A drag speed
of 1" per minute is optimal. As the doctor blade passes over the paste a uniform
layer will form. This step is repeated until the desired paste area is achieved while
leaving a clean 1" edge along the entire Al sheet. After forming the film, it must be

20

cured to remove the NMP and densify the film. The Al sheet and tempered glass
plate are then loaded into the antechamber and heated to 130 °C, followed by
evacuation to 20 inHg. Allen keys should be placed along the clean edges of the
Al foil to prevent it from moving during evacuation. The heating under vacuum
treatment should continue for a minimum of 6 hours - generally overnight.
Once the film is removed from the antechamber it must be cut to size for coin
cell fabrication. Using the die press in the glovebox, cut a 19 mm chip of the active
material. The chip is then pressed with a pressure of 2000 lb/in 2. This chip, along
with an equivalent 19 mm Celgard polyethylene membrane, are placed - active
material below - in a split-cell type setup (Figure 1-4). Approximately 2 mL of
selected electrolyte is then added to the cell, followed by a Li chip (99%, MTI-XTL).
The cell is sealed according to Figure 1-4 and the potentiostat leads are
connected. Review the Gamry software package tutorial for setting up and
initializing experiments.

21

Figure 1-4. Actual coin cell setup used for all electrochemical measurements
performed and described in this thesis. These split cells were purchased from MTIXTL.

22

REFERENCES

23

REFERENCES

(1)

Larson, R. T.; Clift, M. D.; Thomson, R. J. Total synthesis of the Galbulimima
alkaloid (-)-GB17. Angew. Chem. Int. Ed. Engl. 2012, 51, 2481–4.

(2)

Deng, J.; Zhu, B.; Lu, Z.; Yu, H.; Li, A. Total synthesis of (-)-fusarisetin A and
reassignment of the absolute configuration of its natural counterpart. J. Am.
Chem. Soc. 2012, 134, 920–3.

(3)

Michels, T. D.; Dowling, M. S.; Vanderwal, C. D. A synthesis of echinopine
B. Angew. Chem. Int. Ed. Engl. 2012, 51, 7572–6.

(4)

Jakubec, P.; Hawkins, A.; Felzmann, W.; Dixon, D. J. Total synthesis of
manzamine A and related alkaloids. J. Am. Chem. Soc. 2012, 134, 17482–
5.

(5)

Faber, J. M.; Eger, W. A.; Williams, C. M. Enantioselective total synthesis of
the mexicanolides: khayasin, proceranolide, and mexicanolide. J. Org.
Chem. 2012, 77, 8913–21.

(6)

Enomoto, M.; Morita, A.; Kuwahara, S. Total synthesis of the tremorgenic
indole diterpene paspalinine. Angew. Chem. Int. Ed. Engl. 2012, 51, 12833–
6.

(7)

Tian, S.; Zi, W.; Ma, D. Potentially biomimetic total synthesis and relative
stereochemical assignment of (±)-gracilamine. Angew. Chem. Int. Ed. Engl.
2012, 51, 10141–4.

(8)

Diethelm, S.; Carreira, E. M. Total synthesis of (±)-Gelsemoxonine. J. Am.
Chem. Soc. 2013, 135, 8500–3.

(9)

Bonazzi, S.; Cheng, B.; Wzorek, J. S.; Evans, D. A. Total synthesis of (-)nakadomarin A. J. Am. Chem. Soc. 2013, 135, 9338–41.

(10) Mukai, K.; Urabe, D.; Kasuya, S.; Aoki, N.; Inoue, M. A convergent total
synthesis of 19-hydroxysarmentogenin. Angew. Chem. Int. Ed. Engl. 2013,
52, 5300–4.
(11) Tsuna, K.; Noguchi, N.; Nakada, M. Convergent total synthesis of (+)ophiobolin A. Angew. Chem. Int. Ed. Engl. 2011, 50, 9452–5.

24

(12) Frederich, J. H.; Harran, P. G. Modular access to complex prodiginines: total
synthesis of (+)-roseophilin via its 2-azafulvene prototropisomer. J. Am.
Chem. Soc. 2013, 135, 3788–91.
(13) Schaak, R. E.; Mallouk, T. E. Perovskites by Design: A Toolbox of SolidState Reactions. Chem. Mater. 2002, 14, 1455–1471.
(14) Ranmohotti, K. G. S.; Josepha, E.; Choi, J.; Zhang, J.; Wiley, J. B.
Topochemical manipulation of perovskites: low-temperature reaction
strategies for directing structure and properties. Adv. Mater. 2011, 23, 442–
60.
(15) Sivakumar, T.; Wiley, J. Topotactic route for new layered perovskite oxides
containing fluorine: Ln1.2Sr1.8Mn2O7F2 (Ln=Pr, Nd, Sm, Eu, and Gd). Mater.
Res. Bull. 2009, 44, 74–77.
(16) Seddon, J.; Suard, E.; Hayward, M. A. Topotactic reduction of YBaCo 2O5
and LaBaCo2O5: square-planar Co(I) in an extended oxide. J. Am. Chem.
Soc. 2010, 132, 2802–10.
(17) Hayward, M. A. Phase Separation during the Topotactic Reduction of the
Pyrochlore Y2Ti2O7. Chem. Mater. 2005, 17, 670–675.
(18) Muguerra, H.; Colin, C.; Anne, M.; Julien, M.-H.; Strobel, P. Topotactic
synthesis, structure and magnetic properties of a new hexagonal polytype of
silver cobaltate(III) AgCoO2+δ. J. Solid State Chem. 2008, 181, 2883–2888.
(19) Blakely, C. K.; Bruno, S. R.; Poltavets, V. V. Low-temperature solvothermal
approach to the synthesis of La4Ni3O8 by topotactic oxygen deintercalation.
Inorg. Chem. 2011, 50, 6696–700.
(20) Overton, A. J.; Best, J. L.; Saratovsky, I.; Hayward, M. a. Influence of
Topotactic Reduction on the Structure and Magnetism of the Multiferroic
YMnO3. Chem. Mater. 2009, 21, 4940–4948.
(21) Hyett, G.; Barrier, N.; Clarke, S. J.; Hadermann, J. Topotactic oxidative and
reductive control of the structures and properties of layered manganese
oxychalcogenides. J. Am. Chem. Soc. 2007, 129, 11192–201.
(22) Hadermann, J.; Abakumov, A. M.; Adkin, J. J.; Hayward, M. A. Topotactic
reduction as a route to new close-packed anion deficient perovskites:
structure and magnetism of 4H-BaMnO2+x. J. Am. Chem. Soc. 2009, 131,
10598–604.

25

(23) Saratovsky, I.; Lockett, M. A.; Rees, N. H.; Hayward, M. A. Preparation of
Sr7Mn4O13F2 by the topotactic reduction and subsequent fluorination of
Sr7Mn4O15. Inorg. Chem. 2008, 47, 5212–7.
(24) Dixon, E.; Hayward, M. A. The Topotactic Reduction of Sr3Fe2O5Cl2-Square
Planar Fe(II) in an Extended Oxyhalide. Inorg. Chem. 2010, 49, 9649–54.
(25) Tassel, C.; Kageyama, H. Square planar coordinate iron oxides. Chem. Soc.
Rev. 2012, 41, 2025–35.
(26) Toda, K. Room temperature synthesis and characterization of perovskite
compounds. Solid State Ionics 2002, 154-155, 393–398.
(27) Blakely, C. K.; Bruno, S. R.; Poltavets, V. V. Multistep soft chemistry method
for valence reduction in transition metal oxides with triangular (CdI2-type)
layers. Chem. Commun. (Camb). 2014, 2–5.
(28) Tassel, C.; Watanabe, T.; Tsujimoto, Y.; Hayashi, N.; Kitada, A.; Sumida, Y.;
Yamamoto, T.; Kageyama, H.; Takano, M.; Yoshimura, K. Stability of the
infinite layer structure with iron square planar coordination. J. Am. Chem.
Soc. 2008, 130, 3764–5.
(29) Takamatsu, T.; Kato, M.; Noji, T.; Koike, Y. Low-temperature synthesis of
the infinite-layer compound LaNiO2 using CaH2 as reductant. Phys. C
Supercond. 2010, 470, S764–S765.
(30) Ma, R.; Liu, Z.; Takada, K.; Iyi, N.; Bando, Y.; Sasaki, T. Synthesis and
exfoliation of Co2+-Fe3+ layered double hydroxides: an innovative
topochemical approach. J. Am. Chem. Soc. 2007, 129, 5257–63.
(31) Stoltz, C.; Ramesha, K.; Sirchio, S. a; Gönen, Z. S.; Eichhorn, B. W.;
Salamanca-Riba, L.; Gopalakrishnan, J. Topochemical anion metathesis
routes to the Zr2N2S phases and the Na2S and ACl derivatives (A = Na, K,
Rb). J. Am. Chem. Soc. 2003, 125, 4285–92.
(32) Badri, V. Topochemical oxygen deintercalation in oxides using butyllithium.
Phys. C Supercond. 2000, 339, 106–112.
(33) Viciu, L.; Kodenkandath, T. A.; Wiley, J. B. Construction of a double-layered
tetrahedral network within a perovskite host: Two-step route to the alkalimetal-halide layered perovskite, (LixCl)LaNb2O7. J. Solid State Chem. 2007,
180, 583–588.

26

(34) Neiner, D.; Spinu, L.; Golub, V.; Wiley, J. B. Ferromagnetism in
Topochemically Prepared Layered Perovskite Li0.3Ni0.85La2Ti3O10. Chem.
Mater. 2006, 18, 518–524.
(35) Mcintyre, R. A.; Falster, A. U.; Li, S.; Simmons, W. B.; Connor, C. J. O.;
Wiley, J. B. Opening a Perovskite to Valence Manipulation : Two-Step
Topotactic Route to a New Mixed-Valence Titanate, Na1-x+yCax/2LaTiO4. J.
Am. Chem. Soc. 1998, 7863, 217–218.
(36) Choi, J.; Zhang, X.; Wiley, J. B. Building alkali-metal-halide layers within a
perovskite host by sequential intercalation: (A2Cl)LaNb2O7 (A = Rb, Cs).
Inorg. Chem. 2009, 48, 4811–6.
(37) Schaak, R. E.; Mallouk, T. E. Prying Apart Ruddlesden−Popper
Phases: Exfoliation into Sheets and Nanotubes for Assembly of Perovskite
Thin Films. Chem. Mater. 2000, 12, 3427–3434.
(38) Francesconi, M. G.; Greaves, C. Anion substitutions and insertions in copper
oxide superconductors. Phys. C 1997, 10, 29–37.
(39) Clearfield, A. Role of ion exchange in solid-state chemistry. Chem. Rev.
1988, 88, 125–148.
(40) Bucknum, M. J.; Castro, E. A. Kentuckia: A 4,6,8-connected tetragonal
structural pattern in space group P4/nmm. Russ. J. Gen. Chem. 2006, 76,
265–266.
(41) Slater, P. R.; Hodges, J. P.; Francesconi, M. G.; Greaves, C.; Slaskic, M.
Fluorination of the Ruddlesden – Popper type cuprates , LnACuO ( Ln = La
, Nd ; A = Ca , Sr ). Structure 1997, 7, 2077–2083.
(42) Nakayama, R.; Tanaka, Y.; Yui, H.; Yamauchi, H.; Baron, A. Q. R.;
Sasagawa, T. Preparation of infinite layer cuprate compounds
(Sr1−xLax)CuO2 and detailed phonon properties in x = 0. Mater. Sci. Eng. B
2010, 173, 41–43.
(43) Sachdev, S. Where is the quantum critical point in the cuprate
superconductors? Phys. Status Solidi 2010, 247, 537–543.
(44) Iyo, A.; Hirai, M.; Tokiwa, K.; Watanabe, T.; Tanaka, Y. Preparation of
polycrystals with various Tc and single crystal growth of Ba 2Ca3Cu4O8(O1yFy)2 under high pressure. Phys. C Supercond. 2003, 392-396, 140–144.

27

(45) Hadermann, J.; Van Tendeloo, G.; Abakumov, A. M.; Rozova, M. G.;
Antipov, E. V. HREM Study of Fluorinated Nd 2CuO4. J. Solid State Chem.
2001, 157, 56–61.
(46) AI-Mamouri, M.; Edwards, P. P.; Greaves, C.; Slaski, M. Synthesis and
superconducting properties of the strontium copper oxyfluoride Sr2CuO2F2+δ.
Nature 1994, 369, 382–384.
(47) Zhao, H.; Shen, J.-L.; Zhao, Z. X. Proximity effect in layered high Tc
superconductors. Phys. C Supercond. 2000, 333, 242–252.
(48) Choy, J.-H.; Park, N.-G.; Hwang, S.-J.; Kim, D.-H.; Hur, N. H. New
Superconducting Intercalation Compounds: (HgX2)0.5Bi2Sr2CaCu2Oy (X = Br
and I). J. Am. Chem. Soc. 1994, 116, 11564–11565.
(49) Shimizu, S.; Mukuda, H.; Kitaoka, Y.; Iyo, a.; Tanaka, Y.; Kodama, Y.;
Tokiwa, K.; Watanabe, T. Uniform Mixing of Antiferromagnetism and HighTemperature Superconductivity in Electron-Doped Layers of Four-Layered
Ba2Ca3Cu4O8F2: A New Phenomenon in an Electron Underdoped Regime.
Phys. Rev. Lett. 2007, 98, 60–63.
(50) Tsujimoto, Y.; Tassel, C.; Hayashi, N.; Watanabe, T.; Kageyama, H.;
Yoshimura, K.; Takano, M.; Ceretti, M.; Ritter, C.; Paulus, W. Infinite-layer
iron oxide with a square-planar coordination. Nature 2007, 450, 1062–5.
(51) Hayward, M. A.; Green, M. A.; Rosseinsky, M. J.; Sloan, J. Sodium Hydride
as a Powerful Reducing Agent for Topotactic Oxide Deintercalation:
Synthesis and Characterization of the Nickel(I) Oxide LaNiO2. J. Am. Chem.
Soc. 1999, 121, 8843–8854.
(52) Crespin, M.; Isnard, O.; Dubois, F.; Choisnet, J.; Odier, P. LaNiO: Synthesis
and structural characterization. J. Solid State Chem. 2005, 178, 1326–1334.
(53) Crespin, M.; Levitz, P.; Gatineau, L. Reduced Forms of LaNiO3 Perovskite.
1983, 1181–1194.
(54) Shimakawa, Y.; Inoue, S.; Haruta, M.; Kawai, M.; Matsumoto, K.;
Sakaiguchi, A.; Ichikawa, N.; Isoda, S.; Kurata, H. Topotactic Changes in
Thin Films of Brownmillerite SrFeO2.5 Grown on SrTiO3 Substrates to
Infinite-Layer Structure SrFeO2. Cryst. Growth Des. 2010, 10, 4713–4715.
(55) Pralong, V.; Reddy, M. A.; Caignaert, V.; Malo, S.; Lebedev, O. I.;
Varadaraju, U. V; Raveau, B.; Crismat, L.; Mar, B.; Caen, F.-. A New Form

28

of LiNbO3 with a Lamellar Structure Showing Reversible Lithium
Intercalation. Chem. Mater. 2011, 23, 1915–1922.
(56) Leoni, M.; Scardi, P. Nanocrystalline domain size distributions from powder
diffraction data research papers. J. Appl. Crystallogr. 2004, 37, 629–634.
(57) Blakely, C. K.; Bruno, S. R.; Baum, Z. J.; Poltavets, V. V. Effects of ball
milling and thermal annealing on size and strain of ASnO 3 (A = Ba, Sr)
ceramics. Solid State Sci. 2013, 15, 110–114.
(58) Leoni, M.; Confente, T.; Scardi, P. PM2K: a flexible program implementing
Whole Powder Pattern Modelling. Zeitschrift für Krist. Suppl. 2006, 2006,
249–254.
(59) Xu, W.; Read, A.; Koech, P. K.; Xiao, J. Effects of Substitution Positions,
Electrolytes, and Binders on the Performance of Anthraquinone-based
Cathode Materials. Chem. Eng. 2011.
(60) Barsykov, V.; Khomenko, V. The Influence of Polymer Binders on the
Performance of Cathodes for Lithium-Ion Batteries. Sci. J. Riga Tech. Univ.
2010, 21, 67–71.
(61) Chou, S.; Wang, J.; Liu, H.; Dou, S. Rapid Synthesis of Li 4Ti5O12
Microspheres as Anode Materials and Its Binder Effect for Lithium-Ion
Battery. 2011, 16220–16227.
(62) Chu, C.-M.; Liu, C.-Y.; Wang, Y.-Y.; Wan, C.-C.; Yang, C.-R. On the
evaluation of the factors influencing the rate capability of a LiCoO2|Li battery.
J. Taiwan Inst. Chem. Eng. 2012, 43, 201–206.
(63) Yamada, A.; Chung, S. C.; Hinokuma, K. Optimized LiFePO 4 for Lithium
Battery Cathodes. J. Electrochem. Soc. 2001, 148, A224.

29

Chapter 2 : Low-temperature solvothermal reduction method

2.1 Introduction
Conventional high-temperature (≥750 °C) solid state synthetic techniques
result in the preparation of thermodynamically stable products, while the lowtemperature chimie douce (soft chemistry) approach allows for the formation of
new metastable phases with unique structural features. Because extensive parts
of the structure are retained during soft chemistry reactions, these reactions are
topotactic in nature. In spite of the low diffusion coefficients of ions in solids at
lower temperatures, there are many examples of topotactic reactions performed at
or even below room temperature. For example, alkali metal intercalation can be
performed at −78 °C in NH3(l), as in the case of the synthesis of LixMo2SbS2 (2).
Due to a large ionic radius and double charge, the diffusion of O2– can be
expected to be prohibitively slow for any reaction to occur at room temperature;
however, it has been shown that it takes only 12 h to transform SrCoO2.5 into
SrCoO3 by oxidation in a sodium hypobromite solution at 0 °C (3). These reaction
conditions are in sharp contrast with the traditional high oxygen pressure method
of SrCoO3 preparation (2000 bar at 300 °C) (4). This anomaly can be attributed to
the strong oxidation power of sodium hypobromite, as well as the higher stability
of SrCoO3 at lower temperatures. While oxygen intercalation reactions can be
performed at 0 °C, the inverse process – topotactic oxygen deintercalation –
always requires higher temperature. Oxygen deintercalation is conventionally

30

performed by reaction in flowing H2 or NH3 at temperatures above 300 °C (5, 6),
or using metal getters like Zr (7, 8). Recently, metal hydrides were introduced as
powerful low-temperature agents for topotactic oxygen deintercalation (9–12).
These reagents gave access to new metastable compounds inaccessible by
reduction with flowing H2, such as La3Ni2O6 (11), SrFeO2 (13), and CaFeO2 (14).
Solid state reductions have become possible at temperatures as low as 210 °C
due to the high chemical activity of NaH and CaH2 (15). Novel highly reduced
phases with unusual coordination polyhedrons were prepared by this topotactic
method – for example, LaBaCoO4.25 with Co(I) in square planar coordination and
Sr3LaFe1.5Co1.5O7.5 with three-coordinate Fe(II)/Co(II) (15, 16).
Solid state reduction using metal hydrides generally requires intimately grinding
the reagents, and sealing the reaction mixture in glass ampoules, and often
involves intermediate regrinding/resealing steps. Unreacted metal hydrides and
metal hydroxide byproducts must be removed by washing with methanol or
saturated NH4Cl in methanol solution (13).
One can consider performing topotactic oxygen deintercalation in the presence
of organic solvents in order to improve reaction kinetics and eliminate the necessity
of grinding/sealing operations. Reduction using metal hydrides in solvents is a
classical method in organic chemistry; nevertheless, to the best of our knowledge,
this technique has never been employed for the topotactic oxygen deintercalation
of extended oxides. While NaBH4 has been employed for metal oxides reduction
in aqueous solution (17–22) and organic solvents (23), it was used as a source of

31

H2, and reduction did not occur in a topotactic manner. Considering that topotactic
oxygen deintercalation at temperatures below 200 °C is an unexplored area, it is
possible that many new metastable transitional metal oxides can be prepared in
the 50–150 °C temperature range. With this fact in mind, a new, lower temperature
method of oxygen deintercalation is needed. In this chapter, the relative efficiency
of a novel topotactic oxygen deintercalation method that exploits the reductive
power of metal hydrides in common organic solvents is investigated. The results
obtained from this method are compared with more conventional experimental
techniques, including reduction in the solid state using metal hydrides and H2 gas,
as well as metal hydride reduction in organic solvents at ambient pressure.

32

2.2 Description of the novel solvothermal methods
La4Ni3O10 was synthesized by the sol–gel Pechini technique as described
previously (24). The final sintering was performed under oxygen flow at 1150 °C
for 200 hrs with 4 intermittent grindings.

2.2a Experimental methods for the test case of La4Ni3O10
Four different experimental reduction methods - solid state utilizing NaH or H2,
solution, and solvothermal - were used. For solid state reactions where H2 was
used as the reducing agent, 400 mg (0.449 mmol) of La 4Ni3O10 was finely ground
and then spread evenly on the bottom of an alumina boat. The materials were fired
at 200 °C, 250 °C, or 300 °C for up to 15 days under dried H2 flow. In the solid
state reaction employing NaH as a reducing agent, 400 mg (0.449 mmol) of
La4Ni3O10 was ground with NaH (13.47 mmol) under a nitrogen atmosphere in a
glovebox and then loaded into a glass ampoule and covered with a vacuum
adapter. The ampoule was removed from the glovebox, evacuated, and flame
sealed. Samples were annealed at 100–200 °C for up to 10 days with intermediate
regrinding/resealing every 24–48 hrs. Products were washed with anhydrous
methanol under a nitrogen atmosphere to remove any remaining NaH or NaOH
and then dried under vacuum for 1 h.
In the solution-based reactions, high-boiling solvents including triethylene
glycol dimethyl ether (b.p. = 216 °C), diethylene glycol dimethyl ether (b.p. = 162
°C), and ethylene glycol (b.p. = 197 °C) were used. A 200 mg (0.22 mmol) sample

33

of La4Ni3O10 with an excess of NaH (0.66–13.2 mmol) was loaded into a 100 mL
Schlenk flask in a nitrogen glovebox. The flask was then charged with 20 mL of a
high-boiling solvent and a Teflon stir bar, fitted with a reflux condenser, and placed
in a temperature-regulated (25 °C – 200 °C) oil bath. Aliquots were periodically
removed, washed with anhydrous methanol, and analyzed to determine phase
composition.
The solvothermal reactions proceeded as follows: 200 mg (0.22 mmol) of
La4Ni3O10 and NaH (2.22 mmol – 6.6 mmol) were added to a 45 mL Teflon sleeve
to avoid reaction with the steel vessel. The Teflon sleeve was then charged with
20 mL of anhydrous pentane, after which it was inserted into a steel autoclave and
treated solvothermally at either 150 °C or 200 °C for up to 10 days. The products
were washed under nitrogen with multiple aliquots of anhydrous methanol to
remove any unreacted NaH or NaOH and then dried for 1 h under vacuum to
remove any solvent. All samples were stored in a glovebox under nitrogen
atmosphere.
The samples were characterized by powder X-ray diffraction (PXD) on a Bruker
D8 Advanced with a DAVINCI design diffractometer using Cu Kα radiation. The
patterns were recorded at room temperature with a step size of 0.02° (2Θ) on a
zero background sample holder. Relative amounts of La4Ni3O10, La4Ni3O9, and
La4Ni3O8 have been estimated as a ratio of the main peak intensities.
Internal autogenous pressure of the autoclave at a given temperature was
calculated either by the Clausius-Clapeyron relation (Equation 2-1), for

34

temperatures below the critical point of the solvent, or the Van-der-Waals equation
(Equation 2-2), if the reaction temperature exceeded the critical point of the
solvent. The only solvent used at a temperature higher than its critical point was
pentane (critical point = 196 °C); therefore, the Van-der-Waals a and b constants
used for pentane were 19.26 L2bar/mol2 and 0.146 L/mol respectively, as recorded
in the 2013 CRC. Calculated pressures and enthalpies for the solvents used during
reduction are presented in Table 2-1.

ln

-∆Hvap 1 1
P1
=
( - )
P2
R
T2 T1

(P+

n2 a
V2

) (V-nb) = nRT

(2-1)

(2-2)

For the Clausius-Clapeyron relation, P and T refer to pressure in bar and
temperature in Kelvin, R is the gas law constant 8.31447 J/molK, and ∆Hvap is the
enthalpy of formation in units of J/mol. In the Van-der-Waals equation, P and T
also refer to pressure in bar and temperature in Kelvin, n is the number of moles
of solvent present, and V is the volume of the autoclave sleeve in L, while a and b
are Van-der-Waals constants that vary depending on the solvent present. The a
value describes the intermolecular attractive forces between particles while the nb
term describes the total volume per mole occupied by the gas.

35

Table 2-1. Vapor pressure, enthalpy of vaporization, and calculated pressures at
given temperatures for solvents used in solvothermal reduction reactions.
Chemical data obtained from the 2013 CRC handbook online.
Solvent

∆Hvap (kJ/mol)

P150 (atm)

P200 (atm)

-3

48.0

0.6895

2.9155

-6

63.7

0.0873

0.5916

-8

76.9

0.0068

0.0692

-1

26.2

15.65

59.738

-2

36.6

4.3158

12.956

-3

51.5

0.4579

2.1508

Pvapor (atm)

Diglyme

2.10x10

Triglyme

1.31x10

Tetraglyme

2.48x10

Pentane

5.61x10

Heptane

5.82x10

Decane

1.04x10

2.2b Results of the reduction of La4Ni3O10
The crystal structure of the Ruddlesden–Popper (RP) nickelate La4Ni3O10 can
be described as an intergrowth of rock salt and perovskite blocks (Figure 2-1). As
shown earlier, the low-temperature reduction of La4Ni3O10 proceeds through
oxygen deintercalation from the LaO layers in the perovskite blocks of the structure
and results in two distinct line phases: La4Ni3O9 and La4Ni3O8 (24). Apart from the
presence of oxygen vacancies, La4Ni3O9 is structurally similar to La4Ni3O10. On the
contrary, formation of La4Ni3O8 occurs through an oxygen deintercalation followed
by a structural change in the (LaO)2 part of the structure, resulting in a
transformation from rock salt to fluorite arrangement (Figure 2-1) (1, 24). As a
result of this structural transformation, the so-called triple layer T′-type structure is

36

formed with infinite NiO2 planes, where nickel is exclusively in square-planar
coordination. While both reduced nickelates, La4Ni3O9 and La4Ni3O8, can be
prepared by reduction of La4Ni3O10 under H2 flow at 350 °C, different
concentrations of H2 have to be utilized. La4Ni3O9 is produced when 5–10% H2 is
used, while synthesis of La4Ni3O8 requires pure H2 flow. Reaction at higher
temperatures results in complete decomposition of both phases with La2O3 and Ni
as the decomposition products.
In the following experiments, the topotactic reduction of La4Ni3O10 was chosen
as a model system. This selection was based on the existence of two possible
reduction products (La4Ni3O9 and La4Ni3O8), which can qualitatively indicate the
activity of the reducing agents. The progress of each reaction was monitored via
PXD, since the main peaks of all three nickelates can be easily distinguished in
powder diffraction patterns (1).
Reduction of La4Ni3O10 in flowing H2 (100%) at different temperatures has been
utilized as a benchmark for comparison with other techniques. The final
composition of the reactions clearly demonstrates a rapid decrease in H2 activity
with decreasing temperatures (Table 2-1). While the most reduced La4Ni3O8
nickelate can be prepared at 300 °C, only La4Ni3O9 formed at 250 °C. No reaction
was detected at 200 °C, confirmed by the sole presence of La4Ni3O10.
NaH showed high activity in the solid state reduction of La4Ni3O10 at 200 °C,
where La4Ni3O8 was the major product (Table 1, Figure 2a). While unable to
prepare a pure phase sample even after annealing for 10 days with 5 intermediate

37

grindings, a ΔT of −100 °C in comparison with the H2 reaction is rather impressive.
The activity of NaH in solid state reduction reactions declines sharply at lower
temperatures, with no reaction observed at 150 °C.
Surprisingly, the reduction of La4Ni3O10 by NaH in triglyme at 200 °C resulted
in the intermediate nickelate La4Ni3O9 as the most reduced reaction product (Table
2-1). While a pure phase sample can be prepared faster (in 4 days) and easier (no
regrinding/resealing) than in the case of the reported solid state reduction, the
inability to form the most reduced product is essentially a failure of the method.

Figure 2-1. Structure models of La4Ni3O10 (left) and La4Ni3O8 (right) with denoted
layers and structural blocks: P, perovskite; RS, rock salt; IL, infinite layer; F, fluorite.
Yellow, cyan, and red balls represent La, Ni, and O atoms, respectively.

38

At 150 °C, reduction with NaH in triglyme under ambient pressure produces
almost pure La4Ni3O9. By these results, one can see that oxygen diffusion in
nickelates is sufficient at this temperature for oxygen deintercalation to occur.
Since no reaction was observed in the solid state reduction at 150 °C, there must
be a limiting process absent in the presence of solvent.
The reaction of La4Ni3O10 with NaH is summarized below:
La4Ni3O10 + NaH → La4Ni3O9 + NaOH

(2-3)

La4Ni3O9 + NaH → La4Ni3O8 + NaOH

(2-4)

Since the eutectic temperature in the NaH–NaOH system, 450 °C (25), is much
higher than the reaction temperature under consideration, reduction without the
presence of a solvent is a purely solid state process, and involves diffusion of NaH
through NaOH(s) to the nickelate surface. Due to the relatively high solubility of
NaOH in glyme ethers, the nickelate particle surface opens up constantly for
further reaction with NaH. Even at a lower temperature of 100 °C, La4Ni3O9 forms
slowly in the reaction with NaH/diglyme.
Reduction in autoclaves was initially tried as a convenient alternative to
reaction in the presence of solvents under reflux. A large array of solvents can be
used at temperatures above their boiling points, i.e. solvothermal conditions.
Reduction of La4Ni3O10 by NaH in diglyme (bp = 162 °C) at 200 °C led to the most
reduced nickelate, La4Ni3O8, as the major phase. Therefore, at 200 °C, the

39

reducing activity of NaH in diglyme under solvothermal conditions is comparable
with that of solid NaH, and higher than NaH in triglyme at ambient pressure.

Table 2-2. Products of La4Ni3O10 reduction by different techniques. Autogenous
pressures were calculated using either the Clausius-Clapeyron equation or the van
der Waals equation depending on the critical point temperature of the solvent.
reagent/solvent

H2

reaction temperature
(reaction time),
autogenous pressure

relative content (%) of
La4Ni3O10/La4Ni3O9/La4Ni3O8
(normalized to 100% total)

300 °C (100 h)

0/0/100

250 °C (120–360 h)

0/100/0

200 °C (360 h)

no reaction

200 °C (240 h)

0/30/70

150 °C (240 h)

no reaction

200 °C (168 h)

0/100/0

150 °C (120 h)

10/90/0

200 °C (168 h), 2.9 atm

0/30/70

150 °C (240 h)

0/90/10

200 °C (240 h), 59.7 atm

0/0/100

150 °C (72 h), 15.7 atm

0/25/75

NaH

NaH/triglyme

NaH/diglyme

NaH/pentane

40

Figure 2-2. Phase composition versus reaction time during the reduction of
La4Ni3O10 with NaH at 200 °C: (a) solid state reaction, (b) solvothermal conditions.
La4Ni3O10, black squares; La4Ni3O9, red circles; La4Ni3O8, blue triangles. Relative
phase content is normalized to 100% total.

41

The solvothermal reduction performed with NaH in pentane (bp = 36.1 °C) at
200 °C resulted in pure phase La4Ni3O8. While the reaction is kinetically hindered
and takes 10 days to prepare the single phase, in this case, simplicity of procedure
and the purity of the product are advantageous over solid state reduction with NaH.
Representative PXD patterns for the reduction products prepared by four
different experimental methods are shown in Figure 2-3. Pure La4Ni3O8 without
any peaks of admixture phases can be prepared by reduction with H2 at 325 °C or
by reduction with NaH under solvothermal condition at 200 °C. Highly crystalline
products are formed in both cases. The full width at half maximum (FWHM) for the
main La4Ni3O8 reflections, (1 1 0) and (1 0 7), were 0.113(8)° and 0.131(5)° 2Θ for
the H2 reduction product, and 0.111(4)° and 0.146(3)° 2Θ for La4Ni3O8 prepared
by the solvothermal reduction method (Figure 2-4). Cell parameters were refined
as a = 3.9708(1) Å and c = 26.106(1) Å after reaction with H2 at 325 °C, and a =
3.9705(2) Å and c = 26.103(2) Å after solvothermal preparation at 200 °C.
La4Ni3O8 was also obtained as a main phase by the solvothermal reduction
with NaH in pentane at 150 °C. The solubility of both NaH and NaOH was expected
to be much lower (or zero for NaH) in nonpolar solvents like pentane than in the
polar glyme ethers; therefore, the reducing power of NaH in pentane cannot be
attributed to reagent/product solubility. There is a clear correlation between
formation of La4Ni3O8 and autogenous pressure in autoclaves (Table 2-2). The
pressure inside of the autoclaves was calculated by employing the Clausius–
Clapeyron equations. Above the critical temperature of pentane (196.7 °C), the

42

van der Waals equation was used with a = 18.99 L2 atm mol–2 and b = 0.146 L
mol–1 as reported in the 2013 CRC. The effect of pressure could be related to many
factors: changes in relative surface energy of the nickelates, an increase in
solubility of either the reagent or the byproduct, and/or changes in the oxygen
diffusion coefficients. Since the autogenous pressure was important for both polar
and nonpolar solvents, it is not likely that surface energies or NaOH and NaH
solubilities played the determining role.

43

Figure 2-3. Representative PXD patterns for the reduction products prepared by different experimental methods.
La4Ni3O9, red circles; La4Ni3O8, blue triangles.

44

Figure 2-4. Refinement of PXD data for La4Ni3O8 using the TOPAS software suite.
The sample was prepared via the solvothermal reduction route.

2.2c Discussion on the role of pressure
Nucleation of a new phase can be a limiting step for topotactic transformations
at such low reaction temperatures. In a solid–solid nucleation, a change in specific
volume that occurs must be accommodated elastically, leading to a strain energy
effect. Oxygen diffusion coefficients in ab planes can be expected to be much
higher than in the c direction in the nickelates under consideration; therefore, the

45

seeds of reduced nickelates presumably form epitaxially on any crystal facets
except (00l). In previously reported oxygen deintercalation reactions of these
nickelates, large changes in c cell parameters were observed. For example, the c
parameter decreases by 1.9% during reduction of La 4Ni3O10 to La4Ni3O9. Further
oxygen deintercalation of La4Ni3O9 to La4Ni3O8 leads to a striking 4.9% decrease
in the c parameter. Formation of strained coherent interfaces raises the free energy
of the system and can hinder the homogeneous nucleation. It is possible that the
pressure during reduction under solvothermal conditions decreases the nucleation
energy of the reduced nickelates—an effect that should be particularly important
at lower temperatures. While it is not possible to uniquely attribute the pressure
effect to increased reduction activity, the results presented in this chapter give
direction for further development of this low-temperature topotactic oxygen
deintercalation method. Higher pressure is favorable for this reaction to occur. At
the same time, the autogenous pressure becomes lower at lower temperatures;
therefore, reduction under an externally generated pressure of inert gas might
allow extending reaction to lower temperatures.

46

2.3 LaNiO2 – Square Planar Ni1+
2.3a Literature background of LaNiO2
Before proceeding into the experimental methods of the synthesis of LaNiO3, it
is important to emphasize the reasoning behind the choice of LaNiO 3 as a
reduction target to form the infinite-layer type phase LaNiO2. In LaNiO2, the
paramagnetic d9 Ni1+ metal exists in square planar configuration with the dx2-y2
singly filled and the rest of the d orbitals doubly filled (the square planar orbital
manifold diagram is presented in Figure 2-5), and is both isoelectronic and
isostrucural to Cu2+ in the high temperature cuprate superconductors with infinite
layer CuO2 (CaCuO2, Sr2CuO4, YBa2Cu3O7, etc.) (26–28). Considering that the
Cu2+ infinite-layer compounds have been extensively studied, the synthesis of
LaNiO2 would provide a simple model for the study of a monovalent transition metal
d9 ion – something essentially nonexistent in bulk materials (29, 30).

47

Figure 2-5. d orbital manifold diagram of the crystal field splitting for a transition
metal in square planar configuration.

The first literature report of the synthesis of LaNiO2 was published in 1983 by
Crespin and Gatineau at the Université Paris-Sud using a very complex vacuum
system with circulation loop to explicitly control the amount of hydrogen that was
exposed to the LaNiO3 starting material (5). According to the literature preparation,
500 mg of LaNiO3 was introduced into the recirculation loop and oxidatively
cleaned for 15 hrs at 670 K. The loop was then evacuated and filled with an exact
amount of purified and dried H2 gas and held at 397 °C until exactly 1 molar
equivalent of H2 was consumed, as measured by a monometer attached to the
system, resulting in what was believed to be pure phase LaNiO 2. However,
similarities in the K absorptions edge of the as-prepared LaNiO2 and Ni metal

48

highlight the possible Ni0 impurity in the sample, which would be invisible to x-ray
diffraction techniques due to its amorphous nature at low temperature.
Fifteen years after the first reported synthesis of LaNiO2, Hayward published a
seminal work on the reduction of solid state materials to form metastable phases
using NaH as a reducing agent (9). Hayward reacted LaNiO3 with an excess of
NaH in an evacuated glass ampoule between 165 °C and 250 °C to investigate
the reducing power of NaH. When the reaction was performed at 210 °C, a mixture
of LaNiO2, La2Ni2O5, and Ni metal was produced in a ratio of 80:15:5 by PXD peak
intensity. Although this impure sample had the highest LaNiO2:impurity ratio, the
method allowed, for the first time, the reproducible synthesis of LaNiO2 using
simple techniques and common reagents.
The most recent report of the synthesis of LaNiO2 was from the Koike group at
Tohoku University in Japan (31). Rather than using NaH as the reducing agent,
the less active CaH2 was employed; therefore, the reaction temperature had to be
increased to 400 °C before LaNiO2 was produced as the major product. The
experimental setup was identical to that proposed by Hayward - a glass ampoule
was loaded with the starting materials then sealed under vacuum and heated to
300 °C for 24 hrs. The reaction products of the reduction between 150 °C and
400 °C are shown in Figure 2-6 – focus should be drawn to the presence of Ni
metal in every reduction attempt. Just as in previous literature reports, the
presence of Ni metal, and the reason for the over-reduction was mentioned, at
most in passing. This will be further explored in the following section.

49

2.3b Reactions pathways for the formation of LaNiO2
While planning the experimental conditions for the synthesis of LaNiO 2 from
LaNiO3, it was important to understand the multiple different reaction routes that
could occur and to eliminate temperature and pressure regions that modulate the
reaction energetics down the wrong pathway. The reduction of LaNiO3 can be
explained using 4 different pathways:
2LaNiO3 + 2H2 → La2 NiO4 + Ni + 2H2 O

(2-6)

4LaNiO3 + 4H2 → La2 Ni2 O5 + 2Ni + La2 O3 + 4H2 O

(2-7)

2LaNiO3 + 3H2 → La2 O3 + 2Ni + 4H2 O

(2-8)

LaNiO3 + H2 → LaNiO2 + H2 O

(2-9)

As previously discussed, the effects of temperature on the reduction of LaNiO3
with H2 have been explored in the literature (5, 9, 31). Above 425 °C, the
brownmillerite phase La2Ni2O5 forms as oxygen is removed from the crystal lattice
of LaNiO3, resulting in Ni disproportionation to Ni0 metal (Equation 2-6) (8). As the
temperature is increased, the brownmillerite further decomposes to Ni metal and
La2O3 (Equation 2-8). At temperatures above 700 °C, the thermodynamically
stable spinel type structure La2NiO4 is formed (Equation 2-6). Lastly, under
ambient pressure, LaNiO2 is formed via reaction with H2 between 350 °C and
400 °C (Equation 2-9) (32).

50

Figure 2-6. Powder X-ray diffraction patterns of products obtained by the reaction
between LaNiO3 and CaH2 at various reaction temperatures using Cu Kα1 radiation.
Reproduced with permission from (31).

51

The effect of pressure on the reactant energetics and LaNiO2 stabilization has
been much less studied, in which the Hayward and Koike groups were the only
ones previously to investigate both pressure and temperature effects concurrently
on the reduction of LaNiO3. From the Koike group, a simple phase diagram could
be inferred relating pressure and temperature of the reaction of LaNiO3 and CaH2
(PXD patterns show in Figure 2-6) (31). Unfortunately, no real qualitative trends
could be observed due to the lack of different pressures at each temperature point.
That being said, the influence of both pressure and temperature on the reduction
of LaNiO3 to LaNiO2 were investigated.

2.3c Synthesis and characterization of LaNiO2
The parent oxygen-deficient perovskite LaNiO3 adopts a highly symmetric
geometry with 12-coordinate La3+ and corner sharing NiO6 octahedra. Topotactic
oxygen deintercalation of the parent led to the removal of the apical O 2- from the
NiO6 octahedra to form the tetragonal infinite-layer compound LaNiO3 with Ni in
the 1+ oxidation state (Figure 2-7). The Pechini method was used for the synthesis
of LaNiO3: 2.780 g (30 mmol) Ni(OH)2, 4.89 g (15 mmol) La2O3, 30 g citric acid, 5

52

Figure 2-7. Crystal structures of a) LaNiO3 and b) LaNiO2. Grey spheres represent
La3+, teal spheres represent Ni3+ (LaNiO3) or Ni1+ (LaNiO2), and red spheres are
O2-.

mL ethylene glycol, 3 mL nitric acid, and 150 mL water were added to a stirring
250 mL ceramic evaporating dish. The dish was allowed to stir at room
temperature for 48 hours after which the solution became transparent green then
was heated to 300 °C. Any remaining organic matter was burned off by firing the
dish at 600 °C for 2 hours. The ash was then ground with an agate mortar and
pestle and pelleted into 3/4" pellets. The pellets were placed into an alumina boat
and heated to 850 °C for 160 hours under flowing O 2. The resulting material – a
mixture of LaNiO3, La2O3, and NiO – was reground and pelletted, then fired for
another 80 hours. The final product was stored in a vial inside of the glovebox.
A PXD diffractogram of pure LaNiO3 prepared by the above method is
presented in Figure 2-8, with the diffraction lines denoted by magenta tickmarks.

53

TGA analysis under oxygen showed no mass gain, confirming the LaNiO3
stoichiometry compared to an oxygen deficient LaNiO3-x.

Figure 2-8. PXD pattern of the perovskite LaNiO3 after Pechini synthesis and
thermal treatment for 240 hrs at 850 °C under an O2 atmosphere. Diffraction
pattern was taken on a D2 phaser.

With the knowledge that the reduction of LaNiO3 to LaNiO2 occurs at 210 °C
with NaH (9), initial attempts focused on reduction in high boiling solvents in an
attempt to mitigate the formation of La2Ni2O5 and Ni metal as reduction products.
Reaction of LaNiO3 with 30 molar equivalents of NaH in 20 mL of triglyme (b.p.

54

216 °C) at 200 °C for 7 days under N2 led to an approximate 50:50 mixture of
LaNiO3 and La2Ni2O5 by peak intensity. Increasing the temperature to 225 °C and
swapping triglyme for tetraglyme (b.p. 275 °C) to maintain ambient pressure
conditions produced a very similar mixture of LaNiO3 and La2Ni2O5.
To understand the effects of pressure on the reaction, a solvothermal approach
was attempted. LaNiO3 and 30 equivalents of NaH were intimately ground then
loaded into a 45 mL Parr autoclave with 20 mL of anhydrous pentane. The
autoclave was sealed under N2 and placed in the glovebox at 150 °C. The reaction
was duplicated 5 times and the autoclaves were held for 24 hrs, 72 hrs, 120 hrs,
168 hrs and 240 hrs to investigate the reduction products over time – as
summarized in Table 2-3. The reduced intermediate La2Ni2O5 began to appear
after only 24 hours – a much shorter time frame compared to the 15 day analogous
solid-state reduction reaction. Up until 72 hrs, only LaNiO3 and La2Ni2O5 are
present; however, after 120 hrs diffraction peaks representing Bragg reflections
from the tetrahedral space group P4/mmm (LaNiO2) begin to appear. After 10
days, and furthermore, after 1 month, pure phase LaNiO2 is never obtained at 150
°C.

55

Table 2-3. Products of the solvothermal reaction of LaNiO3 with NaH at 150 °C.
Calculated autogenous internal pressure was approximately 10 atm.
Time (hrs)

Phase % (Main peak intensity ratio)
LaNiO3

La2Ni2O5

LaNiO2

24

80

20

0

72

65

35

0

120

15

70

15

168

0

80

20

240

0

25

75

Table 2-4. Products of the solvothermal reaction of LaNiO3 with NaH at 200 °C.
Calculated autogenous internal pressure was approximately 60 atm.
Time (hrs)

Phase % (Main peak intensity ratio)
LaNiO3

La2Ni2O5

LaNiO2

24

55

35

10

72

25

60

15

120

10

40

50

168

0

20

80

240

0

<5

> 95

56

The same reactions were repeated at 200 °C to understand the effect of
temperature on the reduction products. As seen in Table 2-4, LaNiO2 is present in
an appreciable amount after only 24 hours; while after 10 days, the product mixture
is primarily LaNiO2 (Figure 2-9) with a very small (2.5% by Rietveld refinement)
mixture of La2Ni2O5. Although only the temperature was changed between these
sets of reactions, from 150 °C to 200 °C the autogenous pressure inside of the
autoclave increased dramatically from 15 atm to 60 atm. From previous
experiments (reduction of LaNiO3 at 225 °C with tetraglyme and NaH, ambient
pressure) it can be deduced that the increase in reductive power is due to the
increase in experimental pressure rather than temperature.
When a more nucleophilic hydride was used as the reducing agent, LiAlH4, at
125 °C in tetraglyme while stirring for 48 hrs, the major reduction products were
La2Ni2O5, Ni metal, and LaNiO2. Although NaH is a better reducing agent when
described via electronegativity arguments, the solubility of LiAlH4 allows for the
solid-solid reaction mechanism to occur much faster, as will be discussed later.
However, the increase in reduction also lead to a 25% mixture of Ni metal at 125
°C. When the temperature was lowered to 100 °C, very little reduction occurred resulting in a product mixture of 90%:10% LaNiO3:La2Ni2O5.

57

Figure 2-9. PXD pattern of the perovskite LaNiO3 and the reduced infinite-layer
type phase LaNiO2 synthesized via the solvothermal reduction method at 200 °C
for 10 days.

The multi-phase refinement of LaNiO2 obtained solvothermally was especially
difficult due to the peak overlap between LaNiO2 and La2Ni2O5 (Table 2-5 presents
all of the Bragg reflections of LaNiO2 and reflections of La2Ni2O5 that appear at the
same d spacing). As seen in Figure 2-10, the peak profile was difficult to fit,
resulting in residuals at the most intense peaks (fluctuations in the grey difference

58

line below the plot). Initial refinement attempts were made using a Gaussian type
peak profile that gave a refinement Rwp = 7.82 – a poor refinement for high
symmetry compounds and good diffraction statistics. Introduction of a Lorentzian
component to the peak shape lowered the Rwp value to 2.65. A small amount of
preferred orientation was present that could not be appropriately fit, manifested in
the refinement as imperfect intensity fittings of the (h00) peaks of LaNiO2. Although
Ni was included in the plot of the Rietveld refinement (Figure 2-10 – magenta tick
marks), inclusion of the phase in the actual refinement did not statistically improve
the fit; therefore, it was only included as a reference to show that no unique
diffraction peaks from Ni are present. A compilation of cell parameters for LaNiO2
synthesized previously in the literature, as well as in this thesis, is presented in
Table 2-5. Cell parameters obtained from the refinement of LaNiO2 in this work
match closest to the data from Crespin, deemed the most pure sample of the three
in the literature. The larger c parameters in the Hayward and Koike samples may
be due to Ni in-plane disorder resulting from O non-stoichiometry (9, 33).

59

Table 2-5. Cell parameter comparison for LaNiO2. Three literature reported cell
parameter values as well as the cell parameters refined in this thesis are presented.
Source

Space
Group

a (Å)

b (Å)

c (Å)

Volume
(Å3)

Crespin

P4/mmm

3.966

3.966

3.376

53.10

Hayward

P4/mmm

3.958

3.395

3.381

52.96

Koike

P4/mmm

3.964

3.964

3.399

53.40

Blakely

P4/mmm

3.962

3.962

3.378

53.03

2.3d Reinvestigating the magnetic properties of LaNiO2
The magnetic susceptibility of LaNiO2 is presented in Figure 2.11 along with
the inverse magnetic susceptibility. The material can be easily broken down into
three different temperature regimes: a Curie-Weiss paramagnet between 5 K and
60 K, a temperature independent paramagnetic state between 60 K and 225 K,
and a second Curie-Weiss paramagnetic regime with a very large Weiss constant
above 225 K. Between 5 K and 60 K, the magnetization data was modeled with a
simple Curie-Weiss fit (C = 0.385) with Θ = -91.62 K denoting antiferromagnetically
coupled spins in the paramagnetic state. In this temperature regime the μeff is ~
0.5 β lower than the expected 1.73 β for a Ni1+ ion (Figure 2-12).

60

Figure 2-10. Rietveld refinement of LaNiO2 obtained from the solvothermal
reduction of LaNiO3 at 200 °C for 10 days using NaH. Phase analysis lead to a
LaNiO2:La2Ni2O5:Ni (metal) product ratio of 500:12.35:1.

In the temperature independent paramagnetic region (50 K – 225 K), the data
does not follow Curie-Weiss law. The most probable cause for this temperature
independent paramagnetism is a spin glass state with the low temperature spin
glass feature hidden by the Ni impurity signal.
Between 225 K and 400 K, the magnetization data follows Curie-Weiss law with
an exceptionally high Weiss temperature of -302 K – indicative of highly correlated
electrons and spins. The high effective magnetic moment may be due to a mixture

61

of superparamagnetic Ni with very small particle sizes on the order of 10 nm or
less, which would cause an increase in the overall effective moment. Considering
that the critical radius for Ni nanoparticles with a blocking temperature of 50 K is
approximately 30 nm, this assumption is appropriate (34). Using the magnetic data
as a phase purity test, which is not a guide due to the inaccuracies brought on by
the superparamagnetic Ni, while locking the effective magnetic moment to 1.73
bohr magnetons and setting the effective moment of the superparamagnetic Ni to
1.3 bohr magnetons (35), a phase ratio of 1:0.036:0.036 LaNiO2:La2Ni2O5:Ni.

62

Table 2-6. Bragg reflections of LaNiO2 and selected reflections of La2Ni2O5 that
appear at similar d spacing, thereby complicating the multiphase refinement.
LaNiO2

La2Ni2O5

d (Å)

hkl

d (Å)

hkl

3.968

010

3.898

200

3.376

001

3.455

021

2.804

110

2.789

-202

2.571

011

2.551

221

2.157

111

2.171

222

1.983

200

1.954

400

1.773

120

1.778

-402

1.708

021

1.709

-241

1.685

002

1.682

-332

1.569

121

1.563

242

1.444

112

1.447

314

1.401

220

1.409

243

1.285

022

1.281

334

1.175

131

1.175

154

63

Figure 2-11. Susceptibility (closed circles) and inverse susceptibility (open circles)
data collected on a sample of LaNiO2 under a 1000 Oe field using a SQUID
magnetometer.

The magnetic data obtained for LaNiO2 can then be compared with previous
reports in the literature (9). As seen in Figure 2-13, the general shape of the
susceptibility curve is similar for the sample made by Hayward in 1999; however,
the fit is substantially different (9). Due to the large Ni impurity and the assumption
that the Ni impurity was in a ferromagnetic state (an impossibility considering that
the Ni particle size would have to be significantly larger than the critical radius for
a superparamagnetic at 300 K of 40 nm), both fits – represented by b) and c) in

64

Figure 2-13 – required the use of a temperature independent term to achieve
reasonable values (9). The sample of LaNiO2 presented in this thesis is of a much
higher purity based on the magnetic measurements than previous reports (with the
exception of Crespin, who did not characterize the magnetic properties), leading
to the conclusion that the phase purity presented in this thesis is of higher purity
than many of the literature accounts.

Figure 2-12. Effective magnetic moment of LaNiO2 across the temperature range
studied. A horizontal line was drawn in blue at 1.73 µB – the effective magnetic
moment of an S = ½ ion (Ni1+).

65

Figure 2-13. Literature fitted molar paramagnetic susceptibility of LaNiO2 from the
Hayward group. Adapted with permission from (9). Copyright 2013 American
Chemical Society.

2.3e Comparison of solid-solid and solid-gas reaction mechanisms
The mechanism of formation of LaNiO2 can be approached differently than
discussed earlier ((2-6) – (2-9)) by looking at the possible reduction interfaces: a
solid-solid reaction (2-11) or a solid-gas reaction (2-10, 2-12).

66

LaNiO3 (s) + 2NaH(s) → 2Na(s)+ H2 (g) + LaNiO2 (s)
(2-10)
→ NaOH(s) + NaH(s) + LaNiO2 (s)
LaNiO3 (s) + NaH(s) → 2NaH(s) + 1/2O2 (g) + LaNiO2 (s)
(2-11)
→ NaOH(s) + LaNiO2 (s)
LaNiO3 (s) + NaH(s) → NaOH(s) + LaNiO2 (s)

(2-12)

The solid-gas reaction using H2 (2-10) can be ruled out from occurring below
250 °C by the TGA reduction of LaNiO3 in 10% H2, which shows that LaNiO3 is
stable towards reduction until about 250 °C (Figure 2-14). In this TGA, the first
reduction step corresponds to the formation of La2Ni2O5 (350 °C) and the second
reduction step (500 °C) represents decomposition to La2O3 and Ni. When LaNiO3
and the reductant were physically separated during the solvothermal reaction via
a teflon cup in the autoclave, very little reduction occurred until 225 °C when
La2Ni2O5 began to appear. This was anomalous until the decomposition
temperature of NaH(s) into Na(s) and H2(g) was investigated under pressure – and
shown that NaH decomposes between 210 °C and 225 °C resulting in the
reduction of LaNiO3 to La2Ni2O5 by H2. This fact also rules out the solid-gas
reaction presented in (2-10).
The second solid-gas reaction listed (2-11) relies on the release of oxygen from
the crystal lattice of LaNiO3 under a reducing atmosphere. This pathway can be
easily ruled out by the TGA of LaNiO3 under nitrogen (Figure 2-15) which shows a
negligible mass loss; therefore, oxygen is not released upon heating at 200 °C.

67

Since pathways (2-10) and (2-11) are ruled out, the solid-solid reaction mechanism
(2-12) was analyzed. If the reduction of LaNiO3 occurred via a solid-solid route, the
reaction rate would be dependent on the crystallite size of the starting material as
well as the reaction mixture intimacy.
To test this hypothesis LaNiO3 was ball-milled for 30 minutes, 1 hour, and 2
hours, then reduced under the same conditions (200 °C, 10 d, pentane, handground to mix) as previously discussed. Without ball milling, no Ni metal was
present in the PXD pattern; however, after ball-milling for any appreciable amount
of time, a large peak (~10% by intensity) of Ni metal was present, accompanied by
both La2Ni2O5 and La2O3 – essentially following the solid-solid reduction reaction
pathway articulated by (2-7). One final proof of the solid-solid reaction mechanism
is the molar dependence of NaH on the reaction rate. When a fourfold or tenfold
molar excess of NaH was used, the major phase present after 10 days was
La2Ni2O5. On the other hand, when 30 equivalents of NaH were added to the
autoclave the reaction went to completion, resulting in LaNiO2 as the major phase
under the same reaction conditions. Since the reaction rate was dependent on
both the particle size as well as the local reductant concentration, the solid-solid
reaction pathway is the most likely mechanism.

68

Figure 2-14. TGA/DSC data for the reduction of LaNiO3 under a 40:20 H2:N2
atmosphere from 25 – 1000 °C. The sample holder was loaded under N2.

Interestingly, the previous discussion on the solid-solid vs. sold-gas reaction
mechanism only applies when NaH is used as the reducing agent. Below the
decomposition temperature of 210 °C, the basicity of NaH dominates its reactivity
and the hydride ion acts as the reductant rather than H2 gas. The reduction of
LaNiO3 to LaNiO2 reported by Koike at 400 °C using CaH2 presumably follows a
solid-gas reduction pathway considering the decomposition of CaH2 to form Ca (s)
and H2 (g) begins at 240 °C and at 400 °C a large overpressure of H2 would be

69

present (36). This idea is corroborated by the TGA reduction of LaNiO3 in H2
(Figure 2-14) where the first reduction from LaNiO3 to La2Ni2O5 occurs at 275 °C
followed by a negative slope plateau at approximately 400 °C. This discussion
applies to the other reducing agents used, LiAlH4 and NaBH4, as well.

Figure 2-15. TGA/DSC data for the calorimetry experiment of LaNiO3 under a N2
atmosphere from 25 – 1000 °C. The sample holder was loaded under N2.

70

2.3f DFT study on LaNiO2 and CaCuO2 similarities
Pickett and Lee published an in-depth work in the computational study of
LaNiO2 comparing Ni1+ to Cu2+ in the infinite-layer CaCuO2 where the correlation
effects and electronic properties of both were compared (37). In comparing the
density of states (DOS) plots in Figure 2-16, it is apparent that intimate mixing of
the Cu 3d states with the O 2p states occurs just below the Fermi level in CaCuO2;
however, there is a large energy separation between the centers (3eV - 4eV) of
the Ni 3d and O 2p states in LaNiO2. The “pinning” of the Cu 3d orbitals to the O
2p orbitals below the Fermi level – causing a high degree of 2p-3d hybridization –
in the high-temperature superconducting cuprates has been heavily investigated
as the source of the superconductive properties.

71

Figure 2-16. Comparison of calculated LDA projected paramagnetic DOS LaNiO2
(upper panel) and CaCuO2 (lower panel) using DFT. Reproduced with permission
from (37). Copyright 2012 by The American Physical Society.

Apart from the isostrucural and isoelectronic similarities between LaNiO2 and
CaCuO2, the overall magnetic features and behavior of LaNiO2 are markedly
different. LaNiO2 is an apparent bad metal (lower resistivity than expected from the
band diagram due to electron-electron interactions) – the temperature dependence
of the paramagnetic susceptibility as well as the resistivity corroborate with the 3d
orbitals of Ni crossing the Fermi level – as opposed to the insulating nature of
CaCuO2. There is no observed long range ordering of the spins in LaNiO 2 from

72

either the magnetic susceptibility measurements recorded from samples
synthesized using the solvothermal method or those presented in the literature; a
refinement of neutron data by Hayward also failed to elucidate any type of spin
ordering (9, 31). This is markedly different from the essentially universal
antiferromagnetic spin ordering of the high-Tc superconducting cuprates (38–40).
The lack of long range magnetic ordering in LaNiO2 may be a result of two different
scenarios – a small amount of Ni2+ exists in the lattice, resulting in a general
formula of LaNiO2+x where x is small enough to not break the crystal symmetry, or
the exceptionally long Ni – O bonds (1.974(2) Å refined in this work, 1.979 Å
reported (9)). The presence of Ni2+ – most likely from oxidation during sample
transport or measurement preparation – could introduce a high enough number of
hole carriers to suppress long range ordering, as seen in La2-xSrxCuO4, where the
introduction of 3% by mole Cu3+ suppresses the antiferromagnetic long range
ordering, lowering the Nèel temperature from 250 K to 6 K (41). The decrease in
covalency caused by the long Ni-O bonds would lead to the weakening of the
superexchange mechanism between O – Ni – O; therefore, compared to the
exceptionally strong superexchange mechanism (resulting in the antiferromagnetic
spin ground state) in CaCuO2 where the O 2p and Cu 3d orbitals overlap very well,
no long range magnetic ordering is seen in LaNiO2.

73

2.4 Reduction of EuNiO3
All known nickelates with infinite NiO2 layers - Ln3Ni2O6, Ln4Ni3O8, and LnNiO2
(Ln = La, Nd) - contain nickel in formal oxidation states between 1.5+ and 1+.
Considering that there are no perovskite related nickelates in which the
subsequent reduction would result in a Ni2+ IL perovskite phase where only the Ni
oxidation state is reduced, we examined EuNiO3 as a possible precursor for the
formation of the infinite layer compound EuNiO2. In this case, both Eu and Ni are
expected to be reduced from the 3+ to 2+ oxidation state during topotactic oxygen
deintercalation. EuNiO2 would be the first perovskite infinite layer structure with Ni
in the 2+ oxidation state. The high-pressure synthesis presented in the following
sections was done in collaboration with Dr. Jai Prakash at MSU.

2.4a Solvothermal synthesis of EuNiO3
EuNiO3 was synthesized by a flux method using high purity Eu 2O3 (99.99%,
Strem) and NiO (99.8%, Sigma-Aldrich) as reactants and a LiCl (99%, SigmaAldrich) and KCl (ACS grade, Columbus Chemical Industries) mixture (4:6) as the
flux. EuNiO3 and NiO were preheated to remove adsorbed water at 900 °C and
400 °C, respectively. The reactants were weighed according to the stoichiometric
ratio and mixed with a LiCl/KCl flux in a N2-filled glovebox with a ratio of flux to
reactants fixed to 5:1 by weight. The reaction was carried out under an O2 pressure
of 180 bar at 400 °C for 16 days and then cooled to room temperature under 180
bar of oxygen pressure. A high pressure autoclave (Parr model 4652) was used

74

for all of the high pressure experiments. The products were washed with 0.2 M HCl
to remove the flux (LiCl/KCl) and un-reacted oxides, and then dried in air at 50 °C
to afford a black powder.
An attempt at the synthesis of EuNiO3 using lower temperatures was made by
employing a molten NaOH/KOH eutectic mixture as a solvent to dissolve the
otherwise insoluble Eu2O3. The use of fluxes at low temperatures allows for the
stabilization of kinetically metastable phases as compared to the thermodynamic
sinks favored when reacting at high temperatures. Stoichiometric amounts of
Eu2O3 and NiO were ground together under inert atmosphere, placed in an alumina
boat with a twofold excess of dry NaOH/KOH in a 50:50 mixture, and loaded into
a tube furnace in a fused silica tube. The reactants were heated to 300 °C at a
heating rate of 70 °C/min under flowing oxygen and held for 120 hrs. After the
reaction cooled to room temperature, the product mixture was washed with water
to remove the hydroxides, resulting in a mixture of the layered α-NaFeO2 type
phases NaNiO2 and NaxNiO2 (0.80 < x < 1.00), Eu2O3, and NiO. Interestingly,
synthesis of the layered phases NaxNiO2 has only been reported above 700 °C;
however, the changed energetics under molten salt conditions stabilized the
layered phase at a much lower temperature. The same reaction was attempted
solvothermally at 200 °C in the presence of 5 mL of water, resulting in starting
materials only.
To understand whether EuNiO3 can be produced at a lower oxygen pressure
than 180 bar, a 45 mL Parr autoclave was loaded with stoichiometric amounts of

75

Eu2O3 and NiO, as well as a molar equivalent of KMnO4 to generate oxygen
pressure in-situ according to the following reaction above 200 °C under ambient
pressure:
2KMnO4 → K2MnO4(s) + MnO2(s) + O2(g)

(2-13)

The autoclave was heated to 225 °C and held for 72 hours. An autogenous
pressure of 7.35 atm was generated upon decomposition of the permanganate
assuming ideal gas conditions as well as full decomposition based on the study
done by Weissman, neither of which are entirely accurate (42). Since the KMnO4
was physically separated from the Eu2O3 and NiO via a Teflon cup, no separation
techniques were required to isolate the product mixture. Only starting materials
and an unknown, low symmetry phase were present after the reaction.
Reduction of EuNiO3 was attempted through two routes: solid state reduction
using 100% H2, and a solvothermal route similar to the reduction of La 4Ni3O10 to
La4Ni3O8. For the solid state reduction reactions, EuNiO3 was finely ground and
loaded into an alumina boat, which was centered in a 3' quartz tube. Alumina boats
filled with P2O5 were placed at either end of the quartz tube to absorb any released
water and the tube was sealed with glass barbed fittings. Pure hydrogen was
passed over the sample while the reaction temperature was modulated to elucidate
a possible synthetic route to the reduction of EuNiO3 while avoiding complete
decomposition.
The solvothermal reactions were prepared in a nitrogen filled glovebox while
utmost care was taken to avoid exposing any of the metal hydride reducing agents

76

to air or oxidizing solvents. A 45 mL Teflon lined Parr 4744 autoclave was charged
with a 30 molar excess of a reducing agent (CaH2, NaH), 250 mg of finely ground
EuNiO3, and 20 mL of anhydrous pentane. The autoclave was sealed inside of the
glovebox and placed in a top-down furnace. Both time (24 hours to 240 hours) and
temperature (100 °C - 200 °C) were varied in attempts to obtain the infinite layer
EuNiO2 phase via the solvothermal reduction route.

Figure 2-17. Crystal structure of the perovskite EuNiO3 (Pbnm, a = 5.2984(3) Å,
b = 5.4572(2) Å, c = 7.5365(3) Å) obtained via solvothermal synthesis.

The synthesis of EuNiO3 at 400 °C and 1 bar of O2 pressure in a LiCl/KCl flux
was unsuccessful. The PXD pattern of the washed product showed the presence

77

of Eu2O3 and NiO reagents, indicating the necessity for high O 2 pressure during
the synthesis of this compound.
Synthesis of EuNiO3 at 400 °C and 180 bar O2 pressure lead to the formation
of the perovskite phase (Figure 2-17). The products were shiny black in color and
were found to be stable under ambient condition and towards water. PXD pattern
show very sharp diffraction peaks reflecting the good crystalline nature of the
samples. All observed reflections could be indexed on the basis of an orthorhombic
Pbnm structure. Refined lattice parameters of EuNiO3 were found to be a =
5.2984(3) Å, b = 5.4572(2) Å, and c = 7.5365(3) Å, which are in good agreement
with previous reports (43, 44).

2.4b Discussion on the reduction of EuNiO3
During TGA analysis of EuNiO3, the mass of the samples under an O2
atmosphere remains constant until 950 °C, indicating the absence of any oxygen
vacancies, as well as absence of a decomposition process (Figure 2-18). This is
in striking contrast with the high pressure conditions required for these compounds.
Therefore, annealing of EuNiO3 at 950 °C in O2 flow for 3 hours was performed.
PXD of the annealed product confirmed full EuNiO3 decomposition attributing the
TGA experiment results to the relative kinetic stability of EuNiO3 under the
experimental conditions used.
The DSC curve for EuNiO3 showed an endothermic peak at 186.5 °C,
corresponding to the metal-to-insulator transition and agreed well with previous

78

reports (43, 44). TGA analysis of EuNiO3 under a N2 atmosphere showed no
change in the sample weight until 772 °C. Above this temperature, thermal
decomposition occurred in one well-defined step which starts at 772 °C and ends
at 900 °C. The observed weight loss on the TG curve was 3.10%, very close to the
3.09% weight loss of 0.25 O2. To check the final products after decomposition of
EuNiO3 under a N2 atmosphere, the sample was subjected to heat treatment at
950 °C for 3 hrs, resulting in decomposition to Eu2O3 and NiO. TGA and PXD
results for EuNiO3 suggests the following reaction under a N2 environment:
EuNiO3 → 0.5Eu2O3 + NiO + 0.25O2

(2-14)

The TGA of EuNiO3 under a reducing atmosphere of hydrogen (10% H2/N2) is
shown in Figure 2-18. The mass of EuNiO3 remains nearly constant until 384 °C,
above which a clear weight loss step occurs starting from 384 °C and ending at
570 °C. The sample weight above 570 °C remains constant until the final
temperature (950 °C). The observed weight loss (9.04%) corresponds to a
composition of "EuNiO1.5”, indicating the formation of Eu2O3 and Ni. Data obtained
from both TGA and PXD (Figure 2-19) measurements for EuNiO3 suggests the
following chemical reaction under a H2 environment:
EuNiO3 + 1.5H2 → 0.5Eu2O3 + Ni + 1.5H2O

(2-15)

Utilization of the LiCl/KCl flux has allowed for a drastic decrease in the
temperature required to prepare EuNiO3 (from 1000 °C to 400 °C). While pressure
is needed to prepare EuNiO3, a reaction at lower temperature allows for the use of

79

less expensive and commercially available autoclaves with large internal volumes.
The developed method is expected to be suitable for the preparation of other rareearth based nickelates.
We hoped that the reduction of EuNiO3, in a similar manner to the reduction of
LaNiO3 discussed earlier, would lead to the formation of EuNiO2, a new compound
with both Eu and Ni in the 2+ oxidation state. This compound would be the first
infinite layer structure with Ni in the 2+ oxidation state. Unfortunately, the employed
experimental conditions did not indicate the possibility of such phase preparation
due to the metastability of oxygen deficient EuNiO3-x.

Figure 2-18. TGA data for EuNiO3 recorded under different atmospheres.

80

Reduction of EuNiO3 using the solid state method employing H2 as a reducing
agent was attempted in the temperature range of 200 °C to 300 °C with a firing
time of 36 hours and a heating ramp of 4 hours. Between 200 °C and 235 °C, no
decomposition products were observed and EuNiO3 was phase pure after 36 hours.
Beginning at 240 °C, the reduction of EuNiO3 resulted in the presence of small
amounts of Eu2O3 in the PXD pattern. From 260 °C - 280 °C, decomposition was
the primary reaction occurring, while small amounts of EuNiO3 were present after
36 hours. Complete decomposition at 300 °C was confirmed by the presence of
Eu2O3 and Ni without any traces of EuNiO3 in the PXD pattern.
Under solvothermal experimental conditions, decomposition to Eu2O3 and Ni
metal began to occur at 150 °C when EuNiO3 was reduced with NaH after 168
hours. Interestingly, EuNiO3 was also present in this sample, confirming the fact
that if any oxygen-deficient phases of EuNiO3-x exist, they are exceptionally
metastable and might be isolated using other, currently undeveloped, techniques.
As the solvothermal reaction temperature approached 200 °C, complete
decomposition of EuNiO3 was observed. At 100 °C, a small amount of NaOH was
present in the PXD pattern after 168 hours, indicating reduction had occurred;
however, there were no new diffraction peaks associated with either an oxygendeficient perovskite EuNiO3-x structure or the decomposition products Eu2O3 or Ni.

81

Figure 2-19. PXD spectra after decomposition of EuNiO3 at 950 °C in O2, N2, and
H2. Tickmarks refer to cubic Eu2O3 (wine, top), Ni metal (olive, middle), and Ni
(navy, bottom). All other reflections were indexed as monoclinic Eu2O3.

This result can be attributed to a small amount of decomposition of EuNiO3,
resulting in nanoparticles of Eu2O3 and Ni that are in amounts undetectable via the
D2-Phaser diffractometer employed. The autogenous pressures associated with
the solvothermal experiments at 100 °C, 150 °C and 200 °C were 12.38 atm, 15.65
atm, and 59.74 atm, respectively.

82

In conclusion, no oxygen-deficient phases of EuNiO3 were observed under
TGA conditions, solid state reduction conditions, or solvothermal conditions.

83

2.5 Chapter Conclusions
A novel solvothermal topotactic oxygen deintercalation method employing
metal hydrides and organic solvents was described and validated via reduction of
La4Ni3O10 to La4Ni3O8 at 150 °C. The relative reduction activities of different
conventional solid state reduction techniques were compared to determine the
reductive power of the proposed solvothermal method. Introduction of pressure is
favorable and leads to formation of pure La4Ni3O8 at lower temperatures than
previously reported.
Nearly pure phase LaNiO2 was synthesized using the solvothermal method
developed. The material was fully characterized, including a reinvestigation of the
magnetic properties and comparison to previously published results. According to
the Rietveld refinement and magnetic properties, the LaNiO2 sample prepared by
the solvothermal method saw an approximately fourfold increase in purity
compared to literature reports (by ratio of LaNiO2:La2Ni2O5).
Multiple attempts were made to synthesize Ni2+ in square planar configuration
in EuNiO2. The parent phase, EuNiO3, was synthesized via a simple solvothermal
method that was substantially easier than literature methods. Reduction of EuNiO3
to EuNiO2 using multiple methods was unsuccessful as shown by multiple
characterization methods.

84

2.6 Future Work on this Topic
Recent literature has presented other 3d transition metals in square planar
configuration, including Mn in BiMn7O12 (45), Co in YBaCo2O5 (15), and Fe in
SrFeO2 (46). Investigation of the reduction of perovskites and perovskite-related
phases with these metals would be an interesting continuation of this project. Due
to the exploratory type of this project and the high-risk, high-reward nature of
metastable phase synthesis, a full thesis could be written on this topic alone.

85

REFERENCES

86

REFERENCES

(1)

Blakely, C. K.; Bruno, S. R.; Poltavets, V. V. Low-temperature solvothermal
approach to the synthesis of La4Ni3O8 by topotactic oxygen deintercalation.
Inorg. Chem. 2011, 50, 6696–700.

(2)

Lappas, A.; Nuttall, C. J.; Fthenakis, Z. G.; Pomjakushin, V. Y.; Roberts, M.
A. Topotactic Intercalation of a Metallic Dense Host Matrix Chalcogenide
with Large Electron−Phonon Coupling: Crystal Structures and Electronic
Properties of LixMo2SbS2(0 ≤ x < 0.7). Chem. Mater. 2007, 19, 69–78.

(3)

Nemudry, A.; Rudolf, P.; Schoollhorn, R. Topotactic Electrochemical Redox
Reactions of the Defect Perovskite SrCoO2.5+x. 1996, 4756, 2232–2238.

(4)

Taguchi, H.; Shimada, M.; Koizumi, M. The Effect of Oxygen Vacancy on the
Magnetic Properties in the System SrCoO3-delta. J. Solid State Chem. 1979,
225, 221–225.

(5)

Crespin, M.; Levitz, P.; Gatineau, L. Reduced Forms of LaNiO3 Perovskite.
1983, 1181–1194.

(6)

Poeppelmeier, K. R.; Leonowicz, M. E.; Scanlon, J. C.; Longo, J. M.; Yelon,
B. Structure Determination of CaMnO3 and CaMnO2.5 by X-Ray and Neutron
Methods. J. Solid State Chem. 1982, 79, 71–79.

(7)

Yamamoto, T.; Kageyama, H. Hydride Reductions of Transition Metal
Oxides. Chem. Lett. 2013, 42, 946–953.

(8)

Alonso, J. A.; Martinez-Lope, M. J.; Garcia-Munoz, J. L.; Fernandez-Diaz,
M. T. A structural and magnetic study of the defect perovskite LaNiO2.5 from
high-resolution neutron diffraction data. J. Phys. Condens. Matter 1997, 9,
6417–6426.

(9)

Hayward, M. A.; Green, M. A.; Rosseinsky, M. J.; Sloan, J. Sodium Hydride
as a Powerful Reducing Agent for Topotactic Oxide Deintercalation:
Synthesis and Characterization of the Nickel(I) Oxide LaNiO2. J. Am. Chem.
Soc. 1999, 121, 8843–8854.

(10) Hayward, M. A.; Rosseinsky, M. J. Synthesis of the infinite layer Ni(I) phase
NdNiO2+x by low temperature reduction of NdNiO3 with sodium hydride. Solid
State Sci. 2003, 5, 839–850.

87

(11) Poltavets, V. V; Lokshin, K. A.; Dikmen, S.; Croft, M.; Egami, T.; Greenblatt,
M. La3Ni2O6: a new double T’-type nickelate with infinite Ni1+/2+O2 layers. J.
Am. Chem. Soc. 2006, 128, 9050–1.
(12) Poltavets, V. V.; Lokshin, K. A.; Egami, T.; Greenblatt, M. The oxygen
deficient Ruddlesden–Popper La3Ni2O7−δ (δ=0.65) phase: Structure and
properties. Mater. Res. Bull. 2006, 41, 955–960.
(13) Tsujimoto, Y.; Tassel, C.; Hayashi, N.; Watanabe, T.; Kageyama, H.;
Yoshimura, K.; Takano, M.; Ceretti, M.; Ritter, C.; Paulus, W. Infinite-layer
iron oxide with a square-planar coordination. Nature 2007, 450, 1062–5.
(14) Tassel, C.; Pruneda, J. M.; Hayashi, N.; Watanabe, T.; Kitada, A.; Tsujimoto,
Y.; Kageyama, H.; Yoshimura, K.; Takano, M.; Nishi, M.; Ohoyama, K.;
Mizumaki, M.; Kawamura, N.; Iñiguez, J.; Canadell, E. CaFeO2: a new type
of layered structure with iron in a distorted square planar coordination. J.
Am. Chem. Soc. 2009, 131, 221–9.
(15) Seddon, J.; Suard, E.; Hayward, M. A. Topotactic reduction of YBaCo 2O5
and LaBaCo2O5: square-planar Co(I) in an extended oxide. J. Am. Chem.
Soc. 2010, 132, 2802–10.
(16) Bowman, A.; Allix, M.; Pelloquin, D.; Rosseinsky, M. J. Three-coordinate
metal centers in extended transition metal oxides. J. Am. Chem. Soc. 2006,
128, 12606–7.
(17) Zhu, Y. T.; Manthiram, A. A New Route for the Synthesis of Tungsten Oxide
Bronzes. J. Solid State Chem. 1994, 110, 187–189.
(18) Tsang, C.; Lai, S. Y.; Manthiram, A. Reduction of Aqueous Na 2WO4 by
NaBH4 at Ambient Temperatures To Obtain Lower Valent Tungsten Oxides.
Inorg. Chem. 1997, 36, 2206–2210.
(19) Tsang, C.; Dananjay, A.; Kim, J.; Manthiram, A. Synthesis of Lower Valent
Molybdenum Oxides by an Ambient Temperature Reduction of Aqueous
K2MoO4 by KBH4. Inorg. Chem. 1996, 35, 504–509.
(20) Hwang, S.-J.; Choy, J.-H. Local Atomic Arrangement and Electronic
Structure of Nanocrystalline Transition Metal Oxides Determined by X-ray
Absorption Spectroscopy. J. Phys. Chem. B 2003, 107, 5791–5796.
(21) Manthiram, A.; Tsang, C. Synthesis of Amorphous MoO2+x and Its Electrode
Performance in Lithium Batteries. J. Electrochem. Soc. 1996, 143, 10–12.

88

(22) Tsang, C. F.; Manthiram, A. Synthesis of lower-valent molybdenum oxides
in aqueous solutions by reducing Na2MoO4 with NaBH4. J. Mater. Chem.
1997, 7, 1003–1006.
(23) Zhou, C. X.; Wang, Y. X.; Yang, L. Q.; Lin, J. H. Syntheses of hydrated
molybdenum bronzes by reduction of MoO3 with NaBH4. Inorg. Chem. 2001,
40, 1521–6.
(24) Poltavets, V. V; Lokshin, K. A.; Croft, M.; Mandal, T. K.; Egami, T.;
Greenblatt, M. Crystal structures of Ln4Ni3O8 (Ln = La, Nd) triple layer T’type nickelates. Inorg. Chem. 2007, 46, 10887–91.
(25) Gnanasekaran, T. Thermochemistry of binary Na–NaH and ternary Na–O–
H systems and the kinetics of reaction of hydrogen/water with liquid sodium
– a review. J. Nucl. Mater. 1999, 274, 252–272.
(26) Wen, X.-D.; Cahill, T. J.; Gerovac, N. M.; Bucknum, M. J.; Hoffmann, R.
Playing the quantum chemical slot machine: an exploration of ABX2
compounds. Inorg. Chem. 2010, 49, 249–60.
(27) AI-Mamouri, M.; Edwards, P. P.; Greaves, C.; Slaski, M. Synthesis and
superconducting properties of the strontium copper oxyfluoride Sr2CuO2F2+δ.
Nature 1994, 369, 382–384.
(28) Slakle, J. M. S. Crystal chemical substituents and doping of YBa2Cu3Ox and
related superconductors. Mater. Sci. Eng. R Reports 1998, 1–40.
(29) Schlesslnger, R. H.; Burk, D.; Lobo, A. P.; Nordlander, J. E.; Roberts, J. D.;
Gosselink, D. W.; Kaczynski, J. A.; Mckinley, V.; Beck, K. R.; Grutzner, J. B.;
Ferber, G. J.; Kiinck, R. E.; Stothers, J. B. Nickel Phsophine Complexes.
145, 5187–5189.
(30) Hayes, W.; Wilkens, J. An Investigation of the Ni+ Ion in Irradiated LiF and
NaF. Proc. R. Soc. A Math. Phys. Eng. Sci. 1964, 281, 340–365.
(31) Takamatsu, T.; Kato, M.; Noji, T.; Koike, Y. Low-temperature synthesis of
the infinite-layer compound LaNiO2 using CaH2 as reductant. Phys. C
Supercond. 2010, 470, S764–S765.
(32) Crespin, M.; Isnard, O.; Dubois, F.; Choisnet, J.; Odier, P. LaNiO: Synthesis
and structural characterization. J. Solid State Chem. 2005, 178, 1326–1334.

89

(33) Alonso, J. A.; Martínez-Lope, M. J.; García-Muñoz, J. L.; Fernández, M. T.
Crystal structure and magnetism in the defect perovskite LaNiO2.5. Phys. B
Condens. Matter 1997, 234-236, 18–19.
(34) Fonseca, F.; Goya, G.; Jardim, R.; Muccillo, R.; Carreño, N.; Longo, E.;
Leite, E. Superparamagnetism and magnetic properties of Ni nanoparticles
embedded in SiO2. Phys. Rev. B 2002, 66, 104406.
(35) Knickelbein, M. B. Nickel clusters: The influence of adsorbed CO on
magnetic moments. J. Chem. Phys. 2001, 115, 1983.
(36) Tse, J.; Klug, D.; Desgreniers, S.; Smith, J.; Flacau, R.; Liu, Z.; Hu, J.; Chen,
N.; Jiang, D. Structural phase transition in CaH2 at high pressures. Phys.
Rev. B 2007, 75, 134108.
(37) Lee, K.-W.; Pickett, W. Infinite-layer LaNiO2: Ni1+ is not Cu2+. Phys. Rev. B
2004, 70, 165109.
(38) Poltavets, V.; Greenblatt, M.; Fecher, G.; Felser, C. Electronic Properties,
Band Structure, and Fermi Surface Instabilities of Ni1+/Ni2+ Nickelate
La3Ni2O6, Isoelectronic with Superconducting Cuprates. Phys. Rev. Lett.
2009, 102, 046405.
(39) Nakayama, R.; Tanaka, Y.; Yui, H.; Yamauchi, H.; Baron, A. Q. R.;
Sasagawa, T. Preparation of infinite layer cuprate compounds
(Sr1−xLax)CuO2 and detailed phonon properties in x = 0. Mater. Sci. Eng. B
2010, 173, 41–43.
(40) Sachdev, S. Where is the quantum critical point in the cuprate
superconductors? Phys. Status Solidi 2010, 247, 537–543.
(41) Budnick, J. I.; Chamberland, B.; Yang, D. P.; Niedermayer, C.; Golnik, A.;
Recknagel, E.; Rossmanith, M.; Weidinger, A. Dependence of the NèelTemperatures of La2CuO4 on Sr-Doping Studied by Muon Spin Rotation.
Europhys. Lett. 1988, 5, 651–656.
(42) Herbstein, F. H.; Ron, G.; Weissman, A. The thermal decomposition of
potassium permanganate and related substances. Part I. Chemical aspects.
J. Chem. Soc. A Inorganic, Phys. Theor. 1971, 1821.
(43) Alonso, J. A.; Martinze-Lope, M. J.; Rasines, I. Preperation, Crystal
Structure, and Metal-to-Insulator Transition of EuNiO3. J. Solid State Chem.
1995, 120, 170–174.

90

(44) Torrance, J. B.; Lacorre, P.; Nazzal, A. I.; Ansaldo, E.; Niedermayer, C.
Systematic study of insulator-metal transitions in perovskites RNiO3
(R=Pr,Nd,Sm,Eu) due to closing of charge-transfer gap. Phys. Rev. B 1992,
45, 8209–8212.
(45) Okamoto, H.; Imamura, N.; Karppinen, M.; Yamauchi, H.; Fjellvag, H.
Square Coordinated MnO2-Units in BiMn7O12. Solid State Ionics 2010,
8709–8712.
(46) Seinberg, L.; Yamamoto, T.; Tassel, C.; Kobayashi, Y.; Hayashi, N.; Kitada,
A.; Sumida, Y.; Watanabe, T.; Nishi, M.; Ohoyama, K.; Yoshimura, K.;
Takano, M.; Paulus, W.; Kageyama, H. Fe-site substitution effect on the
structural and magnetic properties in SrFeO2. Inorg. Chem. 2011, 50, 3988–
95.

91

Chapter 3 : Soft chemistry synthesis of extended oxyfluoride materials

3.1 Survey of solid-state synthesis of inorganic oxyfluorides
Topotactic reactions, when extensive parts of the original framework are
retained and only minor structural rearrangements take place, allows for greater
control of the final product structures. Tuning the transition metal oxidation state
via cation exchange is a well-studied and exploited method in the literature (1–3).
Modification of the anion lattice primarily focuses on the preparation of oxyfluorides
by fluorination of perovskite and related structures (Dion-Jacobson, RuddlesdenPopper, etc.) where NaCaNb2O6F, PbFeO2F, Sr2LaCu2O5F, and Sr2MnGa(O,F)6
are the most studied examples (4–9). Applications of oxyfluorides include: high
temperature superconductors, electrochromic glasses, next-generation lithium ion
battery materials, and ferroelectrics (10–14). During the replacement of O2- for F1crystal structure motifs are often preserved due to the similarity of ionic radii.
Unfortunately, powder x-ray and neutron diffraction cannot differentiate between
O2- and F1- positions in the structure thus preventing the establishment of O/F
ordering (15).
Solid state synthesis methods employing pure F2 as the fluorination agent are
notoriously demanding and dangerous due to the high oxidizing potential of F2 gas
(15). Recently, fluorination methods involving less severe conditions have been
proposed. Greaves and coworkers have reported oxyfluoride preparations using
mixtures of the starting materials and transition metal fluorides MF2 (M = Cu, Zn,

92

Ni, Ag) utilized as fluorination reagents (16). These reactions result in unwanted
admixtures of the corresponding metal oxide byproducts, thus fluorinated polymers
as sources of fluorine were also studied. Slater et al investigated the potential of
poly(vinylidene fluoride) (PVDF) as a low temperature fluorinating agent in the
synthesis of Sr2TiO3F2 as well as Ca2CuO2F2 from the parent materials Sr2TiO4
and Ca2CuO3, respectively, in a non-oxidative fashion (17, 18). Oxidative fluorine
intercalation has also been performed on the Ln1.2Sr1.8Mn2O7 (Ln = La, Pr, Nd, Sm,
Eu, and Gd) system by decomposing PVDF at 400 °C to form Ln 1.2Sr1.8Mn2O7F1
(19). XeF2 has seen more use in the recent literature as a powerful source of
fluorine without the production of admixture MxOy or MF2 byproducts. Due to the
relatively high vapor pressure of XeF2 , the temperatures required to perform
fluorination using this reagent are in the 100 °C - 300 °C range, allowing for
topotactic intercalation of fluorine into anion vacancies (20).
There are multiple examples of oxygen/fluorine ordering in the literature
including Sr3-xBaxAlO4F, Sr7Mn4O13F2, and Nb3O5F5 (21, 22). When trying to
achieve anion ordering during synthesis, it is important to perform anion
intercalation at as low a temperature as possible to avoid thermal randomization
of the anions (21, 23, 24).
The focus of this chapter is on the preparation of oxyfluoride materials by
multiple different synthetic methods including a multistep fluorination method
employing XeF2 and a novel solvothermal method. Multiple proof-of-principle

93

cases are presented to demonstrate the fluorinating power of each of the
approaches.

3.2 Novel multistep oxyfluoride synthesis procedure
In this section, a novel low temperature fluorination method employing XeF2 for
the synthesis of the oxyfluoride SrFeO2F via the recently discovered infinite layer
compound SrFeO2 is presented. Characterization of SrFeO2F was accomplished
using Rietveld refinement of synchrotron powder diffraction data, Mössbauer
spectroscopy, AC and DC magnetometry, and heat capacity measurements. The
prepared SrFeO2F does not have long range O/F ordering but demonstrates
differences in both magnetic properties and Mössbauer spectrum in comparison
with the earlier reported disordered SrFeO2F, lending credence to the idea that the
chosen synthetic approach has an effect on short range O/F ordering and,
consequently, on the properties of the compound (7).
3.2a Multistep synthetic route and experimental methods
All reagents were of analytical grade purity or higher and were used without
further purification. SrCO3 (99.9%), Fe2O3 (99.99%), and XeF2 (99.9%) were
obtained from Alfa Aesar. CaH2 (99%) was purchased from Sigma Aldrich packed
as a suspension in mineral oil. The CaH2 was washed with anhydrous pentane
under nitrogen to remove the mineral oil, resulting in a fine white powder. Utmost
care should be taken to avoid exposing the XeF2 to moisture, as hazardous HF will
form.

94

Figure 3-1. Multistep synthetic route for the preparation of the SrFeO2F oxyfluoride. The previously known literature
method using PVDF at 400 °C is also highlighted (6).

95

The parent cubic perovskite SrFeO3-x was synthesized by the reaction of SrCO3
and Fe2O3. 0.739 g (5 mmol) of SrCO3 was coarsely mixed with 0.399 g (2.5 mmol)
Fe2O3 then ball milled in agate jars for 30 minutes with four 6 mm agate balls using
a Fritsch Planetary ball mill. The powder was then pelletted in 3/4" pellets under 5
tons pressure and fired at 1000 °C in air for 48 hours. SrFeO 2 was prepared via a
modified literature method (25), described below, where large quantities of sample
were produced per batch to avoid batch inconsistencies. 4.75 g (25 mmol) of
previously synthesized SrFeO3−x was intimately ground with 4.20 g (100 mmol) of
CaH2 in a nitrogen filled glovebox. The precursor mixture was loaded into an
ampoule then connected to a silicon oil bubbler, removed from the glovebox and
lowered through the top of a muffle furnace. The material was fired at 280 °C for
48 hours with no intermittent grindings. The ampoule was connected to a silicon
oil bubbler to release the hydrogen pressure. After completion of the reaction, the
product was washed with four 100 mL aliquots of a saturated NH4Cl in methanol
solution to remove any remaining CaH2 as well as the CaO reduction byproduct.

96

Figure 3-2. Comparison of PXD patterns of the SrFeO3-x perovskite (bottom
pattern), the SrFeO2 infinite-layer compound (middle), and the SrFeO2F
oxyfluoride (top).

The oxyfluoride phase SrFeO2F was prepared via the reaction of the infinitelayer SrFeO2 with XeF2. 500 mg (2.87 mmol) of SrFeO2 was weighed out in a
glovebox and loaded into a teflon sleeve along with 289 mg (1.70 mmol) of XeF2,
previously weighed under N2 using a teflon spoon and a Ni weighing dish. The
sleeve was sealed in a steel autoclave then removed from the glovebox and placed
in a furnace. The mixture was heated for 48 hours at 150 °C, after which the

97

product was stored under N2 due to the extremely moisture sensitive nature of the
oxyfluoride.

3.2b Synthesis and characterization of SrFeO2F
Powder X-ray Diffraction (PXD) patterns were obtained on a Bruker D8
Advanced New diffractometer using Cu radiation with a Kα1 wavelength of 1.5418
Å. The Synchrotron Powder x-ray Diffraction (SPXD) pattern was collected on the
Argonne National Lab Synchrotron BM-11 beam line using a 0.8 mm kapton
capillary. The sample was first sealed in a 0.6 mm glass capillary then placed in
the kapton capillary and held with wax. The synchrotron source wavelength was
0.373822 Å.
Rietveld refinement and structure analysis were completed using the GSAS
software package. Peak positions, unit cell parameters, background parameters,
and profile parameters were refined one by one, followed by a refinement of all
parameters.
Average Fe oxidation state was determined by iodometric titration. The
iodometric titration procedure has been described previously (26). Care was taken
to avoid exposure to oxygen by blowing nitrogen over the flask. Fluorine content
was investigated by ion selective electrode (ISE) technique. ISE measurements
were performed using a Hanna Instruments HI 4110 Fluoride Combination ISE
electrode with an ion-activity buffer (TISAB II) by the calibration curve method.
Careful attention was paid to both the temperature of the material and the electrode

98

equilibration time before recording the voltage response. Inductively coupled
plasma (ICP) spectroscopy was performed on a Vista-MPX CCD Simultaneous
ICP-OES instrument (Varian Inc.).
The parent oxygen-deficient perovskite SrFeO3-x adopts a highly symmetric
cubic geometry with 12-coordinate Sr2+ and corner sharing FeO6 octahedra.
Topotactic oxygen deintercalation of the parent led to the removal of the apical O 2from the FeO6 octahedra to form the tetragonal infinite-layer compound SrFeO2.
While the reaction temperature and time used for the preparation of SrFeO 2
were the same as previously reported by Tsujimoto et al. (25), instead of using a
sealed Pyrex tube the ampoule was connected to a bubbler filled with silicon oil.
Ca(OH)2 formed during reaction with the oxide phase further reacted with CaH 2,
leading to the formation of CaO and H2. Utilizing a bubbler prevents H2 pressure
build up and allows for scaling up of the synthesis in a safe manner; therefore,
multiple grams of SrFeO2 can be prepared in one batch instead of a few hundred
milligrams as previously reported. From this it can be concluded that H 2 pressure
is not required for SrFeO2 formation as the reduction process occurs through the
direct reaction of 2 solids.
Fluorination, the third step in the multistep synthesis route, resulted in the
oxidative insertion of F1- to give the final cubic oxyfluoride phase SrFeO2F. The
reaction temperature was critical in obtaining phase pure oxyfluoride. When fired
at 250 °C, a small admixture of SrF2 was always present regardless of the excess
amount of XeF2 used, which manifested as broad SrF2 reflections in the powder x-

99

ray diffraction pattern due the poor crystallinity of the admixture. Reacting the
starting materials at 150 °C resulted in SrFeO2F with no traces of SrF2. Comparing
this result with the literature data where SrFeO2F was prepared at 400 °C by
fluorination of SrFeO3-x with poly(vinylidene fluoride) we can conclude that the
appearance of SrF2 at 250 °C is not related to the insufficient kinetic stability of the
metastable SrFeO2F product but rather to the energetic ease of SrF2 formation
when the SrFeO2 precursor is used.
PXD patterns indicating pure phases of SrFeO3−x, SrFeO2, and SrFeO2F are
compared in Figure 3-2. The SrFeO2F pattern was readily refined in the cubic
space group Pm3̅ m (a = 3.955 Å), similar to the parent perovskite phase SrFeO 3x,

with an Rwp value of 1.68 %. Observed, calculated, and difference plots for the

Rietveld refinement of SrFeO2F are shown in Figure 3-3. An anion disordered
structure was used in the refinement due to the similar scattering factors of O2- and
F1-. Refinement of the anion sites led to complete occupancy.

100

Figure 3-3. Rietveld refinement of the synchrotron powder x-ray diffraction data of
SrFeO2F showing the observed pattern (black crosses), fitted pattern (red trace),
difference curve (gray trace, below), and Bragg peak positions (magenta
tickmarks, below).

Unit cell parameters for SrFeO3-x, SrFeO2, and SrFeO2F as well as the
structural data for SrFeO2F are presented in Tables 3-1 and 3-2, respectively.
Large thermal parameters for the anion position in SrFeO 2F might indicate a
minute shift of O/F atoms from the special position. However, refinements with
anions in a lower symmetry position did not lead to lower goodness-of-fit
parameters.

101

Due to the presence of trivalent fluorine-complexing ions (Fe3+), the use of a
fluoride ion-selective electrode resulted in underestimated F content values even
when an ion-activity buffer (TISAB II) was added.
Because fluorination reactions can lead to fluorine insertion into interstitial
positions as well as fluorine-for-oxygen substitution, prepared oxyfluorides can be
represented by the general formula SrFeOxFy. Due to the presence of two
unknown variables, Ox and Fy, two different compositional analysis methods must
be corroborated to deduce the actual composition. Using ICP spectroscopy, the
molar mass of the oxyfluoride, 194.5 g/mol, was quantitatively deduced from the
Sr and Fe content and known initial sample mass assuming the presence of 1 Sr
and 1 Fe ion per oxyfluoride formula unit, as in the initial SrFeO3 and intermediate
SrFeO2 compounds. This assumption is assured by the fact that no impurities
containing Sr or Fe were found in the synchrotron data. The average Fe oxidation
state (2.99+) was determined by iodometric titration. As a result of these two
experiments, both Ox and Fy can be solved, resulting in a stoichiometry of
Sr1.00(1)Fe1.00(1)O2.01(1)F0.99(1).
The a parameter increased to 3.955 Å in the oxyfluoride phase relative to 3.851
Å in the parent perovskite SrFeO3-x due to the size difference between Fe3+ and
Fe4+. Direct establishment of O/F ordering by diffraction techniques is not possible
due to the similar X-ray form factors and neutron coherent scattering lengths (O 2= 5.803 fm, F1- = 5.654 fm) for these anions (27). However, there is a subtle
difference in the ionic radii in a 6 coordinate environment: O2- = 1.26 Å while F1- =

102

1.19 Å (28); therefore, layered O/F ordering in SrFeO2F might lead to a tetragonal
distortion with different values for the a and c cell parameters. This hypothesis was
confirmed by DFT calculations as discussed above. No tetragonal distortion was
observed for SrFeO2F by PXD or by SPXD. Therefore, there is no long range
layered anion ordering in the prepared oxyfluoride and an anion disordered
structure was formed in spite of using a precursor with ordered O vacancies
(SrFeO2). As the Sr/Fe cationic sublattice was preserved during fluorination, this
reaction can be considered topotactic; however, O/F mixing in SrFeO2F over the
same anionic crystallographic position indicates that both anions are mobile during
the fluorination procedure. Mobility of the anion sublattice in SrFeO 2 has been
previously investigated, where during O intercalation the intermediate Sr2Fe2O5
brownmillerite phase forms (29, 30). The difference in O vacancy ordering in
SrFeO2 and Sr2Fe2O5 requires substantial anion lattice rearrangement during the
reaction.

Table 3-1. Unit cell parameters for SrFeO3-x, SrFeO2, and SrFeO2F. Previously
reported literature cell parameters for SrFeO2F: Pm3̅ m, a = 3.956 Å (6, 7).
Compound

Space
Group

Unit Cell
Parameters (Å)

Volume (Å3)

SrFeO3-x

Pm3̅ m

a = 3.851

55.11

SrFeO2

P4/mmm

a = 3.991
c = 3.478

55.35

SrFeO2F

Pm3̅ m

a = 3.955(1)

61.86(4)

103

Table 3-2. Atomic positions obtained from the Reitveld refinement of synchrotron
diffraction data for SrFeO2F with refined occupancies.
Atom

Wyckoff
Position

x

y

z

Occ.

102 Uiso (Å2)

Sr

1b

0.5

0.5

0.5

1

0.03(2)

Fe

1a

0

0

0

1

0.13(4)

O/F

3d

0

0

0.5

0.66 / 0.33

0.96(2)

Collected thermograms (Figure 3-4) for SrFeO2F under both O2 and N2 in the
range of 300-1200 K were analyzed based on total weight loss due to the
complexity of the decomposition pathways.
Under an O2 atmosphere, the final mass was 99.47%, giving a total weight loss
of 0.53% as described in reaction (3-1) below, which is dominated by the formation
of SrF2 and the stable parent perovskite SrFeO3−x. When the oxyfluoride was
heated under a reducing N2 environment, a mass loss of 2.09% was observed,
pointing again to the formation of SrF2 as well as the more reduced iron oxide
Fe3O4 and the stable brownmillerite phase Sr2Fe2O5 as shown in reaction (3-2)
below.
2 SrFeO2F → SrFeO3-x + SrF2 + 0.50 Fe2O3

(3-1)

SrFeO2F → 0.25 Sr2Fe2O5 + 0.50 SrF2 + 0.16 Fe3O4 + 0.11 O2

(3-2)

The mass gain under an O2 atmosphere seen from 1000 K to 1100 K followed
by an equal mass loss to give the final mixture is attributed to a higher oxygen

104

activity at 1000 K, leading to fewer oxygen vacancies (lower x value) in SrFeO 3−x.
At 1100 K the oxygen activity decreases and the more stable SrFeO3−x (0.25 > x >
0.15) perovskite forms (31). The majority of the mass loss under an N2 atmosphere
is due to the loss of oxygen at higher temperatures. TGA decomposition products
were confirmed by PXD.
An endothermic peak in the DSC curve under an N2 atmosphere at 510 K might
indicate a local O/F order-disorder transition. This peak is absent in the O2 DSC
curve. The small number of O vacancies formed in SrFeO 2F on heating under N2
flow is apparently important for O/F rearrangement to occur.

3.2c Full magnetic properties investigation
The temperature dependence of DC magnetic susceptibility was measured on
powder samples with a Quantum Design SQUID magnetometer at H = 1000 Oe in
the temperature range 4 - 400 K. Samples were loaded into gelatin capsules under
N2, placed into a plastic sample holder and cooled to 4 K under zero magnetic
field. All measurements were performed by warming the samples in the applied
field after cooling to 4 K in zero field (ZFC, zero field cooling) and by cooling the
samples in the applied measuring field (FC, field cooling).
The field dependence of DC magnetic susceptibility, AC magnetization, and
specific heat measurements were performed on a powder sample with a Quantum
Design PPMS. AC magnetic susceptibility data was collected between 4 K and 310
K by varying the frequency from 50 Hz to 10000 Hz.

105

Figure 3-4. Thermogravimetric analysis of the fluorinated product SrFeO2F under an O2 and N2 atmospheres.
Positive slope in the DSC curve represents endothermic signals, while negative slope represents exothermic signals.

106

The magnetic susceptibility and specific heat measurement results are shown
in Figure 4. The small divergence between ZFC and FC curves at 300 K (Figure
3-5a) is due to SQUID instrument measurement noise as confirmed by the M vs.
H measurement on a PPMS system at 300 K (Figure 3-5b). No magnetization
frequency dependence was observed in the AC susceptibility data above 60 K
(Figure 3-5c, 3-5d). Finally, the specific heat data did not reveal any convincing
transitions in the temperature range studied (Figure 3-5a).
A cusp in the AC magnetization at about 60 K and dependence of the cusp
position on the frequency of the AC susceptibility measurements are characteristic
features of a spin-glass like transition (Figure 3-5c, 3-5d) (32–34). The divergence
of the ZFC and FC susceptibility curves below 60 K (Figure 3-5a) and non-linear
magnetization on magnetic field dependence at 10 K (Figure 3-5b) are in
agreement with the spin-glass like state as well. Above 60 K the magnetic
properties presented here (Figure 3-4a, 3-4b) are drastically different from reported
for SrFeO2F prepared directly from SrFeO3 at 400 °C (7). For the literature sample,
a spin-glass transition at 300 K was reported based on the bifurcation of ZFC and
FC data and a nonlinear change of magnetization vs. magnetic field at 300 K (7).
In our sample, there was no magnetization frequency dependence in the AC
magnetization data between 310 K and 60 K which eliminates the possibility of a
spin-glass state. Considering that the SrFeO2F cell parameters and composition
reported here and in (6, 7) are virtually identical, it is unexpected to see such
distinct magnetic properties. The difference in properties can be related to

107

differences in O/F ordering due to both the preparation approaches (direct vs.
multistep) and synthesis temperature (400 °C vs. 150 °C).

3.2d Local coordination control using 57Fe Mössbauer
57Fe

Mössbauer spectroscopy was carried out utilizing a

57Co-source

in a

rhodium matrix at room temperature. The measured spectrum was evaluated using
Fit;o) - a Mössbauer spectrum fitting program (35).
Local Fe environments in SrFeO2F were studied by Mössbauer spectroscopy.
The room temperature 57Fe Mössbauer spectrum of SrFeO2F is shown on Figure
3-6 and Mössbauer spectrum fitting parameters are given in Table 3-3. It was
necessary to introduce 3 Fe sites in the refinement to fit the spectrum (2= 3.7).
Unlike the previously reported oxyfluoride PbFeO2F (36), no asymmetry of the
main sextet peaks was observed in our case. Nevertheless, attempts to add an
extra Fe site in order to use two Fe sites with close quadrupole splitting parameters
to fit the main sextets similarly to Inaguma et al. were undertaken (36). Such
attempts did not lead to a better fit and the refined area for an extra Fe site was
below 2%, which is within error.

108

Figure 3-5. DC magnetization at H = 1000 Oe (a), DC field dependence (b), AC magnetic susceptibility (c, d) and
specific heat measurement (a) results for SrFeO2F collected on a Quantum Design PPMS system.

109

Table 3-3. SrFeO2F Mössbauer spectrum fit parameters obtained using the Fit:o)
software package (35).
Quadrupole
Hyperfine
Line Width Area  2
(%)
Splitting  0.02 Magnetic Field
 0.02
(mm/s)
(mm/s)
 0.08 (T)

Site

Isomer
Shift  0.02
(mm/s)

Fe1

0.42

0.11

51.43

0.79

41

Fe2

0.33

0.04

49.29

0.80

46

Fe3

0.42

1.04

-

0.81

13

Figure 3-6. Fit (black line) of room temperature

57Fe

Mössbauer spectrum of

SrFeO2F with denoted contributions from Fe sites: Fe1 (blue line), Fe2 (red line),
Fe3 (green line).

110

Mössbauer spectroscopy was utilized to study the local Fe environments, O/F
ordering, and Fe spin states in SrFeO2F prepared by the multistep procedure
reported here. A slightly broader range of isomer shifts (0.33 – 0.42 mm/s) for Fe
positions was found in our sample in comparison with the data (0.34 – 0.39 mm/s)
of Berry (7). The simultaneous presence of Fe sites with zero and non-zero
hyperfine fields in the single phase O/F disordered SrFeO2F sample was reported
earlier between 200 and 317 K (7). All isomer shift values for Fe sites in our fit can
be assigned to Fe3+ (HS) states. For the spherical Fe3+ (3d5) ion, quadrupole
splitting (QS) can only arise due to non-cubic symmetry of the lattice contribution
to the electric field gradient (EFG) (36). Random O/F distribution in Fe oxyfluorides
should lead on average to FeO4F2 polyhedra with 2 possible O/F arrangements:
trans - FeO4F2 with F1- ions at opposite corners of the octahedron and cis - FeO4F2
arrangement with 2 F1- at adjacent octahedron corners. The value of QS for the
non-magnetic Fe3 site (1.04 mm/s) is similar to that (0.96 mm/s) reported for
disordered SrFeO2F (7). It is not possible to uniquely deduce the coordination
nature of the nonmagnetic Fe3 site which can consist of cis-, trans- or, most
probably, a mixture of both anion arrangements. In the presence of a magnetic
hyperfine field (BHF), as is observed for the Fe1 and Fe2 sites, the QS value
depends on the angle () between the principal axis of the EFG (Vzz) and the
direction of BHF:  = ½ 0 (3cos2 - 1), where  and 0 are QS value with and
without a magnetic hyperfine field, respectively. The expected values of QS for
magnetic sites can be estimated following the excellent discussion of Berry (7).

111

Accepting 0 equal to 1.04 mm/s and  values of 45° and 90°, the calculated QS
values for the cis-arrangement are 0.26 mm/s and -0.52 mm/s. The 45°  value
corresponds to the cis-configurations with one of the F1- ion situated on the z axis,
while the 90°  value is for cis-configuration with 2 F1- ions in the equatorial
positions. Similarly, for the trans-FeO4F2 0 equal to -1.04 mm/s (an opposite QS
sign is expected for cis- and trans- configurations) and  values of 0° and 90°, the
calculated QS values are -1.04 mm/s and 0.52 mm/s. The calculated QS values
for the trans-configuration are much larger than the observed ones for the Fe1 and
Fe2 sites. After comparing the absolute values of calculated QS with the fit results
(Table 3-3), the cis-configuration can uniquely be assigned to the Fe1 and Fe2
sites. Even more specifically, the cis-configurations with one of the F1- anions on
Vzz can be selected. The difference between the calculated QS value of 0.26 mm/s
for one of the cis-configurations and the experimentally observed ones (~0.1
mm/s) for Fe1 and Fe2 can be attributed to subtle octahedra distortion resulting in
the angle between Vzz and BHF of ~51° instead of the ideal value of 45°.
Considering the magnetic character of Fe1 and Fe2 sites, as well as an absence
of the divergence of the ZFC and FC susceptibility and linear character of magnetic
field dependence at room temperature, we propose a cluster of spins state above
60 K. The reverse magnetic susceptibility vs. temperature dependence is linear
between 60 and 400 K. The effective magnetic moment µexp per Fe ion calculated
from the Curie-Weiss model was 3.44 µB, which is substantially smaller than the
expected spin only values of 5.92 µB (S = 5/2) for Fe3+ (d5) in high spin

112

configuration. The lower effective magnetic moment is in agreement the proposed
spin cluster model. The presence of some amount of FeO5F and FeO3F3 polyhedra
can be assumed in the sample. Attempts to take into accounts these configurations
by adding extra sites to the fit led to peak areas below 2%, i.e. below the fit error,
and thus were excluded from the fit.
As was discussed in great detail by Berry et al. (7) for a disordered SrFeO2F
phase, the area ratio for cis- and trans- configurations is expected to be 67% :
33%. A close ratio (65% : 35%) was indeed observed experimentally in their case.
In our case, even if we ascribe the Fe3 site completely to trans-configuration, the
cis- to trans- peak area is 81% : 19%; much higher than for disordered oxyfluorides.
The ratio may be even higher since the Fe3 site can be a mixture of cis- and transconfigurations. Thus, the multistep low temperature preparation technique used
resulted in the predominance of cis fluorine configuration in the FeO4F2 polyhedra.

3.2e DFT structure optimization for ordered SrFeO2F
The structure optimization of the ordered SrFeO2F was performed using the
Wien2K 11.1 computational package (37). The tetragonal structural model with
layered O and F ordering was generated and total energy was computed for
structures with varied a/c ratios in order to determine the most stable structure
model. Beginning from the cell parameters for the disordered phase (a = 3.955 Å,
a/c = 1), the a/c parameter ratio was varied by 1% in a grid of 20x20 structures
with the original structure held at the center.

113

Figure 3-7. Left: Potential energy surface of varied a/c structures. The minimum position of the surface is highlighted
by the white circle. Right: Contour plot zoomed in around the minimum energy extracted from the potential surface
shown on the left. The dark blue area represents the minimum energy.

114

The energies of the structures were each calculated with an RKmax of 8, a Gmax
of 15, 5000 k-points, a core separation of -7.0 Ry and a muffin tin sphere reduction
of 10% to allow for expansion and contraction of the spheres during optimization.
Relaxation of the ordered structure of SrFeO2F was also performed using the
Vienna Ab-initio Simulation Package (VASP) 5.2.12 with the projector augmented
wave method and the Perdew-Burke-Enzerholf '96 functional utilizing a conjugategradient algorithm for the ionic relaxation (38–40). The internal positions and cell
parameters were allowed to relax. A gamma point centered pack of 1000 k-points
were used in each calculation with an energy cut-off (Encut) value of 400 eV.
The results of the structure optimization of the ordered SrFeO2F using the
Wien2K package is shown in Figure 3-7 where the energies of structures with
different a and c parameters are plotted. The figure clearly shows a local minimum
at a = 3.775 Å and c = 4.120 Å. VASP was used to verify that the large tetragonal
distortion should be expected for the ordered SrFeO2F with layered O/F ordering.
The cell parameters for the VASP relaxed structure are a = 3.786 Å and c = 4.086
Å, validating the expected tetragonal distortion.
Considering similar ionic radii for O2- and F1- ions, a large theoretically predicted
cell parameter difference is rather unexpected for the Fe 3+ (d5) compound,
especially since half-filled d electronic orbitals possess spherical symmetry around
the atomic nucleus (34). At the same time, an apparent failure of the effective ionic
radii concept in oxyfluorides was shown experimentally previously for LaOF. In
LaOF the La – O distance (2.42 Å) is substantially shorter than for La – F (2.60 Å)

115

(42). Interestingly, the predicted distance difference between Fe – O and Fe – F
bonds in the ordered SrFeO2F (0.17 Å) is similar to La – O and La – F distance
difference in LaOF with metal – fluorine distances longer in both cases.

3.2f Concluding remarks for SrFeO2F
A low temperature multi-step synthesis of the SrFeO2F oxyfluoride using XeF2
was presented. According to Rietveld refinement of SPXD data, the oxyfluoride
crystallizes in a cubic perovskite structure (Pm3̅ m space group, a = 3.955 Å)
without long range O/F ordering. A large thermal parameter for the O/F
crystallographic position indicated local disorder and differences in Fe-O and FeF bond lengths. Unexpectedly, for SrFeO2F with layered O/F ordering, DFT
calculations predict Fe-F bonds significantly longer than Fe-O ones. When
synthesized at 150 °C using the method described herein, a preference for cis
fluorine configuration in local FeO4F2 polyhedra was extrapolated from Mössbauer
spectroscopy data, demonstrating the effects of experimental conditions on local
Fe coordination.

116

3.3 Synthesis of complex oxyfluorides La4Ni3O8Fx (x = 1,2)
The infinite NiO2 square-planar layer is a very rare structural element in solids
in spite of the well-known stability of Ni2+ cations in square-planar coordination. All
nickelates with infinite NiO2 layers are metastable and can be prepared only by low
temperature reduction of the parent phases (43–48). Further information on these
phases can also be found in Chapter 2 of this thesis. Such nickelates belong to the
so-called T-type Lnn+1NinO2n+2 (Ln = La,Nd; n = 2,3,) homologous series (49).The
physical properties of the T-type nickelates are of great interest due to the 3d9/3d8
electronic configuration of Ni1+/Ni2+, the same as that of Cu2+/Cu3+ in the hightemperature superconductors (50, 51). The n = 2 homologue, La3Ni2O6,
demonstrates multiply degenerate magnetic ground state while a spin density
wave transition was observed for the n = 3 phase, La4Ni3O8. (46, 52, 53).
Depending on the homologue synthesized, the Ni oxidation state varies between
1+ and 1.5+. Synthesizing infinite layer phases with higher Ni oxidation states than
1.5+ is difficult due to the limited solubility range of alkaline earth metals in the A
position; however, as discussed in Chapter 2 of this thesis, Eu2+ in EuNiO2 may be
a viable route to synthesizing an infinite-layer phase with Ni in the 2+ oxidation
state. As the direct synthesis of any of the metastable phases presented in this
section is thermodynamically hindered, a multistep route was taken in the
synthesis of the oxyfluorides La4Ni3O8Fx (x = 1,2). The experimental conditions
were carefully optimized to increase the purity an crystallinity of each phase in the
series.

117

Tying in with the general theme of this thesis, the ability to topotactically
intercalate F1- into certain positions in the crystal structure of La4Ni3O8 would allow
for the establishment of a structure-function relationship based on F1stoichiometry. As the oxidation of La4Ni3O8 to La4Ni3O10 is easily performed under
an O2 atmosphere at relatively low temperature, fluorine should also be easily
intercalated into the vacancies. However, oxyfluorides are often only kinetically
stable and decompose to the thermodynamically stable phases as temperature is
increased. Indeed, the reaction of La4Ni3O8 with XeF2 at 350 C led to the formation
of LaOF and an amorphous mixture, demonstrating that the oxyfluoride LaOF is
the more stable product at that temperature. This work was done in collaboration
with Dr. Artem Abakumov (electron diffraction), Dr. Joke Hadermann (electron
diffraction), Dr. Jason Hodges (neutron diffraction), and Dr. Matthew Suchomel
(synchrotron diffraction).

3.3a Solid state synthesis using XeF2
All fluorinated samples were treated with the utmost care to limit exposure to
moisture due to the hazards related to HF. All sample manipulations were done
under an N2 atmosphere in a glovebox. XeF2 was weighed using a nickel boat and
teflon spoon to avoid oxidation of equipment in the glovebox.

118

Figure 3-8. Topotactic transformation of La4Ni3O8 to La4Ni3O8F2. The structure
shown for La4Ni3O8F2 depicts O/F ordering. Colored balls represent the following
atoms: yellow – La, red – O, teal – Ni, green – F.

La2O3 (Alfa Aesar, 99.9%) and Ni(OH)2 (Sigma Aldrich, 99.9%) were both
stored under nitrogen to avoid water absorption. XeF2 (Alfa Aesar, 98%) was used
as received without further purification.

119

La4Ni3O10 was synthesized by the Pechini method: 7.310 g (22.43 mmol) La 2O3
was mixed with 3.12 g (33.6 mmol) Ni(OH)2 in a 200 mL ceramic bowl. An excess
of citric acid, 5 mL of ethylene glycol, and 1 mL of concentrated HNO3 were added.
The bowl was filled with deionized water and allowed to stir for 48 hrs. After a gel
had formed, the sample was fired at 600 °C for 2 hours to remove any remnant
organics. The resulting black mixture was intimately ground, pelleted into 3/4"
pellets then fired at 1150 °C for 468 hours with five intermittent grindings.
3.80 g (4.26 mmol) of finely ground La4Ni3O10 was loaded into an alumina
crucible and fired at 310 °C for 142 hours with one intermittent grinding, resulting
in the pure phase La4Ni3O8. Extreme care was taken during the use of pure
hydrogen due to the extreme flammability of the gas.
For the solid state fluorination of La4Ni3O8 to afford La4Ni3O8F1, a nickel crucible
was charged with 2.00 g (2.3 mmol) of La4Ni3O8 then placed inside a teflon lined
stainless steel autoclave. A second nickel crucible was loaded with 0.220 g XeF2
(1.3 mmol, 2.6 mmol F) using a teflon spoon then placed in the same teflon lined
autoclave. The autoclave was sealed inside the glovebox to limit exposure to air
then placed in a muffle furnace. A 48 hr temperature ramp was set to a final
temperature of 250 °C then held for 72 hrs followed by furnace cooling to room
temperature. The autoclave was then transferred back to the glovebox and was
evacuated to remove any excess F2 gas. After two intermittent grindings and
recharges of XeF2, a pure phase sample was obtained with a total firing time of
460 hrs. A similar experimental method was used in the synthesis of La 4Ni3O8F2

120

as that for La4Ni3O8F2. Two nickel crucibles, one containing 2.00 g (2.325 mmol)
La4Ni3O8 and the other containing 0.473 g (2.79 mmol) XeF2, were loaded into a
teflon lined stainless steel autoclave. The autoclave was fired for a total of 296 hrs
at 250 °C with three intermittent grindings and XeF2 recharges. The final samples
were stored under nitrogen at all times.

3.3b Structural characterization of La4Ni3O8Fx (x = 1,2)
Diffraction patterns, refinements, and other structural features of La4Ni3O10 and
La4Ni3O8 are located in Chapter 2 of this thesis. Curious readers should review the
full description in the first few sections of that chapter.
Initially, a well crystallized product was obtained by fluorination of the T-type
parent phase La4Ni3O8 at 250 C for 48 hrs. This product was investigated by
electron diffraction (ED) and energy dispersive X-ray (EDX) analysis. ED patterns
taken from different crystallites can be separated into two groups according to
distinct c/a ratio: ~7.0 for one phase and ~7.3 for the second phase. These phases
were often present in one crystallite as micro-domains. According to the EDX
analysis, both phases have the same La:Ni ratio and contain some amount of
fluorine – a quantitative analysis of fluorine content is difficult with EDX due to the
low number of electrons. However, taking a qualitative approach, the estimated F
content is clearly higher for the phase with the larger c/a ratio. It can be concluded
that a mixture of two oxyfluorides was obtained by the fluorination of La4Ni3O8.

121

Optimization of the synthetic conditions allowed the preparation of La4Ni3O8F1 and
La4Ni3O8F2 separately.

Figure 3-9. Rietveld refinement profile for the synchrotron powder X-ray diffraction
data of La4Ni3O8F1. Observed intensities (crosses), calculated pattern (solid line),
and difference curve (bottom solid line). Tickmarks represent the following phases:
La4Ni3O8F1 – magenta (top), La4Ni3O8F2 – green (middle), La4Ni3O8 – blue
(bottom).

122

Table 3-4. Atomic parameters of La4Ni3O8F1 from Rietveld refinement of
synchrotron powder diffraction data in the space group I4/mmm with a = 3.931(8)
Å and c = 27.66(1) Å.
Atom

Site

x

y

z

Uiso (102 Å2)

La1

4e

0

0

0.4335(2)

1.16(1)

La2

4e

0

0

0.3022(3)

1.01(2)

Ni1

2a

0

0

0

0.83(1)

Ni2

4e

0

0

0.1356(2)

0.27(2)

O1

4c

0

0.5

0

2.09(8)

O2

4e

0

0.5

0.8627(5)

1.08(4)

O3

8g

0

0

0.2216(8)

1.68(2)

F1

4e

0

0

0.0729(1)

1.89(7)

The space groups of the prepared nickel oxyfluorides were determined by
Rietveld refinement of synchrotron powder X-ray diffraction data (Figure 3-9). An
idealized (space group I4/mmm) structure of an n=3 RP phase was used for a
starting model preparation. The symmetry of the starting model was lowered to
several possible space groups as determined by ED. Typical ED patterns of nickel
oxyfluorides material are shown in Figure 3-10. The ED patterns are characterized
by sharp spots that correlate to an c/a ratio of 7.02 – 7.10 for La4Ni3O8F1 and 7.22
– 7.28 for La4Ni3O8F2. ED measurements of La4Ni3O8F1 point to an I4/mmm lattice
structure with a faint orthorhombic superstructure present. The reflection

123

conditions found for La4Ni3O8F2 were hkl: h+k=2n; 0kl: k,l =2n; h0l: h=2n; h00:
h=2n; 0k0: k=2n; 00l: l=2n; therefore, Cc-- is the only possible extinction symbol.
For Ccc- there is disagreement with h0l: h,l=2n, which would not allow for any of
the weaker reflections to be present, nor could they then be due to double
diffraction or twinning. Ccm21 (36), Cc2m (40) and Ccmm (63) are possible
subgroups of the Cc-- extinction symbol. To distinguish between these, convergent
beam electron diffraction would be needed for determining whether the point group
is mm2 (36), 2mm (40) or mmm (63). Due to observed crystal twinning, the results
from such experiment would be unreliable.
Rietveld refinement of the synchrotron diffraction data for La4Ni3O8F1
converged successfully with a three phase refinement, shown in Figure 3-11.
Fittings parameters of χ2 = 4.82 and Rwp = 2.86 % were obtained, showing a very
good refinement of the high resolution data. Atom positions of La4Ni3O8F1
(I4/mmm, a = 3.931(8), c = 27.66(1) ) are presented in Table 3-4. Although both
O2- and F1- are isoelectronic and cannot be directly distinguished in synchrotron
diffraction data, F1- appears to sit in a unique crystallographic site. When F1- is
introduced shared position with O2- atoms, the refinement quickly diverged or
negative thermal parameters were reported. Furthermore, F1- occupancy was able
to be refined, resulting in a fluorine stoichiometry of F1.11 – slightly higher than the
La4Ni3O8F1 ideal stoichiometry. Other atom occupancies were rounded to the
appropriate value for the parent phase La4Ni3O8. The phases La4Ni3O8F2 and the
parent La4Ni3O8 were both included in the refinement, with a normalized refined

124

phase composition ratio of 0.947:0.041:0.012 for La4Ni3O8F1:La4Ni3O8F2:La4Ni3O8,
respectively.

Figure 3-10. Electron diffraction patterns of La4Ni3O8F2 (left column) and
La4Ni3O8F1 (right column).

125

To further investigate the possibility of O2-/F1- ordering, neutron diffraction
measurements were taken of La4Ni3O8F1 at Oak Ridge National Lab under the
guidance of Dr. Jason Hodges (Figure 3-11). Unfortunately, as a consequence of
the similar neutron cross sections between O2- and F1- and the extremely poor
counting statistics of the data, no further information on anion ordering was
obtained (27).

Figure 3-11. Rietveld refinement of the NPD data of La4Ni3O8F1. Observed
intensities (crosses), calculated pattern (solid line), and difference curve (bottom
solid line). Tickmarks represent the following phases: La4Ni3O8F1 – magenta (top),
La4Ni3O8F2 – green (middle), La4Ni3O8 – blue (bottom).

126

Synchrotron powder diffraction, neutron powder diffraction, and electron
diffraction were all performed on La4Ni3O8F2. Unfortunately, only the SPXD pattern
had good enough counting statistics to use, and only for LeBail refinement to
obtain the cell parameters (54). A very wave-like background and exceptionally
broad peaks were present in the SPXD data. Refinement of the background and
acceptable peak profiles simultaneously resulted divergence of the Rietveld
refinement. The NPD was of similar quality to that of La4Ni3O8F1 presented in
Figure 3-11; however, the lack of an initial model system (as was available for
La4Ni3O8F1 from refinement of the high-resolution synchrotron diffraction data)
greatly hindered any attempts at Rietveld refinement of La 4Ni3O8F2 from neutron
diffraction data. A combination of the LeBail refinement of cell parameters from the
synchrotron data and the c/a ratio obtained from the ED results led to a Ccmm
crystal symmetry assignment with a = 3.977(7) Å, b = 3.898(1) Å, and c = 28.90(4)
Å. As the diffraction positions produced by this space group appeared to correlate
with the broad peaks in the diffraction pattern, a high-resolution neutron diffraction
study may lead to a successful Rietveld refinement of atomic positions.
A comparison of the cell parameters, volumes, and c/a ratios for all four of the
phases prepared in this section is shown in Table 3-5.

127

Table 3-5. Refined cell parameters of the La4Ni3OxFy family of compounds
prepared. Cell parameters obtained for La4Ni3O8F2 via LeBail refinement of
synchrotron data.
Phase

Space Group

La4Ni3O10

Bmab

La4Ni3O8

I4/mmm

La4Ni3O8F1

I4/mmm

La4Ni3O8F2

Ccmm

Cell Parameters (Å)
a = 5.412(2)
b = 5.467(3)
c = 27.96(1)
a = 3.969(2)
c = 26.11(4)
a = 3.931(8)
c = 27.66(1)
a = 3.977(7)
b = 3.898(1)
c = 28.90(4)

Volume (ų)

c/a

827.3(4)

5.168

411.3(1)

6.579

427.4(6)

7.036

445.7(2)

7.271

Nickel oxidation states of the compounds prepared were confirmed using Ni Kedge X-ray absorption spectroscopy measurements (Figure 3-12) were performed.
When the data is viewed at an absorption coefficient of 0.5, it is apparent that the
Ni oxidation state trends as the relative energy of the Ni K-edge increases. For
example, the Ni K-edge energy of La3Ni2O6 is lower in energy compared to LaNiO3
because of the lower oxidation state (Ni1.5+ vs Ni3+) in La3Ni2O6. The Ni K-edge for
the prepared oxyfluoride La4Ni3O8F2 falls near and of higher energy to La3Ni2O7.
From this, a Ni oxidation state of approximately 2+ can be established. As this
oxidation is not exact, it cannot be used to back-calculate F1- stoichiometry.
Quantitative ICP analysis was completed on La4Ni3O8F2, resulting in a
stoichiometry of La4Ni3O8F2.01(1) in which La, Ni, and O stoichiometries were fixed.
The Ni-K XAS results are consistent with Ni oxidation state changes by the

128

multistep synthesis and provide additional evidence of F1- incorporation into the
crystal structure.

Figure 3-12. X-ray absorption spectra for La4Ni3O10, La4Ni3O8, La4Ni3O8F1,
La4Ni3O8F2 as well as applicable standards.

3.3c Discussion on the T to T’ transformation
The layered structures of the RP and the T-type phases can be described as
intergrowth of perovskite and rock- salt (in RP) and infinite layer and fluorite (in Ttype) structural blocks. The formation of the T-type nickelates from the parent RP
phases occurs as a result of an oxygen deintercalation and a structural

129

rearrangement in the (LaO)2 blocks. Oxygen atoms are deintercalated exclusively
from LaO layers between NiO2 planes. Due to the drastic decrease of the distance
between adjacent NiO2 layers the La4Ni3O8 structure does not allow oxygen
vacancies despite its preparation by oxygen deintercalation. During the structural
rearrangement of the (LaO)2 blocks, oxygens shift in between La layers,
transforming the rock-salt type block into a fluorite analog. The driving force for this
T-to-T transformation is attributed to release of structural stress and is discussed
in detail in the literature (49). As a result of the described structural changes, Ni
exclusively adopts square-planar coordination in the infinite NiO2 layers in
La4Ni3O8.

3.3d Concluding remarks for La4Ni3O8Fx (x = 1,2)
The syntheses presented for the oxyfluorides La4Ni3O8F1 and La4Ni3O8F2 is a
rare example of a 4 step synthetic procedure in solid-state chemistry. La4Ni3O8F1
shows a tendency towards O/F ordering based on the refinement of high-resolution
synchrotron data. Rietveld refinement of La4Ni3O8F2 was hindered based on poor
counting statistics and high peak broadening. Electron diffraction measurements
of both phases corroborated the cell parameters refined from the diffraction
patterns. The synthesized phases La4Ni3O8F1 and La4Ni3O8F2 with infinite-layer
NiO2 would be the first phases synthesized with Ni in a higher oxidation state than
1.5+ with this coordination.

130

3.4 Novel solvothermal synthesis of complex oxyfluorides
Although many complex materials can be directly prepared via standard solidstate ceramic methods, compounds with low thermodynamic stability relative to
their formation kinetics must be synthesized using soft chemistry techniques.
Solvothermal chemistry has been employed with much success to produce
metastable phases difficult or impossible to obtain by high-temperature methods
including NaFeSO4F, Rb2Hg3Te4, LaSrMnO4F and LiMnPO4 (16, 19, 48, 55–58).
However, very rarely are solvothermal conditions used to perform topotactic
modifications on extended solids in which the cation and/or anion framework is
preserved. Unlike conventional solid state methods, where direct control over atom
connectivity and crystal structure is largely dictated by thermodynamics, topotactic
strategies allow for the intelligent design of complex oxides beginning from a
precursor host through multiple synthetic steps including ion exchange,
intercalation, substitution, grafting, etc.(1, 59–61)
Multiple reaction pathways for the low-temperature oxidative intercalation of
fluorine into perovskite and perovskite-related hosts have been reported in the
literature, in which a majority focus on the use of F2 gas, NH4F, and MF2 (M = Cu,
Zn, Ni, Ag) as sources of fluorine (21, 62–64). Although relatively effective, the
exceptionally hazardous nature of F2 gas and the unwanted MO by-products
formed during fluorination when using MF2 reagents leaves much to be desired
when performing topotactic fluorination. Very recently, a solvothermal fluorination
method was published by Sheng (65) wherein NH4HF2 in absolute ethanol was

131

used to synthesize the superconductor Sr2Cu(O,F)4+δ with Tc = 38 K. As with
previously reported fluorination techniques, the unwanted by-products SrF2 and
CuO were formed according to the following reaction (65):
Sr2 CuO3 + 2HF → Sr2 Cu(O,F)4+x + SrF2 + CuO + H2 O

(3-3)

The presence of water during the synthesis of oxyfluorides is especially
alarming considering that many metastable oxyfluorides slowly hydrolyse in the
presence of water to make HF. The solvothermal fluorination method employing
XeF2 as the oxidative fluorine source reported herein relies on a much simpler
reaction pathway because of the nature of the reactants:
Sr2 Co2 O5 + 0.5XeF2 → Sr2 Co2 O5 F + 0.5Xe

(3-4)

Assuming the reaction conditions (time, temperature, pressure, XeF2
concentration) are optimized, unwanted by-products can be avoided to afford pure
product. The fluorination of the Brownmillerite Sr2Co2O5 was performed in a steel
autoclave as a validation case for the novel solvothermal fluorination method
proposed.

3.4a Description of the solvothermal fluorination method
The solvothermal method differs from the sole previous report of solvothermal
fluorination in the fluorine source – XeF2. The solubility of XeF2 is fairly high in HF
– 192 g/100 mL at 29.95 °C (66). It is also reasonably soluble in chloroform,
dichloromethane, tetrachloromethane, fluorotrichloromethane, and acetonitrile;
however, both proton-fluorine and chlorine-fluorine exchange was observed in the

132

halogenated solvents while the exchange in acetonitrile was negligible (67, 68).
Acetonitrile was chosen as the solvent for the low reactivity and the
environmentally friendly aspect.

Figure 3-13. Comparison of the two known methods for the preparation of the
oxyfluoride Sr2Co2O5F. Colored balls represent the following atoms: blue – Co, red
– O, white – Sr, green – F, and purple – O/F disorder.

Approximately 250 mg of the parent phase is weighed out in a glovebox and
added to a 45 mL teflon sleeve, followed by 3 mL of anhydrous acetonitrile and
507 mg XeF2 – generating a 1 M solution of XeF2 with approximately 5x molar
excess fluorine in solution. This concentration may need to be changed depending
on the formula weight of the parent phases to maintain the 5x molar excess. The
teflon sleeve was capped and inserted into a stainless steel autoclave, sealed, and
loaded into a temperature controlled oven. Depending on the stability of the phase,
a temperature range between 50 °C and 225 °C can be used. Before performing

133

any of these experiments, the autogenous pressure generated during reaction
from both the solvent and complete decomposition of the XeF2 must be calculated.
Once the reaction is completed, the products were washed three times with
anhydrous acetonitrile to remove any excess XeF2 then dried under vacuum.

3.4b Proof-of-principle: solvothermal synthesis of Sr2Co2O5F
The brownmillerite phase Sr2Co2O5 was synthesized using previous literature
methods (69). Stoichiometric amounts of SrCO3 and Co3O4 were fired for 24 hours
at 1000 °C in air then quenched into liquid nitrogen to avoid the formation of the
thermodynamically stable cubic structure. However, as the resulting product is not
anion ordered, using the cubic perovskite SrCoO3-x as the precursor phase is
feasible and synthetically easier.
The brownmillerite lattice is perovskite-like in structure but differs in local
coordination. In brownmillerite phases, oxygen vacancies are ordered within
planes, generating two different vertex-linked transition metal coordination
environments: tetrahedral chains along the a axis and tetrahedral/octahedral
layers along the c axis, as depicted in Figure 3-13.
Sr2Co2O5F was first synthesized in 2012 by the Greaves group (70). A sample
of undisclosed mass was reacted at 200 °C for 15 minutes under a 10% F 2/N2
stream. The sample was then ground and fired at 220 °C for 90 minutes in static
10% F2/N2. Sr2Co2O5F produced by this method was poisoned with a substancial
amount of the parent brownmillerite, probably due to fluorination on the surface.

134

Figure 3-14. Observed (green trace), calculated (crosses), and difference (grey
trace, below) XRD profiles obtained from Rietveld refinement of PXD data for
Sr2Co2O5F made at 100 °C. Magenta tickmarks denote Bragg reflections.

The refinement of Sr2Co2O5F converged with χ2 = 0.491 and Rwp = 2.21 % with
a very good fit between the calculated and observed patterns as shown in Figure
3-14. The refined cell parameter a = 3.8549(1) Å was very similar to the literature
value of a = 3.8574 Å (70) as well as the literature value for the cubic disordered
perovskite SrCoO3 (a = 3.855 Å) (71). Addition of SrF2 did not result in a change
of any fitting parameters or a relative phase percent of more than 0.34% SrF2. This
finding also validates the fact that the solvothermal method leads to the synthesis

135

of metastable phases without the byproducts generally associated with solid state
fluorination. The refined atomic positions of Sr2Co2O5F are shown in Table 3-6.

Table 3-6. Structure parameters of Sr2Co2O5F based on Rietveld refinement of
powder diffraction data in the space group Pm3̅ m with a = 3.8549(1) Å.
Name

Site

x

y

z

Uiso (102 Å2)

Sr

1a

0

0

0

1.022

Co

1b

0.5

0.5

0.5

0.542

O/F

3c

0

0.5

0.5

1.328

The temperature dependence of DC magnetic susceptibility measured at a
3000 Oe field is presented in Figure 3-15. The increase of magnetic susceptibility
at low temperatures is associated with a canted antiferromagnetism state. A
bifurcation of the FC and ZFC magnetic susceptibility curves occurs at T ~ 30 K,
similar to what was previously reported. The top inset displays an expanded view
of the isothermal magnetization measurement M(H) at 2 K with the magnetic field
H swept between +9 T and -9 T, where a hysteresis loop is obviously observed a feature previously argued to originate from the existence of a ferromagnetic
component in Sr2Co2O5 (70). However, the hysteresis behavior only occurs below
30 K, found by measuring the M(H) curves from 9 T to 0 T at various temperatures
(insets of Figure 3-15). The remnant magnetization decreases with increasing
temperature and becomes zero above 30 K; therefore, instead of being genuinely

136

ferromagnetic, Sr2Co2O5F may form a spin-glass like magnetic state with a glass
transition temperature of 30 K, which is presumably attributed to the magnetic
frustration or disorder originating from the complex magnetic interactions involving
Co4+ and Co3+ mixed states.

Figure 3-15. DC magnetic susceptibility measurements of Sr2Co2O5F. Insets show
the field and temperature dependence on magnetization in terms of magnetic
moment per Co. Data collected on a Quantum Design PPMS system.

3.4c Exploring the importance of experimental conditions
To further understand the fluorination reaction mechanism in this solvothermal
method, a series of experimental setups was devised. The initial hypothesis of a

137

solution based fluorination as opposed to gas based reaction was first adopted.
Following, four different experimental conditions were drawn up, including the
initial solvothermal method described previously. Two solid state reactions - one
using the standard 45 mL autoclave and one with a reduced volume of
approximately 10 mL - as well as a solvothermal method using 20 mL MeCN to
dilute the XeF2 concentration. Each method was tested using an experimental time
of 1, 3, 5, 7, and 10 days at 100 °C. This lower temperature was chosen to subdue
the fluorination rate, which allowed for a more precise comparison over time.
The two solid state type reactions were competed to understand whether the
fluorination occurred via F2 gas as opposed to a soluble fluorine source. For the
standard volume solid state reactions, 250 mg (0.798 mmol) of Sr2Co2O5 and
0.677 mg (4 mmol) XeF2 were loaded into a 45 mL teflon sleeve, capped, and
placed in a stainless steel autoclave followed by thermal treatment at 100 °C. In
the reduced volume solid state reactions, a teflon insert was added to the 45 mL
sleeve after the reactants had been added. The teflon insert displaced
approximately 35 mL of sleeve volume. The sleeve was then capped and placed
in a stainless steel autoclave. The two solvothermal type reactions were completed
using the same parameters as explained above with a change in solvent volume
for the 20 mL case.

138

Figure 3-16. Fluorination rates over time for the fluorination of Sr2Co2O5 using
XeF2 as a fluorine source under different experimental conditions. The blue lines
show the % of Sr2Co2O5F at any given time.

139

From Figure 3-16 and Table 3-7, it is quite clear that fluorination occurs at a
much faster rate under solvothermal conditions than in the solid state reactions. It
appears that the fluorination rates are constant with all four methods. Interestingly,
when lowering the temperature from 125 °C to 100 °C, the fluorination of Sr 2Co2O5
under solvothermal conditions with 3 mL of acetonitrile is still incomplete after 10
days. A cursory glance at the fluorination rates in Figure 3-16 point to a rate first
order in XeF2, an idea that will be discussed further. The small temperature change
results in a very large decrease in reactivity. Neither of the solid state reactions
produced more than approximately 20% of the oxyfluoride Sr 2Co2O5F after 10
days.

Table 3-7. Percent formation of Sr2Co2O5F via the fluorination of Sr2Co2O5 over
time using four different methods. The total percent of unreacted Sr2Co2O5 is the
difference between the listed value below and 100.
Time

Solvothermal
(3 mL)

Solvothermal
(20 mL)

Solid State
(45 mL)

Solid State
(10 mL)

24

10

7

<3

<3

72

24

12

4

5

120

38

28

6

7

168

62

41

9

11

240

88

55

11

17

140

In the solid state reactions, there are two possible pathways for fluorination to
occur as shown in equations 3-5, 3-6 and 3-7, with the first one being much more
likely based on thermodynamic arguments.
XeF2 →Xe+2F•

(3-5)

XeF2 →XeF+ +F-

(3-6)

F• /F- +Sr2 Co2 O5 □1 →Sr2 Co2 O5 F1

(3-7)

The concentration of F• is directly related to the temperature of the reaction as
well as the vapor pressure of XeF2 because it is in equilibrium with XeF2 (g) below
115 °C. The vapor pressure of XeF2 was explicitly measured by Chernick at
Argonne national Lab and can be described by the equation presented in (3-8)
where Pmm is the vapor pressure in mmHg and T is the reaction temperature (72).
logPmm =

-3057.67
-1.23521logT+13.969736
T

(3-8)

For instance, the vapor pressure of XeF2 is 25.34 mmHg and 394.1 mmHg at
50 °C and 100 °C, respectively, using equation (3-8). The formation of atomic
fluorine is much more likely in this reaction because formation of the fluorine anion
requires the presence of a Lewis acid, for example performing the reaction in a
Pyrex flask. The acidic sites (B2O3 and Al2O3) polarize the linear XeF2 molecule,
allowing for reaction with the highly electrophilic XeF+ (73). At higher temperatures,
the equilibrium shifts to purely atomic Xe and F• . Corroborating evidence for the
radical species acting as the fluorination agent is given by the increase in reaction
rate as the experimental volume is decreased from 45 mL to 10 mL, lowering the

141

mean path required for reactants to interact. Using simple ideal-gas law equations,
the partial pressure of F• is 4.5 times higher in the reduced volume case.

Figure 3-17. Possible bonding interaction between XeF2 and two adjacent CoO4
tetrahedra in the brownmillerite structure Sr2Co2O5. A full description is presented
in the text. Bond lengths were obtained from the literature (20, 28, 66, 74, 75).

142

Deducing the mechanism of the solvothermal reactions is a bit more difficult
because of the possible reactants. As mentioned previously, and discussed in
much more detail further, the electrophilic XeF+ is formed in the presence of Lewis
acids. In the case of the fluorination of Sr2Co2O5, Co3+ near the surface can act as
a hard Lewis acid (Figure 3-17), forming a weak bond with one of the fluorine atoms
and generating the electrophile XeF+ while oxidizing the Co3+ to Co3+/Co4+. The
electrophile can then participate in a one electron radical transfer, resulting in the
release of Xe gas and the local abstraction of fluorine into the vacancies in the unit
cell of Sr2Co2O5 (66). This reactivity is visualized in Figure 3-17, where the XeF+
ion interacts with XeF2, forming the bent Xe2F3+ electrophile (66, 75). When the
Xe2F3+ approaches an open face of the CoO4 tetrahedra in Sr2Co2O5, a fluorine
atom can form a bond with the acidic Co3+. This is shown in the picture as the F1Co bond with length 1.78 Å. Because of the geometry of both the bent Xe2F3+ and
the offset tetrahedra, the terminal fluorine in Xe2F3+ is now within about 2 Å of
another Co3+. This Lewis acid site may promote further local formation of Xe 2F3+
with XeF2, essentially creating a cascade fluorination effect. Since this cascade
effect occurs in solution, this mechanism also validates the experimentally
observed results (66, 76, 77). When the volume of acetonitrile was increased from
3 mL to 20 mL while the mass of XeF2 was kept constant in the solvothermal
reactions, the overall concentration of XeF2 and any electrophilic XeF2 species is
lowered by a factor of nearly 7. Refer to the paragraphs on the reactivity of XeF2
in the presence of Lewis acids further in this chapter for more information.

143

Increasing the fluorination power of XeF2 by forming adducts with strong Lewis
acids was also investigated in hopes of reducing reaction times and allowing for
higher throughput. Considering that the chemistry of XeF2 is starkly different than
the chemistry of BF4- or NMe4F, the Lewis acid used had to be strong with
appropriate orbital energies. The F1- anion - for example, in NMe4F - is
exceptionally basic, and for that reason also very rare. When the ion is bound
through Lewis interactions to a good Lewis acid like BF3 or PF5, any possible
secondary basicity interactions are reduced, effectively lowering the fluorinating
power relative to the F1- anion. An even further reduction in basicity occurs in XeF2
where the central XeII atom is electron deficient. Because of this reduced basicity,
XeF2 more readily forms ionic salts with strong Lewis acids (denoted LF5 below),
according to equations 3-9 and 3-10 below.
XeF2 + LF5 → XeF+ + LF6-

(3-9)

2XeF2 + LF5 → Xe2F3+ + LF6-

(3-10)

These two species, XeF+ and Xe2F3+, are both highly electrophilic and easily
reduced by a single electron to form the strong fluorination agents XeF• and Xe2F3•,
respectively.
To test the hypothesis that adding a Lewis acid to a solution of XeF 2 in MeCN
increases the fluorinating power of the solution, reactions were set up using NbF 5
and AlCl3. A 1M solution of XeF2 in acetonitrile was prepared by dissolving 0.845
g (5 mmol) XeF2 in 5 mL of anhydrous acetonitrile in a glass vial in the glovebox.
A teflon stir bar was added to promote dissolution. Half of the solution was then

144

transferred to a second vial using a volummetric pipette. 0.939 g (5 mmol) NbF5 or
0.667 g (5 mmol) AlCl3 were added separately to the vials. After vigorous reaction,
especially for AlCl3, the solutions turned pale yellow. Following complete
dissolution of the Lewis acid, 0.373 g (1 mmol) of Sr2Co2O5 was added. The vials
were then stirred for 1 and 3 days, after which the product was washed three times
with anhydrous acetonitrile and diffractograms were collected. According to the
diffractograms, only pure phase Sr2Co2O5 was present after all reactions with
XeF2-Lewis acid fluorine sources.
There are two possible reasons why the fluorination rates were substantially
suppressed after addition of NbF5 or AlCl3 to XeF2. The first explanation is that
neither of these materials are strong enough Lewis acids to form ionic salts with
XeF2, instead forming more covalent type interactions. This idea is corroborated
by the fact that both AlCl3 and NbF5 form polymeric units that must crack to create
the ionic XeF2 adduct. The other possible explanation is the lack of knowledge on
the stability of highly reactive fluorinating agents in MeCN. Considering that XeF2
adducts with NbF5 and AlCl3 should be colorless, the yellow solution and fierce
bubbling may have represented the oxidation and/or fluorination of MeCN - shown
in equation (3-11).
XeF2 + NbF5 → [XeF]+[NbF6]- + CH3CN → Xe + H+ + NbF6- + CH2FCN

(3-11)

As very little is known on the stability of XeF2 adducts in acetonitrile - many of
these compounds have been intensely studied in dry HF - further experiments
must be first completed to identify the side reactions occurring with the solvent.

145

Alternatively, a transition to using HF as the solvent is possible; however, as the
driving force for the creation of this solvothermal method was to lower the energy
requirements and safety precautions of the experiment, using HF as a solvent
would appear to be moving backwards in that regard.

146

3.5 Chapter Conclusions
In this chapter, multiple different synthetic methods for the preparation of
complex, metastable oxyfluoride materials are discussed. The synthesis of
SrFeO2F resulted in a large thermal parameter for the O/F lattice site, a sure sign
of anion disorder in the structure. However, a large preference for cis-FeO4F2
coordination was demonstrated when SrFeO2F was synthesized using the softchemistry approached described herein, as opposed to the complete
randomization in the previous higher temperature synthesis report. Further studies
on this material, may result in a completely anion ordered structure.
Fluorination of the T’ type La4Ni3O8 phase previously synthesized in Chapter 2
of this thesis using a low temperature, multistep synthetic method was also
discussed. Two distinct oxyfluoride phases, La4Ni3O8F1 and La4Ni3O8F2, were
prepared. Structure characterization and Rietveld refinement was performed on
synchrotron and NPD patterns of La4Ni3O8F2, while high-resolution diffraction
patterns were elusive for La4Ni3O8F2 due to the poor crystallinity. Cell parameters
and space groups were confirmed by electron diffraction studies. Stoichiometry
and Ni oxidation states were investigated using ICP and XANES, respectively.
The chapter concludes with a presentation of a solvothermal fluorination
method that does not result in metal oxide admixtures – a common problem with
solid state fluorination methods. The power of the method is demonstrated with the
synthesis of Sr2Co2O5F at a much lower temperature than previously reported.

147

3.6 Future work on this topic
Due to the robust nature of the solvothermal fluorination method presented, a
future direction of the project could focus on the synthesis of metastable
oxyfluorides tailored for certain properties, including second harmonic generation
or Li-ion cathode materials. Depending on the diffusion pathways available in a
target phase, many different compounds could be synthesized with this method,
especially those structural types with anion vacancies or large interstitial zones
(Brownmillerite, Aurivillius, etc). Applying the ideas of structure vs. function by
modulating the transition metal oxidation states or the crystal structure in these
phases via fluorination may lead to the direct tuning of physical or magnetic
properties of the target phases.
A very different direction of this project could rely on developing methods to
intercalate or exchange sulfur in a topotactic regime. The recent literature has
shown sulfurization to be a very interesting method in synthesizing functional
materials, yet there are no – as of this writing – papers on the topotactic
intercalation of sulfur (78–80). A topotactic sulfurization reaction could be imagined
where triphenylphosphine is used as the sulfur transport reagent, especially for
reactions where the driving force is the formation of triphenylphosphine oxide after
exchange.

148

REFERENCES

149

REFERENCES

(1)

Ranmohotti, K. G. S.; Josepha, E.; Choi, J.; Zhang, J.; Wiley, J. B.
Topochemical manipulation of perovskites: low-temperature reaction
strategies for directing structure and properties. Adv. Mater. 2011, 23, 442–
60.

(2)

Mcintyre, R. A.; Falster, A. U.; Li, S.; Simmons, W. B.; Connor, C. J. O.;
Wiley, J. B. Opening a Perovskite to Valence Manipulation : Two-Step
Topotactic Route to a New Mixed-Valence Titanate, Na1-x+yCax/2LaTiO4. J.
Am. Chem. Soc. 1998, 7863, 217–218.

(3)

Li, R. K.; Greaves, C. Double-layered ruthenate Sr3Ru2O7F2 formed by
fluorine insertion into Sr3Ru2O7. Phys. Rev. B 2000, 62, 3811–3815.

(4)

Troyanchuk, I.; Kasper, N.; Mantytskaya, O.; Shapovalova, E. High-pressure
synthesis of some perovskite — Like compounds with a mixed anion type.
Mater. Res. Bull. 1995, 30, 421–425.

(5)

Yoo, C.-Y.; Kim, S.-J. Dimensional modification of oxyfluoride lattice:
Preparation and structure of A′ANb2O6F (A′=Na, K; A=Ca, Sr). J. Phys.
Chem. Solids 2008, 69, 1475–1478.

(6)

Berry, F. J.; Ren, X.; Heap, R.; Slater, P.; Thomas, M. F. Fluorination of
perovskite-related SrFeO3−δ. Solid State Commun. 2005, 134, 621–624.

(7)

Berry, F. J.; Heap, R.; Helgason, Ö.; Moore, E. a; Shim, S.; Slater, P. R.;
Thomas, M. F. Magnetic order in perovskite-related SrFeO2F. J. Phys.
Condens. Matter 2008, 20, 215207.

(8)

Isobe, M.; Li, J.-Q.; Matsui, Y.; Izumi, F.; Kanke, Y.; Takayama-Muromachi,
E. Synthesis, crystal structures and superconductivity of new copper
oxyfluorides, Sr2RCu2O5F (R = Y,La,Nd,Sm,Gd,Dy,Er and Yb). Phys. C
Supercond. 1996, 269, 5–14.

(9)

Alekseeva, A. M.; Abakumov, A. M.; Rozova, M. G.; Antipov, E. V.;
Hadermann, J. Synthesis and crystal structure of the Sr2MnGa(O,F)6
oxyfluorides. J. Solid State Chem. 2004, 177, 731–738.

(10) Tong, W.; Yoon, W.-S.; Hagh, N. M.; Amatucci, G. G. A Novel Silver
Molybdenum Oxyfluoride Perovskite as a Cathode Material for Lithium
Batteries. Chem. Mater. 2009, 21, 2139–2148.

150

(11) Ravez, J. The Inorganie Fluoride and Oxyfluoride Ferroelectrics. J. Phys. III
1997, 7, 1129–1144.
(12) Mizoguchi, H.; Orita, M.; Hirano, M.; Fujitsu, S.; Takeuchi, T.; Hosono, H.
NbO2F: An oxyfluoride phase with wide band gap and electrochromic
properties. Appl. Phys. Lett. 2002, 80, 4732.
(13) Isobe, M.; Li, J.-Q.; Matsui, Y.; Izumi, F.; Kanke, Y.; Takayama-Muromachi,
E. Synthesis, crystal structures and superconductivity of new copper
oxyfluorides, Sr2RCu2O5F (R = Y,La,Nd,Sm,Gd,Dy,Er and Yb). Phys. C
Supercond. 1996, 269, 5–14.
(14) Amatucci, G. G.; Pereira, N. Fluoride based electrode materials for
advanced energy storage devices. J. Fluor. Chem. 2007, 128, 243–262.
(15) Greaves, C.; Kissick, J. L.; Francesconi, M. G.; Aikens, L. D.; Gillie, L. J.
Synthetic strategies for new inorganic oxide fluorides and oxide sulfates. J.
Mater. Chem. 1999, 9, 111–116.
(16) Aikens, L. D.; Li, R. K.; Greaves, C. The synthesis and structure of a new
oxide fluoride, LaSrMnO4F, with staged fluorine insertion. Chem. Commun.
2000, 2129–2130.
(17) Francesconi, M. G.; Slater, P. R.; Hodges, J. P.; Greaves, C.; Edwards, P.
P.; Al-Mamouri, M.; Slaski, M. Superconducting Sr2−xAxCuO2F2+δ (A=Ca,
Ba): Synthetic Pathways and Associated Structural Rearrangements. J.
Solid State Chem. 1998, 135, 17–27.
(18) Slater, P. R.; Gover, R. K. B. Synthesis and structure of the new oxide
fluoride Sr2TiO3F2 from the low temperature fluorination of Sr2TiO4: an
example of a staged fluorine substitution/insertion reaction. J. Mater. Chem.
2002, 12, 291–294.
(19) Aikens, L. D.; Gillie, L. J.; Li, R. K.; Greaves, C. Staged fluorine insertion into
manganese oxides with Ruddlesden–Popper structures: LaSrMnO4F and
La1.2Sr1.8Mn2O7F. J. Mater. Chem. 2002, 12, 264–267.
(20) Weinstock, B.; Weaver, E. E.; Knop, C. P. The Xenon-Fluorine System.
Inorg. Chem. 1966, 5, 2189–2203.
(21) Saratovsky, I.; Lockett, M. A.; Rees, N. H.; Hayward, M. A. Preparation of
Sr7Mn4O13F2 by the topotactic reduction and subsequent fluorination of
Sr7Mn4O15. Inorg. Chem. 2008, 47, 5212–7.

151

(22) Cordier, S.; Roisnel, T.; Poulain, M. Synthesis and characterization of the
novel Nb3O5F5 niobium oxyfluoride: the term n=3 of the NbnO2n−1Fn+2 series.
J. Solid State Chem. 2004, 177, 3119–3126.
(23) Prodjosantoso, A. .; Kennedy, B. .; Vogt, T.; Woodward, P. . Cation and
anion ordering in the layered oxyfluorides Sr3−xAxAlO4F (A=Ba, Ca). J. Solid
State Chem. 2003, 172, 89–94.
(24) Brink, F. J.; Withers, R. L.; Cordier, S.; Poulain, M. An electron diffraction
and bond valence sum investigation of oxygen/fluorine ordering in
NbnO2n−1Fn+2, n=3. J. Solid State Chem. 2006, 179, 341–348.
(25) Tsujimoto, Y.; Tassel, C.; Hayashi, N.; Watanabe, T.; Kageyama, H.;
Yoshimura, K.; Takano, M.; Ceretti, M.; Ritter, C.; Paulus, W. Infinite-layer
iron oxide with a square-planar coordination. Nature 2007, 450, 1062–5.
(26) Chen, W. M.; Hong, W.; Geng, J. F.; Wu, X. S.; Ji, W.; Li, L. Y.; Qui, L.; Jin,
X. Iodometric titration for determining the oxygen content of samples doped
with Fe and Co. Phys. C Supercond. 1996, 270, 349–353.
(27) Sears, V. F. Neutron scattering lengths and cross sections. Neutron News
1992, 3, 26–37.
(28) Shannon, R. D. Revised effective ionic radii and systematic studies of
interatomic distances in halides and chalcogenides. Acta Crystallogr. Sect.
A 1976, 32, 751–767.
(29) Tassel, C.; Kageyama, H. Square planar coordinate iron oxides. Chem. Soc.
Rev. 2012, 41, 2025–35.
(30) Hayward, M. A.; Rosseinsky, M. J. Cool conditions for mobile ions. Nature
2007, 450, 960.
(31) Beurmann, P. Phase transitions in the SrSnO3–SrFeO3 solid solutions: Xray diffraction and Mössbauer studies. J. Solid State Chem. 2003, 174, 392–
402.
(32) Woo, H.; Tyson, T. a; Croft, M.; Cheong, S.-W. Evidence for spin clusters
and glassy behaviour in Bi1-xCaxMnO3 (x = 0.875). J. Phys. Condens. Matter
2004, 16, 2689–2705.
(33) Hyett, G.; Barrier, N.; Clarke, S. J.; Hadermann, J. Topotactic oxidative and
reductive control of the structures and properties of layered manganese
oxychalcogenides. J. Am. Chem. Soc. 2007, 129, 11192–201.

152

(34) Bhowmik, R. N.; Ranganathan, R. Anomaly in cluster glass behaviour of
Co0.2Zn0.8Fe2O4 spinel oxide. J Mag. Mater. 2002, 248, 101–111.
(35) Hjøllum, J.; Madsen, M. B. Fit:o) - A Mossbauer spectrum fitting program.
arXiv 2011.
(36) Inaguma, Y.; Greneche, J. Structure and Mo 1 ssbauer Studies of F-O
Ordering in PbFeO2F. Chem. Mater. 2005, 1386–1390.
(37) Blaha, P.; Schwarz, K.; Madsen, G. K. H.; Kvasnicka, D.; Luitz. WIEN2k, An
Augmented Plane Wave + Local Orbitals Program for Calculating Crystal
Properties, 2001.
(38) Blochl, P. E. Projector augmented-wave method. Phys. Rev. B 1994, 50,
17953.
(39) Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M.
R.; Singh, D. J.; Fiolhais, C. Atoms, Molecules, Solids, and Surfaces:
Applications of the Generalized Gradient Approximation for Exchange and
Correlation. Phys. Rev. B 1992, 46, 6671.
(40) Kresse, G.; Hafner, J. Ab initio molecular dynamics for liquid metals. Phys.
Rev. B 1993, 47, 558.
(41) Adamson, A. W. Domain representations of orbitals. J. Chem. Educ. 1965,
42, 140.
(42) Urusov, V. S.; Orlov, I. P. State-of-art and Perspectives of the Bond-Valence
Model in Inorganic Crystal Chemistry. Crystallogr. Reports 1999, 44, 736–
760.
(43) Poltavets, V. V; Lokshin, K. A.; Croft, M.; Mandal, T. K.; Egami, T.;
Greenblatt, M. Crystal structures of Ln4Ni3O8 (Ln = La, Nd) triple layer T’type nickelates. Inorg. Chem. 2007, 46, 10887–91.
(44) Poltavets, V. V.; Lokshin, K. A.; Egami, T.; Greenblatt, M. The oxygen
deficient Ruddlesden–Popper La3Ni2O7−δ (δ=0.65) phase: Structure and
properties. Mater. Res. Bull. 2006, 41, 955–960.
(45) Poltavets, V. V; Lokshin, K. A.; Dikmen, S.; Croft, M.; Egami, T.; Greenblatt,
M. La3Ni2O6: a new double T’-type nickelate with infinite Ni1+/2+O2 layers. J.
Am. Chem. Soc. 2006, 128, 9050–1.

153

(46) Poltavets, V.; Greenblatt, M.; Fecher, G.; Felser, C. Electronic Properties,
Band Structure, and Fermi Surface Instabilities of Ni1+/Ni2+ Nickelate
La3Ni2O6, Isoelectronic with Superconducting Cuprates. Phys. Rev. Lett.
2009, 102, 046405.
(47) Hayward, M. A.; Green, M. A.; Rosseinsky, M. J.; Sloan, J. Sodium Hydride
as a Powerful Reducing Agent for Topotactic Oxide Deintercalation:
Synthesis and Characterization of the Nickel(I) Oxide LaNiO2. J. Am. Chem.
Soc. 1999, 121, 8843–8854.
(48) Blakely, C. K.; Bruno, S. R.; Poltavets, V. V. Low-temperature solvothermal
approach to the synthesis of La4Ni3O8 by topotactic oxygen deintercalation.
Inorg. Chem. 2011, 50, 6696–700.
(49) Lacorre, P. Passage from T-type to T′-type arrangement by reducing
R4Ni3O10 to R4Ni3O8 (R = La, Pr, Nd). J. Solid State Chem. 1992, 97, 495–
500.
(50) Wen, X.-D.; Cahill, T. J.; Gerovac, N. M.; Bucknum, M. J.; Hoffmann, R.
Playing the quantum chemical slot machine: an exploration of ABX2
compounds. Inorg. Chem. 2010, 49, 249–60.
(51) AI-Mamouri, M.; Edwards, P. P.; Greaves, C.; Slaski, M. Synthesis and
superconducting properties of the strontium copper oxyfluoride Sr2CuO2F2+δ.
Nature 1994, 369, 382–384.
(52) ApRoberts-Warren, N.; Crocker, J.; Dioguardi, a. P.; Shirer, K. R.; Poltavets,
V. V.; Greenblatt, M.; Klavins, P.; Curro, N. J. NMR evidence for spin
fluctuations in the bilayer nickelate La3Ni2O6. Phys. Rev. B 2013, 88,
075124.
(53) ApRoberts-Warren, N.; Dioguardi, A. P.; Poltavets, V. V.; Greenblatt, M.;
Klavins, P.; Curro, N. J. Critical spin dynamics in the antiferromagnet
La4Ni3O8 from 139La nuclear magnetic resonance. Phys. Rev. B 2011, 83,
014402.
(54) Peterson, V. K. Lattice parameter measurement using LeBail versus
structural (Rietveld) refinement: A caution for complex, low symmetry
systems. Powder Diffr. 2012, 20, 14–17.
(55) Tripathi, R.; Ramesh, T. N.; Ellis, B. L.; Nazar, L. F. Scalable synthesis of
tavorite LiFeSO4F and NaFeSO4F cathode materials. Angew. Chem. Int. Ed.
Engl. 2010, 49, 8738–42.

154

(56) Li, J.; Chen, Z.; Lam, K.; Mulley, S.; Proserpio, D. M.; Strutturale, C.
Rb2Hg3Te4 : A New Layered Compound Synthesized from Solvothermal
Reactions. 1997, 1669, 684–687.
(57) Dong, Y.; Xie, H.; Song, J.; Xu, M.; Zhao, Y.; Goodenough, J. B. The
Prepared and Electrochemical Property of Mg Doped LiMnPO 4 Nanoplates
as Cathode Materials for Lithium-Ion Batteries. J. Electrochem. Soc. 2012,
159, A995–A998.
(58) Choi, D.; Wang, D.; Bae, I.-T.; Xiao, J.; Nie, Z.; Wang, W.; Viswanathan, V.
V; Lee, Y. J.; Zhang, J.-G.; Graff, G. L.; Yang, Z.; Liu, J. LiMnPO4 nanoplate
grown via solid-state reaction in molten hydrocarbon for Li-ion battery
cathode. Nano Lett. 2010, 10, 2799–805.
(59) Schaak, R. E.; Mallouk, T. E. Perovskites by Design: A Toolbox of SolidState Reactions. Chem. Mater. 2002, 14, 1455–1471.
(60) Blakely, C. K.; Bruno, S. R.; Davis, J. D.; Poltavets, V. V. Synthesis and
Electrochemical Properties of Novel Co2+/Co3+ Layered Ordered Rock-Salt
Structure Type Li0.75Ca0.25CoO2. ECS Trans. 2013, 45, 3–10.
(61) Wu, S.-H.; Liu, H.-Y. Preparation of α-LiFeO2-based cathode materials by
an ionic exchange method. J. Power Sources 2007, 174, 789–794.
(62) Slater, P. R.; Hodges, J. P.; Francesconi, M. G.; Greaves, C.; Slaskic, M.
Fluorination of the Ruddlesden – Popper type cuprates , LnACuO (Ln = La ,
Nd ; A = Ca , Sr). Structure 1997, 7, 2077–2083.
(63) Stavber, G.; Zupan, M.; Stavber, S. Solvent-free fluorination of organic
compounds using N–F reagents. Tetrahedron Lett. 2007, 48, 2671–2673.
(64) Berry, F. J.; Ren, X.; Heap, R.; Slater, P.; Thomas, M. F. Fluorination of
perovskite-related SrFeO3−δ. Solid State Commun. 2005, 134, 621–624.
(65) Sheng, J.; Tang, K.; Liang, Z.; Wang, Y.; Wang, D.; Zhang, W. Solvothermal
fluorination: A new chemical fluorination method to insert fluorine into
Sr2CuO3 and NdSr2Cu2O6−δ. Mater. Chem. Phys. 2009, 115, 483–487.
(66) Tramšek, M.; Žemva, B. Synthesis , Properties and Chemistry of Xenon(II)
Fluoride. 2006, 105–116.
(67) Barbara Grüttner, Michael F. A. Dove, Alan F. Clifford, G. J. and H. S.
Inorganic Chemistry in Liquid Hydrogen Cyanide and Liquid Hydrogen
Fluoride; Elsevier, 1971.

155

(68) Dukat, W.; Holloway, J. H.; Hope, E. G.; Townson, P. J.; Powell, R. L. The
reactions of xenon difluoride with “inert” solvents. J. Fluor. Chem. 1993, 62,
293–296.
(69) Sullivan, E.; Hadermann, J.; Greaves, C. Crystallographic and magnetic
characterisation of the brownmillerite Sr2Co2O5. J. Solid State Chem. 2011,
184, 649–654.
(70) Sullivan, E.; Greaves, C. Fluorine insertion reactions of the brownmillerite
materials Sr2Fe2O5, Sr2CoFeO5, and Sr2Co2O5. Mater. Res. Bull. 2012, 47,
2541–2546.
(71) Taguchi, H.; Shimada, M.; Koizumi, M. The Effect of Oxygen Vacancy on the
Magnetic Properties in the System SrCoO3-x. J. Solid State Chem. 1979,
225, 221–225.
(72) Schreiner, F.; McDonald, G. N.; Chernick, C. L. Vapor pressure and melting
points of xenon difluoride and xenon tetrafluoride. J. Phys. Chem. 1968, 72,
1162–1166.
(73) Ramsden, C. A. Xenon difluoride in the organic laboratory: a tale of
substrates, solvents and vessels. Arkivoc 2013, 2014, 109.
(74) Stout, J. W.; Reed, S. A. The Crystal Structure of MnF2 , FeF2 , CoF2 , NiF2
and ZnF2. J. Am. Chem. Soc. 1954, 76, 5279–5281.
(75) Nakajima, T.; Tressaud, A.; Zemva, B. Advanced Inorganic Fluorides.
Handb. Clin. Neurol. / Ed. by P.J. Vinken G.W. Bruyn 2008, 90, IX–X.
(76) Brown, D. R.; Clegg, M. J.; Downs, A. J.; Fowler, R. C.; Minihan, A. R.;
Norris, J. R.; Stein, L. The dixenon(1+) cation: formation in the condensed
phases and characterization by ESR, UV-visible, and Raman spectroscopy.
Inorg. Chem. 1992, 31, 5041–5052.
(77) Nabiev, S. S.; Ostroukhova, I. . Raman study of molecular dynamics of
inorganic fluoroxidizers in nonaqueous solutions- Part 1 . Xenon difluoride in
acetonitrile. Spectrochmica Acta 1993, 49, 1527–1535.
(78) Lu, P.; Lu, N.; Wang, J.; Kim, M. J.; Xia, Y. Site-selective sulfurization of
bromide-capped palladium nanocubes by polysulfide and the underlying
mechanism. Nanotechnology 2014, 25, 014003.
(79) Liu, X.; Liu, Z.; Lu, J.; Wu, X.; Chu, W. Silver sulfide nanoparticles sensitized
titanium dioxide nanotube arrays synthesized by in situ sulfurization for

156

photocatalytic hydrogen production. J. Colloid Interface Sci. 2014, 413, 17–
23.
(80) Tang, X.; Ye, H.; Hu, H. Sulfurization synthesis and photocatalytic activity of
oxysulfide La3NbS2O5. Trans. Nonferrous Met. Soc. China 2013, 23, 2644–
2649.

157

Chapter 4 : MuRE – A multistep topotactic synthetic method
4.1 Literature Background on α-NaFeO2 type phases

Figure 4-1. Different structure variations found in the α-NaFeO2 system based on
the coordination of the alkali metal – O for octahedral and P for trigonal prismatic.
A further description of the naming convention is given in the text.

158

At first glance, the α-NaFeO2 structure presented in Figure 4-1 appears to be a
simple rock-salt superstructure with alternating layers of NaO6 and FeO6
octahedrally coordinated planes. However, the actual crystal structure and
coordination geometry of phases in this family are highly dependent on the
stoichiometry of the sample; for example, Na0.76CoO2 and Na0.75CoO2 adopt
different structural motifs (P2 and O1, respectively) with very similar
stoichiometries (1, 2). Before proceeding, the subtle differences in structures for
the α-NaFeO2 family of compounds must be closely examined.
Originally presented by Delmas, a simple hexagonal packing scheme with three
distinct oxygen positions (A, B, C) can be used to identify the cation geometry
between (MO2)n sheets. Assuming the oxygen in the (MO2)n sheets are in the A
and B positions, only two possibilities exist for the position of the third oxygen
sheet: B or C. When the third oxygen is located in the B position, either ABBA or
ABBCCA stacking motifs can exist depending on the position of the fourth oxygen
(either A or C), resulting in a trigonal prismatic coordination for the alkali metals
between the (MO2)n sheets. On the other hand, when the third oxygen is located
on a C position, an ABCABC stacking pattern is created with the alkali metals
adopting octahedral symmetry. The final case, which will not be further explored
here, exists when the third oxygen lies in the A position leading to an ABAB
stacking pattern that only occurs with A2MO2 stoichiometry and alkali metals in
tetrahedral coordination. As viewed down the c axis, the two and three layer
phases can be easily distinguished. In the two layer phases, all of the M ions lie in

159

the same positions on the a-b plane; however, in the three layer system the M ions
are staggered based on hexagonal packing into three distinct a-b coordinates.
Once the geometry of the alkali cation is established as either O (octahedral)
or P (trigonal prismatic), the number of (MO2)n repeating units per unit cell is
appended to give the structural type. Finally, a prime is added after the geometry
notation to inform of the presence of a monoclinic distortion of the unit cell. Often,
after a monoclinic distortion, the unit cell can be described using a one layer cell
(similar to the P1 structure in Figure 4-1).
The majority of known phases adopting the α-NaFeO2 type structure or one of
its deviations are shown pictorially in Figure 4-2. The dotted line running with slight
curvature towards the top right of the figure separates phases with P type structure
and geometry from those of O type. Different color markers represent different
transition metals. This Rouxel type diagram presents a relationship between bond
ionicity and A atom size that makes it easy to see when the trigonal prismatic
environment is dominant: large alkali/alkaline earth metals in the A position, when
vacancies in the A position are present, and when the lattice is more covalent than
ionic (3). The bond ionicities were calculated using Pauling’s formula (4):
Ib =1 - exp(-0.25(∆X)2 )

(4-1)

Where ∆X refers to the difference in electronegativity between the atoms in the
bond being studied (5). For example, the bond ionicity of NaCoO2 can be
calculated by multiplying the iconicity of Na-O by the ionicity of Co-O. Using the
electronegativity values of Na (0.93), Co (1.88), and O (3.44) in Equation 4-1 gives

160

ionicities of 0.793 for Na-O and 0.456 for Co-O (4). Multiplying these values gives
a total bond iconicity for NaCoO2 of 0.361, as shown in Figure 4-2.
As there are exceptions to the trends shown in Figure 4-2, a theoretical plot of
compounds yet to be produced is a bit forward. For example, in the material
K0.70CrO2, K1+ adopts octahedral coordination, contrary to the bond ionicities.
Similarly, Na0.70MnO2 shows Na1+ in trigonal prismatic coordination while firmly
placed in the octahedral coordination coordinate of the plot. These anomalies are
generally a result of either superstructural ordering or mixing of cations in different
positions.
This is a fairly rudimentary calculation of bond ionicity, which takes advantage
of the direct relationship between electronegativity difference and bond ionicity. A
more rigorous definition would be a self-contained atomic structure calculation
based on atomic radius and charge neutrality.

161

Figure 4-2. Rouxel type plot of bond ionicities vs. alkali metal/alkaline earth metal. Dotted line separates P type (left
side) from O type phases. Bond ionicities were calculated using the Pauling method (Equation 4-1) (4).

162

4.1a Review of α-NaFeO2 type and similar structures
The following sections give a brief survey of the literature as pertaining to
phases with the α-NaFeO2 type structure categorized by where the transition metal
falls in the periodic table.
The preparation and characterization of quantum spin systems on perfectly
geometrically frustrated lattices is a subject of considerable interest (6). NaTiO2 is
a canonical two-dimensional frustrated S=1/2 system, and as such may be a model
for the quantum spin fluid ground state proposed by Anderson (7). NaTiO2 adopts
the α-NaFeO2 structure which is an ordered variant of the rock-salt structure. In
this structure, layers of edge-sharing TiO6 octahedra and edge-sharing NaO6
octahedra alternate along the rock-salt [111] direction. In NaTiO2, there exists a
structural transition from the high-temperature (HT) rhombohedral phase (space
group R3̅ m) to the low-temperature (LT) monoclinic phase (space group C2/m)
through a broad, continuous transition from 220 K and 250 K (8). NaTiO2 formally
contains d1 Ti3+ with S=1/2 arranged on a triangular lattice with a separation of
about 3.04 Å.
The main body of literatures is focused on how different synthetic routes
influence the physical properties of NaxTiO2. A range of synthetic procedures from solid state reactions to soft chemistry - have been applied to prepare this
compound; however, it has been shown that different synthetic pathways result in
a different Na content, causing a change in the Ti3+/Ti4+ ratio (7, 9, 10).

163

Depending on the synthesis route, different starting materials such as Na2O,
Na2O2, Na and n-butyl sodium, and anatase TiO2 or Ti2O3 have been used.
Generally, the high temperature solid-state methods result in sodium deficient
phases due to the high sodium vapor pressure above 700 °C. The analogous Li1+
phase can be prepared by reducing anatase with nButLi in hexane at room
temperature. However, efforts to intercalate the larger Na1+ ion using sodium
naphthalide in tetrahydrofuran failed (7). Independent of synthetic method chosen,
phases in the entire NaxTiO2 oxidation range crystalize in the hexagonal R3̅ m
space group with cell parameters a = 3.00 - 3.05 Å and c = 16.23 - 16.44 Å.
Because of the incredibly low solubility of NaxTiO2 in concentrated acid and HF, a
direct correlation between cell parameters and sodium stoichiometry was
established (7). Due to the sensitivity of the a lattice parameter to the oxidation
state of titanium, sodium concentration can be calculated based solely on the
lattice parameters from high resolution diffractograms. This relationship has been
explored throughout the Na deficient range; a convenient way to prepare
stoichiometrically pure Na1.0TiO2 (a = 3.04365 Å, c = 16.2539 Å) by reacting
sodium deficient NaxTiO2 with sodium metal at 300 °C for an extended period of
time was first reported by Clarke (7). Afterward, additional Na was washed out with
liquid NH3.
The observed χm at room temperature for NaTiO2 is very small and relatively
temperature independent. However, on lowering the temperature, the magnetic
susceptibility starts to decrease steeply at 260 K followed by a gentle tail(8). Since

164

no hysteresis on cooling and heating was observed, the transition is believed to be
second order caused by a shift of the TiO6 and NaO6 planes relative to each other.
Heat capacity measurements were also consistent with a second order phase
transition at 260 K (9). A similar metal-metal bonding derived transition has been
reported for early transition metal oxides, including V2O3 and Ti2O3 (11). If these
transitions are similar in nature, the magnetic features of NaTiO2 are dictated by
correlated electrons, not by the spin frustration present in the structure. The
magnetic susceptibility of off-stoichiometry NaxTiO2 depends strongly on the a
lattice parameter and the synthetic route used to prepare the material. For
example, the magnetic susceptibility of Na1.2TiO2 falls off at 260 K while the
susceptibility of Na0.9TiO2 is temperature independent across all temperature
regions. Further investigations into the dependence of molar susceptibility on alkali
metal stoichiometry would be noteworthy and allow for more concrete
determination of Na1+ content.
The synthesis of NaScO2 has been reported once in the literature by the
Veilhaber group (12). Na2O2 and Sc2O3 were intimately fixed and fired at 600 °C,
resulting in phase pure material. No properties investigation was reported. The
overall lack of literature precedence is most likely due to the extreme scarcity of
scandium on the open market and the difficulty in preparing metallic Sc (13).
V2O3 and VO2 vanadium oxides were intensively studied in the 1970s owing to
the metal to insulator transition resulting from the presence of V3+ (3d2) and V4+
(3d1) ions in the corundum and rutile structures, respectively (14). The possibility

165

to introduce these ions into the layered oxide structure is therefore expected to
lead to a new family of materials with highly correlated electrons. The NaxVO2
system has been researched only very recently, certainly owing to the very
reducing character of the trivalent vanadium ions that makes its synthesis difficult
(15).
Synthesis of NaVO2 required two consecutive steps. Initially, NaVO3 was
synthesized via the solid state reaction of Na2CO3 and V2O5 by heating to the
melting point (630 °C). Afterward, NaVO3 powder was reduced chemically under
H2 at 700 °C for 8 hours. As trivalent vanadium inNaVO2 is hyper-sensitive to both
oxygen and moisture, the synthesis and characterization must be completed with
great care.
Vanadium oxides with layered (α-V2O5 and Na1+xV3O8) or tunnel (β-NaxV2O5)
structures have been demonstrated to reversibly cycle up to 1.6 Na (225 mAh/g),
although the average cell potential is low and capacity retention is poor (16–19).
Layered NaxVO2 (x = 0.7, 1) are reported to cycle 0.5 Na reversibly with a specific
capacity of 130 mAh/g (20). Voltage profiles indicate that sodium vacancy ordering
occurs in the layer as a function of Na content, a phenomena that creates twophase and solid-solution. When NaVO2 is cycled against a Na metal anode above
3V, a phase transition similar to that seen in LiCoO2 occurs that reduces the
specific capacity upon subsequent cycling (20). Recently, the prismatic phase P2Na0.70VO2 has shown tolerance towards cycling at high voltage compared to O3NaVO2 (21). The high electronic conductivity of P2-Na0.70VO2 is believed as the

166

origin of the reduced polarization considering that O3-NaVO2 is electronically
insulating.
Recently, rechargeable batteries with high energy density and low cost are in
great demand as energy sources for various applications. Lithium ion batteries
(LIBs) are being widely used in portable electronics and are considered as
promising power sources for future electric vehicles. However, the large-scale
commercialization of electric vehicles using LIBs might be hampered significantly
due to the relatively high cost and limited resources of lithium. In contrast, the earth
abundance of sodium (2.75%) is about three orders of magnitude higher than that
of lithium (0.002%). Because of this abundance, rechargeable sodium-ion
batteries (SIBs) are being considered as promising energy storage systems for
large-scale and cost sensitive applications. Many layered Na intercalated
compound with general formula NaMO2 (M = V, Cr, Mn, CO, Ni) are considered as
potential cathode materials for SIBs (22–26). Among them NaCrO2 is the most
promising (27). Electrochemical cycling of Na1+ in NaCrO2 shows a flat and smooth
charge/discharge voltage plateau similar to widely commercialized LiCoO 2. In
addition, NaCrO2 has higher thermal stability at the charged state compared to
LiCoO2 and even LiFePO4, providing a safer cathode material that is not prone to
thermal runaway. Interestingly, the analogous Li1+ phase with the same layered
structure is electrochemically inactive, due to the presence of Cr6+ in tetrahedral
coordination upon deintercalation of Li1+.

167

The electrochemical performance of NaCrO2 as a cathode material has been
extensively studied by several groups. The compound shows a reversible capacity
of 110 mAh/g between 2.5 and 3.6 V vs Na1+ / Na) with good capacity retention,
demonstrating that half of Na1+ can be reversibly extracted and inserted back per
NaCrO2 (28). Komaba attributed the high reversibility of NaCrO2 to the mismatch
between the size of Cr6+ ions and the interstitial tetrahedral site of the NaCrO2
structure (29, 30). Most of the research dedicated to NaCrO2 looks at structural
changes during electrochemical Na (de)intercalation (27).
The synthesis of NaCrO2 is straightforward: a stoichiometric mixture of Cr2O3
and Na2CO3 is heated at 900 °C under flowing argon for 5 hours, resulting in a O3
type structure (C2/m, a = 5.0178 Å, b = 2.8828 Å, c = 5.9027 Å and ß = 106.97 °)
(31). Structural changes of Na1-xCrO2 during electrochemical Na deintercalation in
was investigated in real time by in situ diffraction using a specially designed
electrochemical cell (27). According to the variations in Bragg peak positions, the
removal of Na1+ from the layered NaCrO2 results in consecutive structural phase
transitions between three solid solutions. For Na1-xCrO2, when x < 0.08, the solid
solutions crystalize in the O3 type structure with octahedrally coordination Na 1+.
Further Na1+ removal results in the emergence of a second phase with the O’3type structure with a monoclinic structure. In the 0.08 ≤ x < 0.2 range, two line
phases coexist, while for x = 0.25 the dominant phase is Na0.74CrO2. When a third
of the Na1+ is deintercalated from the structure, the P3 type phase begins to come

168

in due to the larger separation of the CrO2 slabs. As x increases to 0.40 and
greater, only the P3-type NaCrO2 is present.
The archetype compound of this crystal structure, α-NaFeO2, is a relatively well
studied compound, including magnetic, electrochemical, and ionic conductivity
measurements (32, 33). NaFeO2 is one of the oldest documented synthesized
solid-state materials, first being described in 1888 by Bernheim (34). The sodium
mobility in this structure has been investigated multiple times through different
types of topotaxy including hydrolysis, oxidation, cation exchange, and proton
exchange (35–38). Just as in other similar layer phases described in these thesis,
the large difference in the covalent M-O bonds and the ionic A-O bonds allows for
topotactic access to the A cations. The vacant tetrahedral positions in the rock
salt structure also provides hopping intermediate positions for the A cations, which
is generally understood to be the source of the high ionic conductivity reported
(32).
The synthesis of α-NaFeO2 is easily accomplished - a sodium precursor (NaOH
or Na2O2) in 10% molar excess is intimately mixed with Fe2O3 then fired in air at
500 °C for 24 hours. Any excess sodium precursor is easily removed by washing
with anhydrous methanol. Oxidation with Br2 in MeCN, a very common oxidation
method used to deintercalate Na from these layered phases, resulted in a
Na0.90FeO2 stoichiometry. Further deintercalation results in decomposition of the
phase.

169

A very nice study on the ionic conductivity of α-NaFeO2 was completed by
Blesa in 1999 (32). The ionic conductivity of the alpha ferrite was compared to a
very common sodium ion conductor NaInO2, where α-NaFeO2 showed a four
orders of magnitude increase in ionic conductivity (5.5 x 10 -5 S/cm) over NaInO2
(7.9 x 10-9 S/cm) (32). As the mechanism of Na1+ conductivity is the same in both
of these compounds - movement through a vacant tetrahedral position between
the two octahedral sites - the increase in ionic conductivity must be caused by a
crystallochemical reason.

The sodium indate would appear to be a better

conductor based on a larger unit cell volume (146.40 ų vs 127.56 ų); however,
the increased crystal radius of In3+ is substantially larger than Fe3+ (0.80 Å vs 0.64
Å), resulting in a smaller total free space for Na1+ movement (32, 39, 40).
The magnetic properties of NaFeO2 are very complicated, with two different low
temperature magnetic phases (33). The rich magnetic properties of this material
can be described by the common vein of many of these layered transition metal
oxides with Cd-I2 type layers: frustrated systems with multiple different competing
couplings of approximately equal magnitude. The 4d analog NaRuO2 has also
been synthesized although very little characterization has been reported (41). In
general, when a less electronegative ion is substituted, the metallic conductivity is
enhanced due to the reduced directionality in the bonding nature of the material.
Of the main group metals and metalloids, NaInO2 and related species are the
most studied by far. Neither Al or B form layered α-NaFeO2 type phases of formula
AMO2 due to their small crystal radii - the tunnel ß-NaFeO2 type phases are more

170

thermodynamically preferred in these cases. Furthermore, slight doping of both
Sn4+ and Pb2+ for In3+ in the AInO2 layered structure has been reported as alkali
solid electrolytes although higher doping led to phase disproportionation or nonstoichiometry (42, 43). The most obvious difference between the literature reports
of NaInO2 and related phases and the transition metal AMO2 phases is the lack of
magnetic investigation - the spin frustration present in the strongly correlated
transition metal materials has been alleviated by diamagnetic In 3+; however, a
somewhat rich suite of properties has been published recently.
Synthesis of NaInO2 is straightforward: equimolar amounts of Na2CO3 and
In2O3 were fired at 900 °C for 1 hr to avoid Na loss due to vaporization, resulting
in the O3 type structure with a = 3.227 Å and c = 16.38 Å (40). The analogous Li1+
containing phase can be produced similarly by mixing Li2CO3 with In2O3 and
sintering at 1000 °C (44). Topotactic cation exchange of H1+ for Na1+, as well as
anion exchange of S2- for O2-, have been reported, giving phases with
H0.6Na0.4InO2 and NaInS2 respectively (40, 45). A rarely used technique in solid
state chemistry, the low temperature facile exchange of H1+ for alkali metals in
extended solids employing molten benzoic acid saw great success in this system.
Trivial isolation of the ion exchange product from the sodium benzoate carboxylate
salt was completed by washing with hot methanol and water, affording the O3 type
H0.6Na0.4InO2 pure phase material with a slightly compressed c parameter (16.35
Å) relative to the parent phase. Attempts at further ion exchange by increasing the
temperature above the melting point of benzoic acid (> 180 °C) led to phase

171

decomposition most likely caused by layer exfoliation as a result of hydrogen
binding to only one oxygen as opposed to the octahedrally coordinated Na 1+ it
replaced (45).
In recent literature reports, NaInX2 (X = O, S) has been heavily investigated as
a photocatalyst for water splitting and organic molecule degradation. A series of
Na1-x(Li,K)xInO,2 (0.2 >= x >= 0) were synthesized by the previously mentioned
solid state route then impregnated with Ru3(CO)12 in tetrahydrofuran solution (46).
After oxidation in air at 400 °C, 1% RuO2 by weight was loaded on the indates.
NaInO2 was chosen as the scaffold for RuO2 impregnation due to its layered
structure, which has been previously reported to allow for ideal RuO 2 particle
dispersion (46, 47). A closed system hooked up to a gas chromatograph was set
up where 250 mg of the solid was suspended in distilled water then illuminated
with a Hg-Xe lamp. The highest water splitting activity was found for the pristine
RuO2 impregnated NaInO2 (0.8 μmol/hr) followed by K1+ doped Na1-xKxInO2 (0.7
μmol/hr). Without RuO2 particles on the surface, no catalytic activity was observed
for NaInO2 under any conditions. Interestingly, Sato et al. also studied the effects
of NaInO2 crystallinity on catalytic activity, finding that a calcination temperature of
1200 °C led to the highest catalytic activity rates due to the highly crystalline nature
of the NaInO2 sample prepared at this temperature. Both higher and lower
calcination temperatures led to the formation of jagged particles which increased
the occurrence of surface trap states, reducing catalytic activity by crippling charge
separation rates. With further optimization, impregnated NaInO2 may be a viable

172

catalyst for water oxidation, paving the way for more widespread use of H 2 as a
green energy source (46).
NaInO2 has also been investigated as a potential organic molecule
decomposition catalyst because of the efficient separation of excitons attributed to
the layered α-NaFeO2 structure. The photocatalytic activity of NaInO2 for
methylene blue decomposition under UV light was studied by Nomani et al (48).
NaiNO2 was impregnated with 0.5% RuO2, NiO, and Pt, then suspended in a
methylene blue solution. Photocatalytic activity was monitored via UV-Vis
absorption spectroscopy due to the strong absorbance of methylene blue at 650
nm. After 60 minutes, all four samples demonstrated methylene blue
decomposition while the Pt impregnated sample completely discolored the
solution. When codoped with RuO2, NiO, or Pt, the NaNiO2 powder showed a
substantial increase in photocatalytic activity due to the enhanced photoinduced
electron hole spatial separation; therefore, the recombination rate of the holeelectron pairs was greatly reduced. Using previous literature reports of the
photocatalytic decomposition of methylene blue, two mechanisms can be
proposed (49). After photoinduced generation of an electron-hole pair the holes in
the valence band can react with H2O to produce ∙OH or the electrons in the
conduction band can react with dissolved O2 to produce superoxide radicals. Either
of these radical species has been shown to quickly decompose methylene blue
under illumination.

173

4.1b Dependence of oxidation state on properties
The most studied system by far, NaxCoO2 demonstrates a rich range of
properties depending on the Na concentration (50). Considering the oxidation state
of Co can be very finely controlled through chemical or electrochemical
deintercalation of Na from the crystal lattice, thereby controlling the hole density,
NaxCoO2 presents a replete system to study the structure-properties relationship
in a frustrated triangular lattice.
In general, NaxCoO2 (x >= 0.60) phases are prepared using classic solid state
methods. Stoichiometric amounts of Co metal and a 10% excess sodium source
(NaOH, Na2O2) were intimately ground under inert atmosphere and fired under O 2
atmosphere. A large body of literature has been devoted to the topotactic methods
of Na intercalation and deintercalation into the NaxCoO2 framework to specifically
tune the Co oxidation state. These tools give great structural control to the
inorganic chemist because it allows for the synthesis of phases with similar
stoichiometries yet different atomic coordinations. For example, Na0.50CoO2 is
easily synthesized, crystallizing in both P'3 and P2 type symmetries depending on
the reaction conditions. When NaCoO2 (O3) is oxidized with I2 in anhydrous MeCN
for two days, Na0.50CoO2 (P'3) is produced. These two structures - O3 and P'3 are easily converted by shifting the CoO2 layers about 1 Å parallel to the a axis.
On the other hand, oxidation with I2 under the same conditions results in the twolayer P2 type Na0.50CoO2 phase. Furthermore, single crystals of P1 type
Na0.51CoO2 were synthesized when NaCoO2 (P2) was chemically oxidized with

174

H2O2. Structural control of the highly oxidized cobaltate, Na0.34CoO2, is similarly
available: oxidation of O3 NaCoO2 with Br2 in MeCN afforded O3 Na0.34CoO2.
Overall phase composition can be further tuned by modifying oxidation time or
intercalating Na using common reducing agents such as Na-benzophenone.
At first glance, it appears that any NaxCoO2 (0.34 <= x <= 1) with either
octahedral or prismatic coordination can be easily chemically synthesized. As seen
in Table 4-1, this is not the case. For the two layer system (P2 type phases), solid
solution phases with trigonal prismatic symmetry are easily synthesized across
essentially the entire range of x values except for a small island of instability at
high Na compositions. As reported by the Cava group, the three layer system
shows much narrower solid solution limits with two major subgroups: octahedral
coordination when x > 0.75 and x < 0.4 sandwiching prismatic P1 type structures
when 0.75 > x > 0.4 (51). Intercalation and deintercalation of Na in the three layer
system is therefore much more complicated, as synthesis of phases outside of the
solid solution limits of each structural type generally leads to biphasic products.
In these frustrated cobaltates, interplanar length seems to play a key role in
material properties - larger distances between CoO2 planes reduce interplanar
interactions relative to intraplanar interactions. Two competing forces dictate how
the interplanar distance changes depending on the stoichiometry of these phases:
Coulombic interactions between Na1+/Co3+/4+ and O, and the O2- - O2- repulsion in
adjacent CoO2 layers. As the NaO2 layer becomes Na1+ deficient, an increase in
repulsion between adjacent CoO2 layers accompanied with a decrease in

175

Coulombic interactions leads to an increase in interplanar spacing in both the twolayer and three-layer systems. This increase is accompanied by shorter intraplanar
Co-Co distances and Co-O bond lengths due to the smaller crystal radii of Co4+
relative to Co3+ (39).

Figure 4-3. Graph of the figure-of-merit Z values for the in NaxCoO2 series of
layered cobaltates. Used with permission from (52).

176

Table 4-1. Summary of structure types, space groups, cation coordination, and the
Co oxidation state in the three layer and two layer families of the NaxCoO2 system.

Composition

A Coordination

Structure
Type

Space
Group

Co
Oxidation
State

Three Layer Family
NaCoO2 (53)

Octahedral

O3

R3̅ m

3+

Na0.92CoO2 (2)

Octahedral
Trigonal
Prismatic
Trigonal
Prismatic
Trigonal
Prismatic
Trigonal
Prismatic
Trigonal
Prismatic
Octahedral

O3

R3̅ m

3.08+

P’1

C2/m

3.25+

P’1

C2/m

3.33+

P’1

C2/m

3.40+

P’1

C2/m

3.48+

P’1

C2/m

3.50+

O3

R3̅ m

3.68+

P2

P63/mmc

3+

P2

P63/mmc

3.15+

P2

P63/mmc

3.20+

P2

P63/mmc

3.24+

P2

P63/mmc

3.29+

P2

P63/mmc

3.37+

P2

P63/mmc

3.47+

P2

P63/mmc

3.50+

P2

P63/mmc

3.66+

Na0.75CoO2 (2)
Na0.67CoO2 (54)
Na0.60CoO2 (2)
Na0.52CoO2 (53)
Na0.50CoO2 (2)
Na0.32CoO2 (2)

Two Layer Family
NaCoO2 (1)
Na0.85CoO2 (1)
Na0.80CoO2 (1)
Na0.76CoO2 (1)
Na0.71CoO2 (53)
Na0.63CoO2 (1)
Na0.53CoO2 (53)
Na0.50CoO2 (55)
Na0.34CoO2 (1)

Trigonal
Prismatic
Trigonal
Prismatic
Trigonal
Prismatic
Trigonal
Prismatic
Trigonal
Prismatic
Trigonal
Prismatic
Trigonal
Prismatic
Trigonal
Prismatic
Trigonal
Prismatic

177

When Co is in the 3+ oxidation state – NaCoO2 – an unremarkable
nonmagnetic, nonmetallic material exists. However, when slightly oxidized,
NaxCoO2 (1 >= x >= 0.80) exhibits highly interesting thermoelectric properties. Lee
et al recently published a comprehensive study of the thermopower (S) and figure
of merit (Z - Equation 4-2), in which they found a direct relation between the
thermoelectric properties and Na doping in the structure (Figure 4-3).
2

Z=S /ρκ

(4-2)

The curves in Figure 4-3 represent different x values in NaxCoO2 according to
the following: 1(x = 0.71), 3(x = 0.80), 4(x = 0.85), 5(x = 0.88), 6(x = 0.89), 7(x =
0.96), 8(x = 0.97). At approximately 60 K, a maximum Z is reached for all samples
with Na0.88CoO2 having the highest Z of 1.8 x 10-3. A large jump in the in-plane
resistivity between samples 6 and 7 is responsible for the dramatic drop in Z for
NaxCoO2 where x > 0.95. However, resistivity curves with metallic like shape were
observed across the doping region studied. The authors propose that the jump in
resistivity in samples close to the ideal NaCoO2 stoichiometry is due to charge
carrier segregation into layers, thus leading to a lower degree of electrical
connection between layers. Understanding the reason for a 40 fold increase in the
figure of merit for Na0.88CoO2 relative to NaCoO2 is far from complete, intimately
depending on both charge carrier interactions as well as the geometric frustration
inherent to the structure.

178

When 25% of the Na is deintercalated from NaCoO2, effectively oxidizing Co to
3.33+, a curious metallic state manifests. The in-plane susceptibility of Na0.75CoO2
follows Curie-Weiss law above 25 K, yet the resistivity is linear with temperature
above 25 K. Essentially, from the susceptibility data, it appears that in Na 0.75CoO2
the magnetic susceptibility is dictated by the total number of positive spin carriers,
whereas in a metal the susceptibility would quickly fall as the temperature
decreased due to lack of population of aligned spins. This mixture of Curie-Weiss
paramagnetic susceptibility with metallic resistivity has garnered Na 0.75CoO2 (and
similar Na composition) the title “Curie-Weiss metal”.
Further oxidation of the sodium cobalt oxide lattice results in the x = 0.5 phase
Na0.50CoO2, an insulator that separates the two metallic regions on either side.
Deintercalation of Na1+ from Na0.74CoO2 gives Na0.50CoO2 with a hexagonal
structure (a = 2.81 Å, c = 11.1 Å, P2 type), with Na1+ occupying both 2b and 2c
sites with near equality. An orthorhombic supercell was established from faint
reflections in the neutron diffraction data corresponding with perfectly ordered Na1+
atoms – a structural feature that defies an obvious explanation. A low temperature
structural study of Na0.50CoO2 identified the reason for the Na1+ ordering below 300
K: induced Na1+ ordering based on charge ordering of the Co3+/4+ ions in the lattice
(55). The Na1+ ordering was confirmed by thermal conductivity measurements. As
in most layered oxides of this structural type, the thermal conductivity is primarily
carried by phonons; therefore, in an ordered material the phonon mean free path
would be much longer than a disordered structure. The in-plane thermal

179

conductivity of Na0.50CoO2 does indeed increase below the disorder to order
transition temperature to 300 W/mK, a similar value to that of quartz and diamond.
Synthesis of the highly oxidized Na0.34CoO2 was accomplished by reacting
Na0.74CoO2 with Br2 in MeCN. The structure is similar to that of Na0.92CoO2 and αNaFeO2 - three layer structure with octahedrally coordinated Na1+ (O3). The
magnetic properties of Na0.34CoO2 were reported in the literature - molar
susceptibility independent of temperature - but were never explained due to a
serendipitous discovery of superconductivity in 2003. A collaboration of materials
science groups in Japan were working on layered cobaltates and measuring the
magnetic properties of Na0.34CoO2 to try and further elucidate the reason for the
independent susceptibility; however, they handled the highly water-sensitive
sample in air, allowing 1.47 water molecules per formula unit to intercalate into the
12k sites of the structure. Magnetization measurements in a field of 1000 Oe
showed a large diamagnetic signal below 5 K (Figure 4-4). This superconductivity
is postulated to occur because of the large increase between CoO2 layers upon
intercalation of two layers of water, effectively localizing the 2d CoO2 layers similar
to the CuO2 square planar superconductors. A direct comparison between these
two structural types is null, as the CoO2 slabs adopt a frustrated triangular lattice
while the CuO2 square planes are ordered antiferromagnetically.

180

Figure 4-4. Magnetic susceptibility (x) of NaxCoO2·yH2O. Filled circles, zero-field
cooling; open circles, field cooling. The susceptibility was measured in a magnetic
field of 20 Oe. The inset shows x measured under various magnetic fields by the
zero-field cooling process. Reused without adaptation from (56).

There have been multiple accounts of directly controlling the Na1+ stoichiometry
in NaxCoO2 by intercalating Na1+ back into the structure after oxidation. This has
been used extensively at higher x values in hopes of increasing the thermopower
of NaxCoO2, and has allowed for the large scale synthesis of Na0.88CoO2.
In an attempt to develop more rugged structures, Wiley et al initially proposed
the aliovalent exchange of a divalent alkaline earth metal for two alkali metals in

181

the NaCoO2 family of compounds (53, 57). Compared to previous reports of
aliovalent exchange predominately performed in aqueous solutions, the ability to
synthesize moisture sensitive materials in molten salt fluxes with controlled
vacancies is obviously useful to the solid state chemist. Lamellar calcium cobalt
oxides were prepared by intimately mixing the sodium precursor NaxCoO2 (x = 1,
0.71, 0.6, 0.53, 0.52) with an excess of anhydrous Ca(NO 3)2 in an evacuated,
sealed Pyrex tube and heated at 310 °C for 48 hrs (Equation 4-3, Table 4-2) (53,
57).
NaxCoO2 + x/2 Ca(NO3)2 → Cax/2CoO2 + xNaNO3

(4-3)

The Cax/2CoO2 phases were isolated by washing out the nitrate byproduct with
anhydrous methanol then stored under inert atmosphere. Interestingly, except for
Ca0.52CoO2, the calcium product phases and the reactant sodium phases were
indexed in the same space groups, demonstrating the topotactic nature of the
reactions. In the case of Ca0.52CoO2, the monoclinic distortion from O3 → P’3 can
be reversed by a shift in the CoO2 layers along the a axis. The reason for this
transformation, originally proposed by Delmas, is due to the enhanced thermal
stability of the A cations in trigonal prismatic coordination as the CoO 2 interlayer
spacing increases.

182

Table 4-2. Unit cell parameters of sodium parent phases and calcium exchange
phases discussed in this chapter (53, 57). Notice the topotactic nature of the
aliovalent exchange preserves the structural type between sodium and calcium
phases.
Precursor

NaCoO2 (O3)

Na0.70CoO2 (P2)
Na0.60CoO2 (P’3)

Na0.53CoO2 (P2)
Na0.52CoO2 (P’3)

Unit cell (Å)
a = 2.886
c = 15.600
a = 2.834
c = 10.882
a = 4.904
b = 2.827
c = 5.709
ß = 106.1 °
a = 2.820
c = 11.05
a = 4.889
b = 2.821
c = 5.751
ß = 106.2 °

Exchange Product
Ca0.52CoO2 (P’3)

Ca0.35CoO2 (P2)
Ca0.30CoO2 (P’3)

Ca0.27CoO2 (P2)
Ca0.26CoO2 (P’3)

Unit cell (Å)
a = 4.881
b = 2.826
c = 5.654
ß = 105.8 °
a = 2.812
c = 10.87
a = 4.870
b = 2.811
c = 5.664
ß = 106.6 °
a = 2.812
c = 10.87
a = 4.961
b = 2.805
c = 5.671
ß = 106.7 °

As shown herein and elsewhere, as a compound with an α-NaFeO2 type
layered structure is oxidized by removing the A cations, approaching the
stoichiometry A0.50MO2, a general crystallinity loss is observed. In the case of Co,
the culprit of this decomposition is the presence of a majority of Co4+, not vacancies
in the AO2 layers. However, since the oxidation state of Co is preserved during
aliovalent exchange, a larger number of vacancies are created in the AO 2 layer
while preserving less than ½ of the Co sites in the Co4+ oxidation state; therefore,
structural deformations due to the presence of Co4+ are avoided. Furthermore, the

183

Ca2+ ions act as structural pillars – the strong interaction between Ca2+ and O2between layers effectively holds the structure together. This observation was
experimentally demonstrated by comparing the CoO2 interplanar distance upon
different Ca2+ concentrations in CaxCoO2 – the interplanar distance (approximately
5.43 Å) is both constant and independent of Ca2+ stoichiometry (53).
Multiple reports of the aliovalent exchange of Sr2+ for Na1+ have recently
surfaced in the arXiv repository, focusing on the phase Sr0.35CoO2 (58, 59).
Polycrystalline samples of Sr0.35CoO2 were synthesized in a similar fashion as the
analogous Ca2+ compounds reported herein and elsewhere: Sr(NO3)2 and
Na0.7CoO2 were intimately mixed and allowed to react at 310 °C for two days in
air. The publications report on an interesting phenomenon occurring in Sr0.35CoO2
where the annealing temperature has a direct correlation on the band structure of
the material. When the sample is annealed at 400 °C under O 2, a metallic state
arises with exceptionally low resistivity; however, when annealed in air at 300 °C,
the material is a standard semiconductor. The authors propose that the difference
in oxygen stoichiometry generated between the two firing conditions is the cause
for this change in resistivity without any further investigation. In reality, and in
agreement with decomposition products found when performing the Ca2+
exchange with NaxCoO2 at T > 350 °C, decomposition of the phase at the grain
boundaries is most likely leading to a decrease in resistivity. Obvious impurities
found in the PXD patterns reported support this idea.

184

Figure 4-5. Schematic of the general four step MuRE method. In the chemical formulae, M represents any 3d
transition metal. Colored spheres represent the following atoms: grey – Na1+, teal – Mx+, blue – Ca2+, yellow –
Na1+/Li1+. MO6 octahedra are shown in teal for clarity.

185

4.2 The multistep reductive exchange (MuRE) method
The general four step MuRE method is depicted in Figure 4-5. The procedure
begins with the synthesis of the NaM3+O2 precursor phase by the appropriate solid
state methods described previously in this review. Sodium containing phases were
chosen over Li1+ or K1+ because all NaM3+O2 phases with 3d transition metals
crystallizing in the α-NaFeO2 type structure are well ordered and easily
synthesized. The second step of the procedure is applicable when a final
stoichiometry other than A0.5A’0.5MO2 is desired because the Na1+ content in the
structure before aliovalent exchange directly determines the final phase
stoichiometry. Depending on the stability of the transition metal in the 3+/4+
oxidation state, different amounts of Na1+ can be deintercalated from the structure
using common oxidants (i.e. I2, Br2, NOBF4, NO2BF4), resulting in NaxCoO2 (1 >=
x > 0). The third step of the procedure is the aliovalent exchange of Ca 2+ for 2 Na1+,
thereby creating vacancies in the AO2 planes of the structure. Multiple groups have
shown success in the exchange of Sr2+ for Na1+ as well, although partial aliovalent
exchange for any alkaline earth cation has not been reported (58). Due to the
metastability of these phases and the desire for topotactic exchange with structure
retention, the aliovalent exchange is performed at as low of a temperature possible
by exploiting the eutectic between Ca(NO3)2 and NaNO3 (214 °C). The final step,
intercalation of alkali metals into the vacancies created during the aliovalent
exchange, results in structures with transition metals in the 2+/3+ oxidation states
while maintaining the layered α-NaFeO2 structure. Because complete exchange of

186

Ca2+ for Na1+ occurs, the final product can contain any alkali metal. An example
set of reactions for the synthesis of AxCa0.25CoO2 is presented in Equations 4-4 –
4-8. The side products after alkali intercalation using

nBuLi

or the sodium

benzophenone ketyl, octane and benzophenone, are written to balance the
reaction only. The actual side products are a much more complex mixture that is
outside of the scope of this review.
1/3 Co3O4 + NaOH + 1/12 O2 → NaCoO2 + 0.5 H2O

(4-4)

NaCoO2 + 0.25 I2 → Na0.5CoO2 + 0.5 NaI

(4-5)

Na0.5CoO2 + 0.25 Ca(NO3)2 → Ca0.25CoO2 + 0.5 NaNO3

(4-6)

Ca0.25CoO2 + 0.75 nBuLi → Li0.75Ca0.25CoO2 + 3/8 C8H18

(4-7)

Ca0.25CoO2 + 0.67 C13H10ONa → Na0.67Ca0.25CoO2 + 0.67C13H10O

(4-8)

The diffractograms of the extremely air sensitive samples Li0.75Ca0.25CoO2 and
Na0.67Ca0.25CoO2 were both measured with the diffractometer in an N2 filled
purgebox at a 0.5 s scan rate to minimize any oxidation or hydrolysis. Complete
filling all vacancies was not possible in some cases – especially when the larger
Na1+ ion was intercalated. This trend will be discussed further but may be caused
by large particle size, which leads to a core-shell type situation where the shell is
much more intercalated than the core resulting in an overall non-stoichiometric
phase.

187

All compounds were prepared in a nitrogen filled glovebox and, if needed,
transported to a furnace in a sealed alumina tube. NaOH (Alfa Aesar, 99%),
Na2CO3 (Sigma Aldrich, 99.9%), Br2 (Jade Chemical, ACS Grade), I2 (Jade
Chemical, ACS), Co metal (Alfa Aesar, 99.99%), Na metal (Mallinckrodt, ACS
Grade), n-BuLi (Sigma Aldrich, 1.6M in hexanes), and benzophenone (Sigma
Aldrich, 99%) were used as received without further purification. Ca(NO 3)2•4H2O
(Sigma Aldrich, 99.9%) was thermally treated at 500 °C in dry air for 2 hours to
afford anhydrous Ca(NO3)2.
Synthesis methods for the precursor phases are presented in the following few
paragraphs. 1.600 g (40 mmol) NaOH and 2.63 g (10.91 mmol) Co3O4 were ball
milled for 30 minutes, then fired at 500 °C for 120 hours under O 2, affording pure
phase NaCoO2 after washing out excess of NaOH with anhydrous methanol.
Na0.74CoO2 was prepared by heat treatment of a mixture of 1.908 g (18 mmol)
Na2CO3 with 3.77 g (15.66 mmol) Co3O4 for 48 hrs at 850 °C under O2 flow. The
more oxidized cobaltates (x = 0.50, 0.34) were synthesized in 5 g batches by room
temperature reaction with a 10 molar excess of I2 and Br2 in acetonitrile for 96 hrs
and 108 hrs, respectively. The NaI/NaBr was removed by washing with dry
acetonitrile. Na0.74CoO2 can also be prepared by the staged deintercalation of Na1+
from NaCoO2 using I2 in acetonitrile; however, the solid state method is more facile.
The sodium cobaltates were stored in a glovebox to prevent hydrolysis over time
from reaction with atmospheric water.

188

The aliovalent exchange of calcium for sodium was completed by intimately
mixing the sodium precursor with 4 equivalents of previously dried Ca(NO 3)2 in a
glovebox, then firing the powder under O2 for 48 hours at 310 °C. The reaction
products were washed with water to remove the nitrate salts, then dried under
vacuum and stored in a nitrogen filled desiccators.
Calcium exchange phases containing Fe as the transition metal were
synthesized similarly with one change – the calcium exchange was performed at
225 °C rather than 310 °C to avoid decomposition.
Topotactic chemical intercalation of Na and Li into the vacancies of the layered
framework was performed using standard organometallic reducing agents at room
temperature in the glove box. A 20 mL glass vial was charged with 250 mg of the
calcium phase and 10 molar excess of either 1.6 M nBuLi in hexanes or 1M Nabenzophenone in THF, depending on the desired product, and a magnetic stir bar.
The reaction was allowed to stir for up to 10 days depending on the phase and
final composition desire. After the reaction was complete, the final products were
washed with 5 20 mL aliquots of pentane for the Li intercalation or THF for the Na
intercalation reactions to remove excess starting material.
NaFeO2 was prepared by grinding 10 g (62.6 mmol) Fe2O3 and 5.1 g (125
mmol) NaOH under N2, heating the mixture in an alumina boat to 330 °C over 5
hours, holding at 330 °C for 10 hours, then raising the temperature to 500 °C and
firing for 48 hours under O2. The product was washed with anhydrous ethanol
under N2 to remove excess NaOH.

189

All Powder X-ray diffraction (PXD) patterns were obtained on a Bruker D8Advanced New from 15°- 75° 2θ with Cu radiation (λ = 1.5418 Å), a 2mm beam
slit, and a step size of 0.02° 2θ. A plastic dome with a lead coated air-scattering
slit was mounted on a zero-background sample holder (Si wafer) to afford the airtight sample holder used during all measurements.
Inductively-Coupled Plasma Atomic Emission Spectroscopy (ICP-OES) was
used to determine the elemental composition of the synthesized compounds on a
Varian Vista-MPX CCD. Standard solutions of 1ppm, 10ppm, and 100ppm were
prepared from stock solutions of 1000 ppm Li, Na, Ca, and Co obtained from Sigma
Aldrich (Puratronic Tracert, standardized). Samples were run in quintuplicate to
afford the emission intensity used to derive concentration from calibration curves
of the standards.
Thermogravimetric Analysis (TGA) and Differential Scanning Calorimetry
(DSC) measurements were performed on a Netsch Jupiter STA-449 F3 TGA/DSC
instrument with ultrapure N2 as a buffer gas. Buoyancy effects were corrected for
using an empty alumina sample holder as a blank. Measurements were run with a
5 °C/min step size.
The average magnetic moment of Co in each of the samples was determined
using a Quantum Design MPMS. Samples were prepared in gel capsules under
N2 in a glovebox to avoid oxidation. Susceptibility vs. temperature (Χ vs. T) was
measured at both 1000 Oe and 50000 Oe while susceptibility vs. magnetic field (Χ

190

vs. H) measurements were taken in a scanning field from -50000 Oe to 50000 Oe
at 5 K and 300 K.
Electrochemical cycling performance of the lithiated calcium cobaltates was
completed using a coin cell type setup with a Celgard 2400 membrane, a 1 M LiPF6
ins EC:DC electrolyte, and a Li metal anode. The cells were prepared in an Ar filled
glovebox, then connected to a potentiostat in galvanostatic mode. The voltage was
scanned from 3 V to 4.5 V with a step rate of 0.042 mV/s, equivalent to a C/10
scanning rate, for 4 cycles and a step rate of 0.42 mV/s (1C rate) for 50 cycles.
Electrochemical performance of the sodium intercalated analogs was also
investigated using the same experimental setup with a Na chip used as the anode
instead of Li.

191

4.3 Fe2.5+ Phases prepared with the MuRE method
The initial phase, NaFeO2, was prepared by ball milling 10 g (62.6 mmol) of
Fe2O3 with 5.01 g (125 mmol) NaOH in agate jars for 1 hour at 400 rpm. The jars
were opened under nitrogen and the contents were placed in an alumina boat. The
boat was loaded into a tube furnace and the following heating profile was initiated
under O2: raise to 330 °C at 65 °C/hour and hold for 10 hours to allow the sodium
hydroxide to melt without crawling up the sides of the boat, heat to 500 °C at 100
°C/hour, and hold for 48 hours. This procedure was modified from the literature
method only in the fact that the sodium hydroxide was allowed to melt before
increasing the temperature. Refinement of phase pure NaFeO2 obtained from this
procedure is presented in Figure 4-6.

192

Figure 4-6. Rietveld refinement PXD data for α-NaFeO2 performed using the
GSAS software package. Peak anisotropy from preferred orientation is shown in
the difference plot (grey line).

The refinement of NaFeO2 converged with χ2 = 1.43 and Rwp = 2.87%. The
refined cell parameters (R3̅m, a = 3.0248(2) Å, c = 16.093(1) Å) matched well with
the previously reported cell parameters (R3̅m, a = 3.0251 Å, c = 16.094 Å) (33). A
small amount of peak anisotropy was present in all peaks – a slight decrease in
the fitting statistics was observed when preferred orientation along the (003) plane
was used in the refinement. The standard hexagonal setting with atomic positions

193

Na (0 0 0), Fe (½ ½ ½), and O (0.2338 0.2338 0.2338) was used. Since the
compound diffracts very well, possible impurities of Fe3O4 or Fe2O3 are not
apparent in the diffractogram. To check for purity of the sample, low temperature
magnetization curves were obtained in a 1000 Oe field. The magnetic data closely
matched previous reports, verifying the purity of the sample.

4.3a Ca0.50FeO2 characterization and intercalation
Initial attempts at Ca-for-Na exchange were unsuccessful because of the
decomposition of the produced phase (Ca0.50FeO2) to Fe2O3 and CaO at 310 °C
under O2, as confirmed in the impure PXD where the main peak for Fe2O3 is seen
approximately 27° 2θ. Once the reaction temperature was lowered to 225 °C – just
above the eutectic temperature of Ca(NO3)2 and NaNO3 - pure phase Ca0.50FeO2
was produced after 48 hours under O2 (60). The powders were intimately ground
in the glovebox before firing, but were not pelletted for the first reaction. When
pelletted, the reaction speed increased dramatically, resulting in pure phase
material after just 24 hours. After the products cooled to room temperature, the
material was ground with a mortar and pestle under anhydrous methanol to
remove the NaNO3 present after exchange. The final stoichiometry was
determined by ICP-OES. PXD patterns of the Fe family of compounds prepared
using the MuRE method are presented in Figure 4-7.

194

Figure 4-7. PXD spectra of the Fe based family of materials prepared using the
MuRE method presented in this chapter. The background (broad hump, peak near
16 °2θ) in Na0.50Ca0.50FeO2 was due to the dome-type sample holder used during
collection.

195

Table 4-3. Cell parameters of Ca0.50FeO2 obtained from the LeBail refinement of
powder X-ray and neutron diffraction data at different temperatures (61). Full
Rietveld refinement was undertaken as well.
Method

a (Å)

b (Å)

c (Å)

ß (°)

PXD

5.1743(8)

5.9532(4)

5.8215(3)

73.21(2)

NPD 75K

5.1453(3)

5.9442(7)

5.8060(1)

73.19(4)

NPD 12 K

5.1321(1)

5.9385(4)

5.7924(2)

73.16(3)

Indexing of Ca0.50FeO2 was successful, resulting in a C2/m monoclinic unit cell
with a = 5.1743(8) Å, b = 5.9532(1) Å, c = 5.8215(3) Å, and ß = 73.21(2) °. Rietveld
refinement of Ca0.50FeO2 was completed on both PXD and NPD patterns (75 K
and 12 K) obtained at Oak Ridge National Lab on the POWGEN beam line in
collaboration with Dr. Jason Hodges. Rietveld refinement of the PXD data
converged with χ2 = 1.79. The cell parameters obtained from refinement of the
diffraction data sets are presented in Table 4-3 – the small discrepancy between
the PXD and NPD derived cell parameters is due to the measurement
temperatures: the PXD spectra were recorded at 298 K, while the NPD data was
recorded at both 75 K and 12 K.

196

Figure 4-8. Rietveld refinement of the PXD data of Ca0.50FeO2 showing the
observed pattern (black crosses), refined fit (red trace), difference curve (grey
below), and Bragg diffraction positions (magenta tickmarks, below).

PXD least squares refinement of the Ca0.50FeO2 diffraction pattern was
straightforward and converged with the following statistics: χ2 = 1.79, Rwp = 5.89
%, and Rp = 6.72 % as seen in Figure 4-8. A small amount of non-systematic peak
anisotropy was observed in the pattern, attributed to a mixture of preferred
orientation and peak broadening as a result of a smaller particle size distribution.
Atomic positions of Ca0.50FeO2 in the C2/m space group were refined as well and

197

are presented in Table 4-4. The large thermal parameters obtained are a
consequence of the relatively poor diffraction pattern statistics. Observed and
calculated Bragg positions are compared in Table 4-4. Due to the unreliable
refinement, the model obtained from this PXD data was further used in the neutron
refinements. It is difficult to directly compare the structures of NaFeO2 and
Ca0.50FeO2 based on cell parameters due to the large differences in unit cell
symmetry; however, the layered structure was preserved as seen from the atomic
positions and diffraction pattern.
Refinement of the neutron data was exceptionally difficult for two reasons: poor
counting statistics at low d spacing and large angle-dependent peak asymmetry.
These problems were overcome by careful profile refinement, which will be
explained over the next few paragraphs.

Table 4-4. Refined atomic positions for Ca0.50FeO2 obtained by Rietveld
refinement of NPD data.
Atom

Wyckoff
Position

X

y

z

Uiso 102 (Å2)

Occ.

Ca2+

2b

0

0

0.5

0.0184(2)

0.50

Fe3+

2a

0

0

0

0.0161(3)

0.50

Fe3+

2c

0.5

0

0

0.0153(1)

0.50

O2-

8j

0.36

0.25

0.2

0.0212(7)

0.34

O2-

4i

0.17

0

0.21

0.0216(8)

0.33

198

Table 4-5. Observed and calculated Bragg reflections for Ca0.50FeO2 in the C2/m
space group as proof of correct indexing. Indexing of the PXD data was done using
the TOPAS software package.
Calculated °2θ

Observed °2θ

15.887

15.897

23.342

23.338

25.112

25.121

29.866

29.815

31.361

31.358

32.144

32.148

34.081

34.082

35.161

35.165

36.323

36.243

40.976

40.978

44.375

44.382

46.823

46.901

47.822

47.815

49.375

49.402

50.197

50.185

52.019

52.017

54.259

54.248

55.535

55.521

56.519

56.498

57.892

57.885

199

Initially, the cell parameters obtained from the PXD refinement of Ca 0.50FeO2
were added as a phase to GSAS, followed by the addition of the two NPD
histograms for Ca0.50FeO2 recorded at 75 K and 12 K. Since the NPD data was
recorded as time of flight, a diffractometer constant of 22778.326 s was used to
convert the time of flight data to d-spacing with Equation 4-9.
t = DIFC*d + DIFA*d2 + ZERO

(4-9)

In the above formula, DIFC can be approximated if the total neutron path length
and the scattering angle are known by combining the Bragg Law with the de Broglie
relationship (where m is the neutron mass, h is Planck’s constant, t is total flight
time, and L is total flight path):
t=(

2m
)Lsinθd
h

(4-10)

When m and h are populated with 939.565 MeV/c2 and 4.13567 x 10-15 eVs,
respectively, the following relationship is obtained:
t = 505.56Lsinθd

(4-11)

From this relationship, when DIFA and ZERO are 0 – common with many
neutron sources – DIFC can be approximated by 505.56Lsinθ. When multiple
histograms are taken at different wavelengths, DIFC must be constantly refined for
the lower resolution histogram and fixed for the higher resolution histogram to
prevent divergence from the high correlation of DIFC to lattice parameters.
Similarly, DIFA can be introduced to the refinement if a large difference in source

200

wavelengths are incorporated – the quadratic dependence of DIFA on the
scattering angle fits the small shift in total flight path between longer wavelength
(longer flight length) and shorter wavelength (less absorption, shorter path). In the
case of Ca0.50FeO2, DIFA was refined after the profile refinement but did not
increase the refinement fit to any appreciable degree. The background was then
refined with a minimum number of terms (6 due to the complexity of the
background at low d-spacing) using the Chebyshev profile.
After the histogram related constants were inserted, atom positions and
occupancies were added according to Table 4-5. The peak profile was then
characterized as pseudo-voight – the weighted average of Gaussian and
Lorentzian peak profiles. Using this profile type required the refinement of six
parameters denoted sig-0, sig-1, sig-2 for the Gaussian part, and gam-0, gam-1,
gam-2 for the Lorentzian part of the profile as show in Equations 4-12 and 4-13,
respectively.
σ2 = σ20 + σ21 d2 + σ22 d4

(4-12)

γ = γ0 + γ1 d +γ2 d2

(4-13)

These parameters are much more important than they initially look – especially
sig-1 and gam-1 due to their linear relationship to d-spacing. In general, with a
stationary detector, peak broadening is related to d-spacing by a constant over the
whole d space. Refinement of sig-1 and gam-1 directly describes this constant,
creating a profile where the peak width naturally increases as the d-spacing

201

decreases. Unfortunately, these parameters alone do not account for anisotropic
peak broadening, which also has a d space dependence. The addition of two terms
to the Lorentzian Equation can account for anisotropic deviations with a linear
relationship to d space.
γ = γ0 + γ1 d +γ2 d2 + (γ1e d+ γ2s d2 )cosθ + γL

(4-14)

Figure 4-9. Rietveld refinement of NPD of Ca0.50FeO2 collected at 75 K showing
the observed pattern (crosses), fit pattern (red trace), difference curve (grey line,
below), and Bragg positions (magenta tickmarks). The refined unit cell is present
in the top right corner showing the topotactically preserved layered structure.

202

Incorporating the γ1e (g1ec), γL (L11), and γ2s (g2ec) parameters adds a
second set of terms similar to the first that are shifted depending on the diffraction
angle. Considering the phase being refined – Ca0.50FeO2 – incorporation of
anisotropic broadening (commonly caused by stacking faults) was scientifically
reasonable. A summary of the profile parameter values used for the refinements
at 12 K and 75 K are given in Table 4-6. Both datasets include diffraction data
collected with 1.066 Å and 3.731 Å radiation to cover a large d space range.

Table 4-6. Values for profile fitting parameters used in the Rietveld refinement of
Ca0.50FeO2 NPD data collected at both 75 K and 12 K.
Parameter

Value

sig-0

0.0000

sig-1

3022.5

sig-2

2081.8

gam-0

0.0000

gam-1

713.12

gam-2

0.0000

g1ec

-77.098

g2ec

41.231

L11

0.0000

203

Even with the extra profile parameters included to describe the peak anisotropy
from stacking faults and other broadening features, the low d space fit at 75 K is
fairly poor (Figure 4-9). As expected, the low d spacing profile fit at 12 K is
substantially better because of the lower lattice thermal energy (Figure 4-10).
There was also a problem refining oxygen positions due to the signal to noise ratio,
resulting in skewed octahedra.

Figure 4-10. Rietveld refinement of NPD of Ca0.50FeO2 showing the observed
pattern (crosses), fit pattern (red trace), difference curve (grey, below), and Bragg
positions (tickmarks, below). The second line of magenta tickmarks represents the
magnetic cell observed. The first few reflections of the magnetic cell are labeled.
204

The appearance of a magnetic cell in the low temperature NPD pattern was not
surprising considering NaFeO2 has two very distinct magnetic cells at low
temperatures (33). However, one NaFeO2 magnetic cell is incommensurate and
one is commensurate with the parent rhombohedral unit cell, a main difference
found in the low temperature magnetic unit cell of Ca0.50FeO2. The
incommensurate nature of the Ca0.50FeO2 magnetic cell made Rietveld refinement
extremely difficult. A LeBail refinement was completed on the magnetic cell,
resulting in the cell parameters presented in Table 4-7 (62). Higher quality NPD
patterns obtained at multiple temperatures between 5 K and 25 K (reason for this
temperature discussed further) are needed to establish the magnetic propagation
vector and moment direction, as well as to investigate the possibility of multiple
magnetic states.

Table 4-7. A comparison of the nuclear and incommensurate magnetic cell
parameters of Ca0.50FeO2 obtained from refinement of NPD data.
Cell
Parameter

Nuclear Cell

Magnetic Cell

% Change

a (Å)

5.1321

8.9378

+74

b (Å)

5.9385

11.435

+93

c (Å)

5.7924

5.7952

0

ß (°)

73.16

73.02

-1.2

Volume (Å3)

168.96

566.47

+330

205

An educated guess at the nature of the magnetic cell would point to a
ferromagnetic nearest-neighbor interaction with coupling constant JNN, an
antiferromagnetic

next-nearest-neighbor

interaction

JNNN,

and

an

antiferromagnetic nearest-neighbor interplanar interaction JIP when viewed looking
down the frustrated triangular lattice. If the next-nearest-neighbor interaction was
not present, the FeO2 layers would be dominated by ferromagnetic interactions;
hence, the JNNN gives rise to the frustration of the lattice. This description of the
magnetic cell is similar to that of one of the magnetic superstructures of NaFeO 2
(36). A corroboration of further NPD and magnetic measurements at low
temperatures (between 5 K and 25 K) could be used to determine the effect of JIP
and how the antiferromagnetic influence changes based on temperature.
The sodium intercalated product of Ca0.50FeO2 – Na0.50Ca0.50FeO2 – was
synthesized by first placing 250 mg Ca0.50FeO2 (2.32 mmol) in a clean 50 mL round
bottom flask in the glovebox with a magnetic stir bar. A 1 M solution of Nabenzophenone in THF was prepared by adding 0.4598 g (20 mmol) Na metal and
3.664 g (20 mmol) benzophenone to 20 mL of THF while stirring. 10 mL of the 1
M solution of Na-benzophenone was added to the round bottom flask and the
reaction was stirred for 10 days. Initially, to check for product purity and
intercalation rates, aliquots were taken every day and analyzed via ICP. A full yearlong reaction scaled up 25x in a 1 L round bottom flask was completed as well,
resulting in pure phase material.

206

Figure 4-11. Refinement of PXD of Na0.50Ca0.50FeO2. Contribution from the
sample holder is the reason for the unindexed peaks/amorphous background.

The background of Na0.50Ca0.50FeO2 is due to the domed sample holder used
to collect the diffractogram of the exceptionally air-sensitive sample. Many of the
diffractograms collected on air-sensitive samples and presented in this thesis were
obtained on a standard low background Si sample holder while the diffractometer
was located in an inert purgebox. When a diffractogram of Na0.50Ca0.50FeO2 was
collected in this way, the pattern showed obvious sample hydrolysis due to
excessive peak broadening and loss of crystallinity as the theta angle increased.

207

The sample was then loaded into a dome sample holder that was assembled inside
of a purified atmosphere glovebox to assure a constant and completely inert
atmosphere. However, the sample holder diffracted as well, leading to the
unindexed peaks and large hump-like background found in the PXD spectrum.
Refinement of PXD data for Na0.50Ca0.50FeO2 was also difficult due to both the
poor counting statistics of lab diffractometer data as well as the large background
contribution from the sample holder (Figure 4-11). The Ca0.50FeO2 model
previously refined was used as a starting basis for the refinement of
Na0.50Ca0.50FeO2. The refinement did converge when all occupancies were fixed
at values obtained from ICP data; however, with very poor refinement criteria of χ2
= 7.85 and Rwp = 19.76%. Thermal parameters were refined then closed as the
final refinement diverged when thermal parameters were open (Table 4-8).
The Li-intercalated product Li0.50Ca0.50FeO2 was also successfully synthesized;
however, according to diffraction data a very small peak (<5%) that could not be
indexed was always present. This peak was not seen in the other diffraction
patterns and may be a sign of short range ordering of Li1+ or Ca2+ in the lattice.
Stoichiometry was confirmed by ICP analysis. Refinement of the PXD data
converged with χ2 = 2.23 and Rwp = 5.64%. Refined atomic positions for
Li0.50Ca0.50FeO2 are presented in Table 4-9. The atomic positions are very similar
to those of Na0.50Ca0.50FeO2 and Ca0.50FeO2, verifying the topotactic nature of the
reaction.

208

Table 4-8. Atomic positions for Na0.50Ca0.50FeO2 from Rietveld refinement of PXD
data.
Atom

Wyckoff
Position

x

y

z

102 Uiso
(Å2)

Occ.

Ca2+

2b

0

0

0.5

0.51(3)

0.5000

Na1+

4i

0.49

0

0.52

3.1(2)

0.5108

Fe3+

2a

0

0

0

0.82(4)

0.5000

Fe3+

2c

0.5

0

0

0.64(1)

0.5000

O2-

8j

0.36

0.23

0.20

2.1(2)

0.335

O2-

4i

0.17

0

0.21

1.8(1)

0.335

Table 4-9. Atomic positions for Li0.50Ca0.50FeO2 from Rietveld refinement of PXD
data
Atom

Wyckoff
Position

X

y

z

102 Uiso(Å2)

Occ.

Ca2+

2b

0

0

0.5

0.86(3)

0.5000

Li1+

8j

0.47

0.08

0.57

2.6(1)

0.4821

Fe3+

2a

0

0

0

0.79(7)

0.5000

Fe3+

2c

0.5

0

0

0.61(4)

0.5000

O2-

8j

0.35

0.22

0.21

1.5(3)

0.3784

O2-

4i

0.15

0

0.20

1.4(2)

0.3025

209

A comparison of the cell parameters of Ca0.50FeO2 and the intercalated
compounds follows similar trends established in the previous cobaltate literature
(Table 4-10). The a-b plane is very similar among all three compounds with a slight
expansion in the Na1+ doped material due to the larger size of Na1+ compared to
Li1+. The c parameter contracted slightly after intercalation of Ca 0.50FeO2, caused
by the attraction between the intercalated alkali metals and the O2- of the CoO2
slabs. FeO2 interplanar distances followed expected trends as well, as shown in
Table 4-11.
The interplanar distances between Fe atoms in the Fe containing phases are
compared in Table 4-11. The distances trend as would be expected from chemical
intuition. After the aliovalent exchange of Ca2+ for Na1+ the interplanar distance
decreased nearly 1 Å due to the larger Coulombic interaction between Ca 2+ and
O2- and the increase of bond covalency over the Na1+ - O2- bonding. After alkali
metal ions were intercalated into the vacancies created during the exchange the
interplanar distance decreased slightly. It appears that the Ca 2+ ions act as pillars
in the structure essentially locking the interplanar distance regardless of further ion
intercalation.

210

Table 4-10. Cell parameters comparison of Ca0.50FeO2 and the intercalated
products Na0.50Ca0.50FeO2 and Li0.50Ca0.50FeO2.
Compound

a (Å)

b (Å)

c (Å)

ß (°)

Ca0.50FeO2

5.1743(8)

5.9532(4)

5.8215(3)

73.21(2)

Na0.50Ca0.50FeO2

5.1796(4)

5.9643(6)

5.8025(2)

73.24(2)

Li0.50Ca0.50FeO2

5.1733(3)

5.9482(7)

5.8044(1)

73.12(4)

Table 4-11. FeO6 interplanar distance comparison in the Fe compounds prepared
with the MuRE method.
Compound

Interplanar Distance (Å)

NaFeO2

5.6354

Ca0.50FeO2

5.5165

Na0.50Ca0.50FeO2

5.5052

Li0.50Ca0.50FeO2

5.4931

211

4.3b Magnetic measurements of Fe2.5+ phases

Figure 4-12. DC magnetic susceptibility data collected for NaFeO2, Ca0.50FeO2,
and Na0.50Ca0.50FeO2 collected under a 1000 Oe field.

The

magnetic

susceptibility

observed

for

NaFeO2,

Ca0.50FeO2,

and

Na0.50Ca0.50FeO2 is presented in Figure 4-12. The NaFeO2 susceptibility curve
shows an antiferromagnetic transition with Neel temperature TN = 17 K, matching
the previous literature reports (33). Above the antiferromagnetic transition, NaFeO2
appears to follow Curie-Weiss law in the range measured; however, after analyzing
the first derivative of the trace, it is apparent that a very broad tail exists above 100
K. This tail may be due to an increase in competition between the coupling

212

constants JNN, JNNN, and JIP at higher temperatures. Similarly to the parent phase
NaFeO2, Ca0.50FeO2 also shows a kink in the susceptibility data at TN = 23 K. The
divergence from Curie-Weiss behavior is even more apparent in the trace of
Ca0.50FeO2. The smaller interplanar distance between FeO2 slabs, in Ca0.50FeO2
relative to NaFeO2 leads to an increase in the JIP coupling. This coupling causes
frustration not only in the FeO2 planes but between planes and manifests in shortrange spin fluctuations that cause the non-CW behavior across the measured
region.

Figure 4-13. Effective magnetic moment for NaFeO2, Ca0.50FeO2, and
Na0.50Ca0.50FeO2 calculated from magnetic susceptibility data.

213

The short-range fluctuations present in Ca0.50FeO2 manifest in the effective
magnetic moment at low temperature (Figure 4-13). The effective moment of 5.92
bohr magnetons expected for an Fe3+ ion was realized at a much lower
temperature in NaFeO2 than Ca0.50FeO2, as expected for a less frustrated system.
The reduced phase Na0.50Ca0.50FeO2 showed very different susceptibility data
than the parent and intermediate phases in its synthesis. A large bifurcation
between the zero-field cooled and field cooled traces at low temperature becomes
apparent below 35 K. This bifurcation is typical of a spin glass state, especially due
to the nearly linear slope of the field cooled curve at low temperature - a sign of a
stochastic temperature dependence on time-frozen spins. The effective magnetic
moment plot of Na0.50Ca0.50FeO2 is even more interesting. As the temperature is
increased, enough thermal energy is present to break any short range spin
fluctuations present in the system. However, above 50 K, the effective magnetic
moment is much higher than the expected μS for a mixed Fe2+/3+ system or the μL
of an Fe2+ system. The high spin value may be a result of the large spin orbitcoupling present in octahedral Fe2+ and a small superparamagnetic Fe or Fe3O4
impurity. Further studies, including AC magnetization or a Langevin plot, can be
used to identify superparamagnetic Fe/Fe3O4. A Langevin fit of the M vs H
presented in Figure 4-14 was completed to investigate the existence of the
superparamagnetic impurity.

214

Figure 4-14. Dependence of magnetization on applied field of Na0.50Ca0.50FeO2
collected at 300 K and 5 K.

The field dependence curve of Na0.50Ca0.50FeO2 is shown at 300 K and 5 K in
Figure 4-14. To further understand the nature of the large bifurcation in the M vs.
T curve as well as the nearly linear M vs. H curve at 5 K, a purely academic
investigation into the possibility of superparamagnetic nanoparticles was
undertaken. The data presented in the following few paragraphs coupled with
Mossbauer spectroscopy would cement the cause of the interesting magnetic
results mentioned. However, a Langevin fit of the 5 K M vs H curve resulted in a
saturation magnetization of 27.02 g/emu and a particle mass of 1.27 x 10-19 g

215

according to Equation 4-15 (63). The variables C and xc represent the saturation
magnetization in g/emu and the particle mass in g respectively. Based on a 5.05
g/mL density for Fe3O4 and a spherical particle assumption, a 2.894 nm particle
diameter was calculated.
y=y0 + C( coth(x-xc ) -

1
)
(x-xc )

(4-15)

From the M vs. H curve, a very small coercivity of approximately 25 Oe is
present. By definition, a superparamagnetic material must have 0 coercivity;
however, as this value is dependent on the instrument response and accuracy,
many materials that shows coercivities less than 50 Oe are reported as
superparamagnetic in the literature. Since blocking temperatures for Fe3O4 are
reported as low as 10 K for 3.8 nm particles, it appears that the positive coercivity
may be a result of the superparamagnetic Fe3O4 particles present with a diameter
less than 3 nm (63, 64). The saturation magnetization value of 27.02 emu/g
observed is very low compared to literature reports of between 36 and 50 emu/g
(65, 66). It is also possible that a complex ferrite or a mixture of nanoparticles are
the sources of the possible superparamagnetism at low temperature, a
phenomena that has been reported previously (67, 68).

4.3c Electrochemical cycling of Li0.50Ca0.50FeO2
In the recent literature, cycling of the Fe2+/3+ oxidation state has been
notoriously difficult due to the insulating nature of the materials that restricted

216

current flow across the cell. This problem was thrown into the spotlight when
LiFePO4 was initially proposed as a Li1+ cathode material (69–72).
In the olivine type phase LiFePO4, Li1+ diffusion occurs along unideminsional
tunnels – generally unacceptable as a feature of a cathode material due to defects
inherent in most structural types (73). However, in the olivine structure, the
diffusion channels are never blocked by point defects (Li/Fe cation disorder) or
stacking faults across crystal lattice when the particle size is sub-micron (74). The
two-phase charging reaction of LiFePO4, given in Equation 4-16, also presents
another materials limitation: the constant charging voltage of 3.45 V limits both the
electronic and ionic conductivities of the material due to collection of the products
of reaction at the surface of the particles (69). This problem has been alleviated
through multiple pathways, including reduction of particle size, while the most
success has been seen by coating the particles in a layer of a Li-permeable
material (i.e. carbon) that allows the Li to be removed from the surface (72, 75).
LiFePO4 → FePO4 + Li1+ + e1-

V = 3.45 V

(4-16)

The electrochemical cycling performance of Li0.50Ca0.50FeO2 was tested using
a coin cell type setup against Li metal. The full experimental details of this setup
can be found in the appendix section of this thesis.

217

Figure 4-15. Partially optimized charge-discharge curve of Li0.50Ca0.50FeO2 cycled
from 2.2 V to 3.8 V vs. a Li-metal anode.

Similar to LiFePO4, initial electrochemical testing of Li0.50Ca0.50FeO2 resulted in
very poor performance. Li0.50Ca0.50FeO2 was prepared for electrochemical
characterization in an Ar-filled glovebox by grinding the powder with 20% by weight
acetylene black (Alfa Aesar, 99.9%) to increase overall conductivity of the sample.
In general, 1 g of sample material was used during sample preparation. A sufficient
amount (4-6 mL) of N-methyl-2-pyrrolidone was added to wet the powder, creating
a very thick consistency. A syringe was used to drop 2 mL of this slurry onto battery
grade aluminum foil and was flattened to a height of approximately 300 microns

218

using a doctor blade tool. The aluminum foil was then heated to 130 °C under a
partial vacuum of 500 mmHg for a minimum of 12 hours to completely remove the
solvent and densify the spread material. A die press was used in the glovebox to
cut 12.7 cm discs from the coated aluminum foil. These discs will now be referred
to as “active material”. The active material was loaded into the bottom of a split
coin cell, followed by a Celgard polyethylene membrane, 1 mL of 1 M LiClO4 in
ethylene carbonate, and a Li-metal anode. The cell was closed, pressed, and
hooked up to a potentiostat as demonstrated in the appendix of this thesis. Cycling
Li0.50Ca0.50FeO2 under these conditions from 2.5 V to 4 V resulted in a total cycled
capacitance of approximately 10 mAh/g.
Initial optimization of the cycling performance of this material was achieved by
altering the sample preparation method. Multiple literature reports have looked at
the importance of intimate contact between the conductive filler (carbon) and the
cathode material being tested (76–78). Taking this into account, Li0.50Ca0.50FeO2
was ball-milled with acetylene black and N-methyl-2-pyrrolidone for 1 hr in agate
jars at 400 rpm. The resulting liquid was similar in consistency to thick paint and
had no indication of clumping or inhomogeneity (as found when the materials were
hand-ground together). Further changes made to the experimental setup included
the substitution of LiClO4 for LiPF6 as the electrolyte, which has been shown to be
much more compatible with Fe-based cathode materials (79, 80), as well as
modifying the electrolyte solvent by replacing ethylene carbonate with a 4:3:3
ethylene carbonate:dimethyl carbonate:diethyl carbonate (MTI-XTL). A custom

219

program module that allowed for easy processing of cyclic voltammetry data led to
more accurate charge-discharge curves as well because it introduced the ability to
add background contributions to the analysis.

Figure 4-16. Secondary electron SEM of Ca0.50FeO2 prepared using the MuRE
method.

Once these optimizations were made, the charge-discharge curves for
Li0.50Ca0.50FeO2 presented in Figure 4-15 were obtained. The large capacitance
demonstrated in the black trace in the figure can be attributed to the formation of
a secondary electrolyte interface (SEI) layer, a phenomenon that has been

220

reported and studied many times in the literature (81–83). After formation of the
SEI layer, a maximum capacitance of 70 mAh/g is achieved, followed by 2%
capacitance degradation per cycle. Reasons for this degradation may be physical
(flaking of the material from the Al foil, large particle size, low conductivity) or
chemical (soluble SEI layer, chemical instability).
The particle size of the sample is an obvious problem that, when optimized,
may lead to a 25% increase in cycled capacity and retention (84). As seen in the
SEM micrograph presented in Figure 4-16, the Ca0.50FeO2 particles are
approximately 1 micron after grinding; however, very large crystals are still present.
Because Ca0.50FeO2 is insulating, only the surface of these large crystals will cycle
Li1+. As these are a large percentage of the total volume with a relatively small
surface area, the presence of these large crystals skews the data substantially.
The synthesis of smaller particles of Ca0.50FeO2 via some kind of microemulsion
or aqueous exchange may take advantage of solvent exfoliation and result in
higher true cycling capacity and retention.

221

4.4 Co2+/3+ phases prepared with the MuRE method
The following few sections describe and characterize the phases prepared with
the MuRE method with Co as the transition metal.

4.4a Structure characterization of AxCayCoO2 phases

Figure 4-17. PXD spectra of the Na0.50CoO2 based family of materials prepared
using the MuRE method presented in this chapter.

222

Intercalation of Li1+ into the P2 type Ca0.25CoO2 led to a slight expansion of the
unit cell in the a-b plane coinciding with a nearly 1 Å contraction in the c direction.
Larger Coulombic attraction between the Li1+ and O2- from the CoO2 layers is the
reason for this contraction. A much smaller contraction in the c axis after Na1+
intercalation was observed due to a combination of the larger size of Na1+ and the
lack of complete filling of all vacancies during intercalation. Both intercalated
phases were indexed in the hexagonal P63/mmc space group, proving the
topotactic nature of the MuRE method (Figure 4-17).

Table 4-12. Refined atomic positions of Li0.75Ca0.25CoO2 based on the Ca0.25CoO2
model. Li1+ and Ca2+ shared atomic sites in the refinement because the statistics
were not high enough to try and separate them into different positions.
Atom

Wycoff
site

x

y

z

102 Uiso(Å2)

Li1/Ca1

2b

0

0

0.25

3.2

Li2/Ca2

2c

0.1650(1)

0.3304(4)

0.25

4.7

Co

2a

0

0

0

2.4

O

4f

1/3

2/3

0

2.8

223

Figure 4-18. PXD patterns of the intercalated materials in the Na0.74CoO2 family
prepared using the MuRE method.

Rietveld refinement of PXD data of Li0.75Ca0.25CoO2 was performed in the
P63/mmc space group with Li1+ and Ca2+ in trigonal prismatic coordination. All
peaks were indexed in this model with very moderate fitting statistics (χ2 = 1.997,
Rwp = 7.86% Rp = 5.99%). Peak profile functions were refined using the LeBail
method followed by Rietveld refinement of atomic positions and thermal

224

parameters. Atomic occupancies were not refined, as the data did not have
sufficiently good statistics to reliably extract the occupancies of lighter elements. A
final unit cell of a = 2.817 Å c = 10.794 Å was produced with atomic positions
presented in Table 4-12. Rietveld refinement of Na0.67Ca0.25CoO2 resulted in a unit
cell of a = 2.828 Å and c = 10.851 Å with similar atomic positions and convergence
criteria (χ2 = 2.297, Rwp = 9.94% Rp = 7.22%). As apparent from the atomic thermal
parameters, the diffraction data was not of high enough quality for proper Rietveld
refinement.
In general, the PXD patterns obtained for the MuRE prepared compounds were
of too poor quality to perform reliable Rietveld refinements. Preferred orientation
and very large anisotropy in the c direction of the phases led to intense main peaks
with the majority of the other Bragg diffractions lost in the background.
Furthermore, PXD patterns were saturated with fluorescence from Co, a common
problem with Cu radiation sources (85). Voltage discrimination was used to
enhance the diffraction quality to a degree where peaks were visible; however,
changing the voltage window too far resulted in anomalous and malformed peaks.
Rietveld refinement diverged easily, especially when trying to include the small
Bragg reflections in the patterns; therefore, LeBail refinement was employed to
obtain the cell parameters of all phases, which are presented in Table 4-13.
Rietveld refinement of atomic positions and occupancies would require much
higher resolution data to understand if any Li1+ ordering is present. The refinement
of Li0.75Ca0.25CoO2 and Na0.67Ca0.25CoO2 was completed for example purposes

225

only and cannot be assumed to be correct due to the poor signal to noise ratio.
Rietveld refinement of the Fe-containing phases was possible because of the
neutron diffraction data and the lesser degree of anisotropy in the crystal structure.

Figure 4-19. PXD spectra of the NaCoO2 based family of materials prepared using
the MuRE method presented in this chapter. The broad background and marked
(*) peaks are due to the sample holder used.

226

PXD patterns for the rest of the Co containing compounds prepared with the
MuRE method are presented in Figures 4-18 and 4-19. It is apparent that the
layered structure is preserved throughout the MuRE series of reactions in all of the
compounds. As some samples were exceptionally air sensitive – showing slow
hydrolysis in the purge box – diffraction patterns were collected using a dome-type
sample holder. Because of the large intensity of the main peak, the amorphous
background is much less prominent in these patterns.

Table 4-13. Cell parameters of the materials with transition metals in the 2+/3+
oxidation state couple prepared with the MuRE method next to the cation
coordination type and Co oxidation state.
Compound

Space
Group

Li0.47Ca0.50Co1.00O2

C2/m

Na0.31Ca0.50Co1.00O2

R3̅ m

Li0.64Ca0.37Co1.00O2

P63/mmc

Na0.59Ca0.37Co1.00O2

P63/mmc

Li0.75Ca0.25Co1.00O2

P63/mmc

Na0.67Ca0.25Co1.00O2

P63/mmc

Unit Cell
(Å, ß in °)
a = 4.789
b = 4.027
c = 5.696
ß = 107.5
a = 2.831
c = 16.288
a = 2.818
c = 10.820
a = 2.856
c = 10.809
a = 2.817
c = 10.794
a = 2.828
c = 10.851

227

Structure
Type

Co Oxidation
State

O’3

+2.53

O3

+2.69

P2

+2.62

P2

+2.67

P2

+2.75

P2

+2.83

Further structural studies of Na0.31Ca0.50CoO2 must be completed, as the unit
cell can also be indexed in the C2/m space group with cell parameters a = 4.873
Å, b = 3.866 Å, c = 5.422 Å, and ß = 89.94 °. Likewise, Li0.47Ca0.50CoO2 can be
indexed in both R3̅ m and C2/m. High resolution diffraction is needed to understand
whether this is a monoclinic distortion or some other type of atomic ordering. The
low intensity peaks that should be present if the space group is monoclinic would
be apparent with a better signal to noise ratio.
Interplanar [CoO2] spacing for the NaxCoO2, Cax/2CoO2, and the intercalated
A1-xCax/2CoO2 are phases are presented in Table 4-14. Since the interplanar
spacings are based on Rietveld refinement of powder diffraction data collected on
a lab diffractometer with rather poor statistics or by interpolating based on the
powder diffraction pattern d spacing (main peaks are all 00c), there is a large
uncertainty in the values. The following discussion is purely academic in nature.
As discussed previously, the layer spacing in the NaxCoO2 materials is completely
dependent on x, increasing nearly linearly with oxidation of Co 3+ to Co4+. After
intercalation of the alkali metals back into the Cax/2CoO2 phases, a decrease in the
[CoO2] separation is observed. As the vacancies created during the aliovalent
exchange step are filled with cations, Coulombic attraction between the Na 1+/Li1+
and Ca2+ and the O2- ions draw the layers together. However, the Ca2+ pillars
appear to lock the interplanar distance into a narrow region (5.429 Å – 5.439 Å)
depending on the alkali metal intercalated and the x value in A 1-x/2Cax/2CoO2.

228

Table 4-14. CoO6 interplanar spacings for all Co-containing layer phases prepared
with the MuRE method
Material

[CoO2] separation (Å)

NaCoO2 (53)

5.200

Na0.71CoO2 (53)

5.441

Na0.53CoO2 (53)

5.529

Ca0.50CoO2 (53)

5.438

Ca0.35CoO2 (53)

5.436

Ca0.25CoO2

5.438

Li0.47Ca0.50Co1.00O2

5.429

Na0.31Ca0.49Co1.00O2

5.431

Li0.64Ca0.37Co1.00O2

5.435

Na0.59Ca0.37Co1.00O2

5.433

Li0.75Ca0.25Co1.00O2

5.437

Na0.67Ca0.25Co1.00O2

5.430

Crystallite size was calculated using the Whole Powder Pattern modeling
(WPPM) method with the PM2K software package. Microscopy measurements
were not performed due to the air and water sensitivity of the samples; however,
previous literature reports demonstrate the efficacy of WPPM in determining
crystallite size distribution for phases of high symmetry (86, 87). A lognormal

229

function was chosen to describe the particle size, given in Equation 4-17.
Thermodynamically this assumption is reasonable as well, considering that small
grains will generally form convex shapes (sphere, cube, etc) that can be easily
described using a single parameter. The lognormal function employed for this size
distribution study is a combination of the lognormal mean 𝜇 and the lognormal
variance σ:
gl (D)=

1

2

1/2

Dσ(2π)

exp[-(lnD-μ) /2σ2 ]

(4-17)

where D is the diameter of the spherical particle. A more in-depth description of
the lognormal function and the Fourier transform of said function used in the PM2K
software suite is given by Scardi and Leoni, but can also be found in Chapter 5 of
this thesis (88).The precursor sodium cobalt oxide phases all had respective size
distributions above 1 micron. After undergoing aliovalent exchange, the particle
size distribution shifted to approximately 500 +/- 100 nm, most likely due to layer
exfoliation, which is common in layered materials. Particle size distributions of the
final phases prepared with the MuRE method where Co is in the 2+/3+ oxidation
state were grouped near 120 +/- 20, showing a large decrease after each topotactic
operation was completed on the structure (Table 4-15). The particle size
distribution given by PM2K is the average platelet height parameter, which agrees
well with SEM (Figure 4-20).

230

Table 4-15. Particle size distributions obtained using the WPPM algorithms
implemented in the PM2K software suite. PM2K results gave the shortest length
(height of platelet) for platelet type particles.
Particle Size
(nm) from PM2K

Particle Size
(nm) from SEM

NaCoO2

Space
Group
R3̅ m

> 1 µm

-

Ca0.50CoO2

C2/m

512 ± 95

-

Li0.47Ca0.50CoO2

C2/m

119 ± 12

-

Na0.31Ca0.50CoO2

R3̅ m

128 ± 21

-

Na0.74CoO2

P63/mmc

> 1 µm

900 - 1100

Ca0.37CoO2

P63/mmc

502 ± 259

650 - 725

Li0.64Ca0.37CoO2

P63/mmc

114 ± 52

-

Na0.59Ca0.37CoO2

P63/mmc

157 ± 17

148 - 205

Na0.51CoO2

P63/mmc

> 1 µm

-

Ca0.25CoO2

P63/mmc

800 ± 235

-

Li0.75Ca0.25CoO2

P63/mmc

130 ± 70

-

Na0.67Ca0.25CoO2

P63/mmc

143 ± 10

-

NaFeO2

R3̅ m

> 1 µm

600 - 800

Ca0.50FeO2

C2/m

> 1 µm

550 - 750

Li0.49Ca0.50FeO2

C2/m

117 ± 29

-

Na0.50Ca0.50FeO2

C2/m

514 ± 244

475 - 575

Material

231

Figure 4-20. SEM image of Na0.59Ca0.37CoO2 using back scattered electrons (top)
and secondary electrons (bottom) showing the large anisotropy in crystal
dimensions resulting in platelets. Clearly, the ridges on the larger particles are
actually many different platelets.

232

4.4b Thermal analysis and stability investigation

Figure 4-21. TGA and DSC curves of Li0.75Ca0.25CoO2 under O2 and N2
atmospheres. Negative signals in the DSC are exothermic.

It is important to note the role of Ca2+ as “pillars” between the MO2 layers,
effectively stabilizing the structures with M2+ ions. Theoretical investigations into
the reasons for ordering in these layered phases have pointed to large differences
in elastic energy as the driving force to overcome the entropy required for ordering.
For example, since the ionic radii of Ca2+ in an octahedral environment (1.15 Å) is
much larger than that of Fe2+ (High spin (HS) - 0.92 Å), the radii ratio R(Fe2+/Ca2+)

233

= 0.80, well below the maximum empirical value of 0.86 required for layered
ordering (39, 89). Since the Fe2+ radii is similar to that of Li1+ (0.90 Å) – generating
a radii ratio R(Fe2+/Li1+) = 1.01 - the direct synthesis of layered compounds without
Ca2+ pillars is impossible because disorder in the resulting structure would be
energetically favored.

Figure 4-22. DSC curves of the Na1+ intercalated products of CaxCoO2 (x = 0.50,
0.37, 0.25) under an N2 atmospheres. Negative signals in the DSC are exothermic.

234

The TGA plot for Li0.75Ca0.25CoO2, shown in Figure 4-21, has a very interesting
mass gain feature under N2 at 425 °C. This feature was ruled out as either
instrumentally anomalous or systematic in nature, considering it is reproducible in
Li0.75Ca0.25CoO2 and was not seen for any Na1+ intercalated phases or in the TGA
of Li0.75Ca0.25CoO2 collected under Ar. As the resulting product of this mass gain
is not discernible using PXD due to the poor spectral quality, the hypothesis that
Li3N is formed during heating is proposed. Previous literature reports of Li metal
under N2 show a very sharp onset reaction and mass gain at approximately 180
°C (90). The DSC curve under O2 may lend some credence towards this
hypothesis, especially since the exothermic peak at 275 °C may represent the
burning of Li in air to form Li2O. NPD or SPXD would lend more insight into the
produced phase, and determine whether phase separation occurs and Li3N is a
separate material or if Li3N is intercalated into the layered structure.
As the DSC plots of the Li1+ intercalated phases were poisoned by the formation
of Li3N, the metastability of the phases was studied using the Na1+ intercalated
compounds. DSC curves for the three Na1+ materials with reduced Co transition
metals are presented in Figure 4-22. The DSC curves of these layered phases with
disordered cations show very broad transitions; however, upon zooming in, more
information on the stability of the materials can be elucidated.

235

Figure 4-23. Magnified differential scanning calorimetry trace of Na0.31Ca0.50CoO2
with the first derivative of the trace.

A magnified view of the DSC curve of Na0.31Ca0.50CoO2 is presented in Figure
4-23 to investigate the true metastability of phases prepared by the MuRE method.
In general, there are two types of solid state materials: high-temperature
(thermodynamic) phases and low-temperature (kinetic) phases. The kinetic
phases can be further split into two subgroups – metastable and low-temperature
– where the metastable phases cannot be directly prepared, while the lowtemperature phases are easily prepared using conventional solid state techniques
at low temperatures. One of the easiest ways to distinguish between these two

236

types of kinetic phases is to look closely at the DSC curves for an exothermic signal
at low temperature. This signal points to the transformation of a metastable phase
into a truly low-temperature phase. Unfortunately, due to the nature of metastable
phases – especially in terms of disorder and artificial defect density – the
exothermic signals can be very broad or faint in magnitude.
For the case of Na0.31Ca0.50CoO2, a clear exothermic signal is apparent at 250
°C in the DSC curve collected under an Ar environment. To appreciate whether
the signal is a manifestation of a slight change in background or a true exothermic
signal, the first derivative of the trace is also presented – again demonstrating that
Na0.31Ca0.50CoO2 is indeed a metastable material. Powder X-ray diffraction of
Na0.31Ca0.50CoO2 after heating to 250 °C under Ar showed a complete loss of
crystallinity – assumed to be an effect of this transition to the low-temperature
phase.

237

4.4c Magnetic properties – completing the phase diagram

Figure 4-24. Molar magnetic susceptibility of the Na1+ intercalated cobaltates
synthesized with the MuRE method obtained under an applied field of 0.1 T.

Investigation into the magnetic properties of the materials with exceptionally
low oxidation states produced by the MuRE method will greatly increase the
knowledge base of these triangular frustrated species and the impact of A cation
stoichiometry

on

their

fundamental

properties.

Preliminary

magnetic

measurements of the Co2+/3+ materials present similar features to the Co3+/4+
NaxCoO2 materials (Figures 4-24 and 4-25). For example, the molar susceptibility

238

of Na0.31Ca0.50Co2.69+O2 is temperature independent, very similar to Na0.30Co3.7+O2.
However, the magnetic susceptibility data between 60 K and 150 K follows CurieWeiss law with theta = 58.4 K and C = 0.08403 cm3K/mol. The positive Weiss
constant is indicative of ferromagnetic interactions in the ground state. Above this
temperature the susceptibility appears to vary from the CW model and is
independent of temperature, indicative of a bad-metal state. Under a 5 T magnetic
field the compound appears to be a standard paramagnet, but at 0.1 T, cusps in
the magnetization at 10 K and 45 K become apparent. The origin of these cusps
is unknown, although a spin glass state of asperomagnets is most likely present
between 10 K and 45 K. This hypothesis is substantiated by the bifurcation
between the ZFC and FC curves and characteristic magnetization curves of a spin
glass at 5 K denoted by the non-linear dependence of magnetization on field at
high field. Under high magnetic field a thermal randomization of the spins destroys
the spin-glass network, resulting in a paramagnetic signal. The uptick in molar
magnetization below 10 K may correlate to local short-range dominance of the
ferromagnetic contribution of the spin glass state. DFT calculations are underway
to investigate the magnetic ground state of this material. The cusps in the
magnetization data may point to charge ordering of the doped electrons similar to
Na0.50CoO2, a charge ordered insulator.

239

Figure 4-25. Magnetization dependence on field traces of the Na1+ intercalated
cobaltates synthesized with the MuRE method obtained at 5 K.

The Co2.83+ material Na0.67Ca0.25CoO2 appears to follow Curie-Weiss law in the
temperature range 60 K - 400 K, but upon further inspection of the first derivative
of the susceptibility, deviation across the whole range becomes obvious. The
combination of the very slight 'S' shape of the field dependence curve with no
hysteresis in the M vs. H loop and the minor bifurcation in the susceptibility data
below 10 K between the FC and ZFC curves suggests the presence of
superparamagnetic cobalt nanoparticles in the sample with an approximate
particle size of 2 nm. No cobalt impurities were seen via powder diffraction. The
sharp transition at 50 K in the susceptibility curve may be caused by charge
240

ordering of the HS Co2+ and LS Co3+ present. Heat capacity and low temperature
diffraction measurements are needed to identify if this transition is a first or second
order phase transition.
Na0.59Ca0.37CoO2, where Co is in the +2.67 oxidation state, is composed of the
nonmagnetic LS Co3+ and the paramagnetic HS Co2+ ions in a 2:1 ratio. Assuming
the simple model where Co2+ carries all of the charge and Co3+ is not in an
intermediate spin state, thus eliminating Co2+/Co3+ coupling, the susceptibility
follows Curie-Weiss law between 80 K and 400 K with a negative Weiss constant
of -117 K. This relatively large negative Weiss constant points to a strong
antiferromagnetic correlation between the spins in Na0.59Ca0.37CoO2. The uptick at
18 K in the susceptibility curve and the slight 'S' shape lacking hysteresis in the
field dependence curve point to a very small impurity of cobalt nanoparticles that
are formed during the reductive intercalation step undetected by diffraction
techniques. Interestingly, considering that only 33% of the Co2+/3+ ions are
magnetic, a possible charge ordering state may exist where the magnetic
frustration in the CoO2 layers is completely relieved by having 2 Co3+ and 1 Co2+
per triangular subunit. Low temperature diffraction may shed light on whether this
scenario exists below 20 K, correlating with the transition present in the
susceptibility data.
To complete the NaxCoO2 magnetic phase diagram presented earlier with
properties observed for the novel Co2+/3+ materials would be premature at this
point; however, heat capacity and resistivity measurements will shed light on the

241

nature of the magnetic features present. Very preliminary additions to the 2.5+/3+
region of the magnetic phase diagram include charge ordering near +2.80, an
asperomagnetic spin glass (possible Mott insulator depending on the resistivity
results) at +2.69, and a Curie-Weiss paramagnet at +2.67. These additions can be
seen in Figure 4-26. Because the Na1+ stoichiometry is easily tuned through
chemical deintercalation, a highly detailed magnetic phase diagram is possible.

Figure 4-26. Low-temperature magnetic phase diagram of layered oxide phases
with Co in different oxidation states

242

4.4d Electrochemical cycling of layered cobaltates
When designing new cathode materials for portable devices, three general
requirements should be adhered to: high operating voltage, multi-dimensional ion
pathways, and long cycle lifetimes (Table 4-16). Other important factors, including
cell safety, are generally considered an engineering problem, although toxic
elements should be avoided whenever possible. Currently, the two most common
Li-ion cathode materials are LiCoO2 and LiFePO4, with specific energies of 0.546
kWh/kg and 0.581 kWh/kg, respectively (91, 92). Both of these materials suffer
from inherent drawbacks: LiCoO2 can only reversibly cycle 0.5 Li1+ per formula unit
and is composed of expensive Co, while the synthesis of LiFePO4 is challenging
and industrially complicated (75).
When LiCoO2 is charged and more than 0.5 Li1+ is removed, a monoclinic
structural modification occurs where CoO2 planes shift relative to LiO2 planes,
leading to a degradation of the cycling capacity (93–95). To add rigidity to LiCoO2,
one could imagine placing ‘chemical pillars’ that connect the CoO2 layers and limit
their movement in the a-b plane. This idea, the topotactic modification of an αNaFeO2 type structure, can be easily achieved using the MuRE method previously
described herein, shown to be successful with Li0.75Ca0.25CoO2 (96). Starting with
the O3 type NaCoO2, the aliovalent exchange of Ca2+ for 2 Na1+ leads to a product
with ‘chemical pillars’ between the CoO2 layers: Ca0.50CoO2. The doubly-charged
Ca2+ ion locks the CoO2 layers together at a fixed distance – shown by the constant
interplanar distance independent of Ca stoichiometry in the CaxCoO2 layered

243

materials (53). Li1+ can then be intercalated into Ca0.50CoO2 while retaining the
‘chemical pillars’, giving a final product of Li0.47Ca0.50CoO2.

Table 4-16. Electrochemical properties of common Li-ion cathode materials
compared to the novel cobaltates presented in this thesis.

Material

Li+
Cycled

Average
Voltage
(V)

Specific
Capacity
(mAh/g)

Specific
Energy
(kWh/kg)

Specific
Energy
vs
LiCoO2

LiCoO2

0.50

3.9

140

0.546

1.00

LiMn2O4

0.50

4.0

148

0.592

1.08

LiFePO4

1

3.3

176

0.581

1.06

LiCo1/3Ni1/3Mn1/3O2

0.50

3.6

150

0.540

1.02

Li0.47Ca0.50CoO2

0.47

3.54

111

0.388

0.71

Li0.63Ca0.37CoO2

0.63

3.54

159

0.562

1.03

Li0.75Ca0.25CoO2

0.75

3.58

190

0.680

1.25

A key difference between LiCoO2 and Li0.47Ca0.50CoO2 is the oxidation state of
Co, having been reduced from 3+ to 2+/3+. When employed as a cathode material,
the average charging voltage vs. lithium for each cathode is given in Equations 418 and 4-19 below, demonstrating the effect of oxidation state on voltage.

244

Figure 4-27. Charge-discharge curves of Li0.47Ca0.50CoO2 showing two separate
discharge events at 2.9 V and 3.2 V vs Li metal.

LiCo3+O2 → Li0.5Co3.5+O2 + 1/2Li

Ecell = 3.9 V

Li0.47Ca0.50Co2.53+O2 → Ca0.50Co3+O2 + 0.47Li

Ecell = 3.5 V

(4-18)
(4-19)

Using charging voltage as a guide, Li0.47Ca0.50CoO2 has a capacity
approximately 10% less than that of LiCoO2; however, due to the added weight of
Ca2+ in the structure, the specific capacity is 22% less, resulting in a total decrease
in specific energy of approximately 30% when 0.47 equivalents of Li are cycled.
Figure 4-27 shows initial cycling results for Li0.47Ca0.50CoO2 at a rate of C/5 in a

245

coin cell type setup. The first cycle capacity of 75 mAh/g decreases upon further
cycling and stabilizes at 50 mAh/g, 45% of the theoretical capacity. As the charging
voltage is essentially fixed, an increase in specific energy must be accomplished
by increasing the number of Li1+ cycled per formula unit or decreasing the molar
mass of the compound. Interestingly, the discharge curve for Li0.47Ca0.50CoO2 has
two well defined intercalation peaks, while the charging curve has only one. It
appears that the presence of these two curves is the manifestation of two redox
couples involved – the cycling of the Co2+/3+ oxidation couple and the Co3+/4+
oxidation couple. This is one of few examples of multiple redox couple cycling in a
Li-ion battery cathode material (97, 98).
Because the MuRE method allows for control over the number of vacancies
created in the A position of the AMO2 structure, a starting material with a large
number of vacancies will translate to a final product with a relatively large amount
of intercalated Li1+. For example, if NaCoO2 is first oxidized to Na0.50CoO2 then the
Na1+ ions are exchanged for Ca2+, followed by Li1+ intercalation, the highly lithiated
phase Li0.75Ca0.25CoO2 is formed (96). The combination of formula mass decrease
between Li0.50Ca0.50CoO2 (114.44 g/mol) and Li0.75Ca0.25CoO2 (106.08 g/mol), as
well as an increase in Li1+ cycled per formula unit, leads to a specific capacity of
190 mAh/g for Li0.75Ca0.25CoO2 and a total specific energy increase of 0.292
kWh/kg. Experimentally, it was demonstrated that Li0.75Ca0.25CoO2 can be charged
with an unoptimized first cycle capacity of approximately 115 mAh/g – well below
the theoretical maximum 190 mAh/g (96). Upon subsequent cycles the charging

246

capacity diminished rapidly, attributed to the formation of an SEI layer with poor
Li1+ conductivity (Figure 4-28). Varying the experimental cycling conditions,
including electrolyte, solvent, loading content, and step rate, is currently underway
to optimize the experimental specific capacity.

Figure 4-28. Per-cycle capacity of Li0.75Ca0.25CoO2 measured at a C/5 rate against
a Li metal anode.

Overall, the highest specific capacity can be achieved by beginning with the
oxidized phase Na0.34CoO2 and employing the MuRE process. After Li1+
intercalation, a product with a theoretical stoichiometry of Li0.83Ca0.17CoO2 is
possible; however, experimentally this proves difficult. Aliovalent exchange of Ca2+

247

for Na1+ on Na0.34CoO2 results in nearly pure phase Ca0.17CoO2 with a Co2O3
admixture. Upon reductive intercalation, the structure quickly decomposes at room
temperature. The high concentration of oxidizing Co4+ in the structure, as well as
the ease of CoO2 plane shifting caused by a lack of interplanar Coulombic
attraction, is attributed to the extreme instability of the phase. Further experiments
may show some viability in electrochemical insertion of Li1+. If synthesized,
Li0.83Ca0.17CoO2 would have a specific capacity of 216 mAh/g and a specific energy
of 0.788 kWh/kg, nearly 1.5x greater than LiCoO2.

4.4e Intercalation kinetics investigation
Intercalation kinetics were monitored over time to understand the relationship
between layer stacking and alkali metal geometry with cation insertion in the host
lattice (Table 4-17). Aliquots of both Li1+ and Na1+ intercalation solutions with
different CaxCoO2 (x = 0.50, 0.37, 0.25) host phases were taken every 24 hours.
Alkali metal content was determined in triplicate with ICP. The smaller Li1+ is more
mobile during intercalation due to fewer in-space interactions with the large and
double-charged Ca2+. As becomes quickly apparent, insertion of Na1+ into the O3
type Ca0.50CoO2 structure is much slower than in the P2 type Ca0.35CoO2 or
Ca0.25CoO2 structures. After 120 hours, both P2 phases had nearly 90% Na1+
occupancy, whereas only 60% of the theoretical limit of Na1+ intercalation into
Ca0.50CoO2 was achieved. In the O3 structural types Na1+ and Ca2+, both share
the same structural position (3b) as opposed to the P2 structure where both the 2b

248

and 2c sites are occupied by cations in the AO2 layer. The two positions in the P2
structure can be differentiated by their relation to the CoO 6 octahedra: prisms in
the 2b site shared edges with CoO6 octahedra while prisms in the 2c site share
faces with CoO6 octahedra. In the majority of P2 type NaxCoO2 phases, Na1+
occupies the edge sharing site with a 2:1 preference. In the O3 stacking type, all
AO6 octahedra share edges with CoO6 octahedra. We believe Na1+ diffusion is
limited in the O3 type phase because of Coulombic repulsion between Na 1+ and
Ca2+ upon intercalation into the lone 3b site. This repulsion occurs to a lesser
degree in the P2 type phases where Na1+ and Ca2+ can occupy two different
structural sites. Na1+ phases prepared by the MuRE method were intercalated for
240 hrs and 360 hrs, in which final compositions were identical between the two
reaction times. Li1+ intercalation into the O3 phase is much less hindered, although
after 120 hours vacancies were still present in Li0.47Ca0.50CoO2.

249

Table 4-17. Compound stoichiometry over intercalation time for the preparation of
the reduced phases presented in the manuscript. Bolded numbers are stopping
points for the preparation each compound.
x after time (hrs)
Compound
24

48

72

96

120

240

LixCa0.50CoO2

0.212

0.371

0.435

0.475

0.489

0.498

LixCa0.37CoO2

0.446

0.523

0.591

0.627

0.642

0.638

LixCa0.25CoO2

0.268

0.313

0.475

0.682

0.751

0.762

NaxCa0.50CoO2

0.062

0.078

0.123

0.153

0.199

0.313

NaxCa0.37CoO2

0.295

0.218

0.275

0.495

0.542

0.593

NaxCa0.25CoO2

0.174

0.183

0.325

0.389

0.472

0.665

250

4.5 Possibility of lower oxidation states with extended MuRE

Figure 4-29. A schematic diagram demonstrating the theoretical extendibility of the MuRE method to produce phases
with transition metal oxidation states between 2+ and 2.5+.

251

By extrapolating on the aforementioned reductive-intercalation approach, it
seems possible to produce layered materials of the α-NaFeO2 type structure with
Co oxidation states below 2.5 – the limit of the reductive-intercalation technique.
As shown in Figure 4-29, aliovalent exchange of Ca2+ for A1+ in the intercalated
compound (Step 5) will lead to a higher Ca2+ content (Ca0.75 when x = 1) than was
possible when starting with Nax=1CoO2 (Ca0.50 when x = 1). Alkali metals could then
be intercalated into the Ca2+ enriched material (Step 6), effectively lowering the
oxidation state of Co to below 2.5+. Theoretically, the aliovalent exchange followed
by intercalation could be repeated ad infinitum as the Co oxidation state
approaches 2+, effectively producing CaCoO2. Similar compounds, Ca3Co4O9 and
Ca2CoO3, which have misfit layered structures that resemble the layered materials
presented herein, are being explored for interesting thermoelectric properties (99,
100).

252

4.6 Chapter conclusions
A multistep, topotactic synthesis method for the preparation of phases with CdI2 type frustrated layers was presented. A series of compounds was synthesized
with transition metals in uniquely low oxidation states for the type of crystal
structure. Structural and stoichiometry characterization was presented to validate
the method and demonstrate the true topotactic nature via the retention of crystal
structure

across

the

multiple

synthetic

steps.

Intercalation

kinetics,

electrochemical characterization, and TGA analysis discussions are presented.
Use of the MuRE method will allow for deeper fundamental insights into
magnetically frustrated systems and how the competing coupling pathways in
these systems can be modulated.

253

4.7 Future work on this topic
The future direction of this research project has the potential to branch into
many other projects. Because the synthetic method is so versatile, and can be
applied to other structural types, a purely fundamental solid state project looking
at the synthesis of new phases and crustal structures could be approached. As
many of the compounds produced with the MuRE method - either the layered αNaFeO2 type or other structural motifs - will exhibit interesting characteristics and
novel properties, full characterization of these materials is necessary.
More specifically, the MuRE method can be used to synthesize new Li-ion
cathode materials with the possibility of 2 electron transfer. Using transition metals
with multiple oxidation states - V or Mn, for example - would facilitate this
chemistry. Furthermore, lighter transition metals will have a larger theoretical
specific capacity.
The MuRE method can also be used as a technique to synthesize materials for
thermoelectric devices. Increasing phonon and electron scattering by introducing
vacancies and multiple different ions in the same crystallographic position are
achievable with the MuRE method. Future direction could focus on using Co and
a 5d transition metal with a mixture of alkali metal and alkaline earth metals in the
A position.
Further structural investigations using high-resolution diffraction techniques either synchrotron or neutron diffraction - are necessary to fully understand the
location of atoms in the phases prepared using the MuRE method. Low

254

temperature neutron diffraction studies may lead to the characterization of longrange atom ordering.

255

REFERENCES

256

REFERENCES

(1)

Huang, Q.; Foo, M.; Pascal, R.; Lynn, J.; Toby, B.; He, T.; Zandbergen, H.;
Cava, R. Coupling between electronic and structural degrees of freedom in
the triangular lattice conductor NaxCoO2. Phys. Rev. B 2004, 70, 184110.

(2)

Viciu, L.; Bos, J.; Zandbergen, H.; Huang, Q.; Foo, M.; Ishiwata, S.; Ramirez,
a.; Lee, M.; Ong, N.; Cava, R. Crystal structure and elementary properties
of NaxCoO2 (x=0.32, 0.51, 0.6, 0.75, and 0.92) in the three-layer NaCoO2
family. Phys. Rev. B 2006, 73, 1–10.

(3)

Rouxel, J.; Danot, M.; Bichon, J. Intercalation Compounds NaxTiS2 General Structural Study of NaxTiS2 and KxTiS2 Phases. Bull. Soc. Chim. Fr.
1971, 3930.

(4)

Pauling, L. The Nature of the Chemical Bond; 3rd ed.; 1948.

(5)

Lucovsky, G. Transition from thermally grown gate dielectrics to deposited
gate dielectrics for advanced silicon devices: A classification scheme based
on bond ionicity. J. Vac. Sci. Technol. A Vacuum, Surfaces, Film. 2001, 19,
1553.

(6)

Mamiya, M.; Kataoka, K.; Kikuchi, S.; Terajima, Y.; Tokiwa, K.; Akimoto, J.
Synthesis and structure analysis of a new titanium oxide having an
intergrowth structure between ramsdellite-type and calcium ferrite-type. J.
Phys. Chem. Solids 2012, 73, 1460–1462.

(7)

Clarke, S. J.; Fowkes, A. J.; Harrison, A.; Ibberson, R. M.; Rosseinsky, M. J.
Synthesis , Structure , and Magnetic Properties of NaTiO2. Society 1998,
4756, 372–384.

(8)

Ouyang, S.-D.; Quan, Y.-M.; Liu, D.-Y.; Zou, L.-J. A Comparative
Investigation on the JT Effect in Triangular Compounds of NaMnO2, NaNiO2
and NaTiO2. Chinese Phys. Lett. 2011, 28, 067102.

(9)

Takeda, K.; Miyake, K.; Takeda, K.; Hirakawa, K. Phase Transition in
NaTiO2: A Candidate for S=1/2 Traingular Heisenberg Antierromagnet. J.
Phys. Soc. Japan 1992, 61, 2156–2157.

(10) Dhariwal, M.; Maitra, T.; Singh, I. Comparative study of NaTiO2 and NaNiO2
using first principle calculations. 2013, 824, 824–825.

257

(11) Reid, A.; Sabine, T.; Wheeler, D. Neutron Diffraction and Other Studies of
Magnetic Ordering in Phases Based on Cr2O3, V2O3, and Ti2O3. 1972, 409,
400–409.
(12) Hoppe, R.; Schepers, B.; Rohrborn, H.-J.; Vielhaber, E. Uber Oxoscandate:
LiScO2 und NaScO2. Zeitschrift fur Anorg. und Allg. Chemie 1965, 339, 130–
143.
(13) US Geological Survey Mineral Commodity Summaries; 2008; pp. 1–2.
(14) Morosan, E.; Natelson, D.; Nevidomskyy, A. H.; Si, Q. Strongly correlated
materials. Adv. Mater. 2012, 24, 4896–923.
(15) McQueen, T.; Stephens, P.; Huang, Q.; Klimczuk, T.; Ronning, F.; Cava, R.
Successive Orbital Ordering Transitions in NaVO2. Phys. Rev. Lett. 2008,
101, 166402.
(16) Ragupathy, P.; Shivakumara, S.; Vasan, H. N.; Munichandraiah, N.
Preparation of Nanostrip V2O5 by the Polyol Method and Its Electrochemical
Characterization as Cathode Material for Rechargeable Lithium Batteries. J.
Phys. Chem. C 2008, 112, 16700–16707.
(17) Sathiya, M.; Prakash, a S.; Ramesha, K.; Tarascon, J. M.; Shukla, A. K.
V2O5-anchored carbon nanotubes for enhanced electrochemical energy
storage. J. Am. Chem. Soc. 2011, 133, 16291–9.
(18) Whittingham, M. S.; Song, Y.; Lutta, S.; Zavalij, P. Y.; Chernova, N. A. Some
transition metal (oxy)phosphates and vanadium oxides for lithium batteries.
J. Mater. Chem. 2005, 15, 3362.
(19) Li, L.; Kim, S.; Wang, W.; Vijayakumar, M.; Nie, Z. A New Vanadium Redox
Flow Battery Using Mixed Acid Electrolytes. Rev. Lit. Arts Am. 2010.
(20) Didier, C.; Guignard, M.; Denage, C.; Szajwaj, O.; Ito, S.; Saadoune, I.;
Darriet, J.; Delmas, C. Electrochemical Na-Deintercalation from NaVO2.
Electrochem. Solid-State Lett. 2011, 14, A75.
(21) Guignard, M.; Didier, C.; Darriet, J.; Bordet, P.; Elkaïm, E.; Delmas, C. P2NaxVO2 system as electrodes for batteries and electron-correlated materials.
Nat. Mater. 2013, 12, 74–80.
(22) Miyazaki, S. Chemical and Electrochemical Deintercalations of the Layered
AMO2, Synthetic Metals 1983, 6, 211-217.

258

(23) Saint, J. A.; Doeff, M. M.; Wilcox, J. Electrode Materials with the Na0.44 MnO2
Structure : Effect of Titanium Substitution on Physical and Electrochemical
Properties. 2008, 3404–3411.
(24) Li, H. H.; Yabuuchi, N.; Meng, Y. S.; Kumar, S.; Breger, J.; Grey, C. P.; Shaohorn, Y. Changes in the Cation Ordering of Layered O3 LixNi0.5Mn0.5O2
during Electrochemical Cycling to High Voltages : An Electron Diffraction
Study. Society 2007, 2551–2565.
(25) Yuan, D.; He, W.; Pei, F.; Wu, F.; Wu, Y.; Qian, J.; Cao, Y.; Ai, X.; Yang, H.
Synthesis
and
electrochemical
behaviors
of
layered
Na0.67[Mn0.65Co0.2Ni0.15]O2 microflakes as a stable cathode material for
sodium-ion batteries. J. Mater. Chem. A 2013.
(26) Cheng, F.; Liang, J.; Tao, Z.; Chen, J. Functional materials for rechargeable
batteries. Adv. Mater. 2011, 23, 1695–715.
(27) Zhou, Y.-N.; Ding, J.-J.; Nam, K.-W.; Yu, X.; Bak, S.-M.; Hu, E.; Liu, J.; Bai,
J.; Li, H.; Fu, Z.-W.; Yang, X.-Q. Phase transition behavior of NaCrO2 during
sodium extraction studied by synchrotron-based X-ray diffraction and
absorption spectroscopy. J. Mater. Chem. A 2013, 1, 11130.
(28) Miyazaki, S.; Kikkawa, S.; Koizumi, M. Chemical and electrochemical
deintercalations of the layered compounds LiMO2 (M = Cr, Co) and NaM’O2
(M' = Cr, Fe, Co, Ni). Synth. Metails 1983, 6, 211–217.
(29) Komaba, S.; Takei, C.; Nakayama, T.; Ogata, A.; Yabuuchi, N.
Electrochemical intercalation activity of layered NaCrO2 vs. LiCrO2.
Electrochem. commun. 2010, 12, 355–358.
(30) Komaba, S.; Nakayama, T.; Ogata, A.; Shimizu, T.; Takei, C.; Takada, S.;
Hokura, A.; Nakai, I. Electrochemically Reversible Intercalation of Layered
NaNi0.5Mn0.5O2 and NaCrO2. In ECS Transactions; ECS, 2009; pp. 43–55.
(31) Elliston, P. R.; Blazey, K. W. Magnetic and Optical Study of NaCrO2. J. Phys.
Chem. Solids 1974, 36, 877–881.
(32) Blesa, M.; Moran, E.; Leon, C.; Santamaria, J.; Tornero, J. D.; Mendendez,
N. α-NaFeO2: ionic conductivity and sodium extraction. Solid State Ionics
1999, 126, 81–87.
(33) McQueen, T.; Huang, Q.; Lynn, J.; Berger, R.; Klimczuk, T.; Ueland, B.;
Schiffer, P.; Cava, R. Magnetic structure and properties of the S=5∕2
triangular antiferromagnet α-NaFeO2. Phys. Rev. B 2007, 76, 1–7.

259

(34) Rousseau, J.; Bernheim, R. Synthesis of NaFeO2. Compt Rend 1888, 170,
240.
(35) Blesa, M. C.; Morán, E.; Menéndez, N.; Tornero, J. D.; Torrón, C. Hydrolysis
of sodium orthoferrite α-NaFeO2. Mater. Res. Bull. 1993, 28, 837–847.
(36) Blesa, M.; Moran, E.; Amador, U.; Andersern, N. H. Crystal and Magnetic
Structures of a Nickel-Rich Ferrite Obtained by Ionic Exchange from αNaFeO2. J. Solid State Chem. 1997, 129, 123–129.
(37) Blesa, M. C.; Morán, E.; Tornero, J. D.; Menéndez, N.; Mata-Zamora, E.;
Saníger, J. M. A new route to γ-Fe2O3 via an intermediate oxyhydroxide. The
reaction of α-NaFeO2 with benzoic acid. J. Mater. Chem. 1999, 9, 227–231.
(38) Blesa, M.; Amador, U.; Moran, E.; Menendez, N.; Torenero, J.; Synthesis
and characterization of nickel and magnesium ferrites obtained from αNaFeO2. Solid State Ionics 1993, 63-65, 429–436.
(39) Shannon, R. D. Revised effective ionic radii and systematic studies of
interatomic distances in halides and chalcogenides. Acta Crystallogr. Sect.
A 1976, 32, 751–767.
(40) Fukuzaki, K.; Kohiki, S.; Matsushima, S.; Oku, M.; Hideshima, T.; Watanabe,
T.; Takahashi, S.; Shimooka, H. Preparation and characterization of NaInO2
and NaInS2. J. Mater. Chem. 2000, 2, 779–782.
(41) Shikano, M.; Delmas, C.; Darriet, J. NaRuO2 and NaxRuO2.yH2O: new oxide
and oxyhydrate with two dimensional RuO2 layers. Inorg. Chem. 2004, 43,
1214–6.
(42) Agrawal, R. C.; Gupta, R. K.; Superionic solids : composite electrolyte phase
– an overview. J. Mater. Sci. 1999, 4, 1131–1162.
(43) Delmas, C.; Maazaz, A.; Fouassier, C.; Reau, J.-M.; Hagenmuller, P.
Influence de l’environnement de l'ion alcalin sur sa mobilite dans les
structures a feuillets Ax(LXM1-x)O2. Mater. Res. Bull. 1979, 14, 329–335.
(44) Kawakami, S.; Sasaki, M.; Tabata, H.; Shimooka, H.; Kohiki, S.;
Matsushima, S.; Oku, M.; Shishido, T. Comparison of electronic structure of
LiInO2 with NaInO2. J. Alloys Compds. 2003, 359, 278–280.
(45) Hernan, L.; Macias, M.; Morales, J. Topotactic exchange of layered NaInO 2
by protons. Mater. Res. Bull. 1990, 25, 1169–1174.

260

(46) Sato, J.; Kobayashi, H.; Saito, N.; Nishiyama, H.; Inoue, Y. Photocatalytic
activities for water decomposition of RuO2-loaded AInO2 (A=Li, Na) with d10
configuration. J. Photochem. Photobiol. A Chem. 2003, 158, 139–144.
(47) Ogura, S.; Kohno, M.; Sato, K.; Inoue, Y. Effects of RuO2 on activity for
water decomposition of a RuO2/Na2Ti3O7 photocatalyst with a zigzag layer
structure. J. Mater. Chem. 1998, 7, 2335–2337.
(48) Chem, W.; Chem, J.; Denki, S.; Mb, T. Photocatalytic activity for methylene
blue decomposition of NaInO2. J. Mater. Sci. 2004, 9, 6367–6370.
(49) Diebold, U. The surface science of titanium dioxide. Surf. Sci. Rep. 2003,
48, 53–229.
(50) Ong, N. P.; Cava, R. J. Electronic frustration on a triangular lattice. Science
2004, 305, 52–3.
(51) Foo, M.; Wang, Y.; Watauchi, S.; Zandbergen, H.; He, T.; Cava, R.; Ong, N.
Charge Ordering, Commensurability, and Metallicity in the Phase Diagram
of the Layered NaxCoO2. Phys. Rev. Lett. 2004, 92, 1–4.
(52) Lee, M.; Viciu, L.; Li, L.; Wang, Y.; Foo, M. L.; Watauchi, S.; Pascal, R. a;
Cava, R. J.; Ong, N. P. Large enhancement of the thermopower in NaxCoO2
at high Na doping. Nat. Mater. 2006, 5, 537–40.
(53) Cushing, B. L.; Wiley, J. B. Topotactic Routes to Layered Calcium Cobalt
Oxides. J. Solid State Chem. 1998, 391, 385–391.
(54) Ono, Y.; Ishikawa, R.; Miyazaki, Y.; Ishii, Y.; Morii, Y.; Kajitani, T. Crystal
Structure, Electric and Magnetic Properties of Layered Cobaltite β-NaxCoO2.
J. Solid State Chem. 2002, 166, 177–181.
(55) Huang, Q.; Foo, M. L.; Lynn, J. W.; Zandbergen, H. W.; Lawes, G.; Wang,
Y.; Toby, B. H.; Ramirez, A. P.; Ong, N. P.; Cava, R. J. Low temperature
phase transitions and crystal structure of Na0.5CoO2. J. Phys. Condens.
Matter 2004, 16, 5803–5814.
(56) Takada, K.; Sakurai, H.; Takayama-muromachi, E. Superconductivity in twodimensional CoO2 layers. 2003, 422, 53–55.
(57) Cushing, B. L.; Falster, A. U.; Simmons, W. B.; Wiley, J. B. A multivalent ion
exchange route to lamellar calcium cobalt oxides. Chem Comm 1996, 2,
2635–2636.

261

(58) Yang, H. X.; Shi, Y. G.; Guo, Y. Q.; Liu, X.; Xiao, R. J.; Physical properties,
strontium ordering and structural modulation in layered hexagonal
Sr0.35CoO2. arXiv 2005, 1–16.
(59) Xiao, R. J.; Yang, H. X.; Xu, L. F.; Zhang, H. R.; Electronic structure and
electron energy-loss spectra of Sr0.35CoO2. arXiv 2006, 1–16.
(60) Gomez, J. C.; Calvet, N.; Starace, A. K.; Glatzmaier, G. C. Ca(NO 3)2 —
NaNO3 - KNO3 Molten Salt Mixtures for Direct Thermal Energy Storage
Systems in Parabolic Trough Plants. J. Sol. Energy Eng. 2013, 135, 021016.
(61) Peterson, V. K. Lattice parameter measurement using LeBail versus
structural (Rietveld) refinement: A caution for complex, low symmetry
systems. Powder Diffr. 2012, 20, 14–17.
(62) Peterson, V. K. Lattice parameter measurement using LeBail versus
structural (Rietveld) refinement: A caution for complex, low symmetry
systems. Powder Diffr. 2012, 20, 14–17.
(63) Parvin, K. Synthesis and magnetic properties of monodisperse Fe 3O4
nanoparticles. J. Appl. Phys. 2004, 95, 7121.
(64) Rumpf, K.; Granitzer, P.; Morales, P. M.; Poelt, P.; Reissner, M. Variable
blocking temperature of a porous silicon/Fe3O4 composite due to interactions
of the magnetic nanoparticles. Nanoscale Res. Lett. 2012, 7, 445.
(65) Sun, J.; Zhou, S.; Hou, P.; Yang, Y.; Weng, J.; Li, X.; Li, M. Synthesis and
characterization of biocompatible Fe3O4 nanoparticles. 2006, 2–10.
(66) Ghandoor, H. El; Zidan, H. M.; Khalil, M. M. H.; Ismail, M. I. M. Synthesis
and Some Physical Properties of Magnetite (Fe3O4). 2012, 7, 5734–5745.
(67) Yuan, H. L.; Wang, Y. Q.; Zhou, S. M.; Liu, L. S.; Chen, X. L.; Lou, S. Y.;
Yuan, R. J.; Hao, Y. M.; Li, N. Low-Temperature Preparation of
Superparamagnetic CoFe2O4 Microspheres with High Saturation
Magnetization. Nanoscale Res. Lett. 2010, 5, 1817–1821.
(68) Lak, A.; Kraken, M.; Ludwig, F.; Kornowski, A.; Eberbeck, D.; Sievers, S.;
Litterst, F. J.; Weller, H.; Schilling, M. Size dependent structural properties
of FeO-Fe3O4 nanoparticles. Nanoscale 2013, 5, 12286–95.
(69) Ouyang, C.; Shi, S.; Wang, Z.; Huang, X.; Chen, L. First-principles study of
Li ion diffusion in LiFePO4. Phys. Rev. B 2004, 69, 1–5.

262

(70) Zaghib, K.; Goodenough, J. B.; Mauger, a.; Julien, C. Unsupported claims
of ultrafast charging of LiFePO4 Li-ion batteries. J. Power Sources 2009,
194, 1021–1023.
(71) Chung, S.-Y.; Bloking, J. T.; Chiang, Y.-M. Electronically conductive
phospho-olivines as lithium storage electrodes. Nat. Mater. 2002, 1, 123–8.
(72) Yuan, L.-X.; Wang, Z.-H.; Zhang, W.-X.; Hu, X.-L.; Chen, J.-T.; Huang, Y.H.; Goodenough, J. B. Development and challenges of LiFePO 4 cathode
material for lithium-ion batteries. Energy Environ. Sci. 2011, 4, 269.
(73) Gu, Y.-J.; Zeng, C.-S.; Wu, H.-K.; Cui, H.-Z.; Huang, X.-W.; Liu, X.-B.; Wang,
C.-L.; Yang, Z.-N.; Liu, H. Enhanced performance and high energy density
of LiFePO4 based lithium ion batteries. Mater. Lett. 2007, 61, 4700–4702.
(74) Wagemaker, M.; Singh, D. P.; Borghols, W. J. H.; Lafont, U.; Haverkate, L.;
Peterson, V. K.; Mulder, F. M. Dynamic solubility limits in nanosized olivine
LiFePO4. J. Am. Chem. Soc. 2011, 133, 10222–8.
(75) Konarova, M.; Taniguchi, I. Synthesis of carbon-coated LiFePO4
nanoparticles with high rate performance in lithium secondary batteries. J.
Power Sources 2010, 195, 3661–3667.
(76) Li, H.; Chen, G.; Zhang, B.; Xu, J. Advanced electrochemical performance
of Li[Ni(1/3−x)FexCo1/3Mn1/3]O2 as cathode materials for lithium-ion battery.
Solid State Commun. 2008, 146, 115–120.
(77) Palomares, V.; Goñi, A.; Muro, I. G. De; de Meatza, I.; Bengoechea, M.;
Cantero, I.; Rojo, T. Conductive additive content balance in Li-ion battery
cathodes: Commercial carbon blacks vs. in situ carbon from LiFePO 4/C
composites. J. Power Sources 2010, 195, 7661–7668.
(78) Wu, N. L. Impact of Current Collector on High- Rate Performance of Li-ion
Battery Electrodes. Society 2011.
(79) Licht, S. A High Capacity Li-Ion Cathode: The Fe(III/VI) Super-Iron Cathode.
Energies 2010, 3, 960–972.
(80) Tarascon, J. M.; Delacourt, C.; Prakash, a S.; Morcrette, M.; Hegde, M. S.;
Wurm, C.; Masquelier, C. Various strategies to tune the ionic/electronic
properties of electrode materials. Dalton Trans. 2004, 2988–94.
(81) Becker, D.; Cherkashinin, G.; Schmid, S.; Hausbrand, R.; Jaegermann, W.
Mechanistic investigation of SEI formation : Study of solvent adsorption on
in-situ prepared thin film cathode surfaces. Surf. Sci. 2011.
263

(82) Cherkashinin, G.; Nikolowski, K.; Ehrenberg, H.; Jacke, S.; Dimesso, L.;
Jaegermann, W. The stability of the SEI layer, surface composition and the
oxidation state of transition metals at the electrolyte-cathode interface
impacted by the electrochemical cycling: X-ray photoelectron spectroscopy
investigation. Phys. Chem. Chem. Phys. 2012, 14, 12321–31.
(83) Mukhopadhyay, A.; Tokranov, A.; Xiao, X.; Guo, F.; Hurt, R. H.; Sheldon, B.
W. Irreversible stress development due to SEI formation and self discharge
kinetics of graphitic thin film anodes in Li-ion battery. Stress Int. J. Biol.
Stress 2011.
(84) Malik, R.; Burch, D.; Bazant, M.; Ceder, G. Particle size dependence of the
ionic diffusivity. Nano Lett. 2010, 10, 4123–7.
(85) Fransen, M. J. 1 and 2 dimensional detection systems and the problem of
sample fluorescence in X-ray diffractometry. Adv. X-ray Anal. 2004, 47, 224–
231.
(86) Blakely, C. K.; Bruno, S. R.; Baum, Z. J.; Poltavets, V. V. Effects of ball
milling and thermal annealing on size and strain of ASnO 3 (A = Ba, Sr)
ceramics. Solid State Sci. 2013, 15, 110–114.
(87) Ahmed, J.; Blakely, C. K.; Bruno, S. R.; Poltavets, V. V. Synthesis of MSnO 3
(M=Ba, Sr) nanoparticles by reverse micelle method and particle size
distribution analysis by WPPM. Mater. Res. Bull. 2012, 47, 2282–2287.
(88) Leoni, M.; Scardi, P. Nanocrystalline domain size distributions from powder
diffraction data research papers. J. Appl. Crystallogr. 2004, 37, 629–634.
(89) Schougaard, S. B.; Bréger, J.; Jiang, M.; Grey, C. P.; Goodenough, J. B.
LiNi0.5+δMn0.5–δO2 High-Rate, High-Capacity Cathode for Lithium
Rechargeable Batteries. Adv. Mater. 2006, 18, 905–909.
(90) Jeppson, D. W.; Ballif, J. L.; Yuan, W. W.; Chou, B. E. Lithium Literature
Review: Lithium’s Properties and Interactions; 1978; p. 111.
(91) Mizushima, K.; Jones, P. C.; Wisemna, P. J.; Goodenough, J. B. LixCoO2
(0<x<=1): A new cathode material for batteries of high energy density. Mater.
Res. Bull. 1980, 15, 783–789.
(92) Goodenough, J. B.; Kim, Y. Challenges for Rechargeable Li Batteries.
Chem. Mater. 2010, 22, 587–603.

264

(93) Gupta, R.; Manthiram, A. Chemical Extraction of Lithium from Layered
LiCoO2. J. Solid State Chem. 1996, 491, 483–491.
(94) Kim, Y.; Veith, G. M.; Nanda, J.; Unocic, R. R.; Chi, M.; Dudney, N. J. High
voltage stability of LiCoO2 particles with a nano-scale Lipon coating.
Electrochim. Acta 2011, 56, 6573–6580.
(95) Jhu, C.-Y.; Wang, Y.-W.; Wen, C.-Y.; Shu, C.-M. Thermal runaway potential
of LiCoO2 and Li(Ni1/3Co1/3Mn1/3)O2 batteries determined with adiabatic
calorimetry methodology. Appl. Energy 2012, 100, 127–131.
(96) Blakely, C. K.; Bruno, S. R.; Davis, J. D.; Poltavets, V. V. Synthesis and
Electrochemical Properties of Novel Co2+/Co3+ Layered Ordered Rock-Salt
Structure Type Li0.75Ca0.25CoO2. ECS Trans. 2013, 45, 3–10.
(97) Park, Y.-U.; Seo, D.-H.; Kim, B.; Hong, K.-P.; Kim, H.; Lee, S.; Shakoor, R.
a; Miyasaka, K.; Tarascon, J.-M.; Kang, K. Tailoring a fluorophosphate as a
novel 4 V cathode for lithium-ion batteries. Sci. Rep. 2012, 2, 704.
(98) Li, X.; He, W.; Xiao, Z.; Peng, F.; Chen, J. Ionothermal synthesis and rate
performance studies of nanostructured Li3V2(PO4)3/C composites as
cathode materials for lithium-ion batteries. J. Solid State Electrochem. 2013,
17, 1991–2000.
(99) Masset, a.; Michel, C.; Maignan, a.; Hervieu, M.; Toulemonde, O.; Studer,
F.; Raveau, B.; Hejtmanek, J. Misfit-layered cobaltite with an anisotropic
giant magnetoresistance: Ca3Co4O9. Phys. Rev. B 2000, 62, 166–175.
(100) Asahi, R.; Sugiyama, J.; Tani, T. Electronic structure of misfit-layered
calcium cobaltite. Phys. Rev. B 2002, 66, 155103.

265

Chapter 5 : Particle size analysis using Whole Powder Pattern Modeling

5.1. Introduction to particle size analysis
As nanostructured materials are more widely studied in the literature, a better
technique to extract physical parameters from powder diffraction data has evolved
- Whole Powder Pattern Modeling (WPPM) (1). Compared to standard line profile
analysis techniques (LPA) like the Scherrer formula (equation 5-1), WPPM
methods do not solely rely on peak width and shape factors in calculating physical
properties like particle size distribution (2). In the Scherrer equation, the average
domain size D is directly related to the dimensionless shape factor K which is
dependent on the shape of the crystallites and the incident x-ray wavelength and
indirectly related to the peak line breadth in radians ß at FWHM and the Bragg
angle θ. This equation always calculates the lower bound for the particle size
because of the intrinsically complicated nature of nanostructure materials. The
width of a peak in an x-ray diffractogram is dependent on both instrumental effects
and crystallite characteristics: size, strain, dislocations, twinning, anti-phase grain
boundaries, and heterogeneities in the chemical makeup. Because the Scherrer
formula only includes crystallite size as the determinant for peak width, it can only
provide a correct particle size distribution when the sample is perfectly crystalline
and dispersed (when every other contribution to peak width is essentially zero).
D=

Kλ
βcosθ

266

(5-1)

Rather than basing particle size analysis on peak breadth, WPPM methods use
a bottom-up approach (1). A least squares analysis - similar in idea to Rietveld
refinement - is used to create a synthetic powder pattern based on the refined
parameters that contribute to peak width, mentioned previously. This pattern is
iterated through and modulated during each refinement cycle until the synthetic
pattern matches the observed pattern. Assuming the diffractogram statistics are
reasonably high, all real contributions to peak width can be quantified using
WPPM.
The PM2K software suite was used in the following studies to employ the
WPPM methods (3). PM2K is a customizable software package that allows plugins to be written and run simultaneously - for example, both WPPM and Rietveld
analysis can be run on the same powder pattern at the same time and each
iteration can be compared. Current plugins that have been created include:
Rietveld-Caglioti, Fundamental Parameter Approach of instrumental broadening,
size broadening including lognormal, delta, and York distributions, strain
broadening to identify dislocations via Wilkens, Green, and Houska functions,
effective broadening using the Harmonics model, Faulting with Warren equations,
antiphase boundary identification with Wilson-soldos models, and grain surface
relaxation. First, an input file is written by the user in a pseudo-visual basic type
code which includes peak locations and reflections as well as sample and
instrument parameters. Then, certain parameters are locked or unlocked

267

depending on which physical phenomena are being investigated, followed by least
squares iteration.
For those interested in learning more about the PM2K software package: a full
user manual with examples has been compiled, printed, and bound. It is located in
Prof. Viktor Poltavets’ office.

268

5.2 Effects of milling and annealing on stannate nanoparticles
Alkaline earth metal stannates ASnO3 (A = Ba, Sr) are actively investigated for
multiple applications. Thin hetero-junction films of BaSnO3 with other oxides have
been studied as gas sensors for CO, CH3SH, and liquefied petroleum gas (4–7).
When structurally similar BaTiO3 and BaSnO3 perovskites are combined,
multifunctional sensors are formed with the capacity to detect temperature,
acetylene, and ethylene under standard conditions (8). BaSnO3 also shows
promise in photocatalysis, has room temperature luminescence properties, and
operate as ceramic dielectrics (9–11). The efficiency of stannates in these
applications is directly linked to the grain size, leading to a push to develop
synthetic methods that result in smaller particle sizes.
The conventional ceramic method results in large, coarse-grained particles with
a non-uniform size distribution. Nano-sized ASnO3 materials have been produced
through low temperature routes including: hydrothermal and sol-gel methods (12),
lyothermal synthesis (8), coprecipitation (13), and a peroxide procedure (14).
Inorganic nitrate, oxalate, and acetate precursors are predominately used in the
low temperature synthesis (150 °C – 200 °C) of nanoparticles due to their low
thermal stability. There are very few systematic size-tuning studies of ternary
oxides in the literature and, to the best of our knowledge, none that focus on the
ASnO3 perovskite materials or the exploration of the effects of ball milling on
particle size and strain.

269

The importance of particle size in cutting-edge devices has recently been
realized in the optimization of the semiconductor charge collector in dye-sensitized
solar cells (DSSCs). Both particle size and crystallite strain are important factors
in the efficiency of a DSSC, where particle size dictates electron diffusion
coefficients (D), electron recombination times (τ), and charge injection efficiency,
while crystallite strain is correlated to surface trap states and an overall decrease
in injection efficiency (15). Thus, there is a need in facile preparation techniques
which would allow for the precise control of both particle size and crystallite strain.
Particle size and strain analysis is commonly performed by studying X-ray
diffraction peak broadening with the Scherrer equation (16, 17). A seminal study
on ZnO samples prepared from different precursor salts clearly defined a
correlation between the chemical nature of the initial precursor (nitrate, oxalate,
acetate, or carbonate) and the microstrutural properties of the obtained ceramic,
including the crystallite shape, size, strain, and stacking faults (18). While widely
employed, the Scherrer equation utilizes only one or two reflections rather than all
applicable reflections in the pattern. A further disadvantage of the Scherrer formula
is the assumption that diffraction peaks are rectangular – an essential shortcoming
that is fixed when the WPPM algorithms are used. Since diffraction peak shape
depends on the combination of particle shape and size distributions, as well as
strain of different origins, improvement of the WPPM model is an ongoing effort.
The WPPM approach models all peaks to extract microstructural features and
information from the entire pattern, rather than a subset, by combining pattern-

270

fitting techniques based on both integral breadth and Fourier methods. Because
the diffraction pattern peaks are modelled directly, the WPPM particle size and
crystallite strain analysis is a powerful and accurate method in microstructure
determination.
In this case study, the preparation of ASnO3 (A = Ba, Sn) nanoparticles by highimpact ball milling and a study of the particle size and strain dependence on milling
time and further annealing temperature is presented. The underlying idea was to
prepare nanoparticles by a simple method and decrease strain by thermal
annealing while preserving small particle size.

5.2a Experimental methods and WPPM input
BaCO3 (Alfa Aesar, 99%), SrCO3 (Alfa Aesar, 99%), and SnO2 (Strem
Chemical, 99%) were used as staring materials without further purification. For the
milling experiments, a Fritsch planetary ball mill was used with 18 mL steel jars
and 4 10mm tungsten carbide balls per jar. Milling was done in five minute
increments with one minute rest at 500 rpm for an appropriate time as described
below. All samples were stored under a N2 atmosphere after preparation.
For the synthesis of BaSnO3 and SrSnO3, stoichiometric amounts of the
appropriate metal carbonate and SnO2 were intimately mixed using an agate
mortar and pestle for 30 minutes, then fired at 1200 °C for 24 hours in air with a 6
hour heating ramp. The material was then reground and fired for an additional 16
hours, resulting in approximately 20 g of pure phase material - enough for all further

271

studies. From this bulk material, 3 g aliquots were ball milled in steel jars according
to the procedure mentioned above for the following specific time durations: 30
minutes, 1 hour, 2 hours, 4 hours, and 8 hours. The 8 hour sample was then split
into 10 equivalent samples, each fired in air with a 3 hour heating ramp to an array
of temperatures (200 °C - 1000 °C with a 100 °C step), and held for 24 hours.
Powder X-ray Diffraction (PXD) patterns were taken of these thermally treated
samples in order to correlate the effect of thermal annealing on grain size and
crystallite strain.
PXD patterns were obtained on a Bruker D8 Advanced New diffractometer
equipped with a LynxEye detector using Cu radiation with a Kα1 wavelength of
1.5418 Å at room temperature. Diffraction patterns were collected from 10 - 80° 2Θ
with a step size of 0.02° at 25 °C for all materials. The instrumental line broadening
profile was developed using the standard reference material LaB6 provided by
NIST measured under identical conditions as the sample material. Due to the high
symmetry of both materials, seven reflections were recorded for BaSnO 3 (Pm3̅ m,
a = 4.1084(2) Å) and six for SrSnO3 (Pm3̅ m, a = 4.0252(3) Å). Rietveld refinement
and structure analysis were completed using the Bruker Topas software suite (19).
Peak positions, areas, unit cell parameters, and crystallite parameters were refined
one by one, then locked, followed by a complete refinement of all parameters to
obtain the final unit cell dimensions. The patterns were initially refined based on
ICSD entries 43138 and 27047 for BaSnO3 and SrSnO3 respectively, using the
Bruker Topas software suite.

272

Thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC)
measurements were performed using a Netsch Jupiter 449 F3 TGA/DSC on 1020 mg samples with a platinum sample holder. Ultra-high purity (99.999%) nitrogen
was used as the protective gas for all experiments. Buoyancy effects were
corrected via subtraction of a blank measurement under identical experimental
conditions and TGA analysis was used to confirm oxygen stoichiometry of the
starting ceramic materials.
Microstructural features of the samples were analyzed using a JEOL-2200FS
high-resolution transmission electron microscope (HRTEM), and elemental
analysis was obtained using an Oxford Inca EDX analyzer attached to the
aforementioned TEM instrument. The PM2K software suite, based on the principle
of WPPM, was used to analyze the particle size distribution and particle strain
features using the assumption of a lognormal distribution of particles and a
Lorentzian peak shape (3).
Necessary parameters for the extraction of size and strain from the powder
diffraction pattern were obtained using both the PM2K and TOPAS software suites.
The major assumptions using this modeling method include: minimum peak
overlap in the diffraction pattern, high counting statistics, and a Lorentzian peak
shape. Considering the highly symmetric nature of the materials studied herein,
the spacing between hkl peaks led to zero line overlap in the pattern, accompanied
by a very high signal-to-noise ratio. A lognormal size distribution was used to
calculate the particle size of the ball-milled and thermally treated samples. There

273

is precedence in the literature for the use of a lognormal distribution of particle
sizes for nanoparticles used in different chemical methods, including catalysis (7).
This assumption is thermodynamically reasonable, considering that small grains
form simple convex shapes (sphere, cube) that can be described using a single
parameter (20).
The lognormal function employed for this size distribution study is a
combination of the lognormal mean μ and the lognormal variance σ :
gl (D)=

1

2

1/2

Dσ(2π)

exp[-(lnD-μ) /2σ2 ]

(5-2)

where D is the diameter of the spherical particle. A more in-depth description of
the lognormal function and the Fourier transform of said function used in the PM2K
software suite is given by Scardi and Leoni (1, 3, 20). The instrument profile
component of the powder x-ray diffraction pattern was modeled with the Caglioti
formula to parametize the FWHM and Lorentzian fractions (21):
FWHM2 = W + Vtan(θ) + Utan2 (θ)

(5-3)

ηLorentz = a + bθdeg + cθ2deg

(5-4)

where θ is half of the Bragg angle 2θ. The W, V, U, a, b, and c parameters were
refined by PM2K to optimize the Caglioti profile function, and then fixed, as we
assumed that the instrumental contribution is constant for the instrument and is not
dependant on the sample (22).
PXD patterns for both BaSnO3 and SrSnO3 were obtained on the same
instrument with identical experimental setups to avoid the introduction of

274

instrument discrepancy. Using the indexing obtained from the respective
refinements, the spectra were loaded into the PM2K program with an input file
specifying the Caglioti parameters previously established, the cell parameters of
the respective phases, a function to refine zero error and sample displacement, a
four parameter Chebyshev background function, and the peak indices and relative
intensities. All variables except for the background, peak intensities, and sample
displacement were initially locked and then released over a series of iterations to
obtain a final whole pattern refinement. Once a suitable refinement fit value was
obtained, a lognormal size distribution function was added to the input file.

5.2b Effect of impact milling on particle size and strain
PXD diffraction patterns of the ceramics BaSnO3 (cubic symmetry, Pm3̅ m
space group, a = 4.1084(2) Å) and SrSnO3 (cubic symmetry, Pm3̅ m space group,
a = 4.0252(3) Å) fired at 1200 °C were refined using the Bruker Topas software
suite and agree with previously reported results (23). TGA and elemental analysis
via EDS both confirmed phase pure oxygen stoichiometric BaSnO 3 and SrSnO3
starting materials.

275

Figure 5-1. Powder x-ray diffraction patterns of a) BaSnO3 and b) SrSnO3, as well
as the corresponding calculated particle size distributions in c) and d), respectively.

Microstructure and particle size analysis of the ceramic compounds were
completed using the PM2K and TOPAS software suites, calculating particles sizes
of 680 ± 400 nm and 201 ± 246 nm, and strain values of 0.0072% and 0.0113%
for un-milled, ceramic BaSnO3 and SrSnO3, respectively. These results were
verified with TEM imaging showing overall cubic particles with a large size

276

variance. The particle size values agree with those previously reported in the
ternary stannate ceramics literature (24).
The ball milling conditions, extracted particle sizes, and calculated crystallite
strain values of ceramic BaSnO3 and SrSnO3 samples milled for different times
ranging from 30 minutes to 8 hours are listed in Table 5-1. As expected, the
extracted average particle size is inversely related to milling time, culminating in
particles of sub 20 nanometer size (Figure 5-1) - ideal for applications in the nanotech industry, such as in solar cells, battery materials, and fuel cells (25).
Microstructural strain associated with the ASnO3 stannates milled for different time
lengths was investigated using the Williamson-Hall plots (Bcos(θ) vs sin(θ)), shown
in Figure 5-2. The slope of the fitted line is directly related to the strain of the
material, given by the equation:
Bcos(θ) =

λ
+ 4εs sin(θ)
Dv

where Dv is the average diameter (nm) and εs is the crystallite strain (26).

277

(5-5)

Table 5-1. Particle sizes and crystallite strain values calculated using the PM2K
software for BaSnO3 and SrSnO3 samples milled for different time periods.
Milling (hr)

Particle Size (nm)

εstr

BaSnO3

SrSnO3

BaSnO3

SrSnO3

0

680 ± 400

201 ± 246

0.00715

0.00817

0.5

147 ± 44

62 ± 16

0.03808

0.04798

1

94 ± 30

56 ± 23

0.04921

0.11225

2

32 ± 17

45 ± 22

0.07343

0.12566

4

24 ± 21

37 ± 19

0.14135

0.14105

8

9.4 ± 2

14 ± 4

0.18133

0.18939

Multiple hkl indices were analyzed for both samples out to 85° 2Θ. There is a
variable amount of scatter away from the fitting lines, especially at longer milling
times, which is indicative of anisotropic microstructural characteristics that can be
a result of high-impact ball milling. The non-superimposible nature of the fitting
lines lend credence to the fact that after milling, non-cuboid particles resulted. The
extracted strain values for the twelve compounds are listed in Table 5-1, clearly
following a constant trend of higher strain under longer milling conditions.

278

Figure 5-2. Williamson-Hall plots of a) BaSnO3 and b) SrSnO3 obtained using the
WPPM methods. The particle strain can be extracted from the slope of the line
using equation (5-5).

5.2c Direct control over size and strain with annealing
After milling for 8 hours, samples of both BaSnO3 and SrSnO3 were subjected
to thermal annealing (200 °C - 1000 °C) for 24 hours. The powder diffraction
patterns were analyzed to extract both average crystallite size and crystallite strain
after thermal annealing (Table 5-2 and Figure 5-3). As the annealing temperature
increased from 200 °C to 600 °C, crystallite strain decreased 38% and 44% for
BaSnO3 and SrSnO3 respectively. From 600 °C to the highest thermal treatment
temperature, 1000 °C, crystallite strain decreased 58% and 28%, resulting in a
total crystallite strain reduction of 71% and 59% for BaSnO3 and SrSnO3,
respectively. Thermal annealing was reproducible and verified by TEM microscopy
measurements, shown in Figure 5-4.

279

Figure 5-3. Williamson-Hall plots of BaSnO3 and SrSnO3 particles after thermal
annealing for 24 hrs at various temperatures. The particle strain can be extracted
from the slope of the line using equation (5-5).

Current DSSCs utilize nanometer size metal-oxide films that have been coated
with a light-absorbing dye to facilitate the conversion of solar energy to electrical
energy via photon capture by the dye, followed by relaxation and elecron injection
into the metal-oxide substrate. In recent literature, the relationships between
particle size and electron diffusion coefficients, as well as recombination lifetimes
have been studied by numerous prominent groups (15, 27, 28). The Yanagida
group in Japan correlated a smaller TiO2 particle size to increased effective light
absorption of the cell due to the larger surface area, leading to an increase in the
number of photogenerated electrons (15). As the TiO2 particle size approached 14
nm, both the charge injection efficiency and a shorter diffusion length caused
efficiency losses in the cell. The ability to control the particle size of the metal-oxide

280

film is paramount when attempting to optimize the electron diffusion coefficients,
recombination lifetimes, and effective light absorption of a DSSC. During the
thermal treatment study, particular attention was paid to the effect of temperature
on both particle size and strain of the crystallites. The information on controlling
particle size and strain gleaned from this study can be used to directly increase the
efficiency of DSSCs.

5.2d Concluding remarks and observations
The ability to specifically control crystallite size has been shown to be an
invaluable tool in the development of cutting-edge materials technology, including
solar cells. In this paper, we have shown that the precise control of the particle size
of ASnO3 (A = Ba, Sn) can be accomplished using simple solid state methods.
Both particle size and crystallite strain were extracted from the powder diffraction
patterns by employing the WPPM algorithm via the PM2K software. This allowed
for a systematic approach to decreasing overall crystallite strain while maintaining
sub-hundred nm particle sizes. This technique can be easily extended to
specifically fabricate other technologically useful materials to investigate the
properties of different particle sizes and strains.

281

Table 5-2. Particle sizes and crystallite strain values calculated using the PM2K
software for BaSnO3 and SrSnO3 samples thermally treated for 24 hours at varied
temperatures.
Temp. (°C)

Particle Size (nm)

Estr

BaSnO3

SrSnO3

BaSnO3

SrSnO3

200

15 ± 6

13 ± 5

0.20068

0.18795

300

14 ± 2

20 ± 12

0.15872

0.14559

400

22 ± 14

49 ± 35

0.18002

0.14657

500

24 ± 10

51 ± 12

0.15574

0.14748

600

51 ± 28

46 ± 15

0.12521

0.10604

700

59 ± 13

63 ± 38

0.12009

0.08103

800

60 ± 25

81 ± 27

0.12910

0.08083

900

58 ± 12

108 ± 56

0.05360

0.07510

1000

63 ± 44

128 ± 95

0.05192

0.07646

282

5
Figure 5-4. The left 6 TEM micrograms represent BaSnO3, while the right 6 vertical panels show SrSnO3 particles.
The labels a) - f) refer to milling times: a) none, b) 30 min, c) 1 hr, d) 2 hr, e) 4hr, and f) 8 hr.

283

5.3 Using WPPM to characterize synthesized nanoparticle
As previously shown, WPPM methods work well to describe particles greater
in size than 15 nm produced by impact milling, which results in very broad peaks
due to the decrease in particle size and the introduction of multiple forms of strain
to the crystallites. This second study looks at the efficacy of WPPM when the
particles are nanosized and essentially free of strain. The following synthesis was
in collaboration with Dr. Jahangeer Ahmed.

5.3a Direct synthesis of stannate nanoparticles
ASnO3 nanoparticles (A = Ba or Sr) were synthesized using the microemulsion
method via two procedures. In procedure I, one microemulsion system was
composed of (1-hexadecyl) trimethylammonium bromide as the surfactant, 1butanol as the co-surfactant, 2,2,4-trimethylpentane as the oil phase, and 0.1M of
Ba(NO3)2 or Sr(NO3)2 and 0.1M of SnCl4 as the aqueous phases. The second
microemulsion system had an identical composition, with the exception of the
aqueous phase, which was NH4OH (28% of NH3). The above two microemulsions
were mixed slowly on a magnetic stirrer before adding 0.5 M ammonium carbonate
drop by drop to maintain the stoichiometry of the loaded reactions at room
temperature. The resulting precipitates were centrifuged and washed several
times with a methanol and chloroform mixture (1:1 ratio) to remove the organic
moieties and dried at 80 °C for 2 hrs. Powder X-ray diffraction patterns show that
ACO3 (A = Ba or Sr) was obtained with orthorhombic crystal structure and

284

amorphous tin based compound. The above product was used a precursor for the
synthesis of ASnO3 (A = Ba or Sr) nanoparticles at 650 °C.
In procedure II, microemulsion one had the same composition as the first
procedure, while the second contained an aqueous phase of 0.5 M (NH4)2CO3
rather than NH4OH (28% of NH3). The resulting precipitate was centrifuged and
washed as given above. These precipitates were used as a precursor for the
synthesis of ASnO3 (A = Ba or Sr) nanoparticles at 650 °C.
The amorphous tin component and barium carbonate obtained after
centrifugation from the first procedure was annealed at 650 °C in air for 60 hrs,
resulting in the formation of BaSnO3 nanoparticles with a small amount of impurity
(3 %). All reflections of the pattern could be indexed on the basis of a cubic unit
cell. The cell parameter was refined using the Bruker Topas software suite, and
the refined lattice parameter of the BaSnO3 nanoparticles was found to be 4.116(3)
Å, which closely matches with the lattice parameter previously reported (14).
Thermal analysis of the synthesized precursors under flowing oxygen shows a
weight loss corresponding to one water molecule and one molecule of carbon
dioxide. The final decomposed product was analyzed as BaSnO3.

285

Figure 5-5. TEM (top, left) and HRTEM (top, right) micrographs of BaSnO3
nanoparticles synthesized using procedure I, and TEM (bottom, left) and HRTEM
(bottom, right) micrographs of BaSnO3 nanoparticles synthesized using procedure
II. Both materials were fired at 650 °C for 60 hrs.

286

5.3b Microscopy investigation of particle size
Figure 5-5 shows the TEM micrographs of the synthesized BaSnO 3
nanoparticles using both procedures. The microscopy studies showed the
formation of spherical shaped nanoparticles with an average size of 15 nm. High
resolution transmission electron microscopic (HRTEM) studies confirm the
formation of nanospheres of BaSnO3 particles with a lattice spacing of 2.916 Å,
which is consistent with that of the (110) plane of BaSnO3. The average size of the
barium stannate nanoparticles obtained at 650 °C via the second procedure was
found to be ~ 40 nm. The HRTEM image of the nanoparticle obtained via the
second procedure with a fringe value was consistent with the 110 plane of BaSnO3.
Similarly, TEM micrographs for SrSnO3 are represented in Figure 5-6. TEM
studies show the formation of SrSnO3 nanoparticles with an average size of 25
nm. The long aging time of the reaction may be responsible for the agglomeration
of the nanoparticles.

287

Figure 5-6. (a) & (b) TEM and HRTEM micrographs of SrSnO3 nanoparticles
synthesized at 650 °C for 72 hr (obtained from procedure I), (c) & (b) TEM and
HRTEM micrographs of SrSnO3 nanoparticles.

288

5.3c Particle size distributions of stannate nanoparticles
PXD patterns for both BaSnO3 and SrSnO3 were obtained on the same
instrument with identical experimental setups. The patterns were initially refined
based on ICSD entries 43138 and 27047 for BaSnO 3 and SrSnO3 respectively,
using the Bruker Topas software suite. Using the indexing obtained from these
refinements, the spectra were loaded into the PM2K program with an input file
specifying the Caglioti parameters previously established, the cell parameters of
the respective phases, a function to refine zero error and sample displacement, a
four parameter Chebyshev background function, and the peak indices and relative
intensities. All variables except for the background, peak intensities, and sample
displacement were initially locked and then released over a series of iterations to
obtain a final whole pattern refinement. Once a suitable refinement fit value was
obtained, a lognormal size distribution function was added to the input file, resulting
in the size distribution shown in Figure 5-7.
Particle size distributions of 15.8 ± 4.2 and 18.4 ± 8.3 were calculated for
BaSnO3 and SrSnO3 respectively, using PM2K. The results obtained from the
WPPM algorithm agreed well with the TEM particle size analysis shown in Figure
X, validating the use of WPPM as a tool for calculating particle size distribution on
cubic as-prepared nanoparticles with simple crystal morphology.

289

Figure 5-7. Particle size distributions of the prepared BaSnO3 and SrSnO3
nanoparticles obtained using the WPPM method.

290

5.4 Efficacy of WPPM on lower symmetry oxides
The previous two case studies on the use of WPPM have focused on highly
symmetric cubic oxide particles of varying sizes. The following study on delafossite
CuAlO2 nanoparticles investigates the modeling functions on much lower
symmetry materials, where preferred orientation can also lead to large peak
asymmetry. The synthesis of the nanoparticles was in collaboration with Dr.
Jahangeer Ahmed.

5.4a Introduction to delafossite type CuAlO2
Delafossite CuAlO2 belongs to a family of transparent conducting oxides
(TCOs) - oxides with low electrical resistivity and high visual light transparency that show promise in multiple applications including liquid crystal displays, touch
screens, gas sensors, thermoelectric devices, solar cells, dilute magnetic
semiconductors, light-emitting diodes, etc (29–34). CuAlO2 nanoparticles have
also been used as thermo-catalysts in the thermal photo-catalytic reactions for the
evolution of hydrogen gas from water splitting (33). The most well-known TCOs,
ZnO and In2O3, are n-type semiconductors; therefore, the synthesis of a p-type
conducting oxide allows for the preparation of p-n junctions - the source of many
interesting semiconductor properties in cutting edge technology (35, 36).
Essentially, a p-n junction photovoltaic using CuAlO2 and In2O3 would allow visible
light to pass through the cell while producing electricity via the absorption of
ultraviolet radiation (37).

291

CuAlO2 has a layered crystal structure with Cu atoms between AlO2 layers
forming O-Cu-O pillars (38). CuAlO2 has indirect and direct band gaps of 1.2 and
3.6 eV respectively, high transparency (80%) in the visible region, and high lightto-electricity conversion efficiencies in solar cells (31, 39). Photocurrents ranging
from 0.02 mA/cm2 to 0.08 mA/cm2 have been observed when CuAlO2
nanoparticles (300 nm) were used as the photocathode materials in p-type DSSCs,
while delafossite CuGaO2 nanoplates produced comparatively high photocurrents
(0.384 mA/cm2) in the region of visible light using polypyridyl Co3+/2+(dtb-bpy) as
the electrolyte and the organic P1 dye (38, 40).
Delafossite CuAlO2 is thermodynamically unstable below 1000 °C in air, easily
decomposing to the Cu2+ compounds CuO and CuAl2O4 under these conditions
(41). Synthesis of polycrystalline CuAlO2 is typically achieved by firing a
stoichiometric mixture of Al2O3 and Cu2O at 1100 °C under argon or air
atmosphere.

292

Figure 5-8. Transmission electron microscopy pictures of CuAlO2 nanoparticles
prepared at 775 °C under a controlled partial oxygen pressure atmosphere (a,b),
800 °C under ultrapure N2 (c,d), and 950 °C under ultrapure N2 (e,f).

Various techniques have been used to produce CuAlO2 nanoparticles including
sol-gel, mechanical alloying, hydrothermal, field emission process, and electrospinning methods (32, 42–44). CuAlO2 nanostructured semiconducting materials
were synthesized by using boehmite (ϒ-AlOOH) nanorods loaded with Cu(II)
acetate as a nanoprecursor at 1150 °C in air (30). CuAlO2 nanowires (50 nm
diameter) and nanoparticles (35 nm) were produced via a porous anodic alumina
oxide (AAO) template at 900 °C and citrate route at 1000 °C, respectively (45).

293

Considering that most syntheses reported require very high reaction temperatures
(> 1000 °C) or complex instrumentation or templating, the production of bulk
quantities of pure phase CuAlO2 nanoparticles is still a serious challenge.

5.4b Synthesis of nanoparticle CuAlO2
Aqueous solutions of 0.1 M Cu(NO3)2∙2.5H2O and 0.1 M Al(NO3)3∙9H2O were
mixed in stoichiometric amounts and then transferred to a porcelain crucible
containing citric acid (0.15 mol) and ethylene glycol (10 ml). Initially, a sky blue
transparent solution was apparent. The solution was allowed to gel for three days
at room temperature before being undergoing thermal treatment at 350 °C for 6 hr
in air to remove the chelating organics, which resulted in a fine black powder of
copper (II) oxide and an amorphous aluminum based compound as determined by
PXD. The fine black powders were then fired at 775 °C for 12 hrs in air, leading to
the formation of CuO and CuAl2O4. This product was further treated at 800 °C in
ultrapure N2 for 48 hr, affording pure CuAlO2 nanoparticles.

5.4c Particle size distribution of CuAlO2 nanoparticles
A lognormal size distribution of spherical nanoparticles was used as previously
described (Equation 5-2). The results obtained from the WPPM algorithm showed
uniform size distribution, as well as strong correlation with the TEM studies (Figure
5-8). The average sizes of the CuAlO2 nanoparticles were calculated to be 35 ± 9
nm for the CuAlO2 nanoparticles prepared at 775 °C under controlled partial

294

oxygen pressure, 39 ± 14 nm for the nanoparticles prepared at 800 °C under
ultrapure N2, and 43 ± 17 nm for the CuAlO2 nanoparticles prepared at 950 °C
under ultrapure N2 (Figure 5-9). These values agree well with the particle sizes
observed from microscopy measurements; however, the WPPM method appears
to give the shortest length for anisotropic crystallite sizes.

Figure 5-9. Particle size distribution of CuAlO2 nanoparticles prepared at 775 °C,
800 °C, and 950 °C. Distribution plots calculated using the PM2K software
package.

295

5.5 Chapter conclusions
The WPPM method as implemented in PM2K is much more effective at
calculating the particle size of both nano- and micro-sized particles than the
Scherrer formula, as shown in the three case studies presented. TEM micrograms
were used as experimental proof that the particle size distributions calculated using
WPPM were accurate. In conclusion, a combination of WPPM and controlled
thermal annealing may allow for very precise control over both particle size and
strain of nanoparticles used in advanced functional materials.

296

REFERENCES

297

REFERENCES

(1)

Leoni, M.; Scardi, P. Nanocrystalline domain size distributions from powder
diffraction data research papers. J. Appl. Crystallogr. 2004, 37, 629–634.

(2)

Patterson, A. The Scherrer Formula for X-Ray Particle Size Determination.
Phys. Rev. 1939, 56, 978–982.

(3)

Leoni, M.; Confente, T.; Scardi, P. PM2K: a flexible program implementing
Whole Powder Pattern Modelling. Zeitschrift für Krist. Suppl. 2006, 2006,
249–254.

(4)

Lampe, U.; Gerblinger, J.; Meixner, H. Carbon-monoxide sensors based on
thin films of BaSnO3. Sensors Actuators B Chem. 1995, 25, 657–660.

(5)

Chu, X. Dilute CH3SH-sensing characteristics of BaSnO3 thick film sensor.
Mater. Sci. Eng. B 2004, 106, 305–307.

(6)

Ostrick, B.; Fleischer, M.; Lampe, U.; Mexines, H. Preparation of
stoichiometric barium stannate thin films: Hall measurements and gas
sensitivities. Sensors Actuators B Chem. 1997, 44, 601–606.

(7)

Hodjati, S.; Vaezzadeh, K.; Petit, C.; Pitchon, V.; Kiennemann, a. NOx
sorption–desorption study: application to diesel and lean-burn exhaust gas
(selective NOx recirculation technique). Catal. Today 2000, 59, 323–334.

(8)

Deepa, A. S.; Vidya, S.; Manu, P. C.; Solomon, S.; John, A.; Thomas, J. K.
Structural and optical characterization of BaSnO3 nanopowder synthesized
through a novel combustion technique. J. Alloys Compd. 2011, 509, 1830–
1835.

(9)

Yuan, Y.; Lv, J.; Jiang, X.; Li, Z.; Yu, T.; Zou, Z.; Ye, J. Large impact of
strontium substitution on photocatalytic water splitting activity of BaSnO3.
Appl. Phys. Lett. 2007, 91, 094107.

(10) Mizoguchi, H.; Woodward, P. M.; Park, C.-H.; Keszler, D. a. Strong nearinfrared luminescence in BaSnO3. J. Am. Chem. Soc. 2004, 126, 9796–800.
(11) Bao, M.; Li, W.; Zhu, P. Study on the dielectric properties of oxide-doped
Ba(Ti,Sn)O3 ceramics prepared from ultrafine powder. J. Mater. Sci. 1993,
28, 6617–6621.

298

(12) Udawatte, C. P.; Yoshimura, M. Preparation of well-crystallized BaSnO3
powders under hydrothermal conditions. Mater. Lett. 2001, 47, 7–10.
(13) Udawatte, C. P.; Kakihana, M.; Yoshimura, M. Preparation of pure
perovskite-type BaSnO3 powders by the polymerized complex method at
reduced temperature. Solid State Ionics 1998, 108, 23–30.
(14) Pfaff, G. Wet chemical synthesis of BaSnO3 and Ba2SnO4 powders. J. Eur.
Ceram. Soc. 1993, 12, 159–164.
(15) Nakade, S.; Saito, Y.; Kubo, W.; Kitamura, T.; Wada, Y.; Yanagida, S.
Influence of TiO2 Nanoparticle Size on Electron Diffusion and Recombination
in Dye-Sensitized TiO2 Solar Cells. Society 2003, 8607–8611.
(16) Hoffmann, M. J.; Hammer, M.; Endriss, A.; Lupascu, D. C. Correlation
between microstructure, strain behavior, and acoustic emission of soft PZT
ceramics. Acta Mater. 2001, 49, 1301–1310.
(17) Audebrand, N.; Auffrédic, J.-P.; Louër, D. An X-ray Powder Diffraction Study
of the Microstructure and Growth Kinetics of Nanoscale Crystallites
Obtained from Hydrated Cerium Oxides. Chem. Mater. 2000, 12, 1791–
1799.
(18) Audebrand, N.; Auffre, J.; Loue, D. X-ray Diffraction Study of the Early
Stages of the Growth of Nanoscale Zinc Oxide Crystallites Obtained from
Thermal Decomposition of Four Precursors. General Concepts on
Precursor-Dependent Microstructural Properties. Society 1998, 4756, 2450–
2461.
(19) Lu, W.; Schmidt, H. Hydrothermal synthesis of nanocrystalline BaSnO3 using
a SnO2•xH2O sol. J. Eur. Ceram. Soc. 2005, 25, 919–925.
(20) Scardi, P.; Leoni, M. Diffraction line profiles from polydisperse crystalline
systems. Acta Crystallogr. Sect. A Found. Crystallogr. 2001, 57, 604–613.
(21) Caglioti, G.; Paoletti, A.; Ricci, F. P. Choice of collimators for a crystal
spectrometer for neutron diffraction. Nucl. Instruments 1958, 3, 223–228.
(22) Diffraction Analysis of the Microstructure of Materials; Mittemeijer, E. J.;
Scardi, P., Eds.; Springer Berlin Heidelberg: Berlin, Heidelberg, 2004; Vol.
68.

299

(23) Azad, A.; Hashim, M.; Baptist, S. Phase evolution and microstructural
development in sol-gel derived MSnO3 (M = Ca , Sr and Ba). Mater. Sci.
2000, 5, 5475 – 5483.
(24) Azad, A.-M.; Liew, L.; Shyan, W.; Yen, P. T. Synthesis, processing and
microstructural characterization of CaSnO3 and SrSnO3 ceramics. J. Alloys
Compd. 1999, 282, 109–124.
(25) Sharma, N.; Shaju, K. M.; Rao, G. V. S.; Chowdari, B. V. R. Anodic behaviour
and X-ray photoelectron spectroscopy of ternary tin oxides. J. Power
Sources 2005, 139, 250–260.
(26) Williamson, G.; Hall, W. X-Ray Line Broadening from Filed Aluminum and
Wolfram. Acta Metall. 1953, 1, 22–31.
(27) Films, T.; Wallenberg, R.; Hagfeldt, A.; Sundstrom, V.; Yartsev, A. P. Particle
Size and Crystallinity Dependent Electron Injection in Fluorescein 27Sensitized TiO2 Films. Phys. Chem. B 2003, 1370–1375.
(28) O’Hayre, R.; Nanu, M.; Schoonman, J.; Goossens, A.; Wang, Q.; Grätzel,
M. The Influence of TiO2 Particle Size in TiO2/CuInS2 Nanocomposite Solar
Cells. Adv. Funct. Mater. 2006, 16, 1566–1576.
(29) Vanaja, K. A.; Ajimsha, R. S.; Asha, A. S.; Jayaraj, M. K. p-type electrical
conduction in α -AgGaO2 delafossite thin films. Appl. Phys. Lett. 2006, 88,
212103.
(30) Thu, T. V.; Thanh, P. D.; Suekuni, K.; Hai, N. H.; Mott, D.; Koyano, M.;
Maenosono, S. Synthesis of delafossite CuAlO2 p-type semiconductor with
a nanoparticle-based Cu(I) acetate-loaded boehmite precursor. Mater. Res.
Bull. 2011, 46, 1819–1827.
(31) Ghosh, C. K.; Popuri, S. R.; Mahesh, T. U.; Chattopadhyay, K. K.
Preparation of nanocrystalline CuAlO2 through sol–gel route. J. Sol-Gel Sci.
Technol. 2009, 52, 75–81.
(32) Li, G.; Zhu, X.; Lei, H.; Jiang, H.; Song, W.; Yang, Z.; Dai, J.; Sun, Y.; Pan,
X.; Dai, S. Preparation and characterization of CuAlO2 transparent thin films
prepared by chemical solution deposition method. J. Sol-Gel Sci. Technol.
2010, 53, 641–646.
(33) Smith, J.; Van Steenkiste, T.; Wang, X.-G. Thermal photocatalytic
generation of H2 over CuAlO2 nanoparticle catalysts in H2O. Phys. Rev. B
2009, 79, 041403.

300

(34) Zheng, X. G.; Taniguchi, K.; Takahashi, A.; Liu, Y.; Xu, C. N. Room
temperature sensing of ozone by transparent p-type semiconductor CuAlO2.
Appl. Phys. Lett. 2004, 85, 1728.
(35) Ferromagnetism, R.; Lu, Y.; Hsu, Y. Cu2+ doped ZnO Nanocrystals: VisibleLight-Driven Photoactivity and. Society 2011, 104, 2418.
(36) Libera, J. A.; Hryn, J. N.; Elam, J. W. Indium Oxide Atomic Layer Deposition
Facilitated by the Synergy between Oxygen and Water. Energy 60439.
(37) Robertson, J.; Gillen, R.; Clark, S. J. Advances in understanding of
transparent conducting oxides. Thin Solid Films 2012, 520, 3714–3720.
(38) Yu, M.; Natu, G.; Ji, Z.; Wu, Y. p-Type Dye-Sensitized Solar Cells Based on
Delafossite CuGaO2 Nanoplates with Saturation Photovoltages Exceeding
460 mV. 2012.
(39) Gillen, R.; Robertson, J. Band structure calculations of CuAlO2, CuGaO2,
CuInO2, and CuCrO2 by screened exchange. Phys. Rev. B 2011, 84,
035125.
(40) Nattestad, A. Dye-sensitized CuAlO2 photocathodes for tandem solar cell
applications. J. Photonics Energy 2011, 1, 011103.
(41) Kumekawa, Y.; Hirai, M.; Kobayashi, Y.; Endoh, S.; Oikawa, E.; Hashimoto,
T. Evaluation of thermodynamic and kinetic stability of CuAlO2 and CuGaO2.
J. Therm. Anal. Calorim. 2009, 99, 57–63.
(42) Deng, Z.; Zhu, X.; Tao, R.; Dong, W.; Fang, X. Synthesis of CuAlO2 ceramics
using sol-gel. Mater. Lett. 2007, 61, 686–689.
(43) Banerjee, A. N.; Maity, R.; Chattopadhyay, K. K. Preparation of p-type
transparent conducting CuAlO2 thin films by reactive DC sputtering. Mater.
Lett. 2004, 58, 10–13.
(44) Zhao, S.; Li, M.; Liu, X.; Han, G. Synthesis of CuAlO2 nanofibrous mats by
electrospinning. Mater. Chem. Phys. 2009, 116, 615–618.
(45) Mo, R.; Liu, Y. Synthesis and properties of delafossite CuAlO2 nanowires. J.
Sol-Gel Sci. Technol. 2010, 57, 16–19.

301