ABSTRACT
Title of dissertation: OXYGEN STORAGE PROPERTIES OF
TERNARY METAL OXIDE SYSTEMS
FOR CHEMICAL LOOPING REACTIONS
Rishvi S. W. A. Jayathilake
Doctor of Philosophy, 2019
Dissertation directed by: Professor Efrain E. Rodriguez
Department of Chemistry
We have studied the reversible uptake and release of oxygen in the layered
metal oxide system AB2O4 to understand their suitability as oxygen storage materi-
als. We examined their structures at their most reduced, oxidized, and intermediate
phases of AFe2O4 for A= Lu, Yb, Y, and In, and studied their structures with
high-resolution synchrotron X-ray diffraction. Under simulated chemical looping
conditions, we monitored their structures and reactivity towards H2 and O2 uti-
lizing in-situ X-ray diffraction, neutron diffraction, and thermogravimetric analysis
measurements. The nature of the trivalent A cation affects the oxidation kinetics,
thermal cycling stability, and oxygen storage capacity (OSC). With the exception
of the A = In analogue, these layered oxides underwent various phase transitions
above 200 ?C that included the creation of a superstructure as oxygen incorporates
until a high temperature phase is established above 400 ?C. To understand trends
in the oxygen incorporation kinetics, we employed bond valence sum analysis of
the Fe-O bonding across the series. The more underbonded the Fe cation, the more
facile the oxygen insertion. During the cycling experiments all samples exhibited re-
versible oxygen insertion at 600 ?C for this series, and displayed OSC values between
0.2-0.27 O2 mol/mol. The Y analogue displayed the fastest kinetics for oxidation,
which may make it the most suitable for oxygen sensing applications. The structure
of the oxidized phase was solved from with simulated annealing and Fourier differ-
ence maps. Structural parameters were reported with combine neutron and X-ray
Rietveld refinement. PDF and XAS were used to confirm the final structural model.
As the final steps experiments were carried out to explore the chemical looping reac-
tivity of AB2O4 layered oxides, with A= Lu, Yb, Y and B=Mn, Fe. We reported the
reactivity with methane of AB2O4 layered oxides for the first time. The RT pristine
structure was regenerated at 600 ?C under methane. Mn substituted compounds
exhibited faster kinetics and also higher oxygen storage capacities. We conclude
that the layered, ternary metal oxide system, AB2O4, is a suitable candidate as an
oxygen storage material for the potential application in chemical looping reactions.
OXYGEN STORAGE PROPERTIES OF TERNARY METAL
OXIDE SYSTEMS FOR CHEMICAL LOOPING REACTIONS
by
Rishvi S. W. A. Jayathilake
Dissertation submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
2019
Advisory Committee:
Professor Efrain E. Rodriguez, Chair
Professor Bryan Eichhorn
Professor Andrei Vedernikov
Professor Ichiro Takeuchi
Professor Jeffery Davis
? Copyright by
Rishvi S. W. A. Jayathilake
2019

Dedication
To my beautiful, forever young mother; the iron woman in my life
For the unconditional love
For all the laughs and tears
For the endless struggles fought on my behalf
For being my courage, my inspiration and my kindred spirit
?My ultimate inspiration comes from my best friend, the dazzling woman
from whom I received my name and my life?s blood, Lorelai Gilmore . . . My mother
never gave me any idea that I couldn?t do whatever I wanted to do or be whomever
I wanted to be. As she guided me through these incredible eighteen years, I don?t
know if she ever realized that the person I most wanted to be was her . . . Thank
you, Mom. You are my guidepost for everything.? ? Rory Gilmore
ii
Acknowledgments
I owe my gratitude to all the people who have made this thesis possible and
because of whom my graduate experience has been one that I will cherish forever.
First and foremost I?d like to thank my adviser, Professor Efrain Rodriguez
for being the best adviser I could ask for, patiently guiding me through out the
past five years, giving me space to transition, grow and thrive. I?m much grateful
to him for giving me an exciting research project that also gave me the opportunity
to work in national laboratories, which is a rare experience for an international
student. Thank you for making my graduate student life a memorable journey with
productive group meetings, numerous group lunches, BBQ get togethers, fun lessons
on Bonsai re-potting (Lady Boltzmann) and interesting discussions on alternative
endings to game of thrones. Simply for being the amazing human being he is.
I would like to thank Dr. Peter Zavalij at the X-ray crystallography center
for his tutelage through out the time I worked as his research assistant. For giving
me the opportunity to learn so much about crystallography. More than anything he
taught me the value of humility and the power of knowledge.
I?d like to extend my sincere gratitude to my dissertation committee Profes-
sor Bryan Eichhorn, Professor Andrei Vedernikov, Professor Ichiro Takeuchi and
Professor Jeffery Davis for sparing their invaluable time reviewing the manuscript.
I thank my former and current lab mates in the Rodriguez lab. Specially
Dan who was my mentor in the first two years in the lab, also Amber, Brandon,
Stephanie, Austin, Zhou, Tianyu, Tim, Lahari, Matt, Lenoard, Huafei and Justin
iii
for their constant support and encouragement. I?ve been extremely fortunate to be
part of such an uplifting team.
All the beamline scientists and staff member including Wenqian Xu, Andrey
Yakovenko, Saul Lapidus and George Sterbinsky at the Argonne National Lab.
Ashfia Huq, Katherine Page at the Oak Ridge National Lab. ?Modern Methods
in Rietveld Refinement for Structural Analysis workshop 2016? and thank all the
instructors.
I?d like to mention my undergraduate advisers professor K.M.N de Silva and
Rohini de Silva for their love and support during the early stages of my research
career. Many thanks to Dr. Dilushan Jayasundara, professor Gehan Amaratunge
and all the other advisers at SLINTEC where I was first employed as a scientist
after my undergraduate studies.
My best friend Anuradha, who stuck with me through thick and thin.
I?d like to remember with much love and thank all of my friends who checked
up on me including Natalia, Brenna, Mary, Ben, Bambi, Mevan, Yasas, Teodora,
Dulith. Apu, Maria, Romi, Aaron, Adi, Kyle, Anjula, Rano, Judith, Isuri, Nimesha,
Sayuri, Ayesha and Ruwandi. Nimasha (co-founder of SLAKE) who single handedly
took care of our organization while I was busy with thesis writing.
All my family members from Sri Lanka
Lastly and most importantly, I express my eternal gratitude and love to my
ever loving mother who stood by me every step of the way with her undying love and
support. It is impossible to remember all, and I apologize to those I?ve inadvertently
left out.
iv
Table of Contents
Dedication ii
Acknowledgements iii
Table of Contents v
List of Tables viii
List of Figures ix
List of Abbreviations xii
1 Introduction 1
1.1 Metal oxides in energy applications . . . . . . . . . . . . . . . . . . . 2
1.2 Chemical looping reactions . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Oxygen storage materials in chemical looping reactions . . . . . . . . 5
1.3.1 Binary metal oxides as oxygen storage materials . . . . . . . 6
1.3.2 Ternary metal oxides as oxygen storage materials . . . . . . . 7
1.3.3 Chemical looping combustion vs chemical looping reforming . 9
1.4 Metrics for designing and characterizing an oxygen storage material . 10
1.4.1 Reaction kinetics . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.4.2 Product selectivity . . . . . . . . . . . . . . . . . . . . . . . . 13
1.4.3 Oxygen storage capacity . . . . . . . . . . . . . . . . . . . . . 16
1.4.4 Long term physical and chemical stability . . . . . . . . . . . 19
1.5 The ternary metal oxide: AB2O4 as an oxygen storage material . . . 21
1.6 Outline of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2 Experimental Methods 25
2.1 Diffraction techniques . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.1.1 Crystalline materials . . . . . . . . . . . . . . . . . . . . . . . 25
2.1.2 Diffraction and Bragg?s Law . . . . . . . . . . . . . . . . . . . 28
2.1.3 X-ray and neutron diffraction . . . . . . . . . . . . . . . . . . 29
2.1.4 Powder diffraction and analysis of powder diffraction patterns 31
2.1.4.1 Le Bail fit . . . . . . . . . . . . . . . . . . . . . . . 32
2.1.4.2 Rietveld refinement . . . . . . . . . . . . . . . . . . 32
2.1.4.3 Determining the quality of a refinement and other
considerations . . . . . . . . . . . . . . . . . . . . . . 36
v
2.1.5 Synchrotron X-Ray and Time-Of-Flight neutron diffraction . . 39
2.1.6 In-situ gas flow diffraction experiments . . . . . . . . . . . . . 43
2.2 Pair Distribution Function analysis (PDF) . . . . . . . . . . . . . . . 47
2.3 Thermogravimetric Analysis . . . . . . . . . . . . . . . . . . . . . . . 51
2.4 X-ray absorption spectroscopy (XAS) . . . . . . . . . . . . . . . . . 52
2.5 ICP-AES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3 Oxygen storage properties of AFe2O4 (A= Yb, Lu, Y, In) 58
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.2.1 Materials synthesis . . . . . . . . . . . . . . . . . . . . . . . . 62
3.2.2 Ex situ high-resolution synchrotron X-ray powder diffraction
(sXPD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.2.3 In situ sXPD . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.2.4 Thermogravimetric analysis (TGA) . . . . . . . . . . . . . . . 65
3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.3.1 Structures of AFe2O4 and AFe2O4+ d from high-resolution
sXPD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.3.2 Structural evolution with temperature: in situ sXPD . . . . . 70
3.3.3 Cycling experiments with in situ sXPD . . . . . . . . . . . . . 73
3.3.4 Thermogravimetric analysis (TGA) . . . . . . . . . . . . . . . 73
3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.4.1 A-site substitution effects on the structure . . . . . . . . . . . 78
3.4.2 Bond valence sum (BVS) calculations . . . . . . . . . . . . . 83
3.4.3 Lattice expansion from oxygen insertion vs. thermal expansion 85
3.4.4 Thermochemical cycling capability and stability . . . . . . . . 88
3.4.5 Comparison of oxygen storage capacity . . . . . . . . . . . . . 92
3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4 Solving the structure of the oxidized phase of AFe2O4 (Yb, Lu) 96
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
4.2.1 Materials synthesis . . . . . . . . . . . . . . . . . . . . . . . . 110
4.2.2 High temperature in situ sXPD characterization . . . . . . . 110
4.2.3 Ex situ high-resolution sXPD and TOF NPD characterization 111
4.2.4 X-ray absorption spectroscopy . . . . . . . . . . . . . . . . . 111
4.2.5 Pair Distribution Function (PDF) Analysis . . . . . . . . . . 112
4.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.3.1 Formation of the oxidized phase, AFe2O4+?, from the pristine
AFe2O4 phase . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.3.2 Solving the structure with simulated annealing (SA) and Fourier
difference map . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
4.3.3 Combined Rietveld refinement and crystal structure . . . . . . 119
4.3.4 Pair Distribution Function (PDF) analysis . . . . . . . . . . . 122
4.3.5 X-ray absorption spectroscopy . . . . . . . . . . . . . . . . . . 125
vi
4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
5 Chemical looping reactivity of AB2O4 (A= Yb, Lu, Y B= Fe, Mn) 129
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
5.2.1 Materials synthesis . . . . . . . . . . . . . . . . . . . . . . . . 131
5.2.2 In situ TOF NPD on POWGEN at SNS . . . . . . . . . . . . 132
5.2.3 In situ sXPD on 17 BM at APS . . . . . . . . . . . . . . . . . 132
5.2.4 Elemental analysis from AES-ICP . . . . . . . . . . . . . . . . 133
5.2.5 Thermogravimetric analysis (TGA) . . . . . . . . . . . . . . . 134
5.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
5.3.1 Material characterization . . . . . . . . . . . . . . . . . . . . . 134
5.3.2 TOF NPD experiments . . . . . . . . . . . . . . . . . . . . . . 136
5.3.3 Methane reactivity of other A-site analogues . . . . . . . . . . 138
5.3.4 Cycling experiments of Mn substituted analogues . . . . . . . 140
5.3.5 Elemental analysis from AES-ICP . . . . . . . . . . . . . . . . 143
5.3.6 Thermogravimetric analysis . . . . . . . . . . . . . . . . . . . 143
5.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
5.4.1 Reactivity towards methane . . . . . . . . . . . . . . . . . . . 145
5.4.2 Elemental substitution effects on kinetics . . . . . . . . . . . . 147
5.4.3 Elemental substitution effects on cycling stability . . . . . . . 147
5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
6 Conclusions and future directions 154
6.1 Study I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
6.2 Study II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
6.3 Study III . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
6.4 Future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
6.4.1 Active oxygen species and product selectivity . . . . . . . . . 155
6.4.2 Improving overall performance as an OSM . . . . . . . . . . . 156
A Resources 160
A.1 Website resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
A.2 TOPAS macro used in this research . . . . . . . . . . . . . . . . . . . 161
A.2.1 TOPAS launch mode . . . . . . . . . . . . . . . . . . . . . . . 161
A.2.2 simulated annealing . . . . . . . . . . . . . . . . . . . . . . . . 161
B History of scientists who contributed to crystallography 164
Bibliography 168
vii
List of Tables
1.1 Oxygen storage capacities of ternary metal oxides reported in literature 19
2.1 The 7 lattice systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.2 Parameters affecting a powder diffraction pattern . . . . . . . . . . . 33
3.1 Structural parameters from combined Rietveld refinement done using
high resolution sXPD for AFe2O4 at 298 K (A=Lu,Yb, Y, In) . . . . 66
3.2 Comparison of bond lengths, unit cell volume and A3+ cationic radii
of the A-site analogues at RT. Q, R and S are labels for the Fe?O
bond lengths as shown in Fig. 3.14. . . . . . . . . . . . . . . . . . . . 83
3.3 OSCs calculated from TGA experiments. . . . . . . . . . . . . . . . . 93
4.1 Software packages available for Simulated annealing and Charge flipping106
4.2 Structural parameters from combined Rietveld refinement done using
high resolution sXPD for AFe2O4 at 298 K (A=Lu,Yb) . . . . . . . . 113
4.3 Structural parameters from combined Rietveld refinement done using
TOF NPD and high resolution sXPD for AFe2O4 at 298 K . . . . . . 121
4.4 Values of fitting variables resulting from a simultaneous fit of the first
Fe-O shell of EXAFS data from YbFe2O4, YbFe2O4+?, and LuFe2O4.
S20 and ?E0 were constrained to be equal for all three compounds. . . 126
5.1 Structural parameters from Rietveld refinement done using 17-BM
sXPD for YbFe2-xMnxO4 at 298 K (x=0, 0.25, 0.75, 1) . . . . . . . . 135
5.2 ICP-AES results forYbFe2-xMnxO4 . . . . . . . . . . . . . . . . . . . 143
5.3 Lattice parameters of the R-3m phase from the Rietveld refinement
done for the last powder pattern at the end of each 1 hour-cycle under
methane at 600?C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
B.1 The scientists who contributed to furthering diffraction . . . . . . . . 165
B.2 The scientists who contributed to furthering diffraction . . . . . . . . 166
B.3 The scientists who contributed to furthering diffraction . . . . . . . . 167
viii
List of Figures
1.1 An illustration of a chemical looping reactor. . . . . . . . . . . . . . 5
1.2 Gibbs free energy change of oxygen carrier reduction with 1 mol of
CO [1] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3 Ellingham diagram for oxygen carrier comparison [1] . . . . . . . . . 17
1.4 Zone of metal oxides for chemical looping [1] . . . . . . . . . . . . . 18
2.1 a) Depicts the periodic arrangement of atoms in a NaCl crystal lattice
b) The smallest repeating unit of a lattice is the unit cell. Every
crystalline material has characteristic unit cell edges (a, b, c) and
angles (?, ?, ?) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.2 The hierarchy of crystallographic classes . . . . . . . . . . . . . . . . 27
2.3 Centering in crystalline lattices . . . . . . . . . . . . . . . . . . . . . 28
2.4 Bragg?s law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.5 Relationship between d-spacing and lattice parameters . . . . . . . . 35
2.6 Instances of bad peak fit during a Rietveld refinement [2] . . . . . . . 38
2.7 How synchrotron X-rays are produced at APS . . . . . . . . . . . . . 41
2.8 How TOF neutrons produced at the Oak Ridge national lab . . . . . 42
2.9 The instrumental set up at 17 BM at the APS . . . . . . . . . . . . 44
2.10 In situ data collection on POWGEN at SNS . . . . . . . . . . . . . . 45
2.11 In situ data collection on 17 BM at the APS . . . . . . . . . . . . . . 46
2.12 Depiction of the total scattering factor or momentum transfer . . . . 48
2.13 Conversion of S(Q) in to PDF . . . . . . . . . . . . . . . . . . . . . 50
2.14 With the team at NOMAD, SNS . . . . . . . . . . . . . . . . . . . . 51
2.15 X-ray absorption phenomenon . . . . . . . . . . . . . . . . . . . . . . 53
2.16 XAS spectra processing with Athena . . . . . . . . . . . . . . . . . . 55
2.17 Experimental set up at 9 BM at APS . . . . . . . . . . . . . . . . . 56
2.18 ICP-AES instrument . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.1 The crystal structure information of AB2O4 . . . . . . . . . . . . . . 60
3.2 Sealing station . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.3 Rietveld refinement of LuFe2O4 and YbFe2O4 . . . . . . . . . . . . . 67
3.4 Rietveld refinement of InFe2O4 and YFe2O4 . . . . . . . . . . . . . . 68
3.5 Monoclinic distortion . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
ix
3.6 Le Bail fit for the oxidized phases Lu, Yb . . . . . . . . . . . . . . . . 71
3.7 Le Bail fit for the oxidized phases In, Y . . . . . . . . . . . . . . . . . 72
3.8 Contour plots of the sXPD Lu Yb . . . . . . . . . . . . . . . . . . . . 74
3.9 Contour plots of the sXPD Y In . . . . . . . . . . . . . . . . . . . . . 75
3.10 Supercell satellite reflections . . . . . . . . . . . . . . . . . . . . . . . 76
3.11 Contour plots of cycling experiments . . . . . . . . . . . . . . . . . . 77
3.12 TGA plots of Lu, Y . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.13 TGA plots of Yb, In . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.14 Linear dependence of lattice parameter on ionic radii . . . . . . . . . 82
3.15 Isolated in situ sXPD patterns of LuFe2O4 . . . . . . . . . . . . . . . 86
3.16 Cell volume evolution under Air Vs He . . . . . . . . . . . . . . . . . 87
3.17 Degradation of InFe2O4 under hydrogen . . . . . . . . . . . . . . . . 89
3.18 The evolution of lattice parameters as a function of temperature in
He and air . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.19 Comparison of unit cell volume change of the Lu and Yb in cycling . 91
3.20 TGA with different ramp rates . . . . . . . . . . . . . . . . . . . . . . 93
4.1 Simulated annealing . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.2 Charge flipping algorithm . . . . . . . . . . . . . . . . . . . . . . . . 107
4.3 Crystal structure AB2O4 . . . . . . . . . . . . . . . . . . . . . . . . . 108
4.4 Rietveld refinement of LuFe2O4 and YbFe2O4 . . . . . . . . . . . . . 114
4.5 Ramping in situ sXPD patterns of Lu Yb . . . . . . . . . . . . . . . . 115
4.6 Steps used in solving the structure with powder diffraction . . . . . . 116
4.7 Structure output from simulated annealing and Fourier difference maps118
4.8 a)Unit cell and b) Crystal structure in polyhedral view of AFe2O4
(A= Lu, Yb) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
4.9 Combined Rietveld refinement done for YbFe2O4+? . . . . . . . . . . 120
4.10 The layer movement due to oxygen uptake . . . . . . . . . . . . . . . 121
4.11 The PDF fit with PDFGui . . . . . . . . . . . . . . . . . . . . . . . . 124
4.12 XANES spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
4.13 EXAFS spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5.1 Rietveld refinement YbFe2-xMnxO4, x=0 (upper) x=0.25(lower) . . . 136
5.2 Cell volume evolution of YbFe2-xMnxO4, x=0 (upper) x=1(lower) . . 137
5.3 TOF NPD patterns of YbFe2O4 from cycling experiments . . . . . . . 138
5.4 Contour plots of YFe2O4 and LuFe2O4 . . . . . . . . . . . . . . . . . 139
5.5 Contour plots of YbFe2-xMnxO4, a) x=0 b) x=0.25 c=0.75 d= x=1 . 141
5.6 In situ sXRD patterns of YbFe2O4 and YbFeMnO4 . . . . . . . . . . 142
5.7 Thermogravimetric analysis done comparing the substitution effects
of the B-site. The TGA plots of the A-site series from Chapter 3 is
also shown for comparison . . . . . . . . . . . . . . . . . . . . . . . . 144
5.8 Cell volume evolution under methane Yb, Lu . . . . . . . . . . . . . . 145
5.9 Reduction kinetics of YbFe2O4 in CH4 Vs H2 . . . . . . . . . . . . . . 146
5.10 Contour plots of YbFe2O4 and YbFeMnO4 . . . . . . . . . . . . . . . 148
5.11 The unit cell volume evolution of YbFe2-xMnxO4, x=0, x=1 . . . . . 149
x
5.12 Rietveld refinements of YbFe2-xMnxO4in the reduction half cycle at
600?C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
5.13 Rietveld refinement for YbFe2-xMnxO4+d, x=0, 0.25 at 600?C . . . . . 151
5.14 Rietveld refinement for YbFe2-xMnxO4+d, x=0.75, 1 at 600?C . . . . . 152
6.1 Chemical looping reaction of YbFe2O4 in CH4 . . . . . . . . . . . . . 157
6.2 Methane reactivity rested forYFe2O4 at 500?C and 600 ?C . . . . . . 159
A.1 Input file in Jedit used in launch mode for TOPAS . . . . . . . . . . 162
A.2 Simulated Annealing macro . . . . . . . . . . . . . . . . . . . . . . . 163
xi
List of Abbreviations
APS Advanced Photon Source
ANL Argonne national laboratory
ADP Atomic displacement parameter
BVS Bond valence sum
CLC Chemical looping combustion
CLR Chemical looping reforming
CO charge ordering
CIF crystallographic information file
EXAFS Extended x-ray absorption fine structure
ICP-AES Inductively coupled plasma-atomic emission spectroscopy
NPD Neutron powder diffraction
OC Oxygen carrier
OSC Oxygen storage capacities
OSM Oxygen storage material
ORNL Oak Ridge national laboratory
PDF Pair distribution function
RT Room temperature
Rwp Weighted prole R-factor
sXPD Synchrotron X-ray powder diffraction
SNS Spallation Neutron Source
SA Simulated Annealing
SOFC Solid oxide fuel cell
TOF Time of flight
TGA Thermogravimetric analysis
XAS X-ray absorption spectroscopy
XRD X-ray diffraction
XANES X-ray absorption near edge structure
xii
Chapter 1: Introduction
As the world?s population is reaching 8 billion, the demand for energy, food
and other resources are increasing rapidly and so does the harmful impact on the
environment. One solution to this crisis is the discovery of functional materials
that are economically and environmentally sustainable. In fields like photovoltaics,
catalysis, electronics, pharmaceuticals, etc. the design and innovation of novel ma-
terials remain the most crucial and the most challenging task. In the past decade,
several families of materials have been the stepping stone for the advancement in
the technological, economical and social developments. For instance, lithium ion
batteries in electrochemical energy storage, silicon in the industry of electronics and
semi-conductors, multi-layered metal oxide thin films in magnetism and composite
graphite materials in consumer products. [3] These materials can be complex mate-
rials or simple materials build into complex structures. In order to provide better,
more effective and efficient materials to the world, one needs to look into the struc-
ture in the atomic scale and understand the structure property relationship of such
materials.
1
1.1 Metal oxides in energy applications
Metal oxides are versatile materials with ubiquitous applications, especially
in energy conversion and storage applications. Lithium ion batteries are prominent
candidates in the electronics and the automotive world. In the past decade, there
has been research carried out on lithium ion batter combined with transition metal
oxides. He et al discusses the usage of LiMO2 (M=Co,Ni, Mn) as cathode materials
in large scale Li-ion battery applications. [4] Transition metal oxides with layered
structures such as Olivine and Spinel have been a highlight in cathode materials.
Super-capacitors are another prominent energy application. Wang et al explores a
range of metal oxides including composites and nanomaterials as super-capacitors.
[5] It states that metal oxides provide super-capacitors with higher energy density
than conventional carbon materials, and improved electrochemical stability than
polymer based capacitors.
Soild oxide fuel cells (SOFC) are an environmentally friendly and an effi-
cient way of producing electricity from hydrogen, natural gas and other renewable
sources. [6] Solid state metal oxides are used as the electrolyte in SOFCs, which
transports oxygen from the cathode (air) to the anode filled with a fuel such as
hydrogen. ZrO2 is one of the most widely used solid state electrolyte together with
dopants such as Yb2O3, Nd2O3, Y2O3, Sc2O3, CaO and MgO. [6] Perovskites are
prominent candidates as mixed ion-electron conductors in SOFCs. Perovskites with
the stoichemetric formula ABO3 ( A= Y, La, Sr and B= Mn, Fe, Co, Cr) are often
studied as SOFC electrolyte materials. [7,8] There has also been a recent interest in
2
Bi2O3 as a SOFC material. [9]
Photovoltaic is another upcoming field as a source for renewable energy, and
metal oxides have been extensively studied as solar energy capture materials. Pho-
tovoltaic cells made from metal oxide semiconductors are chemically stable, nontoxic
and low in cost. [10] Metal oxides such as TiO2 [11] ZnO [12] SnO2 [13] SrTiO3 [14]
and Nb2O5 [15] are recognized as wide band gap electron conductors in photovoltaic
cells. TiO2, WO3 and BiVO4 are also used as the photoanode in water splitting
applications and metal oxides such as Co3O4, NiO, NiFeOx have been utilized as
catalysts in electrochemical water splitting. [16]
1.2 Chemical looping reactions
While the interest for sustainable and renewable energy is on the rise, the
world is still hugely relying on fossil fuel based energy sources. This dependency is
expected to be almost 80% of the world?s energy demand by 2040. [17] Hence, it?s
necessary to explore the ways and means to enhance the efficiency of existing fossil
fuel combustion based energy production and make the process environmentally
sustainable. One of the major drawbacks in such energy production methods is the
emission of CO2, which is one of the major green house gases. This issue can be
tackled by either reducing CO2 evolutions per unit of fuel consumed or by capturing
and sequestering CO2.
Chemical looping reactions are one such technology that was first introduced in
the early 1980s, which enables inherent CO2 capture and sequesteration by Richter
3
and Knoche. [18] Chemical looping reactions involve an oxy-fuel combustion process
with the participation of a metal oxide system that act as the oxygen storage material
(OSM). As shown in Fig.1.1 a chemical looping reactor consist of two chambers that
operate at elevated temperature. In the fuel reactor the metal oxide releases oxygen
oxidizing the fuel, and then the oxygen-depleted metal oxide gets transported and
regenerated in the air reactor. Based on the extent of oxidation, two types of
reactions can take place. 1) Chemical looping combustion (CLC), where a fuel is
completely oxidized in to H2O and CO2. This releases the most heat energy, and
is used in power generation and CO2 capture applications. 2) Chemical looping
reforming (CLR), where a fuel is partially oxidized in to CO and H2 (synthesis
gas). This is then used for other applications such as commercial H2 sequestration,
methanol, ammonia and fertilizer production. [19]
There are several advantages to the CLC method compared to other combus-
tion technique. One is that a pure CO2 form can be sequestered by condensing
water as there are no other gaseous products in the mixture. [19, 20] Furthermore,
since the fuel does not come in to direct contact with gaseous oxygen, it renders
flameless combustion making the overall process safer. Also the elimination of the
costly requirement of cryogenic distillation to purify/isolate oxygen through lique-
faction of atmospheric air. Prevention of NOx gas formation, which in turn prevents
the adulteration of the products. And finally providing selectivity/reactivity choices
for fuel conversions by the usage of specific oxygen carriers that will only release a
specified amount of oxygen enabling desired products formation [21,22] Syngas for-
mation using chemical reforming is in fact a very efficient method as it produces the
4
Depleted CLC CLR
 air CO + H O CO
2 2 + H2
AFe2O4+?
Air Fuel
Reactor Reactor
AFe2O4
Air CH
4
Fuel
Figure 1.1: An illustration of a chemical looping reactor.
desirable H2:CO ratio of 2:1 that is preferred for processes such as the Fischer Trop-
sch synthesis and methanol production. [23] The conventional method to produce
syngas is steam and dry reforming and both are endothermic reactions which are
energetically unfavorable. The main advantage of using chemical looping reaction
over other hydrogen production methods is that it produces a pure stream of hy-
drogen gas. [24,25] These advantages lead to an overall cost reduction and increase
in productivity in industrial applications.
1.3 Oxygen storage materials in chemical looping reactions
An oxygen storage material (OSM) is any chemical compound that has the
ability to reversibly release and uptake gaseous oxygen as a function of an external
5
factor such as temperature and pressure. The storage of oxygen in OSM is achieved
thorough the redox active metal ions that can readily switch between two oxida-
tion states with the participation of lattice oxygen in the structure. A good OSM
must possess high oxygen storage capacity (OSC), fast reactivity under alternating
reducing and oxidizing atmospheres, sufficient oxygen transport capabilities (bulk
to surface within the material), and excellent thermal and mechanical stability for
successive cycles at high temperatures. [17,22] Most of the OSMs studied in the past
have been binary transition metal oxides of Mn, Fe, Co, Ni, and Cu due to their
natural abundance and variable oxidation states. [19,26]
1.3.1 Binary metal oxides as oxygen storage materials
The syngas production through partial oxidation of methane was first achieved
using CeO2 as the oxygen carrier. [27,28] In initial studies, it was found that CeO2
was capable of oxidizing methane to form syngas in the presence of a Pt catalyst at
700?C . [27] Otsuka et al succeeded in increasing the reactivity of CeO2 by doping
it with Zr 4+ (Ce0.8Zr0.2O2), which increased oxygen vacancies in the material,
and thus managed to decrease the operating temperature to 500 ?C with the Pt
catalyst. [28] In a study done using Gd, La and Pr as dopants, it was found that Pr
increases the activity of syngas generation in ceria-zirconia oxides. These authors
further claim that the reactivity towards methane at high temperature is primarily
controlled by the diffusion of lattice oxygen while at low temperatures (lowest 550?C)
its driven by surface bound oxygen. [29] Sadykov et al used Sm3+ as the dopant
6
and showed that the selectivity of methane for syngas generation increases while
maintaining a high stability in redox cycles.citeSadykov2011
A few reports have shown that incorporating Fe3+ into ceria increases the
oxidation reactivity due to the formation of surface defects. [30,30,31] Furthermore,
Fe2O3 has also elevated syngas production with increased adsorption of methane. [32]
CeO2-Fe2O3 system has received much attention due to the high abundance and low
cost. Also an improved performance of selective oxidation of methane and increased
stability resulted from the Fe-doping. It was also discovered that syngas production
from controlled oxidation of methane by the Ce-Fe-O system is strongly influenced
by the specific surface area, and by having a high surface area it has led to the
complete oxidation of methane to carbon dioxide and water. Comparisons have also
been made between Ce-Zr and Ce-Fe systems. In one such study it was observed
that both these systems have similar activity but Ce-Fe has a better selectivity for
syngas production at higher temperatures. [33] The impact of numerous supports
on Ce-Fe-Zr solid solutions have also being investigated. Among such supporting
materials, Al2O3 has induced complete oxidation of methane while SiO2 has reduced
the reactivity and MgO has enhanced both the activity and selectivity for syngas
generation. 27
1.3.2 Ternary metal oxides as oxygen storage materials
In the recent past, ternary metal oxides such perovskites have also drawn a lot
of attention as an OSM owing to their high structural stability and flexibility. [34]
7
Complex crystal structures of lanthanide and transition metal oxides with non-
stoichiometric lattice oxygen have proven to show reversible oxygen uptake/release
properties and has the potential to be used as OSMs. [35, 36] LaFeO3 was the first
perovskite that was successfully researched for oxidation of methane and was also
compared to two other perovskites NdFeO3 and ErFeO3. [37] In this structure Fe3+
was reduced to Fe2+ by methane. It was stated that LaFeO3 showed the best
performance out of the three. That shows even though it is the B cation that
actively takes part in the redox reaction, the A cation has a significant impact on the
reactivity of the material. Numerous studies have been conducted since, with various
elemental substitutions in ?A? and ?B? sites to see the impact towards activity and
selectivity of methane oxidation. Based on an investigation done by several authors,
it was claimed that there are two types of oxygen species in the oxide material that
take part in the oxidation process: 1) Weakly bound lattice oxygen that causes
complete oxidation of the fuel and reacts in lower temperatures.2) strongly bound
lattice oxygen that is capable of partially oxidizing the fuel and suitable for chemical
reforming reactions at higher temperatures. [38] In most cases both chemical looping
combustion and reforming can take place at elevated temperature and it is important
to design materials with higher selectivity. This selectivity has been achieved by
researchers again by different elemental substitutions and combinations. This lead
to the usage of A site and B site ordered perovskites, in which either the A or the B
site is shared by more than one element. And double perovskites where the A layers
alternate between two elements. For example, in an investigation done by Evdou et
al, where the A site ordered- perovskite La1-xSrMO3 (Mn, Ni) were studied, it was
8
concluded that the selectivity towards syngas formation increased with a relatively
low Sr substitution in the structure. [24] Figure 2 depicts the chemical looping cycle
of La1-xSrxFeO3-reported in a study done by Taylor et al. in which Sr content has
shown to increase the reactivity with methane. [39] The selectivity towards syngas
generation is also affected by the Sr content and it has been determined that the
optimal Sr substitution for syngas production is 0.3-0.5 for La1-xSrxFeO3. [40] In
some studies perovskites have also been tested with other catalysts such as Ni/NiO.
Depending on the type of metal used in the perovskite, the presence of a catalyst
has shown to either increase reactivity or have no effect at al. [41]
1.3.3 Chemical looping combustion vs chemical looping reforming
CLC is the process where a fuel is completely oxidized to produce CO2 and H2O
and the OSM should release sufficient oxygen in order to achieve this. The redox
properties of the metal oxide depends on various factors including its thermodynamic
properties. Fig. 1.2 illustrates the Gibbs free energy for the reduction of some
binary metal oxides with carbon monoxide as a function of temperature. This can
be used to evaluate the extent of oxidation of a fuel by a metal oxide system. In this
figure a few metal oxides approach the O2 line above 800 ?C. These metal oxides
can release gaseous O2 to the gas phase above 800 ?. Chemical looping processes
that use metal oxides with the ability to release molecular oxygen are known as
chemical looping oxygen uncoupling (CLOU). [42] CLOU are net exothermic with
higher reaction rates and results in complete combustion of the fuel. Some examples
9
are copper/manganese-based oxides [43] and modified perovskite materials. [44].
CLOU is more often used for solid fuel combustion such as coal. [45] CuO, Co3O4
and Mn2O3 are some examples for oxygen uncoupling metal oxides that give rise
to CLC. On the other hand Fe2O3 is a non-oxygen uncoupling metal oxide. In
NON-CLOU metal oxides, the lattice oxygen take part in the reaction instead of
molecular oxygen. The best way to differentiate between a CLOU metal oxide from
a non-CLOU metal oxide is to heat the metal oxide under an inert atmosphere. The
CLOU metal oxides release molecular oxygen beyond a certain temperature while
the other doesn?t. Fe2O3 is an ideal candidate for non-CLOU reactions. [1] Iron
oxides are a preferable due to its low cost, non-toxicity, high melting points and
high mechanical strengths.
1.4 Metrics for designing and characterizing an oxygen storage ma-
terial
In the world of energy applications, numerous efforts have been made to in-
crease the efficiency of the conversion of carbonaceous fuels such as coal, methane
and biomass using the chemical looping phenomenon. [46]In this process, figuring
out the most reactive, highly stable, cost effective and environmentally friendly OSM
remains to be the most crucial challenge. Extensive research has been carried out
to improve and establish the optimum OSMs. There are several metrics and char-
acteristics that can be used to determine whether a metal oxide system is a good
candidate as an oxygen storage materials. In research studies these metrics are qual-
10
Release O2 to the
gas phase
Figure 1.2: Gibbs free energy change of oxygen carrier reduction with 1 mol of
CO [1]
11
itatively and quantitatively analyzed. 1) The first and the most important metric is
the reversible redox reactivity. In order for the metal oxide to reversibly uptake and
release oxygen, a redox active metal center should be present. The metal ion should
be reactive towards alternating oxidizing (air) and the reducing (fuel) environments
at the operational temperature. 2) Prolonged cycling stability is equally important
for a metal oxide to be used as an OSM. Both physical and chemical stability are
taken in to account. 3) High oxygen storage capacity(OSC) is another metric used
to quantitatively characterize OSMs. The storage capacity refers to the amount of
oxygen that can be stored in a unit mass of material either in oxide form or molec-
ular oxygen form. Both diffraction data and thermogravimetric analysis(TGA) can
be used to calculate the OSC
There are several attributes that need to be considered when designing a metal
oxide system as an oxygen storage material such as reversible redox reactivity, prod-
uct selectivity, high oxygen storage capacity, fast redox kinetics and long term physi-
cal and chemical stability.Other than that these systems should also be economically
feasible and environmentally benign.
1.4.1 Reaction kinetics
Favorable reaction-kinetics in chemical looping reactions ensure fast reactivity
and higher efficiency. Various factors contribute to the fast reactivity such as particle
size and shape, porosity, structural density and composition. Higher surface area
due to smaller particle sizes and higher porosity provide larger reactive area for
12
the solid-gas reaction to take place leading to faster kinetics. When looking into
the structural chemistry of reaction kinetics, ionic diffusion and oxygen vacancy
formation, both the bulk and the surface should be taken in to consideration. This
governs the oxygen mobility rates within the material. Cheng et al discusses the
morphological evolution that takes place over the cause of the reaction due to ionic
diffusion and oxygen vacancy formation. [47]. Metal-oxygen bond energy, crystal
structure and oxygen vacancy formation are some significant factors that affect
oxide ion transport within the structure that ultimately affects the overall reaction
kinetics. [48]CeO2 is a well known OSM with the flourite structure, which favors
the diffusion of oxide ions and oxygen vacancy formation. [49] When more oxygen
vacancies are present, the easier is the oxide ion diffusion. However, it should be a
trade off between enhanced oxygen mobility and oxygen storage capacity.
1.4.2 Product selectivity
The type of metal oxide of choice is one of the significant parameters in predict-
ing the products in methane combustion. CO and H2 are the products in chemical
looping reforming while CO2 and H2O are the main products in chemical looping
combustion as discussed earlier in this chapter. Product selectivity can be fine
tuned by selecting metal oxides that release just the right amount of oxygen that
will give rise to only one of these reactions. This can be predicted by looking at the
relative metal oxide reduction potentials. The Ellingham diagrams provide Gibbs
free energy of reactions at different temperatures from which reduction potential
13
can be determined. Fig. 1.3 depicts Ellingham diagram for several metal oxide
systems [1] and Fig. 1.4 is adopted from this to illustrate the oxidation capability of
some metal oxides. Figure 1.4 is divided in to several zones which correlates to the
product selectivity that can be achieved from these oxygen carriers. Most chemical
looping reactions have operational temperatures that are above 750 ?C. Hence, the
discussion on product selectivity is focused on reactivity in high temperature. There
are three main zones sorted based on the following three reactions 1. 2CO + O2 =
2CO2 (Blue line)
2. 2H2 + O2 = 2H2O (Green line)
3. 2C + O2 = 2CO (Pink line)
Zone A: Region between blue and green lines. Metal oxides in this region
with lower negative Gibbs free energy of oxidation has a higher tendency to release
oxygen and act as strong oxidizers. Hence, these are the most suitable for the
complete oxidation of methane or for CLC. Metal oxides in this region include NiO,
CuO, CoO, Fe2O3,and Fe3O4. Most of the oxygen uncoupling metal oxides, i.e metal
oxides that release molecular oxygen to the gas phase at high temperature, belong
to this region. These zone-A metal oxides are the most suitable for solid fuels such
as carbon. These oxygen carriers can still be used for the partial oxidation (CLR)
of methane under controlled conditions. This can be done by limiting the extent of
reoxidation of the depleted metal oxide in the regeneration step. One instance is
the study done by DieDiego et al using a fluidized bed reactor with Ni based oxygen
carriers. In this report, it was found that by controlling the fraction of Ni reoxidized
in to NiO, a pure stream of syngas could be obtained. According to their findings,
14
the optimum ratio that gave the purest stream of syngas was Ni:NiO = 7:3. This
was done by controlling the air consumed in the air half reactor. [50]. However, this
approach is rather difficult to control in the long term.
Zone B: Region below blue and green and above pink line. Metal oxides in
this region have weak oxidizing properties and will give rise to partial oxidation of
methane. Hence, these oxygen carriers are more suitable for CLR. CeO2 is one such
example. The Ellingham diagram only provides information on product selectivity
based on thermodynamic constraints. However, it should be noted that other fac-
tors such as reaction kinetics, reactant stoichiometry, gas-solid flow dynamics, and
reactor design also play a major role in product selectivity. The end result is a
synergistic effect of all of these parameters
Zone C: Region below pink line Metal oxides in this region are strong reducers
hence fail to release oxygen and act as oxygen storage materials. Cr2O3 and SiO2 are
such inert metal oxides. These materials can however be used as inert supporting
materials or as active materials in combination with zone A and B materials to act
as composite OSMs. Nevertheless, in designing such composite-OSMs, extreme care
should be taken in selecting compatible metal oxides. For instance, if CuO (zone
A) and Al2O3 (zone C ) are combined they will chemically react to form CuAl2O4,
which will not have the same oxygen uncoupling property as CuO. On the other
hand, by combining FeO(zone B) and TiO2(zone C), it will form FeTiO3, which is
a far superior oxygen carrier than any of the individual constituent metal oxides.
Transition zone: Metal oxides that fall between the blue and green lines belong
to this region. These metal oxides may be used for CLR. However, a major drawback
15
is getting H2O as a byproduct, which adulterates the Syngas.
There are also different set reactions named oxidative coupling of methane
(OCM), where methane is slectively oxidized to produce other useful hydrocarbons
such as ethylene. V2O5 is an established OSM used for the selctive oxidation of
methane to produce formaldehyde. [49]. Many studies have been carried out on
what type of metal oxides are the most suitable for OCM and a general conclusion
is that, rare earth metal oxides tend to have a higher selectivity and OSMs with
p-type and oxide conductivity within the band gap of 5-6 eV to have the highest
performance for methane conversion in OCM. [51]
1.4.3 Oxygen storage capacity
In an OSM, the oxygen storage capacity (OSC) is defined as the amount of
oxygen that is released under a reducing atmosphere, or the amount taken up under
an oxidizing atmosphere. [52] The OSC is an important characteristic of OSMs that
defines the methane conversion rate. Typically binary metal oxides such as Fe2O3
have high OSCs, approximately upto 3.4 wt %. [22] However, due to their tendency
to agglomerate and lack of long term stability, this feature is rather overlooked.
Instead, attention is given towards developing high capacity ternary metal oxides,
which are superior in stability to binary metal oxides. The oxygen released in an
OSMs can be either in molecular form of oxygen (chemical looping uncoupling) or in
the oxide form of oxygen. The OSC is typically calculated using thermogravimetric
analysis where the weight gain is monitored under an oxidizing atmosphere as a
16
Figure 1.3: Ellingham diagram for oxygen carrier comparison [1]
17
Figure 1.4: Zone of metal oxides for chemical looping [1]
18
Conditions Oxygen storage capacity
Compound Temperature Cycling atmo- wt mmol Ref
spheres perc O2/g
BaYMn2O5+? 300 to 600 ?C 5 % H2 and 3.7 1.2 3
100% O2
Dy0.7Y0.3MnO3+? 200 to 400 ?C 2.0 0.62 3
air and 100%
O2
HoMnO3+? 300 ?C Air and 100% 1.7 0.54 2
O2
Sr3Fe2O7?? 950 ?C 5% H2and air 2.0 0.62 1
LaxSrxCoxFexO3?? 400 to 600 ?C 3.6 1.1 1
5% H2 and air
BaYCo4O7+? 350 ?C N2 and 3.5 1.1 1
100% O2
LuFe2O4+? 200 to 400 ?C 5% H2 and 2 ? 2.2 0.69 11
10-4 atm pO2
Ca2(AlxMn1- 300 to 700 ?C 100% N2 and 3.0 0.94 1
x)2O55+? 100% O2
Ce0.7Cu0.3O2+? 700 ?C 5% H2 and air 3.2 1.0 2
La1-xSrxFeO3+? 600 to 835 ?C 15% methane 2.3 0.7 11
and air
Table 1.1: Oxygen storage capacities of ternary metal oxides reported in literature
function of time. It is also calculated by using neutron diffraction data where the
oxygen occupancy is calculated through Rietveld refinement in the oxidized form of
the OSM. [53] The table 1.1 depicts some of the reported OSC of various OSMs in
literature. The largest reported so far is for La2O2SO4. [54]
1.4.4 Long term physical and chemical stability
OSMs should be resistant towards physical and chemical degradation in the
face of repeated, numerous cycles in the chemical looping process. Lack of long term
stability decreases the efficiency of the process adding more cost and ultimately
replacement of the oxygen carrier. In terms of physical stability, the metal oxide
particles should be resistant towards attrition. Inactivity of OC can also occur due
to physical processes such as agglomeration and carbon deposition on the materials?
19
surface. The formation of carbon originated from either CH4 decomposition or CO
disproportionation (In the case of CLR). Numerous studies have been carried out to
understand the causes of C formation. It is evident that the type of metal present
in the metal oxide has a clear impact on this. In Ni-based OC systems, it was found
that the carbon formation occurred when 80% of the available oxygen was consumed.
Hence, the carbon formation started towards the end of the reduction cycle where
oxygen is depleted. A similar observation was made for Cu based OSMs. However,
this was in contrast to Fe based systems where carbon formation was almost none
to negligible. [55]
Agglomeration is another major problem that retards the reaction rates. It
minimizes the effective surface area of the metal oxides that comes in to contact with
the gaseous reactants. This is especially a problem in fluidized bed reactors. [22]
The supporting material, the metal content and calcining temperature are three
of the main factors that govern the extent of agglomeration in these metal oxide
systems. In the case of NiO, the usage of supporting materials such as Al2O3, ZrO2
in fluidized bed configuration has shown to reduce agglomeration. [56] However,
agglomeration happened for Mn based OSMs that used Ca, Mg, Ce modified ZrO2
as the supporting material. Interestingly, MgO modified ZrO2 supporting material
did not show any agglomeration even after 70 hours of operation. [57] In Cu based
OSMs, agglomeration seemed more common due to the low melting temperature of
Cu (1085 ?C) [58]
20
1.5 The ternary metal oxide: AB2O4 as an oxygen storage material
Iron based OSMs in chemical looping applications are a popular choice due
to their many inherent attributes, including the low cost and low toxicity with
negligible impact on the environment. The cost of manufacturing an OSM mainly
comes from the cost of raw materials. As a result of high natural abundance and
advanced mining technologies, many precursor materials for Fe-based OSMs come
at a low cost. This is a huge advantage in an industrial set up. Iron is an entirely
safe element, that is also an essential nutrient. Hence many precursor compounds of
iron are considered environmentally benign. Iron oxide has several oxidation states,
and it has been found that the reduction step of Fe2O3 Fe3O4 is faster than that
of Fe3O4 FeO and FeO Fe steps. [59] Several gas and solid fuels have been tested
against Fe2O3, and the reaction rates in general change in the order of hydrogen >
carbon monoxide> methane > solid fuels. [60].
There have been a large number of Fe-based oxygen carriers explored in lit-
erature in chemical looping. Binary metal oxides of Fe, including Hematite are the
most established oxides as OSMs. [46, 59] Ternary metal oxides that contain Fe as
the redox active ion has also being extensively researched including perovskites [53]
and spinel compounds [61]. Compared to these, the ternary metal oxide AFe2O4 is
relatively a new compound as an OSM. This family of compounds are well known for
it?s multiferroic properties. [62] The oxygen storage properties of these compounds
are first reported for LuFe2O4 in 2013 [63]
The crystal structure of this family consists of a triangular lattice of edge
21
sharing AO6 octahedral layers alternating with bilayers of edge sharing BO5 trigonal
bipyramids (Figure 3). Two space groups have been used to describe the crystal
structure so far, one belonging to the trigonal Rm-3 and the other belonging to the
monoclinic C2/m (LuFe2O4).30 The A-site is a trivalent cation (usually rare earth
elements and also Y, In), and the B site consists of a 1:1 mixture of a divalent and a
trivalent cation (usually d-block elements). The AB2O4type material was discovered
with the first reports on LuFe2O4 by two different research groups simultaneously.
[64, 65] The main research focus of these materials in the past was heavily based
on the spin, charge and orbital frustration of the triangular lattice of the transition
metal oxide layer and has given rise to many publications regarding magnetism,
charge ordering, ferroelectricity, linear and non-linear electronic conduction and
crystal chemistry. [66]
In 2013 M. Hervieu et al. reported the first investigation on the oxygen storage
capacity of LuFe2O4. [63] In this study they conducted experiments by heating
LuFe2O4 under a dynamic vacuum while monitoring the weight gain due to oxidation
with thermogravimetric analysis, and studied the structural evolution with in-situ
and ex-situ X-ray diffraction. They managed to regenerate the starting material
by reacting it under hydrogen at 500 ?C stating the suitability of LuFe2O4 for a
potential OSM for the first time.3 Therefore we propose to synthesize and study
other members in this family as an OSM. We have thus far synthesized several
A site analogues of AFe2O4 (A= Lu, Yb, Y, In). And since Yb, Y and In are
cheaper elements than Lu, it will make them more viable for a scaled up industrial
application. The B site contains a redox active metal cation which has an average
22
oxidation state of +2.5. Upon oxidation it become +3 giving the structural formula
of AB2O
2+ 3+
4+?. The coexistence of the Fe and Fe ions has been confirmed through
a Mossbauer spectroscopic analysis done by J. Bourgeois et al. 34 In the next step
we intend to do substitutions for the B site with other transition metals that have
a stable divalent and a trivalent cation such as Co, Mn and Cr.
1.6 Outline of the thesis
The research work in this dissertation focuses on discovering the capabilities,
limitations and structure-property relationship of the ternary metal oxide system
AB2O4 as a novel oxygen storage material.
Chapter 2 describes the experimental techniques used that are used through
out this research. Mostly diffraction based techniques are used for the characteriza-
tion purposes.
Chapter 3 explores four A-site analogues of AB2O4 ( A= Lu, Yb, Y and
In) as oxygen storage materials. Except for Lu, the other analogues have never
been studied for the potential application in chemical looping. We looked at the
reversible oxygen insertion capabilities under reducing-oxidizing atmospheres at high
temperatures. In situ synchrotron X-ray and Time of Flight neutron diffraction
together with thermogravimetric analysis are used to quantify the metrics for oxygen
storage materials.
Chapter 4 is in fact an outcome of an unexpected problem encountered during
the research work discussed in Chapter 3. At the time of this study, the structure of
23
the oxidized phase of these ternary metal oxides (which forms at high temperature
under air) was not completely solved. And in order to understand where the excess
oxygen is getting inserted in to the structure, we explored methods to solve the
crystal structure of the oxidized phase. However, a method to form single crystals
could not be established. Hence, we attempted and successfully solved the crystal
structure using powder diffraction based computational techniques. X-ray absorp-
tion spectroscopy and Pair distribution function analysis are carried out to confirm
the final structural model.
Chapter 5 discusses the reactivity of these metal oxides towards methane.
Once again the methane reactivity has never been reported for this family of metal
oxides before. Gas flow experiments are carried emulating the conditions in a chem-
ical looping reactor with in situ X-ray and neutron diffraction data collection. This
chapter also addresses the B-site substitutions that are carried out to increase the
reactivity towards methane.
24
Chapter 2: Experimental Methods
2.1 Diffraction techniques
2.1.1 Crystalline materials
Crystalline materials have structures with atoms/ions that show long range
ordering and periodicity. This is in contrast to amorphous materials which lack
ordered arrangement of atoms. This periodic structure of a crystalline material is
called a lattice, and in a lattice the smallest repeating structural entity is called
the unit cell (Fig. 2.1 ). If the atoms and their arrangement in the unit cell is
known, the structure of the whole crystal can be known. Hence when characterizing
a crystalline material we merely characterize the contents and the dimensions of
a unit cell. Based on symmetry and dimensions of these unit cell there are seven
crystal systems every crystalline material belongs to, which are listed in Table. The
three lattice centering shown in Fig. 2.3 combines with these 7 crystal systems to
give fourteen Bravais lattices(Fig. 2.2).
25
a
Na
Cl
b
Figure 2.1: a) Depicts the periodic arrangement of atoms in a NaCl crystal lattice
b) The smallest repeating unit of a lattice is the unit cell. Every crystalline material
has characteristic unit cell edges (a, b, c) and angles (?, ?, ?)
26
6 crystal families
7 crystal systems 7 la?ce systems
14 Bravais la?ces
32 point groups
73 crystal classes
230 space groups
Figure 2.2: The hierarchy of crystallographic classes
Lattice system Restriction on cell parameters Centering
1.Triclinic None P
2. Monoclinic ? = ? or ? = 90? P, C
3. Orthorhombic ? = ? = ? = 90? P, I, F, C
4. Tetragonal a = b, ? = ? = ? = 90? P, I
5. Rhombohedral a=b=c (primitive) ? = ? = ? = R
90?or a=b, ? = ? =90?? = 120?
6. Hexagonal a=b (primitive) ? = ?=? = 90 P
7. Cubic a= b=c, ? = ? = ? =90C? P, I, F
Table 2.1: The 7 lattice systems.
27
P F
I C
Figure 2.3: Centering observed in crystalline lattice.a)Primitive (No centering)
b)Body centered c) Face centered d) Base centered.
2.1.2 Diffraction and Bragg?s Law
In early 1900s, Max Von Laue first discovered that the atoms can diffract X-
rays. This is feasible because the atomic dimensions are comparable to the wave
length range of X-rays. This was an important observation that lead to the usage X-
rays for the characterization of crystalline materials. In crystalline materials atoms
are arranged in specific order forming planes of atoms. Henry and Lawrence Bragg
(father and son) figured out a way to map the crystal structure of materials using X-
rays for the first time. In crystalline materials, planes of atoms with specific ordering
have characteristic inter-planar spacing known as d-spacing. And they proposed
that this d-spacing can be correlated to the wave length of X-rays and the angle of
28
? 2dsin? = n?
? ?
d ? ?
dsin ?? ndsi
Figure 2.4: Illustration of Bragg?s law. ?d? is the inter-planar distance between two
adjacent crystal planes belonging to the same Bragg?s family.
diffraction of these X-rays off of these planes. This mathematical correlation was
put forward as the Bragg?s law and Fig. 2.4 illustrates this physical phenomenon.
The Bragg?s law is fundamentally the most important theory that made it possible
to characterize the crystal structures with diffraction.
2.1.3 X-ray and neutron diffraction
Many advancements in the current world of materials sciences were made
possible only after the discovery of X-ray and neutron diffraction. In the history of
Nobel victories, four Nobel prizes were awarded for diffraction based discoveries. [67]
W. C. Ro?ntgen (1891) for the discovery of the X-rays subsequently named
after him. M. von Laue (1914) for his discovery of the phenomenon: Diffraction
29
of X-rays by crystals. W. H. Bragg and W. L. Bragg (1915) for laying down the
foundation for the analysis of crystal structure with X-rays. B. N. Brockhouse
and C. G. Shull (1994) for remarkable contributions to the development of neutron
scattering techniques for studies of condensed matter. In Appendix C, Table B.3
summarizes some of the notable contributions to the field of diffraction.
X-ray and neutron diffraction fundamentally follow the same basic principle
of Bragg?s law. However, owing to some of their complementary characteristics,
simultaneous usage of both techniques render combined advantages. X-rays are
electromagnetic radiation and interact with the electrons in an atom. Hence the
scattering power is directly proportional to the atomic form factor. Neutrons are
particles and they interact with the nucleus of atoms and the scattering power
does not depend on the electron cloud around it. It rather depend on the neutron
scattering length and has no clear trend as the atomic form factors. [68] Neutron
scattering lengths and cross sections can be found on the NIST website.(refer to the
website resources in Appendix B ). As a result, neutron diffraction is more suitable
for characterizing lighter elements such as oxygen. In a compound that has both
heavy and light elements, or having elements with similar atomic numbers, using
both X-ray and neutron diffraction give optimum results.
Another advantage of using neutrons over X-rays is the difference in pene-
tration power. Neutrons have higher penetration power than X-rays, making them
more suitable for bulk analysis. [67] Also due to the presence of a magnetic moment,
neutrons are often used to characterize magnetic materials. Finally inelastic neutron
scattering is an important probe to study the dynamic behavior of materials. This
30
is because the energies of thermal neutrons (10-110 mev) are comparable to that of
phonons in the crystal lattice. This is done by measuring the vibrational density of
states. [67]
2.1.4 Powder diffraction and analysis of powder diffraction patterns
Crystalline materials are mainly of two forms; single crystals and powder/polycrystalline
materials. In this thesis, all materials that are used are powder samples. Hence,
powder diffraction is used as the main form of diffraction technique for data analy-
sis. In powder diffraction, powder diffractometers are used to analyze polycrystalline
samples. In a diffractometer, X-rays are produced in a X-ray tube and scattered
on/through a sample, which are then diffracted and detected by an X-ray detector.
The output is a diffraction pattern that plots scattered intensity vs Bragg angle.
This diffraction pattern can be qualitatively and quantitatively analyzed to reveal
information about the material. In qualitative analysis the diffraction pattern is
matched against a database of patterns (ex: ICSD- Inorganic Structure Database).
This can be done either manually with visual matching of the patterns or by using a
software such as Diffract EVA or JADE. This allows phase identification and deter-
mination of the presence of impurities. Quantitative analysis is a more complicated
matter. There has been many advances in this regard over the years. The next few
topics cover the main quantitative analysis methods used in this research.
31
2.1.4.1 Le Bail fit
The Le Bail method is a whole powder pattern decomposition technique. It
is mostly used for determining the correct space group and lattice parameters of a
crystalline material. The Le Bail method was first put forward by Armel Lebail in
1988 [69]. The Le Bail method is also useful in extracting intensities from a powder
diffraction patterns, which then can be used to solve crystal structures. [70]. In this
research, Le Bail fits are often used for the analysis of in-situ diffraction patterns
to monitor the unit cell parameter evolution as a function of time. The Lebail
method is different from the Rietveld method as it uses fewer parameters, and also
the structure of the material of interest need not be completely known.
2.1.4.2 Rietveld refinement
Rietveld refinement is a full profile quantitative analysis used for powder
diffraction data. It uses a least-squared minimization algorithm, where a calculated
pattern is matched with an experimental diffraction pattern. Then changes to the
calculated pattern are instilled until a global minimum is acquired for the difference
curve. This method was first proposed by Hugo Rietveld in 1967. [71] And since
then, there have been many advances to this method with improved algorithms. The
Rietveld method is different from a Lebail fit mainly due to the fact that it looks at
the content of the unit cell, not just the unit cell edges and angles. This additional
step gives rise to a set of new parameters. In the Rietveld method, the analysis of a
powder diffraction pattern can be segmented into three main categories, 1) The peak
32
Powder pattern Instrument and Sample and material contri-
feature experimental bution
contribution
Peak position Zero error, axial Lattice parameters, sample
divergence of the displacement
beam
Peak shape Optics, radia- Crystallite size, stress,
tion purity strain
Peak intensity The scan time, Preferred orienta-
Radiation tion(sample) Structure
(Lorentz, polar- parameters such as atomic
ization) displacement parame-
ters(temperature effects),
site occupancy,
Table 2.2: Parameters affecting a powder diffraction pattern
positions 2) Peak shape and 3) Peak intensity. Table 2.2 summarizes the parameters
that govern these three segments in a diffraction pattern. All of these parameters
should be refined in the calculated pattern to obtain the optimum fit to the observed
pattern. Mccusker et all extensively discusses some general guidelines that should
be followed when doing a Rietveld refinement for laboratory/ synchrotron X-rays
and constant wavelength/TOF neutron diffraction experiments. [2]
The book by Zavalij et al broadly describes the features and the analysis
of a powder diffraction pattern. [72] The following paragraph briefly summarizes
some vital features and information that can be extracted from a powder diffraction
pattern from the Rietveld method.
The Bragg?s law illustrates the relationship between the interplanar distance
(d) in a crystal structure and the 2? angle in a diffraction pattern. The interplanar
distance is a function of lattice parameters. Fig. 2.5 shows the interplanar distance
and lattice parameter relationship for the 7 crystal systems, where a, b, c are the
33
unit cell edges, ?, ?, ? are the unit cell angles and h, k, l are the miller indices that
define a crystal plane.
Intensity of a powder diffraction pattern mainly comes from the electron den-
sities of the atoms in the lattice, which are described by the structure factors in the
unit cell. Eqn. 2.1 is the structure factor where gj is the population (or occupa-
tion) factor of the jth atom and tj is the Debye-Waller factor, which is a measure
of atomic displacement of atom j due to temperature. Eqn.2.2 gives integrated
intensity combining all of the structural factors.
F hkl = ?g
jtj(s)fgjexp[2?i(hxj + kyj + lzj)] (2.1)
Ihkl = KXphklXL?XP ?XA?T hklXEhklX | F hkl | 2 (2.2)
- K = scale factor
- phkl = Multiplicity factor
- L? = Lorentz multiplier (geometry of diffraction)
- P? = Polarization factor
- A? = Absorption multiplier
- Thkl = Preferred orientation factor
- Ehkl = Extinction multiplier ( In powders nearly always neglected.)
- Fhkl = Structure factor
When it comes to peak shape, mainly two types of functions are defined. It is
34
 
Figure 2.5: The mathematical relationship between d-spacing and lattice parameters
for the 7 lattice systems
35
generally accepted that crystallite size broadening is given by a Lorentzian function,
while microstrain broadening, is better explained with a Gaussian function. [73?75]
The Voigt function, which is a convolution of both Gaussian and Lorentzian func-
tions, is often used to describe the peak shape as most samples contain effects coming
from both the size and the strain. In practice, the pseudo Voigt function, which is
a linear combination of Gaussian and Lorentzian functions, has been introduced to
describe the peak shape of X-ray diffractometers. It approximates the convolution
integral. [76]
Apart from accounting for all of these parameters, the Rietveld refinement
also allows for quantitative phase analysis. In this research work, the Rietveld re-
finements are often used for the basic characterization of the pristine material and
to confirm the phase purity. In a sample with multiple phases, the weight frac-
tion of each phase is determined using the scale factors resulting from the Rietveld
refinement. [77]
2.1.4.3 Determining the quality of a refinement and other consider-
ations
There is a number of metrics that are used in diffraction techniques to assess
the quality of a refinement. B. Toby discusses these different metrics and how these
should be properly utilized. [78] The most commonly used metric is Rwp, which is
depicted in eqn. 2.3. [79] where yc,i is the calculated intensity, yo,i is the observed
intensity,wi is the weighting factor. A weighting factor is used to avoid the overpow-
36
ering of the refinement from intense peaks. An optimum value for Rwp cannot be
deliberately stated as it depends on factors such as X-ray source, flux, instrumental
optics, signal to noise ratio etc. While the objective is to take steps to minimize the
Rwp value during a refinement, a low Rwp value does not necessarily mean it?s a
good fit. For instance, diffraction patterns from a conventional lab diffractometer
give a lower Rwp due to the larger background typically present. While diffraction
patterns from synchrotron sources relatively give higher Rwp values due to their
high signal to noise ratio.
The best way to assess the quality of a refinement is to evaluate the errors
calculated for each parameter refined, and also visually examine the fit and the
difference curve. [78] Fig. 2.6 depicts several instances of getting bad peak fits in a
Rietveld refinement discussed by Mccusker et al. [2] It?s also important to instigate
small changes at a time in a step wise manner to avoid strong correlation between
parameters( ex: Occupancy and atomic displacement values). Otherwise it can lead
to an unstable refinement and get trapped in a local minimum . It?s advisable to
check the correlation matrix often during the refinement, which is present in many
software packages (TOPAS technical manual).
R 2wp = ?iwi(yc,i ? y )2o,i /? 2iwi(yo,i) (2.3)
37
Fig. 4. The observed (circles), calculated (line) and difference 
(bottom) profiles for a peak calculated with too symmetric a 
a peak-shape function. The characteristic difference prorle has 
a `+/-' character and is most pronounced  for the `tails' of the peak.
a b a b
Fig. 2. The observed (circles), calculated (line) and difference Fig. 5. The observed (circles), calculated (line) and 
(bottom) profiles for (a) a good fit of a peak, (b) a calculated difference (bottom) profles for a peak calculated with 
intensity that is too high and (c) a calculated intensity that is 2?  (a) too large and (b) too small. The characteristic
 too low. The characteristic difference profile for an intensity  difference profiles for a 2? mismatch have a ?+/-'
is either positive or negative and concentrated at the centre or a ?-/+' character.
 of the peak.
a b
a b
Fig. 6. The observed (circles), calculated (line) and difference 
Fig. 3. The observed (circles), calculated (line) and (bottom) profiles for some combinations of incorrect profile 
difference (bottom) profiles for a peak calculated parameters (closer to a real refinement situation): (a) an 
with (a) too large an FWHM and (b) too small an FWHM that is too small combined with a peak asymmetry 
FWHM. The characteristic difference profiles for that is too small, and (b) an FWHM that is too small combined 
an FWHM  mismatch have ?-/+/-? or ?+/-/+' characte with an intensity that is too small.
Figure 2.6: Instances of bad peak fit during a Rietveld refinement [2]
38
2.1.5 Synchrotron X-Ray and Time-Of-Flight neutron diffraction
Traditional diffraction techniques used in labs with Cu K? source take several
minutes for a fairly good quality or longer for a high quality diffraction pattern.
In order to study the detailed crystal structure evolution during chemical looping
reactions, and to investigate the reaction mechanism, a faster diffraction data collec-
tion is required. The beamline facilities at the Argonne national laboratories(ANL)
and Oak Ridge national laboratories (ORNL) suffice these requirements. Through
out the research work presented here different beamlines were used for a number of
our experiments. The advanced photon source (APS) at the ANL is commissioned
by the department of energy and produces high-energy, high-flux (brilliance) syn-
chrotron X-ray beams for research in many scientific disciplines. The high energy
synchrotron X-rays are produced by a storage ring, which is shown in Fig. 2.7. Elec-
trons of 7 GeV of energy is injected in to this storage ring, which has a circumference
of 1104 m, and the electrons orbit through a circular path made of aluminum alloy
vacuum chambers. This electron beam is guided by 1000 electromagnets. There
are two kinds of magnet setups. Bending magnets (BM) is the first type where
the electron beam goes through a set of dipole bending magnet which prompts the
electron beam to bend slightly. This change in the direction of electrons produces
X-rays, which then is utilized in different experimental hutches.
We used 9 BM, 11 BM and 17 BM beamlines in our experiments, which are
all bending magnet hutches. The second type of magnetic set up is known as in-
sertion device (ID), which consist of a series of magnets with alternating magnetic
39
poles known as wigglers. When the beam of electrons travels through an insertion
device it prompts it to create a wave (wiggle) and that again produces a beam of
X-rays. Both of these set ups are shown in Fig. 2.7. Electrons for the storage ring
are produced by a hot cathode, which is heated to about 1100 ?C. The electrons are
accelerated to a value closer to the speed of light in a 450 MeV linear accelerator
called the Linac. Once the energy of the electrons reaches 7 GeV they are injected to
the storage ring as mentioned above. [80]. The Spallation neutron source (SNS) at
the ORNL produces Time-Of-Flight (TOF) neutrons. The production of neutrons
and its stages are shown in Fig. 2.8. SNS is known for high flux neutrons. These
high fluxes are generated by a nuclear reactor that uses an accelerated beam of
protons to knock off neutrons from a target metal in a process known as Spallation.
This produces pulses of neutrons at a time with a range of wavelengths known as
Time-Of-Flight (TOF) and these pulsed neutrons are capable of highly efficient, low
background measurements. TOF is an important feature as it allows simultaneous
access of low and high energy neutrons. In the research presented here, this fea-
ture was highly useful as it allowed the characterization of lighter atoms such as
oxygen with the heavier metal ions such as the lanthanides, which was in contrast
to X-rays. [81] Both SNS and APS provide facilities for fast data collection at high
temperature/pressure in a range of sample environments, which is ideal for studying
the structural evolution of materials under chemical looping conditions. Appendix
A presents some further information on conducting experiments in national labs and
writing user proposals to obtain beam time.
40
Figure 2.7: How synchrotron X-rays are produced at APS
41
1
1 2 3
4
5
2 3 4
Figure 2.8: 1) The ion source produce negatively charged protons, which are directed
into a linear accelerator. 2) The linear accelerator know as the Linac consists of
normal conducting and superconducting radio-frequency cavities that accelerate the
beam with magnetic focusing and steering. The proton beam is accelerated upto 1
GeV of energy. Next it goes through a diamond stripping foil where the electrons
are stripped off to produce protons 3) Then this proton beam is accumulated about
1200 rounds in a ring so that enough protons are accumulated to produce high
energy neutrons. The accumulated protons are then released producing a pulse less
than 1 millionth of a second to hit the liquid mercury target at a rate of 60 times
a second. 4) The target is a steel container that holds of 50 tons of moving liquid
mercury. When the high energy protons hit the mercury nucleus 20-30 neutrons
are released, which are also high in energy. Finally these neutrons are guided to a
moderator to decrease their energy so that it?s suitable for research. Water is used
as the moderator for room temperature neutrons while liquid Hydrogen are used for
colder neutrons (20 K) Source:https://neutrons.ornl.gov/content/how-sns-works
42
2.1.6 In-situ gas flow diffraction experiments
In situ diffraction experiments are of utmost importance when studying ki-
netics of a reaction. This allows to monitor the progressive structural changes that
occurs as a function of temperature and atmosphere. In this research work in situ
gas flow diffraction experiments were conducted emulating a chemical looping reac-
tor where the sample was heated under a dynamic flow of gas ( reducing/oxidizing)
while collecting diffraction patterns.These in situ diffraction experiments were used
to extract several important information about these reactions such as how reac-
tive are these oxides towards reduction/oxidation? what are the optimum cycling
temperatures? and how is the cycling stability like?
In situ synchrotron X-ray diffraction (SXRD) experiments were performed in
transmission mode on the 17-BM beamline at the Advanced Photon Source (APS)
at Argonne National Laboratory. 17BM is dedicated to rapid acquisition of powder
diffraction patterns with versatile set-up that can accommodate a wide range of
sample environments. A flow-cell/furnace sample holder is used to control sample
temperature and atmosphere. [82] Sample is loaded in to a quartz capillary of 1 mm
in diameter with a thermocouple placed inside. Two heating elements that provided
Ohmic heat are placed above and below the capillary. This assembly is illustrated
in Fig. 2.9. Usually diffraction patterns are collected every 10-20 seconds. The 2D
diffraction patterns are then integrated into 1D patterns using GSAS II (Fig. 2.11)
In situ TOF neutron diffraction experiments are carried out on the POWGEN
beamline at SNS in the Oak Ridge national laboratory. In situ neutron diffraction
43
Figure 2.9: The instrumental set up at 17 BM at the APS
44
Figure 2.10: In situ data collection on POWGEN at SNS. The quartz sample holder
is shown on the tight.The series of detector panel banks is shown in bottom picture.
The instrument layout is similar to a DebyeScherrer camera with detector panels
positioned at a wide range of scattering angles on either side of the incident beam.
45
Figure 2.11: In situ data collection on 17 BM at the APS with the 2D detector.
Initial 2D images are integrated into a 1D powder pattern with GSAS II
46
data helps us to capture data complementary to in situ X-ray diffraction data.
In this research work specially NPD was crucial as the facile insertion of oxygen
was monitored in the oxygen storage materials as a function of temperature and
atmosphere. The experimental setup used for gas flow experiments on the POWGEN
beamline is shown in Fig. 2.10
2.2 Pair Distribution Function analysis (PDF)
Radiation can interact with matter in many ways. Diffraction only describe
the interaction with the average periodic structure. Radiation can also interact with
defects in a structure and structural domains that are not periodic. Such interac-
tions cause diffuse, inelastic scattering and give rise to the background in a regular
diffraction pattern causing a loss of intensity in peaks to a certain extent. [83]When
closely analyzed, diffuse scattering can give information about the local structure
with short range ordering including disordered vacancies, ionic substitutions, diffu-
sion of atoms, interstitial defects and other structural disorders. [84]
The atomic pair distribution function (PDF) method has recently emerged as
a powerful tool for understanding and analyzing the local structure of materials. It
is a total scattering technique where both Bragg and diffuse scattering is taken into
consideration. PDF primarily provides information on the local atomic structure,
and it gives the probability of finding any two atoms at a given distance. [85] This is
in contrast to the Rietveld analysis of a regular diffraction pattern, which models the
average crystal structure. Total scattering takes every point of a diffraction pattern
47
k - k? = Q
k k?
? 2? Q
d
Figure 2.12: Depiction of the total scattering factor or momentum transfer
into account, including the background, which is usually ignored in a conventional
diffraction experiment. PDF is mainly used to study materials that lack long-
range lattice periodicity such as liquids, amorphous solids and nanomaterials. [86]
Their PDFs show broad features and a PDF from a crystalline material is more
complicated with lot of sharp features.
Before looking into PDF, one needs to understand what the total scattering
vector Q (also known as the momentum transfer) is. Debye put forward a way to
represent intensity of X-rays scattered by electrons either in a crystalline or an an
amorphous material that includes this total scattering vector, Q. [87] Diffraction of
X-rays from a crystalline plane can be depicted in a vector form as shown in Figure
2.10. The total scattering vector (Q) is the difference between the wave vector of the
incoming X-rays (k) and the wave vector of the scattered X-ray beam (k?). It has the
48
dimensionality of an inverse length and is perpendicular to the plane of diffraction.
In an elastic scattering such a Bragg diffraction k=k?. The relationship between
the total scattering vector and the wavelength of the interacting X-rays is given in
equation 2.1. The pair distribution function, g(r) (or G(r)-reduced pair distribution
function) is the Fourier transform of the total scattering structure function, S(Q)
(Fig. 2.13). S(Q) consist of both Bragg and diffuse intensity. The conversion of
S(Q) into PDF is done through a set of mathematical calculations. [3] The PDF
plot which gives the pair distribution function (g(r)) vs distance(r) shows a peak
whenever there an ion pair in the structure at that distance. When r reaches infinity,
g(r) reaches 1. The amplitude of these peaks contain certain structural information
such as atomic displacement due to thermal motion and fall as as r increases. [72]
Q = (4?sin?)/? (2.4)
The experimental set up designed for PDF data collection is much similar
to a typical diffraction experiment. The main difference being that the data for
a PDF analysis should be collected for a wide range of Q (momentum transfer)
and should be in high quality. The quality of the diffraction pattern is of utmost
important for an accurate PDF analysis. Most importantly it?s critical to have data
at high Q. Premature termination of the Q-range causes termination ripples after
the Fourier transformation, which will add to the noise of the pattern. [72] There
are several synchrotron X-ray and spallation neutron high energy sources in the US
that meet these requirements including 11-ID-B at APS and NOMAD at SNS. PDF
49
9 
8 
7 
6 
5 
4 
3 
2 
1 
0  
5 10 15 20 25 30 35 40 45 50  55 60
Q (?-1)
2.5 
2 
1.5 
1 
0.5 
0 
-0.5 
-1 
-1.5 
-2  
5  10 15 20 25 30 35 40 45 50 
R (?)
Figure 2.13: S(Q) is the total scattering structure function which is generated from
a diffraction experiment over a large Q range. Q is the momentum transfer. R is
the real space inter-ionic distance, G(r) is the reduced pair distribution function.
The pair distribution function (g(r)) is generated from G(r)
50
G(r) S(Q)
Figure 2.14: With the team at NOMAD, SNS
experiments for this research work were carried out at the NOMAD beamline in the
Oak Ridge national laboratory. Fig. 2.14 shows the research team at NOMAD who
assisted with all the experiments.
2.3 Thermogravimetric Analysis
In thermogravimetry the weight of the sample is recorded as a function of
time with or without the flow of a gas. In this research TGA was mainly used
for the purpose of determining the oxygen storage capacity of metal oxides. The
instrument is first calibrated using alumina standards for each ramp rate of use.
The instrument consist of two balances, one for the sample and the other for the
reference. Therefore by monitoring the weight change in both pans simultaneously
the extra changes happening in the sample can be attributed to the chemical and
physical reactions taking place in it. TGA is also useful in determining the thermal
decomposition temperature and the melting point of a material.
51
2.4 X-ray absorption spectroscopy (XAS)
X-ray absorption spectroscopy is a versatile technique that can be used to
characterize a wide array of materials from solids to liquids and from crystalline
to amorphous materials. In XAS, a core electron interacts with incoming X-rays
that have energy comparable to the absorption edge of the absorbing element. The
absorbed X-rays excite the core electron, resulting in its emission as a photoelec-
tron. X-ray absorption in the immediate vicinity of the absorption edge is typically
referred to as X-ray absorption near-edge structure (XANES). This accounts for
the region that is approximately 40 eV above the edge. [88] At energies above the
absorption edge, excited photoelectrons undergo constructive or destructive interfer-
ence with the electrons of neighboring atoms, which is described as extended X-ray
absorption fine structure (EXAFS). [89] EXAFS takes into account all the multiple
scattering paths of the photoelectron with its neighboring atoms and is calculated
through a function called the Green?s function. [90]
This physical phenomenon is illustrated in Fig. 2.15 and parts of an X-ray
absorption spectrum corresponding to each stage. XAS can be used to deduce
structural information about a material. XANES mainly provides characteristic
information of the absorbing element such as the valence and the species while
EXAFS provides information on the coordination environment and average bond
distances of the absorber. [91]. EXAFS is particularly useful in validating a crystal
structure by fitting the model to experimental data. In this research work XAS
experiments were done at the 9 BM beamline at APS. The experimental set up
52
Pre-Edge
XANES{ EdgeNear-Edge
EXAFS
?
Energy (keV)
Back scattered wave Constructive
from neighboring interference
electron potentials
Electron densities N1 A N2 EXAFS
of the neighboring 
atoms
Spherical out going 
wave of the photolectron 
from the absorber
e Photoelectron
XANES
X-ray (h?) =  EK.E+EB.E
e
Absorber (A) 1s (K-Edge)
Figure 2.15: The sequential steps in X-ray absorption that give rise to the X-ray
absorption spectrum.
53
used is shown in Fig. 2.17 XAS data analysis was done using the Athena and
Artemis software packages that belongs to the IFEFFIT library. [92] Figure 2.16
depicts briefly how the raw data for EXAFS is processed.
2.5 ICP-AES
Inductively Coupled Plasma Atomic Emission Spectroscopy is a quantitative
analytical method for elemental analysis. ICP-AES is used to quantify the elemental
composition in a sample. First the sample of interest is dissolved in ultra high pure
acid solution such as nitric acid. And then this solution is inserted into the sample
chamber in the instrument where the elements are ionized first. In the process of
ionization electrons are removed and then recombined shortly afterwards. During
recombination of ions and electrons characteristic radiation is emitted with a unique
wavelength for each element. The intensity of these emitted radiation is standardized
by a set of standards. ICP-AES can be used to quantify even trace elements. [93]
54
a
A?er background 
subtrac?on
and normaliza?on 
b
In K-space
c In R-space
Figure 2.16: a) First the raw XAS data is processed by normalizing. This gives the
absorption co-efficient(?) vs energy of the incident X-rays. Normalizing removes the
effect of sample thickness. The sharp rise in energy corresponds to the absorption
edge of the absorbing element. XANES analysis is done using this spectrum b)
Afterwards the data is converted into k-space or reciprocal space by subtracting
the background. c) In the final step data is converted to R-space or real space
through Fourier transform, which gives ? of R vs radial distance. Real distance
corresponds to actual bond distances in the structure. EXAFS analysis is done
using this spectrum
55
Incident X-ray (I0)
Gas inlet
Gas outlet
Sample holder
Tube furnace
Analyzer
Transmitted X-ray (It)
Figure 2.17: Experimental set up at 9 BM at APS
56
Figure 2.18: ICP-AES instrument
57
Chapter 3: Oxygen storage properties of AFe2O4 (A= Yb, Lu, Y, In)
3.1 Introduction
In addition to their unique optical, magnetic and electronic properties, [94]
oxides of transition metals in mixed valence states also display important functions
in energy storage and conversion. [4,95] In particular, their ability to reversibly take
up and release oxygen renders them relevant for key energy applications such as
oxygen storage, energy conversion and oxygen gas sensing. [96, 97] Mixed valence
metal oxides can act as oxygen storage materials (OSMs) releasing (or gaining) oxy-
gen under reducing (or oxidizing) environments, which is often correlated with their
oxygen non-stoichiometries. A challenge in this field is to design materials whereby
oxygen is released and incorporated topotactically and efficiently as the redox active
metal cycles between two oxidation states. Chemical looping combustion (CLC) and
chemical looping reforming (CLR) are two applications that utilize OSMs to ther-
mochemically deliver oxygen from air to fuels. These OSMs can either completely
oxidize a fuel in CLC (to generate CO2 + H2O) or partially oxidize a fuel in CLR (to
generate CO + H2). [22] OSMs eliminate the need to purify and isolate oxygen by
the costly process of cryogenic distillation. Furthermore, they enable easy product
separation and prevent product contami- nation by NOx gas formation. [21,22] Long
58
term stability, fast and reversible redox kinetics, low operating temperatures, and
high oxygen storage capacities are all metrics for designing new OSMs. [17,22] The
materials first studied as OSMs include the binary oxides of Mn, Fe, Co, Ni, and Cu
due to their natural abundance and variable oxidation states. [26] However, their
agglomeration from sintering under thermal working conditions has been a major
drawback. As alternatives to binary metal oxides, ternary metal oxides such as per-
ovskites and spinels have been researched as OSMs owing to their high structural
stability and oxygen storage capacities (OSCs). [37, 38, 98, 99] Two types of oxygen
species in these oxides take part in the oxidation process: (1) weakly bound lattice
oxygen that completely oxidizes the fuel at lower temperatures and (2) strongly
bound lattice oxygen that partially oxidizes the fuel (chemical reforming) at higher
temperatures. [38,98] Hence by fine tuning the type of ternary metal oxide and the
reaction conditions, a well-designed OSM could achieve specific product selectivity.
The layered oxide LuFe2O4 has drawn recent attention as an OSM and a
conductometric gas sensor owing to its 1 : 1 mixed valence states of Fe2+ and
Fe3+. [63, 100, 101] LuFe2O4 does not adopt the typical AB2O4 spinel or ferrite
aristotypes but instead a structure consisting of bilayers where edge-sharing BO5
trigonal bipyramids alternate with edge-sharing AO6 octahedra (Fig. 3.1). Typi-
cally, the A cation is trivalent such as a lanthanide (Ln3+) and B is a transition
metal in a mixed valence state. LuFe2O4 is the second member of the homologous
[AO2]m[BO]n series with m=1 and n=2, [102] and its unit cell is indexed in the
centrosymmetric space group R-3m [103](Fig. 3.1). However, a lowering of symme-
try to P1 or C 2/m has been reported due to the structural distortion arising from
59
Figure 3.1: (a) The crystal structure of AB2O4 and its unit cell in (b) R-3m and
(c) C 2/m space groups. While R-3m is standard, C 2mis used for LuFe2O4 due to
its charge order driven monoclinic distortion below 330 K.
60
charge ordering below 330K that breaks the 3-fold rotational symmetry. [104, 105]
Through electron diffraction studies, Hervieu et al. reported the monoclinic space
group C 2/m to describe the charge ordered phase and its relationship to the hexag-
onal unit cell (Fig. 3.1). [63] However, R-3m is the most commonly reported space
group for YbFe2O4 [106], YFe2O4 [107] and InFe2O4 [108]
The AB2O4-type oxides have been heavily researched for their physical prop-
erties such as multiferroicity, magnetism, and linear/non-linear electronic conduc-
tion. [109?115] However, oxygen storage properties have not been explored as much
in this system. Hervieu et al. found that LuFe2O4 can be successfully cycled be-
tween its reduced (Fe2.5+) and oxidized (Fe3+) states at 500 ?C under hydrogen and
dynamic vacuum. [63] Reversible oxygen intercalation was confirmed through X-ray
diffraction and thermogravimetric analysis. Despite the AFe2O4 family having the
same structure type, their physical properties vary depending on the A-site cation.
For instance, LuFe2O4 exhibits charge ordering below 330 K and ferrimagnetism at
240 K, [116] YFe2O4 displays two metal-insulator transitions at 240 K and 220 K
with antiferromagnetic spin ordering, [65] while InFe2O4 exhibits both charge and
magnetic ordering below room temperature. [66] This study is the first to explore
the impact of elemental substitution in the The AB2O4-system on two properties:
(1) kinetics of reduction/oxidation and (2) overall oxygen storage capacity.
We first report the effects of A-site substitution on the structure of the four
analogues, A = Lu, Yb, Y and In from room temperature to 700 ?C. Next, we re-
late their reactivity and suitability as OSMs as determined by cycling experiments
carried out at 600?C in oxidizing and reducing atmospheres emulating both CLC
61
and CLR conditions. The rapid data collection of in situ synchrotron X-ray pow-
der diffraction (sXPD) allows us to monitor the kinetics of the redox reactions by
analyzing unit cell volume and lattice parameter changes with time. Using thermo-
gravimetric analysis (TGA) experiments we further investigate their oxygen storage
capacities (OSCs) and suitability as OSMs in related oxygen-sensitive applications.
To the best of our knowledge, the In and Y analogues have never been studied for
their reversible oxygen intake/uptake capabilities in O2/H2 atmospheres with in situ
diffraction. A comprehensive comparison of A-site substitution effects on the oxi-
dation kinetics and oxygen storage capability of AFe2O4 materials has never been
pursued either.
3.2 Experimental
3.2.1 Materials synthesis
All samples were prepared using solid-state reactions. 99.9% (Alfa Aesar)
A2O3 (A = Yb, Lu, In, Y), 99% (Sigma Aldrich) Fe2O3 and 99.5% (Sigma Aldrich)
Fe powder were ground in an agate mortar and pestle in 1/2 : 5/6 : 1/3 ratio
respectively to prepare 0.25 g of the target compound, and pressed into pellets of
13 mm diameter. The pellets were placed in a 2 mL alumina crucible and sealed
inside evacuated 8 mm diameter quartz ampoules (Fig. 3.2). Before sealing, the
ampoules were flushed with N2 gas and evacuated several times to obtain the re-
quired oxygen partial pressure (PO2). [117] These compounds are highly sensitive
to PO2, and A2Fe3O7 often forms as an impurity if PO2 is not below approximately
62
Figure 3.2: Sealing station
10-7 Torr. The samples were sintered at 1180 ?C (850 ?C for In) for 12 hours and
then quenched to room temperature. To prepare the oxidized phases, AFe2O4+d (ex
situ), the as-synthesized samples were oxidized at 500 ?C under air and cooled to
room temperature.
3.2.2 Ex situ high-resolution synchrotron X-ray powder diffraction
(sXPD)
Ex situ experiments with high-resolution synchrotron X-ray powder diffraction
(sXPD) were performed on the 11 BM beamline at the Advance Photon Source
(APS) at Argonne National Laboratory. An average wavelength of 0.41458 A? was
used, covering a Q-range of approximately 0.7 A?-1 to 10 A?-1. Powder patterns were
collected at ambient temperature and pressure for both the pristine compounds
(AFe2O4) and their oxidized phases (AFe2O4+d) prepared ex situ. The powder
63
patterns were analysed by Rietveld refinements for the pristine phases and Le Bail
fits for the oxidized phases using TOPAS 5. [118]
3.2.3 In situ sXPD
In situ sXPD experiments were performed in transmission geometry on the 17
BM beamline at the APS. A 2D PerkinElmer Si flat panel detector was used with an
average wavelength of 0.45336 A?. We collected diffraction patterns every 0.33 seconds
and integrated the diffraction images with GSAS-II [119] to yield patterns with a
Q-range of approximately 0.30 A?-1 to 6.6A?-1. A constant gas flow was maintained
at a rate of 15 mL/min through a quartz capillary sample holder which was loaded
with the sample, and a residual gas analyzer (RGA) was used to monitor the gaseous
products. Two types of experiments were carried out for all the A-site analogues.
First, we ramped the temperature of the sample environment up to 700 ?C under 20%
O2/He (air), holding at every 100 ?C for 20 minutes. We repeated this temperature
profile in pure helium to differentiate between temperature and oxidation driven
changes. Second, we performed cycling experiments to emulate a chemical looping
reactor, by switching the atmosphere between 20% O2/He and 3.5% H2/He. The
samples were heated to 600 ?C in He and held at 600 ?C while cycling between the
two atmospheres. Automated sequential refinements [120] were carried out using
TOPAS 5 to study the lattice volume evolution with time.
64
3.2.4 Thermogravimetric analysis (TGA)
TGA was conducted using an SDT Q600 equipped with a TA Discovery
MKS104-S0212004 Micron Vision 2 Mass Spectrometer. The instrumentwas cali-
brated prior to the experiment using alumina standards for each ramp rate. Ramp
rates of 5 ?C/min, 10 ?C/min, and 25 ?C/min were used with a flow rate of 100mL/min.
The sampleswere heated to 700 ?C in air, holding at every 100 ?C for 10 minutes
to allow for maximum weight gain. An experiment was also done by heating the
samples up to 600 ?C in air at 10 ?C/min and holding the temperature for 3 hours
at 600 ?C to monitor the stability in the weight gained.
3.3 Results
3.3.1 Structures of AFe2O4 and AFe2O4+ d from high-resolution
sXPD
We successfully fitted a structural model [103] for pristine AFe2O4 phases (be-
fore oxidation) to the high-resolution sXPD pattern collected at room temperature
(Figure 4.4 and 3.4). A pseudo-Voigt function was used for the peak shape profile
and the intensities were corrected for Lorentz-polarization and absorption effects.
The refined structural parameters are listed in Table 1. All four A-site analogues are
isostructural with relative changes in intensities, predominantly for the (003), (006)
and (009) reflections, due to the difference in the A-site atomic scattering factors.
We observed the characteristic splitting of the (015) reflection in R-3m symmetry to
65
LuFe2O4 R-3m a = 3.4559(3) A? c = 25.108(1) A?
Atom Site x y z Occ Beq (A?2) U11 U33
Lu 3a 0 0 0 1 0.31(5) 0.0006(7) 0.015(1)
Fe 6c 0 0 0.2147(1) 1 0.32(9)
O1 6c 0 0 0.1448(12) 1 11.01(94)
O2 6c 0 0 0.2896(9) 1 2.56(47)
YbFe2O4 R-3m a = 3.45765(6) A? c = 25.2374(7) A?
Atom Site x y z Occ Beq (A?2) U11 U33
Yb 3a 0 0 0 1 0.83(3) 0.0019(3) 0.0321(8)
Fe 6c 0 0 0.2155(1) 0.875 0.41(4)
O1 6c 0 0 0.1280(2) 1 0
O2 6c 0 0 0.2933(2) 1 0
YFe2O4 R-3m a = 3.4561(1) A? c = 25.4959(6) A?
Atom Site x y z Occ Beq (A?2) U11 U33
Y 3a 0 0 0 1 0.85(3) 0.0053(3) 0.0332(7)
Fe 6c 0 0 0.2156(1) 0.625 0.26(3)
O1 6c 0 0 0.1276(2) 0 0
O2 6c 0 0 0.2924(2) 0 0
InFe2O4 R-3m a = 3.45728(3) A? c = 25.61599(4) A?
Atom Site x y z Occ Beq (A?2) U11 U33
In 3a 0 0 0 1 0.85(3) 0.0039(3) 0.0294(6)
Fe 6c 0 0 0.21568(5) 0.5 2.43(3)
O1 6c 0 0 0.1276(2) 0.84(1) 0
O2 6c 0 0 0.2924(2) 0.60(1) 0
Table 3.1: Structural parameters from combined Rietveld refinement done using
high resolution sXPD for AFe2O4 at 298 K (A=Lu,Yb, Y, In)
the (1-1-2) and (201) reflections in C2/m symmetry for the Yb and Lu compounds
but not the Y and In analogues as shown in Figure 3.5. However, since our focus is
to compare A-site substitution effects, R-3m (no. 166) was used as the space group
for all four compounds.
During the Rietveld analysis, we introduced anisotropic atomic displacement
parameters (ADPs) for the A-site, which we found to significantly improve the fit (for
A = Yb Rwp from 18.17% to 14.04%). Table 1 shows the refinement results when
anisotropic ADPs were used with the A cation occupying the 3a special position with
66
25 LuFe2O4 (RT,R-3m) Ycal  
Lu O  (Ia-3) 0.87 wt.% Yobs-Ycalc2 3
20 Rwp = 13.68 % Yobs
15
10
5
0
?5
YbFe2O4 (RT,R-3m)
Ycal  
FeO (Fm-3m) 1.41 wt. %        Yobs-Ycalc
Yb2O3 (Ia-3) 0.53 wt.% Yobs
Rwp = 14.04 %
1.32 2.64 3.96 5.28 6.60
Q (?-1)
Figure 3.3: Observed, calculated, and difference curves of the LuFe2O4 and
YbFe2O4 crystal structures at room temperature (RT) with high-resolution syn-
chrotron X-ray powder diffraction patterns. A wavelength of 0.41458 A?A was used.
67
Intensity (Arbitrary Units)
25 YFe2O4 (RT,R-3m) Ycal  
FeO (Fm-3m) 0.56 wt.% Yobs-Ycalc
Rwp = 13.92 % Yobs
20
15
10
5
0
?5
InFe2O4 (RT,R-3m) Ycal  
FeO (Fm-3m) 3.46 wt.% Yobs-Ycalc
Rwp = 11.53 % Yobs
1.32 2.64 3.96 5.28 6.60
Q (?-1)
Figure 3.4: Observed, calculated, and difference curves of the InFe2O4 and YFe2O4
crystal structures at room temperature (RT) with high-resolution synchrotron X-ray
powder diffraction patterns. A wavelength of 0.41458 A?A was used.
68
Intensity (Arbitrary Units)
5000
YFe2O4
4000
3000
LuFe2O4
YbFe2O4
InFe2O4
2000
1000
9.10 9.20 9.30 9.40
2?(Degrees,?= 0.414580 ?)
Figure 3.5: Monoclinic distortion observed in the high resolution sXPD patterns of
Lu and Yb analogues as evidenced by the splitting of the (0 1 5) reflection of R-3m
symmetry into the (11?2?) and (2 0 1) reflections of C2/m symmetry.
69
Relative Intensity(Arbitrary Units)
100% occupancy. However, we could similarly improve the fit when we employed
isotropic ADPs for the A-site while it occupied the 6c site (0, 0, z) at 50% occupancy.
The site occupancy of A = In slightly deviated from unity, which likely occurred due
to the high volatility of In. We therefore sintered InFe2O4 at a lower temperature
than the rest. We performed Le Bail fits for the high-resolution sXPD patterns of
the maximally oxidized phase AFe2O4+d prepared ex situ (Fig. 3.6 and 3.7). P- 3
was used for A =Yb, Lu and Y [121] and C2/m was used for A = In [122] as the
space group, and the refined lattice parameters are shown as an inset.
3.3.2 Structural evolution with temperature: in situ sXPD
With in situ sXPD, we explored the thermal stability, optimal operating tem-
peratures and associated structural phase transitions of the title compounds. Fig.
3.8 and 3.9 shows the contour plots of the in situ sXPD patterns of AFe2O4 as
they are heated under air up to 700 ?C. Below 200 ?C we did not observe any new
reflections, and the room temperature phases of AFe2O4, which we label R1, were
retained. Upon further heating, extra satellite reflections appeared around 230 ?C
for all compounds except In, which indicated the formation of the supercell struc-
ture (R2 phase). These supercell reflections gradually disappear around 350 ?C for
Y and 480 ?C for Lu and Yb. In the 250?450 ?C region, the diffraction peaks are not
well defined, and the patterns display a large amount of phase co-existence, most
apparently for Y. Upon further heating, a major structural phase transition occurs
and the maximally oxidized phase is reached, which we label the R3 phase. This
70
3
30 10
LuFe2O4+d (RT)
25 P-3
a = 3.47140(2) ?
20 c = 8.44756(8) ?
Rwp = 11.52 %
15
10
5
0
?5
30
YbFe2O4+d (RT)
25 P-3
a = 3.4871(1) ?
20 c = 8.4295(1) ?
Rwp = 11.75 %
15
10
5
0
?5
5 10 15 20 25 30
Q
Figure 3.6: Le Bail fit for the oxidized phases of LuFe2O4 and YbFe2O4 crystal
structures at room temperature (RT) with high-resolution synchrotron X-ray powder
diffraction patterns. A wavelength of 0.41458 A? was used.
71
Intensity (Arbitrary Units)
103
25 YFe2O4+d (RT)
P-3
20 a = 3.5386(1) ?
c = 8.3635(1)  ?
15 Rwp = 12.99 %
10
5
0
?5
25 InFe2O4+d (RT)
20 C2/ma = 5.8562(1) ?
b = 3.3838(1) ?
15 c = 10.4133(3) ?
? = 124.191(2)
Rwp = 11.39 %
10
5
0
?5
5 10 15 20 25 30
Q
Figure 3.7: Le Bail fit for the oxidized phases of InFe2O4 and YFe2O4 crystal struc-
tures at room temperature (RT) with high-resolution synchrotron X-ray powder
diffraction patterns. A wavelength of 0.41458 A? was used.
72
Intensity (Arbitrary Units)
transition is complete by 430 ?C for Y, 500 ?C for Lu and Yb, and 600 ?C for In
(Fig. 3.8 and 3.9). Evidently, the A-site substi- tution has an impact on the ox-
idation temperatures, controlling the window of stability for the high-temperature
R3 phase. All four compounds were stable until 700 ?C. The in situ diffraction
experiments performed in air were repeated in He to isolate temperature driven
changes from those driven by oxidation. We observed different trends in the lattice
parameter evolution under the two atmospheres.
3.3.3 Cycling experiments with in situ sXPD
The optimal operating temperature for OSMs in chemical looping reactions
is a trade-off between fast redox kinetics of oxygen release/uptake and prolonged
structural stability. To achieve fast reaction rates while preserving the structural
stability, we chose 600 ?C as the cycling temperature for the in situ sXPD experi-
ment. To this end, cycling experiments were performed under oxidizing (20%O2/He)
and reducing (3.5%H2/ He) atmospheres. Fig. 3.17 shows the structural evolutions
observed in this experiment as a contour plot of some major reflections. The cycling
stability and reactivity were almost identical for the Lu and Yb analogues. The In
analogue, however, started to degrade once it was exposed to H2(Fig. ??
3.3.4 Thermogravimetric analysis (TGA)
To correlate weight changes upon oxidation to crystallographic parameters,
we performed thermogravimetric analysis (TGA) experiments to mimic the in situ
73
a 700
? 640600 R3 002
500 480
400 R3+R2 320
300
R2 ? * *
200 024
160
R1 ?
100 006
b 0700
R3 ?
960
600 000022
500 720
400 R3+R2 480
300 R2 ?00264* *
200 024
240
R1 ?
100 00066 0
 0.9  1.4  1.9 2.4 2.9
Q (?)-1
Figure 3.8: Contour plots of the sXPD data as AFe2O4 is heated in air (20%O2
in He) from 100 ?C to 700 ?C for Lu (upper) and Yb (lower). The yellow labels
and blue dashed lines indicate phase co-existence. The white arrows and asterisks
mark some of the reflections: (0 0 6) belongs to the R1 phase (AFe2O4) in R-3m
symmetry, (0 2 4) is one of the supercell reflections in the R2 phase that develops
above 230 ?C, and (0 0 2) belongs to the R3 oxidized phase (AFe2O44+d)
74
Temperature (?C)
a 700
?
600 R3 002
1600
500 1200
400 800
300
R2 ? ** 024 400
200
R1 ?
100 006 0
b 700
R3 ?
?
0000
4280
600 2
2
500 R3+ R1 3210
400 2140
300 ?006
024 1070200 R1 ?? 6
100 006 0
 0.9  1.4  1.9 2.4 2.9
Q (?)-1
Figure 3.9: Contour plots of the sXPD data as AFe2O4 is heated in air (20%O2 in
He) from 100 ?C to 700 ?C for Y (upper) and In (lower) . The yellow labels and blue
dashed lines indicate phase co-existence. The white arrows and asterisks mark some
of the reflections: (0 0 6) belongs to the R1 phase (AFe2O4) in R-3m symmetry, (0
2 4) is one of the supercell reflections in the R2 phase that develops above 230 ?C,
and (0 0 2) belongs to the R3 oxidized phase (AFe2O4+d). As illustrated the R2
supercell structure was not observed for In
75
Temperature (?C)
2000
1800 LuFe2O4
YbFe2O4
1600 YFe2O4
1400
1200
1000
? *         
800 024 *         
600 ?024 *         *         
400
?
200 024 *         *         
0
1.6 1.8 2
Q (?-1)
Figure 3.10: The supercell satellite reflections in the isolated powder patterns of
the R2 supercell structure at the end of the 20 min hold at 300 ?C for Lu, Yb and
Y analogues.
76
Intensity
Figure 3.11: Contour plots of synchrotron X-ray diffraction patterns of the AFe2O4
series during cycling between oxidizing (20 % O2 in He) and reducing (3.5 % H2 in
He) atmospheres at 600 ?C. (a) LuFe2O4,(b) YbFe2O44,(c) YFe2O4 and (d) InFe2O4.
Three consecutive cycles were carried outfor Lu andYband twocycles forYandIn. The
Lu,Yband Y analogues showstable cyclability. The degradation of InFe22O4 under
H2 is clearly evidenced by the diminishing intensities of the Bragg reflection
77
sXPD. Fig. 3.12 and Fig. 3.13 display the TGA curves where samples were heated
at 10 ?C/min equilibrating every 100 ?C. The R2 supercell structure observed to
form above 200 ?C from the sXPD data corresponds to a weight gain of 0.80-1.0%
(for Lu, Yb, Y) from TGA data. The total weight gain from the R1 phase to the
R3 phase varies from 2.2 % to 2.6 % among the four oxides. Calculations of OSC
values based on TGA data are presented in section 4.4. The derivative of the weight
change is also plotted in Fig. 3.12and 3.13 (black curve), and the highest weight
gain due to oxidation occurs between the 300 ?C - 400 ?C region for all samples.
This rapid weight gain, however, is most pronounced for the Y analogue, indicating
that this analogue displays the fastest kinetics for oxide insertion. The derivative
curve also demonstrates that the In analogue is the most sluggish of the series, which
corroborates our in situ sXPD data. This is further confirmed in TGA experiments
done with different ramp rates.
3.4 Discussion
3.4.1 A-site substitution effects on the structure
In the AFe2O4 series, the transition metal is the redox active centre while the
A-site cation does not directly participate in the reaction. While this is the case
in many OSMs consisting of ternary metal oxides, the A-site cation has shown to
play key roles in determining the stability of the compound, the concentration of
oxygen vacancies, and the oxidation state of the transition metal. [113] Shen et al.
compared A-site substitution effects of several AFeO3 perovskites and found that
78
Weight change derivative Time
104 200
3
4 700 ?C
2.5
600 ?C 150
3
500 ?C 2
2 400 ?C 1.5
100
300 ?C 1
1 200 ?C 50
0.5 LuFe O
100 ?C 2 4
0 0 0
104 2003
4 700 ?C
2.5
600 ?C 150
3 2
500 ?C
1.5 100
2 400 ?C
300 ?C 1
1 200 ?C 500.5 YFe2O4
100 ?C 
0 0 0
8.5 9 9.5 10 0 100 200 300 400 500 600 700
2 ?  (?, ? = 0.45336 ?) Temperature (?C)
Figure 3.12: Thermogravimetric analysis (TGA) depicting the weight change as a
function of time and temperature for LuFe2O4 (Upper) and YFe2O4 (Lower). The
corresponding synchrotron powder diffraction patterns are shown on the left of each
TGA.
79
Intensity (Arbitrary units) Intensity (Arbitrary units)
Weight change(%) Weight change(%)
Time (min) Time (min)
Weight change derivative Time
10 4 200
3
4 700 ?C
2.5 150
600 ?C
3 2
500 ?C
400 ?C 1.5
100
2
300 ?C 1
1 50
200 ?C 0.5 YbFe2O4
0 100 ?C 0 0
104 2003
4 700 ?C
2.5
600 ?C 150
3 500 ?C 2
400 ?C 1.5 100
2
300 ?C 1
1 200 ?C 50
0.5 InFe2O4
100 ?C 
0 0 0
8.5 9 9.5 10 0 100 200 300 400 500 600 700
2?  (  ?, ?= 0.45336 ?) Temperature (?C)
Figure 3.13: Thermogravimetric analysis (TGA) depicting the weight change as a
function of time and temperature for YbFe2O4 (Upper) and InFe2O4 (Lower). The
corresponding synchrotron powder diffraction patterns are shown on the left of each
TGA.
80
Intensity (Arbitrary units) Intensity (Arbitrary units)
Weight change(%) Weight change(%)
Time (min) Time (min)
the oxygen desorption capability varies with the A cation?s radius. [123] In this
study Ba and Sr were used as the A-site cations, and the Ba analogues showed
higher oxygen desorption amounts due to the larger lattice volumes. In a similar
study of three hexagonal AMnO3 perovskites, Abughayada et al. demonstrated that
the oxygen storage capacity is highly sensitive to the A cation?s radius, [124] where
the adsorption/desorption of the oxygen content increased with larger A cations.
Hence, we anticipated compa- rable results for the AFe2O4 series. Based on our
Rietveld refinements, both structural and cell parameters are affected by the nature
of the A-site in the title compounds. The a-lattice parameters follow the order Y3+
?Yb3+ ?Lu3+ ?In3+, while the c-lattice parameters behave conversely. The ionic
radii of the A-site cation clearly determine these trends (Fig. 3.14).
Rietveld refinements of the R1 phase also revealed significant A-site disorder
from the special position 3a. This characteristic disorder has been reported previ-
ously for LuFe2O4.13,18 Infrared spectroscopy has also revealed a splitting of the
Lu?O vibrational modes, [125] which arises from the interactions between the polar
FeO5 bilayers and the LuO6 octahedral layers. The polarity in FeO5 bilayers is at-
tributed to charge ordering of the Fe3+ (electron-deficient) and Fe2+ (electron-rich)
cations. In our study, we observed this anisotropic disorder along the z axis for the
A cations in all four compounds. Upon heating in air, we observed the supercell
structure of the R2 phase around 250 ?C (Fig. 3.8 and 3.9). Hervieu et al. found
the same supercell reflections by X-ray and electron diffraction, indexed them in the
rhombohedral R-3m space group,citeHervieu2013 and attributed their appearance
to oxygen insertion in a manner commensurate with the metal sublattice. The re-
81
Figure 3.14: (a) The linear dependence of a- and c-lattice parameters on the radii
of the A cations. The size of the spheres is to scale. (b) Displacement of Fe from
the center of the trigonal bipyramid towards the AO6 octahedral layer. As a result,
there are three distinctive Fe?O bond lengths which are denoted as Q, R and S.
82
A- BL Fe-O(Q) Fe-O(R) Fe-O(S) Bond va- Shannon
site (A?) lence sum radii [128]
cation BV for Fe of A site
(v.u) cation (A?)
In3+ BL 1.9521 (2) 2.294 (2) 1.991 (2) 2.392 (3) 1.06
BV 0.5515 (3) 0.237 (1) 0.501 (2)
Lu3+ BL 2.0079 (5) 2.155 (4) 1.936 (4) 2.349 (7) 1.117
BV 0.4803 (6) 0.334 (3) 0.574 (6)
Yb3+ BL 2.0164 (5) 2.155 (4) 1.919 (4) 2.343 (7) 1.125
BV 0.4703 (6) 0.333 (3) 0.599 (6)
Y3+ BL 2.0392 (2) 2.153 (2) 1.985 (2) 2.177 (3) 1.159
BV 0.4445 (2) 0.335 (2) 0.508 (3)
Table 3.2: Comparison of bond lengths, unit cell volume and A3+ cationic radii of
the A-site analogues at RT. Q, R and S are labels for the Fe?O bond lengths as
shown in Fig. 3.14.
fined powder pattern of the R2 phase for LuFe2O4, isolated at 300 ?C, is shown in
Fig. 3.15 along with the refined unit cell parameters. Interestingly, we observed the
formation of the R2 phase for the Lu, Yb and Y analogues but not for the case of
A = In.
3.4.2 Bond valence sum (BVS) calculations
To extend our understanding of the A-site substitution effects on the structure
of the title compounds, we carried out bond valence (BV) calculations using bond
lengths from the X- ray diffraction results. BVs were calculated using the formula
of Brown and Altermatt, [126] as shown in Eqn. 3.1and the empirical values R0 and
B were adopted from Gagn?e et al. (R0=1.712 B=0.4035 ). [127]
S = exp(Ro?R/B) (3.1)
Bond lengths and BVs of the Fe?O bonds are tabulated alongside the A-site
83
ionic radii in Table 2. In the FeO5 trigonal bipyramidal polyhedron, the Fe atoms are
off-centre and shifted towards the AO6 octahedral layers to create two unequal Fe?O
axial bond lengths (Fig. 6b). While the axial Fe?O bonds showed no significant
trend with the choice of the A-site cation, the equatorial Fe?O bond lengths showed a
direct correlation with the A cationic radii. This is also reflected in the overall bond
valence sum (BVS) of Fe?O bonds. The BVS of Fe?O in the Y and In analogues
depicts the highest and lowest deviation from the average valence state of 2.5 in Fe.
Hence, in the Y analogues, iron is the most under-bonded, and in the case of In, it
is the most over-bonded before the oxidation step. The mixed valence state (+2.5)
of the R1 phase has been observed by Fe-2p, O-1s X-ray absorption spectroscopy
(XAS) and Mossbauer spectroscopy, [63,116] and other studies have confirmed that
when forming the R3 phase, the mixed valence state goes fully to Fe3+.38,46 One
noticeable difference across our series is the temperature at which oxidation to the
R3 phase is completed. As mentioned in Section 3.2 (Fig. 3), Y is the first to fully
transform into the R3 phase (430 ?C) while In is the last (600 ?C). This difference
in reactivity towards oxidation clearly reflects the difference in the deviation in the
BVS of Fe from the mixed valence (+2.5) state. In the Y analogue, Fe is found in
the most under-bonded environment, and hence oxidizes the fastest. The reverse is
found for the In analogue, where the sluggish kinetics of oxygen insertion is likely
due to Fe being over- bonded.
84
3.4.3 Lattice expansion from oxygen insertion vs. thermal expansion
Upon heating the samples in He, we did not observe any of the above discussed
phase transitions, confirming that they are completely driven by oxygen insertion.
To further isolate the thermal expansion effects from oxidation ones, the evolution
of refined unit cell parameters with time was analyzed in air versus He. Fig. 3.18
depicts the evolution of cell parameters for Lu and Yb in both atmospheres.
In air, anisotropic cell volume changes were observed as opposed to the isotropic
thermal expansion under He. The a-lattice parameter gradually increased as a func-
tion of temperature while the c-lattice parameter decreased. A similar temperature
dependence for lattice parameterswas reported previously for the Yb and Tm ana-
logues in the temperature range of 50?400 K.47 In our samples, the overall cell
volume change was almost equal in both environments as the a-parameter expan-
sion rate in air was higher than that in He, compensating for the decrease in the
c-parameter in air (Fig. 3.16). The increase in the a-parameter can be attributed
to the bilayer shift mechanism in the ab-plane due to O2 insertion, which has been
evidenced from electron diffraction experiments.14 The decrease in the c-parameter
can be attributed to the decreasing ionic radii of iron upon oxidation. Approxi-
mately 35?44% of Fe-sites are already oxidized at 300 ?C (in air) in the Lu and Yb
compounds based on the ionic radii change around (0.1A?) and c-lattice parameter
change (0.035?0.044A?).
85
Figure 3.15: Isolated in situ sXPD patterns of LuFe2O4, collected while ramping
under air from RT to 700 ?C shown with the refined unit cell parameters. (a) The
structure of LuFe2O4 at RT. (b) The supercell structure at 300 ?C. The satellite
reflections marked with red arrows are due to the O2 driven modulations. (c) The
maximally oxidized structure of LuFe2O4+d at 600 ?C.
86
262
YbFe 2O4-He 300
261.5 YbFe 2O4-Air
Temp. 250
261
200
260.5
150
260
259.5 100LuFe 2O4-He
259 LuFe 2O4-Air 50
Error Bar
258.5 0
0 20 40 60 80 100
Time (min)
Figure 3.16: Cell volume evolution under Air Vs He
87
Unit Cell Volume (?3) 
Temperature (?C)
3.4.4 Thermochemical cycling capability and stability
The only study14 to test the oxygen storage capacity and cycling stability of
LuFe2O4 utilized 500 ?C as the cycling temperature where the reaction kinetics were
slow. Hence, we chose 600 ?C as the cycling temperature, which in fact increased
the reaction rate, noticeably for the reduction step. Fig. 9 depicts the refined cell
volume change with time for the Lu and Yb analogues under air and H2. The
rate of cell volume change in air is relatively faster than that in H2 illustrating the
faster reaction kinetics for oxidation. This is a clear indication of the difference
in the thermodynamic stability of the two phases. We also found the rate of cell
volume change for both compounds to be nearly the same. This is promising since
Yb is less expensive than Lu, and the Yb analogue could therefore be proposed as
a cost-effective alternative to Lu. While YFe2O4 appeared stable towards cycling
(Fig. 4c), it was rather difficult to calculate the accurate cell volume change due to
highly disordered oxidized phases. Under air, we observed that these layered oxides
were quite stable but started to decompose above 700 ?C. This observation agrees
with the report by Hervieu et al.14 However, the stability under H2 at 600 ?C was
not the same for all the samples. The In analogue started to degrade once it was
exposed to the reducing atmosphere. Although the A-site cation typically does not
take part in the redox reaction in many OSMs, In3+ reacts with H2 to destabilize
the structure and therefore leads to phase degradation (3.17).
88
a InFe2O4  94.58%
FeO  3.38%
? Yobs InxFe3 - x O4  2.04%
Ycalc Rw p  7.79%
Ydiff
b InFe2O4  91.98%
FeO  4.66%
InxFe3 - x O4  1.21%
In2O3  2.14%
Rw p  8.21%
  
c InFe2O4  86.05%
FeO  6.38%
InxFe3 - x O4  3.23%
In2O3  4.35%
Rw p  13.3%
5 10 15 20 25
2? ( Degrees, ? = 0.45367 ?)
Figure 3.17: Rietveld refinement done for the sXPD patterns of InFe2O4 at a) RT
b) End of 1st cycle c) end of 2nd cycle at 600 ?C. With each cycle it can be clearly
seen that InFe2O4 starts degrading into multiple phases
89
Relative Intensity(Arbitrary Units)
YbFe2O4 - He LuFe2O4 - He
YbFe2O4 - Air LuFe2O4 - Air
Temp. Error Bar
a 300
3.47
250
3.46 200
150
3.45
100
3.44 50
25.3 0
b 300
25.25 250
200
25.2
150
25.15
100
25.1 50
25.05 0
0 20 40 60 80 100
Time (min)
Figure 3.18: Sequential refinements data for Lu and Yb. While thermal expansion
is observed under He, an anisotropic lattice parameter change is observed under air
due to oxidation driven effects. The error bars are shown and are smaller than the
marker.
90
  c Parameter (?) a Parameter (?) 
Temperature (?C) Temperature (?C)
Cycle 1 Cycle 2 Cycle 3
90
89
P3?
LuFe2O4
YbFe2O4
88
Air Error
0 5 10 38 43 48 75 80 85
267
H2
266
R3?m
265
264
263
26210 20 30 50 60 70 85 95 105
Time (minutes)
Figure 3.19: Comparison of unit cell volume change of the Lu and Yb analogues for 3
consecutive cycles. Refined unit cell volumes from sequential refinements are plotted
with the error bar. Both compounds exhibit a similar stability and reversibility
throughout the three cycles. Error bars are smaller than the markers
91
 Unit Cell Volume (??)
3.4.5 Comparison of oxygen storage capacity
The maximum oxygen non-stoichiometry (d) that has been reported in this
series is 0.5 for LuFe2O4+d. [63] Fig. 3.20 show the change of d with temperature,
calculated with TGA measurements performed with three different ramp rates. The
maximum d obtained is close to 0.5, which is reached by 400 ?C for Y and by 600 ?C
for the rest of the oxides. Further- more, across all ramp rates, the maximum OSCs
vary between 2.20 and 3.13 wt% (0.7?0.98 O2 mmol/ g) at 600 ?C, which are listed
in Table 3.3. Taylor et al. has reported the OSC of the La1-xSrxFeO3 perovskite
system together with OSCs of other metal oxides from the literature.49 In these
oxide systems the OSCs range from 1.7?3.7 wt%. Therefore, AFe2O4 compounds in
this study have comparable OSCs to these perovskites, and structural modifications
could be explored to further increase their OSCs. Fig. 3.20 shows that once the
maximum oxygen gain is reached at 600 ?C, the OSCs remain stable over a time
span of 3 hours. While oxidation kinetics correlate with A-site ionic radii, the final
OSCs do not. However, variations between OSCs could be complicated by other
parameters such as particle size and surface area. However, beyond the scope of this
study, these parameters should be analyzed to better understand how to improve
OSCs in the AFe2O4 system.
3.5 Conclusions
We have studied four A-site analogues of AFe2O4 (A = Lu, Yb, Y, In) as
OSMs. The Y and In compounds have never been explored as OSMs before, and we
92
L u F e 2O 4 Y b F e 2O 4 Y F e 2O 4 In F e 2O 4 T e m p e ra tu re
0 .6 0 .6
5?C/min 7 0 0 10?C/min
7 0 0
0 .5 6 0 0 0 .5 6 0 0
0 .4 5 0 0 0 .4 5 0 0
4 0 0 4 0 0
0 .3 0 .3
3 0 0 3 0 0
0 .2 0 .2
2 0 0 2 0 0
0 .1 1 0 0 0 .1 1 0 0
0 .0 0 0 .0 0
5 0 1 0 0 1 5 0 2 0 0 2 0 4 0 6 0 8 0 1 0 0 1 2 0 1 4 0
0 .6 0 .6
25?C/min 7 0 0 6 0 0
0 .5 6 0 0 0 .5 5 0 0
0 .4 5 0 0 0 .4 4 0 0
4 0 0
0 .3 0 .3 5?C/min 3 0 0
3 0 0 Dwell-3hrs
0 .2 0 .2 2 0 0
2 0 0
0 .1 1 0 0 0 .1 1 0 0
0 .0 0 0 .0 0
2 0 4 0 6 0 8 0 1 0 0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0
T im e  (m in )
Figure 3.20: Thermogravimetric analysis done for the four A-site analogues a) 5
?C/min b) 10 ?C/min c) 25 ?C/min equilibrating every 100 ?C for 10 minutes d)
Heated at 10 ?C/min till 600 ?C and then held at 600 ?C for 3 hours.
Ramp 5 10 15
rate ?C/min ?C/min ?C/min
OSC wt % O2 wt % O2 wt % O2
mmol/g mmol/g mmol/g
Lu 2.29 0.72 2.20 0.69 2.25 0.70
Yb 2.41 0.75 2.36 0.73 2.21 0.69
Y 2.99 0.93 3.07 0.96 3.13 0.98
In 2.64 0.82 2.60 0.81 2.38 0.74
Table 3.3: OSCs calculated from TGA experiments.
93
O x y g e n  N o n - s t o i c h e o m e t r y  (? )
Temperature (?C)
have found that Y and In exhibit the fastest and the slowest kinetics, respectively, for
oxygen uptake across all ramp rates. This study affords us the ability to understand
how the non-redox active metal center can affect important properties such as oxygen
insertion kinetics and cycling stability. The A-site cationic radii demonstrate a clear
correlation with the lattice parameter change of the pristine structures and with the
iron valence state as evidenced by the BVS analysis. The deviation of the BVS of
Fe from the ideal 2.5 state is directly proportional to the A-cationic radii, while
the lattice parameters show an anisotropic correlation. When heated under air all
four analogues undergo a phase transition above 400 ?C from R? 3m (R1 phase)
to P? 3or C2/m symmetry (R3-phase). For Lu, Yb and Y, the formation of a
supercell structure is observed around 250 ?C as evidenced by satellite reflections in
sXPD data. Sequential refinements of in situ diffraction data indicate an anisotropic
cell volume change in air contrary to the isotropic thermal expansion under He.
As the deviation of the BVS of Fe increases, the temperature at which the phase
transition from R1/R3 completes in air decreases due to the more under-bonded
Fe. These layered oxides exhibit reversible oxygen insertion at 600 ?C in alternating
20% O2/He and 3.5%H2/He atmospheres transitioning between the two phases (R1
and R3). Increasing the cycling temperature from 500 ?C to 600 ?C increased the
rate of reaction, and the rate of oxidation was faster than that of reduction.While
Lu, Yb and Y showed stable cycling, the In analogue showed signs of degradation
under H2. OSCs calculated based on thermogravimetric analysis with different ramp
ratesvaryinthe rangeof2.20?3.13 wt% at 600 ?C, but no clear trend with the A-
cation size could be established for this metric. However, the oxidation kinetics
94
showed a direct correlation with the ionic radii across all ramp rates. Based on these
gathered metrics,the Yb and Lu analogues have comparable reaction rates, cycling
stability and OSCs, making Yb a cost-effective alternative to Lu. Considering the
fast oxidation kinetics of the Y analogue at low temperatures along with its ability
to cycle, it could be a suitable candidate for applications such as gas sensing in
addition to chemical looping.
95
Chapter 4: Solving the structure of the oxidized phase of AFe2O4
(Yb, Lu)
4.1 Introduction
Knowing the correct crystal structure of functional materials is crucial for
understanding their structure-property relationships. In our most recent study [129]
we reported the reversible oxygen storage capability of AFe2O4 (A = Y, Yb, In,
Lu). The layered crystal structure of AFe2O4 consist of edge-sharing BO5 trigonal
bipyramids alternate with edge-sharing AO6 octahedra (Fig. 4.3a). The A site is
a trivalent cation, typically a lanthanide (Ln3+) and B site is a transition metal
in a mixed valence state of +2.5, and its unit cell has the centrosymmetric space
group R-3m (Fig. 4.3b). While this family of layered materials has been extensively
been studies as a multiferroic material [109, 130], its oxygen storage properties are
recent discovery. Upon heating of this pristine structure above 500 ?C, a major
structural phase transition occurs and the maximally oxidized phase, AB2O4+? is
formed. From our thermogravimetric analysis we found the ? value is close to be
0.5. P -3 was used as the space group for the Lebail fit for this oxidized phase.
While there have been numerous studies done to index the unit cell [103,107],
96
only one study [131] has been done so far to solve the complete structure of this
oxidized phase, which was published in November, 2017. In this study maximum
entropy method was used together with electron, X-ray and neutron diffraction
experiments to confirm the final structure. Nevertheless we have also been inde-
pendently attempting to solve this structure using high resolution synchrotron X-
ray and Time-Of-Flight neutron powder diffraction data with simulated annealing,
Pair distribution function analysis and X-ray absorption spectroscopy. This current
report describes extensively on how these techniques were utilized in successfully
solving the crystal structure of AB2O4+? with powder diffraction.
Structure determination using powder diffraction data (SDPD) has become a
vital technique for materials for which single crystals are hard to synthesize or do
not exist. In some industries such as pharmaceutical and heterogeneous catalysis,
polycrystalline samples are a necessity. However, using powder diffraction (PD) data
to solve a structure is more challenging due to the collapsing of a three-dimensional
crystal structure information in to a one-dimensional diffraction pattern by spheri-
cal averaging, consequentially losing phase information from peak overlap. In early
days only simple, high symmetry structures could be solved with PD. Nevertheless,
due to recent advancements in instrumentation (e.g synchrotron sources) and com-
putational methods,the true potential of SDPD is realized, and it has become less
complicated and comparable in accuracy to single crystal methods.
History of powder diffraction and solving structures
There are several milestones in powder diffraction history which paved the path
97
to present day advances. In the history of crystallography many notable contribu-
tions were made towards solving structures using powder diffraction data. The very
first experiments carried out by Debye and Scherrer to solve the crystal structure
of LiF [132] in 1916 and by Albert Wallace Hull to solve the structure of Fe [133]
in 1917 using powder diffraction data paved the path to a whole new era in powder
diffraction. While many of these structures seem quite simple and straight forward
for today, the techniques and methods used back then show remarkable intuition for
their time. Among such studies, the crystal chemical study of the 5-f element series
carried out by W. H Zachariasen stands out. Infact, William Houlder Zachariasen
goes down in history as the first person to solve the crystal structure of a non-cubic
binary system: UCl3 in 1948, which was part of the research done at the Manhat-
tan project. [134] This series of publications consisted of several articles that solved
the structures of many elements and compounds with powder diffraction data. In
these compounds the positions of light atoms were deduced solely based on intensity
calculations or based upon the analysis of inter-atomic distances. [135]
In almost all of these studies powder diffraction photographs were obtained
in transmission geometry with a small amount of the sample loaded in to a thin
walled capillary using Cu K radiation filtered through nickel foil. [136]. Systematic
absences and reflections with high intensity were taken in to account in deducing the
positions of heavy elements. His work done on plutonium used differential thermal
expansion to resolve Bragg peak overlap, which was ingenious. [137] In their study
they figured out for the first time that reflections that were superimposed at one
temperature can be resolved in to individual components at a different temperature
98
due to the anisotrpy in the thermal expansion of the metal. The other significant
SDPD study worth mentioning around this time period is the one carried out by
Debets in 1968. In this study the structure of Uranyl Chloride was solved using
Patterson methods. [138] Parallel to X-ray diffraction, neutron powder diffraction
experiments also came in to light for solving crystal structures. The powder pattern
solved using neutron diffraction was first reported for NaCl that was done during
the Manhattan project at the Oak Ridge national laboratory (known as the Clinton
pile back then). [139]
A boost in computational methods in SDPD was clearly evidenced after the
first publication of Hugo Rietveld on his whole powder pattern fitting, least squared
method in 1967 used for tungsten oxide. [71] In this short communication he showed
that one can refine completely overlapping reflections in powder diffraction with the
assumption of a Gaussian profile for neutron Bragg reflections. Finally in 1969 Ri-
etveld published his famous paper ?A profile refinement method for nuclear and
magnetic structures? giving birth to the modern Rietveld refinement method. [140].
Thanks to the Rietveld method technique of structure solution with powder diffrac-
tion data expanded from high symmetry materials to ones with high complexity.
Furthermore, the extension of the Rietveld method to X-rays [141] from neutrons
and then to synchrotron X-rays [142] and Time-Of-Flight neutrons [143] created
an even bigger platform for SDPD. Many of the early studies done to solve zeolite
crystal structures were the stepping stones to the concepts that are implemented in
the present day computer algorithms. In the study done by Olsen et al in 1978 trial
model building refined with distance least-squares (DLS) program was used with di-
99
rect methods to solve the crystal structure of the synthetic zeolite, ZSM-5. [144,145].
DLS was a program built where the chemical connectivity and bond distances (1st
and 2nd neighbor) of a system was readily defined, and which provided a structural
model combined with crystal symmetry and symmetry constrainsts. [146] The whole
powder pattern decomposition methods developed by G. S. Pawley in 1981 [147] and
Armel Lebail in 1988 [69] showed how to extract integrated peak intensities in pow-
der diffraction data. In the Le Bail algorithm, the problem of intensity variations in
overlapped peaks is overcome since it intrinsically tends to equipartition intensities
when peaks overlap. The LeBail and Pawley methods boosted the progress of SDPD
as accurate extraction of intensities in powder diffraction patterns is a crucial step in
today?s structure determination in both direct and reciprocal space methods. This
is the step when a pseudo-single crystal data set is generated from PD data.
Recent advances and computational methods in SDPD
The introduction of synchrotron X-ray sources enabled the success rate in SDPD
due to its high flux radiation and excellent vertical collimation that resulted in high
resolution data sets. The orthorhombic structure of CrPO4 was the first structure
that was solved with synchrotron X-ray powder diffraction (sXPD) using Patterson
methods with a vector search procedure. It was a unit cell with 8 atoms and 68
peaks were used. [148] In 1994 a structure with 60 atoms was solved with sXPD
using a direct method and Different Fourier syntheses demonstrating the powerful
impact of synchrotron sources. [149]
Once the unit cell is indexed, the space group is figured out and the whole
100
powder pattern decmposition is done, numerous computational algorthms and ap-
proaches can be used to solve the structure with powder diffraction data. SDPD
is mainly of three forms, Reciprocal-space, Direct-space and Dual-space methods.
Reciprocal-space methods in SDPD are the conventional crystallographic (single
crystal) structure solutions methods such as Direct [150], Patterson [151], Maximum
entropy methods [152, 153], which are modified to overcome the inherent problems
in powder diffraction. [154] Usually SDPD in reciprocal space yields more success
when high resolution powder diffraction patterns with atomic resolution are used
that are free from preferred orientation. [150]. Commonly used direct method soft-
ware are SHELXS-86 and MULTAN-84 (find newer ones). While these direct and
Patterson methods can be used to solve crystal structures from powder diffraction
data, these techniques are based on knowing the intensity of each individual reflec-
tion. Therefore, they are more sensitive to the loss of phase information that might
occur during extraction of intensities in powder diffraction data.
SDPD is mainly of three forms, Reciprocal-space, Direct-space and Dual-space
methods. Reciprocal-space methods in SDPD are the conventional crystallographic
(single crystal) structure solutions methods such as Direct [150], Patterson [151],
Maximum entropy methods [152, 153], which are modified to overcome the inher-
ent problems in powder diffraction. [154] Usually SDPD in reciprocal space yields
more success when high resolution powder diffraction patterns with atomic reso-
lution are used that are free from preferred orientation. [150]. While these direct
and Patterson methods can be used to solve crystal structures from powder diffrac-
tion data, these techniques are based on knowing the intensity of each individual
101
reflection. Therefore, they are more sensitive to the loss of phase information that
might occur during extraction of intensities in powder diffraction data. Simulated
annealing [155,156], genetic algorithm [157] and Monte-Carlo [158] search are some
of well-known direct space methods. The main principle in direct space methods in-
volve finding the global minimum by optimizing cost functions of a model structure,
such as atomic fractional coordinates and crystal energy. Cerny et al. has listed
down several computer programs that uses direct space methods for SDPD. [159].
Simulated Annealing
In nature atoms and molecules arrange in the lowest energy (most stable) con-
figuration over time. Metals with defects such as dislocations of atoms are high in
energy and thermodynamically unstable. Even though over time these defects are
expected to disappear, the process of annealing is used to remove these defects in
metals in a shorter period. Annealing is a thermally activated process where the
metal is heated to a higher temperature and then cooled down gradually. In doing
so, solid state diffusion of defects will either remove them completely or rearrange
them in a lower energy configuration hence minimizing the over al energy of the sys-
tem. [160] Simulated annealing mimics the annealing process of a metal in searching
for an energy minimum state. SA was proposed as a global minimization technique
in 1983 by Kirkpatrick et al. by using it for the famous travelling-saleman prob-
lem. [161] It?s an iterative, least squared minimization algorithm derived from sta-
tistical mechanics where the global minimum of a system is achieved by instigating
random changes analogous to ?temperature? in the atomic coordinates system (real
space), that in turn minimizes a certain function corresponsing to energy. [156] The
102
acceptance of this new system is determined by a Monte-Carlo like algorithm. [162]
A phase is an ensemble of states with different energies and the energy of any
state can be linked to the temperature by Boltzmann statistics. According to a
Boltzmann distribution the energy difference of two states at high temperature will
be less, where as at low temperatures the probability of the states with low energy
will be high. This physical property of a system is utilized in the SA algorithm
and temperature is used as the control for finding the global minimum. In SA the
?states? are the different configuration of atoms in the model structure and ?energy?
is a measure of how agreeable are the changes to the constraints and the observed
diffraction pattern. [155] (Monte Carlo method is employed in SA in every fixed
temperature, so MC is like the building block for SA) While minimization of the
cost function (energy function driven by ?temperature? changes) is the ultimate
goal, it is allowed to increase when needed in order to overcome local minima (Fig.
4.1). The temperature of the system here simply refers to the relative probability
of two successive states rather than in a thermodynamic context. [163]
This algorithm where mostly a downhill route is followed with occasional uphill
movements is known as the Metropolis algorithm. [164] According to the metropolis
algorithm, if the initial energy of the system/model structure is Einitial, the tem-
perature of the system is randomly varied until the condition shown in equation
1 is met where Enew is energy of the system after random change corresponding
temperature change has been done, which then becomes an acceptable change. The
same condition is shown in equation 2 where T is the temperature, K is a constant,
rand(0,1) represents a random number between 0 and 1. This is what determines
103
whether the random change made to the system is accepted or not. The metropolis
scheme is also shown in scheme 2.
Enew?Einitial (4.1)
?Enew ? Einitialexp[ ] < rand(0, 1) (4.2)
KT
This is in contrast to a traditional gradient based optimization method where
the direction is only downhill causing to be trapped in the closest local minimum.
The reliability of these simulations depend on both the simulation methodologies
utilized and the quality of the potential energy function employed. Another factor
that contributes to the success and reliability of SA, is the accuracy of the initial
model. In scheme 2 the need go back and forth between step 3 and 1 is because of
this.
In the SA algorithm, the search of the global minimum is controlled by several
parameters. The initial temperature T, the rate of cooling/annealing, and the mag-
nitude of the change in atomic coordinates in response to the temperature, which is
approximated by the law of equipartition of energy, are the most significant factors
out of all. [164]
Many programs using SA for structure solutions had the common problem of
taking too long to find the global minimum. Enhancements were done to over come
this by Coelho. Three modes of minimization are used in iteration with appropriate
penalty functions (ex: anti-bumping penalty, potential energy penalty), penalty
104
S = Starting Point  Current state T3<T1< T2
Energy Energy Energy
T2
T1
S
T3
L.M L.M L.M
G.M G.M G.M
    L.M = Local Minimum     G.M = Global Minimum
Figure 4.1: Simulated annealing
weighting and potential energy terms, and the magnitude of displacement in the
coordinate system is related to the average bond lengths. The choice of temperature
regime for the structure solution depends on the chemistry of the unit cell, expected
bond lengths and the ability of the minimization method to return to the global
minimum after a randomization. In this study the discussed SA method was used
on three compound; AlVO4 K2HCr2AsO10 [Co(NH3)5CO3]NO3.H20. it was found
that the increase in penalty weighting, using linear minimization, and calculating
the first and second derivative of the penalty functions increases the success rate of
finding the global minimum.
Toraya et al used SA to solve the crystal structure of CaSO4 polymorph. [165]
The re-annealing technique introduced by Ingber in 1989 showed how to avoid get-
ting trapped in a local minimum. [166]
Charge Flipping
An iterative algorithm which uses both direct and reciprocal space. First de-
105
Simulated Annealing Charge flipping
TOPAS, DASH, GSAS II PLATON, BayMEM, TOPAS,
Olex2, JANA2006, WinGX,
Crystals, FullProf
Table 4.1: Software packages available for Simulated annealing and Charge flipping
scribed by Oszla?nyi and Su?to? in 2004. Required input: lattice parameters and
reflection intensities (Fobs). No use of chemistry or trial structure models. Struc-
tures can be (partially) solved in a few seconds up to a few minutes (generally faster
than is takes to set up a trial structure model for simulated annealing).
Crystal structure AFe2O4
Knowing the correct crystal structure of functional materials is crucial for un-
derstanding their structure-property relationships. The AFe2O4 family, the primary
focus of this study, consists of well know multiferroic materials. The A site is a
trivalent cation, typically a lanthanide (Ln3+), and the B site is a transition metal
in a mixed valence state due to the coexistence of +2 and +3 states. The layered
crystal structure of AFe2O4 consists of edge-sharing BO5 trigonal bipyramids that
alternate with edge-sharing AO6 octahedra (Fig. 4.3 a). The unit cell of this trian-
gular lattice has the centrosymmetric space group R-3m and in some cases C 2/m.
While this family of layered materials has been extensively studied as a multifer-
roic material [109,130], its oxygen storage properties are a recent discovery. [167] In
our most recent study [129] we reported the reversible oxygen storage capability of
AFe2O4 (A = Y, Yb, In, Lu).
Upon heating of the pristine structure above 500 ?C, a major structural phase
transition occurs and the maximally oxidized phase, AB2O4+? is formed. Not many
106
Figure 4.2: Charge flipping algorithm
107
a b
A (Lu, Yb)
Fe
AFe2O4
O R-3m
AO6 octahedral
mono layer
FeO5 trigonal
bipyramidal cR
bi-layer
Z
X
Z
Y
X aR
Figure 4.3: a) The crystal structure of AB2O4 and its b) unit cell in R-3m space
group.
studies have been done on the oxidized phase in contrast to its multiferroic reduced
phase, hence the crystal structure remained unsolved up until very recetly. Ge?rardin
et al. published about the trigonal symmetry of this phase for the first time in
1982. [121] Since then only one study [131] has been done so far to solve the complete
structure of this oxidized phase, which was published in November, 2017. In this
study, the maximum entropy method was used together with electron, X-ray and
neutron diffraction experiments to confirm the final structure. As an emerging,
novel oxygen storage material confirming and validating the accurate structure of
this material is important in order to better understand its functional properties.
We have been independently attempting to solve this structure as an extension
of our previous work. [129] In the current study, we discuss how the structure of the
108
hyperstoicheometric oxidized phase of AFe2O4 (A = Yb, Lu) is solved with simu-
lated annealing and Difference Fourier maps using high resolution synchrotron X-ray
powder diffraction (sXPD) and Time-Of-Flight (TOF) neutron powder diffraction
(NPD) data. We have used simulated annealing algorithm discussed by Coelho [156]
with TOPAS 5 to locate the atoms in the unit cell, which is discussed in detail in
the next section. While SA and Difference Fourier maps were used in solving the
structure, X-ray absorption spectroscopy and Pair Distribution Function analysis
were used to confirm the final model combined with Rietveld analysis. Solving and
confirming the crystal structure of AFe2O4+? (A= Lu, Yb) with these techniques
has never been reported before.
109
4.2 Experimental
4.2.1 Materials synthesis
All samples were prepared using solid-state reactions. 99.9 % (Alfa Aesar) A2O3
(A=Yb, Lu), 99 % (Sigma Aldrich) Fe2O3 and 99.5 % (Sigma Aldrich) Fe powder
were ground in an agate mortar and pestle in 1/2: 5/6: 1/3 ratio respectively to
prepare 0.25 g of the target compound, and pressed into pellets of 13-mm diameter.
The pellets were placed in a 2 ml alumina crucible and sealed inside evacuated 8 mm
diameter quartz ampoules. Before sealing, the ampoules were flushed with N2 gas
and evacuated several times to obtain the required oxygen partial pressure. [117]
The samples were sintered at 1180 ?C for 12 hours and then quenched to room
temperature. To make the oxidized phases, AFe2O4+? (ex situ), the as synthesized
samples were oxidized at 500 ?C under air and cooled to room temperature.
4.2.2 High temperature in situ sXPD characterization
In situ sXPD experiments were performed in transmission geometry on the 17
BM beamline at the APS. A 2D PerkinElmer Si flat panel detector was used with
an average wavelength of 0.45336 A?. We collected diffraction patterns every 0.33
seconds and integrated the diffraction images with GSAS-II [119] to yield patterns
with a Q-range of 0.30 A?-1 to 6.6 A?-1. The powder samples were loaded in to a
quartz capillary sample holder through which, a constant gas flow of 15 mL/ min
was maintained. A residual gas analyzer (RGA) was used to monitor the gaseous
110
products. During the experiment we heated the sample environment up to 700 ?C
under 20
4.2.3 Ex situ high-resolution sXPD and TOF NPD characterization
Ex situ experiments with high-resolution synchrotron X-ray powder diffraction
(sXPD) were performed on the 11 BM beam line at the Advance Photon Source
(APS) at Argonne National Laboratory. An average wavelength of 0.41458 A? was
used, covering a Q-range of approximately 0.7 A?-1 to 10 A?-1. Powder patterns were
collected at ambient temperature and pressure for both the pristine compounds
(AFe2O4) and their oxidized phases (AFe2O4+?) prepared ex situ. The powder pat-
terns were analysed by Rietveld refinements for the pristine phases and Le Bail fits
for the oxidized phases using TOPAS 5.15
4.2.4 X-ray absorption spectroscopy
Iron K-edge X-ray absorption spectra were collected at beamline 9-BM at the
Advanced Photon Source (APS) at Argonne National Laboratory. Measurements
were performed in transmission geometry using gas filled ion chambers as detectors
on powder samples diluted with boron nitride and pressed into pellets. Athena and
Artemis software were used for data processing and fitting. [92] The analysis of the
EXAFS spectra were carried out by Dr. Sterbinsky at the 9-BM at the advanced
photon source.
111
4.2.5 Pair Distribution Function (PDF) Analysis
TOF neutron diffraction for PDF analysis was done at the NOMAD beam line
at Spallation Neutron Source at the Oak Ridge National Laboratory. PDF was
calculated using TOF NPD collected on the NOMAD beam line at the Oak Ridge
national laboratory. The sample-to-detector distance was calibrated with a dia-
mond standard and a vanadium can was used as a standard for incoherent neutron
scattering. Measurements were done for the pristine and the oxidized samples of
LuFe2O4 in vanadium cans at RT. Following were the instrumental parameters used
for the refinement Qmax= 37 A?,Qdamp= 0.017659, and Qbroad= 0.0191822. [168]
The parameter Qbroad defines the instrumental contribution to the peak broaden-
ing in the neutron pair distribution function (nPDF) while the parameter Qdamp
defines the instrumental contribution to the dampening envelope of the nPDF.
4.3 Results and discussion
4.3.1 Formation of the oxidized phase, AFe2O4+?, from the pristine
AFe2O4 phase
The Rietveld refinement done for the pristine AFe2O4 phases using high resolu-
tion sXPD are shown in Fig. 4.4 and the structural parameters are given in Table
4.2. The Yb and Lu analogues are iso strcutural. We used anisotrpic displacement
parameters, and both A-site ions show high displacement along the z axis with U33
= 0.0341(2) A?2 and 0.0379(3) A?2 for Lu and Yb respectively. Oxidation of these
112
LuFe2O4 R-3m a = 3.4559(3) A? c = 25.108(1) A?
Atom Site x y z Occ Beq (A?2) U11 U33
Lu 3a 0 0 0 1 0.31(5) 0.0006(7) 0.015(1)
Fe 6c 0 0 0.2147(1) 1 0.32(9)
O1 6c 0 0 0.1448(12) 1 11.01(94)
O2 6c 0 0 0.2896(9) 1 2.56(47)
YbFe2O4 R-3m a = 3.45765(6) A? c = 25.2374(7) A?
Atom Site x y z Occ Beq (A?2) U11 U33
Yb 3a 0 0 0 1 0.83(3) 0.0019(3) 0.0321(8)
Fe 6c 0 0 0.2155(1) 0.875 0.41(4)
O1 6c 0 0 0.1280(2) 1 0
O2 6c 0 0 0.2933(2) 1 0
Table 4.2: Structural parameters from combined Rietveld refinement done using
high resolution sXPD for AFe2O4 at 298 K (A=Lu,Yb)
pristine phases are rarely looked into, as the attractive multiferroic features dispear
upon oxidation. Only a few studies are reported on the in situ oxidation [63, 129]
and on the structure of the oxidized phase. [121,131]
We conducted in situ diffraction experiments to monitor the oxidation behavior
of AFe2O4 (A = Yb, Lu) compounds by heating it in air. Fig. 4.5 illustrates
the evolution of structure from the stoichiometric AFe2O4 to hyper stoichiometric
AFe2O4+?. Upon oxidation, the space group symmetry graduallly changes from R-
3m to P -3. The stoichiometric, reduced phase undergoes several modulations due
to oxygen insertion and due to the melting of the charge ordering of Fe2+/Fe3+
ions. [63, 169] Up until 200 ?C we didn?t observe any new reflections, between 200-
500 C ?new satelite reflections were observed forming a supercell structure. Once
the temperature reaches 500 ?C the oxidized P -3 phase is fully formed. In this
transition the oxidation state of Fe changes from +2.5 to +3, and these oxidation
states were confirmed though our XANES analysis in the section below. We have
113
25 LuFe2O4 (RT,R-3m) Ycal  
Lu O  (Ia-3) 0.87 wt.% Yobs-Ycalc2 3
20 Rwp = 13.68 % Yobs
15
10
5
0
?5
YbFe2O4 (RT,R-3m)
Ycal  
FeO (Fm-3m) 1.41 wt. %        Yobs-Ycalc
Yb2O3 (Ia-3) 0.53 wt.% Yobs
Rwp = 14.04 %
1.32 2.64 3.96 5.28 6.60
Q (?-1)
Figure 4.4: Observed, calculated, and difference curves of the LuFe2O4 and
YbFe2O4 crystal structures at room temperature (RT) with high-resolution syn-
chrotron X-ray powder diffraction patterns. A wavelength of 0.41458 A?A was used.
114
Intensity (Arbitrary Units)
a 4 b 4
4.5 10 10 P3
4 Lu Yb 1
4 / C3 
3.5 6 0 0  ? C
3 3
2.5 R3m
2 2
1.5
1 1 C
0.5
R T
0 0
8.5 9 9.5 10 8.5 9 9.5 10
2? (?, ? = 0.45336 ?) 2? (?, ? = 0.45336 ?)
Figure 4.5: In situ sXPD patterns collected while samples were heated in air from
RT to 600 ?C
extensively discussed the lattice parameter evolution and structural changes with
temperature in our previous study. [129] The fully oxidized phase remains stable
up to 700 ?C, and after cooling to room temeprature. This transition is reversible
under a reducing environment such as H2, which makes these materials excellent
candidates as oxygen storage materials. [63, 129]
4.3.2 Solving the structure with simulated annealing (SA) and Fourier
difference map
Many programs using SA for structure solutions had the common problem of
taking too long to find the global minimum. Enhancements were done to overcome
this by Coelho. [156] and incorporated in the software TOPAS, which is what was
used in this study. Three modes of minimization are used in iteration with appro-
priate penalty functions (ex: anti-bumping penalty, potential energy penalty) and
115
Intensity (Arbitrary Units)
Intensity (Arbitrary Units)
1. Simulated annealing with sXPD
2. Fourier differnce maps with TOF 
NPD
3. Combined X-ray-neutron Rietveld 
refinemnt
No
Good fit?
Yes
4. Confirmation of structure with PDF 
and EXAFS
Figure 4.6: Steps used in solving the structure with powder diffraction
potential energy terms, and the magnitude of displacement in the coordinate system
is related to the average bond lengths. The choice of temperature regime for the
structure solution depends on the chemistry of the unit cell, expected bond lengths
and the ability of the minimization method to return to the global minimum after a
randomization. It was stated by Coelho that the increase in penalty weighting, using
linear minimization, and calculating the first and second derivative of the penalty
functions increases the success rate of finding the global minimum. [156]
To solve the structure of the oxidized phase with SA , high resolution sXPD
and TOF NPD data collected at room temperature was used to decouple any tem-
perature dependent variations. The unit cell of this hyperstoichiometric phase,
AFe2O4+?, has been indexed in different space groups that varies on the type A-site
cation, [121, 170] however P -3 remains to be the most commonly used spacegroup
116
for the Yb [131], Lu [121] analogues. As the first step to SA, a Pawley fit was
performed to determine the accurate lattice parameters and background with the
high resolution sXPD data. The possibility to use step intensity data (maximum in-
tensities) in TOPAS instead of intensity extraction, makes structure determination
viable even from poor quality data. Next macros specific for SA that initiate the
Monte-Carlo process and carry out the global minimum search in parameter space
was introduced in launch mode with suitable constraints and restraints. The macros
used are given in Appendix C with further information.
Initially only the metal sites were introduced excluding the lighter oxygen sites
as this is the best approach to solving the structure. [171] One site for each metal
ion was introduced. The initial coordinates were given the value of zero and set to
refine while all other variables obtained through the Pawley fit (lattice parameters,
background etc) were fixed. Once the input file was executed in TOPAS, the SA
algorithm promptly located the metal ion positions, which are shown in Fig. 4.7a.
The 1a site for Yb/Lu occupying the corners of the unit cell and the 2d site for
Fe occupying the (2/3, 1/3, z) positions. This step was repeated several times and
each time the same positions were obtained. This shows the precision of the global
minimum obtained as opposed to a false local minima. Finally Fourier difference
maps were used with TOF neutron powder diffraction data in TOPAS 5 to deduce
the residual electron densities, which should correspond to the remaining oxygen
sites. The Fourier difference map output is shown Fig 4.7b. Finally the occupancies
were adjusted based on the number of formula units in the cell generating the average
structural model.
117
a b
b
6
1 1
A 1
1
Fe 3
Residual 
electron 4
densities 
2 2 22
5
Figure 4.7: The direct output of a) Simulated annealing and b) Fourier difference
map AFe2O4+? (A= Lu, Yb) The residual electron densities are shown as white
spheres as these were recognized as carbon atoms by default in TOPAS.We recog-
nized sites numbered from 1-6 are oxygen sites while the residual electron densities
on top of the Yb/Lu sites were due to the atomic displacement of the metal ions.
In the 6 types of oxygen atoms, 1-2, 3-4, 5-6 pairs belong to O3(2c), O2(2d) and O1
(2d) sites.
a b
Lu Fe O1
O2
O3
Z Z
x X Y x
X
Y
Figure 4.8: a)Unit cell and b) Crystal structure in polyhedral view of AFe2O4 (A=
Lu, Yb)
118
4.3.3 Combined Rietveld refinement and crystal structure
The CIF was constructed for the oxidized phase of YbFe2O4 using the new
atomic positions. As this structure contains both light (O) and heavy elements(Yb,
Lu), using neutron and X-ray diffraction data simultaneously in a combined Rietveld
refinement will yield more accurate structural parameters. Therefore, a combined
Rietveld refinement using both sXPD and TOF NPD was performed as illustrated
in Fig. 4.9 in launch mode of TOPAS. A pseudo-Voigt function was used for the
peak shape profile and the intensities were corrected for Lorentz-polarization and
absorption effect. The structural factors are given in table 4.3. All oxygen sites are
under occupied while O3 site being the least occupied site. Occupancy value for the
O3 site is close 0.5 for both Lu and Yb analogues. Isotropic atomic displacement
parameters (ADPs) were first used for the refinement. Refinement was then repeated
using anisotropic ADPs, which are shown in Table 1. As per symmetry constraints
for the space group P -3, U11=U33=2U12 and U13=U23=0. [68]. The anisotrpy was
higher for oxygen compared to the metal ions. The relatively large U11 of oxygen
sites suggests high disorder in the x-y plane in the oxygen sub lattice. In the report
by Nicoud et al [131] this oxygen site was introduced as an interstitial site.
In the final structure the coordination of Fe is changed from 5-fold to 6-fold,
converting the the trigonal bi-pyramid in to an octahedron.
119
10x2
12 a NPD  YbFe2O4+? (RT,P-3)
10 YcalYobs-Ycalc
Yobs
8
RWP = 12.87 %
6
4
2 * * *
0
?2
1.0 1.5 2.0 2.5 3.0
10x3 b sXPD  YbFe2O4+? (RT,P-3)
25
20
15
10
5
0
?5
1 2 3 4 5 6 7 8 9
d-spacing (?)
Figure 4.9: Combined Rietveld refinement done for YbFe2O4+? using TOF neutron
powder diffraction (NPD) and synchrotron X-ray powder diffraction (sXPD). This
oxidized phase was prepared by heating the pristine phase ex-situ at 600?C. The
star marks shows several remnant peaks from the pristine (R-3m) phase, which
were invisible to lab X-rays.
120
Intensity ( Arbitrary units)
YbFe2O4 P -3 a = 3.48657(3) A? c = 8.42591(9) A?
Atom Site x y z Occ Beq (A?2) U11 U33
Yb 1a 0 0 0 1 1.89(3) 0.030(1) 0.080(1)
Fe 2d 1/3 2/3 0.3445(4) 1 2.33(3) 0.058(1) 0.040(1)
O1 2d 2/3 1/3 0.8768(7) 0.83(2) 2.62(12) 0.153(4) 0.007(1)
O2 2d 2/3 1/3 0.4033(6) 0.69(1) 2.03(11) 0.173(5) 0.0002(12)
O3 2c 0 0 0.6523(07) 0.47(1) 3.64(22) 0.106(5) 0.009(3)
LuFe2O4 P -3 a = 3.47166(2) A? c = 8.44945(9) A?
Atom Site x y z Occ Beq (A?2) U11 U33
Lu 1a 0 0 0 1 1.8(1) 0.023(1) 0.069(2)
Fe 2d 1/3 2/3 0.3426(3) 1 2.43(3) 0.066(1) 0.032(1)
O1 2d 2/3 1/3 0.8864(5) 0.84(1) 4.0(1) 0.180(5) 0.023(1)
O2 2d 2/3 1/3 0.3937(2) 0.60(1) 1.6(1) 0.104(5) 0.003(1)
O3 2c 0 0 0.6653(10) 0.42(1) 3.1(1) 0.075(4) 0.001(2)
Table 4.3: Structural parameters from combined Rietveld refinement done using
TOF NPD and high resolution sXPD for AFe2O4 at 298 K
Lu Fe O1 O2 O3
Oxygen 
B ] insertion CA B
B A
C C
B B
C A
A Layer ] CmovementC BZ
A A
Y x
Figure 4.10: The layer movement due to oxygen uptake during the phase transition
from the reduced phase (left) to the oxidized phase (right) of AFe2O4 (A= Lu,
Yb). In the reduced phase the cation layering show ACA.CBC.BAB stacking, which
changes to ABC after the layer shift
121
4.3.4 Pair Distribution Function (PDF) analysis
PDF of LuFe2O4 was obtained by the Fourier transformation of neutron powder
diffraction data collected at the NOMAD beamline. NOMAD is the ideal instrument
for these measurements because it has the high neutron flux and gas flow capabilities
(ILL furnace) needed for PDF analysis throughout this experiment. The nanoscale
nature of the structural modification with oxygen uptake limits the ability of tra-
ditional diffraction experiments in studying this system. Furthermore, it is already
known that there is a charge ordering transition between RT and the active oxy-
gen storage temperature ( 500-600 ?C), and it is therefore necessary to study this
sample in-situ at elevated temperatures. PDF primarily provides information on
the local atomic structure, specifically it gives the probability of finding any two
atoms at a given distance. [85] We used PDF analysis as a tool to further confirm
the structural model of the oxidized phase of LuFe2O4 .Furthermore, since PDF
provides information on the local environment and disorder, useful information that
cannot be seen in a regular diffraction pattern can be inferred. PDFGui was used
to perform the fit using the model structure. A Q-max of 31.42A? was used. The
optimum Q-range could not be used for Lu due to its resonance effect observed at
high Q, which truncated the usable Q-range. In the fit all the atomic displacement
parameters were refined iso-tropically. Fig. 4.11 depicts the fits done for both the
reduced and oxidized phase obtained ex-situ for LuFe2O4. While the fit for the
oxidized phase was reasonable overall, a few misfits were observed specially at low
?r? region. When these misfits were closely analyzed, they corresponded to metal-
122
oxygen and oxygen-oxygen bond distances. In PDF, errors can be caused due to
several sources such as inaccurate corrections for experimental artifacts, termination
of the Fourier transform due to diffraction data with low a Q-range, low instruments
resolution and so on. [85] However, we believe these factors had a minimum effect
on our data. Hence, the misfits corresponds to a certain disordering in the oxygen
sub-lattice. In fact the atomic displacement parameters for oxygen obtained from
combined Rietveld analysis showed high values (Table 1), specifically for the O3 site.
Nevertheless the good agreement in the fit at high r-region confirms the accuracy of
the average structure.
123
a c
6
LuFe2O4 (RT, R-3m) G-Cal 4 LuFe2O4+? (RT,P-3) Fe - OG-Diff Lu - O
4 G-Obs O - O3 Total
2 2
1
0
0
-2
?1
-4 ?2
?3
-6
2 3 4 5 6 7 8 9 10
b d
LuFe2O4+d (RT,P-3) G-Cal Lu Fe O16 G-Diff O2
G-Obs O3
4
2
0
-2
-4
-6
-8 Z
10 20 30 40 50 x X Y
r (?)
Figure 4.11: The PDF fit done for the pristine(top) and the oxidized phase (bottom)
of LuFe2O4 using TOF neutron powder diffraction (NPD) collected on the NOMAD
beam line
124
G(r) G(r)
G(r)
4.3.5 X-ray absorption spectroscopy
X-ray absorption spectroscopy (XAS) is a versatile technique for characterizing
materials from liquids to solids and from amorphous to crystalline materials. In
this study, Fe K-edge x-ray absorption spectra were measured, as both the oxida-
tion state and the coordination environment of Fe change upon oxygen insertion
in AFe2O4(A= Lu, Yb). Fig.4.12 depicts the XANES spectra of the reduced and
oxidized phases of YbFe2O4 overlapped with Fe-standards with different oxidation
states. As illustrated, the absorption edge of the reduced phase of YbFe2O4 lies in
between that of FeO and Fe2O3 while the absoprtion edge of the fully oxidized phase,
YbFe2O4+? overlaps with that of Fe2O3. This confirms that the reduced phase has
both Fe2+ and Fe3+ while the oxidized phase only has Fe3+ ions. The Fe valence
of these layered materials have been previously reported by various experimental
probes such as Fe-2p, O-1s X-ray absorption spectroscopy (XAS), and Mo?ssbauer
spectroscopy. [64,172]
The EXAFS spectrum provides information on the local atomic structure of
the absorbing atom. [89] Hence EXAFS was used as a probe to validate the model
structure of the oxidized phase of AFe2O4 obtained from simulated annealing. The
Fe K-edge XAS was normalized, and a background was subtracted to isolate the
EXAFS part of the spectrum using the Athena computer program. [92] The energy
axis was transformed to k-space using the maximum of the first derivative of the Fe
K-edge XAS as E . [89] The kn0 weighted EXAFS data, where n = 1, 2, and 3 was
Fourier transformed from 2.5 to 11.0 A??1 using a Hanning window with transition
125
Compound A ?2 S20 ?E0 (eV)
YbFe2O4 0.957(4) 0.002(1) 0.41(3) ?6.8(9)
YbFe2O4+? 1.058(4) 0.006(1) 0.41(3) ?6.8(9)
LuFe2O4 0.961(5) 0.002(1) 0.41(3) ?6.8(9)
Table 4.4: Values of fitting variables resulting from a simultaneous fit of the first
Fe-O shell of EXAFS data from YbFe2O
2
4, YbFe2O4+?, and LuFe2O4. S0 and ?E0
were constrained to be equal for all three compounds.
ranges of 2.0 to 3.0 A??1 and 10.5 to 11.5 A??1.
Fourier transformed EXAFS data sets from YbFe2O4, YbFe2O4+?, and LuFe2O4
were fit simultaneously with the phase and amplitude functions of the Fe-O pho-
toelectron scattering paths calculated using Feff from crystal structures determined
using simulated annealing. The fit was done between 1.0 and 2.35 A? to isolate the
first Fe-O coordination shell using the Artemis software [92]. For each data set, all
three k-weights were fit. Because the first shells contain multiple Fe-O bond lengths,
an isotropic expansion factor (A) was fit for each compound to reduce the number of
independent bond lengths. A single Debye-Waller factor (?2) was fit for each com-
pound, which assumes the same disorder for all Fe-O bonds in a given material. A
single passive electron reduction factor (S20) and correction to the threshold energy
(?E0) were fit for all scattering paths, while the coordination numbers were held
fixed. For the fit, the values of the R-factor and ?2? were 0.0215 and 460, respec-
tively. The resulting fits of the YbFe2O4 and YbFe2O4+? EXAFS are shown in Fig.
4.13, and the resulting fitting parameters are given in Table 4.4.
126
Fe2O3
FeO
1 YbFe2O4 ( R3-m)
YbFe2O4+? (P-3)
0.8
0.6
0.4
0.2
0
7090 7100 7110 7120 7130 7140
Energy (eV)
Figure 4.12: XANES spectra of the reduced and oxidized phases of YbFe2O4 over-
lapped with iron standards.
a b
4 YbFe2O 4 
5
Ycal YbFe2O 4+?  Ycal
(R T,R -3m ) Yexp 4 (R T,P -3) Yexp
3
3
2
2
1 1
0 0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
R  (? )
Figure 4.13: Fourier transform magnitudes of the EXAFS spectra of the reduced
and oxidized phases of YbFe2O4 and fits
127
| ? (R )| (?3)
Normalized ?(E)
4.4 Conclusion
We successfully solved the average structure of AFe2O4+? (A= Lu, Yb) with
powder diffraction data. Simulated annealing was used with sXPD data to find the
metal sites while Fourier difference maps with TOF, NPD was used to solve the
oxygen sites. Combined X-ray and neutron Rietveld refinement was performed for
the final model structure. Finally XAS and PDF data was used to confirm the final
structure.
128
Chapter 5: Chemical looping reactivity of AB2O4 (A= Yb, Lu, Y
B= Fe, Mn)
5.1 Introduction
Chemical looping combustion is a cyclic process where a metal oxide is used
to deliver oxygen from an air chamber into a chamber of fuel for oxidation. It?s
an energy efficient method that reduces the energy penalty for CO2 capture as the
metal oxide act as an inherent filter, separating nitrogen and other gases to isolate
oxygen in air. Selecting the optimum OSM remain the most crucial challenge in
chemical looping applications. When designing a metal oxide system, the choice of
metal ion being used governs its overall performance. When a single metal oxide
system is used more often it fails to meet the requirement as an optimum OSM. For
instance CuO is exothermic during the redox reaction but it?s low melting point and
the tendency to agglomerate are major drawbacks. [173] While CoO has reasonable
reactivity high cost and toxicity makes it undesirable in an industrial set up. [174]
Fe2O3 is cost effective and environmentally friendly, however, has a low oxygen
storage capacity. [175] MnO is also less expensive and non-toxic but tend to to be
low in reactivity. Therefore, it makes sense to combine two or more of these metal
129
ions to get the maximum combined performance.
Creating mixed metal oxides by adding another metal ion to the structure can
often alter the existing structural and functional properties. This tunability is even
more feasible for ternary metal oxides due to their high tolerance for structural
modification withstanding degradation. When doping an existing metal site with
another metal few things should be taken into consideration to ensure the compati-
bility. Ionic radii is one such parameter. The ionic radii of the existing metal ion and
the new metal ion should be comparable so that it would not bring about extreme
alterations destabilizing the crystal structure. In a study done, alkali metal(Na,
K, Cs) doping of Fe2O3/Al2O3 system increased the prolonged reactivity over 50
cycles. [21] In a different study done by Wang et al, Zr doping increased oxygen
releasing capacity in Cu-based OSMs. [176] Mohamed et al reported the enhance-
ment of CO conversion in NiO/Al2O3 with Ce doping. [177] In another investigation,
Chen et al showed that tungsten based OSM when modified with Ni improved the
overall performance including oxygen availability, methane conversion and syngas
yield. [178] So it?s clear that elemental substitution is one significant way improve
and fine tune the performance of OSM as the composites delivers the synergistic
effect of all the components.
In this study, up until this point the reducing atmosphere that was used for the
chemical looping reaction was hydrogen. However, in order for the AFe2O4 series to
be used in a chemical looping reactor its reactivity towards methane should be tested
as that is the most commonly used fuel. The methane reactivity for this family of
materials has never been reported before. In our previous experiments in Chapter
130
3, we observed that the reduction rates were slower compared to that of oxidation.
Hence, ways to increase reduction kinetics were also explored. As the Fe-ion is the
redox active center it made sense to investigate elemental substitutions and doping
for Fe to improve the overall performance. Mn was chosen as the dopant as its
ionic radii is comparable to Fe. As the Yb analogue was the best overall performer,
YbFe2O4+? was chosen to synthesize the Mn-Fe solid solution series. Unsubstituted
Lu and Y compounds were also tested for the methane reactivity.
5.2 Experimental
5.2.1 Materials synthesis
All samples were prepared using solid-state reactions. 99.9% (Alfa Aesar) A2O3
(A = Yb, Lu,Y), 99% (Sigma Aldrich) B2O3 ( B= Fe, Mn) and 99.5% (Sigma
Aldrich) Fe powder were ground in an agate mortar and pestle in 1/2 : 5/6 : 1/3
ratio respectively to prepare 0.25 g of the target compound, and pressed into pellets
of 13 mm diameter. The pellets were placed in a 2 mL alumina crucible and sealed
inside evacuated 8 mm diameter quartz ampoules. Before sealing, the ampoules
were flushed with N2 gas and evacuated several times to obtain the required oxygen
partial pressure (PO2). [117] These compounds are highly sensitive to PO2, and
A2Fe3O7 often forms as an impurity if PO
-7
2 is not below approximately 10 Torr.
The samples were sintered at 1180 ?C for 12 hours and then quenched to room
temperature.
131
5.2.2 In situ TOF NPD on POWGEN at SNS
Time-of-flight (TOF) neutron powder diffraction (NPD) data were collected
at the Spallation Neutron Source (SNS) at the Oak Ridge National Laboratory
using the BL-11A POWGEN beamline.In situ ramping and cycling experiments
were carried out in the ILL furnace from RT to 600 ?C. A quartz basket was used as
the samples holder. Prior to the experiment blank diffraction patterns were collected
at every 100 ?C from RT to 600 ?C. Experiments were carried out by first heating the
samples under N 2 from RT to 600 ?C and then alternating the atmosphere between
20% O 2/N 2 (air) and 100% CH 4/N 2 (methane) with mass flow controllers while
holding the temperature at 600 ?C. 3-4 g of sample was used for each experiment to
compensate for low intensity of the neutron source.
5.2.3 In situ sXPD on 17 BM at APS
In situ sXPD experiments were performed in transmission geometry on the 17
BM beamline at the APS. A 2D PerkinElmer Si flat panel detector was used with an
average wavelength of 0.24108 A?. We collected diffraction patterns every 0.33 seconds
and integrated the diffraction images with GSAS-II [119] to yield patterns with a
Q-range of approximately 0.30 A?-1 to 6.6A?-1. A constant gas flow was maintained
at a rate of 15 mL/min through a quartz capillary sample holder which was loaded
with the sample, and a residual gas analyzer (RGA) was used to monitor the gaseous
products. Two types of experiments were carried out for all the A-site analogues.
First, we ramped the temperature of the sample environment up to 700 ?C under 20%
132
O2/He (air), holding at every 100 ?C for 20 minutes. We repeated this temperature
profile in pure helium to differentiate between temperature and oxidation driven
changes. Second, we performed cycling experiments to emulate a chemical looping
reactor, by switch- ing the atmosphere between 20% O2/He and 3.5% H2/He. The
samples were heated to 600 ?C in He and held at 600 ?C while cycling between the
two atmospheres. Automated sequential refinements were carried out using TOPAS
5 to study the lattice volume evolution with time. [120]
5.2.4 Elemental analysis from AES-ICP
Inductively Coupled Plasma-Atomic Emission Spectroscopy was carried out us-
ing a Shimadzu ICPE-9000 ICP Atomic Emission Spectrometer. First, 6-8 standards
were made for a calibration curve from a standard solution with the respective ele-
ments to be analyzed. The sample of interest is then dissolved in diluted ultra-high
purity acid solution such as 5% nitric acid. The instrument requires at least 10
mL of sample that is ideally within the 5 ppb to 10 ppm concentration range. The
standards and samples are then inserted into the sample chamber and ionized using
the inductively coupled plasma (2 mL at a time). Each measurement is repeated
three times to ensure precision. The intensity of the emitted characteristic radiation
from the sample is then standardized by the set of standards. The standards used
had the concentration series of 0.01 ppm, 0.1 ppm, 0.5 ppm, 2 ppm, 4 ppm, 6 ppm,
8 ppm, and 10 ppm.
133
5.2.5 Thermogravimetric analysis (TGA)
TGA was conducted using an SDT Q600 equipped with a TA Discovery MKS104-
S0212004 Micron Vision 2 Mass Spectrom- eter. The instrumentwas calibrated prior
to the experiment using alumina standards for each ramp rate. Ramp rate of 10
?C/min was used with a flow rate of 100mL/min. The samples were heated to 700
?C in air to allow for maximum weight gain. TGA experiments were only conducted
for the two end members of YbFe2-xMnxO4(x=0, 1) in order compare the effect of
Mn substitution.
5.3 Results
5.3.1 Material characterization
Rietveld refinements were done for all the four samples in the series YbFe2-xMnxO4
(x=0, 0.25, 0.75, 1) using the synchrotron XRD patterns collected at RT. All sam-
ples were pure phase except for the x=0.75 sample which had Yb2Fe3O7 as an
impurity but the impurity phase fraction accounted for less than 3%. There was
an anisotropy present for the Yb ion in the z-direction,hence anisotropic atomic
displacement paramteres were used for the refinement. The refined structural pa-
rameters are presented in table 5.1 and the plots are illustrated in Figure 5.1 and
5.2
134
YbFe2O4 R-3m a = 3.4559(3) A? c = 25.108(1) A?
Atom Site x y z Occ Beq (A?2) U11 U33
Yb 3a 0 0 0 1 0.31(5) 0.0006(7) 0.015(1)
Fe 6c 0 0 0.2147(1) 1 0.32(9)
O1 6c 0 0 0.1448(12) 1 11.01(94)
O2 6c 0 0 0.2896(9) 1 2.56(47)
YbFe1.75Mn0.25O4 R-3m a = 3.45765(6) A? c = 25.2374(7) A?
Atom Site x y z Occ Beq (A?2) U11 U33
Yb 3a 0 0 0 1 0.83(3) 0.0019(3) 0.0321(8)
Fe 6c 0 0 0.2155(1) 0.875 0.41(4)
Mn 6c 0 0 0.2155(1) 0.125 0.41(4)
O1 6c 0 0 0.1280(2) 1 0
O2 6c 0 0 0.2933(2) 1 0
YbFe1.25Mn0.75O4 R-3m a = 3.4561(1) A? c = 25.4959(6) A?
Atom Site x y z Occ Beq (A?2) U11 U33
Yb 3a 0 0 0 1 0.85(3) 0.0053(3) 0.0332(7)
Fe 6c 0 0 0.2156(1) 0.625 0.26(3)
Mn 6c 0 0 0.2156(1) 0.375 0.26(3)
O1 6c 0 0 0.1276(2) 0 0
O2 6c 0 0 0.2924(2) 0 0
YbFeMnO4 R-3m a = 3.45728(3) A? c = 25.61599(4) A?
Atom Site x y z Occ Beq (A?2) U11 U33
Yb 3a 0 0 0 1 0.85(3) 0.0039(3) 0.0294(6)
Fe 6c 0 0 0.21568(5) 0.5 2.43(3)
Mn 6c 0 0 0.21568(5) 0.5 2.43(3)
O1 6c 0 0 0.1276(2) 0.84(1) 0
O2 6c 0 0 0.2924(2) 0.60(1) 0
Table 5.1: Structural parameters from Rietveld refinement done using 17-BM sXPD
for YbFe2-xMnxO4 at 298 K (x=0, 0.25, 0.75, 1)
135
250000 YbFMnO4 (RT,R-3m)
Ycal
Yobs-Ycalc
Yobs
200000 Rwp = 8.79 %
150000
100000
50000
0
2 4 6 8 10 12 14
Two theta (Degrees, ? = 0.24108 ? )
Figure 5.1: The Observed, calculated, and difference curves of the YbFe2-xMnxO4,
x=0 (upper) x=0.25(lower) for at room temperature (RT) with synchrotron X-ray
powder diffraction patterns. A wavelength of 0.24108 A? was used.
5.3.2 TOF NPD experiments
The first experiments to test the methane reactivity were carried out on the
POWGEN beamline at the Oak Ridge national lab. The methane reactivity of the
YbFe2-xMnxO4 series has never been reported before. The preliminary experiments
were first carried out for the unsubstituted YbFe2O4. The samples were heated
to 600 ?C in an inert atmosphere and then held under 20 % O2 for 1 hour for
the oxidation half cycle. Next the sample holder was flushed with an inert gas
and then held under the reducing atmosphere of methane (100 %). As methane
reactivity was being tested for the first time, an abundant time period was given for
the reduction half cycle. Figure 5.3 illustrates the last powder diffraction pattern
136
Intensity (Arbitrary Units)
250000 YbFe1.25Mn0.75O4 (RT,R-3m)
Ycal
Yobs-Ycalc
Yobs
200000 Rwp = 7.71 %
150000
100000
50000
0
YbFeMnO  (RT,R-3m) Ycal250000 4 Yobs-Ycalc
Yobs
Rwp = 7.89 %
200000
150000
100000
50000
0
2 4 6 8 10 12 14
Two theta (Degrees, ? = 0.24108 ? )
Figure 5.2: The observed, calculated, and difference curves of the YbFe2-xMnxO4,
x=0.75(upper) x-1(lower) at room temperature (RT) with synchrotron X-ray powder
diffraction patterns. A wavelength of 0.24108 A? was used.
137
Intensity (Arbitrary Units)
R-3m
2 0 0 600 ?C in Methane 
Cycle
1 5 0 P-3 600 ?C in Air  2
R-3m 600 ?C in Methane
1 0 0
P-3 600 ?C in Air
Cycle
 1
5 0
RT
0
4 0 0 0 0 5 0 0 0 0 6 0 0 0 0
T im e  o f F lig h t (? s )
Figure 5.3: TOF NPD patterns of YbFe2O4 from cycling experiments carried out
on the POWGEN beamline at SNS.
from each half cycle. Under air as expected the R1 phase ( R3- m) gets oxidized
into the R3 phase ( P3-). And then under methane the pristine reduced phase
was successfully regenerated. Two consecutive cycles were done, and in each cycle
methane regenerated to the R1 phase.
5.3.3 Methane reactivity of other A-site analogues
The methane cycling experiments were also carried out for the Y and Lu ana-
logues ( 5.4). Both these compounds successfully regenerated under methane. In-
terestingly the Y analogues showed faster reaction rates than the Lu and Yb com-
pounds, similar to that was observed for hydrogen in chapter 3.
138
In te n s ity  (A rb itra ry  U n its )
YFe2O4
360
400
R-3m 320
350
280
300
240
250 P-3
200
200
150 R-3m
160
120
100
40
50 P-3
0
4.0 4.5 5.0 5.5 6.0
LuFe2O4
400 360
R-3m
350 320
300 280
250 P-3 240
200 200
150 R-3m 160
100 120
50 40P-3
0
4.0 4.5 5.0 5.5 6.0
2?, ?= 0.24108 ?
Figure 5.4: Cycling experiments done at 600 ?C under alternating atmospheres of
air and methane. Two cycles were done for each compound
139
Data sequence
Air Methane Air Methane Air Methane Air Methane
5.3.4 Cycling experiments of Mn substituted analogues
Cycling experiments emulating a chemical looping reactor were carried on the
17 BM beamline at the Argonne national lab. All reaction parameters such as the
gas flow rate, reaction time under each atmosphere were equally provided for all
samples. The amount of sample used in the capillary was approximated to be the
same. Figure 5.11 shows the comparison of cell volume evolution during the cycling
experiments done for the two end member of the series, YbFe2-xMnxO4(x=0,1). And
it?s evident that the Mn substitution has significantly increased the reduction kinet-
ics and attains the reduced cell volume at a faster rate. This is further confirmed
by the Rietveld refinement done for the in situ powder diffraction patterns in the
reduction cycle as illustrated in Figure 5.12. The last diffraction pattern at the
end of each 1 hour reduction cycle was used for the Rietveld refinement. In the
unsubstituted analogue (x=0) there is about 10% unreacted R3 phase left while
the Mn substituted analogues are completely regenerated under methane. This an
interesting outcome that will improve the overall reaction kinetics.
Cycling stability, which is another important metric as an OSM, was also as-
sessed in this experiment. For each analogue we carried out two full cycles. Figure
5.5 depicts the contour plots showing the diffraction peak evolution as a function of
the atmosphere at 600 ?C. It is evident that by the consistent intensity of the diffrac-
tion peaks none of the phases are degrading under either atmosphere. To further
confirm the phase stability Rietveld refinement was performed for the diffraction
pattern under methane for each cycle. The consistent lattice parameter values be-
140
c d
400 400
R-3m CH4 R-3m CH 244
300 300 18
P-3 Air P-3 Air
200 200 12
R-3m CH4 R-3m CH4
100 100 6
P-3 Air P-3 Air 0
a b
350
400
24
R-3m R-3m CHCH 44
250 300 18
P-3 Air
P-3 Air 200 12
150
R-3m R-3m CHCH 44
100 6
50 P-3 Air P-3 Air 0
2      3 4     5  6 2      3 4     5  6
2? (Degrees, ? = 0.24108 ?) 
Figure 5.5: Cycling experiments done at 600 ?C under alternating atmospheres of
air and methane. Two cycles were done for each compound.
141
Data sequence
Pristine structure Under air -10 min Under methane - 30 min
YbFe2O4 R1 - R3m YbFeMnO4140 R1 - R3m
120
Cycle 2
100 R3 - P3 R3 - P3 at 600 ?C 
80
60 R1 - R3m R1 - R3m
40 Cycle 1
at 600 ?C 
20 R3 - P3 R3 - P3
0
R1 - R3m R1 - R3m
2 4 6 8 10 2 4 6 8 10
2?  (?)
Figure 5.6: In situ sXRD patterns of YbFe2O4 and YbFeMnO4 at the end of each cy-
cle under alternating oxidizing (air) and reducing (methane) environments. 0.24108
A? was used.
142
In tensity (Arbitrary units)
Sample Element Nominal Experimental
x=0.25 Mn 0.25 0.25(1)
Fe 1.75 1.75 (1)
x=0.75 Mn 0.75 0.72(1)
Fe 1.25 1.28(1)
x=1 Mn 1 0.96(1)
Fe 1 1.01(1)
Table 5.2: ICP-AES results forYbFe2-xMnxO4
tween the two cycles confirms the cycling stability. However, reduction kinetics were
not the same.
5.3.5 Elemental analysis from AES-ICP
The results of this experiment are presented in Table 5.1. For each of the
considered elements (La, Sr, and Fe) the experimentally determined value was within
error of the targeted nominal composition.
5.3.6 Thermogravimetric analysis
Fig. 5.7 illustrates the TGA plots for AB2O4 (A= Yb, Lu, Y, In B= Fe, Mn).
Comparisons are being made for the two end members of the series YbFe2-xMnxO4(x=0,1).
And it is evident that the OSC of the Mn substituted compound is greater than
that of the unsubstituted analogue. This could be due to the fact that Mn has the
feasibility to obtain oxidation states greater than +3.
143
a
0 .6 6 0 0
0 .5 5 0 0
0 .4 4 0 0
0 .3 3 0 0
YbFe2O4
0 .2 YbFeMnO4 2 0 0
T e m p e ra tu re
0 .1 1 0 0
0 .0 0
5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0
T im e  (m in )
b
0 .6
L u F e O 7 0 02 4
Y b F e
0 .5 2
O 4 6 0 0
Y F e 2O 4
In F e 2O0 .4 4 5 0 0
4 0 0
0 .3
3 0 0
0 .2
5?C/min 2 0 0
0 .1 1 0 0
0 .0 0
5 0 1 0 0 1 5 0 2 0 0
Figure 5.7: Thermogravimetric analysis done comparing the substitution effects of
the B-site. The TGA plots of the A-site series from Chapter 3 is also shown for
comparison
144
O x y g e n  n o n -s to ic h e o m e try  (d )
O x y g e n  N o n - s t o i c h e o m e t r y  (? )
T e m p e ra tu re (?C ) T e m p e ra tu re (?C )
Yb C ycle 1
269 Yb C ycle 2
268 Lu Cycle 1
Er ro r
267
266
265
264
263
262
1 2 3 4 5
T im e (hour)
Figure 5.8: Cycling experiments carried out on POWGEN, SNS. The lower plot
illustrates the cell volume evolution under methane. Cell volume decreases due to
reduction and returns to its initial state
5.4 Discussion
5.4.1 Reactivity towards methane
In the 3rd chapter it was discussed how the unit cell volume expands under
air due o oxygen insertion and how the cell volume reduces under hydrogen due to
the removal of oxygen atoms. Figure 5.8 depicts the cell volume evolution of the
reduced phase under methane. It takes about 1.5 hrs for the reaction to complete
and obtain the final reduced cell volume. Similar to the experiments presented in
Chapter 3, we observed that the reduction rates were slower than that of oxidation.
However, the bigger sample sizes used in NPD are not technically ideal to grasp an
idea about the kinetics. Therefore, we also carried out the same experiments on the
17 BM beamline at the Argonne national lab. As a capillary sample holder is being
145
U nit ce ll vo lum e (? 3)
YbFe2O4
0.0 Methane
 0.1
 0.2
 0.3
 0.4
 0.5
 0.6
 0.7
Hydrogen
0.0
 0.2
 0.4
 0.6
 0.8
00                   260                 1420                1680               2840   
DTaitma es e(mquine)nce
Figure 5.9: Cell volume evolution under methane Vs hydrogen for YbFe2O4 at 600
?C
used the rapid data collection at APS is more suitable to gauge an understanding
about the reaction rates.
Figure 5.9 compares reduction rates in the two atmospheres; hydrogen and
methane for YbFe2O4. Interesting the cell volume evolution was faster in hydrogen
than that in methane. In this experiment diffraction patterns were collected every
20 seconds (the x-axis gives number of diffraction patterns). Under hydrogen it can
be seen within the first 6 minutes (20/3) the cell volume reduces rapidly and then
slowly increases and level up. In the case of methane, the reduction in cell volume is
146
Percentage volume change
clearly slower and reaches a plateau after about 30 min (80/3). In both atmosphere
the ultimate cell volume change is the same, which is about 0.7%.
5.4.2 Elemental substitution effects on kinetics
In many Fe-based oxygen carriers Mn is often a common dopant due to their
ionic radii compatibility. The Fe-Mn system has been explored a many times in
metal binary oxides mainly due to its favorable thermodynamics. [179] However
their behavior in ternary metal oxide system as ours was yet to be explored. In
a study done by Azimi et al a series of Mn-Fe ratios were reported which was
done to explore oxygen uncoupling properties of the metal oxide system. [180, 181]
They concluded that the mixed metal composition of Mn:Fe=2:1 exhibited the best
uncoupling behavior in terms of redox reactivity and stated that doping increased
the lattice oxygen mobility.
5.4.3 Elemental substitution effects on cycling stability
To confirm the cycling stability, the lattice parameters of the oxidized phase
at the end of each cycle was compared. The table 5.3 lists down these lattice
parameters and the Figures 5.13 and 5.14 depicts the Rietveld refinement done for
the plots of the oxidized phase at the end of each cycle. According to the resulting
lattice parameters, the values were quite consistent between the two consecutive
cycles, confirming the cycling stability. Furthermore, the diffraction patterns did
not show any additional phases due to to degradation (Fig. 5.13 and 5.14)
147
YbFe2O4 x103
140 27
24
120 21
18
100 R-3m 15
80 12
9
60 P-3 6
3
40 0
3
YbFeMnO x104
27
140
24
120 21
18
100 R-3m 15
12
80
9
6
60 P-3
3
40 0
4.5 5.0 5.5 6.0
2?, ?= 0.24108 ?
Figure 5.10: In situ sXRD patterns of YbFe2O4 and YbFeMnO4 at the end of
each cycle under alternating oxidizing (air) and reducing (methane) environments.
0.41458 A?A was used.
YbFe2 ? Cycle 1: Cycle 1: Cycle 2: Cycle 1:
xMnxO4 a c a c
x=0 3.4813(4) 25.232(5) 3.4723(2) 25.258(3)
x=0.25 3.4728(1) 25.376(1) 3.715(1) 25.394(1)
x=0.75 3.4708(1) 25.640(1) 3.4700(1) 25.654(1)
x=1 3.4727(1) 25.745(1) 3.4722(1) 25.751(1)
Table 5.3: Lattice parameters of the R-3m phase from the Rietveld refinement done
for the last powder pattern at the end of each 1 hour-cycle under methane at 600?C.
148
Data Sequence
Air Methane Air Methane
YbFe O
264.0 2 4e rro r
263.5
263.0
262.5
YbFeM nO
272 4error
271
270
269
0 5 10 15 20 25 30
Time (minutes)
Figure 5.11: The unit cell volume evolution of YbFe2-xMnxO4, x=0(upper)
x=1(lower) as a function of time. The data analyzed here is collected during the
cycling experiment where the material was heated to 600 ?C and then switched
between the two alternating atmospheres.
149
Cell volume (?3)
a YbFe2O4  (P-3)= 88.35% Ycal
300000 YbFe2O4 + d (P-3)= 12.57% Yobs-Ycalc
Yobs
200000 Rwp = 10.13 %
100000
0
2 3 4 5 6 7 8 9 10
250000
b YbFe1 .75Mn0 .25O4  (P-3)= 100% Ycal
200000 Yobs-Ycalc
Yobs
150000
Rwp = 13.41 %
100000
50000
0
?50000
2 3 4 5 6 7 8 9 10
c YbFe1 .25Mn0 .75O4  (P-3)= 100% Ycal
200000 Yobs-Ycalc
Yobs
150000
Rwp = 8.51 %
100000
50000
0
2 3 4 5 6 7 8 9 10
250000
d YbFeMnO4  (P-3)= 100 % Ycal
200000 Yobs-Ycalc
Yobs
150000 Rwp = 8.22 %
100000
50000
0
?50000
2 3 4 5 6 7 8 9 10
2?, ?= 0.24108 ?
Figure 5.12: Rietveld refinements for the last pattern in the reduction half cycle
under methane at 600?C. YbFe2-xMnxO4, a) x=0 b) x=0.25 c=0.75 d= x=1. All
analogues were fully regenerated by the end of the half cycle except for x=0, which
had 10% of the unconverted oxidized phase
150
Intensity
7000
YbFe2O4 +?(RT,P-3) Ycal
6000 Yobs-YcalcYobs
Rwp = 10.13 %
5000
4000
3000
2000
1000
0
3000
YbFe1.75Mn0.25O4+ ?(RT,P-3) YcalYobs-Ycalc
2500 Yobs
Rwp = 13.02 %
2000
1500
1000
500
0
?500
2 4 6 8 10 12 14
2?, ?= 0.24108 ?
Figure 5.13: Rietveld refinement for the oxidized phases at 600?C. YbFe2-xMnxO4,
x=0 (upper) x= 0.25 (lower)
151
Intensity
14000
YbFe1.25Mn0.75O4+ ?(RT,P-3) YcalYobs-Ycalc
12000 Yobs
Rwp = 12.97 %
10000
8000
6000
4000
2000
0
14000 YbFeMnO4+ ?( (RT,P-3) Ycal
Yobs-Ycalc
12000 YobsRwp = 7.72 %
10000
8000
6000
4000
2000
0
2 4 6 8 10 12 14
2?, ?= 0.24108 ?
Figure 5.14: Rietveld refinement for the oxidized phases at 600?C. YbFe2-xMnxO4,
x=0.75 (upper) x=1 (lower)
152
Intensity
5.5 Conclusion
Chemical looping reactivity of AB2O4 (A= Yb, Lu, Y B= Fe, Mn) was explored
by conducting cycling experiments at 600 ?C under alternating oxidizing(air) and
reducing atmosphere(methane). TGA experiments were also carried to deduce the
oxygen storage capacity. Both the reaction kinetics and the oxygen storage capacity
were increased with Mn substitution. ICP-AES experiments confirmed the exper-
imental Mn, Fe ratios. The AFe2O4 (A= Lu, Y) analogues were also successfully
regenerated under methane. The results in this chapter are not-conclusive. TOF-
NPD experiments are scheduled to be carried out at SNS. To determine if there is a
linear dependence of Mn substitution on the rates o reactions, further analysis and
sequential refinements need to be carried out. Bond valence sum calculations are
required to compare the Mn/Fe-O bonds to see if it has an effect on the increased
reaction rates. OSC calculations need to be carried out as well with more TGA
experiments for all the analogues.
153
Chapter 6: Conclusions and future directions
In this research work we explored the oxygen storage properties of the novel
metal oxide system, AB2O4 (A= Lu, Yb, Y, In and = Fe, Mn), for the potential ap-
plication in chemical looping reactions. None of these materials have been explored
for its reactivity with methane before. The research work presented in this thesis
can be divided into three main studies. The following list down the main findings
that were made from our experiments.
6.1 Study I
We studies 4 A-site analogues of AB2O4 for the first time. These AB2O4 layered
oxides exhibit reversible oxygen insertion at 600 ?C in alternating 20% O2/He and
3.5% H2. Oxidation kinetics were governed by the A-site cationic radii. The greater
the ionic radii the greater the deviation of the bond valence sum of Fe-O bonds were.
Lu, Yb and Y showed stable cycling while In degraded under H2. OSCs calculated
based on TGA vary in the range of 2.20?3.13 wt% at 600 ?C, and showed no clear
trend with the size of A-cation.
154
6.2 Study II
The structure of the oxidized phase was solved from with simulated annealing
and Fourier difference maps. Structural parameters were reported with combine
neutron and X-ray Rietveld refinement. PDF and XAS were used to confirm the
final structural model.
6.3 Study III
Studied the chemical looping reactivity of AB2O4 layered oxides, with B=Mn,
Fe. Reported reactivity with methane of AB2O4 layered oxides for the first time.
The RT pristine structure was regenerated at 600 ?C under methane. Mn substituted
compounds exhibited faster kinetics and also higher oxygen storage capacities.
We conclude that the layered, ternary metal oxide system, AB2O4, is a suitable
candidate as an oxygen storage material for the potential application in chemical
looping reactions.
6.4 Future directions
6.4.1 Active oxygen species and product selectivity
Further experiments should be carried out to explore the active oxygen species
and product selectivity of these systems. The types of oxygen species in metal oxides
that take part in chemical looping reactions are mainly of two forms : 1)weakly
bonded lattice oxygen that will get released easily and cause complete oxidization
155
of a fuel. 2) strongly bound lattice oxygen that partially oxidizes a fuel (chemical
reforming), typically at higher temperatures. There are several ways to characterize
what type of reactive oxygen is present in the metal oxide. Dai et al discusses
how this can be done through XPS. [37] Some materials may possess both types of
oxygen species. For the YbFe2O4 family the product selectivity is yet to be tested.
Certain preliminary tests were done in an emulated fixed bed reactor. Fig. 6.1 shows
the product analysis during a chemical looping reaction. According to the plot it
does complete combustion as it gives CO 2 under methane half cycle. However,
the it deactivates after several cycles most probably due to coke formation. At the
end of this experiment, a black residue was observed which strongly suggest coke
formation. So new experiments should be carried on ways to reduce coke formation.
A combination avenues could be tried such as optimum packing density, lowering
flow rate of methane and increasing particle size to improve reactive surface area
etc. Also further experiments should be carried to confirm the purity of the product
stream to ensure product selectivity.
6.4.2 Improving overall performance as an OSM
More elemental substitutions should be carried out for this family of materials
and be tested for oxygen carrying capabilities. To the best of our knowledge AB2O4,
A= Lu, Yb and B = Fe are the only analogues that have been reported with regard
to oxygen storage studies (apart from our own study . [129]). Even in our studies
more experiments should be carried out for for the Y analogue. As it showed the
156
7 0
6 0
5 0
4 0
3 0
2 0
1 0
0
0 .0 2 .5 5 .0 7 .5 1 0 .0 1 2 .5 1 5 .0 1 7 .5
T im e  (h r)
N 2/C O -m in o r O /C H 4 N 2 / C O N /C H 2 A r
C O2 O 2 H 2 C (C O  o r C O 2) H 2 O
Ar CH4 Ar Air Ar CH4 Ar Air Ar CH4 Ar Air Ar CH4 Ar Air2 .0 0
1 .7 5
1 .5 0
1 .2 5
1 .0 0
0 .7 5
0 .5 0
0 .2 5
0 .0 0
0 1 2 3 4 5 6
T im e  (h r)
Figure 6.1: Chemical looping reaction of YbFe2O4 in CH4
157
OSC (O2  mmol/g) OSC (O2  mmol/g)
fastest oxidation kinetics, exploring it?s suitability as an oxygen sensor at lower
temperature is worth looking into. However, in our experiments the Y analogues
did not get regenerated at 500 ?C(6.2) under methane. Maybe Mn substitutions for
the Y analogue could activate it at lower temperatures. The following list down
some of the features that should be explored in future to improve the performance
of AB2O4as an OSM.
- Increase oxygen storage capacity and reactivity towards methane
- Increasing surface area
- Determine the efficiency of methane conversion
- Improve long term cycling stability
- Trying lower cycling temperatures for the Mn substituted analogues
158
140 135
CH4
105
100 Air
75
60 CH4
45
P-3 Air
20 15
36
350 R-3m CH4 28
250 P-3 Air 20
150 R-3m CH 124
50 P-3 Air 4
3.5      4    4.5   5    5.5
2?
Figure 6.2: Methane reactivity rested forYFe2O4 at 500?C and 600 ?C
159
Data sequence
Appendix A: Resources
A.1 Website resources
Neutron scattering lengths and cross sections https://www.ncnr.nist.gov/
resources/n-lengths/
Space group symmetry and Wycoff postions http://www.cryst.ehu.es/
Configuration of Jedit with TOPAS https://community.dur.ac.uk/john.evans/
topas_academic/jedit_setup.htm
Online tutorials for TOPAS in launch mode https://community.dur.ac.uk/
john.evans/topas_academic/topas_main.htm http://topas.dur.ac.uk/topaswiki/
doku.php
International crystallography tables https://it.iucr.org/
Latest updates about beamlines at APS https://www.aps.anl.gov/Machine-Status
Shannon radii http://abulafia.mt.ic.ac.uk/shannon/ptable.php
Bond valence parameters https://www.iucr.org/resources/data/data-sets/
bond-valence-parameters
Free structure solution software https://www.ccdc.cam.ac.uk/solutions/csd-materials/
components/dash/
Rietveld mailing list http://www.ccp14.ac.uk/ccp/web-mirrors/hugorietveld/
160
stxnews/riet/intro.htm
A.2 TOPAS macro used in this research
A.2.1 TOPAS launch mode
Everything that you can do in the GUI and more are possible in launch mode
including -Structure determination (simulated annealing; Fourier analysis)
-Anisotropic refinement models
-Magnetic structure refinement
-Single crystal data
-Constrained phase/ elemental composition
- Introduce your own peak-shape functions and convolutions
-Rigid and soft bodies with all parameters refineable
- Introduce user-defined equations and macros
Please refer to the TOPAs technical manual and John Evans online website for
more information
A.2.2 simulated annealing
1. Perform a Pawley fit for synchrotron X-ray data.
2. Add atomic sites and coordinates and set to zero, set reasonable values for occu-
pancy and ADPs. The positions of heavy elements included first. (lighter elements
161
Figure A.1: Input file in Jedit used in launch mode for TOPAS
like oxygen commented out)
3. Fix all parameters except scale and atomic coordinates.-
4. Include macros for simulated annealing.-(A.2)
5. Run simulated annealing.
162
Auto_T(10) 
r_wp  11.503 r_exp  5.979 r_p  7.837 r_wp_dash  17.193 r_p_dash  13.292 r_exp_dash  8.936 weighted_Durbin_Watson  0.258 
gof  1.924 
chi2_convergence_criteria 0.001 
do_errors 
xdd 11bmb_2147_AA0033_LuFeO_Ox.xye 'add your file and location 
bkg   86.6164399`_0.195360967  31.7396275`_0.356451355 -35.469626`_0.27495905 -863291839`_0.238974431' 
Zero_Error(, 0.00106`_0.00010) 
LP_Factor( 90) 
start_X 2        finish_X 34 
 Rp 9999  Rs 1000    
x_calculation_step 0.0001 
Simple_Axial_Model( , 3.87323`_0.11549) 
Lam  ymin_on_ymax  0.0005 
la  1 lo  0.41458 lh  0.475828 
str 
space_group "P-3" 
 Hexagonal(!lpa  3.47140`_0.00003, !lpc  8.44756`_0.00009)'lattice parameters fixed after the initial Lebail fit 
 cell_volume  88.159932`_0.001587149 
 'CS_G(crysG, 98.87492_0.29001) 
 prm !u  0.09732`_0.00485     prm !v  0.14727`_0.00188   prm !w  0.00306`_0.00018 
peak_type pv pv_lor   0.481182586`_0.0026196415 pv_fwhm = u * ((2 Th)^2) + v * (2 Th) + w; 'PSEUDO VOIGT 
                                   site Lu num_posns 0 occ Lu 0.5 beq  1 x @ 0 y @ 0 z @ 0 
  site Fe num_posns 0 occ Fe 0.5 beq  1 x @ 0 y @ 0 z @ 0 
  site O1 num_posns 0 occ O  0.5 beq  1 x @ 0 y @ 0 z @ 0 
  site O2 num_posns 0 occ O  0.5 beq  1 x @ 0 y @ 0 z @ 0 
  site O3 num_posns 0 occ O  0.5 beq  1 x @ 0 y @ 0 z @ 0 
  site O4 num_posns 0 occ O  0.5 beq  1 x @ 0 y @ 0 z @ 0 
occ_merge Lu*  occ_merge_radius .6   
 occ_merge O*  occ_merge_radius .6 
   occ_merge Fe* occ_merge_radius .6 
   
view_structure   normalize_FCs  append_fractional 
Figure A.2: Simulated Annealing macro
163
Appendix B: History of scientists who contributed to crystallogra-
phy
164
Scientist Discovery Year
1.Francesco Coined the word diffraction. Observed light 17th
Maria Grimaldi going through a pinhole to form a cone century
2.Wilhelm Con- Discovery of X-rays 1895
rad Rontgen
3.Max Von Laue Discovered that X-rays are diffracted by 1912
atoms
4.Walter Carried out first recorded powder diffraction 1912
Friedrich and experiment by crushing a CuSo4 single crys-
Paul Knipping tal
5.Lawrence and Bragg?s Law. Figured out how to determine 1912
Henry William crystal structure from X-ray diffraction
Bragg
6.Paul Knipping Diffraction rings around the primary beam 1913
were observed for the first time. Experiment
carried out using Beeswax
7.Peter De- They explained different hkl?s can contribute 1916
bye and Paul to the same observed reflection (multiplici-
Scherrer ties). They successfully deconvoluted the dif-
ferent intensities of the LiF diffraction rings.
Table B.1: The scientists who contributed to furthering diffraction
165
Scientist Discovery Year
8.Albert Wallace Solved the crystal structure of Fe using pow- 1917,
Hull der diffraction. Provided a very similar 1919
experiment like Debye and Scherrer pretty
much giving the same explanation simulta-
neously. He used Mo radiation. First person
to use the Zr filter as a monochromator. First
person to carry out in situ powder diffraction
experiments. Measure Fe at liquid air tem-
perature, at RT and at 1273 K. Proposed
the phenomenon of identifying chemical sub-
stances using powder diffraction data. Even
for mixtures.
9.Eugen Simon 1st crystal structure of a solidified gas Ar 19
and Clara Von crystal structure and HCl cubic crystal struc-
Simon ture
10.Johannes Introduced Bragg-Brentano geometry. Dis- Around
Christian tance tube-sample and sample-detector slit 1920
Michael is the same.
Brentano
Table B.2: The scientists who contributed to furthering diffraction
166
Scientist Discovery Year
11.William Parrish Published an article on constructing a 1949
diffractometer, which he had previously
patented.
12.Joseph Donald Formulated the first powder diffraction 1936-38
Hanawalt, Harold W. database Published an article in which they
Rinn,Ludo K. Frevel supplied all intensity ratios and d-spacings of
most diffraction lines for 1000 compounds
13.William Houlder First person determine the crystal structure 1944
Zachariasen of a non-cubic binary system; UCl3
14.Hugo Rietveld Proposed the Rietveld method for powder 1967
diffraction data analysis
15.Pawley Pawley method for whole powder pattern de- 1981
composition
16.Bob Vondrele et al Application of Rietveld method with TOF 1982
neutron data
17. Cox et al Application of Rietveld method with syn- 1983
chrotron X-rays
18.Armel Lebail Lebail method for whole powder pattern de- 1987
composition
Table B.3: The scientists who contributed to furthering diffraction
167
Bibliography
[1] Siwei Luo, Liang Zeng, and Liang-Shih Fan. Chemical Looping Technology:
Oxygen Carrier Characteristics. Annu. Rev. Chem. Biomol. Eng., 6(1):53?75,
jul 2015.
[2] L B Mccusker, R B Von Dreele, D E Cox, D Loue E? R D, and P Scardi.
Rietveld refnement guidelines. Int. Union Crystallogr. J. Appl. Crystallogr. J.
Appl. Cryst, 32:36?50, 1999.
[3] Takeshi Egami and Simon Billinge. Chapter 3 The method of total scattering
and atomic pair distribution function analysis. In Underneath Bragg Peaks
Struct. Anal. Complex Mater., volume 16, chapter 3, pages 55?99. Elsevier,
2003.
[4] Ping He, Haijun Yu, De Li, and Haoshen Zhou. Layered lithium transition
metal oxide cathodes towards high energy lithium-ion batteries. J. Mater.
Chem., 22(9):3680, 2012.
[5] Guoping Wang, Lei Zhang, and Jiujun Zhang. A review of electrode materials
for electrochemical supercapacitors. Chem. Soc. Rev., 41(2):797?828, 2012.
[6] A Boudghene Stambouli and E Traversa. Solid Oxide Fuel Cells (SOFCs): A
Review of an Environmentally clean and efficient source of energy, 2002.
[7] Ana B. Mun?oz-Garc??a, Andrew M. Ritzmann, Michele Pavone, John A. Keith,
and Emily A. Carter. Oxygen Transport in Perovskite-Type Solid Oxide
Fuel Cell Materials: Insights from Quantum Mechanics. Acc. Chem. Res.,
47(11):3340?3348, nov 2014.
[8] Zulma Liliana Moreno Botello, Alberto Caneiro, Pascal Roussel, and Gilles
Gauthier. Synthesis and preliminary study of pure and Zr-doped YMnO3
compounds as Solid Oxide Fuel Cells electrode. J. Alloys Compd., 690:348?
355, jan 2017.
168
[9] Adam G. Jolley, Rishvi Jayathilake, and Eric D. Wachsman. Optimizing rhom-
bohedral Bi2O3 conductivity for low temperature SOFC electrolytes. Ionics
(Kiel)., 25(8):3531?3536, aug 2019.
[10] Sven Ru?hle, Assaf Y. Anderson, Hannah Noa Barad, Benjamin Kupfer, Yaniv
Bouhadana, Eli Rosh-Hodesh, and Arie Zaban. All-oxide photovoltaics. J.
Phys. Chem. Lett., 3(24):3755?3764, 2012.
[11] Prashant V. Kamat. TiO 2 Nanostructures: Recent Physical Chemistry Ad-
vances. J. Phys. Chem. C, 116(22):11849?11851, jun 2012.
[12] Juan A Anta, Elena Guille?n, and Ramo?n Tena-Zaera. ZnO-Based Dye-
Sensitized Solar Cells. J. Phys. Chem. C, 116(21):11413?11425, may 2012.
[13] S. Ferrere, A. Zaban, and B. A. Gregg. ChemInform Abstract: Dye Sensiti-
zation of Nanocrystalline Tin Oxide by Perylene Derivatives. ChemInform,
28(39):no?no, aug 2010.
[14] Shelly Burnside, Jacques-E. Moser, Keith Brooks, Michael Gra?tzel, and David
Cahen. Nanocrystalline Mesoporous Strontium Titanate as Photoelectrode
Material for Photosensitized Solar Devices: Increasing Photovoltage through
Flatband Potential Engineering. J. Phys. Chem. B, 103(43):9328?9332, oct
1999.
[15] Frank Lenzmann, Jessica Krueger, Shelly Burnside, Keith Brooks, Michael
Gra?tzel, Doron Gal, Sven Ru?hle, and David Cahen. Surface Photovoltage
Spectroscopy of Dye-Sensitized Solar Cells with TiO 2 , Nb 2 O 5 , and SrTiO
3 Nanocrystalline Photoanodes: Indication for Electron Injection from Higher
Excited Dye States. J. Phys. Chem. B, 105(27):6347?6352, jul 2001.
[16] Gongming Wang, Yi Yang, Dongdong Han, and Yat Li. Oxygen defective
metal oxides for energy conversion and storage, 2017.
[17] J. Ada?nez, L. F. de Diego, F. Garc??a-Labiano, P. Gaya?n, A. Abad, and J. M.
Palacios. Selection of Oxygen Carriers for Chemical-Looping Combustion.
Energy & Fuels, 18(2):371?377, mar 2004.
[18] Horst J. Richter and Karl F. Knoche. Reversibility of Combustion Processes.
pages 71?85. nov 1983.
[19] Anders Lyngfelt and Tobias Mattisson. Synthesis gas generation by chemical-
looping reforming in a continuously operating laboratory reactor. Fuel,
85:1631?1641, 2006.
[20] Y Zeng, S Tamhankar, N Ramprasad, F Fitch, D Acharya, and R Wolf. A novel
cyclic process for synthesis gas production. Chem. Eng. Sci., 58(3-6):577?582,
feb 2003.
169
[21] Lu Liu and Michael R. Zachariah. Enhanced Performance of Alkali Metal
Doped Fe 2 O 3 and Fe 2 O 3 /Al 2 O 3 Composites As Oxygen Carrier
Material in Chemical Looping Combustion. Energy & Fuels, 27(8):4977?4983,
aug 2013.
[22] Juan Adanez, Alberto Abad, Francisco Garcia-Labiano, Pilar Gayan, and
Luis F. de Diego. Progress in Chemical-Looping Combustion and Reform-
ing technologies. Prog. Energy Combust. Sci., 38(2):215?282, apr 2012.
[23] Lori Nalbandian, Antigoni Evdou, and Vassilis Zaspalis. La1-xSrxMyFe1-yO3-
? perovskites as oxygen-carrier materials for chemical-looping reforming. Int.
J. Hydrogen Energy, 36(11):6657?6670, jun 2011.
[24] A. Evdou, V. Zaspalis, and L. Nalbandian. La1-xSrxFeO3-? perovskites as
redox materials for application in a membrane reactor for simultaneous pro-
duction of pure hydrogen and synthesis gas. Fuel, 89(6):1265?1273, jun 2010.
[25] Fernando Fresno, Roc??o Ferna?ndez-Saavedra, M. Bele?n Go?mez-Mancebo, Al-
fonso Vidal, Miguel Sa?nchez, M. Isabel Rucandio, Alberto J. Quejido, and
Manuel Romero. Solar hydrogen production by two-step thermochemical cy-
cles: Evaluation of the activity of commercial ferrites. Int. J. Hydrogen Energy,
34(7):2918?2924, 2009.
[26] Alberto Abad, Juan Ada?nez, Francisco Garc??a-Labiano, Luis F. de Diego, Pilar
Gaya?n, and Javier Celaya. Mapping of the range of operational conditions for
Cu-, Fe-, and Ni-based oxygen carriers in chemical-looping combustion. Chem.
Eng. Sci., 62(1-2):533?549, 2007.
[27] Kiyoshi Otsuka, Ye Wang, Eiyuh Sunada, and Ichiro Yamanaka. Direct Par-
tial Oxidation of Methane to Synthesis Gas by Cerium Oxide. J. Catal.,
175(2):152?160, apr 1998.
[28] Kiyoshi Otsuka, Ye Wang, and Masaki Nakamura. Direct Conversion of
Methane to Synthesis Gas through Gas-Solid Reaction Using CeO2-ZrO2 Solid
Solution at Moderate Temperature. Appl. Catal. A Gen., 183(2):317?324,
1999.
[29] Vladislav A. Sadykov, Nathalia N. Sazonova, Aleksei S. Bobin, Vitalii S.
Muzykantov, Elena L. Gubanova, Galina M. Alikina, Anton I. Lukashe-
vich, Vladimir A. Rogov, Eugenia N. Ermakova, Ekaterina M. Sadovskaya,
Nathalia V. Mezentseva, Ekaterina G. Zevak, Sergei A. Veniaminov, Martin
Muhler, Claude Mirodatos, Yves Schuurman, and Andre C. Van Veen. Partial
oxidation of methane on Pt-supported lanthanide doped ceria-zirconia oxides:
Effect of the surface/lattice oxygen mobility on catalytic performance. Catal.
Today, 169(1):125?137, 2011.
170
[30] Hong Lv, Heng yong Tu, Bin yuan Zhao, Yu ji Wu, and Ke ao Hu. Synthesis
and electrochemical behavior of Ce1-xFexO2-?? as a possible SOFC anode
materials. Solid State Ionics, 177(39-40):3467?3472, 2007.
[31] Jin Yong Luo, Ming Meng, Jin Song Yao, Xin Gang Li, Yu Qing Zha, Xitao
Wang, and Tian Yong Zhang. One-step synthesis of nanostructured Pd-doped
mixed oxides MOx-CeO2 (M = Mn, Fe, Co, Ni, Cu) for efficient CO and C3H8
total oxidation. Appl. Catal. B Environ., 87(1-2):92?103, 2009.
[32] S Fukuda, T Hino, and T Y A S Ina. Desorption processes of hydrogen and
methane from clean and metal-deposited graphite irradiated by hydrogen ions.
J. Nucl. Mater. 1, 164:997?1003, 1989.
[33] Kongzhai Li, Hua Wang, Yonggang Wei, and Dongxia Yan. Direct conversion
of methane to synthesis gas using lattice oxygen of CeO2?Fe2O3 complex
oxides. Chem. Eng. J., 156(3):512?518, feb 2010.
[34] Daniel D Taylor, Nathaniel J Schreiber, Benjamin D Levitas, Wenqian Xu,
Pamela S Whit, and Efrain E Rodriguez. Oxygen Storage Properties of La 1
- x Sr x FeO 3 - ? for Chemical-Looping Reactions - An In Situ Neutron and
Synchrotron X - ray Study. Chem. Mater., 28:3951?3960, 2016.
[35] Jan Kas? and Paolo Fornasiero. Nanostructured materials for advanced au-
tomotive de-pollution catalysts. J. Solid State Chem., 171:19?29, 2003.
[36] M Karppinen, H Yamauchi, S Otani, T Fujita, T Motohashi, Y.-H. Huang,
M. Valkeapa?a?, and H. Fjellv?ag. Oxygen Nonstoichiometry in YBaCo 4 O 7+ ?
: Large Low-Temperature Oxygen Absorption/Desorption Capability. Chem.
Mater., 18(2):490?494, jan 2006.
[37] Xiao Ping Dai, Ran Jia Li, Chang Chun Yu, and Zheng Ping Hao. Unsteady-
State Direct Partial Oxidation of Methane to Synthesis Gas in a Fixed-Bed
Reactor Using AFeO 3 (A = La, Nd, Eu) Perovskite-Type Oxides as Oxygen
Storage. J. Phys. Chem. B, 110(45):22525?22531, nov 2006.
[38] Alexander A. Greish, Lev M. Glukhov, Elena D. Finashina, Leonid M. Kustov,
Jae-Suk Sung, Ko-Yeon Choo, and Tae-Hwan Kim. Oxidative coupling of
methane in the redox cyclic mode over the catalysts on the basis of CeO2 and
La2O3. Mendeleev Commun., 20(1):28?30, jan 2010.
[39] Daniel D Taylor, Nathaniel J Schreiber, Benjamin D Levitas, Wenqian Xu,
Pamela S. Whitfield, and Efrain E Rodriguez. Oxygen Storage Properties of
La 1? x Sr x FeO 3-? for Chemical-Looping Reactions?An In Situ Neutron
and Synchrotron X-ray Study. Chem. Mater., 28(11):3951?3960, jun 2016.
[40] N Kiniizuka. A SERIES OF NEW COMPOUNDS A3+ Fe2O4 ( A = Ho , Er
, Tm , Yb , AND Lu ). Solid State Commun., 15:1321?1323, 1974.
171
[41] ? Evrard O Tanni?eres N and Aubry J. No Title. R. Acad. Sci. Paris, 278:241,
1974.
[42] Tobias Mattisson, Henrik Leion, and Anders Lyngfelt. Chemical-looping with
oxygen uncoupling using CuO/ZrO2 with petroleum coke. Fuel, 88(4):683?
690, apr 2009.
[43] Tobias Mattisson, Anders Lyngfelt, and Henrik Leion. Chemical-looping with
oxygen uncoupling for combustion of solid fuels. Int. J. Greenh. Gas Control,
3(1):11?19, jan 2009.
[44] Peter Hallberg, Malin Ka?lle?n, Dazheng Jing, Frans Snijkers, Jasper van Noyen,
Magnus Ryde?n, and Anders Lyngfelt. Experimental Investigation of CaMnO3-
d Based Oxygen Carriers Used in Continuous Chemical-Looping Combustion.
Int. J. Chem. Eng., 2014(1):1?9, 2014.
[45] Arya Shafiefarhood, Amy Stewart, and Fanxing Li. Iron-containing mixed-
oxide composites as oxygen carriers for Chemical Looping with Oxygen Un-
coupling (CLOU). Fuel, 139:1?10, jan 2015.
[46] Zhongliang Yu, Yanyan Yang, Song Yang, Qian Zhang, Jiantao Zhao, Yitian
Fang, Xiaogang Hao, and Guoqing Guan. Iron-based oxygen carriers in chem-
ical looping conversions: A review. Carbon Resour. Convers., 2(1):23?34, apr
2019.
[47] Zhuo Cheng, Lang Qin, Jonathan A. Fan, and Liang-Shih Fan. New Insight
into the Development of Oxygen Carrier Materials for Chemical Looping Sys-
tems. Engineering, 4(3):343?351, jun 2018.
[48] Xizheng Wang, Rishvi Jayathilake, Daniel D. Taylor, Efrain E. Rodriguez,
and Michael R. Zachariah. Study of C/Doped ?-Bi 2 O 3 Redox Reactions by
in Operando Synchrotron X-ray Diffraction: Bond Energy/Oxygen Vacancy
and Reaction Kinetics Relationships. J. Phys. Chem. C, 122(16):8796?8803,
apr 2018.
[49] Zongxian Yang, Tom K. Woo, Micael Baudin, and Kersti Hermansson. Atomic
and electronic structure of unreduced and reduced CeO2 surfaces: A first-
principles study. J. Chem. Phys., 120(16):7741?7749, apr 2004.
[50] Luis F. de Diego, Mar??a Ortiz, Francisco Garc??a-Labiano, Juan Ada?nez, Al-
berto Abad, and Pilar Gaya?n. Hydrogen production by chemical-looping re-
forming in a circulating fluidized bed reactor using Ni-based oxygen carriers.
J. Power Sources, 192(1):27?34, jul 2009.
[51] Jean-Luc Dubois and Charles J. Cameron. Common features of oxidative
coupling of methane cofeed catalysts. Appl. Catal., 67(1):49?71, jan 1990.
172
[52] Gargi Dutta, Umesh V Waghmare, Tinku Baidya, M S Hegde, K R Priolkar,
and P R Sarode. Origin of Enhanced Reducibility/Oxygen Storage Capacity
of Ce 1 - x Ti x O 2 Compared to CeO 2 or TiO 2. Chem. Mater., 18(14):3249?
3256, jul 2006.
[53] Daniel D. Taylor, Nathaniel J. Schreiber, Benjamin D. Levitas, Wenqian Xu,
Pamela S. Whitfield, and Efrain E. Rodriguez. Oxygen Storage Properties
of La1-xSrxFeO3-?for Chemical-Looping Reactions - An in Situ Neutron and
Synchrotron X-ray Study. Chem. Mater., 28(11):3951?3960, 2016.
[54] Masato Machida, Kiyotaka Kawamura, Kazuhiro Ito, and Keita Ikeue. Large-
Capacity Oxygen Storage by Lanthanide Oxysulfate/Oxysulfide Systems.
Chem. Mater., 17(6):1487?1492, mar 2005.
[55] Paul Cho, Tobias Mattisson, and Anders Lyngfelt. Carbon Formation on
Nickel and Iron Oxide-Containing Oxygen Carriers for Chemical-Looping
Combustion. Ind. Eng. Chem. Res., 44(4):668?676, feb 2005.
[56] Luis F. de Diego, Mar??a Ortiz, Juan Ada?nez, Francisco Garc??a-Labiano, Al-
berto Abad, and Pilar Gaya?n. Synthesis gas generation by chemical-looping
reforming in a batch fluidized bed reactor using Ni-based oxygen carriers.
Chem. Eng. J., 144(2):289?298, oct 2008.
[57] Marcus Johansson, T. Mattisson, and A. Lyngfelt. Investigation of Mn3O4
with stabilized ZrO2 for chemical-looping combustion, 2006.
[58] Magnus Ryde?n, Dazheng Jing, Malin Ka?lle?n, Henrik Leion, Anders Lyngfelt,
and Tobias Mattisson. CuO-Based Oxygen-Carrier Particles for Chemical-
Looping with Oxygen Uncoupling ? Experiments in Batch Reactor and in
Continuous Operation. Ind. Eng. Chem. Res., 53(15):6255?6267, apr 2014.
[59] Hongsheng Chen, Zhong Zheng, Zhiwei Chen, and Xiaotao T. Bi. Reduction
of hematite (Fe 2 O 3 ) to metallic iron (Fe) by CO in a micro fluidized bed
reaction analyzer: A multistep kinetics study. Powder Technol., 316:410?420,
jul 2017.
[60] Mohammad M Hossain and Hugo I. de Lasa. Chemical-looping combustion
(CLC) for inherent CO2 separations. Chem. Eng. Sci., 63(18):4433?4451, sep
2008.
[61] Antigoni Evdou, Vassilios Zaspalis, and Lori Nalbandian. Ferrites as redox
catalysts for chemical looping processes. Fuel, 165:367?378, 2016.
[62] C N R Rao, A Sundaresan, and Rana Saha. Multiferroic and Magnetoelectric
Oxides: The Emerging Scenario. J. Phys. Chem. Lett., 3(16):2237?2246, aug
2012.
173
[63] M Hervieu, A Guesdon, J Bourgeois, E Elka??m, M Poienar, F Damay, J Rou-
quette, A Maignan, and C Martin. Oxygen storage capacity and structural
flexibility of LuFe2O4+x (0-x-0.5). Nat. Mater., 13(1):74?80, jan 2014.
[64] K.-T. Ko, H.-J. Noh, J.-Y. Kim, B.-G. Park, J.-H. Park, A Tanaka, S. B.
Kim, C. L. Zhang, and S-W. Cheong. Electronic Origin of Giant Magnetic
Anisotropy in Multiferroic LuFe2O4. Phys. Rev. Lett., 103(20):207202, nov
2009.
[65] M. Tanaka, J. Akimitsu, Y. Inada, N. Kimizuka, I. Shindo, and K. Siratori.
Conductivity and specific heat anomalies at the low temperature transition in
the stoichiometric YFe2O4. Solid State Commun., 44(5):687?690, nov 1982.
[66] Kengo Oka, Masaki Azuma, Naoaki Hayashi, Shigetoshi Muranaka, Ya-
suo Narumi, Koichi Kindo, Shinya Ayukawa, Masatsune Kato, Yoji Koike,
Yuichi Shimakawa, and Mikio Takano. Charge and Magnetic Orderings in
the Triangular-Lattice Antiferromagnet InFe 2 O 4. J. Phys. Soc. Japan,
77(6):064803, jun 2008.
[67] Zhong Lin Wang, Yi Liu, and Ze Zhang, editors. X-ray and Neutron Scattering,
pages 344?371. Springer US, Boston, MA, 2003.
[68] Paul Niggli and Carl Hermann. International Tables for Crystallography, vol-
ume C of International Tables for Crystallography. International Union of
Crystallography, Chester, England, oct 2006.
[69] A. Le Bail, H. Duroy, and J.L. Fourquet. Ab-initio structure determination
of LiSbWO6 by X-ray powder diffraction. Mater. Res. Bull., 23(3):447?452,
mar 1988.
[70] Armel Le Bail. Whole powder pattern decomposition methods and applica-
tions: A retrospection. Powder Diffr., 20(04):316?326, 2005.
[71] H. M. Rietveld. Line profiles of neutron powder-diffraction peaks for structure
refinement. Acta Crystallogr., 22(1):151?152, jan 1967.
[72] P. Y. Pecharsky, V. K.; Zavalij. Fundamentals of Powder Diffraction and
Structural Characterization of Materials. Springer US, Boston, MA, 2nd editio
edition, 2009.
[73] R. K. Nandi and S. P. Sen Gupta. The analysis of X-ray diffraction profiles
from imperfect solids by an application of convolution relations. J. Appl.
Crystallogr., 11(1):6?9, feb 1978.
[74] Th. H. de Keijser, J. I. Langford, E. J. Mittemeijer, and A. B. P. Vogels. Use of
the Voigt function in a single-line method for the analysis of X-ray diffraction
line broadening. J. Appl. Crystallogr., 15(3):308?314, jun 1982.
174
[75] N. C. Halder and C. N. J. Wagner. Analysis of the broadening of powder pat-
tern peaks using variance, integral breadth, and fourier coefficients of the line
profile. In Gavin R. Mallett, Marie J. Fay, and William M. Mueller, editors,
Advances in X-Ray Analysis, pages 91?102, Boston, MA, 1966. Springer US.
[76] G. K. Wertheim, M. A. Butler, K. W. West, and D. N. E. Buchanan. Deter-
mination of the Gaussian and Lorentzian content of experimental line shapes.
Rev. Sci. Instrum., 45(11):1369?1371, nov 1974.
[77] R. J. Hill and C. J. Howard. Quantitative phase analysis from neutron powder
diffraction data using the Rietveld method. J. Appl. Crystallogr., 20(6):467?
474, dec 1987.
[78] Brian H. Toby. R factors in Rietveld analysis: How good is good enough?
Powder Diffr., 21(01):67?70, mar 2006.
[79] D. E. Cox. The Rietveld method. (IUCr Monograph on Crystallography, No.
5) edited by R. A. Young. J. Appl. Crystallogr., 27(3):440?441, jun 1994.
[80] D. E. Moncton, E. Crosbie, and G. K. Shenoy. Overview of the advanced
photon source (invited). Rev. Sci. Instrum., 60(7):1403?1405, jul 1989.
[81] Stephen E. Nagler, Alexandru D. Stoica, Grigoreta M. Stoica, Ke An,
Harley D. Skorpenske, Orlando Rios, David B. Hendin, and Nathan W. Bower.
Time-of-Flight Neutron Diffraction (TOF-ND) Analyses of the Composition
and Minting of Ancient Judaean ?Biblical? Coins. J. Anal. Methods Chem.,
2019:1?18, mar 2019.
[82] Peter J. Chupas, Karena W. Chapman, Charles Kurtz, Jonathan C. Hanson,
Peter L. Lee, and Clare P. Grey. A versatile sample-environment cell for non-
ambient X-ray scattering experiments. J. Appl. Crystallogr., 41(4):822?824,
2008.
[83] John M. Cowley. Diffuse scattering and absorption effects. In Diffr. Phys.,
pages 257?285. Elsevier, 1995.
[84] Silvia C Capelli. Structure of Complex Materials. In Exp. Methods Phys. Sci.,
volume 44, pages 321?352. Elsevier, 2013.
[85] B. H. Toby and T. Egami. Accuracy of pair distribution function analysis
applied to crystalline and non???crystalline materials. Acta Crystallogr. Sect.
A, 48(3):336?346, 1992.
[86] Egami Takeshi and Simon J.L. Billinge. Structure of Amorphous Materials. In
Pergamon Mater. Ser., volume 16, chapter 12, pages 455?465. Elsevier, 2012.
[87] Peter Debye. Dispersion of R?ontgen rays. Ann. Phys. (N. Y)., 46:809, 1915.
175
[88] B D Shrivastava. X-ray absorption fine structure (XAFS) spectroscopy using
synchrotron radiation. J. Phys. Conf. Ser., 365(365):012002, may 2012.
[89] P. A. Lee, P. H. Citrin, P. Eisenberger, and B. M. Kincaid. Extended x-ray
absorption fine structure?its strengths and limitations as a structural tool.
Rev. Mod. Phys., 53(4):769?806, oct 1981.
[90] C A Ashley and S Doniach. Theory of extended x-ray absorption edge fine
structure (EXAFS) in crystalline solids. Phys. Rev. B, 11(4):1279?1288, feb
1975.
[91] Junko Yano and Vittal K. Yachandra. X-ray absorption spectroscopy. Photo-
synth. Res., 102(2):241?254, 2009.
[92] B. Ravel and M. Newville. ATHENA , ARTEMIS , HEPHAESTUS : data
analysis for X-ray absorption spectroscopy using IFEFFIT. J. Synchrotron
Radiat., 12(4):537?541, jul 2005.
[93] H.P. Longerich, G.A. Jenner, B.J. Fryer, and S.E. Jackson. Inductively coupled
plasma-mass spectrometric analysis of geological samples: A critical evalua-
tion based on case studies. Chem. Geol., 83(1-2):105?118, jun 1990.
[94] Melvin B. Robin and Peter Day. Mixed Valence Chemistry-A Survey and
Classification. pages 247?422. 1968.
[95] Ke-Yu Lai and Arumugam Manthiram. Phase Stability, Oxygen-Storage Ca-
pability, and Electrocatalytic Activity in Solid Oxide Fuel Cells of (Y, In,
Ca)BaCo 4? y Ga y O 7+?. Chem. Mater., 28(24):9077?9087, dec 2016.
[96] G. Korotcenkov. Metal oxides for solid-state gas sensors: What determines
our choice? Mater. Sci. Eng. B, 139(1):1?23, apr 2007.
[97] Chengxiang Wang, Longwei Yin, Luyuan Zhang, Dong Xiang, and Rui Gao.
Metal Oxide Gas Sensors: Sensitivity and Influencing Factors. Sensors,
10(3):2088?2106, mar 2010.
[98] S. K.; Li, R. J.; Yu, C. C.; Ji, W. J.; Shen. Methane oxidation to synthesis gas
using lattice oxygen in La1-?SrxFeO3 perovskite oxides instead of molecular
oxygen. Stud. Surf. Sci. Catal., 147:199?204, 2004.
[99] J. Bourgeois, M. Hervieu, M. Poienar, A. M. Abakumov, E. Elka??m, M. T.
Sougrati, F. Porcher, F. Damay, J. Rouquette, G. Van Tendeloo, A. Maig-
nan, J. Haines, and C. Martin. Evidence of oxygen-dependent modulation in
LuFe2O4. Phys. Rev. B, 85(6):064102, feb 2012.
[100] Shi Cao, Jun Li, Zhen Wang, Huanfang Tian, Yuanbin Qin, Lunjie Zeng, Chao
Ma, Huaixin Yang, and Jianqi Li. Extreme chemical sensitivity of nonlinear
conductivity in charge-ordered LuFe2O4. Sci. Rep., 2(1):330, dec 2012.
176
[101] Saptarshi Ghosh, Ujjal Chowdhury, Somenath Roy, and Rajib Bandyopad-
hyay. Detection of low ppm carbon monoxide with charge ordered LuFe 2 O
4 gas sensor ? A novel sensing mechanism. Ceram. Int., 42(13):14944?14948,
oct 2016.
[102] M Hervieu, F Damay, A Maignan, and C Martin. Rare earth ferrites LuFe
2 O 4?x polymorphism, polytypism and metastable phases. Solid State Sci.,
48:A1?A16, oct 2015.
[103] M. Isobe, N Kimizuka, J Iida, and S Takekawa. Structures of LuFeCoO4 and
LuFe2O4. Acta Crystallogr. Sect. C Cryst. Struct. Commun., 46(10):1917?
1918, oct 1990.
[104] M. Angst, R. P. Hermann, A. D. Christianson, M. D. Lumsden, C. Lee, M.-H.
Whangbo, J.-W. Kim, P. J. Ryan, S. E. Nagler, W. Tian, R. Jin, B. C. Sales,
and D. Mandrus. Charge Order in LuFe2O4: Antiferroelectric Ground State
and Coupling to Magnetism. Phys. Rev. Lett., 101(22):227601, nov 2008.
[105] Makoto Iwata and Yoshihiro Ishibashi. Possible Commensurate Phases in
LuFe 2 O 4. J. Phys. Soc. Japan, 81(3):035003, mar 2012.
[106] Noboru Kimizuka and Eiji Takayama. Ln(GaM2+)O4 and Ln(AlMn2+)O4
compounds having a layer structure [Ln = Lu, Yb, Tm, Er, Ho, and Y, and M
= Mg, Mn, Co, Cu, and Zn]. J. Solid State Chem., 41(2):166?173, feb 1982.
[107] T. Matsumoto, N. Mo?ri, J Iida, M Tanaka, K Siratori, F Izumi, and H Asano.
Crystal structures of the two dimensional antiferromagnets RFe2O4 (R = Y,
Er) and their magnetic properties under pressure. Phys. B Condens. Matter,
180-181:603?605, jun 1992.
[108] Noboru Kimizuka and Takahiko Mohri. Spinel, YbFe2O4, and Yb2Fe3O7
types of structures for compounds in the In2O3 and Sc2O3-A2O3-BO systems
[A: Fe, Ga, or Al; B: Mg, Mn, Fe, Ni, Cu, or Zn] at temperatures over 1000?C.
J. Solid State Chem., 60(3):382?384, dec 1985.
[109] Naoshi Ikeda, Hiroyuki Ohsumi, Kenji Ohwada, Kenji Ishii, Toshiya Inami,
Kazuhisa Kakurai, Youichi Murakami, Kenji Yoshii, Shigeo Mori, Yoichi
Horibe, and Hijiri Kito?. Ferroelectricity from iron valence ordering in the
charge-frustrated system LuFe2O4. Nature, 436(7054):1136?1138, aug 2005.
[110] Kenji Yoshii, Naoshi Ikeda, Yuka Okajima, Yasuhiro Yoneda, Yoji Matsuo,
Yoichi Horibe, and Shigeo Mori. Magnetic and Dielectric Properties of InFe 2
O 4 , InFeCuO 4 , and InGaCuO 4. Inorg. Chem., 47(14):6493?6501, jul 2008.
[111] J. de Groot, K Marty, M D Lumsden, A D Christianson, S E Nagler, S Adiga,
W J H Borghols, K Schmalzl, Z Yamani, S R Bland, R. de Souza, U Staub,
W Schweika, Y Su, and M Angst. Competing Ferri- and Antiferromagnetic
Phases in Geometrically Frustrated LuFe2O4. Phys. Rev. Lett., 108(3):037206,
jan 2012.
177
[112] R.J. Cava, A.P. Ramirez, Q Huang, and J.J. Krajewski. Compounds with the
YbFe2O4Structure Type: Frustrated Magnetism and Spin-Glass Behavior. J.
Solid State Chem., 140(2):337?344, nov 1998.
[113] Naoshi Ikeda, Tomoko Nagata, Jun Kano, and Shigeo Mori. Present status
of the experimental aspect of R Fe 2 O 4 study. J. Phys. Condens. Matter,
27(5):053201, feb 2015.
[114] Midori Tanaka, Hiroko Iwasaki, Kiiti Siratori, and Isamu Shindo. Mo?ssbauer
Study on the Magnetic Stracture of YbFe2O4 A Two-Dimensional Antiferro-
magnet on a Triangular Lattice.pdf. Phys.Soc. Japan, 58:1433, 1989.
[115] Hiroshi Mizoguchi, Arthur W Sleight, and M A Subramanian. New Oxides
Showing an Intense Blue Color Based on Mn 3+ in Trigonal-Bipyramidal
Coordination. Inorg. Chem., 50(1):10?12, jan 2011.
[116] K.-T. Ko, H.-J. Noh, J.-Y. Kim, B.-G. Park, J.-H. Park, A. Tanaka, S. B.
Kim, C. L. Zhang, and S-W. Cheong. Electronic Origin of Giant Magnetic
Anisotropy in Multiferroic LuFe2O4. Phys. Rev. Lett., 103(20):207202, nov
2009.
[117] Charles M. Brooks, Rajiv Misra, Julia A. Mundy, Lei A. Zhang, Brian S
Holinsworth, Kenneth R. O?Neal, Tassilo Heeg, Willi Zander, J. Schubert,
Janice L. Musfeldt, Zi-Kui Liu, David A. Muller, Peter Schiffer, and Darrell G.
Schlom. The adsorption-controlled growth of LuFe 2 O 4 by molecular-beam
epitaxy. Appl. Phys. Lett., 101(13):132907, sep 2012.
[118] R. W. Cheary and Alan Coelho. A fundamental parameters approach to X-ray
line-profile fitting. J. Appl. Crystallogr., 25(2):109?121, apr 1992.
[119] Brian H. Toby and Robert B. Von Dreele. GSAS-II : the genesis of a modern
open-source all purpose crystallography software package. J. Appl. Crystal-
logr., 46(2):544?549, apr 2013.
[120] John S.O. Evans. Advanced Input Files & Parametric Quantitative Analysis
Using Topas. Mater. Sci. Forum, 651:1?9, may 2010.
[121] O.; Ge?rardin, R.; Nodari, I.;Aqachmar, H. & Evrard. Mise en e?vidence de
deux nouvelles familles de ferrites de lanthano??des Ln2Fe4O9 et Ln4Fe6O15.
C. R. Acad. Sci., 295:863, 1982.
[122] I. Nodari, A. Alebouyeh, J.F. Brice, R. Ge?rardin, and O. Evrard. Caracteri-
sation de nouveaux ferrites d?indium: In2Fe4O9 et InFeO3. Mater. Res. Bull.,
23(7):1039?1044, jul 1988.
[123] Qiuwan Shen, Ying Zheng, Cong Luo, Ning Ding, Chu-Guang Zheng, and
Marcus Thern. Effect of A/B-site substitution on oxygen production perfor-
mance of strontium cobalt based perovskites for CO 2 capture application.
RSC Adv., 5(50):39785?39790, 2015.
178
[124] C Abughayada, B Dabrowski, S Kolesnik, D E Brown, and O Chmaissem.
Characterization of Oxygen Storage and Structural Properties of Oxygen-
Loaded Hexagonal R MnO 3+? ( R = Ho, Er, and Y). Chem. Mater.,
27(18):6259?6267, sep 2015.
[125] X. S. Xu, J. de Groot, Q.-C. Sun, B. C. Sales, D. Mandrus, M. Angst, A. P.
Litvinchuk, and J. L. Musfeldt. Lattice dynamical probe of charge order and
antipolar bilayer stacking in LuFe2O4. Phys. Rev. B, 82(1):014304, jul 2010.
[126] I. D. Brown and D. Altermatt. Bond-valence parameters obtained from a
systematic analysis of the Inorganic Crystal Structure Database. Acta Crys-
tallogr. Sect. B Struct. Sci., 41(4):244?247, aug 1985.
[127] Olivier Charles Gagne? and Frank Christopher Hawthorne. Comprehensive
derivation of bond-valence parameters for ion pairs involving oxygen. Acta
Crystallogr. Sect. B Struct. Sci. Cryst. Eng. Mater., 71(5):562?578, oct 2015.
[128] R. D. Shannon. Revised effective ionic radii and systematic studies of in-
teratomic distances in halides and chalcogenides. Acta Crystallogr. Sect. A,
32(5):751?767, sep 1976.
[129] R. S. Jayathilake, B. D. Levitas, and Efrain E Rodriguez. In situ diffraction
studies on reversible oxygen uptake and release in AFe 2 O 4 + ? (A= Lu,
Yb, Y, and In). J. Mater. Chem. A, 6(11):4801?4810, 2018.
[130] A D Christianson, M D Lumsden, M Angst, Z Yamani, W Tian, R Jin, E A
Payzant, S E Nagler, B C Sales, and D Mandrus. Three-Dimensional Magnetic
Correlations in Multiferroic LuFe2O4. Phys. Rev. Lett., 100(10):107601, mar
2008.
[131] Sarah Nicoud, Marielle Huve?, Olivier Hernandez, Alain Pautrat, Mathieu
Duttine, Alain Wattiaux, Claire Colin, Houria Kabbour, and Olivier Mentre?.
Comprehensive Study of Oxygen Storage in YbFe 2 O 4+ x ( x - 0.5): Un-
precedented Coexistence of FeO n Polyhedra in One Single Phase. J. Am.
Chem. Soc., 139(47):17031?17043, nov 2017.
[132] Peter Debye and Paul Scherrer. X-ray scattering interference patterns of par-
ticles oriented at random II. Nachr. Ges. Wiss. Goettingen, Math.-Phys. Kl.,
1916b:1?15, 1916.
[133] A. W. Hull. A New Method of X-Ray Crystal Analysis. Phys. Rev., 10(6):661?
696, dec 1917.
[134] W. H. Zachariasen. The UCl 3 Type of Crystal Structure. J. Chem. Phys.,
16(3):254?254, mar 1948.
[135] W. H. Zachariasen. Crystal chemical studies of the 5f-series of elements. I.
New structure types. Acta Crystallogr., 1(5):265?268, nov 1948.
179
[136] W. H. Zachariasen. Crystal chemical studies of the 5f-series of elements. II.
The crystal structure of Cs2PuCl6. Acta Crystallogr., 1(5):268?269, nov 1948.
[137] W. H. Zachariasen and F. H. Ellinger. The crystal structure of beta plutonium
metal. Acta Crystallogr., 16(5):369?375, may 1963.
[138] P. C. Debets. The structures of uranyl chloride and its hydrates. Acta Crys-
tallogr. Sect. B Struct. Crystallogr. Cryst. Chem., 24(3):400?402, mar 1968.
[139] Clifford G. Shull. Early development of neutron scattering. Rev. Mod. Phys.,
67(4):753?757, oct 1995.
[140] H. M. Rietveld. A profile refinement method for nuclear and magnetic struc-
tures. J. Appl. Crystallogr., 2(2):65?71, jun 1969.
[141] G. Malmros and J. O. Thomas. Least-squares structure refinement based
on profile analysis of powder film intensity data measured on an automatic
microdensitometer. J. Appl. Crystallogr., 10(1):7?11, feb 1977.
[142] By J B Hastings, W Thomlinson, and D E Cox. Synchrotron X-ray Powder
Diffraction. J. Appl. Crystallogr., 17:85?95, 1983.
[143] R. B. Von Dreele, J. D. Jorgensen, and C. G. Windsor. Rietveld refine-
ment with spallation neutron powder diffraction data. J. Appl. Crystallogr.,
15(6):581?589, dec 1982.
[144] D. H. Olson, G. T. Kokotailo, S. L. Lawton, and W. M. Meier. Crystal struc-
ture and structure-related properties of ZSM-5. J. Phys. Chem., 85(15):2238?
2243, jul 1981.
[145] G. T. KOKOTAILO, S. L. LAWTON, D. H. OLSON, and W. M. MEIER.
Structure of synthetic zeolite ZSM-5. Nature, 272(5652):437?438, mar 1978.
[146] W. M. MEIER and H. Villiger. No Title. Zeitschrift fur Krist., 129:411?423,
1969.
[147] G. S. Pawley. Unit-cell refinement from powder diffraction scans. J. Appl.
Crystallogr., 14(6):357?361, dec 1981.
[148] W. I. F. David, K. Shankland, L. B. McCusker, and Ch. Baerlocher, edi-
tors. Structure Determination from Powder Diffraction Data. Oxford Science
Publications, 2002.
[149] Russell E. Morris, Jonathan J. Owen, Judith K. Stalick, and Anthony K.
Cheetham. Determination of Complex Structures from Powder Diffraction
Data: The Crystal Structure of La3Ti5Al15O37. J. Solid State Chem.,
111(1):52?57, jul 1994.
180
[150] C. Giacovazzo. Direct Methods and Powder Data: State of the Art and Per-
spectives. Acta Crystallogr. Sect. A Found. Crystallogr., 52(3):331?339, may
1996.
[151] C. C. Wilson and J. W. Wadsworth. Crystal structure determination from
low-resolution X-ray powder diffraction data. Acta Crystallogr. Sect. A Found.
Crystallogr., 46(4):258?262, apr 1990.
[152] G. Bricogne and C. J. Gilmore. A multisolution method of phase determination
by combined maximization of entropy and likelihood. I. Theory, algorithms
and strategy. Acta Crystallogr. Sect. A Found. Crystallogr., 46(4):284?297,
apr 1990.
[153] C. J. Gilmore, K. Shankland, and G. Bricogne. Applications of the Maximum
Entropy Method to Powder Diffraction and Electron Crystallography. Proc.
R. Soc. A Math. Phys. Eng. Sci., 442(1914):97?111, jul 1993.
[154] C. Giacovazzo. Direct phasing in crystallography. Oxford Science Publications,
Giacovazzo1998, 1998.
[155] M Newsam, M W Deem, and C M Freeman. Data. Accuracy Powder Diffr.
II, NIST Spec. Publ, 846:80?91, 1992.
[156] A. A. Coelho. Whole-profile structure solution from powder diffraction data
using simulated annealing. J. Appl. Crystallogr., 33(3):899?908, jun 2000.
[157] K. D. M. Harris, R. L. Johnston, and B. M. Kariuki. The Genetic Algorithm:
Foundations and Apllications in Structure Solution from Powder Diffraction
Data. Acta Crystallogr. Sect. A Found. Crystallogr., 54(5):632?645, sep 1998.
[158] Kenneth D. M. Harris, Maryjane Tremayne, Philip Lightfoot, and Peter G.
Bruce. Crystal Structure Determination from Powder Diffraction Data by
Monte Carlo Methods. J. Am. Chem. Soc., 116(8):3543?3547, apr 1994.
[159] Radovan C?erny? and Vincent Favre-Nicolin. Direct space methods of structure
determination from powder diffraction: principles, guidelines and perspectives.
Zeitschrift fu?r Krist. - Cryst. Mater., 222(3-4):105?113, jan 2007.
[160] F. J. Humphreys and M. Hatherly. Recrystallization and Related Annealing
Phenomena. Elsevier, 1995.
[161] S. Kirkpatrick, C. D. Gelatt, and M. P. Vecchi. Optimization by Simulated
Annealing. Science (80-. )., 220(4598):671?680, may 1983.
[162] Nicholas Metropolis, Arianna W. Rosenbluth, Marshall N. Rosenbluth, Au-
gusta H. Teller, and Edward Teller. Equation of State Calculations by Fast
Computing Machines. J. Chem. Phys., 21(6):1087?1092, jun 1953.
181
[163] C. R. A. Catlow. Computer Modeling in Inorganic Crystallography. Academic
press, 1996.
[164] DH Smith, SM Allen, S Hurley, and WJ Watkins. Frequency assignment:
Methods and algorithms. In Proceedings of the NATO RTA SET/ISET
symposium on frequency assignment, sharing and conservation in systems
(aerospace), Aalborg, Denmark, pages K1?K18. Citeseer, 1998.
[165] H. Toraya and S. Yamazaki. Simulated annealing structure solution of a new
phase of dicalcium silicate Ca 2 SiO 4 and the mechanism of structural changes
from ?-dicalcium silicate hydrate to ? L ?dicalcium silicate via the new phase.
Acta Crystallogr. Sect. B Struct. Sci., 58(4):613?621, aug 2002.
[166] L. Ingber. Very fast simulated re-annealing. Math. Comput. Model., 12(8):967?
973, 1989.
[167] M Hervieu, A Guesdon, J Bourgeois, E Elka??m, M Poienar, F Damay, J Rou-
quette, A Maignan, and C Martin. Oxygen storage capacity and structural
flexibility of LuFe2O4+x (0-x-0.5). Nat. Mater., 13(1):74?80, jan 2014.
[168] Jo?rg Neuefeind, Mikhail Feygenson, John Carruth, Ron Hoffmann, and Ken-
neth K. Chipley. The Nanoscale Ordered MAterials Diffractometer NOMAD
at the Spallation Neutron Source SNS. Nucl. Instruments Methods Phys. Res.
Sect. B Beam Interact. with Mater. Atoms, 287:68?75, sep 2012.
[169] J. Bourgeois, M. Hervieu, M. Poienar, A. M. Abakumov, E. Elka??m, M. T.
Sougrati, F. Porcher, F. Damay, J. Rouquette, G. Van Tendeloo, A. Maignan,
J. Haines, and C. Martin. Evidence of oxygen-dependent modulation in LuFe
2O 4. Phys. Rev. B - Condens. Matter Mater. Phys., 85(6), 2012.
[170] I Nodari, A Alebouyeh, J F Brice, R Ge?rardin, and O Evrard. Caracterisa-
tion de nouveaux ferrites d?indium: In2Fe4O9 et InFeO3. Mater. Res. Bull.,
23(7):1039?1044, 1988.
[171] A. A. Coehlo. Whole-profile structure solution from powder diffraction data
using simulated annealing. J. Appl. Crystallogr., 33(1953):899?908, 2000.
[172] Bong Kyu Bang, Sung Baek Kim, S-W. Cheong, and Chul Sung Kim.
Mo?ssbauer study of iron ordering in mixed valence system LuFe2O4. Phys.
status solidi, 244(12):4566?4569, dec 2007.
[173] John S. Dennis and Stuart A. Scott. In situ gasification of a lignite coal and
CO2 separation using chemical looping with a Cu-based oxygen carrier. Fuel,
89(7):1623?1640, jul 2010.
[174] Lee Jin-Bae, Park Chu-Sik, Choi Sang-II, Song Young-Wook, Kim Young-
Ho, and Yang Hyun-Soo. Redox Characteristics of Various Kinds of Oxygen
Carriers for Hydrogen Fueled Chemical-Looping Combustion. J. Ind. Eng.
Chem., 11(1):96?102, 2005.
182
[175] Jing-biao Yang, Ning-sheng Cai, and Zhen-shan Li. Reduction of Iron Oxide
as an Oxygen Carrier by Coal Pyrolysis and Steam Char Gasification Inter-
mediate Products. Energy & Fuels, 21(6):3360?3368, nov 2007.
[176] Minjun Wang, Jing Liu, Jianbo Hu, and Feng Liu. O 2 ?CO 2 Mixed Gas
Production Using a Zr-Doped Cu-Based Oxygen Carrier. Ind. Eng. Chem.
Res., 54(40):9805?9812, oct 2015.
[177] Shamseldin A. Mohamed, Mohammad R. Quddus, Shaikh A. Razzak, Moham-
mad M. Hossain, and Hugo I. de Lasa. Fluidizable NiO/Ce-?Al 2 O 3 Oxygen
Carrier for Chemical Looping Combustion. Energy & Fuels, 29(9):6095?6103,
sep 2015.
[178] Sai Chen, Liang Zeng, Hao Tian, Xinyu Li, and Jinlong Gong. Enhanced
Lattice Oxygen Reactivity over Ni-Modified WO3-Based Redox Catalysts for
Chemical Looping Partial Oxidation of Methane. ACS Catal., 7(5):3548?3559,
2017.
[179] Golnar Azimi, Magnus Ryde?n, Henrik Leion, Tobias Mattisson, and Anders
Lyngfelt. (Mn z Fe 1-z ) y O x combined oxides as oxygen carrier for chemical-
looping with oxygen uncoupling. AIChE J., 59(2):582?588, feb 2013.
[180] Golnar Azimi, Henrik Leion, Tobias Mattisson, and Anders Lyngfelt.
Chemical-looping with oxygen uncoupling using combined Mn-Fe oxides, test-
ing in batch fluidized bed. Energy Procedia, 4:370?377, 2011.
[181] Golnar Azimi, Henrik Leion, Magnus Ryde?n, Tobias Mattisson, and An-
ders Lyngfelt. Investigation of Different Mn?Fe Oxides as Oxygen Carrier
for Chemical-Looping with Oxygen Uncoupling (CLOU). Energy & Fuels,
27(1):367?377, jan 2013.
183