University of Southern California Dissertations and Theses

Understanding intercalation-driven structural transformations in energy storage materials

(USC Thesis Other)

Understanding intercalation-driven structural transformations in energy storage materials

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Understanding Intercalation-Driven Structural Transformations in Energy
Storage Materials
by
Jessica L. Andrews
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(CHEMISTRY)
December 2024
Copyright 2024 Jessica L. Andrews

Acknowledgments
Reflecting on my academic journey, I firmly believe that I would not have completed or even
started my PhD without the support and encouragement of my family, friends, and mentors.
Finishing a PhD is a balancing act, and a long one at that (you know, the whole marathon
not a sprint thing), and I know that it would have been exponentially more difficult for me
to balance everything that a PhD entails without my support system.
First, I want to thank my husband, Shay. From high school, into college, and throughout
my entire time at USC you have always pushed me to do my absolute best, but also to be
kinder to myself. You’ve given me balance in the form of binge-watching Netflix/HBO/Hulu
together, long weekends camping in Joshua Tree, introducing me to Animal Crossing, etc.,
etc. (I could go on and on). You’ve also been there to carry more of the load when I couldn’t,
like doing more of the cooking and cleaning so that I could write papers, work on my Quals,
and, more recently, compile this enormous document. Thank you for always listening to me
complain, making me laugh, and making sure there is ice cream in the freezer.
I also want to thank my family, especially my Mom, Dad, and sister, Hannah. Mom
and Dad, thank you for raising me to believe in myself. You taught us never to doubt if we
could do something, but rather, to simply put in the work and make the things we wanted
to happen. Encouraging us to play sports, pushing us to do well in school, and ensuring
we also always made time for fun taught me the time-management skills that kept me sane
and carried me through this PhD. Hannah, thank you for being my best friend and the best
sister anyone could ask for. You have always been my role model and one of my biggest
supporters. I’m so grateful to have a sister who I look up to, pushes me to do my best,
constantly teaches me new things, and is always willing to listen (I don’t think I could’ve
commuted for as long as I have without sister phone calls). Thank you also for giving me a
brother and built-in financial advisor by marrying Tor!
ii

Thank you to all of my research mentors, past and present. Brent, thank you for all of
your support and mentorship over the past 5 years. Thank you for being an advisor who
encourages work-life balance, values our scientific opinions, and cooks amazing barbecue.
I’m so grateful that you always were available to sit down and explain a concept or plan an
experiment with me while simultaneously pushing me to become a more independent and
confident scientist. Hayden, thank you for convincing me to give the whole PhD thing a try.
Working with you as an undergraduate student not only taught me so much about materials
science and research but was also always so much fun. I’m thankful that we are friends (and
collaborators if we ever write that paper) to this day. Molleigh, thank you for taking me
under your wing during my senior year and introducing me to battery research, then taking
me under your wing again and teaching me the ways of beam time during my many solo
SSRL trips. Ram, thank you for being an amazing undergraduate research advisor. Being
a part of your research group as an undergraduate sparked my interest in graduate school,
and I’m so grateful for your support through SCALAR during my PhD.
I’m also incredibly thankful to call my lab mates friends and consider them a part of
my support system. Michael, thank you for the many walks to get lunch and/or coffee, the
lab jam sessions and concerts, for letting me talk your ear off about just about everything
(including Rietveld refinements), and for being my partner in crime throughout our PhDs.
Eric, thank you for being both a friend and a mentor, for teaching me so much about
crystallography, for struggling through the early days of fluoride ion with me, and for making
delicious scones. Megan and Gemma, I’m so grateful to have started and completed my PhD
alongside two other determined women. Gemma, thank you for being my first new friend
at USC (throwback to visit weekend!), for walks back from happy hours, and our recent
swimming adventures together. Megan, thank you for the many conversations about baking,
your always thoughtful gifts, and your understanding of my love of Coke Zero (and enjoying
many with me). Thank you to Moy and Zahra, who manage to make me laugh daily and
who I am so grateful to have become such good friends with so fast. Thank you also to
iii

the friends I made in the Chemistry department from other research groups, especially my
former roommate Sydney—I could not have made it through COVID without you and I’m
so thankful for our many nights of cooking and drinking wine together.
I also would not be here without my friends away from USC. Bianca and Fallon, I’m so
grateful that UCSB gave me two smart, funny, and hard-working best friends who constantly
inspire me. Bianca, thank you for taking a chance and asking me to be your roommate after
knowing me for about a week. There’s no one else I would have rather slept 3 feet away
from every night for 3 years. Fallon, thank you for being our honorary roommate, eating
many meals with me while Bianca studied, and my COVID picnic buddy. Samara, while I
consider you a friend “away from USC”, it did bring us back together, and I’m so grateful
for that. Thank you for study dates (it always helped to have someone sitting across from
me also getting their ass kicked by grad school, even if it was a totally different subject)
and Bachelorette nights where we mostly talked over the show. Thank you to my Wildwood
friends—Australia, Squirrel, Thunder, and Swan. Although your names will make no sense
to anyone else who reads this, I’m so grateful for our serendipitous friendship and recordlength DnD campaign. Thank you also to my very large group of friends from the group
chat with an ever-changing name that makes no sense (you know who you are) for many
nights of Jack Box, barbecues, Friendsgivings, pool parties, and so much more.
Thank you also to the many collaborators I have had the pleasure of working with. From
the professors who offered additional mentorship to fellow graduate students who helped
with measurements and became friends, so much of my PhD research (especially the fluoride
ion work) could not have been done without your help. In particular, I want to say thank you
to the SCALAR PIs (especially Sarah, Bruce, and Sri), Will, and Simon. Finally, thank you
to all of the undergraduate and high school students, especially Kenny and Sophie, whom
I had the pleasure of mentoring during my PhD. Thank you for helping me in the lab and
acting like what I was saying made any sense—I truly believe that teaching you all taught
me so much about the fundamental science behind my research.
iv

Table of Contents
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
Curriculum Vitae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Chapter 1:
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 A Brief Overview of the Energy Crisis . . . . . . . . . . . . . . . . . . . . . 1
1.2 Intercalation-based Energy Storage . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.1 Li-ion Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.2 F-ion Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3 Operando X-ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.4 Differential Electrochemical Mass Spectrometry . . . . . . . . . . . . . . . . 18
1.5 Relevant Crystal Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.5.1 Pyrochlore . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.5.2 Tetragonal Tungsten Bronze-type . . . . . . . . . . . . . . . . . . . . 22
1.5.3 Ruddlesden-Popper . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.6 Dissertation Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Chapter 2:
Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6 29
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.2 Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.2.1 Material Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.2.2 Powder X-ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.2.3 Zero-field 57Fe M¨ossbauer spectroscopy . . . . . . . . . . . . . . . . . 34
2.2.4 Scanning Electron Microscopy . . . . . . . . . . . . . . . . . . . . . . 34
2.2.5 X-ray Photoelectron Spectroscopy . . . . . . . . . . . . . . . . . . . . 35
2.2.6 Electrochemical Characterization . . . . . . . . . . . . . . . . . . . . 35
2.2.7 X-ray Absorption Spectroscopy . . . . . . . . . . . . . . . . . . . . . 36
2.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
v

2.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.5 Supplemental Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.5.1 Synchrotron X-ray Diffraction, Scanning Electron Microscopy, and
Zero-Field 57Fe M¨ossbauer Spectroscopy . . . . . . . . . . . . . . . . 49
2.5.2 Energy Dispersive Spectroscopy . . . . . . . . . . . . . . . . . . . . . 51
2.5.3 X-ray Photoelectron Spectroscopy . . . . . . . . . . . . . . . . . . . . 52
2.5.4 Synchrotron X-ray Diffraction Rietveld Refinement Results and Ex
Situ X-ray Diffraction Analysis . . . . . . . . . . . . . . . . . . . . . 53
2.5.5 Electrochemical Characterization . . . . . . . . . . . . . . . . . . . . 62
2.5.6 X-ray Absorption Spectroscopy . . . . . . . . . . . . . . . . . . . . . 64
Chapter 3:
On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type
Nb8W9O47 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.2 Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.2.1 Material Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.2.2 Powder X-ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.2.3 Scanning Electron Microscopy . . . . . . . . . . . . . . . . . . . . . . 69
3.2.4 Electron Probe Microanalysis . . . . . . . . . . . . . . . . . . . . . . 70
3.2.5 X-ray Photoelectron Spectroscopy . . . . . . . . . . . . . . . . . . . . 70
3.2.6 Electrochemical Characterization . . . . . . . . . . . . . . . . . . . . 71
3.2.7 Operando X-ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . 71
3.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3.5 Supplemental Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
3.5.1 Powder X-ray Diffraction Rietveld Refinement Results . . . . . . . . 83
3.5.2 Additional Electrochemical Cycling Data . . . . . . . . . . . . . . . . 86
3.5.3 Electron Probe Microanalysis . . . . . . . . . . . . . . . . . . . . . . 87
3.5.4 X-ray Photoelectron Spectroscopy . . . . . . . . . . . . . . . . . . . . 88
3.5.5 Operando X-ray Diffraction Rietveld Refinement Results and Additional Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
Chapter 4:
Chemical Fluorination and Structural Evolution of Ruddlesden-Popper
BaLa2Fe2O7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
4.2 Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
4.2.1 Material Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
4.2.2 Powder X-ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . 115
4.2.3 Electron Probe Microanalysis . . . . . . . . . . . . . . . . . . . . . . 115
4.2.4 Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy 116
4.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
4.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
4.5 Supplemental Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
vi

4.5.1 Powder X-ray Diffraction Rietveld Refinement Results . . . . . . . . 126
4.5.2 Electron Probe Microanalysis Results . . . . . . . . . . . . . . . . . . 142
4.5.3 Additional ATR-FTIR Spectroscopy Results . . . . . . . . . . . . . . 147
Chapter 5:
Differential Electrochemical Mass Spectrometry for F-ion Batteries . . . 149
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
5.2 Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
5.2.1 Material Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
5.2.2 Powder X-ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . 153
5.2.3 Differential Electrochemical Mass Spectrometry . . . . . . . . . . . . 153
5.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
5.3.1 Cell-to-MS Interface Iteration 1 . . . . . . . . . . . . . . . . . . . . . 156
5.3.2 Cell-to-MS Interface Iteration 2 . . . . . . . . . . . . . . . . . . . . . 160
5.3.3 Cell-to-MS Interface Iteration 3 . . . . . . . . . . . . . . . . . . . . . 162
5.3.4 Investigating New F-ion Systems . . . . . . . . . . . . . . . . . . . . 165
5.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
5.5 Supplemental Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
5.5.1 Powder X-ray Diffraction and Rietveld Refinement Results . . . . . . 169
5.5.2 Additional Differential Electrochemical Mass Spectrometry Results . 171
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
vii

List of Tables
2.1 EDS Results for CsMnFeF6 Products . . . . . . . . . . . . . . . . . . . . . . 51
2.2 Rietveld Refinement Results for Hydrothermal CsMnFeF6 . . . . . . . . . . 54
2.3 Rietveld Refinement Results for Ceramic CsMnFeF6 . . . . . . . . . . . . . . 55
2.4 Rietveld Refinement Results for Mechanochemical CsMnFeF6 . . . . . . . . . 56
2.5 Rietveld Refinement Results for CsMnFeF6 Electrode Mix . . . . . . . . . . 56
2.6 Rietveld Refinement Results for Ex Situ CsMnFeF6: Cycle 1 . . . . . . . . . 57
2.7 Rietveld Refinement Results for Ex Situ CsMnFeF6: Cycle 2 . . . . . . . . . 57
2.8 Rietveld Refinement Results for Ex Situ CsMnFeF6: Cycle 3 and 4 . . . . . 58
3.1 Rietveld Refinement Results for Nb8W9O47: Transition Metal Sites . . . . . 84
3.2 Rietveld Refinement Results for Nb8W9O47: Oxygen Sites . . . . . . . . . . 85
3.3 EPMA Results for Nb8W9O47 . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.4 Sequential Pawley Fit Results for Operando XRD Scans 0–16 . . . . . . . . . 93
3.5 Sequential Pawley Fit Results for Operando XRD Scans 7–37 . . . . . . . . . 94
3.6 Sequential Rietveld Refinement Results (TM Positions Locked) for Operando
XRD Scans 0–16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.7 Sequential Rietveld Refinement Results (TM Positions Locked) for Operando
XRD Scans 17–37 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.8 Sequential Rietveld Refinement Results (TM Positions Refined) for Operando
XRD Scans 0–16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
viii

3.9 Sequential Rietveld Refinement Results (TM Positions Refined) for Operando
XRD Scans 17–37 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
3.10 Operando XRD Refined TM Positions: Sites 1–3, Scans 0–16 . . . . . . . . . 99
3.11 Operando XRD Refined TM Positions: Sites 1–3, Scans 17–37 . . . . . . . . 100
3.12 Operando XRD Refined TM Positions: Sites 4–6, Scans 0–16 . . . . . . . . . 101
3.13 Operando XRD Refined TM Positions: Sites 4–6, Scans 17–37 . . . . . . . . 102
3.14 Operando XRD Refined TM Positions: Sites 7–9, Scans 0–16 . . . . . . . . . 103
3.15 Operando XRD Refined TM Positions: Sites 7–9, Scans 17–37 . . . . . . . . 104
4.1 Rietveld Refinement Results for BaLa2Fe2O7 . . . . . . . . . . . . . . . . . . 127
4.2 Rietveld Refinement Results for BaLa2Fe2O7 + 1.0 PVDF: Main Phase . . . 128
4.3 Rietveld Refinement Results for BaLa2Fe2O7 + 1.0 PVDF: Impurities . . . . 129
4.4 Rietveld Refinement Results for BaLa2Fe2O7 + 1.5 PVDF: Main Phase . . . 130
4.5 Rietveld Refinement Results for BaLa2Fe2O7 + 1.5 PVDF: Impurities . . . . 131
4.6 Rietveld Refinement Results for BaLa2Fe2O7 + 2.0 PVDF . . . . . . . . . . 132
4.7 Rietveld Refinement Results for BaLa2Fe2O7 + 1.0 NH4HF2: Main Phase . . 133
4.8 Rietveld Refinement Results for BaLa2Fe2O7 + 1.0 NH4HF2: Impurities . . . 134
4.9 Rietveld Refinement Results for BaLa2Fe2O7 + 1.5 NH4HF2 . . . . . . . . . 135
4.10 Rietveld Refinement Results for BaLa2Fe2O7 + 2.0 NH4HF2: Main Phase . . 136
4.11 Rietveld Refinement Results for BaLa2Fe2O7 + 2.0 NH4HF2: Impurities . . . 137
4.12 Rietveld Refinement Results for BaLa2Fe2O7 + 2.5 NH4HF2: Main Phases . 138
4.13 Rietveld Refinement Results for BaLa2Fe2O7 + 2.5 NH4HF2: Impurities . . . 139
4.14 Rietveld Refinement Results for BaLa2Fe2O7 + 3.0 NH4HF2 . . . . . . . . . 140
4.15 EPMA Results for BaLa2Fe2O7 . . . . . . . . . . . . . . . . . . . . . . . . . 143
ix

4.16 EPMA Results for BaLa2Fe2O7 + 1.0 PVDF . . . . . . . . . . . . . . . . . . 143
4.17 EPMA Results for BaLa2Fe2O7 + 1.5 PVDF . . . . . . . . . . . . . . . . . . 144
4.18 EPMA Results for BaLa2Fe2O7 + 2.0 PVDF . . . . . . . . . . . . . . . . . . 144
4.19 EPMA Results for BaLa2Fe2O7 + 1.0 NH4HF2 . . . . . . . . . . . . . . . . . 145
4.20 EPMA Results for BaLa2Fe2O7 + 1.5 NH4HF2 . . . . . . . . . . . . . . . . . 145
4.21 EPMA Results for BaLa2Fe2O7 + 2.0 NH4HF2 . . . . . . . . . . . . . . . . . 146
4.22 EPMA Results for BaLa2Fe2O7 + 2.5 NH4HF2 . . . . . . . . . . . . . . . . . 146
5.1 Rietveld Refinement Results for ReO3 . . . . . . . . . . . . . . . . . . . . . . 170
5.2 Rietveld Refinement Results for CsMnFeF6 . . . . . . . . . . . . . . . . . . . 170
x

List of Figures
1.1 Li-ion Battery Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 LiCoO2 to CoO2 Structural Distortion . . . . . . . . . . . . . . . . . . . . . 8
1.3 LiFePO4 to FePO4 Structural Distortion . . . . . . . . . . . . . . . . . . . . 9
1.4 ReO3 to Li2ReO3 Structural Distortion . . . . . . . . . . . . . . . . . . . . . 10
1.5 TiNb2O7 Crystal Structure Before and After Lithiation . . . . . . . . . . . . 11
1.6 F-ion Battery Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.7 Operando Powder X-ray Diffraction Cell Schematic . . . . . . . . . . . . . . 17
1.8 Differential Electrochemical Mass Spectrometry Cell Schematic . . . . . . . . 20
1.9 Pyrochlore Crystal Structure Depiction . . . . . . . . . . . . . . . . . . . . . 22
1.10 TTB to TTB-type Crystal Structure Depiction . . . . . . . . . . . . . . . . . 23
1.11 Ruddlesden-Popper Crystal Structure Depiction . . . . . . . . . . . . . . . . 24
2.1 CsMnFeF6 Crystal Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2 Refined Mechanochemical CsMnFeF6 XRD Data and SEM Image . . . . . . 37
2.3 GCPL and PEIS of Mechanochemical CsMnFeF6 . . . . . . . . . . . . . . . 38
2.4 GCPL and Differential Capacity of Mechanochemical CsMnFeF6 . . . . . . . 40
2.5 M¨ossbauer Spectroscopy of Mechanochemical CsMnFeF6 . . . . . . . . . . . 42
2.6 Ex Situ CsMnFeF6 XRD Data . . . . . . . . . . . . . . . . . . . . . . . . . . 43
xi

2.7 Ex Situ CsMnFeF6 XAS Data . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.8 Refined Hydrothermal CsMnFeF6 XRD Data, SEM Image, and M¨ossbauer
Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.9 Refined Ceramic CsMnFeF6 XRD Data, SEM Image, and M¨ossbauer Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.10 Representative CsMnFeF6 EDS Specta . . . . . . . . . . . . . . . . . . . . . 51
2.11 Representative Hydrothermal CsMnFeF6 XPS Specta . . . . . . . . . . . . . 52
2.12 Representative Ceramic CsMnFeF6 XPS Specta . . . . . . . . . . . . . . . . 52
2.13 Representative Mechanochemical CsMnFeF6 XPS Specta . . . . . . . . . . . 52
2.14 Refined Ex Situ CsMnFeF6 XRD Data: Cycles 1 and 2 . . . . . . . . . . . . 59
2.15 Refined Ex Situ CsMnFeF6 XRD Data: Cycles 3 and 4 . . . . . . . . . . . . 60
2.16 Full Q Range Ex Situ CsMnFeF6 XRD Data . . . . . . . . . . . . . . . . . . 60
2.17 Phase-Matching of Ex Situ CsMnFeF6 XRD Data . . . . . . . . . . . . . . . 61
2.18 Pawley Fit of Oxidation 3 Ex Situ CsMnFeF6 XRD Data . . . . . . . . . . . 61
2.19 GCPL, PEIS, and Differential Capacity of Ceramic CsMnFeF6 . . . . . . . . 62
2.20 GCPL, PEIS, and Differential Capacity of Hydrothermal CsMnFeF6 . . . . . 63
2.21 Ex Situ CsMnFeF6 Mn K-edge XAS Data . . . . . . . . . . . . . . . . . . . 64
2.22 Ex Situ CsMnFeF6 Fe K-edge XAS Data . . . . . . . . . . . . . . . . . . . . 64
3.1 Nb8W9O47 Crystal Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.2 Refined Nb8W9O47 XRD Data and SEM Image . . . . . . . . . . . . . . . . 73
3.3 Variable-Rate GCPL of Nb8W9O47 . . . . . . . . . . . . . . . . . . . . . . . 74
3.4 Nb8W9O47 GCPL Voltage Cutoff Comparison . . . . . . . . . . . . . . . . . 75
3.5 Cyclic Voltammetry b-value Analysis of Nb8W9O47 . . . . . . . . . . . . . . 76
xii

3.6 In-House Operando XRD of Nb8W9O47, Comparison of Fitting Method Qualities, and Lattice Parameter Evolution during Lithiation . . . . . . . . . . . 78
3.7 Depiction of the Structural Distortion in Nb8W9O47 during Lithiation . . . . 79
3.8 BVSE Mapping of Nb8W9O47 . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.9 Window-Opening Cyclic Voltammetry of Nb8W9O47 . . . . . . . . . . . . . . 86
3.10 Nb8W9O47 GCPL Voltage Cutoff Comparison and Differential Capacity . . . 86
3.11 Representative Nb8W9O47 Survey XPS Spectrum . . . . . . . . . . . . . . . 88
3.12 Representative Nb8W9O47 Nb 3d XPS Spectrum . . . . . . . . . . . . . . . . 89
3.13 Representative Nb8W9O47 W 4f XPS Spectrum . . . . . . . . . . . . . . . . 89
3.14 In-House Operando XRD of Nb8W9O47: 1.2 V Cutoff . . . . . . . . . . . . . 90
3.15 In-House Operando XRD of Nb8W9O47: 1.0 V Cutoff . . . . . . . . . . . . . 91
3.16 In-House Operando XRD of Nb8W9O47: 1.2 V and 1.0 V Cutoff Comparison 92
3.17 Comparison of Lattice Parameter Evolution Trends from Different Fitting
Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
3.18 Comparison of Fitting Method Qualities . . . . . . . . . . . . . . . . . . . . 105
3.19 Comparison of Trend in Rwp for Different Fitting Methods . . . . . . . . . . 106
3.20 Refined Transition Metal Positions from Operando XRD of Nb8W9O47 . . . . 107
3.21 Crystal Structure of Nb8W9O47 with Transition Metal Sites Labeled . . . . . 108
3.22 Structural Distortion of Nb8W9O47 Viewed Down c-axis . . . . . . . . . . . 109
3.23 Structural Distortion of Nb8W9O47 Viewed Down a-axis . . . . . . . . . . . 109
4.1 BaLa2Fe2O7 Crystal Structure . . . . . . . . . . . . . . . . . . . . . . . . . . 113
4.2 Refined Pristine BaLa2Fe2O7 XRD Data . . . . . . . . . . . . . . . . . . . . 116
4.3 Comparison of XRD Data from PVDF Series and NH4HF2 Series . . . . . . 117
xiii

4.4 Proposed Fluorinated BaLa2Fe2O7 Crystal Structures . . . . . . . . . . . . . 118
4.5 Offset Stoichiometry Comparison of Refined XRD Data . . . . . . . . . . . . 120
4.6 Comparison of Extracted Lattice Parameters . . . . . . . . . . . . . . . . . . 121
4.7 Comparison of ATR-FTIR Data from PVDF Series and NH4HF2 Series . . . 123
4.8 Offset Stoichiometry Comparison of ATR-FTIR Data . . . . . . . . . . . . . 124
4.9 Refined BaLa2Fe2O7 + PVDF Oxyfluoride XRD Data . . . . . . . . . . . . . 141
4.10 Refined BaLa2Fe2O7 + NH4HF2 Oxyfluoride XRD Data: Set 1 . . . . . . . . 141
4.11 Refined BaLa2Fe2O7 + NH4HF2 Oxyfluoride XRD Data: Set 2 . . . . . . . . 141
4.12 Comparison of Extracted Lattice Parameters with Additional NH4HF2 Data 142
4.13 Graphical Representation of O:F Ratio Trends from EPMA . . . . . . . . . . 142
4.14 Comparison of ATR-FTIR Data from PVDF Series and NH4HF2 Series: Full
Wavenumber Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
5.1 Basic DEMS Cell Schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
5.2 DEMS Cell-to-MS Interface: Design Iteration 1 . . . . . . . . . . . . . . . . 157
5.3 ReO3 vs. Cu DEMS in Cell-to-MS Interface Design Iteration 1 . . . . . . . . 159
5.4 DEMS Cell-to-MS Interface: Design Iteration 2 . . . . . . . . . . . . . . . . 160
5.5 ReO3 vs. Cu DEMS in Cell-to-MS Interface Design Iteration 2 . . . . . . . . 162
5.6 ReO3 vs. Cu DEMS in Cell-to-MS Interface Design Iteration 3 . . . . . . . . 164
5.7 CsMnFeF6 vs. Bi/BiF3 DEMS in Cell-to-MS Interface Design Iteration 3 . . 166
5.8 ReO3 and CsMnFeF6 Crystal Structures . . . . . . . . . . . . . . . . . . . . 169
5.9 Refined ReO3 and CsMnFeF6 XRD Data . . . . . . . . . . . . . . . . . . . . 169
5.10 ReO3 vs. Cu DEMS in Cell-to-MS Interface Design Iteration 3: All Tracked
Mass Fragment Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
xiv

Curriculum Vitae
Education
2019-2024 Ph.D. in Chemistry
National Science Foundation Graduate Research Fellow
Department of Chemistry
University of Southern California, Los Angeles, CA
2015-2019 B.S. in Chemistry, Highest Honors
Department of Chemistry and Biochemistry
University of California Santa Barbara, Santa Barbara, CA
Publications
15. M. J. Brady,* J. L. Andrews,* E. T. McClure, S. S. Kim, R. C. Monger, K. K.
Jew, K. A. See, and B. C. Melot, Chemical Fluorination and Structural Evolution of
Ruddlesden-Popper BaLa2Fe2O7. (In preparation).
14. M. J. Brady, J. L. Andrews, R. C. Monger, E. Truong, A. Squires, Y.-Y. Hu, D.
O. Scanlon, B. C. Melot, Gallium Disorder Enables Facile Lithium-ion Transport in
Fluoride Garnets. (In preparation).
13. J. L. Andrews, M. J. Brady, C. T. Morrell, K. K. Jew, S. Sloan, K. A. See, and B.
C. Melot, On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type
Nb8W9O47. (Under Review at ACS Energy Lett.).
12. M. J. Brady,* J. L. Andrews,* A. Zambotti, D. Zhang, X. Yuan, X. Duan, Y. Li, J.
Nelson Weker, A. R. Balakrishna, K. A. See, R. Seshadri, A. Van der Ven, B. S. Dunn,
S. H. Tolbert, and B. C. Melot, Multiscale Approaches for Optimizing the Impact of
Strain on Na-ion Battery Cycle Life. (Under Review at Mat. Res. Bull.).
11. L. E. Robinson, J. Wang, H. Asare, J. L. Andrews, B. Tripathi, R. Katiyar, B. C.
Melot, R. J. Messinger, S. C. Jones, and W. C. West, Development of Fluoride-ion
Primary Batteries: The Electrochemical Defluorination of CFx. J. Phys. Chem. C
128 (2024) 14195–14205. [doi]
xv

10. S. S. Kim, D. N. Agyeman-Budu, J. J. Zak, J. L. Andrews, J. Li, A. Van der Ven, B.
C. Melot, J. Nelson Weker, and K. A. See, Effect of Metal d Band Position on Anion
Redox in Alkali-Rich Sulfides. Chem. Mater. 36 (2024) 6454–6463. [doi]
9. S. S. Kim, D. A. Kitchaev, E. S. Patheria, C. T. Morrell, J. L. Andrews, Q. Yan,
S.-T. Ko, J. Luo, B. C. Melot, A. Van der Ven, and K. A. See, Cation Vacancies Enable
Anion Redox in Li Cathodes. J. Am. Chem. Soc. 146 (2024) 20951–20962. [doi]
8. X. Li, S. S. Kim, M. D. Qian, E. S. Patheria, J. L. Andrews, C. T. Morrell, B.
C. Melot, and K. A. See, Reducing Voltage Hysteresis in Li-rich Sulfide Cathodes by
Incorporation of Mn. Chem. Mater. 36 (2024) 5687–5697. [doi]
7. J. L. Andrews, E. T. McClure, K. K. Jew, M. B. Preefer, A. Irshad, M. J. Lertola,
D. D. Robinson, C. Z. Salamat, M. J. Brady, L. F. J. Piper, S. H. Tolbert, J. Nelson
Weker, B. F. Chmelka, B. S. Dunn, S. R. Narayan, W. C. West, and B. C. Melot, Room
Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6. ACS Energy
Lett. 7 (2022) 2340–2348. [doi]
6. J. L. Andrews,* M. J. Brady,* E. T. McClure,* and B. C. Melot, Impact of Structural
Deformations on the Performance of Li-ion Insertion Hosts. Chem. Mater. 34 (2022)
4809–4820. [doi]
5. R. C. Vincent, Y. Luo, J. L. Andrews, A. Zohar, Y. Zhou, Q. Yan, E. M. Mozur,
M. B. Preefer, J. Nelson Weker, A. K. Cheetham, K. Luo, L. Pilon, B. C. Melot,
B. S. Dunn, and R. Seshadri, High-Rate Lithium Cycling and Structure Evolution in
Mo4O11. Chem. Mater. 34 (2022) 4122–4133. [doi]
4. J. L. Andrews, J. P. de los Rios, M. Rayaluru, S. Lee, L. Mai, A. Schusser, and C.
H. Mak, Experimenting with At-Home General Chemistry Labs During the COVID-19
Pandemic. J. Chem. Educ. 97 (2020) 1887–1894. [doi]
3. H. A. Evans, Z. Deng, I. E. Collings, Y. Wu, J. L. Andrews, K. Pilar, J. M. Tuffnell,
G. Wu, J. Wang, S. E. Dutton, P. D. Bristowe, R. Seshadri, and A. K. Cheetham,
Polymorphism in M (H2PO2)3 (M = V, Al, Ga) compounds with the perovskite-related
ReO3 structure. Chem. Commun. 55 (2019) 2964–2967. [doi]
2. H. A. Evans, J. L. Andrews; D. H. Fabini, M. B. Preefer, G. Wu, F. Wudl, A. K.
Cheetham, and R. Seshadri, The capricious nature of iodine catenation in I2 excess,
perovskite-derived hybrid Pt(IV) compounds. Chem. Commun. 55 (2019) 588–591.
[doi]
1. H. A. Evans, D. H. Fabini, J. L. Andrews; M. Koerner, M. B. Preefer, G. Wu,
F. Wudl, A. K. Cheetham, and R. Seshadri, Hydrogen Bonding Controls Structural
Evolution in Perovskite-Related Hybrid Platinum (IV) Iodides. Inorg. Chem. 57
(2018) 10375–10382. [doi]
xvi

Abstract
Identifying and understanding the structural deformations induced by the (de)intercalation
and diffusion of ions is critical for the development of improved energy storage materials.
This dissertation is comprised of a collection of research studies that investigate both cationic
and anionic intercalation-driven structural transformations in host crystal structures, toward
the goal of understanding their impact on overall battery performance.
First, I present a study on the electrochemical (de)fluorination of CsMnFeF6 from a
liquid electrolyte to investigate the potential of the cubic pyrochlore crystal structure as
a fluoride intercalation host. Up to one fluoride ion can be reversibly (de)inserted into
mechanochemically synthesized CsMnFeF6 for multiple cycles at room temperature. The
structural transformation during cycling begins as a subtle, but clearly reversible, expansion
and contraction of the CsMnFeF6 cubic lattice on fluoride insertion and removal, respectively.
As cycling progresses, a topotactic transformation of CsMnFeF6 from the original defect
pyrochlore structure into a related, orthorhombic polytype is observed. While the distorted
material continues to reversibly cycle fluoride ions for multiple cycles, significant capacity
fade is observed soon after, raising the question of whether the structural distortion or other
cell components (i.e., electrolyte stability) negatively impact the cycling performance.
This is followed by an investigation into the structural evolution of tetragonal tungsten bronze-type Nb8W9O47 during lithium (de)intercalation to identify the origin of fast Li
cycling in bronze and bronze-derived materials. Operando X-ray diffraction studies reveal
that the space group is unchanged and the unit cell parameters vary anisotropically during
discharge. Although higher capacities are obtained by cycling to lower voltages, this incurs
significantly greater unit cell volume expansion and negatively impacts the reversibility. Furthermore, the variation in refined transition metal positions indicates charge compensation
occurs through off-centering of the metals within their octahedra, suggesting that rigid crysxvii

tal structures, like that of Nb8W9O47, exhibit such impressive cycling capabilities because
they undergo displacive rather than rotational deformations during (de)lithiation.
I then revisit fluoride (de)intercalation with a study on the chemical fluorination of
Ruddlesden-Popper BaLa2Fe2O7, focused on the impact of the fluorinating agent identity
and the fluorination conditions on the structural evolution of the oxide and product distribution. By investigating the structural distortions induced by chemical fluorination, this study
offers insight into the changes in structure that might be observed when BaLa2Fe2O7 is electrochemically fluorinated. Two oxyfluoride series were synthesized by reacting BaLa2Fe2O7
with either polyvinylidene fluoride (PVDF) or ammonium bifluoride (NH4HF2). A symmetry lowering from the pristine oxide’s tetragonal unit cell to an orthorhombic unit cell
is observed at low levels of fluorinating agent, however, the original tetragonal symmetry
is regained with increasing amounts of fluorine. Both fluorinating agents induce the same
structural transformation, but NH4HF2 produces lower purity samples and larger amounts of
NH4HF2 are required to impart the same structural changes as PVDF, underscoring the importance of optimized synthetic conditions when leveraging chemical fluorination techniques
to investigate fluoride intercalation mechanisms.
Finally, I describe the design and testing of an F-ion compatible continuous flow differential electrochemical mass spectrometry (DEMS) cell and cell-to-mass spectrometer interface.
Developing operando techniques, like DEMS, is crucial for the advanced characterization of
electrochemical reactions in F-ion batteries, especially for those hypothesized to be limiting
the advancement of these systems. The electrochemical fluorination of ReO3 from a liquid
electrolyte is first used as a “control” reaction for cell testing and adaptation because it
consistently results in oxidative H2 evolution due to electrolyte degradation. Then, the finalized DEMS set-up is used to investigate the reversible, electrochemical (de)fluorination
of CsMnFeF6, which is found to similarly exhibit H2 evolution on oxidation, indicative of
electrolyte instability in the employed voltage window.
xviii

Chapter 1
Introduction
1.1 A Brief Overview of the Energy Crisis
The constantly growing global demand for energy and the inherent issues of conventional,
non-renewable energy sources like fossil fuels (i.e., coal, natural gas, petroleum) are continually accelerating the ongoing energy crisis. Along with the critical problem of having a finite
supply of non-renewable resources despite the ever-increasing demand, the greenhouse gas
emissions produced from burning fossil fuels have induced changes to our climate and ecosystems that are unparalleled in recent human history. The 2023 Intergovernmental Panel on
Climate Change (IPCC) report notes that every 0.5 ◦C the global temperature rises above
the current warming level will noticeably increase the occurrence and severity of dangerous
weather events, such as heat extremes, regional droughts, and periods of heavy rainfall.1
The IPCC report also emphasizes that human influence on the climate is undeniable, urging
that the key to avoiding further environmental degradation and securing a net-zero carbon
emissions future is to rapidly shift away from burning fossil fuels and transform our energy
systems to rely on renewable resources.1
Thus, the United States is in the midst of a major transition from reliance on fossil fuels
towards the utilization of renewable resources, such as solar, wind, and tidal sources. While
renewable resources offer the promise of unlimited supply to meet our increasing energy de1

Introduction
mands, the intermittent nature of these energy sources (i.e., no power from solar panels at
night) means power generation fluctuates significantly throughout a given day. This presents
the critical impossibility of synchronizing peak energy production with peak demand, emphasizing the pressing need to be able to safely store this energy when it is produced so
that it can be used at any time.2 Developing and implementing advanced, widespread energy storage systems is the best path forward to effectively utilizing renewable energy whilst
decarbonizing the energy production and transportation sectors to avoid continued climate
change.
Rechargeable batteries, particularly lithium-ion (Li-ion) batteries, already play a critical
role in storing energy and supplying electricity as we begin to address the challenge of
replacing carbon-dependent energy systems with those that rely on renewable resources.3
While Li-ion batteries offer ideal properties, such as large operating potentials and slow
self-discharge rates,4
the scarcity and geopolitics associated with metals critical to their
production (i.e., Li, Ni, Co) will make larger scale deployment of Li-ion batteries extremely
challenging.5 Furthermore, the increasing demand for Li-ion batteries—fueled by the pressure
of needing grid-scale energy storage systems and the growing popularity of electric vehicles—
has resulted in an increase in the price of Li-ion cells for the first time after many years of
steady decline. This can be attributed, in large part, to the price of essential Li-containing
precursors (i.e., Li2CO3, LiOH) skyrocketing past that of the transition metal redox-centers
(i.e., Ni, Co). Furthermore, the vast geopolitical and human rights issues associated with the
mining of transition metals—for example, child labor and slavery in the cobalt mines of the
Democratic Republic of Congo6—emphasize the need to implement more widely available
metals like iron and manganese in developing rechargeable batteries for grid-scale energy
storage.
With economics and the environment in mind, there has been a tremendous research
effort dedicated to replacing Li with more abundant cations like Na,7 Mg,8 and even K.9
However, the possibility of leveraging negatively charged anions, like fluoride, has largely
2

1.1. A Brief Overview of the Energy Crisis
been overlooked.10 Fluoride is mainly produced from the mineral fluorspar, CaF2, which
is widely available throughout North America, Asia, Europe, and Africa, thereby offering
significant supply chain stability.11 According to the 2024 US Geological Survey report, the
United States alone is estimated to have over 72 million tons of fluoride available domestically
in the form of fluorspar compared to 14 million tons of lithium, presumably stored in ores and
geothermal brines.12,13 In 2024, the price of Li2CO3 remains high at $46,000 per metric ton,
whereas the price of acid-grade fluorspar remains steady near $430 per metric ton.12,13 Given
the vast abundance of fluoride and its comparable electrochemical properties to lithium,
fluoride-ion batteries are a promising alternative to pursue in the advancement of energy
storage systems.
While electrode material advances, cell engineering breakthroughs, and improvements
in manufacturing methods have vastly improved the modern Li-ion battery compared to
the original technology, current Li-ion cells are also quickly approaching their theoretical
limits.14 Furthermore, demands for longer cycling lifespans, improved durability, and fastcharging capabilities are rapidly increasing as we continue to incorporate Li-ion batteries
into the transportation sector and our overall energy infrastructure.15 Fundamental battery
research is critical to developing Li-ion batteries and next-generation alternatives that can
meet these growing demands for improved performance metrics as well as easily slot into
the expanding number of applications for rechargeables. A combined effort between the
government and industry will clearly be required to: (1) Secure the raw materials needed
to create and implement battery-based energy storage systems, and (2) Fund or perform
research to increase the variety of electrode materials and new types of chemistries (i.e.,
Na-ion or F-ion) used in rechargeables to ensure more efficient raw material usage.
3

Introduction
1.2 Intercalation-based Energy Storage
The diffusion of ions through the solid state is critical to numerous modern technologies,
making it a heavily studied topic in the materials chemistry community. Of the technologies
that rely on ionic diffusion, one of the most important is arguably the rechargeable battery.16 Enabling the reversible storage and delivery of energy, modern rechargeable batteries,
namely lithium-ion (Li-ion) batteries, are a critical component in portable electronics, electric vehicles, and systems for large-scale storage of energy produced by renewable sources,
like solar.17 In fact, in 2019, the significance of the Li-ion battery was acknowledged with
the awarding of a Nobel Prize in chemistry to Wittingham, Goodenough, and Yoshino for
the development of this critical technology.17,18
1.2.1 Li-ion Batteries
In the most general sense, a battery cell is comprised of an anode (negative electrode),
cathode (positive electrode), separator, and electrolyte.19,20 The separator serves as an
electrically insulating barrier between the anode and cathode but is soaked in electrolyte
to enable ionic conduction between the two electrodes. Current state-of-the-art cathodes
in Li-ion batteries are composed of crystalline layered oxide materials, such as LiCoO2,
21
LiNi1/3Co1/3Mn1/3O2,
22 and LiNi0.70Co0.15Al0.15O2,
23 or crystalline polyanionic compounds,
like LiFePO4,
24,25 blended with conductive carbon and a polymer binder and adhered to a
metal foil current collector. Commercial Li-ion battery anodes consist of graphitic carbon
similarly blended with a polymer binder and adhered to a metal current collector. The separator is typically a polymer film or glass fiber pad soaked in the liquid electrolyte composed
of a Li-containing salt, such as LiPF6, dissolved in carbonate or ethereal solvents.21,26,27 A
simple depiction of a generalized Li-ion cell—with a layered oxide cathode, graphite anode,
and a separator soaked in electrolyte—is shown in Figure 1.1.
When a Li-ion battery is discharged, Li+ ions diffuse from the anode to the cathode
4

1.2. Intercalation-based Energy Storage
Figure 1.1: A diagram of a rechargeable Li-ion battery with a graphite anode, a layered
oxide cathode, and a separator soaked in electrolyte. This depiction shows the cell undergoing
discharge, wherein Li+ ions (yellow) diffuse through the separator/electrolyte from the anode
to the cathode and electrons flow through the external circuit to power a load.
through the separator/electrolyte while the corresponding electrons flow through an external circuit, thereby providing electrical power.17 Conversely, to charge a Li-ion battery, an
electrical current forces electrons to flow in the opposite direction through the external circuit while Li+ ions simultaneously diffuse from the cathode to the anode.17 The diffusion of
Li+ ions and flow of current corresponds to redox reactions occurring within the electrodes,
wherein a redox-active center changes in oxidation state by releasing or accepting electrons.
An example of this is described by the following reactions:
charge LiM3+O2 −→ M4+O2 + Li+ + e
−
discharge M4+O2 + Li+ + e
− −→ LiM3+O2
5

Introduction
where M is the redox-active transition metal site in a generic layered oxide cathode, LiMO2.
Li-ion batteries are rechargeable, meaning they can reversibly go through many of these
charge–discharge “cycles” and the capacity of a given electrode material (charge per unit
mass or volume) depends on the number of Li+ ions that are inserted (discharge) or removed
(charge). Thus, an important metric is the number of these cycles that can be performed
reversibly with minimal capacity fade, dubbed the “cycle life”, which is heavily influenced
by the mechanism of Li+ intercalation into the host crystal structure.
1.2.1.1 Cationic Intercalation
Energy storage via the diffusion of Li+ ions between electrodes can occur via multiple different
reaction mechanisms, including conversion,28 alloying,29 and intercalation.30 While all three
mechanisms are of interest to the battery research community, this dissertation focuses on
materials that undergo intercalation.
In an intercalation reaction, a guest species is reversibly inserted into a vacancy in a
solid host material, driven by either chemical or electrical forces.31 Intercalation reactions
played a critical role in the development of rechargeable batteries because they are reversible,
meaning the guest ions can be deintercalated and the host material can be, at least partially,
restored to its original structure and oxidation state. For intercalation-based batteries this
is an ionic species—such as Li+ in a Li-ion battery—that occupies a vacant crystallographic
site within the host and concurrently results in a change in oxidation state at the host’s redox
active center. For example, in the generic layered oxide MO2, the M4+ is reduced to M3+
upon Li+ ion intercalation (see the discharge reaction described previously). The sites into
which ions are intercalated vary between different materials, but are often found as motifs
like the space between layers that forms the channels found in the layered transition metal
oxides21,31 and graphite32,33 or the windows and tunnels found in spinels,34,35 polyanionic
materials,29,36,37 Wadsley-Roths,38–42 and bronzes.31,41,43,44
The changes in oxidation state that occur upon ionic (de)intercalation lead to changes
6

1.2. Intercalation-based Energy Storage
in bond lengths, bond angles, and electrostatic environments around the redox-active site,
yielding structural distortions in the host materials.45 Furthermore, ionic species must also
satisfy their own bonding requirements, and as a result, the insertion, diffusion, and removal
of ions often also induce host crystal structure changes of varying degrees depending on the
size and charge of the ion. With the insertion or removal of ions also comes the insertion or
removal of the electrons coupled to these ions, thereby changing the occupancy of electronic
states in the host material and, in turn, the associated properties (i.e., electronic conductivity) as a function of the amount of intercalant.46,47 All of these changes can significantly
impact the performance of electrode materials, therefore, developing a deeper understanding
of them is critical to the development of advanced rechargeable batteries.
1.2.1.2 Examples of Intercalation-Driven of Structural Distortions
As introduced in the previous section, the distortions that a host crystal structure experiences
during lithium (de)intercalation can significantly impact rate performance and reversibility.45
The previous section also stated very simply that “in an intercalation reaction, a guest species
is reversibly inserted into a vacancy in a solid host.” However, it is not always this simple.
Solid-state ionic diffusion is often depicted as the migration of ions between vacancies in a
crystal lattice,48,49 but this hopping is strongly influenced by the electrostatic interactions
between the intercalant (Li+ ions) and the framework (lattice anions).50–52 Therefore, the
vacancy must be able to solvate the intercalated ion without binding it so tightly that it
cannot diffuse through and out of the structure. With this in mind, just any empty space
within a crystal structure is not necessarily capable of facilitating ionic transport, and even if
intercalation and diffusion can occur, sometimes these processes trigger irreparable structural
distortions.
The layered rock-salt oxide LiCoO2, the canonical Li-ion battery cathode material, is
composed of slabs of edge-sharing CoO6 octahedra separated by layers of positively charged
lithium ions that screen the electrostatic repulsion between anions and provide channels
7

Introduction
Figure 1.2: As lithium is extracted from the layered rock-salt oxide cathode material, LiCoO2,
shown in (a), the layers slip, producing the polymorph shown in (b) that exhibits limited
lithium re-insertion. In this depiction, lithium is yellow, cobalt is blue, and oxygen is gold.
for facile lithium diffusion (Fig. 1.2a). However, as lithium is deintercalated, electrostatic
repulsion between the layers leads to an elongation of the c-axis (stacking direction) of the
unit cell.53 This expansion continues until all the lithium is removed, at which point the
layers slip. This produces a polymorph that is unlikely to undergo lithium intercalation
because the vacant lithium site now shares parallel faces with the CoO6 octahedra, which
is an energetically unfavorable coordination environment due to strong repulsion between
lithium and cobalt at these small distances (Fig. 1.2b).54 While this problematic distortion
is a crucial failure mechanism for layered oxide cathodes, it can be minimized by limiting
the extent of delithiation, albeit at the cost of lower practical capacity.
Following the emergence of the layered oxide cathodes, researchers turned their attention towards polyanionic compounds. These materials are advantageous for the following
reasons: (1) The robust crystal structures are stable over extended cycling; (2) Inductive
effects from the polyanionic groups increase redox potentials; (3) Versatile frameworks provide access to a wide range of compositions.55–57 Most relevant to this dissertation, the
contrast between the rigid (strongly covalent) main group polyhedra and the deformable
(more ionic) transition metal polyhedra creates hinges around which polyhedra pivot during
lithium (de)intercalation.
The canonical polyanionic electrode, LiFePO4, was first reported by Goodenough and
co-workers in 1997 and exhibits exceptionally fast rate performance despite undergoing a
8

1.2. Intercalation-based Energy Storage
Figure 1.3: As lithium is extracted from the polyanionic cathode material, LiFePO4, shown
in (a), polyhedral rotations occur parallel to the lithium diffusion channel, producing the
structure shown in (b). In this depiction, lithium is yellow, iron is rust red, phosphorous is
lavender, and oxygen is gold.
notable structural distortion during cycling (Fig. 1.3).24,25,58 The extraction of lithium from
LiFePO4 results in cooperative rotation of the polyanionic PO2−
4
groups and a unit cell
volume contraction.58,59 While this is a significant change, the polyhedral rotations occur
parallel to the lithium diffusion pathways and therefore do not interrupt the movement of
Li+ ions through the structure, showing that distortions do not always lead to cell failure.
However, correlated polyhedral rotations often manifest as less reversible transformations that can severely hinder lithium diffusion. A key example of this scenario is lithium
(de)intercalation in ReO3. Crystallizing in a defect perovskite structure, the structure of
ReO3 is composed of a cubic network of corner-sharing octahedra with a vacancy on the A
cation site that creates a 3D network of interstitials (Fig. 1.4a). In 1981, Murphy and coworkers demonstrated the chemical lithiation of ReO3, intercalating up to two lithium ions
per formula unit.60 More recently, Bashian et al. investigated electrochemical (de)lithiation
of ReO3 and observed unit cell volume contraction and a significant structural distortion
upon lithiation, driven by correlated rotations of the octahedra (Fig. 1.4b,c).61 These rotations occur to better coordinate the small lithium ions within the large A-site vacancy,
emphasizing the importance of ideal interstitial size for a given intercalant.60–62 While it is
reversible, the magnitude of this distortion is detrimental to the electrochemical performance
of ReO3, which is unsurprising given large volume changes can damage electrodes.63–65
9

Introduction
Figure 1.4: As lithium is inserted into perovskitic ReO3, shown in (a), correlated polyhedral
rotations occur to better coordinate lithium ions within the vacant A-site, shown in (b) and
(c). In this depiction, lithium is yellow, rhenium is maroon, and oxygen is gold.
As early as 1983, Cava et al. acknowledged the negative impact of this type of transformation on ionic mobility and suggested that crystallographic shearing could stabilize
the corner-shared ReO3 framework against severe distortion by avoiding octahedral rotations at the edge-sharing boundaries of the shear plane.38 Therefore, it is unsurprising that
shear structures have attracted an increasing amount of attention as Li-ion intercalation
hosts.39,42,43,47,66,67 One family of shear phases derived from ReO3 are the “Wadsley–Roth”
structures, where the removal of a plane of the oxide ions (the crystallographic shear) yields
alternating layers of edge-sharing and corner-sharing octahedral units.68–76 For example,
Wadsley-Roth TiNb2O7 is composed of blocks of corner-connected octahedra, where the
blocks are three octahedra wide by three octahedra long and are infinitely connected in
each direction by crystallographic shear planes (Fig. 1.5).77 TiNb2O7 exhibits robust cycling
capabilities, including impressive capacity retention at high rates, at least in part because
the orthogonal shear planes limit structural distortions like correlated polyhedral rotations
during (de)lithiation.43,78,79
Nb3O7F, a shear-derivative of perovskitic NbO2F, is another example of a Wadsley-Roth
phase with crystallographic shearing along a single crystallographic axis.80–85 Bashian et
al. compared the structure and electrochemistry of the two phases and found that the
lithiation of NbO2F results in octahedral rotation, whereas the shear planes in Nb3O7F
suppress rotation.66 While the crystallographic shear stabilizes Nb3O7F relative to NbO2F,
10

1.2. Intercalation-based Energy Storage
Figure 1.5: (a) The crystal structure of Wadsley-Roth TiNb2O7, where the alternating 3×3
corner-sharing blocks (ReO3-type) are highlighted with pink and blue squares. (b) The crystal structure of lithiated TiNb2O7. Corner-sharing blocks are still highlighted to show the
lack of distortion upon lithium insertion. In this depiction, lithium is yellow, titanium/niobium sites are green, and oxygen is gold.
the 1D nature of the shear leaves the structure susceptible to stacking distortions that
disrupt the lithium diffusion channels, suggesting multi-dimensional shearing is needed to
fully stabilize ReO3-derived structures.
An alternative avenue for imparting structural rigidity in ReO3-derived phases is the
“pentagonal column” structural motif seen in variations of the tetragonal tungsten bronze
crystal structure.86–88 Griffith et al. reported on the fast cycling of a bronze-type Nb2O5
polymorph, attributing the impressive rate performance to structural rigidity imbued by the
pentagonal columns upon (de)lithiation.40 The pentagonal column motif will be discussed
further in Section 1.5 of this chapter, and, how this crystallographic motif imparts structural
stability and enables fast Li-ion cycling will be discussed in detail in Chapter 3.
1.2.2 F-ion Batteries
Historically, research on intercalation-based energy storage has heavily focused on understanding and developing systems that leverage cationic intercalants, especially Li+. Anionic
species can also be intercalated into solids, however, and the concept of anionic intercalation
is not new. In fact, it was studied as early as 1938 when R¨udorff and Hoffman reported the
insertion of bulky anionic groups, such as SO2−
4
into graphite.89 But, in comparison to the
research on cationic species, significantly less work has been done to identify mechanisms of
11

Introduction
anionic intercalation and how it impacts host crystal structures.10
Thus far, the smallest anion on the periodic table, fluoride (F−), shows the most promise
as a candidate anionic intercalant for next-generation energy storage applications. Fluorideion (F-ion) batteries are of particular interest because of their potential to match the energy
density of Li-based chemistries whilst reducing or eliminating our dependence on strained
resources required for commercialized Li-ion batteries (i.e., Li, Ni, Co).90 Furthermore, fluorine is extremely abundant in the earth’s crust, primarily found as fluorspar (CaF2) alongside
limestone deposits throughout four different continents: North America, Asia, Europe, and
Africa.11
Despite the dramatic shift in chemistry inherent to moving negatively charged carriers
rather than positive ones, F-ion batteries still retain many of the fundamental properties that
make Li-ion batteries so useful. For example, Li-ion batteries are stable at high voltages
due to, in part, the high reductive stability of Li+. Being the ionic species of the most
electronegative element, F− ions exhibit high oxidative stability and can therefore be used in
large voltage windows and with higher-voltage redox pairs. Much like lithium, fluorine is also
the lightest and smallest in its group on the periodic table, resulting in higher energy densities
and faster ionic transport compared to other anionic intercalants, like Cl−.
91,92 Given the vast
abundance of fluorine and its potential for comparable electrochemical properties to lithium,
the benefit of pursuing F-ion batteries as a low-cost alternative to Li-ion technologies is clear.
1.2.2.1 Conversion-Based Systems
Thus far, the vast majority of reported F-ion batteries rely on conversion reactions, in which
a binary metal fluoride cathode is reversibly reduced to the corresponding metal and a metal
anode is reversibly oxidized to its fluoride salt.93–96 In 2011, Reddy and Fichtner published
the first report of a rechargeable F-ion battery, which relied on this type of conversion-based
chemistry.93 Although conversion reactions are technically reversible, the process of transforming the electrode material from a metal to a metal fluoride, and vice versa, typically
12

1.2. Intercalation-based Energy Storage
incurs dramatic structural changes that involve bond breaking/forming and unit cell expansion/contraction. As a result, repeated cycling of conversion-based materials typically
results in large electrode volume changes, which in turn results in poor cycling stability and
irreversibility over multiple cycles due to electrode cracking and damage. This issue of short
cycle life, as well as low practical capacities, plague conversion-based F-ion batteries and
have hereto precluded the application of such cells.
1.2.2.2 Intercalation-Based Systems
F-ion batteries based on fluoride intercalation, wherein the host material may experience
crystal structure changes but should broadly maintain its original topology, have more recently been reported. Compared to the metal/metal fluoride conversion reactions, fluoride
intercalation requires significantly less atomic rearrangement. Therefore, it is not surprising that intercalation-based systems typically offer improved battery performance, such as
better rate performance and longer cycling lifetimes.
Clemens and co-workers presented some of the first reports of intercalation-based Fion batteries with their demonstration of electrochemical fluoride (de)intercalation into the
Ruddlesden-Popper oxides LaSrMnO4 and La2CoO4.
97,98 Ruddlesden-Popper phases were
chosen by the authors because the crystal structure contains channels hypothesized to be
ideal for fluoride (de)intercalation and diffusion. Figure 1.6 presents a simple depiction of a
generalized, intercalation-based F-ion cell with a Ruddlesden-Popper oxide anode, a metal
fluoride cathode, and a separator/electrolyte.
This brings up a key limitation for F-ion batteries: the electrolyte. The aforementioned
intercalation-based work of Nowroozi et al.,
97,98 and even the conversion-based work of Reddy
and Fichtner,93 relied on a solid-state electrolyte (Ba-doped LaF3) in an all-solid-state cell
geometry. The issue lies in the poor room-temperature fluoride ion conductivity of Ba-doped
LaF3, which requires operating temperatures of 150 ◦C or higher to exhibit sufficient conductivity for fluoride ion cycling.99–101 While there are more recent reports of solid electrolytes
13

Introduction
Figure 1.6: A diagram of a rechargeable F-ion battery with a Ruddlesden-Popper oxide
anode, a metal fluoride cathode, and a separator soaked in electrolyte or a solid electrolyte.
This depiction shows the cell undergoing discharge, wherein F− ions (blue) diffuse through
the electrolyte from the cathode to the anode and electrons flow through the external circuit
to power a load.
that exhibit impressive fluoride conductivity at ambient temperatures, namely PbSnF4 and
BaSnF4, they have only been reported in conjunction with conversion-based electrodes that
suffer from significant capacity fade in the first few cycles.102–104
Thus, poor reversibility and high operating temperature requirements have precluded any
practical application of these promising intercalation-based F-ion batteries, emphasizing the
need to develop electrolytes that exhibit fluoride mobility at ambient temperatures alongside electrode materials that can reversibly (de)intercalate fluoride. In 2018, Davis et al.
made significant progress in the field of F-ion electrolytes, reporting reversible and stable
cycling of conversion-based CuF2@LaF3 core–shell nanoparticles in a liquid electrolyte based
on a branched quaternary amine salt, N,N,N -trimethyl-N -neopentylammonium fluoride or
“Np1F”, dissolved in a fluoroether solvent.91 While this work focused on conversion-based
14

1.3. Operando X-ray Diffraction
chemistry, it presents an avenue to overcome the barriers that have thus far limited the study
of fluoride (de)intercalation. However, the Np1F salt is not commercially available, and obtaining the pure, dry salt requires a complex and expensive synthesis. Many recent reports
have focused on alternatively utilizing solutions of commercially available salts instead, such
as tetra-n-methylammonium fluoride (TMAF)105–107 and tetra-n-butylammonium fluoride
(TBAF).108,109 Chapter 2 will specifically focus on our report of fluoride (de)intercalation at
room temperature utilizing a TBAF-based liquid electrolyte,109 while Chapter 5 investigates
the stability of this electrolyte under electrochemical conditions. Together, these works provide some of the first substantial evidence that fluoride can be leveraged electrochemically
similar to lithium.
Considering the impact of Li-ion batteries on our modern technologies—a direct result of
focused research on cationic intercalation—the potential of leveraging anionic intercalation
for next-generation energy storage systems is extremely promising and intriguing. Expanding
the types of intercalants utilized in batteries would in turn expand the library of possible
host structures to explore for electrode active materials. However, while anionic intercalants
provide an exciting, alternative frontier to explore in developing new battery chemistries,
their larger ionic radii and negative charge mean diffusion through densely packed solids
requires significantly different conditions than for cations.110 Therefore, a pivotal starting
point in the exploration of anionic intercalation is identifying key structural features that
enable and promote solid-state anionic diffusion.
1.3 Operando X-ray Diffraction
Throughout this dissertation, the structural changes induced by the (de)intercalation of
cations or anions into host materials are investigated using multiple different characterization methods. Methods such as powder X-ray diffraction, X-ray absorption and emission
spectroscopy, Raman spectroscopy, and neutron scattering can yield valuable insight into
15

Introduction
the average and local structural changes that occur in electrode materials during cycling,
providing atomic-level insight into intercalation processes and their impact on host crystal
structures. Post-cycling or “ex situ” analysis—wherein the material of interest is cycled to a
specific state of charge, harvested from the cell, then studied using one of the aforementioned
techniques (or others not listed)—can be an incredibly powerful tool in the investigation of
intercalation-driven structural transformations. However, preparing electrodes for ex situ
analysis involves disassembling cells and rinsing the harvested electrodes with organic solvents, often followed by days to weeks before measurement. This can be problematic as the
active material in electrochemically cycled and harvested electrodes may be highly reactive
or kinetically unstable and could degrade during this wait time, making it difficult to know
if the measured sample accurately represents the structure as it is within the battery during
cycling.
This issue can be avoided using operando or in situ measurements that make use of
electrochemical cells specially designed to perform a given measurement during operation,
enabling the direct correlation of electrochemical behavior with structural changes. For
example, in-house operando powder X-ray diffraction (XRD) can be performed in a custommade electrochemical cell equipped with a beryllium window, as shown in Figure 1.7. For a
typical lab diffractometer in reflection geometry, the electrode containing the active material
of interest is placed directly onto the beryllium, and the cell is oriented so that the window
is the topmost component of the cell stack (closest to the X-ray source, Fig. 1.7a). Beryllium is both conductive and light enough in atomic mass to allow for X-ray penetration,
allowing it to serve as both the current collector and an X-ray transparent window so that
diffraction patterns are collected continuously during electrochemical cycling. This provides
a real-time, detailed picture of the structural changes that occur in the active material of
interest in response to ionic (de)intercalation, thereby providing more information during
the intermediate phases of cycling, avoiding the risk of reactivity and/or relaxation in ex
situ samples, and sometimes even capturing phases that may not otherwise be seen due to
16

1.3. Operando X-ray Diffraction
Figure 1.7: Operando powder X-ray diffraction cell schematics showing (a) the cell stack
when used in a reflection geometry, (b) the cell stack when used in a transmission geometry,
and (c) the cell stack within a CAD representation of the custom electrochemical cell.
these issues. A combination of operando and ex situ measurement methods were employed
in the work reported in this dissertation, both of which helped to understand the structural
impacts of ionic (de)intercalation and develop a greater understanding of electrode material
design principles.
In addition to in-house methods, operando XRD measurements are often performed at
synchrotron X-ray sources, such as those available at the Advanced Photon Source (APS) at
Argonne National Lab and the Stanford Synchrotron Radiation Lightsource (SSRL) at SLAC
National Accelerator Laboratory. While in-house measurements are incredibly convenient,
the data obtained from synchrotron sources is far superior for advanced structural charac17

Introduction
terization. For example, the higher flux of the synchrotron source produces higher intensity
diffraction patterns, which in turn can provide more information on structural transformations, especially those that are very subtle and might be missed in lab-resolution diffraction
data. Synchrotron sources also provide superior time resolution than lab sources, with
high-quality patterns collected in seconds rather than minutes, allowing for more accurate
correlation of the voltage profile and structural transformations, especially for fast-cycling
systems.
The previously described cell used for in-house, reflection measurements can easily be
adapted to be used in transmission mode, where the X-ray beam passes directly through
the cell stack (Fig. 1.7b), and taken to SSRL for operando synchrotron XRD experiments.
As shown in Figure 1.7, to go from reflection to transmission the cell is rotated 90◦
so that
the active material of interest is closest to the detector (Fig. 1.7b). In addition to this
reorientation, the steel current collector (counter electrode side) is replaced with a second
beryllium window. The additional beryllium window, as well as a hole bored through the
plunger, creates an aperture through which the X-ray beam enters the cell and passes through
the cell stack. At the APS, operando cells called “AMPIX” (Argonne Multi-Purpose In situ
X-ray) cells are provided and also designed to operate in transmission mode, but with glassy
carbon windows rather than beryllium.111
1.4 Differential Electrochemical Mass Spectrometry
In order to develop high-performance rechargeable batteries, the four key components—
cathode, anode, electrolyte, and separator—and their interactions must be understood and
optimized.112 The electrochemical reactions occurring in and between these components are
responsible for the properties exhibited by Li-ion batteries and emerging alternatives, like
F-ion batteries. Therefore, developing and adapting methods for the real-time observation
of electrochemical reactions in next-generation batteries based on new chemistries will be
18

1.4. Differential Electrochemical Mass Spectrometry
pivotal for realizing systems that can outperform state-of-the-art Li-ion batteries.3
As discussed previously, ex situ analyses have been performed for decades but the cell
disassembly and sample preparation steps are fraught with challenges. To continue advancing
Li-ion batteries and alternative battery technologies, operando methods must be developed
and applied to truly understand the electrochemical behavior observed in a given cell during
operation. It is arguable that the development of Li-ion alternatives, like F-ion batteries,
is impeded by an insufficient understanding of how ion storage behavior and electrolyte
stability might correlate to gas evolution reactions in novel electrochemical cells.
Differential electrochemical mass spectrometry (DEMS), alternatively known as online
electrochemical mass spectrometry, combines electrochemistry with mass spectrometry to detect gaseous reactants, reaction intermediates, or products that form during electrochemical
reactions.113 Over 50 years ago, Bruckenstein and Gadde reported the collection of gaseous
products from electrochemical reactions using a vacuum and their subsequent detection and
identification using mass spectrometry.114 In a modern, continuous flow DEMS measurement,
an inert carrier gas flows through a specially designed electrochemical cell during cycling to
collect and analyze evolved gases using mass spectrometry with very little time delay. This
allows for mass-resolved identification of gaseous or volatile species and direct correlation
of the voltage profile to gas evolution events. Clearly, this operando technique can provide
insight into the interactions between cell components under electrochemical conditions and
how they might impact cycling performance. For example, DEMS can help identify deleterious side reactions, such as electrolyte degradation, offering more information on these
processes than electrochemical techniques alone.
A typical DEMS setup is comprised of an electrochemical cell equipped for gas flow in and
out, a potentiostat, and a mass spectrometer, wherein the two instruments are connected
to a computer for data collection during experimentation. A schematic of the custommade, continuous flow DEMS cell used for the work reported in this dissertation, inspired
by the cell design reported by Berkes et al.,
115 is depicted in Figure 1.8. In this setup, the
19

Introduction
Figure 1.8: A schematic of the custom-made differential electrochemical mass spectrometry
cell and the cell/mass spectrometer interface design (MFC=mass flow controller, BPR=back
pressure regulator).
electrochemical cell and potentiostat are used to conduct electrochemical experiments on the
battery materials of interest. As shown in Figure 1.8, the cell inlet is connected to a line
of the carrier gas of choice, and the gas flow rate is regulated using a mass-flow controller
(MFC). Any volatile or gaseous species generated during the electrochemical reaction are
picked up by the carrier gas and flow through the cell outlet into the mass spectrometer,
where they are measured and identified based on their mass-to-charge ratio (m/z ).
Operando gas evolution analysis can play a critical role in identifying and understanding
relationships between the cell components and battery performance for the development
and eventual optimization of Li-ion alternatives. However, many of the DEMS cells in the
literature, like that of Berkes et al., are designed to be compatible with Li-ion chemistries
(i.e. resistant to carbonate-based electrolytes).115 Therefore, Chapter 5 focuses on the design
of a custom-made continuous flow DEMS cell, testing the cell with a “control” F-ion reaction,
and the iterative adaptation of the cell and cell/mass spectrometer interface for compatibility
20

1.5. Relevant Crystal Structures
with F-ion battery chemistries.
1.5 Relevant Crystal Structures
While a vast number of crystal structures are actively being explored for use in battery
electrodes, this dissertation will examine only a few different structure types: pyrochlore,
tetragonal tungsten bronze-type, and Ruddlesden-Popper. All of these structures feature
highly crystalline ordering creating either multi-dimensional channels or layers for ion diffusion and/or vacant crystallographic sites for ion intercalation.
1.5.1 Pyrochlore
The general formula for the pyrochlore crystal structure is A2B2X6Y, where the A and B
sites are filled by metal cations and X and Y are crystallographically distinct anionic sites.
Crystallizing in the cubic space group F d¯3m (#227), the “ideal” pyrochlore crystal structure
can be described as two interpenetrating, three-dimensional frameworks of B2X6 and A2Y
(Fig. 1.9).116 The B2X6 substructure is comprised of corner-sharing BX6 octahedra that form
a three-dimensional network of six-membered rings and, as a result, produce large hexagonal
cavities in the structure (Fig. 1.9b). The A2Y substructure is comprised of four-coordinate
Y anions and two-coordinate A cations (Fig. 1.9c), and penetrates the B2X6 framework so
that the A cations occupy the centers of the hexagonal cavities. Notably, there is also a
third, unoccupied anionic site, creating an ordered vacancy in the crystal structure.
The larger, eight-coordinate A site is typically occupied by an alkali or rare-earth metal
cation, and the smaller, six-coordinate B site is often occupied by a transition metal cation.
The X and Y anionic sites can be filled by a variety of anions (i.e., oxygen, fluorine, chlorine,
hydroxide, sulfide), and the two sites can be filled by the same anion or two different anions.
Calculations of the interaction energy between the B2X6 and A2Y networks show that the
B2X6 substructure is more stable than the A2Y substructure.118 As a result, various defect
21

Introduction
Figure 1.9: (a) Crystal structure of a generic cubic pyrochlore, A2B2X6Y, viewed down the
[110] direction. In this depiction, the A site is purple, B is gray, X is gold, and Y is blue. (b)
The B2X6 substructure. (c) The A2Y substructure. Crystal structures are depicted using
VESTA.117
pyrochlores form readily because this network can be absent, as in the pyrochlore-type WO3
polymorph, W2O6 (B2X6),119 or partially occupied, as in CsMnFeF6 (ABB’X6).120 Notably,
ionic diffusion in pyrochlores is ascribed to vacancy hopping through a continuous path
of second-nearest-neighbor X sites.121,122 The required vacancies are thought to occur via
Frenkel defects, wherein an X site is vacant and the neighboring anionic vacancy is filled,
and the prevalence of these defects is reported to be increased by introducing cationic disorder
in the structure.121,123 Therefore, Chapter 2 of this dissertation investigates mixed B site,
defect pyrochlores (ABB’X6) as fluoride intercalation hosts based on the presence of anionic
vacancies intercalation sites and the hypothesis that the cationic disorder should correlate
to higher than expected room temperature fluoride conductivity.
1.5.2 Tetragonal Tungsten Bronze-type
The tetragonal tungsten bronze (TTB) crystal structure is named for the bronze phase
KxWO3, first reported in 1824.124 More generally, this family of materials is comprised of
non-stoichiometric compounds with the general formula AxMO3, where A is typically an
alkali metal (0 < x < 1) and M is a transition metal) crystallizing in a tetragonal space
group (Fig. 1.10a).125 The TTB crystal structure is a framework of regular and/or distorted,
corner-sharing MO6 octahedra, where variation in the M–O–M angles produces tunnels of
22

1.5. Relevant Crystal Structures
different sizes and shapes that are oriented along the short crystallographic axis (c-direction
in Fig. 1.10a) and occupied by the A cations.
Figure 1.10: (a) An example generic TTB crystal structure viewed down the [001] direction.
In this depiction, the A site is purple, M is gray, and O is gold. (b) The structure after tripling
the b lattice parameter. (c) The structure after filling 1/3 of the available pentagonal tunnel
sites. (d) The resulting generic TTB-type superstructure viewed down the [001] direction.
Crystal structures are depicted using VESTA.117
Chapter 3 of this dissertation presents an investigation of TTB superstructures, dubbed
“TTB-type” phases, as Li-ion intercalation hosts. The TTB-type superstructure of interest
can be derived by tripling the b-axis of a typical TTB unit cell (Fig. 1.10b), then filling
1/3 of the pentagonal tunnels in the new unit cell (Fig. 1.10c). Filling these pentagonal
tunnels with a transition metal, as in the composition Nb8W9O47 investigated in Chapter
3, creates 7-coordinate transition metal sites that are edge-sharing in the ab-plane with 5
other MX6 octahedra and corner-sharing with other MX7 bipyramids in the c-direction.
The result is often described as one-dimensional, infinite “pentagonal columns” that run
through the crystal structure (Fig. 1.10c).86 These columns are thought to provide increased
structural stability during lithium (de)intercalation by locking out correlated rotations of
the corner-connected polyhedra comprising the TTB-type framework, which should in turn
23

Introduction
enable highly reversible cycling.
1.5.3 Ruddlesden-Popper
Ruddlesden-Popper phases have the general formula An+1BnO3n+1, where the A site is filled
by an alkaline earth or rare earth metal cation and the B site is occupied by a transition
metal cation. Crystallizing in the tetragonal space group I4/mmm (#139), the RuddlesdenPopper crystal structure is best viewed as alternating perovskite (ABO3) and rock-salt (AO)
units, where the n in the general formula represents the number of layers of octahedra
in each perovskite block (Fig. 1.11a). The compositions studied in this dissertation are
n = 2 Ruddlesden-Popper phases, translating to a general formula of A3B2O7 and 2 layers
of octahedra in the perovskite blocks.
Figure 1.11: (a) Crystal structure of a generic, n =2 Ruddlesden-Popper oxide showing
the alternating rock-salt and perovskite slabs, viewed down the [100] direction. In this
depiction, the A site is purple, B is gray, and O is gold. (b) A depiction of the structural
effect of fluorination on a generic, n =2 Ruddlesden-Popper oxide. Crystal structures are
depicted using VESTA.117
Ruddlesden-Popper oxides are ideal candidates for reversible fluoride-ion intercalation
because the interweaving of perovskite and rock-salt layers within the crystal structure produces channels of anionic interstitial sites. Figure 1.11b depicts the two anionic sites into
which fluorine is expected to insert into an n = 2 Ruddlesden-Popper oxide: (1) replacing the
24

1.6. Dissertation Overview
apical oxygen of the BO6 polyhedra and (2) as interstitials within the rock-salt slabs. The
Ruddlesden-Popper structure type also allows for a large degree of compositional tunability,
allowing one to study the impact of different cations (i.e., different sizes, polarizability, and
redox couples) on fluoride intercalation into the structure. Clemens et al. have successfully
shown electrochemical intercalation of fluoride into Ruddlesden-Popper oxides oxyfluorides
utilizing all-solid-sate cell geometries.97,98,126 There are also many studies on the chemical
fluorination of Ruddlesden-Popper oxides, including Chapter 4 of this dissertation, using
fluorine gas127,128 or through the thermal decomposition of fluorinated polymers129–136 and
other solid fluorinating agents (i.e., ammonium bifluoride)137,138 for increased safety and
greater control over the fluorine content of the final product.
1.6 Dissertation Overview
This dissertation encompasses a set of complementary research studies focused on investigating ion intercalation-driven structural transformations in rechargeable battery electrode
materials. More specifically, the presented work investigates and describes the structural
distortions induced by the intercalation of cations or anions into host crystal structures,
with the ultimate goal of correlating these changes to specific structural motifs to establish electrode materials design rules. Furthermore, this dissertation includes work toward
the development of operando characterization techniques for the advanced analysis of F-ion
batteries, an emerging alternative battery chemistry.
The second chapter presents an investigation into the reversible electrochemical
(de)fluorination of CsMnFeF6 using a liquid electrolyte at room temperature, the first report
of its kind. After three galvanostatic cycles, approximately one fluoride ion can be reversibly
(de)inserted into mechanochemical CsMnFeF6 for multiple cycles. Decreased cell impedance
after one galvanostatic cycle suggests the formation of fluoride vacancies in early cycles
generates mixed-valent Fe and enhances the material’s conductivity. Ex-situ synchrotron
25

Introduction
powder X-ray diffraction reveals subtle, but clearly reversible, expansion and contraction
of the CsMnFeF6 cubic lattice on fluoride insertion and removal, respectively, during the
first two cycles. New reflections appear in the cycle 3 ex-situ diffraction, corresponding to a
topotactic transformation of CsMnFeF6 from the original defect pyrochlore structure into a
related, orthorhombic polytype that continues to reversibly cycle fluoride ions. In parallel,
ex-situ X-ray absorption spectroscopy confirms that both Mn2+ and Fe3+ are redox active
during electrochemical cycling. This report is a critical step forward in the evolution of F-ion
batteries as a viable Li-ion alternative, demonstrating that no fundamental restrictions are
preventing the development of this next-generation energy storage technology.
In the third chapter, the structural evolution of TTB-type Nb8W9O47 upon lithium
(de)intercalation is investigated, with particular attention to how it relates to the material’s fast-cycling capabilities. Electrochemical cycling shows that Nb8W9O47 can achieve
greater than one Li+ per transition metal at rates of C/2 or slower, and maintains a capacity equivalent to 0.65 Li+ per transition metal at a rate of 20C. The space group is
unchanged during lithiation, however, sequential Rietveld refinements of in-house operando
X-ray diffraction data reveal anisotropic unit cell parameter changes during discharge and
suggest charge compensation occurs through off-centering of the transition metals within
their octahedra. Although higher capacities can be achieved by cycling lower than 1.2 V,
cells cycled down to 1.0 V exhibit a ∼2% greater expansion of the unit cell volume, and in
turn, poorer reversibility. This work clearly shows that rigid crystal structures, like that of
Nb8W9O47, undergo displacive rather than rotational deformations, a pivotal step forward in
our understanding of the structural origin of fast-cycling in bronze and bronze-type phases.
The fourth chapter revisits fluoride (de)intercalation with a study on the chemical fluorination of the Ruddlesden-Popper oxide, BaLa2Fe2O7, focused on the impact of fluorination
conditions on the structural evolution with increasing fluorine content. Two analogous series of oxyfluorides were synthesized by reacting BaLa2Fe2O7 with two fluorinating agents:
polyvinylidene fluoride (PVDF) and ammonium bifluoride (NH4HF2). Analysis of the pow26

1.6. Dissertation Overview
der X-ray diffraction data uncovers a symmetry lowering from the pristine oxide’s tetragonal
unit cell to an orthorhombic unit cell at low levels of fluorinating agent, then a return to
the tetragonal structure with increasing amounts of fluorine. While both fluorinating agents
induce the same structural transformation, more NH4HF2 is always required to impart the
same structural change observed for a given PVDF-fluorinated sample. Elemental analysis
and the refined impurity concentrations suggest that NH4HF2 also produces lower purity
samples, with the samples exhibiting decreasing purity as the amount of fluorinating agent
is increased. The presented work not only demonstrates successful chemical fluorination
of BaLa2Fe2O7 but also offers insight into the structural evolution that might be observed
electrochemically. These results also underscore the importance of a follow-up investigation
on the impact of synthetic conditions on the final products (i.e., extent of fluorine incorporation, impurity formation) to better understand the mechanism(s) of fluoride intercalation
in Ruddlesden-Popper oxides.
The fifth and final chapter describes the design of an F-ion compatible, continuous flow
DEMS electrochemical cell and cell-to-mass spectrometer interface. Developing operando
gas analysis methods for F-ion batteries is critical to investigating some of the hypothesized
issues plaguing the development of these systems, including intercalation host stability, liquid
electrolyte compatibility, and counter electrode reversibility. The electrochemical fluorination of ReO3 from a liquid, organic electrolyte (1.0 M TBAF in THF) is used as a “control”
reaction for the testing and adaptation of the DEMS setup because it offers a consistent
result of H2 evolution on oxidation due to electrolyte degradation within the first 10 h of
cycling. Using the tentatively finalized continuous flow DEMS setup, gas evolution during
the reversible, electrochemical (de)fluorination of CsMnFeF6 from 1.0 M TBAF in THF is
then investigated and found to also exhibit H2 evolution on oxidation. These results suggest
that electrolyte instability in the voltage windows of interest is limiting the advancement of
F-ion batteries, therefore, developing improved electrolytes and continuing to study these
systems with DEMS will be critical to their continued development.
27

Introduction
28

Chapter 2
Room Temperature Electrochemical
Fluoride (De)Insertion into CsMnFeF6
2.1 Introduction
While research on cation intercalation has thrived and had an extensive impact on modern
technologies, like high energy density Li-ion batteries, little work has been done to understand
the mechanism of anionic intercalation and its effects on a host’s structure.139,140 Given
the numerous, impactful discoveries from research on cation intercalation, the potential
for anionic intercalation is extremely promising. The possibility of anions serving as charge
carriers for electrochemical energy storage provides an alternative frontier to explore a variety
of new rechargeable battery materials. However, due to their larger ionic radii and negative
charges, the diffusion of anions through densely packed solids requires significantly different
conditions than that of mobile cations.110 Developing an understanding of the structural
changes induced by cation intercalation and diffusion in the solid state proved crucial to
augmenting the performance of Li-ion batteries. Therefore, a pivotal starting point in the
exploration of anionic intercalation is identifying the key structural features that enable and
promote anionic diffusion in the solid state.
Thus far, the electrochemistry of smaller anions, such as fluoride, has been mostly limited
Adapted from Andrews et al. ACS Enery Lett. 2022 7 2340–2348. [doi] 29

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
to conversion-based systems, in which the electrode materials react during oxidation/reduction to form entirely new products, typically with drastically different structures and
chemistries.93–96. Fichtner and co-workers employed these conversion reactions to produce
a rechargeable fluoride-ion (F-ion) battery, utilizing a solid electrolyte in tandem with a
CuF2 cathode cycled against a La metal film anode.141 When electrochemically cycled, this
results in the reduction of the cathode to produce Cu metal and conversion of the anode
to yield LaF3. Although these are reversible reactions, F-ion cycling of conversion-based
electrodes typically incurs large volume changes, resulting in poor cycling stability due to
limited reversibility over multiple cycles.
More recently, systems that leverage fluoride intercalation, in which the host material may
experience crystal structure changes but should broadly maintain its original topology, have
been reported. For example, Clemens and co-workers successfully showed intercalation of
fluoride into LaSrMnO4 and La2CoO4 using all-solid-state cell configurations.97,98. However,
the Ba-doped LaF3 solid electrolyte utilized in these all-solid-state cells suffers from poor
room-temperature fluoride conductivity, thereby requiring operating temperatures of 150 ◦C
or higher.99–101 While there are also recent reports of systems that incorporate the solid
electrolytes PbSnF4 and BaSnF4 to cycle fluoride reversibly at room temperature, they are
based on conversion electrodes and suffer from significant capacity fade in the first few
cycles.102–104
Thus, poor reversibility and high operating temperature requirements have limited
any practical use of the reported F-ion systems. This emphasizes the need to develop
intercalation-based electrodes that can reversibly (de)insert fluoride paired with electrolytes
that exhibit significant fluoride mobility at ambient temperatures. Christe reported one of
the first advances toward developing a room-temperature electrolyte with the stabilization of
fluoride using quaternary ammonium salts in tetrahydrofuran (THF).142 Davis et al. recently
reported that solutions of the salt N,N,N -trimethyl-N -neopentylammonium fluoride (Np1F)
dissolved in fluoroether solvents enabled stable cycling of conversion cathodes composed of
30

2.1. Introduction
a b
c
= Cs
= Mn/Fe
= F
Figure 2.1: Crystal structure of the defect fluoride pyrochlore, CsMnFeF6, viewed down the
[110] direction. Crystal structure depicted using VESTA.117
CuF2@LaF3 core–shell nanoparticles.91 These works provided some of the first substantial
evidence that small anions, like fluoride, can be leveraged electrochemically much like lithium
cations can be, presenting an avenue to overcome the barriers that have thus far limited the
study of anionic (de)intercalation.
Building off of these advances, our group recently reported on the electrochemical fluorination of an oxide host, ReO3, from a liquid fluoride electrolyte, tetra-n-butylammonium
fluoride (TBAF) dissolved in THF, at room temperature.108 Although clear evidence for
fluoride incorporation was obtained through a combination of electrochemical cycling, 19F
NMR techniques, and operando X-ray diffraction studies, reversible cycling of fluoride at
room temperature was not achievable due to the instability of fluorinated ReO3.
108
Following this work, we began to explore the defect pyrochlore CsMnFeF6 (Fig. 2.1).143
Oxide pyrochlores are known to display high ionic conductivities121,123 and the crystal structure contains an ordered anionic vacancy,116 which is promising for electrochemical fluoride
intercalation. Ideal pyrochlores have the general formula A2B2X6Y, where the A and B sites
are filled by metal cations, and the X and Y sites are filled by anions, like oxygen or fluorine.
The structure is best viewed as two interpenetrating, three-dimensional frameworks of B2X6
and A2Y.116 The B2X6 substructure is comprised of corner-sharing BX6 octahedra, forming
a three-dimensional network of six-membered rings and, as a result, large hexagonal cavities.
31

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
The A2Y substructure resembles anticristobalite Cu2O, with four-coordinate Y anions and
two-coordinate A cations, and penetrates the B2X6 framework so that the A cations occupy
the centers of the hexagonal cavities. As a result, there are three distinct anionic sites in
the pyrochlore crystal structure: X, Y, and an ordered anionic vacancy. Ionic diffusion in
pyrochlores is typically ascribed to vacancy hopping through a continuous path of secondnearest-neighbor X sites.121,122 Since all of these sites must be occupied to maintain charge
neutrality, diffusion through the network requires the presence of Frenkel defects, wherein
an X site is vacant and an ordered vacancy site is filled by the missing X anion.121,122 These
pairs can be realized by introducing disorder at the B site, which produces disorder between
the X site and the vacant site.121,122
Appreciable ionic conductivity in pyrochlores therefore relies heavily on disorder within
the B2X6 substructure, which is typically achieved via isovalent or aliovalent substitution.121,123 In addition, various defect pyrochlores readily form because the A2Y substructure
can be absent or partially occupied.118 Therefore, we hypothesized that the mixed B site,
defect pyrochlore CsMn2+Fe3+F6 (ABB’X6, Fig. 2.1) would make an ideal fluoride intercalation host as it possesses an ordered anionic vacancy and should display relatively high room
temperature fluoride conductivity due to disorder in the MnFeF6 substructure.
2.2 Experimental Methods
2.2.1 Material Synthesis
Caution! Hydrofluoric acid is toxic and corrosive and must be handled with extreme caution
and the appropriate protective gear. If contact with the liquid or vapor occurs, proper
treatment procedures should be followed immediately.144
CsMnFeF6 was synthesized via hydrothermal, ceramic, and mechanochemical routes. All
of the solid reagents used were dried under vacuum at 110 ◦C for 24 h, then stored and
weighed in an argon glove box. For the hydrothermal route, stoichiometric amounts of CsF
32

2.2. Experimental Methods
(Alfa Aesar, 99.9%), MnF2 (Sigma-Aldrich, 98%), and FeF3 (Sigma-Aldrich, 47.1% Fe by
Na2S2O3 titration) were combined with a twenty-fold molar excess of hydrofluoric acid (Alfa
Aesar, 40%) in an FEP Teflon (0.005” thick) pouch. The pouch was heat-sealed and placed
in a PTFE Teflon-lined Parr pressure vessel with a deionized H2O backfill (33% pressure
vessel volume). The pressure vessel was heated to 150 ◦C for 24 h, then cooled to room
temperature for 20 h (0.1 ◦C per min). The pouch was opened in air, and the contents
were vacuum-filtered and rinsed copiously with deionized H2O, then acetone, to isolate a
pale green powder. For the ceramic route, stoichiometric amounts of CsF, MnF2, and FeF3
were weighed and mixed intimately using an agate mortar and pestle, and then pressed into
10 mm pellets, all under an Ar atmosphere. The pellets were placed in an Inconel metal
tube, which was crimped shut under the Ar atmosphere, and then welded shut using a TIG
welder. The sealed Inconel tube was heated to 500 ◦C for 12 h, then air cooled in a muffle
furnace. These conditions yielded soft, pale brown pellets which produced a taupe powder
once ground. For the mechanochemical route, stoichiometric amounts of CsF, MnF2, and
FeF3 were weighed and ground under an argon atmosphere, using a SPEX Sample Prep
8000D Mixer/Mill high-energy ball mill. The mixture of starting materials was ground for
3 h, in 30 min increments followed by 10 min rest periods, with a 28:1 weight ratio of steel
balls to powder. These conditions produced a taupe powder, very similar in color to the
ceramic product.
2.2.2 Powder X-ray Diffraction
Laboratory powder X-ray diffraction (XRD) patterns were collected on a Bruker D8 Advance
diffractometer with a Cu Kα source (λ1=1.5406 ˚A, λ2=1.5444 ˚A), equipped with a LynxEye
XE-T detector. High-resolution synchrotron powder XRD data was collected at beamline 11-
BM at the Advanced Photon Source (APS) at Argonne National Laboratory using an average
wavelength of 0.458175 ˚A. Discrete detectors covering an angular range from −6 to 28◦ 2θ
were scanned over a 34◦ 2θ range, with data points collected every 0.001◦ 2θ and scan
33

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
speed of 0.1◦ per s. High-resolution synchrotron powder XRD data was also collected at
beamline 2-1 at the Stanford Synchrotron Radiation Lightsource (SSRL) using an average
wavelength of 0.7282313–0.7282346 ˚A. The resulting XRD patterns were refined against
published structures using the Rietveld method as implemented in the TOPAS-Academic
suite.145
2.2.3 Zero-field 57Fe M¨ossbauer spectroscopy
Zero-field 57Fe M¨ossbauer spectroscopy was performed on a SEE Co Model W304 resonant
gamma-ray spectrometer equipped with a Janis Research Model SVT-400 cryostat system.
The source material produced by Ritverc was 57Co in Rh with an activity of 100 mCi ± 10%.
The source line width was less than 0.12 mm/s for a 25-µm α-iron foil standard sample.
Isomer shifts are reported relative to α-iron foil at room temperature. Each spectrum was
collected at room temperature on approximately 30 mg of a powdered sample packed into
a polyethylene sample holder. The acquisition time for each sample was approximately one
day. Absorbance data were fitted using a custom Igor Pro (Wavemetrics) macro package as
previously reported.146,147
2.2.4 Scanning Electron Microscopy
Scanning electron microscopy (SEM) images and Energy Dispersive Spectroscopy (EDS)
spectra were obtained using a model JEOL JSM-6700F field emission electron microscope.
Imaging was conducted with a 5 kV accelerating voltage and lower secondary electron detector configuration. EDS spectra were taken with a 15 kV accelerating voltage and 15 mm
working distance. Samples were spread onto double-sided copper tape and sputter-coated
with gold for 60 s.
34

2.2. Experimental Methods
2.2.5 X-ray Photoelectron Spectroscopy
X-ray photoelectron spectroscopy (XPS) was performed on a Kratos Axis Ultra DLD spectrometer with a monochromatic Al Kα radiation source. Samples were placed on carbon
tape for the measurements. To control the charging of the samples, a charge neutralizer
filament was used. For these high-definition scans, a pass energy of 20 eV was used with
a 0.1 step size and 300 ms dwell time. All spectra were calibrated using the advantageous
carbon 1s peak at 284.8 eV. Analysis was performed on CASA XPS software using the CASA
XPS library.
2.2.6 Electrochemical Characterization
All F-ion cell components and electrodes were dried under vacuum at 110 ◦C for at least 10 h
before cell assembly. All cell assembly was performed in an argon glovebox. Stainless steel
Swagelok cells were utilized as the electrochemical test cells and borosilicate glass fiber pads
(Whatman GF/D) were utilized as separators. A 1.0 M solution of TBAF in THF (Acros
Organics) was used as the liquid electrolyte. Loose powder working electrodes were prepared
by ball milling 20 wt % of Ketjen Black EC60JD conductive carbon (AkzoNobel) and 80 wt %
of active material, CsMnFeF6, for 10 minutes in a SPEX Sample Prep 8000D Mixer/Mill
high-energy ball mill. The electrodes had a typical mass loading of 5 to 10 mg/cm2
. To
produce the Bi/BiF3 composite counter electrodes, Bi metal powder (Sigma-Aldrich, 99%)
and anhydrous BiF3 (Strem, 99.99%) were mixed with Super P conductive carbon and
poly(vinylidene fluoride) (MTI) in a 40:40:10:10 wt % ratio in a minimal amount of Nmethyl-2-pyrrolidone (MTI) solvent to form a slurry. The slurry was then cast on aluminum
foil to 50 µm thick and dried at 110 ◦C under vacuum overnight.
35

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
2.2.7 X-ray Absorption Spectroscopy
Mn and Fe K-edge X-ray absorption spectra of pristine and cycled powders were collected
at beamline 4-1 at the Stanford Synchrotron Radiation Lightsource (SSRL) equipped with
a double-crystal Si(220) monochromator, ϕ = 90◦
. The samples were isolated inside of
a He-filled chamber throughout data collection and were measured in fluorescence mode
using a Passivated Implanted Planar Silicon (PIPS) detector. The vertical beam size was
0.5 mm, and the horizontal beam size varied between 4 and 10 mm depending on the signal
obtained for each sample. Three sweeps per sample were individually reference-corrected
using spectra of Mn or Fe foils, collected simultaneously. In all cases, the edge is defined
as the zero-crossing of the second derivative, where the Mn K-edge is 6539.0 eV and the Fe
K-edge is 7112.0 eV. The data were reference-corrected, merged, and normalized using the
Athena software of the Demeter package.148
2.3 Results and Discussion
CsMnFeF6 crystallizes in the cubic space group F d¯3m (#227), as shown in Figure 2.1.143
CsMnFeF6 was synthesized via three different methods: “hydrothermal”, “ceramic”, and
“mechanochemical”, described in detail in the Experimental Methods section. Rietveld
refinements of the structure against synchrotron powder X-ray diffraction (XRD) data were
performed to evaluate the phase purity obtained from each synthetic method, with the
resulting structural parameters given in Tables 2.2–2.4. While the mechanochemical reaction
was found to yield a high degree of purity (Fig. 2.2), the hydrothermal and ceramic materials
contained mixtures of CsMnFeF6, CsMnF3, MnF2, and CsMnFe2F9 (Fig. 2.8a and 2.9a). It
is important to note that none of these impurities were seen using in-house X-ray diffraction,
and it was only the high intensity of the synchrotron source that revealed the secondary
phases.
Scanning electron microscopy (SEM) images show that of the three methods, the hy36

2.3. Results and Discussion
10 µm
6 9 12 15 18 21 24 27
2θ (deg.) [λ=0.4582 Å]
Intensity (arb. units)
observed
calculated
difference
Figure 2.2: Rietveld refinement of synchrotron XRD data for mechanochemical CsMnFeF6.
Inset: SEM image of mechanochemical CsMnFeF6.
drothermal method yields the largest, most uniformly shaped particles, the ceramic technique yields the smallest, and the mechanochemical approach yields the least uniformly
shaped and sized particles. The hydrothermal CsMnFeF6 particles were roughly ∼10 µm
in diameter and exhibited an octahedral habit with broad (111) facets, similar to previous
reports (Fig. 2.8a inset).149 The ceramic CsMnFeF6 method produces the smallest particles,
between 1 and 5 µm in diameter, with no clearly defined shape (Fig. 2.9a inset). In contrast,
the particles from the mechanochemical method are irregularly shaped and exhibit the widest
range of sizes, anywhere from <1 µm to >10 µm (Fig. 2.2 inset). Despite the variation in
particle size and morphology observed in the SEM, X-ray photoelectron spectroscopy (XPS)
results indicate all three synthetic methods produce particles with surfaces terminated by
mostly Cs and F (Fig. 2.11–2.13).
Materials prepared with each method were electrochemically cycled using Bi/BiF3 composite counter electrodes, which provide a more stable counter electrode over long-term
cycling compared to MFx electrodes.94–97,150 While these composite-style Bi/BiF3 counter
electrodes show enhanced cyclability, it is crucial to acknowledge that they still impose the
limitations of conversion-based electrochemistry (i.e., rate performance, reversibility) on the
37

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
whole cells studied in the present work. Glass fiber separators (Whatman GF/D) were soaked
in a commercial solution of 1.0 M TBAF in THF, which served as the electrolyte. TBAF was
chosen based on prior work that showed a high fluoride conductivity at room temperature
over a relatively wide electrochemical window.108 Loose powder working electrodes, prepared
by ball milling conductive carbon and CsMnFeF6, were utilized.
As seen in Figure 2.3a, the first oxidation of the mechanochemical sample exhibits a
smoothly sloped voltage curve that gradually flattens towards the upper voltage limit of
1.4 V vs. Bi/BiF3 [Eo
c
(Bi3+/Bi) = 3.4 V vs. Li+/Li].91 This first oxidation results in a
specific capacity of 33 mAh/g, which exceeds 40% of the theoretical capacity (74.9 mAh/g),
as calculated for the (de)insertion of one equivalent of fluoride per formula unit of CsMnFeF6.
The voltage profile of the first reduction begins to exhibit four distinct regions of subtle
differences in the voltage slope. In the first region, there is a steep decrease from 1.4 to 1.0 V.
0 100 200 300 400 500
Z’ (Ohm)
0
100
200
300
400
500
-Z’’ (Ohm)
pre-cycling
post-ox1
post-red1
0 15 30 45 60 75 90
Capacity (mAh g-1)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
V vs. Bi/Bi3+
cycle 1
cycle 2
cycle 3
(a)
(b)
Figure 2.3: (a) Cycles 1–3 from galvanostatic cycling of a F-ion cell, with a working electrode
of mechanochemically synthesized CsMnFeF6 and a Bi/BiF3 composite counter electrode,
cycled at room temperature at a rate of C/20 between 0.0 and 1.4 V vs. Bi/BiF3. (b) PEIS
performed before cycling, after the first oxidative cycle (ox 1), and after the first reductive
cycle (red 1).
38

2.3. Results and Discussion
From 1.0 to 0.75 V, the curve flattens slightly to form a short plateau before steepening from
0.75 to 0.4 V. The final region shows the most significant flattening of the voltage profile
with a plateau from 0.4 to 0.0 V, for an overall capacity of 41 mAh/g. The second oxidation
yields a profile that more closely resembles the reversal of the first reduction curve, rather
than the limited capacity of the first oxidation.
By the third cycle, the voltage profile is more well-defined with the capacity exceeding
80% of the theoretical value on both oxidation and reduction and only ∼300 mV of polarization. The voltage profiles for all subsequent cycles exhibit a similar profile (Fig. 2.4a), with
sloped plateaus during oxidation (reduction) at approximately 0.6 V (0.4 V) and 1.15 V
(0.85 V), confirmed by peaks in the differential capacity plot (Fig. 2.4b). The reversible
capacity reaches 70 mAh/g by the fourth cycle, which is very near the maximum theoretical
capacity for one fluoride ion, and is maintained through the ninth cycle with an average
Coulombic efficiency of 98% (cycles 4–9) after which a gradual fade is seen.
The ceramic material cycles similarly to the mechanochemical material, although it shows
less reversibility (Fig. 2.19). The entire first cycle exhibits a low capacity relative to the
theoretical limit, and the faradaic features during the first reduction are very subtle. In
contrast with the mechanochemical material’s voltage profile, the ceramic material does
not exhibit well-resolved faradaic features until the fourth reduction/fifth oxidation. This
is unsurprising, as the ceramic sample contained several impurities that are unlikely to
be redox active, so the poor performance may be attributable to a delayed onset of clear
electrochemistry from fluoride (de)insertion into CsMnFeF6.
From the fifth cycle onward, the galvanostatic cycling exhibits the clearly defined fourregion voltage profile with sloped plateaus during oxidation (reduction) at approximately
0.7 V (0.3 V) and 1.3 V (0.9 V) (Fig. 2.19). The capacity also gradually increases, reaching
a maximum of 58 mAh/g on the fifth reduction and 61 mAh/g on the sixth oxidation,
corresponding to reversible fluoride (de)insertion but only achieving ∼80% of the theoretical
capacity. This average capacity is maintained through the eighth cycle, after which it begins
39

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
(a)
(b)
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
V vs. Bi/Bi3+
-150
-75
0
75
150
dQ/dV (mAhg-1
V-1)
cycle 6
cycle 7
0 15 30 45 60 75 90
Capacity (mAh g-1)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
V vs. Bi/Bi3+
cycle 6
cycle 7 Figure 2.4: (a) Cycles 4–9 from galvanostatic cycling of a F-ion cell, with a working electrode
of mechanochemically synthesized CsMnFeF6 and a Bi/BiF3 composite counter electrode,
cycled at room temperature at a rate of C/20 between 0.0 and 1.4 V vs. Bi/BiF3. Cycles six
and seven are shown in orange and blue, respectively. (b) Differential capacity plot derived
from the sixth and seventh cycles.
to gradually fade. The ceramic material also exhibits a slightly larger voltage polarization
(∼400 mV) compared to that of the mechanochemical material (∼300 mV). Therefore, the
ceramic samples still cycle relatively well, but the maximum capacity is lower and fades more
quickly than that of the mechanochemical samples.
In contrast, the hydrothermal samples do not cycle nearly as well as the others (Fig. 2.20).
Despite exhibiting a similar voltage profile, they show the largest polarization and the least
defined faradaic features. Similar to the mechanochemical and ceramic materials, the first
cycle exhibits a very low capacity compared to the theoretical value. Akin to the ceramic
material, the hydrothermal material does not exhibit well-resolved faradaic features until the
fourth/fifth cycle. This is unsurprising because the hydrothermal material also contained
redox inactive impurity phases, leading to a slower onset of clear electrochemistry from
fluoride (de)insertion into CsMnFeF6. After the fifth cycle, a sloping plateau is observed
40

2.3. Results and Discussion
during oxidation (reduction) at approximately 0.8 V (0.3 V). However, the second plateau
seen at higher voltages in the cycling of the other two materials is not observable in the
hydrothermal cycling (Fig. 2.20). To explore whether the polarization of the cells was shifting
this high voltage feature out of the cycling window, the upper limit was expanded to 1.45 V
vs. Bi/BiF3 (Fig. 2.20). While better reversibility was achieved, the higher voltage feature
still could not be resolved, and opening the window further led to irreversible reactions that
may correspond to fluorination of the carbon additive or attack of the electrolyte.104,108,151,152
The local bonding environments of the Fe atoms within CsMnFeF6 were analyzed directly
using zero-field 57Fe M¨ossbauer spectroscopy, which is sensitive to the types of coordinating
ligands and symmetry of Fe atoms based on the absorption of gamma rays by Fe nuclei in solid
materials. M¨ossbauer spectra were acquired at room temperature for the three CsMnFeF6
materials, a representative example of which is shown in Figure 2.5 for the mechanochemical
material; the spectra for the other materials are shown in Figures 2.8b and 2.9b. Each
of the M¨ossbauer spectra exhibit a dominant quadrupole doublet with an isomer shift, δ,
of 0.3 mm/s, which is comparable to previously reported values for high-spin, octahedral
Fe3+ in similar materials.153,154 Identical quadrupole splittings, |∆EQ|, of 0.3 mm/s are also
observed for this dominant signal in all three spectra. The quadrupole splitting value, which
tends to increase with the distortion of the electronic environment around high-spin Fe3+
ions, is lower than reported previously for similar pyrochlore materials,120,153 reflecting more
symmetric Fe3+ environments in the materials investigated here.
The spectra for the ceramic and mechanochemical CsMnFeF6 samples exhibited a second,
less intense quadrupole doublet with an isomer shift of 0.7 mm/s and a quadrupole splitting
of 1.3 mm/s. This larger isomer shift is indicative of an Fe2+ species,154 which suggests
that the ceramic and mechanochemical synthetic methods produce a small amount (∼9%)
of divalent Fe-sites, whereas the hydrothermal method does not. Overall, the consistent
isomer shift and quadrupole splitting values for the dominant signal among the spectra
for all three samples indicate that similar local Fe environments result from the different
41

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
-4 -3 -2 -1 0 1 2 3 4
Velocity (mm/s)
-5
-4
-3
-2
-1
0
Absorbance (%)
observed
fit 1
fit 2
δ = 0.7 mm/s
|∆EQ| = 1.3 mm/s
91%
9%
δ = 0.3 mm/s
|∆EQ| = 0.3 mm/s
Figure 2.5: M¨ossbauer spectrum of mechanochemical CsMnFeF6, collected at 298 K over
approximately one day.
synthetic methods. The presence of a single doublet in the spectrum of the hydrothermally
prepared CsMnFeF6 suggests that fluoride is the predominant and probably the only type of
coordinating ligand near the Fe-sites in these materials.155 If any F− ligands were replaced
with OH− in the hydrothermal synthesis, then a second doublet with a larger quadrupole
splitting would be present in the room-temperature M¨ossbauer spectrum, corresponding
to differently coordinated Fe-sites.155 Therefore, the M¨ossbauer spectroscopy results indicate
that there is likely no OH− impurity in the hydrothermal product and that all three materials
contain a majority of octahedral Fe3+.
Potentiostatic electrochemical impedance spectroscopy (PEIS) and galvanostatic cycling
were performed on identical cells to investigate if the observed voltage profile changes in the
initial cycles were reflected as changes in the total impedance of the cell (Fig. 2.3b, 2.19a
and 2.20a insets). PEIS was conducted with a voltage amplitude of 10 mV, measured between frequencies of 1 MHz and 100 mHz, on each cell at three different points: (1) before
any galvanostatic cycling, (2) after one oxidation, and (3) after one full cycle. Of the three
samples, the hydrothermal material exhibits the highest initial impedance, whereas the ceramic and mechanochemical materials are very similar to one another. This is in agreement
with trends in the M¨ossbauer spectroscopy, where samples prepared using the ceramic and
42

2.3. Results and Discussion
mechanochemical methods exhibit significant amounts of Fe2+, whereas the hydrothermal
samples exclusively contain Fe3+. This mixed valency likely boosts the electronic conductivity of the mechanochemical and ceramic materials compared to the hydrothermal material.
This is supported by the fact that all three materials show a minor increase in the impedance
after one oxidation, which would remove the Fe2+ from the sample, thereby decreasing the
electrical conductivity. Indeed, following the first reduction, more Fe3+ is converted to Fe2+,
and this mixed valency likely enhances the total conductivity regardless of the synthetic
method employed.
Ex-situ synchrotron XRD was performed to identify and study the structural impacts of
electrochemical fluoride (de)insertion into CsMnFeF6. Electrodes from the mechanochemical
1.92 1.95 1.98 2.01
Intensity (arb. units)
2.6 2.8 3.0 3.2 3.4
Q (Å-1)
ox 3
red 3
ox 4
red 4
(b)
(a)
1.92 1.95 1.98 2.01
Q (Å-1)
Intensity (arb. units)
C-coated
ox 1
red 1
ox 2
red 2
2.6 2.8 3.0 3.2 3.4
Figure 2.6: Zoomed-in regions of ex-situ synchrotron XRD results showing (a) subtle shifts
of the CsMnFeF6 reflections corresponding to expansion on oxidation and contraction on
reduction (collected at 11-BM, APS), and (b) the growth of new reflections corresponding
to a transformation of pyrochlore-type CsMnFeF6 into a related polytype and the reversible
modulation of these reflections with continued fluoride (de)insertion (collected at 2-1, SSRL).
43

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
method were stopped at four points during cycling: after the first oxidation (ox 1), after
one full cycle (red 1), after one cycle and one oxidation (ox 2), and after two full cycles
(red 2). Synchrotron XRD was collected at beamline 11-BM at the APS. The results of
Rietveld refinements (given in Tables 2.5–2.7) show a subtle lattice expansion on oxidation
(insertion) and contraction on reduction (removal) in the first two cycles (Fig. 2.6a, 2.14).
Specifically, the cubic lattice parameter remains nearly constant on ox 1, contracts 0.12% on
red 1, expands 0.08% on ox 2, and contracts 0.09% on red 2. This corresponds to a unit cell
volume contraction of 0.34% on red 1, an expansion of 0.25% on ox 2, and a contraction of
0.28% on red 2.
New, low-intensity reflections appear in the red 2 pattern, with the most intense new
feature around ∼1.93 ˚A−1
(Fig. 2.6, 2.16). Therefore, additional ex-situ synchrotron XRD
was performed at beamline 2-1 at SSRL to see if the lattice expansion/contraction of the
pyrochlore phase continues and if the additional peaks evolve further in the later cycles.
Mechanochemical CsMnFeF6 electrodes were stopped at four more points during cycling:
after two full cycles and one oxidation (ox 3), after three full cycles (red 3), after three cycles
and one oxidation (ox 4), and after four full cycles (red 4); the resulting XRD data is shown
in Figures 2.6b and 2.14b. Simulated patterns from CIFs in the Inorganic Crystal Structure
Database of (1) the expected binary or ternary metal fluoride phases from a conversion
reaction (i.e., Mn–F, Fe–F, Cs–Fe–F, Cs–Mn–F) or (2) Bi-containing species from dissolution
and crossover at the counter electrode (i.e., Bi, BiF3) did not agree well with the new
reflections (Fig. 2.17).
These results suggest that a conversion reaction does not dominate the observed electrochemistry and the evolution of the new peaks is likely due to a topotactic transformation
of CsMnFeF6 from the original pyrochlore structure to a related polytype. This conclusion
is further supported by the presence of the new peaks in both the oxidized and reduced
samples, as well as the slight changes in their shape and intensity between each process—if
a conversion reaction were occurring, the new peaks would be expected to entirely disappear
44

2.3. Results and Discussion
and/or shift between each sample. Rietveld refinements of the cubic pyrochlore structure
against the cycle 3 and 4 ex-situ data were also performed and indicated that the cubic lattice parameter expands 0.035% on ox 3, contracts 0.034% on red 3, and expands 0.015% on
ox 4 (Table 2.8, Fig. 2.15). Therefore, the expansion and contraction of the cubic pyrochlore
lattice on oxidation (insertion) and reduction (removal), respectively, diminish as the new
peaks intensify.
The modulation of the peaks belonging to the orthorhombic phase, namely the changing shoulder shapes and differing intensities between oxidation (broadening, decreasing)
and reduction (narrowing, increasing) suggests that this new phase continues to reversibly
(de)insert fluoride ions. Indexing of the new peaks yields a unit cell in the orthorhombic
space group I222 (#23). A multi-phase Pawley fit of the ox 3 pattern performed using
the reported F d¯3m cell for the original pyrochlore phase and the indexed I222 cell for the
new phase sufficiently described the experimental data, shown in Figure 2.18. Notably, the
lattice parameters of the orthorhombic phase from the Pawley fit are extremely similar to
those reported in the Inorganic Crystal Structure Database for fluoride “weberite” phases.
The pyrochlore and weberite crystal structures are polytypes, therefore this suggests that
the defect pyrochlore structure could be transforming into a defect weberite-type structure
(the orthorhombic phase). A structure solution of the orthorhombic phase is being actively
pursued, and a more detailed description of this electrochemically induced topotactic transformation from the original pyrochlore symmetry into the orthorhombic polytype will be
reported in a forthcoming structural study.
Ex-situ Mn K-edge and Fe K-edge X-ray absorption spectroscopy (XAS) experiments
were performed at beamline 4-1 at SSRL to identify the redox couple(s) responsible for
the observed electrochemical activity. In addition to measuring pristine and carbon-coated
mechanochemical CsMnFeF6, electrodes were harvested at two points during cycling: after
two cycles and one oxidation (ox 3) and after three full cycles (red 3). The reported edge
energies were defined using the zero-crossings of the second derivatives (Fig. 2.21, 2.22).
45

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
The Mn K-edge of pristine CsMnFeF6 (6547.1 eV) is at a similar energy as that of the
MnF2 standard (6547.8 eV), suggesting the composition is predominantly in the Mn2+ state
before cycling, as expected (Fig. 2.7a, 2.21). In the sample collected after the third oxidation,
the Mn K-edge shifts to a higher energy (6551.8 eV), towards the MnF3 standard’s Mn Kedge (6554.1 eV), indicating that the oxidation state of the Mn this sample is closer to 3+
than in the pristine material. After the subsequent reduction, the Mn K-edge shifts back
to a lower energy (6547.8 eV), re-aligning with the Mn K-edge of the pristine material and
confirming that the oxidation is reversible. However, the shape of the X-ray absorption near
edge structure (XANES) is clearly very different for the pristine material compared to the
two ex-situ samples, indicative of changes in the Mn coordination environment with the
(de)insertion of fluoride. A more comprehensive characterization of these structural changes
will be reported in a forthcoming study of the local structure.
The Fe K-edge of pristine CsMnFeF6 (7129.4 eV) is at a similar energy as that of the
FeF3 standard (7129.3 eV), suggesting the composition is predominantly in the Fe3+ state
before cycling, as expected (Fig. 2.7b, 2.22). In the post-oxidation sample (ox 3), the Fe Kedge is observed at 7129.5 eV. This energy is similar to that of the pristine material and the
FeF3 standard, confirming that the Fe in CsMnFeF6 returns to the 3+ state after oxidation.
In the post-reduction sample (red 3), the Fe K-edge shifts to a lower energy (7124.7 eV),
towards the Fe K-edge of the FeF2 standard (7122.8 eV), indicating that the the oxidation
state of the Fe this sample is closer to 2+ than in the pristine material. The shape of the
XANES is relatively consistent between the pristine and cycled materials, suggesting the Fe
coordination environment does not change significantly with the (de)insertion of fluoride.
Therefore, the shifts in the Mn and Fe K-edges confirm that upon fluoride (de)insertion into
CsMnFeF6, both Mn2+ and Fe3+ are redox active.
The incomplete shift of the Mn and Fe K-edges observed for the ex-situ samples is consistent with the reversible capacity achieved. While the maximum capacity observed in the
galvanostatic cycling was ∼75 mAh/g, the ex-situ XAS samples were collected during/after
46

2.3. Results and Discussion
(a)
(b)
7112 7120 7128 7136 7144 7152
Energy (eV)
Intensity (arb. units)
CsMnFeF6
oxidation 3
reduction 3
Fe K-edge
6536 6544 6552 6560 6568 6576
Energy (eV)
Intensity (arb. units)
CsMnFeF6
oxidation 3
reduction 3
Mn K-edge
Figure 2.7: Ex-situ X-ray absorption spectroscopy results showing (a) the shifting Mn Kedge and (b) shifting Fe K-edge upon fluoride (de)insertion.
cycle 3, which typically only reaches 80% of the theoretical capacity. The observed redox
activity of both Mn2+ and Fe3+ during cycling suggests that, theoretically, the capacity
could be doubled to 149.9 mAh/g (two fluoride ions per formula unit). In other words, if
Fe3+ could be fully reduced to Fe2+ on reduction and if Mn2+ were fully oxidized to Mn3+
on oxidation, multi-electron redox could be realized. Toward this exciting prospect, we are
actively continuing to optimize our room-temperature F-ion cell geometry.
47

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
2.4 Conclusions
In summary, we present the first report of reversible fluoride (de)insertion into CsMnFeF6
using a liquid electrolyte at room temperature. Using a combination of galvanostatic cycling and electrochemical impedance spectroscopy, we propose that an increase in fluoride
vacancies and mixed valency at the Fe-sites occurs in the early cycles. Ex-situ XRD shows
that CsMnFeF6 undergoes a small but clearly reversible expansion on oxidation (fluoride
insertion) and contraction following reduction (fluoride removal) during the first two cycles. From the third cycle onward, the pyrochlore-related reflections decrease in intensity
while new reflections corresponding to an orthorhombic phase begin to intensify, signifying
a topotactic transformation of CsMnFeF6 from the original defect pyrochlore structure into
a related polytype that continues to reversibly cycle fluoride ions. In parallel, ex-situ XAS
indicates that both the Fe3+/2+ and Mn3+/2+ redox couples are active during the cycling of
CsMnFeF6. This report clearly demonstrates there are no fundamental restrictions preventing the development of F-ion batteries as a viable alternative energy storage technology, and
therefore represents a pivotal step forward in their evolution.
48

2.5. Supplemental Information
2.5 Supplemental Information
2.5.1 Synchrotron X-ray Diffraction, Scanning Electron Microscopy, and Zero-Field 57Fe M¨ossbauer Spectroscopy
10 µm
(a)
(b)
-4 -3 -2 -1 0 1 2 3 4
Velocity (mm/s)
-6
-5
-4
-3
-2
-1
0
Absorbance (%)
observed
fit
δ = 0.3 mm/s
|∆EQ| = 0.3 mm/s
100%
6 9 12 15 18 21 24 27
2θ (deg.) [λ=0.4582 Å]
Intensity (arb. units)
observed
calculated
difference
Figure 2.8: (a) Rietveld refinement of hydrothermal CsMnFeF6 synchrotron XRD data; the
hydrothermal product was refined to a mix of two CsMnFeF6 phases with slightly differing
unit cell parameters (orange and purple reflections), CsMnFe2F9 (pink), and MnF2 (black).
Inset: SEM image of hydrothermal CsMnFeF6. (b) M¨ossbauer spectrum of hydrothermal
CsMnFeF6, collected at 298 K over approximately one day.
49

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
10 µm
10 µm
(a)
(b)
-4 -3 -2 -1 0 1 2 3 4
Velocity (mm/s)
-6
-5
-4
-3
-2
-1
0
Absorbance (%)
observed
fit 1
fit 2
δ = 0.7 mm/s
|∆EQ| = 1.3 mm/s
91%
9%
δ = 0.3 mm/s
|∆EQ| = 0.3 mm/s
6 9 12 15 18 21 24 27
2θ (deg.) [λ=0.4582 Å]
Intensity (arb. units)
observed
calculated
difference
Figure 2.9: (a) Rietveld refinement of ceramic CsMnFeF6 synchrotron XRD data; the ceramic product was refined to a mix of two CsMnFeF6 phases with slightly differing unit cell
parameters (orange and purple reflections), CsMnF3 (pink), and MnF2 (black). Inset: SEM
image of ceramic CsMnFeF6. (b) M¨ossbauer spectrum of ceramic CsMnFeF6, collected at
298 K over approximately one day.
50

2.5. Supplemental Information
2.5.2 Energy Dispersive Spectroscopy
Cs
Cs
Cs
0 2 4 6 8 10
Energy (keV)
Intensity (arb. units) Mn
Mn Mn
Fe
F
Cs
Cu
Cu
Fe
Fe
Cu
Au
0 2 4 6 8 10
Energy (keV)
Intensity (arb. units) Mn
Mn Mn
Fe
F
Cs
Cs
Cs
Cs
Cu
Fe
Fe Cu
Au
0 2 4 6 8 10
Energy (keV)
Intensity (arb. units) Mn Mn Mn
Fe
F
Cs
Cs
Cs
Cs
Cu
Fe
Fe Cu
Au
(a) (b) (c)
Figure 2.10: Representative energy dispersive spectra for (a) hydrothermally, (b) ceramically,
and (c) mechanochemically synthesized CsMnFeF6 powders.
Table 2.1: EDS elemental quantification results for the three different CsMnFeF6 powders.
Elemental ratios are normalized to the Mn value and the quantifications are averaged across
multiple spots for each sample.
Hydrothermal Ceramic Mechanochemical
Cs 0.97 ± 0.05 1.00 ± 0.05 1.22 ± 0.08
Mn 1.00 ± 0.06 1.00 ± 0.05 1.00 ± 0.06
Fe 1.07 ± 0.06 1.18 ± 0.05 1.03 ± 0.06
51

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
2.5.3 X-ray Photoelectron Spectroscopy
The Fe 2p spectra overlap with more intense Cs signals, so they were not included.
(a) (b) (c)
720 723 726 729 732 735 738 741
Binding Energy (eV)
1000
2000
3000
4000
Intensity (counts/s)
data
fit
Cs 3d5/2
Cs 3d3/2
678 680 682 684 686 688 690 692
Binding Energy (eV)
500
1000
1500
2000
2500
3000
3500
Intensity (counts/s)
data
F 1s fit
635 640 645 650 655 660
Binding Energy (eV)
50
100
150
200
250
300
350
400
Intensity (counts/s)
data
fit
Mn 2p3/2
Mn 2p1/2
Figure 2.11: Representative (a) Cs 3d, (b) Mn 2p, and (c) F 1s X-ray photoelectron spectra
for hydrothermally synthesized CsMnFeF6 powder.
(a) (b) (c)
720 723 726 729 732 735 738 741
Binding Energy (eV)
1000
2000
3000
4000
Intensity (counts/s)
data
fit
Cs 3d5/2
Cs 3d3/2
678 680 682 684 686 688 690 692
Binding Energy (eV)
500
1000
1500
2000
2500
Intensity (counts/s)
data
fit
F 1s
635 640 645 650 655 660
Binding Energy (eV)
100
150
200
250
300
350
400
Intensity (counts/s)
data
Mn 2p fit 3/2
Mn 2p1/2
Figure 2.12: Representative (a) Cs 3d, (b) Mn 2p, and (c) F 1s X-ray photoelectron spectra
for ceramically synthesized CsMnFeF6 powder.
720 723 726 729 732 735 738 741
Binding Energy (eV)
1000
2000
3000
4000
5000
Intensity (counts/s)
data
fit
Cs 3d5/2
Cs 3d3/2
635 640 645 650 655 660
Binding Energy (eV)
100
150
200
250
300
350
400
Intensity (counts/s)
data
fit Mn 2p3/2
Mn 2p1/2
678 680 682 684 686 688 690 692
Binding Energy (eV)
1000
2000
3000
4000
Intensity (counts/s)
data
fit
F 1s
(a) (b) (c)
Figure 2.13: Representative (a) Cs 3d, (b) Mn 2p, and (c) F 1s X-ray photoelectron spectra
for mechanochemically synthesized CsMnFeF6 powder.
52

2.5. Supplemental Information
2.5.4 Synchrotron X-ray Diffraction Rietveld Refinement Results
and Ex Situ X-ray Diffraction Analysis
High resolution powder X-ray diffraction patterns of the as-synthesized hydrothermal, ceramic, mechanochemical, and ex situ CsMnFeF6 were collected at beamline 11-BM at the
APS and beamline 2-1 at SSRL. The published structures were refined against the resulting
patterns using the Rietveld method, as implemented in TOPAS-Academic v6,145 to evaluate
phase purity of the materials. Relevant results are listed in Tables 2.2–2.8.
The hydrothermal product was found to contain CsMnFeF6 in the expected F d¯3m space
group (No. 227) as the majority phase. Upon close inspection of the hydrothermal XRD
data, we observed that every reflection attributed to CsMnFeF6 exhibits a small shoulderlike feature on the right-hand side. As this peak asymmetry appears on every CsMnFeF6-
associated reflection, the possibility of stacking faults was ruled out. Instead, these shoulders
were attributed to compositional inhomogeneity in the samples prepared hydrothermally, and
the data was refined to include a secondary CsMnFeF6 phase, also in the F d¯3m space group
but with a slightly smaller cubic unit cell parameter. The hydrothermal XRD data was
also refined to include two minor impurity phases: 3.2 wt % hexagonal tungsten bronzetype CsMnFe2F9 (P63/mcm, No. 193; Fig. 2.8a) and 0.9 wt % MnF2 (P42/mnm, No.
136; Fig. 2.8a). The ceramic product contained a majority phase of CsMnFeF6 in the F d¯3m
space group, and exhibited the same asymmetry on the CsMnFeF6 reflections. Therefore, the
shoulder-like features were again attributed to compositional inhomogeneity and were refined
to a secondary CsMnFeF6 phase with a smaller cubic unit cell parameter. Additionally, the
ceramic XRD data was refined to include two minor impurity phases: 9.2 wt % hexagonal
perovskite CsMnF3 (P63/mmc, No. 194; Fig. 2.9a) and 2.9 wt % MnF2 (P42/mnm, No.
136; Fig. 2.9a). The mechanochemical product exhibited only CsMnFeF6 in the F d¯3m space
group and no impurity phases (Fig. 2.2a).
53

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6 Table 2.2: Results of the Rietveld refinement of pristine hydrothermal CsMnFeF6 against the synchrotron powder diffraction data. Parameter CsMnFeF6 (1) CsMnFeF6 (2) CsMnFe2F9 MnF2 Space Group F d¯3m F d¯3m P
63/mcm P
42/mnm
a 10.55914(8) ˚A 10.5243(1) ˚A 7.5443(2) ˚A 4.8741(4) ˚A
b – – – –
c – – 7.7712(3) ˚A 3.3091(8) ˚A
α 90
◦ 90
◦ 90
◦ 90
◦
β – – – –
γ – – 120 –
Cs position (0.375, 0.375, 0.375) (0.375, 0.375, 0.375) (0, 0, 0) –
Mn position (0, 0, 0) (0, 0, 0) (0.483, 0, 0.25) (0, 0, 0)
Fe position (0, 0, 0) (0, 0, 0) (0.483, 0, 0.25) –
Fe2 position – – (0.483, 0, 0.25) –
F position (0.3187(2) 0.125, 0.125) (0.3170(4) 0.125, 0.125) (0.5, 0, 0) (0.3053, 0.3053, 0)
F2 position – – (0.4213, 0.2071, 0.25) –
weight % 66.60 29.01 3.30 1.09
mol % 65.06 28.35 2.45 4.14
Rwp 15.495 – – –
54

2.5. Supplemental Information
Table 2.3: Results of the Rietveld refinement of pristine ceramic CsMnFeF6 against the synchrotron powder diffraction data.
Parameter CsMnFeF6 (1) CsMnFeF6 (2) CsMnF3 MnF2
Space Group F d¯3m F d¯3m P
63/mmc P
42/mnm
a 10.54936(6) ˚A 10.5322(1) ˚A 6.23041(5) ˚A 4.86947(9) ˚A
b – – – –
c – – 15.1139(2) ˚A 3.31153(8) ˚A
α 90
◦ 90
◦ 90
◦ 90
◦
β – – – –
γ – – 120 –
Cs position (0.375, 0.375, 0.375) (0.375, 0.375, 0.375) (0, 0, 0.25) –
Cs2 position – – (0.3333, 0.6667, 0.9748) –
Mn position (0, 0, 0) – (0, 0, 0) (0, 0, 0)
Mn2 position – – (0.3333, 0.6667, 0.8485) –
Fe position (0, 0, 0) (0, 0, 0) – –
F position (0.3175(2), 0.125, 0.125) (0.3199(3) 0.125, 0.125) (0.766, 1.532, 0.25) (0.3053, 0.3053, 0)
F2 position – – (0.8193 0.6385, 0.7152) –
weight % 55.72 31.91 9.75 2.61
mol % 49.72 28.46 12.67 9.15
Rwp 12.044 – – –
55

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
Table 2.4: Results of the Rietveld refinement of pristine mechanochemical CsMnFeF6 against
the synchrotron powder diffraction data.
Parameter CsMnFeF6
Space Group F d¯3m
a 10.5448(1) ˚A
b –
c –
α 90◦
β –
γ –
Cs position (0.375, 0.375, 0.375)
Mn position (0, 0, 0)
Fe position (0, 0, 0)
F position (0.31751(1), 0.125, 0.125)
Rwp 6.593
Table 2.5: Results of the Rietveld refinement of carbon-coated mechanochemical CsMnFeF6
against the synchrotron powder diffraction data.
Parameter CsMnFeF6
Space Group F d¯3m
a 10.5471(1) ˚A
b –
c –
α 90◦
β –
γ –
Cs position (0.375, 0.375, 0.375)
Mn position (0, 0, 0)
Fe position (0, 0, 0)
F position (0.31721(1), 0.125, 0.125)
Rwp 6.656
56

2.5. Supplemental Information
Table 2.6: Results from the Rietveld refinements of ex situ mechanochemical CsMnFeF6
electrodes against the synchrotron powder diffraction data.
Parameter ox 1 red 1
Space Group F d¯3m F d¯3m
a 10.5441(2) ˚A 10.5320(2) ˚A
b – –
c – –
α 90◦ 90◦
β – –
γ – –
Cs position (0.375, 0.375, 0.375) (0.375, 0.375, 0.375)
Mn position (0, 0, 0) (0, 0, 0)
Fe position (0, 0, 0) (0, 0, 0)
F position (0.3166(1), 0.125, 0.125) (0.3158(2), 0.125, 0.125)
Rwp 7.817 7.963
Table 2.7: Additional results from the Rietveld refinements of ex situ mechanochemical
CsMnFeF6 electrodes against the synchrotron powder diffraction data (Fig. 2.14).
Parameter ox 2 red 2
Space Group F d¯3m F d¯3m
a 10.5408(3) ˚A 10.5308(3) ˚A
b – –
c – –
α 90◦ 90◦
β – –
γ – –
Cs position (0.375, 0.375, 0.375) (0.375, 0.375, 0.375)
Mn position (0, 0, 0) (0, 0, 0)
Fe position (0, 0, 0) (0, 0, 0)
F position (0.3159(3), 0.125, 0.125) (0.3166(2), 0.125, 0.125)
Rwp 11.593 9.710
57

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6 Table 2.8: Additional results from the Rietveld refinements of ex situ mechanochemical CsMnFeF6 electrodes against the synchrotron powder diffraction data (Fig. 2.15). Parameter ox 3 red 3 ox 4 Space Group F d¯3m F d¯3m F d¯3
m
a 10.5346(5) ˚A 10.531(2) ˚A 10.533(1) ˚A
b – – –
c – – –
α 90
◦ 90
◦ 90
◦
β – – –
γ – – – Cs position (0.375, 0.375, 0.375) (0.375, 0.375, 0.375) (0.375, 0.375, 0.375)
Mn position (0, 0, 0) (0, 0, 0) (0, 0, 0)
Fe position (0, 0, 0) (0, 0, 0) (0, 0, 0)
F position (0.3205(3), 0.125, 0.125) (0.3190(8), 0.125, 0.125) (0.3190(4), 0.125, 0.125)
Rwp 9.808 16.716 12.387
58

2.5. Supplemental Information
(a)
(c)
(b)
(d)
1.6 2.4 3.2 4.0 4.8 5.6 6.4
Q (Å-1)
Intensity (arb. units)
observed
calculated
difference
1.6 2.4 3.2 4.0 4.8 5.6 6.4
Q (Å-1)
Intensity (arb. units)
observed
calculated
difference
1.6 2.4 3.2 4.0 4.8 5.6 6.4
Q (Å-1)
Intensity (arb. units)
observed
calculated
difference
1.6 2.4 3.2 4.0 4.8 5.6 6.4
Q (Å-1)
Intensity (arb. units)
observed
calculated
difference
Figure 2.14: Rietveld refinements of ex situ mechanochemical CsMnFeF6 synchrotron XRD
data collected at beamline 11-BM at the APS. Electrode after (a) one oxidation (ox 1), (b)
one full cycle (red 1), (c) one full cycle and a second oxidation (ox 2), and (d) two full cycles
(red 2). The unrefined, sharp peaks in the ox 2 pattern correspond to residual TBAF and
associated organic products from electrolyte degradation. Orange reflections correspond to
the refined cubic pyrochlore CsMnFeF6 structure.
59

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
1.6 2.4 3.2 4.0 4.8 5.6 6.4
Q (Å-1)
Intensity (arb. units)
observed
calculated
difference
(a)
1.6 2.4 3.2 4.0 4.8 5.6 6.4
Q (Å-1)
Intensity (arb. units)
observed
calculated
difference
(b)
1.6 2.4 3.2 4.0 4.8 5.6 6.4
Q (Å-1)
Intensity (arb. units)
observed
calculated
difference
(c)
Figure 2.15: Rietveld refinements of ex situ mechanochemical CsMnFeF6 synchrotron XRD
data collected at beamline 2-1 at SSRL. Electrode after (a) two cycles and one oxidation (ox
3), (b) three full cycles (red 3), and (c) three full cycles and one oxidation (ox 4). Orange
reflections correspond to the refined cubic pyrochlore CsMnFeF6 structure.
1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0
Q (Å-1)
Intensity (arb. units)
C-coated
ox 1
red 1
ox 2
red 2
1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0
Q (Å-1)
Intensity (arb. units)
ox 3
red 3
ox 4
red 4
(b)
(a)
Figure 2.16: Comparison of ex situ synchrotron XRD patterns from (a) cycles 1 and 2,
collected at beamline 11-BM at the APS, and (b) cycles 3 and 4, collected at beamline 2-1
at SSRL. Gray boxes indicate regions of Q that Figure 6 of the main text is zoomed into.
60

2.5. Supplemental Information
1.5 2.0 2.5 3.0 3.5
Q (Å-1)
FeF2
2.0 2.5 3.0 3.5
MnF2
Intensity (arb. units)
FeF3 red 3
CsMnFeF6
MnF3
1.5 2.0 2.5 3.0 3.5
Q (Å-1)
BiF3 (trigonal)
2.0 2.5 3.0 3.5
BiF3 (orthorhombic)
Intensity (arb. units)
Bi red 3
CsMnFeF6
BiF3 (cubic)
(a) (b)
Figure 2.17: Phase-matching of the synchrotron XRD pattern of the ex situ CsMnFeF6 sample collected after reduction 3 at beamline 2-1 at SSRL. (a) Comparison of the experimental
“red 3” pattern to VESTA117 simulated patterns of CsMnFeF6 and the binary fluorides
MnF2, MnF3, FeF2, and FeF3. (b) Comparison of the experimental pattern to VESTA117
simulated patterns of CsMnFeF6 and Bi metal, the cubic polymorph of BiF3, orthorhombic
BiF3, and trigonal BiF3.
1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0 4.4
Q (Å-1)
Intensity (arb. units)
observed
Fd3m (pyrochlore)
I222 (indexed)
difference
Figure 2.18: Multi-phase Pawley fit of the ox 3 ex situ pattern including the cubic pyrochlore
phase (F d¯3m, a = 10.5346 ˚A) in orange and the orthorhombic indexed phase (I222, a =
6.5323 ˚A, b = 9.2318 ˚A, c = 7.1258 ˚A) in green.
61

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
2.5.5 Electrochemical Characterization
0.0 0.5 1.0 1.5
V vs. Bi/Bi3+
-100
-50
0
50
100
dQ/dV (mAhg-1
V-1)
cycle 6
cycle 7
0 15 30 45 60 75
Capacity (mAh g-1)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
V vs. Bi/Bi3+
0 100 200 300 400
Z’ (Ohm)
0
100
200
300
400
-Z’’ (Ohm)
pre-cycling
post-ox1
post-red1
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
V vs. Bi/Bi3+
cycle 1
cycle 2
cycle 3
cycle 4
(a)
(b)
Figure 2.19: Galvanostatic cycling of a F-ion cell, with a working electrode of ceramically
synthesized CsMnFeF6 and a Bi/BiF3 composite counter electrode, cycled at room temperature at a rate of C/20 between 0.0 and 1.4 V vs. Bi/BiF3. (a) Cycles 1–4. Inset: PEIS
performed before cycling, after the first oxidative cycle “ox 1”, and after the first reductive
cycle “red 1”. (b) Cycles 5–9, cycles six and seven are shown in orange and blue, respectively.
Inset: Differential capacity plot derived from the sixth and seventh cycles.
62

2.5. Supplemental Information
0.0 0.5 1.0 1.5
V vs. Bi/Bi3+
-75
-50
-25
0
25
50
75
dQ/dV (mAhg-1
V-1)
cycle 6
cycle 7
0 10 20 30 40 50 60
Capacity (mAh g-1)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
V vs. Bi/Bi3+
0 400 800 1200
Z’ (Ohm)
0
400
800
1200 -Z’’ (Ohm)
pre-cycling
post-ox1
post-red1
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
V vs. Bi/Bi3+
cycle 1
cycle 2
cycle 3
cycle 4
(a)
(b)
Figure 2.20: Galvanostatic cycling of a F-ion cell, with a working electrode of hydrothermally synthesized CsMnFeF6 and a Bi/BiF3 composite counter electrode, cycled at room
temperature at a rate of C/20 between 0.0 and 1.4 V vs. Bi/BiF3. (a) Cycles 1–4. Inset:
PEIS performed before cycling, after the first oxidative cycle “ox 1”, and after the first reductive cycle “red 1”. (b) Cycles 5–9, cycles six and seven are shown in orange and blue,
respectively. Inset: Differential capacity plot derived from the sixth and seventh cycles.
63

Room Temperature Electrochemical Fluoride (De)Insertion into CsMnFeF6
2.5.6 X-ray Absorption Spectroscopy
Energy (eV)
6540 6545 6550 6555 6560
0
d
2
µ(E)/dE
2 Intensity (arb. units)
MnF2
MnF3
CsMnFeF6
0
d
2
µ(E)/dE
2
Mn K-edge
(a)
(b)
(c)
(d)
(f)
(e)
6540 6544 6548 6552 6556 6560
Energy (eV)
0
d
2
µ(E)/dE
2 Intensity (arb. units)
MnF2
MnF3
ox 3
red 3
0
d
2
µ(E)/dE
2
Mn K-edge
Figure 2.21: X-ray absorption spectroscopy results showing pristine and ex situ samples
compared to MnF2 and MnF3 standards (a,d). Plots of the second derivatives of the spectra
have the zero-crossings used to define the Mn K-edges marked with green ticks (b, c, e).
7116 7120 7124 7128 7132 7136
Energy (eV)
0
d
2
µ(E)/dE
2
0
d
2
µ(E)/dE
2 Intensity (arb. units)
FeF2
FeF3
ox 3
red 3
Fe K-edge
(d)
(e)
(f)
7116 7120 7124 7128 7132 7136
Energy (eV)
0
d
2
µ(E)/dE
2
0
d
2
µ(E)/dE
2 Intensity (arb. units)
FeF2
FeF3
CsMnFeF6
Fe K-edge
(a)
(b)
(c)
Figure 2.22: X-ray absorption spectroscopy results showing pristine and ex situ samples
compared to FeF2 and FeF3 standards (a,d). Plots of the second derivatives of the spectra
have the zero-crossings used to define the Fe K-edges marked with green ticks (b, c, e).
64

Chapter 3
On the Structural Origin of Fast Liion Cycling in Tetragonal Bronze-type
Nb8W9O47
3.1 Introduction
Fast-cycling electrode materials are critical for the continued development of Li-ion batteries
and the modern technologies that rely on them. However, despite the massive body of work
studying Li-ion insertion hosts, many questions remain about how to identify new materials
that can match or exceed the performance of graphite and the rock salt family. The distortions that crystal structures experience during the (de)insertion of lithium ions impact the
rate performance and reversibility of the electrode material.45 Therefore, an understanding of
how to mitigate or leverage these structural transformations must be developed to establish
design rules for new electrode materials.
For example, Bashian et al. recently showed that the electrochemical lithiation of ReO3
results in highly correlated polyhedral rotations.61 While these distortions are reversible, their
magnitude negatively affects the electrochemical performance of ReO3 (i.e., large voltage polarizations, rapid capacity fade). In fact, in 1983, Cava, Murphy, and Zahurak discussed the
65

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
detrimental impact of this type of transformation on lithium ion mobility in a host crystal lattice.38 They suggested that the introduction of crystallographic shear planes into the
corner-shared ReO3 framework could stabilize the structure against severe distortions by limiting octahedral rotations at the edge-sharing boundaries, thereby enhancing (de)lithiation
kinetics and reducing hysteresis.38 Therefore, it is no surprise that the Wadsley-Roth phases,
derived from ReO3 by introducing shear planes to yield alternating layers of edge-sharing
and corner-sharing octahedral units,68–76 have attracted an increasing amount of attention as
Li-ion intercalation hosts.39,42,43,47,66,67,156 TiNb2O7 is an example of one such Wadsley–Roth
shear phase that exhibits robust cycling capabilities due to the presence of orthogonal crystallographic shear planes that efficiently limit structural distortions during lithium insertion
and removal.43,78,79 The structure is composed of blocks of corner-connected Ti/Nb–O octahedra (ReO3-type), where the blocks are three octahedra wide by three octahedra long and
are infinitely connected in each plane by crystallographic shear planes, denoted (3×3)∞.
77
TiNb2O7 boasts a practical capacity of 388 mAh/g (5 Li+ per f.u.), and the structural rigidity is believed to be, at least in part, responsible for its ability to maintain most of this
capacity even at high rates.43,78,79
An alternative avenue for imparting structural rigidity in ReO3-derived phases is the
“pentagonal column” structural motif seen in variations of the tetragonal tungsten bronze
(TTB) crystal structure.86–88 Take, for example, the composition that this report focuses
on: Nb8W9O47 (4Nb2O5·9WO3), first reported in 1965 by Wadsley and Roth.70–73 The orthorhombic crystal structure of Nb8W9O47 (P21212, #18) is often referred to as “TTB-type”,
and the unit cell is best described as a TTB superstructure (Fig. 3.1).157 This superstructure can be derived by tripling the b-axis of a typical TTB unit cell and then filling 1/3 of
the pentagonal tunnels with a transition metal. Filling these tunnels creates 7-coordinate
transition metal sites that are edge-sharing in the ab-plane with 5 other MX6 octahedra and
corner-sharing with other MX7 bipyramids in the c-direction, producing one-dimensional, infinite “pentagonal columns” that run through the crystal structure (Fig. 3.1).86 Other than
66

3.1. Introduction
these columns, the TTB-type Nb8W9O47 crystal structure is a relatively open framework
featuring vacant 3-sided, 4-sided, and 5-sided tunnels running along the c-direction.
Figure 3.1: Crystal structure of TTB-type Nb8W9O47 viewed down the [001] direction. All
transition metal sites in the unit cell are mixed Nb and W. Blue polyhedra depict the sites
with approximately 50:50 Nb–W occupancy and gray polyhedra are used for sites with a
majority W occupancy. The 7-coordinate transition metal sites depicted without polyhedra
form the 1D pentagonal columns. Crystal structure depicted using VESTA.117
Griffith et al. reported on the fast cycling of metastable, bronze-type polymorphs of
Nb2O5, owing the high capacity and impressive rate performance to the stability of the
unique “room-and-pillar” framework structure, where the “pillars” are akin to the pentagonal
columns.40 Griffith et al. followed this study with a report on multi-electron redox and fast
cycling in the TTB-type composition Nb18W16O93, which notably exhibits these impressive
cycling capabilities in micron-scale active material particles.41 In this work, the authors
hypothesized that the reversibility and high-rate capabilities could be, at least in part, due
to the pentagonal column motifs preventing polyhedral rotations within the corner-connected
regions of the structure that surround the vacant tunnels.41 More recently, Luo et al. reported
67

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
on similarly impressive fast cycling and the structural evolution of TTB-type Nb2W3O14.
44
While the crystal structure of Nb8W9O47 is similar to that of Nb18W16O93, studied by Griffith
et al.,
41 and Nb2W3O14, studied by Luo et al.,
44 these compositions differ in the distribution
of filled pentagonal tunnels and in their Nb:W ratios.
Chemical and electrochemical insertion of lithium into Nb8W9O47 has been previously
reported.158,159 These initial reports include ex situ powder X-ray diffraction of the lithiated
phase, showing what appears to be a dramatic, albeit reversible, structural change upon
lithiation. Yan et al. followed these reports with a study on a surface-modified variant of
Nb8W9O47 (nitrided nanofibers), where they presented extended electrochemical cycling and
observed the same structural change during (de)lithiation.160 However, the existing literature on Nb8W9O47 as a Li-ion intercalation host lacks an in-depth analysis of the structural
transformation that occurs during the (de)intercalation of lithium, which is critical to understanding why the transformation is so reversible, even at high rates. While many reports have
emphasized the importance of suppressing structural transformations during (de)lithiation,
it is important to acknowledge that the transition metal is still undergoing redox and has to
satisfy its valence somehow (i.e., bond lengthening or contraction). Probing the structural
change more carefully will help improve our understanding of the structural motifs, like pentagonal columns, that might permit the necessary local distortions for lithium insertion and
diffusion but preclude more problematic coherent, long-range structural transformations.
Toward this goal, we report on the synthesis and electrochemical properties of TTB-type
Nb8W9O47 and operando powder X-ray diffraction to investigate the structural changes upon
lithium (de)intercalation.
68

3.2. Experimental Methods
3.2 Experimental Methods
3.2.1 Material Synthesis
Nb8W9O47 was synthesized via a rapid, microwave-assisted technique, based on methods previously described by Levin et al.161 Stoichiometric amounts of dried Nb2O5 (Sigma-Aldrich,
99.99%) and WO3 (Acros Organics, 99+%) were weighed and mixed intimately using an
agate mortar and pestle, then pressed into 13 mm pellets (3.0 tons, 30 s). The pellets were
sealed in silica ampules under vacuum, placed in a bed of carbon susceptor (activated charcoal) within an alumina crucible, and heated in a Panasonic 1200 W household microwave
oven set to 50% power for 18 min. These conditions yielded dark blue pellets which produced
a fine, blue powder once ground.
3.2.2 Powder X-ray Diffraction
Laboratory powder X-ray diffraction (XRD) patterns were collected on a Bruker D8 Advance
diffractometer in Bragg-Brentano geometry equipped with a Cu Kα source (λ1=1.5406 ˚A,
λ2=1.5444 ˚A) and a LynxEye XE-T detector. High-resolution synchrotron powder XRD data
was collected at beamline 2-1 at the Stanford Synchrotron Radiation Lightsource (SSRL)
using an average wavelength of 0.729213 ˚A. The resulting XRD patterns were refined against
published structures using the Rietveld method, as implemented in the TOPAS-Academic
suite.145
3.2.3 Scanning Electron Microscopy
Scanning electron microscopy (SEM) images were obtained using a Nova NanoSEM 450 field
emission scanning electron microscope. Imaging was conducted with a 5 kV accelerating
voltage and 5 mm working distance in secondary electron detector configuration. Samples
were spread onto double-sided carbon tape (Ted Pella) and sputter-coated with carbon for
69

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
20 s before measurement.
3.2.4 Electron Probe Microanalysis
Elemental analysis was performed using electron probe microanalysis (EPMA) to confirm the
synthesized materials had a composition (Nb:W ratio) close to the targeted Nb:W ratio of
8:9 for Nb8W9O47. EPMA data were collected using a JEOL JXA-8200 system with a 10 kV
and 5 nA focused beam. The focused electron beam was 150 nm in diameter. Standards
for analysis were W metal and NbO, and data were collected on 3 spots per standard.
Approximately 40 mg of sample powder was pressed with a hydraulic press for 5 minutes at
2 tons into a 6 mm diameter pellet. Carbon was sputtered onto pellets using graphite under
vacuum, and transferred into the instrument. Data were collected on 10 spots per sample.
3.2.5 X-ray Photoelectron Spectroscopy
X-ray photoelectron spectroscopy (XPS) was used to evaluate the Nb and W oxidation states
in the Nb8W9O47 product. Pristine WO3 (Acros Organics, 99+%), NbO2 (Strem Chemicals,
99+%), and Nb2O5 (Sigma-Aldrich, 99.99%) standards were also measured to benchmark
the expected W6+, Nb4+, and Nb5+ binding energies. Samples were prepared by grinding
each material in a 1:1 wt % mixture with Super P carbon (Alfa Aesar) for 5 minutes in
an alumina mortar and pestle. The resulting mixtures were affixed to carbon tape (Ted
Pella) on the XPS sample holder. XPS data were collected using a Kratos Axis Ultra
system at a pressure < 3×10−9
torr. Samples were irradiated with a monochromatic Al-Kα
source (1486.7 eV) at 150 W. Low-resolution survey spectra were acquired between binding
energies of 1 and 1200 eV at a pass energy of 40 eV (Fig. 3.11). High-resolution scans were
collected on the Nb 3d and W 4f lines at a pass energy of 20 eV (Fig. 3.12, 3.13). The XPS
data were analyzed using the CasaXPS software, and individual peaks were fit with Shirely
backgrounds and mixed Gaussian–Lorentzian line shapes (GL(30)). High-resolution spectra
were referenced to adventitious carbon (285.0 eV).162
70

3.2. Experimental Methods
3.2.6 Electrochemical Characterization
All Li-ion cell components and electrodes were dried under vacuum at 110 ◦C for at least
10 h before cell assembly. Cell assembly was performed in an argon glovebox. Stainless
steel Swagelok cells with two borosilicate glass fiber (Whatman GF/D) separators were
used as electrochemical test cells. A 1.0 M solution of LiPF6 in a 1:1 mixture of EC:DMC
(Sigma-Aldrich, battery grade) was used as the liquid electrolyte. Polished lithium foil
(Sigma-Aldrich) was punched as 12 mm discs and used as the combined counter/reference
electrode.
Two different types of working electrodes were prepared: (1) slurry cast on Cu current
collectors or (2) free-standing pellets. An active material/conductive carbon mixture was
prepared by grinding Nb8W9O47 and Super P conductive carbon (Alfa Aesar) in an agate
mortar and pestle until homogeneous. For slurry cast electrodes, the active material/conductive carbon mixture was then combined with polyvinylidene fluoride (PVDF, MTI), for
an active–carbon–binder weight ratio of 75:20:5, in a in a minimal amount of N-methyl-2-
pyrrolidone (≥99.0%, Sigma-Aldrich) to form a slurry. The slurry was cast on copper foil
to 10 µm thick and dried at 40 ◦C overnight, then punched as 12 mm electrodes and dried
at 110 ◦C under vacuum for at least 10 h before use. The slurry cast electrodes had a
typical active mass loading of ∼1 to 3 mg/cm2
. For the free-standing pellet electrodes, the
active material/conductive carbon mixture was combined with a different polymer binder,
polytetrafluoroethylene (PTFE, MTI), for an active–carbon–binder weight ratio of 75:20:5.
Free-standing pellets were pressed in a 10 mm die at 1.0 ton for 15 s and had a typical active
mass loading of ∼8 to 12 mg/cm2
.
3.2.7 Operando X-ray Diffraction
Operando XRD data were collected in-house on the previously described Bruker D8 Advance
diffractometer using a custom-made electrochemical cell with a PEEK body, stainless steel
71

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
electrical contacts, and an X-ray transparent Be window (SPI Supplies, 0.25 mm thick). The
Be window served as a current collector and allowed for X-ray penetration so that diffraction
patterns could be collected continuously while cycling galvanostatically.
Free-standing pellet electrodes (described previously) were cycled against polished Li foil
using a BioLogic SP-200 potentiostat at a C/10 rate with two Whatman glass fiber separators
(GF/D) soaked with 1 M LiPF6 in 1:1 EC:DMC electrolyte (Sigma-Aldrich, battery grade).
Patterns were collected approximately every 20 min over an angle range of 13.5◦
to 40.5◦ 2θ
throughout the electrochemical cycling.
3.3 Results and Discussion
Phase pure Nb8W9O47, which adopts the orthorhombic space group P21212 (#18), was synthesized via a rapid, microwave-assisted method, previously unreported for this material
and described in detail in the Experimental Methods. Rietveld refinement of the structure
against synchrotron powder X-ray diffraction (XRD) data was performed on the product
(Fig. 3.2), with the resulting structural parameters given in Tables 3.1 and 3.2. Scanning
electron microscopy (SEM) images show that this synthesis yields rod-like particles, a common morphology for bronze materials due to preferential growth in the c-direction,163 with
particle sizes ranging from ∼1–8 µm (Fig. 3.2 inset). Electron probe microanalysis (EPMA)
was performed to assess the Nb:W elemental ratio of the Nb8W9O47 product (Table 3.3).
EPMA results indicate an average Nb:W ratio of approximately 0.836, which is suggestive
of a small Nb deficiency but is overall very close to the target ratio of 0.889 and the ratio of 0.829 obtained from the synchrotron XRD Rietveld refinement. X-ray photoelectron
spectroscopy (XPS) was performed to assess Nb and W oxidation states in the Nb8W9O47
product and indicated that, at the penetration depth of the technique, Nb exists in the 5+
oxidation state and W in the 6+ oxidation state, as expected (Fig. 3.12, 3.13). Pristine
WO3, NbO2, and Nb2O5 standards were measured concurrently to benchmark the expected
72

3.3. Results and Discussion
W6+, Nb4+, and Nb5+ binding energies.
1 2 3 4 5
Q (Å-1)
Intensity (arb. units)
observed
calculated
difference
*
*
Figure 3.2: Rietveld refinement of Nb8W9O47 synchrotron powder XRD data. Black tick
marks indicate reflections of Nb8W9O47 in the P21212 space group. Peaks marked with a
green star correspond to diamond powder diluent. Inset: SEM image of pristine Nb8W9O47
particles.
Nb8W9O47 was cycled using two different types of working electrodes depending on the
analytical technique: (1) slurry cast on Cu current collectors for general cycling or (2)
free-standing pellets for operando XRD. Working electrodes were cycled versus polished Li
metal combined counter-reference electrodes with glass fiber separators (Whatman GF/D)
soaked in 1.0 M LiPF6 in a 1:1 (v/v) mixture of ethylene carbonate (EC) and dimethyl
carbonate (DMC). First, a window-opening cyclic voltammetry (CV) experiment, where
the upper voltage limit was set to 3.0 V and the lower voltage limit was lowered with
each successive sweep from 1.4 V to 1.0 V, was performed to identify the optimal voltage
window for cycling (Fig. 3.9a). Akin to previous reports of similar TTB-type phases,43,44 this
experiment revealed an irreversible anodic feature below 1.2 V indicative of decomposition,
irreversible structural changes, or both.
Galvanostatic cycling with potential limitation (GCPL) was performed at various rates
between 3.0 and 1.2 V vs. Li/Li+ (Fig. 3.3). The capacity of Nb8W9O47 obtained at rates
between C/10 and C/2 is greater than the theoretical capacity of 145 mAh/g calculated for
73

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Li
+ per TM
0 5 10 15 20 25 30
cycle number
0
40
80
120
160
Capacity (mAh/g)
discharge
charge
0 25 50 75 100 125 150
Capacity (mAh/g)
1.5
2.0
2.5
3.0
V vs. Li/Li
+
C/5
C/5
C/2
C/2
1C
1C
2C
2C
5C
5C
10C
10C
20C
20C
40C
40C
60C
60C
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Li+ per TM in Nb8
W9
O47
C/5
(a)
(b)
Figure 3.3: Variable-rate galvanostatic cycling of Nb8W9O47 vs. Li/Li+ at room temperature,
shown in terms of specific capacity and lithium per transition metal (TM) in Nb8W9O47.
the transfer of 1 electron per transition metal in the formula unit (17 total). Therefore, at
these rates, more than one Li ion can be inserted per transition metal and multi-electron
redox is achievable. At rates faster than C/2, the capacity dips below the theoretical maximum, but over 65% of this capacity is maintained at rates up to 20C (3-minute discharge).
The discharge profile of Nb8W9O47 features two significant changes in slope from (1) 2.2 to
2.0 V and (2) 1.75 to 1.55 V (Fig. 3.3b, 3.4). The first, higher voltage region features a
very flat plateau and the second, lower voltage region is marked by a more gradual, sloping
plateau. These plateaus are maintained on charge with a small polarization of ∼60–70 mV
(C/10, 1.2 V cutoff; Fig. 3.4a,c), and the overall voltage profile is retained for multiple cycles, suggesting the (de)lithiation of Nb8W9O47 is a highly reversible process. Notably, while
cycling down to 1.0 V achieves higher capacities, lower first cycle Coulombic efficiencies and
larger voltage polarizations are observed (Fig. 3.4b,d). These results further suggest that
decomposition or an irreversible structural change occurs below 1.2 V and negatively impacts
the cell performance over multiple cycles.
CV sweep rate dependence studies were conducted to investigate the kinetic properties
74

3.3. Results and Discussion
0.0 0.4 0.8 1.2
Li+ per TM in Nb8
W9
O47
1.0
1.5
2.0
2.5
3.0
V vs. Li/Li
+
cycle 1
cycle 2
0 60 120 180
Capacity (mAh/g)
1.0
1.5
2.0
2.5
3.0
V vs. Li/Li
+
0.0 0.4 0.8 1.2
cycle 1
cycle 2
0 60 120 180
(a)
(c) (d)
(b)
Figure 3.4: A comparison of the first two galvanostatic cycles performed at a rate of C/10
in a 1.2 to 3.0 V (a,c) window and a 1.0 to 3.0 V (b,d) window to show the differences in
capacity, voltage polarization, and Coulombic efficiency.
of Nb8W9O47 (Fig. 3.5). During a CV experiment, the peak current response (Ip) is related
to the sweep rate (v) according to the equation: Ip = avb
, where a is a constant value
and b is the power-law exponent, hereafter referred to as the “b-value.”164 A b-value of 0.5
indicates the active material’s charge storage kinetics are bulk diffusion-controlled, whereas
a b-value of 1.0 suggests the kinetics are surface diffusion-controlled.44,164 For this study, CV
sweeps were performed at multiple, increasing rates in a 1.2 to 3.0 V window. Analysis of
the sweeps between 0.2 and 1.0 mV/s according to the power law described above yields
an average b-value of 0.79 for the peaks centered at 2.1 V and 0.94 for the peaks centered
at 1.65 V (Fig. 3.5a). These results, similar to those obtained by Luo et al. for TTBtype Nb2W3O14,
44 indicate that the charge storage kinetics of Nb8W9O47 are mainly surface
diffusion-controlled. However, the lower b-value for the peaks around 2.1 V suggests that
some bulk diffusion-controlled character also impacts the kinetics of this redox reaction.
Prior work on TTB-type compositions utilized operando XRD coupled with structural
analysis methods that only refine lattice and peak shape parameters (i.e., Pawley or Le Bail
fits). While tracking the changes in these metrics is incredibly informative, these meth75

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
0.93
0.94
0.76
0.82
(a)
(b)
1.2 1.5 1.8 2.1 2.4 2.7 3.0 3.3
V vs. Li/Li+
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
Current Density (A/g)
-0.8 -0.4 0.0
log(ν) (mV/s)
-1.0
-0.5
0.0
log(I) (A/g)
cathodic 1
anodic 1
cathodic 2
anodic 2
1.2 1.5 1.8 2.1 2.4 2.7 3.0
V vs. Li+
-3
-2
-1
0
1
2
3
Current Density (A/g)
0.2 mV/s
0.4 mV/s
0.6 mV/s
0.8 mV/s
1.0 mV/s
2.0 mV/s
4.0 mV/s
Figure 3.5: Cyclic voltammetry b-value analysis. (a) Nb8W9O47 vs. Li+ CV at 0.2 mV/s
with the major anionic and cationic redox peaks labeled with their respective, calculated
b-values. Inset: b-value fitting of peaks. (b) Nb8W9O47 vs. Li+ CV at rates varying from
0.2 to 4.0 mV/s.
ods offer a limited understanding of the structural evolution upon (de)lithiation as they do
not provide insight into atomic motion occurring within the unit cell during charge and
discharge. Rietveld refinements, which take the atomic “contents” of the unit cell into account during the fitting process, are therefore critical to better understanding relationships
between structure and electrochemical properties. Tracking transition metal motion by refining atomic positions during (de)lithiation should offer more insight into the impressive
electrochemical properties of Nb8W9O47 and other TTB-type phases.
Operando XRD, using Cu Kα radiation in-house, was performed at a rate of C/10 in
a custom-made electrochemical cell equipped with an X-ray transparent beryllium window.
76

3.3. Results and Discussion
Two individual cells were cycled down to 1.2 and 1.0 V to compare the structural evolution of Nb8W9O47 during electrochemical (de)lithiation and, in particular, investigate the
irreversibility seen below 1.2 V in the electrochemical studies. While significant peak shifts
and intensity changes are observed in the operando diffraction during discharge, all of the
changes are entirely reversible, as the original diffraction pattern is recovered following the
subsequent charge cycle (Fig. 3.6a, 3.16). These changes are also reversible for multiple
cycles, no matter the lower voltage cutoff, as shown in Figures 3.14 and 3.15.
A sequential Pawley fit was first performed to identify lattice parameter trends and any
symmetry changes that occur during cycling. Despite the dramatic changes to the diffraction
pattern over the course of discharge, this initial analysis suggests the space group (P21212)
remains unchanged during electrochemical cycling. Notably, the higher voltage plateau (2.2
to 2.0 V) appears to exhibit two-phase behavior in the operando XRD, while the rest of the
cycling can be described as solid-solution behavior. The sequential Pawley fit corroborates
this observation, requiring the addition of a second phase of the same space group, but with
slightly expanded a and b parameters and a contracted c parameter, to satisfactorily fit the
diffraction from the high voltage plateau.
Next, a sequential Rietveld refinement where only the lattice parameters were allowed
to refine was performed and showed clear issues with the quality of fit, especially at higher
lithium contents (Fig. 3.6b, 3.18, 3.19) This was followed by a second sequential Rietveld
refinement where the atomic coordinates of the transition metal sites were also refined.
Allowing the transition metals to displace improves the fit noticeably, particularly at the
end of discharge where the initial refinement falls short (Fig. 3.6b, 3.18). While this adds 25
additional parameters, allowing the transition metals to move is clearly necessary to obtain a
more realistic picture of the structural evolution during (de)lithiation. Oxygen positions were
not refined because the lower scattering power of oxygen makes refinement of their atomic
coordinates unreliable and may falsely improve the fit without any chemical meaning.
The lattice parameter trends tracked over the first discharge reveal anisotropic unit cell
77

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
1.6 1.8 2.0 2.2
Q (Å-1)
Intensity (arb. units)
1.0 2.0 3.0
V vs. Li/Li+
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Li per TM in Nb
W
8
O
9 47
(a) (b)
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Li+ per TM in Nb8
W9
O47
1.0
1.5
2.0
2.5
V vs. Li/Li
+
0.0
3.0
6.0
∆vol (%)
-2.0
0.0
2.0
∆c (%)
0.0
1.0
2.0
∆b (%)
0.0
2.0
4.0
∆a (%)
C/10
1.2 1.5 1.8 2.1 2.4 2.7
Q (Å-1)
Rietveld (TM open)
Intensity (arb. units)
Rietveld (TM locked)
obs
Pawley
diff
(c)
Figure 3.6: (a) In-house operando XRD of Nb8W9O47 discharged to 1.0 V then charged
to 3.0 V at a C/10 rate vs. Li/Li+. (b) A comparison of the sequential Pawley fit (top
panel), Rietveld refinement with locked transition metal positions (middle), and Rietveld
refinement with refined transition metal positions (bottom) of the operando XRD pattern
of Li18.1Nb8W9O47 (∼1.1 Li+ per TM). (c) Lattice parameter changes extracted from the
sequential Rietveld refinement with refined transition metal positions, correlated to the first
discharge voltage profile of the 1.0 V cutoff operando XRD.
changes upon lithiation (Fig. 3.6c, 3.17). As shown in Figure 3.17 and Tables 3.4–3.9,
the extracted lattice parameter trends from the sequential Pawley fit, Rietveld refinement
with locked transition metal positions, and Rietveld refinement with refined transition metal
positions all align with one another. The unit cell of phase 1 first expands in the a and b
directions and remains steady in the c direction. Upon entering the two-phase region, the
a–b plane of phase 1 continues to expand while c is unchanged. Phase 2 also undergoes a–b
plane expansion, but exhibits contraction in the c direction. Increasing a and b alongside
decreasing c is common in bronze-derived structures, suggestive of lithium first occupying
channels in the a–b plane.41,44 After the two-phase region (>0.25 Li+ per TM), all of the
lattice parameters and the unit cell volume steadily increase and the b ≈ 3a superstructure
relationship is maintained until 0.55 Li+ per TM, at which point the b parameter contracts
significantly and the unit cell volume decreases. Although the unit cell volume begins to
increase again around 0.7 Li+ per TM, the volume expansion at 1.0 Li+ per TM is only
3.2%, equivalent to the volume at 0.55 Li+ per TM. The small volume change compared to
unlithiated Nb8W9O47 is promising, as it should help mitigate electrode cracking and improve
78

3.3. Results and Discussion
long-term cycling performance. However, the unit cell volume increases significantly as the
lithium content exceeds 1.0 Li+ per TM, exhibiting a 5.3% volume expansion at the end
of the discharge down to 1.0 V. On the other hand, the volume expansion is limited to
3.3% if the discharge is stopped at 1.2 V. The rapid 2% unit cell volume expansion between
1.2 and 1.0 V is more than likely detrimental for longer term cycling—as seen in the lower
first cycle Coulombic efficiency, faster capacity fade, and larger voltage polarization of the
cell cycled to 1.0 V—and clearly outweighs the benefit of the higher capacity.
In addition to tracking changes in the refined lattice parameters from the sequential Rietveld refinement, crystal structure visualization software can be utilized to assess trends
in transition metal motion over the course of lithiation. Transition metal displacements in
Figure 3.7: (a) A fraction of the pristine Nb8W9O47 unit cell viewed down the [001] direction
with the TM sites labeled numerically (as in the Rietveld refinement, Table S1, S6–S8).
(b) The fraction after discharge down to 1.0 V (1.25 Li+ per TM), with TM site 5 highlighted
in blue to emphasize the TM motion. (c) The same fraction of the pristine unit cell viewed
down the [100] direction with the TM sites labeled numerically. (d) The fraction after
discharge down to 1.0 V (1.25 Li+ per TM), with one layer highlighted in blue to emphasize
the TM motion. Crystal structures are depicted using VESTA.117
the ab-plane can be visualized by viewing the crystal structure down the [001] direction
(Fig. 3.7a,b, 3.22). From this view, significant off-centering of the transition metals within
their octahedra, and corresponding M–O bond lengthening and contraction, occurs during
79

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
lithiation. Comparing the crystal structure of pristine Nb8W9O47, shown in Figure 3.7a, to
that of the fully lithiated phase (Li21.2Nb8W9O47), shown in Figure 3.7b, it is clear that the
transition metals within all-corner-sharing octahedra undergo some of the most significant
distortions (i.e., TM site 5 in Fig. 3.7b, 3.22). Alternatively, viewing the crystal structure
down the [100] direction provides a perspective of transition metal motion in the bc-plane
(Fig. 3.7c,d, 3.23). In the crystal structure of pristine Nb8W9O47 (Fig. 3.7c) the transition
metals reside above and below the ab-plane, or the (00l) plane. As lithiation occurs, the transition metals move in the c-direction, shifting into the plane and aligning with one another, as
highlighted in Figure 3.7d and 3.23. The most significant movement of transition metals into
the ab-plane is observed for the partially lithiated phase, Li9.6Nb8W9O47 (0.55 Li+ per TM),
which corresponds to the point at which the b parameter contracts (Fig. 3.6c, 3.23d). Overall, similar displacements in the c-direction are observed for all of the transition metal sites.
In other words, the extent of transition metal movement into the ab-plane during lithiation
does not seem to be impacted by the octahedral connectivity (corner- versus edge-sharing)
like it is within this plane.
Notably, a strain parameter was also included in the sequential Pawley fit, Rietveld refinement with locked transition metal positions, and Rietveld refinement with refined transition
metal positions. Similar trends in the strain parameter are observed for all three fitting procedures (Fig. 3.17). Overall, the strain parameter remains relatively steady for all three fits
but exhibits two significant and unsurprising increases. The first increase in strain occurs
during the two-phase region, after which it levels out and decreases slightly. The second
strain increase occurs during the b-axis and unit cell volume contraction, which is again
followed by a small decrease and leveling out of the strain parameter’s value.
In their study on the relationship between structure and electrochemical properties in
TTB-type phases, Yao et al. hypothesized that the impressive rate performance could be
attributed to the presence of highly interconnected or multidimensional diffusion pathways
for fast Li+ ion conduction.165 Therefore, to further investigate the structural origins of
80

3.3. Results and Discussion
fast cycling in Nb8W9O47, bond-valence sum energy (BVSE) mapping was performed to
analyze the probable Li+ ion conduction pathways using the softBV software package and
the structural model from Rietveld refinement as the input structure.166–168 BVSE mapping
utilizes the bond valence site energy approach to estimate energy barriers for ionic diffusion
within a given host crystal lattice.169 For this analysis, the interaction energy between Li+
and un-lithiated Nb8W9O47 is calculated over a grid of locations in the crystal structure. This
yields an estimate of the relative hopping energy barriers for Li+ in Nb8W9O47, as well as a
representation of the energy landscape and the probable conduction pathways for Li+. Values
of ∆V between 0.05 and 0.2 valence units are typically considered the most realistic for BVSE
mapping of light, mobile ions (like Li+) at room temperature.169 Based on the extensive and
highly interconnected Li+ conduction pathways visible at a ∆V = 0.1 valence unit cutoff
depicted in Figure 3.8, the BVSE mapping suggests low-energy barriers for Li+ ion migration
in both the ab-plane and along the c-direction in Nb8W9O47.
Figure 3.8: Bond-valence sum energy map of Li+ as the mobile ion in Nb8W9O47, viewed
(a) down the [001] direction and (b) down the [100] direction with the probable Li+ ion
conduction pathways (∆V =0.1 valence units) shown in yellow. Crystal structures are
depicted using VESTA.117
81

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
3.4 Conclusions
In summary, we report on the rapid, microwave-assisted synthesis and electrochemical properties of TTB-type Nb8W9O47. Micron-scale particles of Nb8W9O47 galvanostatically cycled
at rates of C/2 or slower achieve multi-electron redox, reversibly (de)intercalating more
than 1.0 Li+ per TM, and maintain reversible cycling of 0.65 Li+ per TM at 20C. In-house
operando XRD was employed to investigate the structural origin of the material’s fast Li-ion
cycling capabilities. Rietveld refinements of the operando XRD data indicate the structure
of the lithiated phase can still be described by the original space group symmetry. While
cells cycled below 1.2 V achieve higher capacities, operando XRD analysis shows that cycling
down to 1.0 V results in a 5.3% increase in the unit cell volume, with 2% of the expansion
occurring below 1.2 V. Our structural analysis also reveals anisotropic unit cell parameter
changes during lithiation and indicates the charge compensation in Nb8W9O47 is based on
localized transition metal displacement.
One of the critical electrode materials design principles for high-rate, long-term cycling
is reducing long-range structural deformation. As evidenced by Nb8W9O47, and likely other
TTB-type materials, this can be achieved through displacive mechanisms rather than problematic, coherent distortions between polyhedra. However, the presence of crystallographic
motifs that imbue the crystal structure with rigidity (i.e., shear planes, pentagonal columns)
and prevent polyhedral bending/twisting is crucial, as they force the transition metal to rely
on bond lengthening and contraction when it undergoes redox. In other words, the chemical
identity of the redox active site is critical because certain transition metal properties, like
Jahn-Teller activity, make these kinds of mechanisms feasible.
Why these materials exhibit such impressive cycling abilities in micron-scale particles
is still not clear. For example, ball-milled electrode mixes of Nb8W9O47 with sub-micron
particle sizes exhibited larger voltage polarizations and faster capacity fade, suggesting that
decreasing particle size post-synthesis may actually disrupt the material’s cycling ability.
82

3.5. Supplemental Information
Coupling the structural analysis presented in this work with synchrotron XRD capabilities
to probe structural evolution during fast cycling and in electrodes prepared with different
processing methods (like ball milling) will be meaningful studies to further develop our understanding of the electrochemical cycling capabilities of bronze and bronze-derived electrode
materials.
3.5 Supplemental Information
3.5.1 Powder X-ray Diffraction Rietveld Refinement Results
High-resolution, synchrotron powder X-ray diffraction of pristine Nb8W9O47 was collected
through the mail-in program at beamline 2-1 at SSRL. The published structure was refined against the resulting diffraction pattern using the Rietveld method, as implemented
in TOPAS-Academic v6,145 to evaluate the symmetry and phase purity of the material.
Relevant results are listed in Tables 3.1 and 3.2.
Values reported here without error (i.e., O atomic coordinates, thermal parameters) were
locked to previously reported values and not refined. Notably, Nb8W9O47 is very strongly
X-ray absorbing (due to the presence of Nb and W), causing issues with the transmission
experiments typical to the BL 2-1 mail-in program. Therefore, to obtain higher quality data,
pristine Nb8W9O47 was diluted with diamond powder to reduce absorption. The diamond
powder was refined as a secondary phase but is not reported in the pristine Nb8W9O47
refinement tables (Tables 3.1, 3.2).
83

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
Table 3.1: Rietveld refinement results for pristine Nb8W9O47 synchrotron powder XRD data.
This table includes the refined atomic coordinates, occupancies, and thermal parameters of
the transition metal sites only. See Table 3.2 for refined atomic coordinates, occupancies,
and thermal parameters of the oxygen sites. Diamond powder was added to the sample as
a diluent to reduce X-ray absorption.
Space Group a (˚A) b (˚A) c (˚A) α = β = γ Rwp
P21212 12.24351(7) 36.5963(2) 3.94460(2) 90◦ 3.226
Atom Wyck. Pos. X Y Z Occ. B
Nb1 2b 0 1/2 0.440(1) 0.20(1) 1.5(1)
W1 2b 0.80(1)
Nb2 4c 0.5750(2) 0.57032(8) 0.556(1) 0.47(1) 2.2(1)
W2 4c 0.53(1)
Nb3 4c 0.0701(2) 0.59752(8) 0.5731(9) 0.34(1) 0.59(7)
W3 4c 0.66(1)
Nb4 4c 0.2089(3) 0.69357(8) 0.528(1) 0.45(1) 1.7(1)
W4 4c 0.55(1)
Nb5 4c 0.7943(3) 0.63914(8) 0.432(1) 0.516(8) 0.7(1)
W5 4c 0.484(8)
Nb6 4c 0.5143(2) 0.67326(6) 0.5820(6) 0.07(1) 0.70(5)
W6 4c 0.93(1)
Nb7 4c 0.4206(3) 0.76812(8) 0.448(1) 0.47(1) 0.9(1)
W7 4c 0.53(1)
Nb8 4c 0.2929(2) 0.52153(7) 0.564(1) 0.45(1) 0.67(9)
W8 4c 0.55(1)
Nb9 4c 0.3318(3) 0.6099(1) 0.445(1) 0.976(6) 4(2)
W9 4c 0.024(6)
84

3.5. Supplemental Information
Table 3.2: Additional Rietveld refinement results for pristine Nb8W9O47 synchrotron XRD
data. This table includes the refined atomic coordinates, occupancies, and thermal parameters of the oxygen sites only. See Table 3.1 for refined atomic coordinates, occupancies, and
thermal parameters of the transition metal sites. Diamond powder was added to the sample
as a diluent to reduce X-ray absorption.
Space Group a (˚A) b (˚A) c (˚A) α = β = γ Rwp
P21212 12.24351(7) 36.5963(2) 3.94460(2) 90◦ 3.226
Atom Wyck. Pos. X Y Z Occ. B
O1 4c 0.5123 0.6718 1.023 1.0 1.405
O2 4c 0.0758 0.5976 1.017 1.0 1.137
O3 2b 0 1/2 0.995 1.0 1.342
O4 4c 0.2962 0.5220 1.010 1.0 1.042
O5 4c 0.7941 0.6374 0.989 1.0 1.974
O6 4c 0.4223 0.7660 0.997 1.0 1.366
O7 4c 0.2127 0.6903 1.006 1.0 1.240
O8 4c 0.5732 0.5699 1.008 1.0 0.971
O9 4c 0.3315 0.61009 0.9935 1.0 0.671
O10 4c 0.7116 0.5938 0.5024 1.0 1.050
O11 4c 0.2904 0.7363 0.4999 1.0 0.924
O12 4c 0.2154 0.5708 0.5046 1.0 0.513
O13 4c 0.5060 0.7222 0.516 1.0 1.113
O14 4c 0.4972 0.6189 0.5049 1.0 0.568
O15 4c 0.3572 0.6657 0.5007 1.0 0.679
O16 4c 0.3537 0.8117 0.505 1.0 1.271
O17 4c 0.0072 0.5512 0.508 1.0 1.413
O18 4c 0.9419 0.6227 0.504 1.0 1.413
O19 4c 0.1792 0.63733 0.5026 1.0 0.568
O20 4c 0.4119 0.55894 0.4960 1.0 0.553
O21 4c 0.3838 0.4798 0.5036 1.0 0.892
O22 4c 0.0681 0.7087 0.5055 1.0 0.892
O23 4c 0.1584 0.4991 0.5031 1.0 0.742
O24 4c 0.6609 0.6660 0.504 1.0 1.034
85

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
3.5.2 Additional Electrochemical Cycling Data
1.0 1.5 2.0 2.5 3.0
V vs. Li/Li+
-120
-80
-40
0
40
80
120
cycle 1
2
5
10
15
20
Sweep Rate: 0.1 mV/s
(b)
C
urre
nt D
e
n
sity (m
A
/
g)
1.2 1.5 1.8 2.1 2.4 2.7 3.0
V vs. Li/Li+
-150
-100
-50
0
50
100
150
cycle 1
2
5
10
15
20
Sweep Rate: 0.1 mV/s
(c)
C
urre
nt D
e
n
sity (m
A
/
g)
1.0 1.5 2.0 2.5 3.0
V vs. Li/Li+
-200
-150
-100
-50
0
50
100
150
200
1.4 V
1.3 V
1.2 V
1.1 V
1.0 V
Sweep Rate: 0.1 mV/s
(a)
C
urre
nt D
e
n
sity (m
A
/
g)
Figure 3.9: (a) Window opening cyclic voltammetry experiment. (b) Cyclic voltammetry
performed in a 1.0 to 3.0 V window showing poor reversibility when cycled to this lower
voltage limit. (c) Cyclic voltammetry performed in a 1.2 to 3.0 V window showing superior
cycling stability and reversibility.
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Li+ per TM in Nb8
W9
O47
1.0
1.5
2.0
2.5
3.0
V vs. Li/Li
+
1.0 V cutoff
1.2 V cutoff
1.0 1.5 2.0 2.5 3.0
V vs. Li/Li
+
-6.0
-3.0
0.0
3.0
6.0
dQ/dV (mAh/gV)
Figure 3.10: A comparison of galvanostatic cycling performed at a rate of C/10 in a
1.0 to 3.0 V window and a 1.2 to 3.0 V window to show the difference in voltage polarization and Coulombic efficiency in the two windows. Associated differential capacity plots
are shown in the bottom panel. Cycling is taken from operando cells cycled with free-standing
pellet electrodes.
86

3.5. Supplemental Information
3.5.3 Electron Probe Microanalysis
Table 3.3: EPMA elemental quantification results for the Nb and W formula atoms of a
sample of microwave-prepared Nb8W9O47 powder.
Nb W Nb W
Spot Element Element Element Atomic Atomic Atomic Nb:W
# Percent Percent Total Percent Percent Total Ratio
1 20.57 49.14 91.39 12.01 14.50 100.00 0.828
2 22.58 50.24 95.66 12.50 14.06 100.00 0.889
3 21.80 48.92 92.87 12.44 14.11 100.00 0.882
4 21.53 49.63 93.39 12.25 14.28 100.00 0.858
5 19.62 49.55 90.56 11.62 14.83 100.00 0.784
6 19.55 48.73 89.42 11.71 14.75 100.00 0.794
7 22.22 48.73 93.24 12.60 13.97 100.00 0.902
8 19.45 49.54 90.30 11.56 14.88 100.00 0.777
9 20.55 49.01 91.19 12.02 14.48 100.00 0.830
10 20.29 49.05 90.88 11.92 14.57 100.00 0.819
average 0.836
stdev 0.045
87

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
3.5.4 X-ray Photoelectron Spectroscopy
Figure 3.11: Representative survey X-ray photoelectron spectra of pristine Nb8W9O47 powder (a,b) and NbO2 (c), Nb2O5 (d), and WO3 (e) standards.
88

3.5. Supplemental Information
Figure 3.12: Representative Nb 3d X-ray photoelectron spectra of pristine Nb8W9O47 powder
(a,b) and NbO2 (c), Nb2O5 (d), and WO3 (e) standards.
Figure 3.13: Representative W 4f X-ray photoelectron spectra of pristine Nb8W9O47 powder
(a,b) and WO3 (c), NbO2 (d), and Nb2O5 (e) standards.
89

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
3.5.5 Operando X-ray Diffraction Rietveld Refinement Results
and Additional Data
1.5 1.8 2.1 2.4 2.7
Q (Å-1)
Intensity (arb. units)
1.2 1.8 2.4 3.0
V vs. Li/Li+
0.0
0.4
0.8
0.0
0.4
0.8
0.4
0.8
0.0
0.4
0.8
Li
+ per TM in Nb
W
8
O
9
4
7
0.4
0.8
0.0
0.4
0.8 Figure 3.14: In-house operando XRD results from Nb8W9O47 cycled vs. Li/Li+ at a rate of
C/10 in a 1.2 to 3.0 V window for three cycles. Cycling begins at the bottom of the figure
(cycle number increases going up the y-axis).
90

3.5. Supplemental Information
1.5 1.8 2.1 2.4 2.7
Q (Å-1)
Intensity (arb. units)
1.0 2.0 3.0
V vs. Li/Li+
0.0
0.5
1.0
0.5
1.0
Li
+ per TM in Nb
W
8
O
9
4
7
0.5
1.0
0.0
0.5
1.0
0.0
Figure 3.15: In-house operando XRD results from Nb8W9O47 cycled vs. Li/Li+ at a rate of
C/10 in a 1.0 to 3.0 V window for two cycles. Cycling begins at the bottom of the figure
(cycle number increases going up the y-axis).
91

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
(a)
C/10
1.2 V cutoff
1.5 1.6 1.7 1.8 1.9
Q (Å-1)
Intensity (arb. units)
1.2 1.8 2.4 3.0
V vs. Li/Li+
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
Li per TM in Nb
W
8
O
9 47
(b)
C/10
1.0 V cutoff
1.5 1.6 1.7 1.8 1.9
Q (Å-1)
Intensity (arb. units)
1.0 2.0 3.0
V vs. Li/Li+
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Li per TM in Nb
W
8
O
9 47
Figure 3.16: Comparison of in-house operando XRD of Nb8W9O47 vs. Li/Li+ at a rate of
C/10 in a (a) 1.2 to 3.0 V window and (b) 1.0 to 3.0 V window. Both plots show only the
first cycle and are zoomed into the most intense diffraction peaks. Cycling begins at the
bottom of the figure (cycle number increases going up the y-axis).
0.0 0.2 0.4 0.6 0.8 1.0 1.2
1.0
1.5
2.0
2.5
V vs. Li/Li
+
0.0
0.3
0.6
strain
0.0
3.0
6.0
∆vol (%)
-2.0
0.0
2.0
∆c (%)
0.0
1.0
2.0
∆b (%)
0.0
2.0
4.0
∆a (%)
(a)
Pawley
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Li+ per TM in Nb8
W9
O47
Rietveld (TM locked)
(b)
Rietveld (TM open)
0.0 0.2 0.4 0.6 0.8 1.0 1.2
(c)
Figure 3.17: Comparison of refined unit cell and strain parameter trends from the (a) Pawley
fit, (b) Rietveld refinement with TM positions locked, and (c) Rietveld refinement with TM
positions refined from in-house operando XRD of Nb8W9O47 vs. Li/Li+ in a 1.0 to 3.0 V
window.
92

3.5. Supplemental Information
Table 3.4: Sequential Pawley fit results for scans 0–16 of the operando XRD of Nb8W9O47
vs. Li/Li+ in a 1.0 to 3.0 V window. See Table 3.5 for the Pawley fit results for scans 17–37.
Note that scans 4–8 represent the two-phase region and therefore have two entries for each
parameter.
Scan # a (˚A) b (˚A) c (˚A) strain Rwp
0 12.2539(5) 36.6239(18) 3.94363(6) 0.245(5) 1.933
1 12.2648(4) 36.6390(14) 3.94367(6) 0.195(5) 1.872
2 12.2760(4) 36.6612(13) 3.94262(6) 0.158(6) 1.960
3 12.2882(4) 36.6941(12) 3.93978(6) 0.100(9) 2.072
4 12.2989(8) 36.7290(16) 3.93494(9) 0.10(2) 1.768
12.39(2) 37.00(8) 3.900(7) 0.8(3)
5 12.304(2) 36.752(4) 3.9342(2) 0.27(2) 1.742
12.39(1) 37.08(2) 3.892(4) 0.4(1)
6 12.313(8) 36.760(7) 3.9357(7) 0.40(3) 1.770
12.408(7) 37.07(2) 3.890(3) 0.43(8)
7 12.332(6) 36.746(15) 3.940(1) 0.37(4) 1.734
12.419(6) 37.112(15) 3.882(4) 0.36(5)
8 12.354(4) 36.706(8) 3.941(1) 0.10(7) 1.801
12.430(4) 37.134(1) 3.882(1) 0.36(3)
9 12.4465(10) 37.167(4) 3.87745(11) 0.304(9) 2.138
10 12.4584(9) 37.199(4) 3.87915(10) 0.280(8) 2.108
11 12.4715(8) 37.235(3) 3.88223(9) 0.258(7) 2.001
12 12.4806(8) 37.264(3) 3.88636(9) 0.248(7) 2.119
13 12.4909(8) 37.292(3) 3.89132(9) 0.237(7) 2.111
14 12.5032(8) 37.329(3) 3.89640(9) 0.223(7) 2.062
15 12.5135(8) 37.351(3) 3.90130(9) 0.219(7) 2.097
16 12.5197(8) 37.348(3) 3.90471(9) 0.218(8) 2.203
93

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
Table 3.5: Sequential Pawley fit results for scans 17–37 of the operando XRD of Nb8W9O47
vs. Li/Li+ in a 1.0 to 3.0 V window. See Table 3.4 for the Pawley fit results for scans 0–16.
Scan # a (˚A) b (˚A) c (˚A) strain Rwp
17 12.5283(7) 37.262(3) 3.9051(1) 0.24(1) 2.494
18 12.543(1) 37.029(7) 3.9046(6) 0.32(1) 2.788
19 12.5575(10) 36.880(6) 3.9036(3) 0.32(1) 2.650
20 12.5696(9) 36.817(5) 3.9042(2) 0.32(1) 2.449
21 12.579(1) 36.798(4) 3.9065(2) 0.32(1) 2.496
22 12.587(1) 36.789(4) 3.91011(15) 0.300(9) 2.568
23 12.592(1) 36.764(4) 3.9146(1) 0.286(9) 2.636
24 12.594(1) 36.722(4) 3.9198(1) 0.275(9) 2.839
25 12.594(1) 36.686(4) 3.9252(1) 0.269(9) 2.750
26 12.594(1) 36.672(4) 3.93095(11) 0.250(9) 2.847
27 12.5955(13) 36.681(3) 3.9369(1) 0.24(1) 2.763
28 12.597(1) 36.704(3) 3.9428(1) 0.225(10) 2.853
29 12.602(1) 36.729(4) 3.9486(1) 0.24(1) 2.877
30 12.606(2) 36.765(4) 3.9548(1) 0.26(1) 2.801
31 12.613(2) 36.787(4) 3.9610(2) 0.30(1) 2.663
32 12.608(3) 36.818(5) 3.9672(3) 0.31(1) 2.608
33 12.615(2) 36.844(4) 3.9720(3) 0.28(1) 2.535
34 12.623(2) 36.868(4) 3.9766(5) 0.26(1) 2.526
35 12.631(2) 36.884(4) 3.9808(6) 0.257(9) 2.500
36 12.642(2) 36.910(4) 3.9846(6) 0.248(8) 2.475
37 12.652(2) 36.938(4) 3.9884(7) 0.247(8) 2.560
94

3.5. Supplemental Information
Table 3.6: Sequential Rietveld refinement (TM positions locked) results for scans 0–16 of
the operando XRD of Nb8W9O47 vs. Li/Li+ in a 1.0 to 3.0 V window. See Table 3.7 for the
Rietveld refinement (TM positions locked) results for scans 17–37. Scans 4–8 represent the
two-phase region and therefore have two entries for each parameter.
Scan # a (˚A) b (˚A) c (˚A) strain Rwp
0 12.2520(6) 36.632(2) 3.94330(8) 0.278(7) 3.431
1 12.2643(5) 36.6466(15) 3.94341(8) 0.214(8) 3.490
2 12.2757(5) 36.667(1) 3.94234(8) 0.175(9) 3.555
3 12.2884(5) 36.697(1) 3.93950(8) 0.100(15) 3.838
4 12.3004(5) 36.731(2) 3.93487(9) 0.10(2) 3.390
12.371(7) 37.03(2) 3.8795(9) 0.33(6)
5 12.308(1) 36.754(3) 3.9340(1) 0.30(1) 3.401
12.397(2) 37.064(6) 3.8793(3) 0.26(2)
6 12.318(2) 36.778(7) 3.9348(3) 0.47(2) 3.549
12.4065(15) 37.074(4) 3.8794(2) 0.304(15)
7 12.315(10) 36.86(3) 3.9373(8) 0.61(5) 3.823
12.417(1) 37.097(4) 3.8788(2) 0.32(1)
8 12.32(5) 36.95(15) 3.943(1) 0.2(1) 3.964
12.430(1) 37.132(3) 3.87774(15) 0.30(1)
9 12.4466(9) 37.174(3) 3.8776(1) 0.301(9) 4.060
10 12.4607(9) 37.211(3) 3.8793(1) 0.271(9) 3.993
11 12.4728(8) 37.245(3) 3.8824(1) 0.260(9) 4.067
12 12.4831(8) 37.273(3) 3.8865(1) 0.244(8) 4.086
13 12.4929(8) 37.299(2) 3.8914(1) 0.228(8) 4.127
14 12.5040(7) 37.327(2) 3.8965(1) 0.213(8) 4.037
15 12.5138(8) 37.354(2) 3.9013(1) 0.209(8) 4.110
16 12.5222(7) 37.350(2) 3.9046(1) 0.197(8) 4.148
95

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
Table 3.7: Sequential Rietveld refinement (TM positions locked) results for scans 17–37 of
the operando XRD of Nb8W9O47 vs. Li/Li+ in a 1.0 to 3.0 V window. See Table 3.6 for the
Rietveld refinement (TM positions locked) results for scans 0–16.
Scan # a (˚A) b (˚A) c (˚A) strain Rwp
17 12.5300(8) 37.257(3) 3.9044(2) 0.225(10) 4.571
18 12.540(1) 37.050(4) 3.9034(2) 0.35(1) 4.804
19 12.550(1) 36.933(4) 3.9047(3) 0.36(1) 4.815
20 12.559(1) 36.890(4) 3.9061(2) 0.34(1) 5.064
21 12.566(1) 36.878(4) 3.9087(3) 0.34(1) 5.465
22 12.575(1) 36.868(4) 3.9123(3) 0.321(15) 5.912
23 12.582(1) 36.810(4) 3.9164(3) 0.30(2) 6.293
24 12.585(1) 36.737(4) 3.9206(3) 0.287(15) 6.570
25 12.585(1) 36.697(4) 3.9257(3) 0.293(15) 6.629
26 12.5849(15) 36.680(4) 3.9311(3) 0.29(2) 6.744
27 12.5857(15) 36.685(4) 3.9368(3) 0.295(16) 6.807
28 12.5863(15) 36.709(4) 3.9426(3) 0.29(2) 6.842
29 12.5893(15) 36.737(4) 3.9483(3) 0.30(2) 6.709
30 12.592(2) 36.771(4) 3.9542(3) 0.31(2) 6.738
31 12.596(2) 36.792(4) 3.9600(3) 0.32(2) 6.458
32 12.602(2) 36.815(4) 3.9660(3) 0.33(2) 6.533
33 12.613(2) 36.837(4) 3.9711(3) 0.310(15) 6.372
34 12.624(2) 36.858(4) 3.9758(3) 0.30(1) 6.337
35 12.636(2) 36.876(4) 3.9801(3) 0.30(1) 6.213
36 12.647(2) 36.898(4) 3.9840(3) 0.30(1) 6.134
37 12.657(2) 36.922(4) 3.9879(3) 0.30(1) 6.071
96

3.5. Supplemental Information
Table 3.8: Sequential Rietveld refinement (TM positions refined) results for scans 0–16 of
the operando XRD of Nb8W9O47 vs. Li/Li+ in a 1.0 to 3.0 V window. See Table 3.9 for
the Rietveld refinement (TM positions refined) results for scans 17–37. Scans 4–8 represent
the two-phase region and therefore have two entries for each parameter. Refined transition
metal site atomic coordinates can be found in Tables 3.10–3.15.
Scan # a (˚A) b (˚A) c (˚A) strain Rwp
0 12.2525(6) 36.632(2) 3.94331(8) 0.275(7) 3.142
1 12.2648(5) 36.646(1) 3.94341(8) 0.214(8) 3.191
2 12.2761(4) 36.667(1) 3.94232(8) 0.174(9) 3.214
3 12.2888(4) 36.697(1) 3.93946(8) 0.10(1) 3.458
4 12.2995(5) 36.735(2) 3.93478(8) 0.100(15) 2.897
12.36(1) 37.01(4) 3.876(1) 0.51(7)
5 12.3075(10) 36.762(3) 3.93375(14) 0.31(1) 2.761
12.396(2) 37.070(7) 3.8779(4) 0.27(2)
6 12.320(2) 36.796(7) 3.9346(3) 0.48(2) 2.810
12.4075(15) 37.081(5) 3.8785(2) 0.295(14)
7 12.328(7) 36.85(2) 3.9372(7) 0.56(4) 2.944
12.417(1) 37.102(4) 3.8784(2) 0.31(1)
8 12.32(2) 36.98(7) 3.942(1) 0.27(8) 3.020
12.430(1) 37.142(3) 3.8774(1) 0.28(1)
9 12.4463(8) 37.179(2) 3.87745(10) 0.297(7) 3.044
10 12.4600(7) 37.216(2) 3.8791(1) 0.271(7) 2.992
11 12.4723(7) 37.249(2) 3.88221(9) 0.260(7) 2.960
12 12.4823(7) 37.279(2) 3.88639(9) 0.246(6) 2.996
13 12.4920(7) 37.305(2) 3.89132(9) 0.231(6) 3.032
14 12.5033(6) 37.332(2) 3.89641(9) 0.217(6) 2.940
15 12.5130(6) 37.359(2) 3.90124(9) 0.212(6) 2.980
16 12.5211(6) 37.355(2) 3.90456(9) 0.201(6) 2.947
97

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
Table 3.9: Sequential Rietveld refinement (TM positions refined) results for scans 17–37 of
the operando XRD of Nb8W9O47 vs. Li/Li+ in a 1.0 to 3.0 V window. See Table 3.8 for the
Rietveld refinement (TM positions refined) results for scans 0–16. Refined transition metal
site atomic coordinates can be found in Tables 3.10–3.15.
Scan # a (˚A) b (˚A) c (˚A) strain Rwp
17 12.5292(7) 37.269(3) 3.9046(1) 0.230(8) 3.340
18 12.5415(9) 37.056(4) 3.9033(2) 0.33(1) 3.755
19 12.5531(9) 36.925(4) 3.9044(2) 0.338(9) 3.714
20 12.5623(9) 36.870(3) 3.9055(2) 0.320(9) 3.713
21 12.5711(9) 36.831(3) 3.90771(15) 0.31(1) 3.850
22 12.5803(9) 36.806(3) 3.91101(15) 0.29(1) 3.940
23 12.5856(9) 36.778(3) 3.9151(1) 0.287(9) 3.944
24 12.587(1) 36.735(3) 3.9199(1) 0.287(9) 4.069
25 12.5857(9) 36.701(3) 3.9251(1) 0.283(9) 3.937
26 12.5855(9) 36.685(3) 3.9307(1) 0.267(9) 3.986
27 12.587(1) 36.691(3) 3.9365(1) 0.258(9) 3.989
28 12.587(1) 36.716(3) 3.94225(14) 0.246(1) 4.008
29 12.5902(9) 36.744(3) 3.94787(15) 0.25(1) 3.908
30 12.5933(9) 36.780(3) 3.95358(15) 0.26(1) 3.853
31 12.5966(9) 36.805(3) 3.9594(2) 0.276(9) 3.601
32 12.6018(9) 36.825(3) 3.9652(2) 0.281(8) 3.465
33 12.6118(9) 36.848(2) 3.9705(2) 0.263(7) 3.340
34 12.621(1) 36.868(2) 3.9755(2) 0.259(7) 3.317
35 12.633(1) 36.884(2) 3.9798(2) 0.261(7) 3.248
36 12.643(1) 36.908(2) 3.9839(2) 0.258(7) 3.217
37 12.653(1) 36.935(3) 3.9878(2) 0.264(7) 3.241
98

3.5. Supplemental Information
Table 3.10: Refined atomic coordinates for TM sites 1–3 and scans 0–16 from the sequential Rietveld refinement (TM positions
open) of the operando XRD of Nb W8 9O47 vs. Li/Li+ in a 1.0 to 3.0 V window. See Table 3.11 for results for TM sites 1–3 and
scans 17–37.
Scan # X1 Y 1 Z1 X2 Y 2 Z2 X3 Y 3 Z3
0 0 0.5 0.46(1) 0.572(4) 0.568(1) 0.576(8) 0.074(2) 0.5975(11) 0.526(6)
1 0 0.5 0.45(1) 0.572(4) 0.568(1) 0.583(8) 0.073(2) 0.597(1) 0.527(6)
2 0 0.5 0.45(1) 0.572(4) 0.568(1) 0.573(8) 0.075(2) 0.597(1) 0.526(6)
3 0 0.5 0.47(1) 0.573(4) 0.568(1) 0.570(9) 0.074(3) 0.598(1) 0.517(7)
4 0 0.5 0.45(2) 0.571(4) 0.5682(15) 0.53(2) 0.074(3) 0.597(1) 0.50(2)
0 0.5 0.45(12) 0.57(3) 0.57(1) 0.64(2) 0.08(2) 0.601(8) 0.50(8)
5 0 0.5 0.45(3) 0.572(6) 0.568(2) 0.55(2) 0.074(4) 0.595(2) 0.50(3)
0 0.5 0.52(3) 0.57(1) 0.563(4) 0.54(7) 0.074(6) 0.600(3) 0.50(5)
6 0 0.5 0.40(5) 0.57(1) 0.577(4) 0.63(3) 0.08(1) 0.599(2) 0.50(3)
0 0.5 0.41(3) 0.568(7) 0.566(2) 0.57(3) 0.073(4) 0.598(2) 0.50(2)
7 0 0.5 0.38(9) 0.56(2) 0.578(2) 0.64(2) 0.08(2) 0.604(5) 0.50(6)
0 0.5 0.40(2) 0.569(6) 0.565(2) 0.58(2) 0.072(4) 0.597(1) 0.50(2)
8 0 0.5 0.39(15) 0.55(4) 0.57(2) 0.6(1) 0.095(7) 0.606(2) 0.50(8)
0 0.5 0.40(2) 0.568(5) 0.566(2) 0.56(2) 0.071(3) 0.597(1) 0.50(2)
9 0 0.5 0.40(2) 0.567(4) 0.5645(13) 0.51(2) 0.072(3) 0.595(1) 0.51(2)
10 0 0.5 0.407(15) 0.567(3) 0.566(1) 0.51(2) 0.072(3) 0.5950(9) 0.52(1)
11 0 0.5 0.41(1) 0.567(3) 0.566(1) 0.50(2) 0.070(3) 0.5948(9) 0.53(2)
12 0 0.5 0.419(15) 0.568(3) 0.566(1) 0.50(2) 0.070(3) 0.5944(9) 0.52(1)
13 0 0.5 0.422(15) 0.569(3) 0.566(1) 0.49(2) 0.070(3) 0.5944(9) 0.535(17)
14 0 0.5 0.427(15) 0.570(3) 0.566(1) 0.49(2) 0.071(3) 0.5950(9) 0.54(2)
15 0 0.5 0.41(1) 0.569(3) 0.566(1) 0.50(2) 0.071(3) 0.5943(9) 0.52(2)
16 0 0.5 0.45(2) 0.572(3) 0.566(1) 0.49(3) 0.071(2) 0.5942(9) 0.51(3)
99

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47 Table 3.11: Refined atomic coordinates for TM sites 1–3 and scans 17–37 from the sequential Rietveld refinement (TM positions open) of the operando XRD of Nb
W8 9O47 vs. Li/Li
+ in a 1.0 to 3.0 V window. See Table 3.10 for results for TM sites 1–3 and
scans 0–16.
Scan #
X
1
Y
1
Z
1
X
2
Y
2
Z
2
X
3
Y
3
Z
3
17 0 0.5 0.52(4) 0.574(3) 0.565(1) 0.50(4) 0.069(2) 0.5902(9) 0.50(4)
18 0 0.5 0.37(2) 0.578(3) 0.562(1) 0.54(2) 0.069(3) 0.589(1) 0.51(2)
19 0 0.5 0.385(16) 0.579(3) 0.567(1) 0.56(2) 0.066(3) 0.591(1) 0.50(2)
20 0 0.5 0.44(3) 0.581(3) 0.571(1) 0.54(3) 0.068(3) 0.591(1) 0.50(5)
21 0 0.5 0.51(8) 0.580(3) 0.571(1) 0.50(7) 0.064(2) 0.5916(9) 0.50(5)
22 0 0.5 0.51(9) 0.582(3) 0.572(1) 0.50(8) 0.063(2) 0.5902(9) 0.50(6)
23 0 0.5 0.50(8) 0.582(3) 0.572(1) 0.50(8) 0.060(2) 0.5894(9) 0.50(7)
24 0 0.5 0.50(15) 0.580(3) 0.573(1) 0.5(1) 0.059(2) 0.5894(9) 0.5(1)
25 0 0.5 0.5(2) 0.580(3) 0.573(1) 0.5(2) 0.059(2) 0.5894(9) 0.5(1)
26 0 0.5 0.5(2) 0.578(4) 0.574(1) 0.5(2) 0.060(3) 0.589(1) 0.5(1)
27 0 0.5 0.5(3) 0.579(4) 0.575(1) 0.5(2) 0.060(3) 0.589(1) 0.5(2)
28 0 0.5 0.5(3) 0.580(4) 0.577(1) 0.5(2) 0.059(3) 0.589(1) 0.5(2)
29 0 0.5 0.5(2) 0.576(4) 0.577(1) 0.5(2) 0.063(3) 0.589(1) 0.5(2)
30 0 0.5 0.49(15) 0.576(4) 0.577(1) 0.5(2) 0.062(3) 0.589(1) 0.50(15)
31 0 0.5 0.48(8) 0.575(4) 0.577(1) 0.50(9) 0.064(3) 0.589(1) 0.50(8)
32 0 0.5 0.47(5) 0.572(4) 0.578(1) 0.50(8) 0.063(3) 0.589(1) 0.50(6)
33 0 0.5 0.46(4) 0.574(3) 0.5765(13) 0.50(7) 0.066(3) 0.589(1) 0.50(5)
34 0 0.5 0.44(2) 0.575(3) 0.577(1) 0.51(4) 0.064(3) 0.589(1) 0.51(4)
35 0 0.5 0.45(3) 0.572(3) 0.577(1) 0.50(5) 0.066(3) 0.589(1) 0.51(4)
36 0 0.5 0.44(2) 0.572(3) 0.577(1) 0.51(4) 0.065(3) 0.5894(9) 0.51(4)
37 0 0.5 0.44(2) 0.570(4) 0.578(1) 0.51(4) 0.064(3) 0.589(1) 0.51(4)
100

3.5. Supplemental Information
Table 3.12: Refined atomic coordinates for TM sites 4–6 and scans 0–16 from the sequential Rietveld refinement (TM positions
open) of the operando XRD of Nb W8 9O47 vs. Li/Li+ in a 1.0 to 3.0 V window. See Table 3.13 for results for TM sites 4–6 and
scans 17–37.
Scan # X4 Y 4 Z4 X5 Y 5 Z5 X6 Y 6 Z6
0 0.506(3) 0.6707(9) 0.603(5) 0.797(4) 0.641(1) 0.403(9) 0.204(4) 0.692(1) 0.512(7)
1 0.504(3) 0.670(1) 0.601(5) 0.798(4) 0.641(1) 0.399(9) 0.204(4) 0.692(1) 0.511(7)
2 0.503(3) 0.670(1) 0.605(5) 0.798(4) 0.642(1) 0.401(9) 0.205(3) 0.692(1) 0.509(7)
3 0.502(4) 0.669(1) 0.611(5) 0.797(4) 0.642(1) 0.395(10) 0.204(4) 0.692(1) 0.502(8)
4 0.511(3) 0.673(1) 0.60(1) 0.801(4) 0.643(1) 0.43(2) 0.205(4) 0.6915(12) 0.46(2)
0.49(3) 0.665(9) 0.51(8) 0.79(3) 0.641(9) 0.36(6) 0.20(3) 0.693(10) 0.45(8)
5 0.508(5) 0.674(1) 0.61(1) 0.811(6) 0.644(2) 0.45(3) 0.202(6) 0.692(2) 0.45(3)
0.512(7) 0.671(3) 0.51(4) 0.79(1) 0.641(4) 0.39(3) 0.216(9) 0.691(3) 0.45(4)
6 0.4(1) 0.665(4) 0.59(3) 0.815(17) 0.645(3) 0.46(4) 0.20(1) 0.692(4) 0.47(3)
0.518(5) 0.673(2) 0.51(2) 0.797(6) 0.642(3) 0.42(3) 0.213(5) 0.691(2) 0.45(3)
7 0.49(2) 0.665(6) 0.62(5) 0.818(7) 0.647(5) 0.47(7) 0.21(2) 0.693(6) 0.54(4)
0.514(4) 0.6722(15) 0.51(2) 0.797(5) 0.642(2) 0.44(2) 0.213(5) 0.6925(16) 0.45(2)
8 0.49(3) 0.665(11) 0.58(8) 0.82(4) 0.643(9) 0.4(1) 0.23(5) 0.69(1) 0.606(8)
0.514(3) 0.6725(13) 0.51(2) 0.797(5) 0.642(2) 0.46(2) 0.215(4) 0.692(1) 0.48(2)
9 0.510(2) 0.673(1) 0.52(1) 0.800(4) 0.642(1) 0.509(3) 0.216(3) 0.693(1) 0.59(1)
10 0.512(2) 0.6737(9) 0.52(1) 0.803(4) 0.642(1) 0.51(2) 0.215(3) 0.693(1) 0.58(1)
11 0.512(2) 0.6741(9) 0.51(1) 0.803(4) 0.642(1) 0.51(2) 0.215(3) 0.693(1) 0.59(1)
12 0.513(2) 0.6739(9) 0.510(15) 0.805(3) 0.642(1) 0.51(2) 0.216(3) 0.693(1) 0.58(1)
13 0.513(2) 0.6744(9) 0.509(15) 0.805(4) 0.642(1) 0.51(2) 0.216(3) 0.692(1) 0.58(1)
14 0.513(2) 0.6745(8) 0.509(15) 0.804(3) 0.643(1) 0.51(2) 0.216(3) 0.6926(9) 0.58(1)
15 0.512(2) 0.6743(9) 0.51(1) 0.806(4) 0.643(1) 0.51(2) 0.214(3) 0.692(1) 0.575(13)
16 0.514(2) 0.6748(8) 0.51(2) 0.807(4) 0.643(1) 0.51(3) 0.210(3) 0.692(1) 0.558(15)
101

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47 Table 3.13: Refined atomic coordinates for TM sites 4–6 and scans 17–37 from the sequential Rietveld refinement (TM positions open) of the operando XRD of Nb
W8 9O47 vs. Li/Li
+ in a 1.0 to 3.0 V window. See Table 3.12 for results for TM sites 4–6 and
scans 0–16.
Scan #
X
4
Y
4
Z
4
X
5
Y
5
Z
5
X
6
Y
6
Z
6
17 0.516(2) 0.678(1) 0.541(15) 0.812(4) 0.6473(9) 0.50(5) 0.206(3) 0.6928(9) 0.47(3)
18 0.510(3) 0.676(1) 0.509(15) 0.808(4) 0.647(3) 0.509(3) 0.211(4) 0.695(1) 0.56(2)
19 0.509(3) 0.676(1) 0.51(2) 0.812(4) 0.647(1) 0.51(2) 0.206(4) 0.6970(8) 0.54(2)
20 0.509(4) 0.678(1) 0.51(3) 0.815(4) 0.647(1) 0.51(4) 0.203(4) 0.6951(9) 0.51(4)
21 0.513(3) 0.681(3) 0.53(2) 0.818(9) 0.6474(9) 0.50(7) 0.211(3) 0.6946(8) 0.48(5)
22 0.514(3) 0.6815(8) 0.52(3) 0.818(3) 0.647(1) 0.5(1) 0.214(3) 0.6946(8) 0.49(7)
23 0.515(3) 0.6815(8) 0.52(3) 0.818(3) 0.647(1) 0.5(1) 0.215(3) 0.6947(8) 0.49(8)
24 0.514(3) 0.6815(8) 0.52(4) 0.818(3) 0.647(1) 0.50(15) 0.216(3) 0.6949(8) 0.5(1)
25 0.513(3) 0.6811(8) 0.51(6) 0.818(3) 0.647(1) 0.5(2) 0.214(3) 0.6934(9) 0.5(2)
26 0.512(3) 0.6801(9) 0.51(5) 0.818(3) 0.647(1) 0.5(2) 0.212(3) 0.6934(9) 0.5(2)
27 0.513(3) 0.679(1) 0.51(7) 0.818(3) 0.647(1) 0.5(2) 0.209(3) 0.6941(9) 0.5(3)
28 0.513(3) 0.678(1) 0.51(7) 0.818(3) 0.647(1) 0.5(2) 0.209(3) 0.693(1) 0.5(3)
29 0.515(3) 0.677(1) 0.51(7) 0.816(3) 0.647(1) 0.5(2) 0.205(3) 0.693(1) 0.5(2)
30 0.513(3) 0.678(1) 0.52(6) 0.818(3) 0.647(1) 0.5(1) 0.205(3) 0.692(1) 0.5(1)
31 0.511(3) 0.6775(10) 0.53(4) 0.815(3) 0.647(1) 0.51(8) 0.202(4) 0.691(1) 0.51(6)
32 0.509(3) 0.678(1) 0.53(3) 0.815(3) 0.647(1) 0.51(6) 0.202(4) 0.689(1) 0.51(5)
33 0.510(3) 0.6785(9) 0.52(3) 0.812(4) 0.647(1) 0.51(6) 0.200(4) 0.688(1) 0.52(5)
34 0.511(3) 0.6785(9) 0.51(2) 0.808(4) 0.647(1) 0.51(4) 0.198(4) 0.688(1) 0.52(3)
35 0.508(3) 0.6780(9) 0.51(3) 0.808(4) 0.6474(15) 0.51(4) 0.199(4) 0.6866(15) 0.52(4)
36 0.508(4) 0.6785(9) 0.51(2) 0.808(4) 0.647(1) 0.51(4) 0.199(4) 0.688(1) 0.52(3)
37 0.507(4) 0.6775(10) 0.51(2) 0.806(4) 0.647(1) 0.51(4) 0.197(4) 0.688(1) 0.51(4)
102

3.5. Supplemental Information
Table 3.14: Refined atomic coordinates for TM sites 7–9 and scans 0–16 from the sequential Rietveld refinement (TM positions
open) of the operando XRD of Nb W8 9O47 vs. Li/Li+ in a 1.0 to 3.0 V window. See Table 3.15 for results for TM sites 7–9 and
scans 17–37.
Scan # X7 Y 7 Z7 X8 Y 8 Z8 X9 Y 9 Z9
0 0.425(4) 0.769(1) 0.49(1) 0.524(1) 0.607(8) 0.339(4) 0.339(4) 0.613(1) 0.46(1)
1 0.424(4) 0.768(1) 0.482(9) 0.5235(13) 0.602(8) 0.339(4) 0.339(4) 0.612(1) 0.46(1)
2 0.426(4) 0.768(1) 0.495(10) 0.524(1) 0.604(8) 0.338(3) 0.338(3) 0.612(1) 0.465(12)
3 0.424(4) 0.767(1) 0.50(1) 0.5241(15) 0.589(9) 0.338(3) 0.338(3) 0.612(1) 0.46(1)
4 0.426(4) 0.7671(15) 0.52(2) 0.5239(15) 0.54(2) 0.338(4) 0.338(4) 0.6125(13) 0.44(2)
0.42(3) 0.769(8) 0.52(2) 0.525(8) 0.65(2) 0.34(3) 0.34(3) 0.613(1) 0.52(2)
5 0.427(6) 0.766(2) 0.52(3) 0.525(2) 0.53(3) 0.338(7) 0.338(7) 0.612(2) 0.44(2)
0.43(1) 0.773(4) 0.52(8) 0.525(4) 0.61(4) 0.34(1) 0.34(1) 0.611(5) 0.48(5)
6 0.425(13) 0.762(4) 0.52(4) 0.525(4) 0.54(4) 0.34(1) 0.34(1) 0.614(4) 0.41(3)
0.427(7) 0.770(2) 0.52(2) 0.524(2) 0.57(3) 0.336(7) 0.336(7) 0.612(3) 0.522(6)
7 0.41(2) 0.760(6) 0.52(6) 0.526(6) 0.51(6) 0.35(3) 0.35(3) 0.615(7) 0.37(5)
0.428(6) 0.770(2) 0.52(2) 0.525(2) 0.55(3) 0.337(6) 0.337(6) 0.613(2) 0.522(5)
8 0.40(5) 0.77(2) 0.52(8) 0.530(2) 0.6(1) 0.356(7) 0.356(7) 0.615(12) 0.4(1)
0.429(5) 0.770(2) 0.52(2) 0.524(2) 0.55(2) 0.336(5) 0.336(5) 0.613(2) 0.522(5)
9 0.426(4) 0.769(1) 0.52(2) 0.5245(13) 0.49(2) 0.337(4) 0.337(4) 0.613(2) 0.48(2)
10 0.424(3) 0.768(1) 0.52(2) 0.524(1) 0.49(2) 0.336(4) 0.336(4) 0.6131(15) 0.47(2)
11 0.426(3) 0.768(1) 0.51(2) 0.524(1) 0.49(2) 0.335(4) 0.335(4) 0.613(1) 0.461(15)
12 0.424(3) 0.768(1) 0.51(2) 0.524(1) 0.49(2) 0.335(4) 0.335(4) 0.613(1) 0.47(2)
13 0.424(3) 0.7675(12) 0.50(2) 0.524(1) 0.49(2) 0.335(4) 0.335(4) 0.613(1) 0.465(18)
14 0.424(3) 0.767(1) 0.50(2) 0.524(1) 0.49(2) 0.334(3) 0.334(3) 0.6125(13) 0.47(2)
15 0.423(3) 0.768(1) 0.52(2) 0.524(1) 0.49(2) 0.334(3) 0.334(3) 0.613(1) 0.49(2)
16 0.425(3) 0.767(1) 0.51(3) 0.524(1) 0.49(3) 0.333(3) 0.333(3) 0.613(1) 0.49(3)
103

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47 Table 3.15: Refined atomic coordinates for TM sites 7–9 and scans 17–37 from the sequential Rietveld refinement (TM positions open) of the operando XRD of Nb
W8 9O47 vs. Li/Li
+ in a 1.0 to 3.0 V window. See Table 3.14 for results for TM sites 1–3 and
scans0–16.
Scan #
X
7
Y
7
Z
7
X
8
Y
8
Z
8
X
9
Y
9
Z
9
17 0.431(3) 0.769(1) 0.51(3) 0.525(1) 0.49(4) 0.333(3) 0.333(3) 0.614(1) 0.522(5)
18 0.429(4) 0.771(1) 0.52(3) 0.522(1) 0.53(2) 0.346(3) 0.346(3) 0.6175(11) 0.44(2)
19 0.432(4) 0.7705(13) 0.52(3) 0.517(1) 0.54(2) 0.352(4) 0.352(4) 0.617(1) 0.43(2)
20 0.435(3) 0.7672(15) 0.52(3) 0.518(1) 0.51(5) 0.351(4) 0.351(4) 0.618(1) 0.49(5)
21 0.432(3) 0.761(1) 0.52(3) 0.5172(9) 0.49(6) 0.350(3) 0.350(3) 0.6145(10) 0.51(8)
22 0.431(3) 0.760(1) 0.52(4) 0.5165(9) 0.49(8) 0.350(3) 0.350(3) 0.614(1) 0.5(1)
23 0.428(3) 0.7598(9) 0.52(3) 0.5160(9) 0.49(8) 0.349(3) 0.349(3) 0.613(1) 0.5(1)
24 0.427(4) 0.760(1) 0.51(9) 0.5147(9) 0.5(1) 0.349(3) 0.349(3) 0.613(1) 0.5(1)
25 0.425(4) 0.760(1) 0.5(1) 0.5147(9) 0.5(1) 0.347(3) 0.347(3) 0.614(1) 0.51(15)
26 0.424(4) 0.760(1) 0.5(1) 0.5141(9) 0.5(1) 0.347(4) 0.347(4) 0.614(1) 0.5(1)
27 0.425(4) 0.760(1) 0.5(2) 0.5137(9) 0.5(1) 0.347(4) 0.347(4) 0.614(1) 0.5(2)
28 0.423(4) 0.760(1) 0.5(1) 0.514(1) 0.5(1) 0.347(4) 0.347(4) 0.615(1) 0.5(2)
29 0.424(4) 0.760(1) 0.50(15) 0.515(1) 0.5(1) 0.347(4) 0.347(4) 0.615(1) 0.5(1)
30 0.423(4) 0.760(1) 0.5(1) 0.516(1) 0.5(1) 0.347(4) 0.347(4) 0.614(1) 0.5(1)
31 0.423(4) 0.760(1) 0.51(7) 0.518(1) 0.49(8) 0.347(4) 0.347(4) 0.613(1) 0.52(4)
32 0.420(4) 0.760(1) 0.52(5) 0.519(1) 0.49(7) 0.348(4) 0.348(4) 0.612(1) 0.52(4)
33 0.423(4) 0.760(1) 0.51(6) 0.520(1) 0.49(6) 0.346(3) 0.346(3) 0.6110(10) 0.52(4)
34 0.423(4) 0.760(1) 0.50(4) 0.520(1) 0.49(4) 0.345(3) 0.345(3) 0.610(1) 0.522(5)
35 0.421(4) 0.760(1) 0.51(5) 0.521(1) 0.49(5) 0.345(3) 0.345(3) 0.611(1) 0.522(6)
36 0.421(4) 0.760(1) 0.51(4) 0.519(1) 0.49(4) 0.347(4) 0.347(4) 0.610(1) 0.522(5)
37 0.421(4) 0.760(1) 0.50(4) 0.520(1) 0.49(4) 0.350(4) 0.350(4) 0.611(1) 0.52(4)
104

3.5. Supplemental Information
(a)
Li1.1Nb8W9O47
1.2 1.5 1.8 2.1 2.4 2.7
Q (Å-1)
Intensity (arb. units)
Rietveld (TM open)
Rietveld (TM locked)
obs
Pawley
diff
(e)
Li12.4Nb8W9O47
1.2 1.5 1.8 2.1 2.4 2.7
Q (Å-1)
Intensity (arb. units)
Rietveld (TM open)
Rietveld (TM locked)
obs
Pawley
diff
Li4.2Nb8W9O47
(b)
1.2 1.5 1.8 2.1 2.4 2.7
Q (Å-1)
Rietveld (TM open)
Rietveld (TM locked)
obs
Pawley
diff
(f)
Li15.2Nb8W9O47
1.2 1.5 1.8 2.1 2.4 2.7
Q (Å-1)
Rietveld (TM open)
Rietveld (TM locked)
obs
Pawley
diff
Li6.8Nb8W9O47
(c)
1.2 1.5 1.8 2.1 2.4 2.7
Q (Å-1)
Rietveld (TM open)
Rietveld (TM locked)
obs
Pawley
diff
(g)
Li18.1Nb8W9O47
1.2 1.5 1.8 2.1 2.4 2.7
Q (Å-1)
Rietveld (TM open)
Rietveld (TM locked)
obs
Pawley
diff
(d)
Li9.6Nb8W9O47
1.2 1.5 1.8 2.1 2.4 2.7
Q (Å-1)
Rietveld (TM open)
Rietveld (TM locked)
obs
Pawley
diff
(h)
Li21.2Nb8W9O47
1.2 1.5 1.8 2.1 2.4 2.7
Q (Å-1)
Rietveld (TM open)
Rietveld (TM locked)
obs
Pawley
diff Figure 3.18: A comparison of the quality of fit from a Pawley fit (blue/top), a Rietveld
refinement with TM positions locked (green/middle), and a Rietveld refinement with TM
positions refined (yellow/bottom). Selected diffraction patterns are from the operando XRD
of Nb8W9O47 vs. Li/Li+ in a 1.0 to 3.0 V window.
105

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
Figure 3.19: Comparison of the trend in Rwp (quality of fit) over the course of sequential
fitting for the Pawley fit, Rietveld refinement with TM positions locked, and Rietveld refinement with TM positions refined from the operando XRD of Nb8W9O47 vs. Li/Li+ in a
1.0 to 3.0 V window.
106

3.5. Supplemental Information
TM 1 (a)
-0.05
0.00
0.05
x
0.45
0.50
0.55
y
0.40
0.50
z
(b) TM 2
0.55
0.60
x
0.55
0.60
y
0.50
0.60
z
(c) TM 3
0.05
0.10
x
0.55
0.60
y
0.50
0.55
z
TM 4 (d)
0.20
0.25
x
0.65
0.70
0.75
y
0.50
0.60
z
(e) TM 5
0.80
0.85
x
0.60
0.65
y
0.40
0.50
z
(f) TM 6
0.50
0.55
x
0.65
0.70
y
0.50
0.60
z
TM 7 (g) (h) (i) TM 8 TM 9
0.40
0.45
x
0.75
0.80
y
0.45
0.50
0.55
z
0.0 0.3 0.6 0.9 1.2
Li+ per TM in Nb8
W9
O47
1.0
2.0
3.0
V vs. Li/Li
+
0.30
0.35 x
0.60
0.65
y
0.40
0.50
z
0.0 0.3 0.6 0.9 1.2
Li+ per TM in Nb8
W9
O47
1.0
2.0
3.0
V vs. Li/Li
+
0.25
0.30 x
0.50
0.55
y
0.50
0.60
z
0.0 0.3 0.6 0.9 1.2
Li+ per TM in Nb8
W9
O47
1.0
2.0
3.0
V vs. Li/Li
+
Figure 3.20: A graphical visualization of refined transition metal positions from the operando
XRD of Nb8W9O47 vs. Li/Li+ in a 1.0 to 3.0 V window. The values for the refined transition
metal site atomic coordinates can be found in Tables 3.10–3.15.
107

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
Figure 3.21: Crystal structure of pristine Nb8W9O47 depicted (a) in polyhedral view down
the [001] direction and (b) with no polyhedra drawn, also viewed down the [001] direction
with a smaller portion of the unit cell highlighted in blue. (c) A view of the smaller portion
of the unit cell down the [001] direction with transition metal sites labeled numerically (as
labeled in the refinement tables). (d) A view of the smaller portion of the unit cell down
the [100] direction with transition metal sites numbered. Crystal structures depicted using
VESTA.117
108

3.5. Supplemental Information
Figure 3.22: Visualization of the transition metal motion occurring in Nb8W9O47 with increasing lithium content, viewed down the [001] direction. Selected CIFs are from the Rietveld refinement of the operando XRD of Nb8W9O47 vs. Li/Li+ in a 1.0 to 3.0 V window.
Crystal structures depicted using VESTA.117
Figure 3.23: Visualization of the transition metal motion occurring in Nb8W9O47 with increasing lithium content, viewed down the [100] direction. Selected CIFs are from the Rietveld refinement of the operando XRD of Nb8W9O47 vs. Li/Li+ in a 1.0 to 3.0 V window.
Crystal structures depicted using VESTA.117
109

On the Structural Origin of Fast Li-ion Cycling in Tetragonal Bronze-type Nb8W9O47
110

Chapter 4
Chemical Fluorination and Structural Evolution of Ruddlesden-Popper
BaLa2Fe2O7
4.1 Introduction
As Li-ion batteries approach their theoretical capacity limits, the continued advancement of
rechargeable batteries will require a shift to new energy storage paradigms. One relatively
unexplored alternative is anionic intercalation, wherein small anions, like F−, act as the
mobile charge carriers instead of cations, like Li+. Considering the massive impact of Liion batteries on modern technologies, the potential of leveraging anionic intercalation to
develop next-generation energy storage systems is extremely promising and intriguing. While
anionic intercalants offer an exciting new battery chemistry to investigate and develop, the
(de)intercalation and diffusion of anions require significantly different conditions than cations
due to their larger ionic radii and negative charges.110 Therefore, a pivotal starting point in
exploring anionic intercalation for energy storage is identifying the structural features that
enable and promote anionic diffusion in densely packed solids.
Fluoride-ion (F-ion) batteries offer the potential to match the energy density of Li-based
111

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
chemistries while also reducing dependence on the limited and strained resources used in
commercial Li-ion cells (i.e., Li, Ni, Co).90 The majority of F-ion battery chemistries reported
in the literature rely on conversion reactions, in which a binary metal fluoride cathode is
reduced to the corresponding metal and a metal anode is oxidized to the fluoride salt.93–96
Although these reactions are reversible, they incur dramatic structural changes that involve
bond breaking and formation, as well as significant unit cell expansion and contraction.
These unit cell level changes correlate to large changes in the physical electrode volume,
which in turn results in electrode damage and leads to poor cycling stability and reversibility
over multiple cycles.
More recently, F-ion batteries based on fluoride intercalation, wherein the host material
may experience crystal structure changes but should broadly maintain its original topology,
have been reported. Intercalation requires significantly less atomic rearrangement than conversion, and as a result, intercalation-based cells typically offer improved reversibility and
longer cycling lifetimes compared to conversion-based cells. Clemens and co-workers presented some of the first reports of intercalation-based F-ion batteries with their demonstration of electrochemical fluoride (de)intercalation into the n = 1 Ruddlesden-Popper oxides
La2SrMnO4 and La2CoO4 and the n = 1 Ruddlesden-Popper oxyfluoride La2NiO3F2.
97,98,126
Ruddlesden-Popper phases have the general formula An+1BnO3n+1, where the A site is
filled by an alkaline earth or rare earth metal cation and the B site is occupied by a transition
metal cation. The crystal structure can be viewed as alternating perovskite (ABO3) and rocksalt (AO) units, where n in the general formula equals the number of layers of octahedra in
each perovskite block. This interweaving of perovskite and rock-salt layers within the crystal
structure produces channels of anionic interstitial sites, presumably an ideal structural motif
for reversible fluoride intercalation and diffusion. The Ruddlesden-Popper structure also
allows for a great degree of compositional tunability, offering the potential to investigate the
impact of the properties of different A and B cations (i.e., size, polarizability, and redox
activity) on fluoride insertion and mobility in the host crystal lattice.
112

4.1. Introduction
Significant breakthroughs in Li-ion batteries were afforded by investigating the structural
changes induced by chemical lithiation, therefore, chemical fluorination should be similarly
leveraged to study the mechanisms of fluoride (de)intercalation and the structural changes
that could be observed under electrochemical conditions. Indeed, there are already many
reports on the chemical fluorination of Ruddlesden-Popper phases and similar layered oxides using fluorine gas directly,127,128 fluorinated polymers,129–136 and other solid fluorinating
agents, like ammonium fluoride or bifluoride.137,138
In the present work, we investigate the structural changes incurred by the chemical fluorination of the n = 2 Ruddlesden-Popper oxide phase, BaLa2Fe2O7, with a mixed, redox
inactive A site (mixed La3+, Ba2+) and a redox-active B site (Fe3+), shown in Figure 4.1.
Two analogous series of oxyfluoride compounds were synthesized from the reaction of the
Figure 4.1: Crystal structure of BaLa2Fe2O7 viewed down the [100] direction. All A sites are
mixed Ba (blue) and La (green) but are shown as ordered in this depiction based on their
preferred A site for simplicity. Crystal structure depicted using VESTA.117
starting oxide and two different fluorinating agents: polyvinylidene fluoride (PVDF) and
ammonium bifluoride (NH4HF2). PVDF and NH4HF2 were chosen because they offer increased safety and control over the fluorine content of the final product compared F2 gas,
and do not require high fluorination temperatures (≤400 ◦C). In this report, we use a com113

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
bination of powder X-ray diffraction, Rietveld refinement, elemental analysis, and infrared
spectroscopy to investigate whether the two fluorinating agents impart different structural
changes over the course of fluorination, as well as how the chosen fluorinating agent and the
associated synthetic conditions impact the extent of fluorination and phase selectivity in the
final product.
4.2 Experimental Methods
4.2.1 Material Synthesis
BaLa2Fe2O7 was synthesized via a traditional ceramic route, similar to the method previously
reported by Gurusinghe et al.127 Stoichiometric amounts of BaCO3 (Sigma-Aldrich, 99.98%),
La2O3 (Sigma-Aldrich, 99.99%), and Fe2O3 (Sigma-Aldrich, ≥96%), in which the metals
reside in the desired oxidation states, were intimately mixed using an agate mortar and
pestle then pressed into a 13 mm pellet. The pellet was placed in a zirconia crucible and
first heated at 900 ◦C for 1 h in a muffle furnace to decompose the carbonate. The sample was
then re-mixed intimately and re-pressed, heated at 1375 ◦C for 36 h in a muffle furnace, then
naturally cooled to yield polycrystalline BaLa2Fe2O7. These conditions produced pellets
with brown-black exteriors and brick red to red-brown interiors, yielding a brick red to redbrown powder once ground. Notably, BaCO3 and Fe2O3 were used directly from the bottle,
whereas La2O3 was dried in a zirconia crucible at 700 ◦C for 15 h in a muffle furnace before
use. If the La2O3 is not dried before weighing, the resulting off-stoichiometry will result in
a major reaction product of perovskitic LaFeO3 instead of the desired Ruddlesden-Popper
BaLa2Fe2O7 phase.
The two series of oxyfluorides were synthesized by mixing BaLa2Fe2O7 with stoichiometric
ratios ranging from 1.0 to 3.0 molar equivalents of the chosen fluorinating agent, PVDF
(MTI) or NH4HF2 (Alfa Aesar, 99.999% trace metals basis). The molecular weight of the
PVDF monomer (CH2CF2) was used for stoichiometric calculations and 1.0 equivalent of
114

4.2. Experimental Methods
PVDF was presumed to provide 2.0 equivalents of fluorine. For the PVDF series, BaLa2Fe2O7
was ground with PVDF in the desired stoichiometric ratio in an agate mortar and pestle
until homogeneous, then placed in a zirconia crucible as a loose powder and heated at 400 ◦C
for 12 h in a muffle furnace. For the NH4HF2 series, BaLa2Fe2O7 was ground with NH4HF2
in the desired stoichiometric ratio in an agate mortar and pestle until homogeneous, then
placed in a zirconia crucible as a loose powder and heated at 400 ◦C for 12 h in a tube
furnace under flowing Ar (∼25 mL/min). Both fluorinating agents and the corresponding
synthetic conditions yielded brick red to bright orange powders, where increasing amounts
of fluorinating agent enhanced the orange hue of the product (least F− = brick red, most
F
− = bright orange). It is also worth noting that higher degrees of impurity are observed
when the BaLa2Fe2O7/fluorinating agent reaction mixture is pressed into pellets and heated,
rather than heated as a loose powder, emphasizing the gas-mediated nature of these reactions.
4.2.2 Powder X-ray Diffraction
Laboratory powder X-ray diffraction (XRD) patterns were collected on a Bruker D8 Advance
diffractometer with a Cu Kα source (λ1=1.5406 ˚A, λ2=1.5444 ˚A), equipped with a LynxEye
XE-T detector. The resulting XRD patterns were refined against published structures using
the Rietveld method as implemented in the TOPAS-Academic suite.145
4.2.3 Electron Probe Microanalysis
Elemental analysis was performed using electron probe microanalysis (EPMA) to investigate
the extent of fluorination in the two series of oxyfluorides. EPMA data were collected using
a JEOL JXA-8200 system with a 10 kV and 5 nA focused beam. The focused electron
beam was 150 nm in diameter. Approximately 50 mg of sample powder was pressed with a
hydraulic press for 5 minutes at 2 tons into a 6 mm diameter pellet. Carbon was sputtered
onto pellets using graphite under vacuum and transferred into the instrument. Data were
collected on 5 to 6 spots per sample.
115

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
4.2.4 Attenuated Total Reflectance Fourier Transform Infrared
Spectroscopy
Attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy was performed using a ThermoScientific Nicolet iS5 FT-IR spectrometer equipped with an iD7 Diamond ATR. The ATR-IR spectra were collected on the pristine powders at room temperature
over a wavenumber range of 3600 to 400 cm−1
.
4.3 Results and Discussion
BaLa2Fe2O7, which adopts the tetragonal space group I4/mmm (#139, Fig. 4.1), was synthesized according to a traditional ceramic method described in detail in the Experimental
Methods section.127 Rietveld refinements of the structure against powder X-ray diffraction
(XRD) data were performed to evaluate the symmetry and phase purity of the sample, with
the resulting structural parameters given in Table 4.1. As shown in Figure 4.2, the prodFigure 4.2: Rietveld refinement of pristine BaLa2Fe2O7 XRD data.
uct was found to contain BaLa2Fe2O7 in the expected I4/mmm space group as the major
116

4.3. Results and Discussion
phase, as well as two small impurities: an orthorhombic perovskite phase, LaFeO3 (P bnm,
6.6 wt %), and un-reacted La2O3 starting material (<1 wt %).
The two series of oxyfluorides were synthesized by reacting the starting oxide,
BaLa2Fe2O7, with molar ratios ranging from 1.0 to 3.0 equivalents of PVDF or NH4HF2,
as described in detail in the Experimental Methods section. Overall, the evolution of the
diffraction patterns observed for both series of oxyfluorides is suggestive of successful fluorination of BaLa2Fe2O7. Peak shifting to lower angles relative to the pristine oxide is
observed for both series, with larger amounts of fluorinating agent corresponding to larger
shifts (Fig. 4.3). These shifts suggest the unit cell volume increases as the amount of fluorinating agent is increased, presumably due to increasing fluorine incorporation into the
crystal structure. However, the extent of peak shifting for the NH4HF2 series appears to
20 30 40 50 60 70 80
2θ (deg.) [λ=1.5418 Å]
Intensity (arb. units)
BaLa2
Fe2
O7
+ 1.0 PVDF (2F)
+ 1.5 PVDF (3F)
+ 2.0 PVDF (4F)
(a) (b)
20 30 40 50 60 70 80
2θ (deg.) [λ=1.5418 Å]
Intensity (arb. units)
BaLa2
Fe2
O7
+ 1.0 NH4
HF2 (2F)
+ 1.5 NH4
HF2 (3F)
+ 2.0 NH4
HF2 (4F)
Figure 4.3: A comparison of the powder XRD patterns from the (a) PVDF oxyfluoride series
and (b) NH4HF2 oxyfluoride series.
consistently fall slightly behind that of the PVDF series. For example, the shifts observed
for the 1.5 equiv. NH4HF2 sample appear to be more similar to those of the 1.0 equiv. PVDF
sample. The differences in raw peak intensity and broadness in the diffraction patterns—
which were obtained using identical scan parameters—also suggest the PVDF oxyfluorides
retain a higher degree of crystallinity than those produced with NH4HF2.
117

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
The pristine oxide, BaLa2Fe2O7, and the “fully fluorinated” phase, BaLa2Fe2O5F4, both
crystallize in the tetragonal I4/mmm space group.127,128 However, no crystal structures for
oxyfluorides between these end members have been reported. Therefore, Rietveld refinements of an I4/mmm unit cell against the XRD data of the intermediate oxyfluorides were
performed to begin evaluating the structural evolution with increasing fluorine content. This
initial analysis quickly revealed that the tetragonal symmetry provides a poor structural description of the oxyfluorides synthesized with ≤1.5 equiv. PVDF and ≤2.0 equiv. NH4HF2
due to significant peak splitting that cannot be fit by the I4/mmm reflections. Peak splitting
often originates from lowered symmetry, therefore, MAXSUB (Maximal Subgroups of Space
Groups) from the Bilbao Crystallographic Server 170 was employed to identify subgroups of
I4/mmm that could capture this change in the diffraction pattern. TRANSTRU (TransFigure 4.4: (a) The pristine I4/mmmm BaLa2Fe2O7 unit cell, (b) the Fmmm intermediate
oxyfluoride unit cell, and (c) the “fully fluorinated” I4/mmmm BaLa2Fe2O5F4 unit cell, all
viewed down the [100] direction. Crystal structure depicted using VESTA.117
form Structure)170 was then used to transform the I4/mmm unit cell into a lower symmetry
unit cell with the chosen subgroup symmetry. Using this process, the orthorhombic space
group Fmmm (#69) was identified as the highest symmetry subgroup that captures the ob118

4.3. Results and Discussion
served peak splitting. The resulting Fmmm unit cell is related to the pristine BaLa2Fe2O7
I4/mmm unit cell by an approximately √
2 expansion of the ab–plane (Fig. 4.4). Notably,
a tetragonal to orthorhombic symmetry lowering upon chemical fluorination of a similar
Ruddlesden-Popper oxide was reported previously.134 Fluorine incorporates into the lower
symmetry Fmmm unit cell as interstitials within the rock salt slabs, as shown in Figure 4.4b.
In Figure 4.4c, fluorine is shown as inserted into the higher symmetry I4/mmm unit cell in
two ways: (1) by replacement of apical oxygens in the FeO6 polyhedra and (2) as interstitials
within the rock-salt slabs. It is important to note, however, that X-rays cannot distinguish
between oxygen and fluorine. As a result, differentiating between oxygen and fluorine at
the apical or equatorial anionic sites is not feasible using XRD alone, therefore only the
interstitial fluorine occupancies are refined.
Rietveld refinements of either the Fmmm or I4/mmm unit cell against the XRD data of
the two series oxyfluorides were repeated, with the refinement results provided in Tables 4.2–
4.14. For the PVDF series, the oxyfluorides synthesized with 1.0 and 1.5 equiv. are best
described by the Fmmm unit cell, whereas the 2.0 equiv. sample is fit best by an I4/mmm
unit cell, exhibiting a return to the original, higher symmetry structure type. On the other
hand, samples produced with 1.0, 1.5, and 2.0 equiv. NH4HF2 were best fit by the Fmmm
unit cell. Therefore, higher equivalents of NH4HF2 (2.5, 3.0) were tested to investigate
whether the transition back to I4/mmm would occur. The 2.5 equiv. NH4HF2 sample was
refined as a mixture of Fmmm and I4/mmm, and the 3.0 equiv. NH4HF2 sample was
best described by an I4/mmm unit cell (Fig. 4.11). Therefore, these refinement results
strongly support the prior observation that the fluorination of BaLa2Fe2O7 requires at least
0.5 equiv. more NH4HF2 to undergo the same structural evolution as the analogous PVDF
sample (Fig. 4.5).
The refined lattice parameters can also be extracted and their trends analyzed to gain
more insight into the structural transformation upon fluorination. The evolution of the
“pseudo-cubic” a lattice parameter (a in tetragonal samples and the average of a/
√
2 and
119

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
Figure 4.5: A comparison of the Rietveld refinements for two selected oxyfluoride samples
from each series to emphasize the offset in fluorinating agent amounts required to observe
analogous structural evolution. (a) 1.0 equiv. PVDF versus 1.5 equiv. NH4HF2 and (b)
2.0 equiv. PVDF versus 3.0 equiv. NH4HF2. The NH4HF2 data is shown at 2x the scale of
the PVDF data for clarity.
b/
√
2 in orthorhombic samples), the c lattice parameter, and the unit cell volume are plotted
versus increasing nominal fluorine content in Figure 4.6. For the PVDF series, the extracted
unit cell volume steadily increases over the course of fluorination, exhibiting nearly a 10%
volume increase from the starting oxide to the 2.0 equiv. sample (Fig. 4.6a). The c lattice
parameter also increases as the amount of fluorinating agent is increased, suggestive of
fluoride intercalation within the rock salt slab (Fig. 4.6c). Notably, the largest changes in
volume and c occur in the samples exhibiting tetragonal symmetry, while the metrics of the
orthorhombic samples remain nearly unchanged. In contrast, the pseudo-cubic a parameter
initially increases with increasing fluorinating agent and the symmetry lowering, then slightly
decreases (Fig. 4.6e). Aligning with the previous observations, the lattice parameter trends
for the NH4HF2 samples fall behind those of the PVDF samples. For example, while the unit
cell volume, c lattice parameter, and pseudo-cubic a parameter all also increase initially, the
second significant change (c/volume increase and a decrease) is not observed in the 2.0 equiv.
NH4HF2 sample like it is for the analogous PVDF sample (Fig. 4.6b,d,f). As the amount of
120

4.3. Results and Discussion
0 1 2 3 4
Nominal fluorine content
3.92
3.96
4.00
a-axis (Å)
20.8
21.6
22.4
c-axis (Å)
315
330
345
360
volume (Å
3
)
0 1 2 3 4
PVDF NH4
HF2 (a)
(c)
(e)
(b)
(d)
(f)
Figure 4.6: The refined values of the unit cell volume (a,b), the c lattice parameter (c,d),
and the pseudo-cubic a lattice parameter (e,f) plotted versus the nominal fluorine content
(stoichiometric amount of fluorine provided based on the molar equivalent of fluorinating
agent used).
NH4HF2 is increased further (≥2.5 equiv.) the changes akin to those seen with 2.0 equiv. of
PVDF are observed, including the ∼10% volume expansion (Fig. 4.12b,d,f).
Another notable difference between the PVDF and NH4HF2 oxyfluoride series derived
from the Rietveld refinements is the product distribution. Both series exhibit impurities
alongside the targeted Ruddlesden-Popper oxyfluoride phases, however, the samples prepared
with NH4HF2 consistently contain higher impurity contents compared to the PVDF series.
While increasing the amount of NH4HF2 induces the return to the higher symmetry I4/mmm
structure type, the refined weight percents of the impurities increase concurrently. The
1.0 equiv. NH4HF2 sample contains a 9.1 wt. % BaF2 impurity and no LaF3, whereas the
2.0 equiv. sample contains a 3.3 wt. % LaF3 impurity and the BaF2 content increases to
15.1 wt. %. Finally, the 3.0 equiv. NH4HF2 sample exhibits 30.6 wt. % BaF2 and 5.1 wt. %
LaF3. Although the samples prepared with PVDF also contain impurities, the impurity
content does not grow as the amount of PVDF is increased. The small La2O3 impurity
observed in the pristine BaLa2Fe2O7 sample appears to transform into LaOF in the 1.0 equiv.
121

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
PVDF sample and into LaF3 in the 1.5 and 2.0 equiv. samples, with the refined fraction
remaining below 3.0 wt. % for all of the samples. This result suggests that PVDF does
not produce impurities, rather, it fluorinates the existing impurities, whereas the increasing
binary fluoride content in the NH4HF2 series suggests the fluorinating conditions are directly
related to impurity formation.
Electron probe microanalysis (EPMA) was performed to assess the Ba:La:Fe ratio of
the pristine oxide and compare the O:F ratios of the two series of oxyfluoride samples (Tables 4.15–4.22). EPMA results for the pristine BaLa2Fe2O7 sample indicate an average
Ba:La:Fe ratio of 1.0:2.1:2.0, which is overall very close to the target ratio of 1.0:2.0:2.0.
This EPMA data also confirms fluorination of BaLa2Fe2O7, as only oxygen is detected in
the pristine oxide sample but both oxygen and fluorine are detected in all of the oxyfluoride samples. The NH4HF2-fluorinated samples consistently exhibit lower O:F ratios (higher
fluorine content) than the analogous PVDF samples. Notably, the O:F ratios of the PVDFfluorinated samples track very closely with the “provided” ratios (Fig. 4.13), calculated based
on the expected amount of oxygen in the sample if all of the provided fluorine is incorporated
into the structure without changing the iron oxidation state. However, this elemental analysis technique cannot be used to decisively determine the extent of fluorination of BaLa2Fe2O7
due to the high quantities of fluorine-containing impurities in the samples, especially those
synthesized with NH4HF2. Considering the NH4HF2 oxyfluorides structurally trail behind
the PVDF analogs according to the XRD analysis, the lower O:F ratios of the NH4HF2 series
are attributed to the higher binary fluoride impurity contents rather than greater fluorine
incorporation into BaLa2Fe2O7.
Infrared spectroscopy (ATR-FTIR) was utilized to probe the local coordination environments to further confirm fluorine incorporation into BaLa2Fe2O7 and compare the fluorination progress between the PVDF and NH4HF2 series. We hypothesized that changes in the
metal–ligand absorption band shapes and/or the number of bands should be observed with
increasing fluorination, as replacing the apical oxygen with fluorine should alter the symme122

4.3. Results and Discussion
try of the FeO6 octahedra and fluorine interstitials in the rock-salt slabs should change the
La/Ba coordination environment. However, the most intense Fe–O bands are often observed
12001400 8001000 400600
Wavenumber (cm-1)
Intensity (arb. units)
BaLa2
Fe2
O7
+ 1.0 PVDF (2F)
+ 1.5 PVDF (3F)
+ 2.0 PVDF (4F)
(a) (b)
12001400 8001000 400600
Wavenumber (cm-1)
Intensity (arb. units)
BaLa2
Fe2
O7
+ 1.0 NH4
HF2 (2F)
+ 1.5 NH4
HF2 (3F)
+ 2.0 NH4
HF2 (4F)
Figure 4.7: A comparison of the ATR-FTIR spectra from the (a) PVDF oxyfluoride series
and (b) NH4HF2 oxyfluoride series.
between 470 and 570 cm−1
,
171,172 while Ba–O bands are typically found around 480 and
500 cm−1
,
172 La–O bands are expected from 490 to 515 cm−1
,
173, and Fe–F bands should be
present between 450 and 540 cm−1
.
174 In short, the bands of interest are likely to overlap
significantly, so directly assigning the signals will be difficult but tracking their evolution
with increasing fluorinating agent should still be informative.
As shown in Figure 4.7, the FTIR spectrum of pristine BaLa2Fe2O7 exhibits two absorption bands at 575 cm−1 and 520 cm−1
. For the PVDF series, the 1.0 equiv. sample exhibits
an asymmetric band at 510 cm−1
, potentially composed of two superimposed bands with
the more intense band maxima at 510 cm−1 and the second band near 530 cm−1
(Fig. 4.7a).
The 1.5 equiv. PVDF sample exhibits a switch in the asymmetry of this band, or possibly
an intensity flip in the two bands with the more intense band maxima now at 535 cm−1 and
the second band near 505 cm−1
. Finally, the most intense band in the 2.0 equiv. PVDF
sample is centered around 530 cm−1 and appears significantly more symmetric and intense
compared to the less fluorinated samples. Overall, the evolution of the FTIR spectrum with
increasing PVDF is indicative of changes in the metal coordination environments, likely due
123

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
to the incorporation of fluorine as interstitials and in place of oxygen in the FeO6 polyhedra.
The FTIR spectra of the oxyfluoride samples prepared with NH4HF2 exhibit the previously observed trend of falling behind the PVDF series, needing at least 0.5 equiv.
more NH4HF2 to impart the same local structure changes observed for the PVDF samples (Fig. 4.7b). Figure 4.8 emphasizes this difference by comparing the FTIR spectra for
analogous PVDF and NH4HF2 samples (a,b) and for offset samples that exhibit the same
crystal structure symmetry in their Rietveld refinements (c,d). Finally, the FTIR spectra
were also utilized to assess whether the samples contained any residual carbon- or nitrogencontaining impurities from the fluorinating agents that may be amorphous and therefore not
visible in the XRD.175,176 As shown in Figure 4.14, none of the oxyfluoride samples exhibit
signals corresponding to residual organic species associated with either of the fluorinating
agents over the measured wavenumber range (3600–400 cm−1
).
2.0 NH4
HF2
2.0 PVDF
750 450600
3.0 NH4
HF2
2.0 PVDF
1.0 NH4
HF2
1.0 PVDF
750 450600
Wavenumber (cm
-1)
Intensity (arb. units)
1.5 NH4
HF2
1.0 PVDF
(a) (b)
(c) (d)
Figure 4.8: A comparison of the ATR-FTIR for selected oxyfluoride samples from each series
to emphasize the offset in fluorinating agent equivalents required to observe analogous structural changes. (a) 1.0 equiv. PVDF versus 1.0 equiv. NH4HF2, (b) 2.0 equiv. PVDF versus
2.0 equiv. NH4HF2, (c) 1.0 equiv. PVDF versus 1.5 equiv. NH4HF2, and (d) 2.0 equiv. PVDF
versus 3.0 equiv. NH4HF2.
124

4.4. Conclusions
4.4 Conclusions
In summary, this report demonstrates successful chemical fluorination of the RuddlesdenPopper oxide BaLa2Fe2O7 using two different fluorinating agents: PVDF and NH4HF2.
Rietveld refinements of power XRD data reveal a structural transformation to a lower symmetry orthorhombic unit cell at low levels of fluorinating agent, followed by a return to the
original tetragonal symmetry as the amount of fluorine is increased. While both fluorinating agents induce this structural evolution of BaLa2Fe2O7, more NH4HF2 is always needed
to impart the changes observed from PVDF. The trend in the O:F ratio from EPMA coupled with the extracted impurity concentrations from the Rietveld refinements also indicate
that fluorination with NH4HF2 produces binary fluoride impurities whereas PVDF produces
higher-purity oxyfluoride products.
These results emphasize the need to further investigate how fluorination conditions affect
the final product distribution, namely the extent of fluorination and impurity formation.
For example, we hypothesize that the carbon present in PVDF might reduce the pristine
oxide to produce oxygen vacancies under the reaction conditions. However, since the PVDF
fluorinations are done in air, atmospheric O2 fills vacancies that remain after fluorination.
On the other hand, NH4HF2 might react with oxygen in the material to form NH3 and H2O,
also producing oxygen vacancies. However, these fluorinations are carried out under Ar, so
there is no O2 to repair vacancies, leaving the material susceptible to Ba extraction to form
BaF2. So, while the target oxyfluoride can be obtained by adding more NH4HF2, the greater
concentration of fluorinating agent may also exacerbate the hypothesized impurity formation
mechanism. Therefore, investigating how NH4HF2 breaks down in Ar and how PVDF breaks
down in air—namely, identifying the fluorine-rich gaseous using mass spectrometry—would
be a beneficial future study to better understand how the chosen fluorinating agent and its
stability (or lack thereof) affects the final product distribution and fluorination mechanism.
125

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
4.5 Supplemental Information
4.5.1 Powder X-ray Diffraction Rietveld Refinement Results
Laboratory powder X-ray diffraction (XRD) patterns were collected in-house using a Bruker
D8 Advance diffractometer, described in detail in the Experimental Methods section. The
resulting XRD patterns were then refined against published structures using the Rietveld
method, as implemented in the TOPAS-Academic suite,145 to evaluate the phase purity and
symmetry. Relevant results are listed in Tables 4.1–4.14.
Notably, the interweaving of perovskite (ABO3) and rock-salt (AO) produces two different
A sites in n = 2 Ruddlesden-Popper phases: (1) 12-coordinate within the perovskite block
and (2) 9-coordinate in the rock-salt layers. In a mixed composition like BaLa2Fe2O7, the
larger cations (Ba2+) are expected to prefer the 12-coordinate site, whereas smaller cations
(La3+) will likely occupy the 9-coordinate site. However, the similar atomic weights of Ba
and La mean they are not easily distinguishable using X-rays alone. Therefore, rather than
refining site occupancies the 12-coordinate site was modeled as all Ba, and the 9-coordinate
site was modeled as all La for the reported refinements. Similarly, oxygen and fluorine cannot
be distinguished using X-ray techniques. Therefore, in the oxyfluoride samples, the apical
anionic site occupancies were not refined, rather, they were modeled as a 50:50 mixture of
oxygen and fluorine (labeled “ap.” in the following refinement tables). On the other hand,
the interstitial anionic sites that are only expected to be filled by fluorine were modeled as
fluorine in the oxyfluoride samples and their occupancies were refined (labeled “int.” in the
following refinement tables).
126

4.5. Supplemental Information
Table 4.1: Rietveld refinement results for pristine BaLa2Fe2O7.
Space Group a b c α = β = γ
BaLa2Fe2O7 I4/mmm 3.93051(2) ˚A 3.93051(2) ˚A 20.8476(1) ˚A 90◦
Atom X Y Z Occupancy B
Ba1 (2b) 0 0 1/2 1.0 1.375(31)
La1 (2b) 0 0 1/2 0.0 1.375(31)
Ba2 (4e) 0 0 0.31804(4) 0.0 1.17(2)
La2 (4e) 0 0 0.31804(4) 1.0 1.17(2)
Fe 0 0 0.0948(1) 1.0 1.28(4)
O1 0 1/2 0.1025(3) 1.0 1.8(1)
O2 0 0 0.2024(4) 1.0 2.4(2)
O3 0 0 0 1.0 2.1(3)
weight % 92.44
Space Group a b c α = β = γ
LaFeO3 P bmn 5.5489(3) ˚A 5.5823(3) ˚A 7.8512(5) ˚A 90◦
Atom X Y Z Occupancy B
La 0.988(2) 0.0165(21) 1/4 1.0 1.0
Fe 0 1/2 0 1.0 1.0
O1 0.7190 0.3020 0.0290 1.0 1.0
O2 0.0800 0.4850 1/4 1.0 1.0
weight % 6.57
Space Group a b c α = β ̸= γ
La2O3 P¯3m 3.94085(68) ˚A 3.94085(68) ˚A 6.13555(215) ˚A 90◦
,120◦
Atom X Y Z Occupancy B
La 1/3 2/3 0.24604 1.0 1.0
O1 1/3 2/3 0.6464 1.0 1.0
O2 0 0 0 1.0 1.0
weight % 0.99
Rwp 3.647
127

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
Table 4.2: Rietveld refinement results for the main phase from the reaction of BaLa2Fe2O7
+ 1.0 equiv. of PVDF. See Table 4.3 for details on minor and impurity phases.
Space Group a b c α = β = γ
BaLa2Fe2O7 Fmmm 5.59722(5) ˚A 5.710135(49) ˚A 21.194(2) ˚A 90◦
+ 2 F
Atom X Y Z Occupancy B
Ba1 (2b) 0 0 1/2 1.0 1.69(4)
La1 (2b) 0 0 1/2 0.0 1.69(4)
Ba2 (2d) 1/2 0 0 1.0 1.69(4)
La2 (2d) 1/2 0 0 0.0 1.69(4)
Ba3 (4i) 0 0 0.31962(5) 0.0 1.82(3)
La3 (4i) 0 0 0.31962(5) 1.0 1.82(3)
Ba4 (4j) 1/2 0 0.81962(5) 0.0 1.82(3)
La4 (4j) 1/2 0 0.81962(5) 1.0 1.82(3)
Fe1 0 0 0.09165(13) 1.0 1.599(45)
Fe2 1/2 0 0.59165(13) 1.0 1.599(45)
O1 1.24(1) 0.252(8) 0.1021(4) 1.0 2.2(2)
O2 0 0 0.1958(6) 1.0 8.0(4)
O3 1/2 0 0.6958(6) 1.0 8.0(4)
O4 0 0 0 1.0 4.2(5)
O5 1/2 0 1/2 1.0 4.2(5)
F (int.) 0.777(9) 1/4 1/4 0.55(17) 1.0
weight % 86.62
Rwp 3.415
128

4.5. Supplemental Information
Table 4.3: Rietveld refinement results for the minor and impurity phases from the reaction
of BaLa2Fe2O7 + 1.0 equiv. of PVDF. See Table 4.2 for details on the main phase.
Space Group a b c α = β = γ
LaFeO3 P bmn 5.557(2) ˚A 5.592(2) ˚A 7.8726(36) ˚A 90◦
Atom X Y Z Occupancy B
La 0.9855(27) 0.027(2) 1/4 1.0 1.0
Fe 0 1/2 0 1.0 1.0
O1 0.7190 0.3020 0.0290 1.0 1.0
O2 0.0800 0.4850 1/4 1.0 1.0
weight % 10.95
Space Group a b c α = β = γ
LaOF Fm¯3m 5.817(2) ˚A 5.817(2) ˚A 5.817(2) ˚A 90◦
Atom X Y Z Occupancy B
La 1/4 1/4 1/4 1.0 1.0
O 0 0 0 0.5 1.0
F 0 0 0 0.5 1.0
weight % 2.43
Rwp 3.415
129

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
Table 4.4: Rietveld refinement results for the main phase from the reaction of BaLa2Fe2O7
+ 1.5 equiv. of PVDF. See Table 4.5 for details on minor and impurity phases.
Space Group a b c α = β = γ
BaLa2Fe2O7 Fmmm 5.6412(1) ˚A 5.6553(1) ˚A 21.7131(7) ˚A 90◦
+ 3 F
Atom X Y Z Occupancy B
Ba1 (2b) 0 0 1/2 1.0 2.58(5)
La1 (2b) 0 0 1/2 0.0 2.58(5)
Ba2 (2d) 1/2 0 0 1.0 2.58(5)
La2 (2d) 1/2 0 0 0.0 2.58(5)
Ba3 (4i) 0 0 0.32097(6) 0.0 2.162(35)
La3 (4i) 0 0 0.32097(6) 1.0 2.162(35)
Ba4 (4j) 1/2 0 0.82097(6) 0.0 2.162(35)
La4 (4j) 1/2 0 0.82097(6) 1.0 2.162(35)
Fe1 0 0 0.0894(2) 1.0 2.07(5)
Fe2 1/2 0 0.5894(2) 1.0 2.07(5)
O1 1.228(8) 0.273(8) 0.0979(4) 1.0 2.3(2)
O2 0 0 0.1795(5) 1.0 5.21(35)
O3 1/2 0 0.6795(5) 1.0 5.21(35)
O4 0 0 0 1.0 7.6(6)
O5 1/2 0 1/2 1.0 7.6(6)
F (int.) 0.7776(65) 1/4 1/4 0.71(1) 1.0
weight % 87.22
Rwp 3.529
130

4.5. Supplemental Information
Table 4.5: Rietveld refinement results for the minor and impurity phases from the reaction
of BaLa2Fe2O7 + 1.5 equiv. of PVDF. See Table 4.4 for details on the main phase.
Space Group a b c α = β = γ
LaFeO3 P bmn 5.560(3) ˚A 5.590(3) ˚A 7.880(5) ˚A 90◦
Atom X Y Z Occupancy B
La 0.993 0.0297 1/4 1.0 1.0
Fe 0 1/2 0 1.0 1.0
O1 0.7190 0.3020 0.0290 1.0 1.0
O2 0.0800 0.4850 1/4 1.0 1.0
weight % 10.16
Space Group a b c α = β ̸= γ
LaF3 P¯3c1 7.188(3) ˚A 7.188(3) ˚A 7.359(4) ˚A 90◦
,120◦
Atom X Y Z Occupancy B
La 0.3401 0 1/4 1.0 1.0
F1 0.312 -0.005 0.581 1.0 1.3
F2 1/3 2/3 0.313 1.0 1.0
F3 0 0 1/4 1.0 1.7
weight % 2.62
Rwp 3.529
131

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
Table 4.6: Rietveld refinement results for BaLa2Fe2O7 + 2.0 equiv. of PVDF.
Space Group a b c α = β = γ
BaLa2Fe2O7 I4/mmm 3.97524(3) ˚A 3.97524(3) ˚A 22.3797(2) ˚A 90◦
+ 4 F
Atom X Y Z Occupancy B
Ba1 (2b) 0 0 1/2 1.0 1.584(45)
La1 (2b) 0 0 1/2 0.0 1.584(45)
Ba2 (4e) 0 0 0.32322(5) 0.0 1.87(3)
La2 (4e) 0 0 0.32322(5) 1.0 1.87(3)
Fe 0 0 0.0862(1) 1.0 1.68(5)
O1 0 1/2 0.09879(35) 1.0 2.2(2)
O2 (ap.) 0 0 0.1764(5) 0.5 6.0(4)
F1 (ap.) 0 0 0.1764(5) 0.5 6.0(4)
O3 0 0 0 1.0 1.5(4)
F (int.) 0 1/2 1/4 1.0(3) 2.85(40)
weight % 92.65
Space Group a b c α = β = γ
LaFeO3 P bmn 5.584(2) ˚A 5.6352(15) ˚A 7.872(2) ˚A 90◦
Atom X Y Z Occupancy B
La 1.00(2) 0.026(3) 1/4 1.0 1.0
Fe 0 1/2 0 1.0 1.0
O1 0.7190 0.3020 0.0290 1.0 1.0
O2 0.0800 0.4850 1/4 1.0 1.0
weight % 5.20
Space Group a b c α = β ̸= γ
LaF3 P¯3c1 7.188(2) ˚A 7.188(2) ˚A 7.359(4) ˚A 90◦
,120◦
Atom X Y Z Occupancy B
La 0.357(5) 0 1/4 1.0 1.0
F1 0.312 -0.005 0.581 1.0 1.3
F2 1/3 2/3 0.313 1.0 1.0
F3 0 0 1/4 1.0 1.7
weight % 2.15
Rwp 3.662
132

4.5. Supplemental Information
Table 4.7: Rietveld refinement results for the main phase from the reaction of BaLa2Fe2O7
+ 1.0 equiv. of NH4HF2. See Table 4.8 for details on minor and impurity phases.
Space Group a b c α = β = γ
BaLa2Fe2O7 Fmmm 5.5946(1) ˚A 5.7077(1) ˚A 21.1800(6) ˚A 90◦
+ 2 F
Atom X Y Z Occupancy B
Ba1 (2b) 0 0 1/2 1.0 1.785(55)
La1 (2b) 0 0 1/2 0.0 1.785(55)
Ba2 (2d) 1/2 0 0 1.0 1.785(55)
La2 (2d) 1/2 0 0 0.0 1.785(55)
Ba3 (4i) 0 0 0.31947(6) 0.0 1.745(37)
La3 (4i) 0 0 0.31947(6) 1.0 1.745(37)
Ba4 (4j) 1/2 0 0.81947(6) 0.0 1.745(37)
La4 (4j) 1/2 0 0.81947(6) 1.0 1.745(37)
Fe1 0 0 0.0912(2) 1.0 1.67(6)
Fe2 1/2 0 0.5912(2) 1.0 1.67(6)
O1 1.2154(35) 0.258(4) 0.10031(45) 1.0 1.5(2)
O2 0 0 0.1945(8) 1.0 9.1(5)
O3 1/2 0 0.6945(8) 1.0 9.1(5)
O4 0 0 0 1.0 2.75(52)
O5 1/2 0 1/2 1.0 2.75(52)
F (int.) 0.75(25) 1/4 1/4 0.578(15) 1.0
weight % 62.01
Rwp 3.200
133

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
Table 4.8: Rietveld refinement results for the minor and impurity phases from the reaction
of BaLa2Fe2O7 + 1.0 equiv. of NH4HF2. See Table 4.7 for details on the main phase.
Space Group a b c α = β = γ
BaLa2Fe2O7 I4/mmm 3.9397(1) ˚A 3.9397(1) ˚A 20.886(1) ˚A 90◦
Atom X Y Z Occupancy B
Ba1 (2b) 0 0 1/2 1.0 2.1(2)
La1 (2b) 0 0 1/2 0.0 2.1(2)
Ba2 (4e) 0 0 0.3185(2) 0.0 1.40(9)
La2 (4e) 0 0 0.3185(2) 1.0 1.40(9)
Fe 0 0 0.0932(6) 1.0 1.86(15)
O1 0 1/2 0.105(1) 1.0 3.12(55)
O2 0 0 0.200(2) 1.0 2.0(8)
O3 0 0 0 1.0 1.0
weight % 28.89
Space Group a b c α = β = γ
BaF2 Fm¯3m 5.966(2) ˚A 5.966(2) ˚A 5.966(2) ˚A 90◦
Atom X Y Z Occupancy B
Ba 0 0 0 1.0 1.0
F 1/4 1/4 1/4 1.0 1.0
weight % 9.10
Rwp 3.200
134

4.5. Supplemental Information
Table 4.9: Rietveld refinement results for BaLa2Fe2O7 + 1.5 equiv. of NH4HF2.
Space Group a b c α = β = γ
BaLa2Fe2O7 Fmmm 5.6076(2) ˚A 5.69655(16) ˚A 21.309(1) ˚A 90◦
+ 3 F
Atom X Y Z Occupancy B
Ba1 (2b) 0 0 1/2 1.0 1.99(7)
La1 (2b) 0 0 1/2 0.0 1.99(7)
Ba2 (2d) 1/2 0 0 1.0 1.99(7)
La2 (2d) 1/2 0 0 0.0 1.99(7)
Ba3 (4i) 0 0 0.32008(8) 0.0 2.03(5)
La3 (4i) 0 0 0.32008(8) 1.0 2.03(5)
Ba4 (4j) 1/2 0 0.82008(8) 0.0 2.03(5)
La4 (4j) 1/2 0 0.82008(8) 1.0 2.03(5)
Fe1 0 0 0.0923(2) 1.0 1.94(7)
Fe2 1/2 0 0.5923(2) 1.0 1.94(7)
O1 1.197(4) 0.254(2) 0.1014(6) 1.0 1.3(4)
O2 0 0 0.19375(87) 1.0 6.2(6)
O3 1/2 0 0.69375(87) 1.0 6.2(6)
O4 0 0 0 1.0 5.55(90)
O5 1/2 0 1/2 1.0 5.55(90)
F (int.) 0.75(17) 1/4 1/4 0.785(27) 1.0
weight % 87.21
Space Group a b c α = β = γ
BaF2 Fm¯3m 5.979(2) ˚A 5.979(2) ˚A 5.979(2) ˚A 90◦
Atom X Y Z Occupancy B
Ba 0 0 0 1.0 1.0
F 1/4 1/4 1/4 1.0 1.0
weight % 12.79
Rwp 3.754
135

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
Table 4.10: Rietveld refinement results for the main phase from the reaction of BaLa2Fe2O7
+ 2.0 equiv. of NH4HF2. See Table 4.11 for details on minor and impurity phases.
Space Group a b c α = β = γ
BaLa2Fe2O7 Fmmm 5.6319(2) ˚A 5.65815(19) ˚A 21.7775(14) ˚A 90◦
+ 4 F
Atom X Y Z Occupancy B
Ba1 (2b) 0 0 1/2 1.0 2.35(8)
La1 (2b) 0 0 1/2 0.0 2.35(8)
Ba2 (2d) 1/2 0 0 1.0 2.35(8)
La2 (2d) 1/2 0 0 0.0 2.35(8)
Ba3 (4i) 0 0 0.32083(9) 0.0 2.27(6)
La3 (4i) 0 0 0.32083(9) 1.0 2.27(6)
Ba4 (4j) 1/2 0 0.82083(9) 0.0 2.27(6)
La4 (4j) 1/2 0 0.82083(9) 1.0 2.27(6)
Fe1 0 0 0.0889(3) 1.0 1.75(7)
Fe2 1/2 0 0.5889(3) 1.0 1.75(7)
O1 1.26(15) 0.263(15) 0.0935(7) 1.0 2.5(4)
O2 0 0 0.18125(75) 1.0 3.2(5)
O3 1/2 0 0.68125(75) 1.0 3.2(5)
O4 0 0 0 1.0 5.95(91)
O5 1/2 0 1/2 1.0 5.95(91)
F (int.) 0.7925(79) 1/4 1/4 0.71(3) 1.0
weight % 81.61
Rwp 3.311
136

4.5. Supplemental Information
Table 4.11: Rietveld refinement results for the minor and impurity phases from the reaction
of BaLa2Fe2O7 + 2.0 equiv. of NH4HF2. See Table 4.10 for details on the main phase.
Space Group a b c α = β = γ
BaF2 Fm¯3m 5.992(1) ˚A 5.992(1) ˚A 5.992(1) ˚A 90◦
Atom X Y Z Occupancy B
Ba 0 0 0 1.0 1.0
F 1/4 1/4 1/4 1.0 1.0
weight % 15.13
Space Group a b c α = β ̸= γ
LaF3 P¯3c1 7.209(7) ˚A 7.209(7) ˚A 7.23(1) ˚A 90◦
,120◦
Atom X Y Z Occupancy B
La 0.2973(45) 0 1/4 1.0 1.0
F1 0.312 -0.005 0.581 1.0 1.3
F2 1/3 2/3 0.313 1.0 1.0
F3 0 0 1/4 1.0 1.7
weight % 3.26
Rwp 3.311
137

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
Table 4.12: Rietveld refinement results for the main phases from the reaction of BaLa2Fe2O7
+ 2.5 equiv. of NH4HF2. See Table 4.13 for details on the impurity phases.
Space Group a b c α = β = γ
BaLa2Fe2O7 Fmmm 5.6260(5) ˚A 5.6353(3) ˚A 22.1781(35) ˚A 90◦
+ 5 F
Atom X Y Z Occupancy B
Ba1 (2b) 0 0 1/2 1.0 2.35(8)
La1 (2b) 0 0 1/2 0.0 2.35(8)
Ba2 (2d) 1/2 0 0 1.0 2.35(8)
La2 (2d) 1/2 0 0 0.0 2.35(8)
Ba3 (4i) 0 0 0.32083(9) 0.0 2.27(6)
La3 (4i) 0 0 0.32083(9) 1.0 2.27(6)
Ba4 (4j) 1/2 0 0.52083(9) 0.0 2.27(6)
La4 (4j) 1/2 0 0.52083(9) 1.0 2.27(6)
Fe1 0 0 0.0889(3) 1.0 1.75(7)
Fe2 1/2 0 0.5889(3) 1.0 1.75(7)
O1 1.261(15) 0.263(15) 0.0935(7) 1.0 2.5(4)
O2 0 0 0.18125(75) 1.0 3.2(5)
O3 1/2 0 0.68125(75) 1.0 3.2(5)
O4 0 0 0 1.0 5.95(91)
O5 1/2 0 1/2 1.0 5.95(91)
F (int.) 0.7925(79) 1/4 1/4 0.71(3) 1.0
weight % 24.34
Space Group a b c α = β = γ
BaLa2Fe2O7 I4/mmm 3.9757(2) ˚A 3.9757(2) ˚A 22.374(1) ˚A 90◦
+ 5 F
Atom X Y Z Occupancy B
Ba1 (2b) 0 0 1/2 1.0 1.9(1)
La1 (2b) 0 0 1/2 0.0 1.9(1)
Ba2 (4e) 0 0 0.3217(1) 0.0 2.09(8)
La2 (4e) 0 0 0.3217(1) 1.0 2.09(8)
Fe 0 0 0.0863(3) 1.0 1.8(1)
O1 0 1/2 0.09985(61) 1.0 1.0
O2 (ap.) 0 0 0.178(1) 0.5 7.9(7)
F1 (ap.) 0 0 0.178(1) 0.5 7.9(7)
O3 0 0 0 1.0 1.0
F (int.) 0 1/2 1/4 1.0(8) 8(2)
weight % 50.20
Rwp 3.502
138

4.5. Supplemental Information
Table 4.13: Rietveld refinement results for the impurity phases from the reaction of
BaLa2Fe2O7 + 2.5 equiv. of NH4HF2. See Table 4.12 for details on the main phases.
Space Group a b c α = β = γ
BaF2 Fm¯3m 5.9987(7) ˚A 5.9987(7) ˚A 5.9987(7) ˚A 90◦
Atom X Y Z Occupancy B
Ba 0 0 0 1.0 1.0
F 1/4 1/4 1/4 1.0 1.0
weight % 21.82
Space Group a b c α = β ̸= γ
LaF3 P¯3c1 7.275(8)) ˚A 7.275(8) ˚A 7.21(1) ˚A 90◦
,120◦
Atom X Y Z Occupancy B
La 0.325(11) 0 1/4 1.0 1.0
F1 0.312 -0.005 0.581 1.0 1.3
F2 1/3 2/3 0.313 1.0 1.0
F3 0 0 1/4 1.0 1.7
weight % 3.64
Rwp 3.502
139

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
Table 4.14: Rietveld refinement results for BaLa2Fe2O7 + 3.0 equiv. of NH4HF2.
Space Group a b c α = β = γ
BaLa2Fe2O7 I4/mmm 3.97523(6) ˚A 3.97523(6) ˚A 22.40145(51) ˚A 90◦
+ 6 F
Atom X Y Z Occupancy B
Ba1 (2b) 0 0 1/2 1.0 1.595(65)
La1 (2b) 0 0 1/2 0.0 1.595(65)
Ba2 (4e) 0 0 0.32308(7) 0.0 1.395(44)
La2 (4e) 0 0 0.32308(7) 1.0 1.395(44)
Fe 0 0 0.0862(2) 1.0 1.31(7)
O1 0 1/2 0.0982(4) 1.0 1.0
O2 (ap.) 0 0 0.1801(7) 0.5 4.89(35)
F1 (ap.) 0 0 0.1801(7) 0.5 4.89(35)
O3 0 0 0 1.0 1.0
F2 (int.) 0 1/2 1/4 1.000(35) 3.1(6)
weight % 64.30
Space Group a b c α = β = γ
BaF2 Fm¯3m 6.0082(4) ˚A 6.0082(4) ˚A 6.0082(4) ˚A 90◦
Atom X Y Z Occupancy B
Ba 0 0 0 1.0 1.0
F 1/4 1/4 1/4 1.0 1.0
weight % 30.61
Space Group a b c α = β ̸= γ
LaF3 P¯3c1 7.217(3) ˚A 7.217(3) ˚A 7.3702(55) ˚A 90◦
,120◦
Atom X Y Z Occupancy B
La 0.3285(97) 0 1/4 1.0 1.0
F1 0.312 -0.005 0.581 1.0 1.3
F2 1/3 2/3 0.313 1.0 1.0
F3 0 0 1/4 1.0 1.7
weight % 5.09
Rwp 3.132
140

4.5. Supplemental Information
Figure 4.9: Rietveld refinements of the oxyfluorides produced by reaction of BaLa2Fe2O7
and (a) 1.0 equiv. PVDF, (b) 1.5 equiv. PVDF, and (c) 2.0 equiv. PVDF.
Figure 4.10: Rietveld refinements of the oxyfluorides produced by reaction of BaLa2Fe2O7
and (a) 1.0 equiv. NH4HF2, (b) 1.5 equiv. NH4HF2, and (c) 2.0 equiv. NH4HF2.
Figure 4.11: Rietveld refinements of the oxyfluorides produced by reaction of BaLa2Fe2O7
and (a) 2.5 equiv. NH4HF2 and (b) 3.0 equiv. NH4HF2.
141

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
0 1 2 3 4
Nominal fluorine content
3.92
3.96
4.00
a-axis (Å)
20.8
21.6
22.4
c-axis (Å)
315
330
345
360
volume (Å
3
)
0 2 4 6
PVDF NH4
HF2 (a)
(c)
(e)
(b)
(d)
(f)
Figure 4.12: The refined values of unit cell volume (a,b), c lattice parameter (c,d), and
pseudo-cubic a lattice parameter (e,f) plotted versus the nominal fluorine content (stoichiometric amount of fluorine provided based on the molar equivalent of fluorinating agent used).
4.5.2 Electron Probe Microanalysis Results
Figure 4.13: A graphical representation of the O:F ratio trends derived from EPMA. The
gray “provided” line represents the O:F ratio calculated from the expected amount of oxygen
(based on all of the provided fluorine incorporating into the material without changing the
iron oxidation state) and the molar equivalent of fluorine provided.
142

4.5. Supplemental Information
Table 4.15: EPMA elemental quantification results for pristine BaLa2Fe2O7.
Ba La Fe O F
Spot Atomic Atomic Atomic Atomic Atomic Atomic Ba:La:Fe
# % % % % % Totals Ratio
1 9.23 18.41 17.80 54.56 0.00 100.00 1.0:2.0:1.9
2 8.78 18.42 18.10 54.70 0.00 100.00 1.0:2.1:2.1
3 9.18 18.53 18.04 54.25 0.00 100.00 1.0:2.0:2.0
4 9.09 18.80 18.06 54.05 0.00 100.00 1.0:2.1:2.0
5 7.66 19.25 16.89 56.20 0.00 100.00 1.0:2.5:2.2
average 8.79 18.68 17.78 54.75 0.00 — 1.0:2.1:2.0
stdev 0.65 0.35 0.51 0.85 0.00 — —
Table 4.16: EPMA elemental quantification results for BaLa2Fe2O7 + 1.0 equiv. of PVDF.
Ba La Fe O F
Spot Atomic Atomic Atomic Atomic Atomic Atomic O:F
# % % % % % Totals Ratio
1 8.76 16.68 15.79 42.80 15.96 100.00 2.682
2 8.82 17.63 16.03 42.32 15.19 100.00 2.785
3 9.14 18.20 17.40 40.47 14.80 100.00 2.734
4 8.91 19.66 16.84 41.66 12.92 100.00 3.225
5 8.57 17.20 16.72 42.10 15.42 100.00 2.731
average 2.831
stdev 0.223
143

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
Table 4.17: EPMA elemental quantification results for BaLa2Fe2O7 + 1.5 equiv. of PVDF.
Ba La Fe O F
Spot Atomic Atomic Atomic Atomic Atomic Atomic O:F
# % % % % % Totals Ratio
1 7.98 16.97 16.56 37.49 20.99 100.00 1.786
2 7.89 17.05 16.60 37.95 20.51 100.00 1.850
3 7.86 16.41 15.44 38.34 21.95 100.00 1.747
4 8.48 17.41 17.19 35.78 21.14 100.00 1.693
5 8.64 17.98 17.37 35.30 20.72 100.00 1.704
6 8.17 17.22 16.29 36.06 22.26 100.00 1.620
average 1.733
stdev 0.080
Table 4.18: EPMA elemental quantification results for BaLa2Fe2O7 + 2.0 equiv. of PVDF.
Ba La Fe O F
Spot Atomic Atomic Atomic Atomic Atomic Atomic O:F
# % % % % % Totals Ratio
1 7.71 16.43 14.59 31.01 30.26 100.00 1.025
2 7.87 16.82 14.81 30.67 29.84 100.00 1.028
3 8.04 16.25 15.93 31.50 28.29 100.00 1.114
4 8.10 16.17 15.78 31.88 28.07 100.00 1.136
5 7.74 15.78 14.70 32.20 29.57 100.00 1.089
6 7.78 15.80 15.62 33.16 27.65 100.00 1.199
average 1.098
stdev 0.067
144

4.5. Supplemental Information
Table 4.19: EPMA elemental quantification results for BaLa2Fe2O7 + 1.0 equiv. of NH4HF2.
Ba La Fe O F
Spot Atomic Atomic Atomic Atomic Atomic Atomic O:F
# % % % % % Totals Ratio
1 7.96 17.09 15.92 37.51 21.52 100.00 1.743
2 7.98 17.26 16.12 37.97 20.67 100.00 1.837
3 8.08 16.64 15.91 38.05 21.32 100.00 1.784
4 8.33 17.35 16.94 38.56 18.82 100.00 2.049
5 8.35 17.41 17.04 39.13 18.08 100.00 2.164
6 8.14 16.67 16.62 38.47 20.10 100.00 1.914
average 1.915
stdev 0.163
Table 4.20: EPMA elemental quantification results for BaLa2Fe2O7 + 1.5 equiv. of NH4HF2.
Ba La Fe O F
Spot Atomic Atomic Atomic Atomic Atomic Atomic O:F
# % % % % % Totals Ratio
1 8.32 17.13 16.60 36.29 21.65 100.00 1.677
2 8.17 16.88 16.37 36.12 22.47 100.00 1.607
3 8.09 16.73 15.94 36.20 23.03 100.00 1.572
4 7.93 16.36 16.64 36.84 22.22 100.00 1.658
5 8.16 16.60 16.94 35.16 23.14 100.00 1.520
6 8.00 16.66 16.81 35.77 22.76 100.00 1.571
average 1.601
stdev 0.059
145

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
Table 4.21: EPMA elemental quantification results for BaLa2Fe2O7 + 2.0 equiv. of NH4HF2.
Ba La Fe O F
Spot Atomic Atomic Atomic Atomic Atomic Atomic O:F
# % % % % % Totals Ratio
1 7.87 16.34 15.95 31.42 28.43 100.00 1.105
2 7.30 16.05 15.79 31.57 29.29 100.00 1.078
3 7.54 16.22 14.77 27.98 33.48 100.00 0.836
4 7.46 16.01 15.53 32.84 28.16 100.00 1.166
5 7.33 16.90 14.05 27.04 34.67 100.00 0.780
average 0.993
stdev 0.173
Table 4.22: EPMA elemental quantification results for BaLa2Fe2O7 + 2.5 equiv. of NH4HF2.
Ba La Fe O F
Spot Atomic Atomic Atomic Atomic Atomic Atomic O:F
# % % % % % Totals Ratio
1 7.42 15.96 15.86 25.84 34.92 100.00 0.740
2 7.61 15.95 15.19 24.52 36.73 100.00 0.668
3 7.80 16.42 15.77 24.46 35.55 100.00 0.688
4 6.88 14.67 15.83 25.55 37.07 100.00 0.689
5 6.75 14.39 15.74 25.01 38.11 100.00 0.656
average 0.688
stdev 0.032
146

4.5. Supplemental Information
4.5.3 Additional ATR-FTIR Spectroscopy Results
(a) (b)
3500 25003000 15002000 5001000
Wavenumber (cm-1)
Intensity (arb. units)
BaLa2
Fe2
O7
+ 1.0 NH4
HF2 (2F)
+ 1.5 NH4
HF2 (3F)
+ 2.0 NH4
HF2 (4F)
3500 25003000 15002000 5001000
Wavenumber (cm-1)
Intensity (arb. units)
BaLa2
Fe2
O7
+ 1.0 PVDF (2F)
+ 1.5 PVDF (3F)
+ 2.0 PVDF (4F)
Figure 4.14: A comparison of the ATR-FTIR spectra from the (a) PVDF oxyfluoride series
and (b) NH4HF2 oxyfluoride series over the entire measured wavenumber range.
147

Chemical Fluorination and Structural Evolution of Ruddlesden-Popper BaLa2Fe2O7
148

Chapter 5
Differential Electrochemical Mass
Spectrometry for F-ion Batteries
5.1 Introduction
The four key components of a battery—cathode, anode, electrolyte, separator—and their
interactions with one another under electrochemical conditions must be understood and optimized to continue advancing rechargeable battery technologies.112 Both the desirable and
problematic properties of rechargeable batteries correspond directly to the electrochemically driven reactions that occur within and between the cell components. Thus, it is no
surprise that developing methods for real-time observation and characterization of these reactions has propelled our understanding and the improvement of modern Li-ion batteries. As
next-generation or “beyond Li-ion” battery chemistries emerge, adapting existing operando
characterization techniques that enable in-depth studies of electrochemical reactions occurring during cell operation will be pivotal to realizing systems that can match or outperform
Li-ion batteries.3
Operando analytical methods played a critical role in establishing design rules and best
practices for Li-ion systems. Along with new battery chemistries come new materials, therefore, the four key rechargeable battery components must be reconsidered and their chemistry
149

Differential Electrochemical Mass Spectrometry for F-ion Batteries
must be re-evaluated in these new electrochemical environments. The advancement of many
emerging Li-ion alternatives is arguably impeded by the lack of investigation into potential
gas evolution reactions that might be associated with the novel ion storage mechanisms and
changed cell components. Differential electrochemical mass spectrometry (DEMS) combines
electrochemistry with mass spectrometry to identify gaseous species that evolve during a
given electrochemical reaction, making it an operando characterization technique that can
play a critical role in the development of new battery chemistries.113
In a continuous flow DEMS measurement, an inert carrier gas (i.e., He or Ar) flows
through a specially designed cell during electrochemical cycling to collect and analyze evolved
gases using mass spectrometry with very little time delay. Thus, this technique enables
operando mass-resolved determination of gaseous reactants, intermediates, or products. This
allows for the direct correlation of changes in the voltage profile to gas evolution events,
providing more information on these processes than electrochemical techniques alone. As a
result, DEMS is particularly useful for identifying interactions between cell components under
electrochemical conditions—for example, identifying deleterious side reactions like electrolyte
degradation—and gaining insight into how these might impact cycling performance.
A typical DEMS setup is comprised of an electrochemical cell equipped with a gas inlet
and outlet that is connected to a potentiostat and a mass spectrometer, both of which are
controlled by a computer for experiment programming and data collection. The carrier gas
flows through the inlet so that any evolved gases are picked up and flow through the cell outlet
into the mass spectrometer, where they are then identified based on their mass-to-charge ratio
(m/z ). Therefore, DEMS is an essential technique for investigating how the reactivity of the
cell components under electrochemical conditions correlates to overall battery performance,
which will in turn guide the continued development of Li-ion alternatives, like F-ion batteries.
Bashian et al. recently reported the electrochemical fluorination of ReO3 from a liquid
electrolyte, tetra-n-butylammonium fluoride (TBAF) dissolved in THF, at room temperature.108 Although the authors obtained clear evidence of fluoride incorporation into ReO3
150

5.1. Introduction
through a combination of galvanostatic cycling, operando X-ray diffraction, and 19F NMR
spectroscopy, reversible cycling was not achievable due to the instability of the fluorinated
ReO3. In their study ReO3 was cycled against Cu metal, which provides a consistent reference potential but cannot be reduced, raising the critical question of the identity of the
counter-reaction for the observed oxidative fluoride insertion. Therefore, DEMS was employed to gather more insight into the electrochemical reactions occurring on oxidation in
ReO3 versus Cu cells. Only H2 evolution was observed in substantial quantities during
the DEMS experiment, therefore, the counter-reaction was presumed to be the hydrogen
evolution reaction (HER) from electrolyte attack.
This result implies some of the major limitations that may be plaguing the development
of room-temperature F-ion batteries are (1) intercalation host stability in the fluorinated and
de-fluorinated state, (2) voltage window and liquid electrolyte compatibility, and (3) counter
electrode reversibility. To begin to understand and address these limitations, investigating
a system capable of reversible fluoride (de)insertion is critical. Following the ReO3 study,
we reported on reversible fluoride (de)insertion in CsMnFeF6 at room temperature from the
same liquid electrolyte, TBAF in THF.109 In this study, we utilized Bi/BiF3 composite counter/reference electrodes to incorporate a species that can be reduced and because M/MFx
composite electrodes are reported to be more stable over long-term cycling compared to pure
MFx.
94–97,150
CsMnFeF6 cycles 1.0 fluoride ion reversibly for up to 10 cycles, after which the capacity steadily fades. The Bi/BiF3 electrode ostensibly provides a counter-reaction for both
oxidation and reduction, so the capacity fade may be due to the limited reversibility of
conversion-based electrodes. On the other hand, the same organic liquid electrolyte is used
as in the ReO3 versus Cu cells. If HER via electrolyte degradation still contributes to the
counter-reaction on oxidation, this could limit the number of cycles based on the amount
of electrolyte available. Clearly, DEMS will be crucial in identifying and understanding the
electrochemical reactions that occur in F-ion batteries containing organic, liquid electrolytes.
151

Differential Electrochemical Mass Spectrometry for F-ion Batteries
This operando gas evolution analysis technique can help uncover the cell component(s) that
might be limiting reversible cycling lifetimes, thereby guiding the improvement and eventual
optimization of room-temperature F-ion batteries.
While the ultimate goal of this work is to investigate how electrochemically-induced
gas evolution reactions impact room-temperature F-ion battery performance, a compatible
DEMS cell must first be designed. ReO3 versus Cu is an ideal test reaction for designing a
fluoride-compatible DEMS setup because it offers a consistent result of HER during oxidation
within the first ∼10 h of cycling. Therefore, the present work will overview the design of a
continuous flow DEMS setup, wherein the electrochemical fluorination of ReO3 using 1.0 M
TBAF in THF is used as a “control” reaction for the testing and adaptation for F-ion battery
chemistry compatibility. After tentative finalization of the DEMS setup, gas evolution during
the reversible, electrochemical (de)fluorination of CsMnFeF6 is investigated.
5.2 Experimental Methods
5.2.1 Material Synthesis
ReO3 was synthesized according to the reduction and precipitation route previously reported
by Chong et al.177 and Bashian et al.108 In a typical synthesis of ReO3, 0.05 mmol of Re2O7
(Sigma-Aldrich, ≥99.9% trace metals basis) was dissolved in 0.5 mL of methanol (MilliporeSigma) in a 10 mL round bottom flask. This solution was then heated to 250 ◦C and held
at this temperature for approximately 5 min, or until the methanol completely evaporated.
This yielded a shiny, brick-red film of ReO3 on the walls of the round bottom flask, which
was then collected and ground in an agate mortar and pestle to yield a brick-red powder
with a metallic luster. ReO3 crystallizes in the cubic space group Pm¯3m (#221),178 shown
in Figure 5.9a. Structural parameters from the Rietveld refinement of the structure against
powder X-ray diffraction (XRD) data are given in Table 5.1.
CsMnFeF6 was synthesized according to the mechanochemical method previously re152

5.2. Experimental Methods
ported by Andrews et al.109 All of the solid reagents used were dried under vacuum at
110 ◦C for 24 h, then stored and weighed in an argon glove box. Stoichiometric amounts of
CsF (Alfa Aesar, 99.9%), MnF2 (Sigma-Aldrich, 98%), and FeF3 (Sigma-Aldrich, 47.1% Fe
by Na2S2O3 titration) were weighed and ground under an argon atmosphere using a SPEX
Sample Prep 8000D Mixer/Mill high-energy ball mill. The mixture of starting materials was
ground for 3 h, in 30 min increments followed by 10 min rest periods, with a 28:1 weight
ratio of steel balls to powder. These conditions yield a taupe, powdery product. CsMnFeF6
crystallizes in the cubic space group F d¯3m (#227),143 shown in Figure 5.9b. Structural
parameters from the Rietveld refinement of the structure against XRD data are given in
Table 5.2.
5.2.2 Powder X-ray Diffraction
Laboratory powder XRD patterns were collected on a Bruker D8 Advance diffractometer
with a Cu Kα source (λ1=1.5406 ˚A, λ2=1.5444 ˚A), equipped with a LynxEye XE-T detector. High-resolution synchrotron powder XRD data was collected at beamline 11-BM
at the Advanced Photon Source (APS) at Argonne National Laboratory using an average
wavelength of 0.458175 ˚A. Discrete detectors covering an angular range from −6 to 28◦ 2θ
were scanned over a 34◦ 2θ range, with data points collected every 0.001◦ 2θ and scan speed
of 0.1◦ per s. The resulting XRD patterns were refined against published structures using
the Rietveld method as implemented in the TOPAS-Academic suite.145
5.2.3 Differential Electrochemical Mass Spectrometry
The DEMS cell design and the interface between the cell, the potentiostat, and the mass
spectrometer are described in detail in the Results and Discussion section.
153

Differential Electrochemical Mass Spectrometry for F-ion Batteries
5.2.3.1 Electrode Preparation
Free-standing ReO3 working electrodes were prepared by hand grinding the active material,
Super P conductive carbon (Alfa Aesar), and polytetrafluoroethylene (PTFE, MTI) binder
in an active–carbon–binder weight ratio of 80:10:10. Free-standing CsMnFeF6 working electrodes were prepared by first ball milling the active material and Ketjen Black EC60JD
conductive carbon (AkzoNobel) for 10 minutes in a SPEX Sample Prep 8000D Mixer/Mill
high-energy ball mill. The CsMnFeF6/conductive carbon mixture was then combined with
PTFE (MTI) binder for an active–carbon–binder weight ratio of 75:20:5. These mixtures
were then pressed into 13 mm pellet electrodes (∼1.0 ton for 10 s) with a typical mass loading
of 10 to 15 mg/cm2
.
The copper metal counter electrodes were prepared by punching Cu foil into 19 mm discs
with a gas flow hole of ∼0.5 mm punched in the center. The Bi/BiF3 composite counter
electrodes were prepared by mixing Bi metal powder (Sigma-Aldrich, 99%), anhydrous BiF3
(Strem, 99.99%), Super P conductive carbon, and polyvinylidene fluoride (PVDF, MTI)
binder in a Bi–BiF3–carbon–binder weight ratio of 40:40:10:10 in a minimal amount of Nmethyl-2-pyrrolidone (MTI) to form a slurry. The slurry was then cast on aluminum foil to
50 µm thick, dried at 80 ◦C overnight, punched into 19 mm discs with a ∼0.5 mm gas flow
hole punched in the center, then dried at 110 ◦C overnight under vacuum before use.
5.2.3.2 Electrochemical Characterization
All cell components and electrodes were dried at 110 ◦C under vacuum for at least 10 h
before cell assembly. All cell assembly was performed in an argon glovebox. The custommade DEMS cell, with a PTFE body and stainless steel current collectors, was utilized as
the electrochemical test cell. A borosilicate glass fiber pad (Whatman GF/D) was used as
the separator and a 1.0 M solution of TBAF in THF (Beantown Chemical) served as the
liquid electrolyte. The reported galvanostatic cycling was performed using a Biologic SP-200
single-channel potentiostat.
154

5.3. Results and Discussion
5.2.3.3 Mass Spectrometry
A Hiden HPR-40 DEMS bench-top quadrupole mass spectrometer equipped with a Faraday/Electron Multiplier detector, a scan range of 1–100 AMU, and a 250 µL/min heated
capillary inlet for gas sampling was used for the reported gas analysis. Argon carrier gas
flowed through the electrochemical cell at a rate of 1 mL/min set by a mass flow controller.
DEMS experiments were only run if atmospheric gases (28 AMU/N2, 32 AMU/O2) corresponded to <0.5% of the total pressure for an empty cell at a 1 mL/min Ar carrier gas flow
rate.
5.3 Results and Discussion
The custom-made, electrochemical DEMS cell described in this report is largely inspired by
the continuous flow cell design reported by Berkes et al. in 2015.115 The cell is composed
of four main parts: (1) the cell body made of chemically resistant polytetrafluoroethylene
(PTFE); (2) the upper current collector, made of stainless steel and engraved with a “gas
flow field” reminiscent of a fuel cell bipolar plate; (3) the bottom current collector, also
made of stainless steel; and (4) the top assembly that compresses the cell stack, seals the
cell, and houses the gas inlet and outlet (Fig. 5.1). The upper current collector (2) includes
gas flow holes along the sides, a hole in the center, and the engraved gas flow field to allow
for effective gas flow into and out of the cell (CAD design shown in Fig. 5.1). The upper
current collector seals the cell stack with an O-ring around the outer diameter above the
gas flow holes, and the gas outlet is sealed from mixing with the inlet stream with a small
O-ring on the top of this current collector around the outlet hole. Following the direction of
the red arrows in Figure 5.1, carrier gas enters the cell through the inlet of the top assembly,
then passes through the side holes of the upper current collector to arrive at the cell stack.
Any evolved gases are collected and carried out of the cell by the carrier gas through the
hole in the center of the upper current collector, flowing through the outlet into the mass
155

Differential Electrochemical Mass Spectrometry for F-ion Batteries
Figure 5.1: A basic schematic of the DEMS cell design. (1) The chemically resistant cell
body, (2) the upper current collector engraved with a “gas flow field”, (3) the bottom current
collector, and (4) the top assembly that seals the cell, compresses the cell stack, and contains
the gas inlet and outlet.
spectrometer. Notably, a hole must also be punched in the top electrode so that gas flow
through the outlet is not blocked. It is also important to note that it is not possible to
distinguish at which electrode the gases evolve using this cell design because a porous glass
fiber is employed as the separator. The bottom current collector (3) seals the interior of the
cell from the atmosphere with an O-ring along its outer diameter and a second O-ring that
is compressed between the bottom of the current collector and the cell body (1). Finally,
the top assembly (4) is secured to the cell body by tightening 6 screws until it compresses
an O-ring placed between the body and top assembly, simultaneously applying pressure to
the cell stack and forming an air-tight seal.
5.3.1 Cell-to-MS Interface Iteration 1
With the basic DEMS electrochemical cell design established, an interface between the carrier
gas source, cell, and mass spectrometer (MS), as well as an interface between the cell and
potentiostat, also needed to be designed. Interfacing the cell with a potentiostat is achieved
simply by tapping 2 mm diameter holes for banana plugs in the side of the top assembly and
156

5.3. Results and Discussion
Figure 5.2: A schematic of the first iteration of the DEMS cell-to-MS interface design.
in the base of the bottom current collector that protrudes from the cell body (Fig. 5.2). In
the first iteration of the cell-to-MS interface design, depicted in Figure 5.2, the carrier gas
passes through a mass flow controller (MFC) directly from a cylinder of the chosen gas before
reaching the cell. The incorporated MFC is equipped for flow rates of <10 mL/min and the
gas cylinder outlet is regulated to ∼80 psi. While stainless steel tubing and fittings are ideal
for ensuring leak-tight seals, MFCs use electrical signals to set gas flow rates. Therefore, a
section of PEEK tubing was incorporated in between the MFC and the cell to insulate the
two components from one another. Otherwise, all gas lines depicted in Figure 5.2 are 1/16”
stainless steel tubing. The gas lines before the cell inlet are hereafter referred to as the “Ar
line” and the gas lines after the cell outlet as the “MS line.”
To easily connect and disconnect the cell to the Ar line and the MS line, Swagelok
miniature quick connects were employed at the cell inlet and outlet. Female quick connects
are always leak-tight, but only special male quick connects with double-end shutoff (DESO)
stems close to the atmosphere when uncoupled. DESO male quick connects were integrated
at the end of the Ar line and the beginning of the MS line and female quick connects were
157

Differential Electrochemical Mass Spectrometry for F-ion Batteries
integrated into the top assembly as the gas inlet and outlet ports so that the interface and
cell are leak-tight even when not connected. However, the mass spectrometer constantly
pulls vacuum, so the cell cannot be connected to the MS line before it is connected to the
Ar line and gas is flowing. Therefore, a 3-way ball valve, labeled “valve” in Figure 5.2, was
integrated into the MS line so it can be connected to the cell outlet but closed to the cell
(open to the atmosphere) until the Ar line is connected and gas is flowing. After the valve,
the MS line includes an open tee, labeled “vent” in Figure 5.2, because 1 mL/min of carrier
gas will flow into the cell during experiments but the mass spectrometer is equipped with a
250 µL/min heated capillary inlet.
With the cell-to-MS interface designed, constructed, and tested for leak tightness using
the empty DEMS cell, the first ReO3 versus Cu control reaction could be run. A free-standing
ReO3 pellet electrode was cycled versus Cu foil with a glass fiber separator (Whatman GF/D)
soaked in 1.0 M TBAF in THF electrolyte. The ReO3 pellet served as the bottom electrode
and the Cu foil served as the top electrode and had a ∼0.5 mm diameter hole punched in
the center to avoid clogging the gas outlet. Galvanostatic cycling with potential limitation
(GCPL) was performed at a C/10 rate between 0.0 and 2.0 V vs. Cu/Cu2+ (Fig. 5.3). The
cell exhibited a relatively normal voltage profile for the system, with an OCV between −150
to −450 mV and a rapid increase to ∼1.0 V with the introduction of positive current into an
upward-sloping plateau. After cycling for approximately 4.75 h (x = 4.75 in FxReO3), the
cell reached the upper voltage limit of 2.0 V and exhibited the expected irreversibility upon
the introduction of negative current. Notably, uncharacteristic dips in voltage were observed
early on in the ∼1.0 V plateau. When tested with a multi-meter, the heated capillary inlet
exhibits a low voltage, suggesting it employs resistive heating and that the observed voltage
bumps may be electrical noise because the cell is not insulated from the capillary.
Several mass fragments were monitored during cycling, including 2 AU (H2), 28 AU (N2,
CO), 32 AU (O2), 40 AU (Ar), and 44 AU (CO2). These were chosen based on the observation of HER upon oxidation in the reported ReO3 versus Cu DEMS experiment108 and to
158

5.3. Results and Discussion
0.0
1.0
2.0
V vs. Cu
0.0 1.0 2.0 3.0 4.0 5.0 6.0
Time (h)
1.0×10-8
2.0×10-8
3.0×10-8
4.0×10-8
MS Response (torr)
44 AU (CO2
)
2 AU (H2
)
OCV GCPL (C/10) clogged outlet
Figure 5.3: Room-temperature, galvanostatic cycling of ReO3 at a C/10 rate between 0.0
and 2.0 V vs. Cu/Cu2+ (top panel) in the continuous flow DEMS cell using iteration 1 of the
cell-to-MS interface. GCPL data (top panel) and the corresponding m/z signal evolution
(bottom panel) are both plotted versus time.
monitor the cell seal during cycling by comparing the ratio of atmospheric gases to the Ar
carrier gas. Atmospheric gas-related fragments are not included in the main text figures for
clarity. As shown in Figure 5.3, the 2 AU (H2) signal quickly increases with the onset of
oxidation, but is plagued by repeated and random spikes and drops as the cycling proceeds.
Interestingly, the 44 AU (CO2) signal also increases, but during the rest period rather than
upon the introduction of positive current, suggesting this gaseous species evolves from the
cell components with or without electrochemical influence. After just over 3.0 h of cycling,
both the 2 AU and 44 AU show dramatic fluctuations (Fig. 5.3, highlighted in red) that also
correspond to significant disturbances in the atmospheric gas-related fragments, suggestive
of outlet clogging and/or issues with the cell seal. While the previous result of HER on oxidation was successfully replicated and the rapid response time between the start of cycling
and observation of gas evolution is extremely promising, these results strongly suggest the
outlet and interior cell seals need to be reconsidered for improved F-ion compatibility.
159

Differential Electrochemical Mass Spectrometry for F-ion Batteries
5.3.2 Cell-to-MS Interface Iteration 2
A schematic of the second iteration of the cell-to-MS interface design is depicted in Figure 5.4. In this design, the Ar line is unchanged from the first iteration of the interface.
Deconstruction of the first cell revealed that while one of the key issues was indeed outlet
clogging, arguably the biggest problem was damage to the outlet and MS line quick connects upon contact with the electrolyte. As carrier gas flows through the cell and the mass
spectrometer pulls vacuum, some of the electrolyte is inevitably carried through the outlet
as well. This is particularly problematic because the electrolyte employed in this study is
THF-based and miniature quick connects rely on interior Viton O-rings for sealing, but Viton
exhibits very poor chemical resistance to THF. To resolve this issue, the outlet quick connect
was replaced with the 3-way ball valve, labeled “valve” in Figure 5.4. Quick connects were
still incorporated after the valve to retain the ease of connecting and disconnecting the cell
and the MS line, but switching the position of the valve and quick connects makes it less
likely that the interior O-rings will be exposed to significant amounts of THF.
Figure 5.4: A schematic of the second iteration of the DEMS cell-to-MS interface design.
160

5.3. Results and Discussion
A 20 psi back pressure regulator, labeled “BPR” in Figure 5.4, was integrated into the MS
line after the valve to maintain a pressure of 20 psi in the cell and vent any excess pressure
in an attempt to mitigate the outlet clogging. The back pressure regulator is made entirely
of PEEK, so its integration into the MS line also serves the secondary purpose of insulating
the cell from the heated capillary inlet’s stray electrical signal. The remainder of the MS
line is unchanged from the first iteration of the cell-to-MS interface. Finally, all of the cell
O-rings were originally Viton and degraded significantly upon exposure to the electrolyte,
therefore, they were replaced with either Kalrez or EPDM. Kalrez is vastly more chemically
resistant to THF than EPDM, but the latter is available in a wider range of diameters and
thicknesses and is still better suited for use with THF than Viton so it can be employed
when needed for specific dimensions.
Once the second iteration of the cell-to-MS interface was constructed and leak tested, another ReO3 versus Cu control DEMS experiment was run. An identical cell stack—composed
of a ReO3 pellet bottom electrode, a Cu foil top electrode, and a glass fiber separator soaked
in 1.0 M TBAF in THF—was cycled under the same conditions so that the only changes
being tested were those to the cell-to-MS interface. The GCPL data shows similar but
improved cycling behavior from the first cell, with a slightly higher OCV between −50 to
−250 mV, the same rapid increase to ∼1.0 V with positive current, and an upward-sloping
plateau until reaching the upper voltage limit of 2.0 V (Fig. 5.5). Notably, the uncharacteristic dips in voltage are no longer observed during the plateau with the introduction of the
PEEK back pressure regulator between the cell and the MS.
The same mass fragments were monitored during cycling as in the first control reaction,
again with particular attention to 2 AU (H2) and 44 AU (CO2). As shown in the bottom panel
of Figure 5.5, the new cell-to-MS interface still exhibits rapid response time. The 2 AU (H2)
signal once again quickly increases with the onset of oxidation, but is still plagued by large
spikes in the MS response. While a spiky signal is not necessarily an issue for qualitative
analysis of the results, it precludes quantitative data analysis. Notably, this cell was also
161

Differential Electrochemical Mass Spectrometry for F-ion Batteries
0.0
1.0
2.0
V vs. Cu
0.0 1.0 2.0 3.0 4.0 5.0
Time (h)
5.0×10-9
1.0×10-8
1.5×10-8
2.0×10-8
MS Response (torr)
44 AU (CO2
)
2 AU (H2
)
OCV GCPL (C/10)
Figure 5.5: Room-temperature, galvanostatic cycling of ReO3 at a C/10 rate between 0.0
and 2.0 V vs. Cu/Cu2+ (top panel) in the continuous flow DEMS cell using iteration 2 of the
cell-to-MS interface. GCPL data (top panel) and the corresponding m/z signal evolution
(bottom panel) are both plotted versus time.
monitored after cycling and the 2 AU (H2) mass fragment signal immediately decreases when
the cell returns to rest after oxidation, confirming that HER occurs in ReO3 versus Cu cells
specifically under oxidative potentials (right side of Fig. 5.5, boxed in gray). Similar to the
first control DEMS, the 44 AU (CO2) signal increases during the rest period and remains
steady during and after cycling. These results indicate that the updates to the cell-to-MS
interface design vastly improved the outlet and interior cell seals, but the persistent spiking
of the 2 AU (H2) signal suggests a cell-related issue may still need to be resolved.
5.3.3 Cell-to-MS Interface Iteration 3
The spiky signal was hypothesized to originate from H2 bubble formation and/or gas flow
field clogging, the latter of which was investigated first. A previously undiscussed feature of
the DEMS cell design is the upper electrode support. Sealing the cell and compressing the
cell stack pushes the upper electrode and the separator into the gas flow field so that they
take the form of the channels and block them. Therefore, a thin stainless steel disc with
162

5.3. Results and Discussion
the same diameter as the upper electrode current collector and a hole punched in the center
was included between the upper electrode and the engraved gas flow field to help the upper
electrode keep its shape and prevent this type of blockage. The original upper electrode
support (used in the first two tests) was handmade rather than CNC machined because it
is a simple part that was needed quickly. However, the inner hole of this support is slightly
oversized compared to the actual outlet size, and as a result, a small fraction of the gas flow
field was still exposed to the cell stack. Thus, to test if the spiky signal originates from gas
flow field clogging an improved upper electrode support, CNC machined to have an identical
outlet hole size as the upper current collector, was incorporated into the cell stack. The
remainder of the interface (Ar line MS line, and cell inlet/outlet) is unchanged from the
second iteration of the cell-to-MS interface depicted in Figure 5.4.
With the new counter electrode support in place, a third replicate of the ReO3 versus Cu
control DEMS cell (described in detail in the previous sections) was constructed and cycled.
The GCPL data reflects that of the second cell, with an OCV between −50 to −250 mV,
a nearly instant increase to ∼1.0 V with the introduction of positive current, and the same
upward sloping plateau up to 2.0 V (Fig. 5.6). Interestingly, this cell also achieved a slightly
higher capacity than the previous two iterations, cycling for approximately 6.0 h (x = 6.0 in
FxReO3) before reaching the upper voltage limit.
In addition to the mass fragments monitored in the first two trials, the 42 AU (THF)
mass fragment was added because the 44 AU (CO2) trend suggests it could be related to an
inherently volatile cell component: the electrolyte. The bottom panel of Figure 5.6 shows
that the rapid response time is maintained, with the 2 AU (H2) signal quickly increasing at
the start of oxidation. Most notable is the lack of 2 AU (H2) signal spiking observed in this
trial. Other than the very clear initial increase, the 2 AU (H2) signal smoothly increases
until reaching approximately 1.5 × 10−8
torr, then remains steady until the cell returns to
rest after cycling and the signal promptly decreases (Fig. 5.6). Both the 44 AU (CO2) and
42 AU (THF) signals increase during the rest period and remain steady during and after
163

Differential Electrochemical Mass Spectrometry for F-ion Batteries
0.0
1.0
2.0
V vs. Cu
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
Time (h)
5.0×10-9
1.0×10-8
1.5×10-8
2.0×10-8
MS Response (torr)
44 AU (CO2
)
2 AU (H2
)
OCV GCPL (C/10)
Figure 5.6: Room-temperature, galvanostatic cycling of ReO3 at a C/10 rate between 0.0
and 2.0 V vs. Cu (top panel) in the continuous flow DEMS cell using iteration 3 of the
cell-to-MS interface. GCPL data (top panel) and the corresponding m/z signal evolution
(bottom panel) are both plotted versus time.
cycling, but the MS response of the 42 AU (THF) signal in torr is nearly two orders of
magnitude higher than that of the 44 AU (CO2) signal (Fig. 5.10). The analogous trends
in these two fragments and the greater prevalence of the 42 AU (THF) signal suggest that
CO2 is not being evolved, rather, the fragments are both associated with the evaporation of
THF from the electrolyte.
Overall, these results indicate that the continuous flow DEMS cell and the cell-to-MS
interface are compatible with F-ion chemistries and can be used to produce data of high
enough quality for quantitative analysis rather than just qualitative observations. However,
while it is not surprising considering the constant flow of carrier gas through the cell, the THF
evaporation observed in the ReO3 DEMS control experiments cannot be ignored. Electrolyte
evaporation will inevitably be a larger issue as we explore reversible F-ion systems that cycle
for days to weeks or that utilize organic, liquid electrolytes based on more volatile solvents.
A potential solution to this issue is introducing a method of replenishing the electrolyte
solvent in the form of a gas bubbler (or a “humidifier”) the solvent in the Ar line before
164

5.3. Results and Discussion
the cell.115 Passing the carrier gas through the bubbler before the cell would help maintain
a constant composition of electrolyte solution in the DEMS cell during long measurements.
However, the cell inlet may also need to be adapted for experiments that require a bubbler,
as the Viton O-rings in the inlet quick connect likely will not be able to withstand prolonged
exposure to the solvent-enriched carrier gas.
5.3.4 Investigating New F-ion Systems
With the most critical issues of adapting the DEMS cell and cell-to-MS interface to be F-ion
compatible resolved, gas evolution during the reversible, electrochemical (de)fluorination of
CsMnFeF6 was investigated. A free-standing CsMnFeF6 pellet electrode was cycled versus
Bi/BiF3 on Al foil, the preparation of which is described in detail in the Experimental
Methods section, with a glass fiber separator (Whatman GF/D) soaked in 1.0 M TBAF in
THF. The CsMnFeF6 pellet served as the bottom electrode and the Bi/BiF3 on Al foil served
as the top electrode and had a ∼0.5 mm diameter hole punched in the center to avoid clogging
the gas outlet. GCPL was performed at a C/20 rate between 0.0 and 1.4 V vs. Bi/Bi3+
(Fig. 5.7). As seen in the top panel of Figure 5.7, the first oxidation exhibits a smoothly
sloped voltage curve that gradually flattens towards the upper voltage limit of 1.4 V and
achieves <50% of the theoretical capacity (75 mAh/g) calculated for one fluoride per formula
unit of CsMnFeF6. The voltage profile of the first reduction exhibits four distinct regions:
(I) a steep decrease from 1.4 to 1.15 V, (II) a short, sloping plateau from 1.15 to 0.85 V, (III)
a second steep decrease from 0.85 to 0.5 V, and (IV) a sloping plateau from 0.5 to 0.0 V,
again achieving <50% of the theoretical capacity. The voltage profile of the second oxidation
closely resembles the reversal of the first reduction curve, and by the third cycle, the voltage
profile is well-defined and achieves a capacity of around 75% of the theoretical value on both
oxidation and reduction. The voltage profiles for subsequent cycles similarly exhibit sloped
plateaus during oxidation (reduction) at approximately 0.7 V (0.5 V) and 1.2 V (0.9 V).
Notably, the DEMS cell does not achieve higher than ∼75% of the theoretical capacity and
165

Differential Electrochemical Mass Spectrometry for F-ion Batteries
0.0
0.5
1.0
1.5
V vs. Bi/Bi3+
0.0 15.0 30.0 45.0 60.0 75.0 90.0
Time (h)
3.0×10-9
6.0×10-9
9.0×10-9
1.2×10-8
MS Response (torr)
44 AU (CO2
)
2 AU (H2
)
Figure 5.7: Room-temperature, galvanostatic cycling of CsMnFeF6 at a C/20 rate between
0.0 and 1.4 V vs. Bi/Bi3+ (top panel) in the continuous flow DEMS cell using iteration 3 of
the cell-to-MS interface. GCPL data (top panel) and the corresponding m/z signal evolution
(bottom panel) are both plotted versus time.
it begins to fade in cycle 6, whereas a typical Swagelok cell with the same cell stack can
achieve and maintain nearly 100% of the theoretical capacity from cycles 4 through 10 before
capacity fade begins.109
Mass fragments 2 AU (H2), 28 AU (N2, CO), 32 AU (O2), 40 AU (Ar), and
44 AU (CO2, THF) were monitored during electrochemical cycling. These were chosen
to investigate whether HER occurs on oxidation (2 AU), to monitor THF evaporation over
long-term cycling (44 AU), and to monitor the cell seal during cycling by comparing the ratio of atmospheric gases to the Ar carrier gas (28, 32, 40 AU). The atmospheric gas-related
fragments are not included in the main text figures for clarity. As shown in Figure 5.7,
there is no clear trend in the 2 AU (H2) signal correlated to the voltage profile until a sharp
increase is observed with the onset of the third oxidation. The signal remains steady during
oxidation, then immediately decreases with the onset of negative current (reduction). This
clear H2 evolution on oxidation and decrease upon reduction continues for the remainder of
the DEMS experiment, suggesting the oxidative electrolyte degradation observed for ReO3
could be occurring in this cell as well.
166

5.3. Results and Discussion
Similar to the ReO3 DEMS control experiments, the 44 AU (CO2, THF) signal increases
during the rest period and levels out early on in the cycling. However, around the 10.0 h
mark of cycling in the CsMnFeF6 versus Bi/BiF3 cell, the 44 AU (CO2, THF) signal decreases
sharply and steadily declines for the remainder of the experiment. While this signal decrease
was never observed in the ReO3 control cells, those experiments also never surpassed 8.0 h
of cycling before reaching the upper voltage limit. This indicates the flow of carrier gas
through the cell does indeed lead to high levels of THF evaporation, therefore, adding an
electrolyte bubbler into the cell-to-MS interface will be necessary for these longer-duration
experiments. The cell seal was maintained for the entirety of this nearly 100.0 h experiment,
suggesting that despite the issue of electrolyte evaporation, the cell design itself is compatible
for long-term, F-ion DEMS experiments to investigate the gas evolution over many cycles.
We originally hypothesized that although Bi/BiF3 counter electrodes should provide an
oxidative and reductive counter-reaction, the observed capacity fade in CsMnFeF6 versus
Bi/BiF3 cells could be due to the limited reversibility of these conversion-based counter
electrodes. However, the CsMnFeF6 versus Bi/BiF3 DEMS experiment in the adapted continuous flow cell reveals H2 evolution upon oxidation when 1.0 M TBAF in THF is employed
as the liquid electrolyte. These results suggest that HER via electrolyte degradation still
contributes to the counter-reaction on oxidation, and as a result, the amount of electrolyte
available may be limiting the cycling lifetime of these cells. Thus, a clear follow-up experiment is quantitative DEMS until cell failure. If the amount of electrolyte is truly limiting
cell performance due to oxidative electrolyte degradation, then H2 evolution should decline
as the capacity fades and the amount of H2 evolved should align with the amount of electrolyte added. The observed THF evaporation, a phenomenon that is likely unique to the
DEMS cell due to carrier gas flow through the cell, also likely contributes to the observed
early onset of capacity fade. Therefore, an electrolyte bubbler will be integrated into the
cell-to-MS interface to more accurately investigate the capacity fade observed for CsMnFeF6
versus Bi/BiF3 using DEMS.
167

Differential Electrochemical Mass Spectrometry for F-ion Batteries
5.4 Conclusions
In summary, this report describes the design of a continuous flow DEMS electrochemical cell
and cell-to-mass spectrometer interface compatible with F-ion battery chemistries. The cell
and interface design component of the presented work emphasizes that one of the biggest
issues that will be faced in adapting operando techniques established for Li-ion systems for
use with F-ion systems is actually very simple: ensuring “inactive” cell components are
chemically inert to the new electrolytes. In other words, all of the non-electrode components
of the cell that we have determined to be stable in Li-ion electrolytes are not necessarily
going to work for F-ion electrolytes.
This report also underscores the importance of identifying F-ion electrolytes with wider
voltage stability windows, particularly under oxidative potentials. The electrochemical fluorination of ReO3 from a liquid electrolyte, 1.0 M TBAF in THF, was utilized as a control
reaction to guide our testing and adaptation of the DEMS setup and consistently exhibited the expected HER on oxidation along with a rapid response time. The tentatively
finalized continuous flow DEMS setup was then employed to investigate gas evolution during the reversible, electrochemical (de)fluorination of CsMnFeF6, also from 1.0 M TBAF in
THF. The observation of H2 evolution on oxidation in the CsMnFeF6 versus Bi/BiF3 cell,
presumably from electrolyte degradation, suggests that electrolyte instability is one of the
largest factors limiting the advancement of F-ion batteries. While developing improved electrolytes is critical, we must concurrently work towards identifying other materials capable
of F-ion intercalation to realize a full intercalation-based cell and avoid the limitations of
conversion-based counter electrodes. Overall, the presented work emphasizes the potential
of utilizing DEMS to better understand the electrochemical reactions that occur in F-ion
batteries containing organic, liquid electrolytes and identify the cell component(s) limiting
their performance.
168

5.5. Supplemental Information
5.5 Supplemental Information
5.5.1 Powder X-ray Diffraction and Rietveld Refinement Results
Figure 5.8: (a) Crystal structure of ReO3. Rhenium cations are colored brown-red and
oxygen anions are colored gold. (b) Crystal Structure of CsMnFeF6. Cesium cations are
colored light teal, mixed manganese/iron cationic sites are colored maroon, and fluorine
anions are colored blue. Crystal structures are depicted using VESTA.117
1.6 2.4 3.2 4.0 4.8 5.6
Q (Å-1)
Intensity (arb. units)
observed
calculated
difference
1.6 2.4 3.2 4.0 4.8 5.6
Q (Å-1)
Intensity (arb. units)
observed
calculated
difference
(a) (b)
ReO3 CsMnFeF6
Figure 5.9: Rietveld refinements of (a) pristine ReO3 lab XRD data and (b) pristine
CsMnFeF6 synchrotron XRD data. Rietveld refinements performed in TOPAS.145
169

Differential Electrochemical Mass Spectrometry for F-ion Batteries
Table 5.1: Results from the Rietveld refinement of pristine ReO3 against lab XRD data.
Parameter ReO3
Space Group Pm¯3m
a = b = c 3.74887(8) ˚A
α = β = γ 90◦
Re position (0, 0, 0)
O position (1/2, 0, 0)
BRe 1.70(2)
BO 1.60(8)
Rwp 5.376
Table 5.2: Results from the Rietveld refinement of pristine CsMnFeF6 against synchrotron
XRD data.
Parameter CsMnFeF6
Space Group F d¯3m
a = b = c 10.54511(9) ˚A
α = β = γ 90◦
Cs position (3/8, 3/8, 3/8)
Mn/Fe position (0, 0, 0)
F position (0.31773(6), 1/8, 1/8)
BCs 3.054(9)
BMn 1.328(8)
BF e 1.328(8)
BF 3.31(2)
Rwp 5.844
170

5.5. Supplemental Information
5.5.2 Additional Differential Electrochemical Mass Spectrometry
Results
0.0
1.0
2.0
V vs. Cu
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
Time (h)
1.0×10-10
1.0×10-9
1.0×10-8
1.0×10-7
1.0×10-6
1.0×10-5
MS Response (torr)
36 AU (Ar)
42 AU (THF)
28 AU (N2
, CO)
2 AU (H2
)
44 AU (CO2
)
32 AU (O2
)
OCV GCPL (C/10)
Figure 5.10: Room-temperature, galvanostatic cycling of ReO3 at a C/10 rate between 0.0
and 2.0 V vs. Cu (top panel) in the continuous flow DEMS cell using iteration 3 of the
cell-to-MS interface. GCPL data (top panel) and the corresponding m/z signal evolution
(bottom panel) are both plotted versus time.
171

Differential Electrochemical Mass Spectrometry for F-ion Batteries
172

References
(1) Lee, H.; Calvin, K.; Dasgupta, D.; Krinner, G.; Mukherji, A.; Thorne, P.; Trisos,
C.; Romero, J.; Aldunce, P.; Barret, K., et al. IPCC, 2023: Climate Change 2023:
Synthesis Report, Summary for Policymakers. Contribution of Working Groups I, II
and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate
Change. IPCC, Geneva, Switzerland. 2023.
(2) Kerestes, R. J.; Reed, G. F.; Sparacino, A. R. In 2012 IEEE Power and Energy
Society General Meeting, 2012, pp 1–9.
(3) Kim, S.; Kim, H.-S.; Kim, B.; Kim, Y.-J.; Jung, J.-W.; Ryu, W.-H. Advanced Energy
Materials 2023, 13, 2301983.
(4) Tarascon, J.-M.; Armand, M. Nature 2001, 414, 359–367.
(5) Tarascon, J.-M. Nature Chemistry 2010, 2, 510.
(6) Banza Lubaba Nkulu, C.; Casas, L.; Haufroid, V.; De Putter, T.; Saenen, N. D.;
Kayembe-Kitenge, T.; Musa Obadia, P.; Kyanika Wa Mukoma, D.; Lunda Ilunga,
J.-M.; Nawrot, T. S., et al. Nature sustainability 2018, 1, 495–504.
(7) Wang, M.; Wang, Q.; Ding, X.; Wang, Y.; Xin, Y.; Singh, P.; Wu, F.; Gao, H.
Interdisciplinary Materials 2022, 1, 373–395.
(8) Song, J.; Sahadeo, E.; Noked, M.; Lee, S. B. The Journal of Physical Chemistry
Letters 2016, 7, 1736–1749.
(9) Dhir, S.; Wheeler, S.; Capone, I.; Pasta, M. Chem 2020, 6, 2442–2460.
(10) Nazri, G. A.; Pistoia, G., Lithium Batteries: Science and Technology; Springer US:
2008.
(11) Bulatovic, S. M. In Handbook of Flotation Reagents: Chemistry, Theory and Practice;
Elsevier: Amsterdam, 2015, pp 57–76.
(12) Survey, U. S. G. Mineral Commodity Summaries 2024 - Fluorspar, https://pubs.
usgs.gov/periodicals/mcs2024/mcs2024-fluorspar.pdf, Accessed July 1, 2024.
(13) Survey, U. S. G. Mineral Commodity Summaries 2024 - Lithium, https://pubs.
usgs.gov/periodicals/mcs2024/mcs2024-lithium.pdf, Accessed July 1, 2024.
(14) Whittingham, M. S. Chemical Reviews 2014, 114, 11414–11443.
173

References
(15) Larcher, D.; Tarascon, J.-M. Nature chemistry 2015, 7, 19–29.
(16) Goodenough, J. B. Accounts of chemical research 2013, 46, 1053–1061.
(17) Whittingham, M. S.; Yoshino, A. The Royal Swedish Academy of Sciences 2019.
(18) Kamat, P. V. ACS Energy Letters 2019, 4, 2757–2759.
(19) Deng, D. Energy Science & Engineering 2015, 3, 385–418.
(20) Talaie, E.; Bonnick, P.; Sun, X.; Pang, Q.; Liang, X.; Nazar, L. F. Chemistry of
Materials 2017, 29, 90–105.
(21) Mizushima, K.; Jones, P. C.; Wiseman, P. J.; Goodenough, J. B. Mater. Res. Bull.
1980, 15, 783–789.
(22) Yabuuchi, N.; Ohzuku, T. J. Power Sources 2003, 119–121, 171–174.
(23) Guilmard, M.; Pouillerie, C.; Croguennec, L.; Delmas, C. Solid State Ion. 2003, 160,
39–50.
(24) Padhi, A. K.; Nanjundaswamy, K. S.; Goodenough, J. B. J. Electrochem. Soc. 1997,
144, 1188–1194.
(25) Padhi, A. K.; Nanjundaswamy, K. S.; Masquelier, C.; Okada, S.; Goodenough, J. B.
J. Electrochem. Soc. 1997, 144, 1609.
(26) Yoshino, A.; Sanechika, K.; Nakajima, T. Secondary battery, US Patent 4,668,595,
1987.
(27) Brandt, K Solid State Ionics 1994, 69, 173–183.
(28) Yu, S.-H.; Feng, X.; Zhang, N.; Seok, J.; Abru˜na, H. D. Accounts of Chemical Research
2018, 51, 273–281.
(29) Zhang, W.-J. Journal of Power Sources 2011, 196, 13–24.
(30) Ellis, B. L.; Lee, K. T.; Nazar, L. F. Chemistry of Materials 2010, 22, 691–714.
(31) Whittingham, M. S. Intercalation chemistry 1982, 1–18.
(32) Guerard, D; Herold, A Carbon 1975, 13, 337–345.
(33) Shu, Z.; McMillan, R.; Murray, J. Journal of The Electrochemical Society 1993, 140,
922.
174

References
(34) Thackeray, M. M.; David, W. I. F.; Bruce, P. G.; Goodenough, J. B. Mater. Res.
Bull. 1983, 18, 461–472.
(35) Thackeray, M. M. Progress in Solid State Chemistry 1997, 25, 1–71.
(36) Yang, S.; Zavalij, P. Y.; Whittingham, M. S. Electrochemistry Communications 2001,
3, 505–508.
(37) Malik, R.; Abdellahi, A.; Ceder, G. Journal of the electrochemical society 2013, 160,
A3179.
(38) Cava, R. J.; Murphy, D. W.; Zahurak, S. M. J. Electrochem. Soc. 1983, 130, 2345–
2351.
(39) Cheng, Q.; Liang, J.; Zhu, Y.; Si, L.; Guo, C.; Qian, Y. J. Mater. Chem. A 2014, 2,
17258–17262.
(40) Griffith, K. J.; Forse, A. C.; Griffin, J. M.; Grey, C. P. J. Am. Chem. Soc. 2016, 138,
8888–8899.
(41) Griffith, K. J.; Wiaderek, K. M.; Cibin, G.; Marbella, L. E.; Grey, C. P. Nature 2018,
556, 556–563.
(42) Deng, Q.; Fu, Y.; Zhu, C.; Yu, Y. Small 2019, 15, 1804884.
(43) Griffith, K. J.; Seymour, I. D.; Hope, M. A.; Butala, M. M.; Lamontagne, L. K.;
Preefer, M. B.; Ko¸cer, C. P.; Henkelman, G.; Morris, A. J.; Cliffe, M. J.; Dutton,
S. E.; Grey, C. P. J. Am. Chem. Soc. 2019, 141, 16706–16725.
(44) Luo, Y.; Le Calvez, E.; Zhou, Y.; Gautron, E.; Quarez, E.; Preefer, M.; Crosnier,
O.; Weker, J. N.; Pilon, L.; Brousse, T.; Dunn, B. Chemistry of Materials 2023, 35,
8675–8685.
(45) Andrews, J. L.; Brady, M. J.; McClure, E. T.; Melot, B. C. Chemistry of Materials
2022, 34, 4809–4820.
(46) Montoro, L.; Abbate, M; Almeida, E.; Rosolen, J. M. Chemical Physics Letters 1999,
309, 14–18.
(47) Preefer, M. B.; Saber, M.; Wei, Q.; Bashian, N. H.; Bocarsly, J. D.; Zhang, W.; Lee,
G.; Milam-Guerrero, J.; Howard, E. S.; Vincent, R. C.; Melot, B. C.; Van der Ven,
A.; Seshadri, R.; Dunn, B. S. Chem. Mater. 2020, 32, 4553–4563.
(48) Park, M.; Zhang, X.; Chung, M.; Less, G. B.; Sastry, A. M. J. Power Sources 2010,
195, 7904–7929.
175

References
(49) Iton, Z. W. B.; See, K. A. Chem. Mater. 2022, 34, 881–898.
(50) Kalinin, S.; Balke, N.; Jesse, S.; Tselev, A.; Kumar, A.; Arruda, T. M.; Guo, S.;
Proksch, R. Mater. Today 2011, 14, 548–558.
(51) Green, P. F., Kinetics, Transport, and Structure in Hard and Soft Materials; CRC
Press: 2005.
(52) Diffusion in Solid Metals and Alloys; Mehrer, H., Ed.; Springer: 1990.
(53) Kikkawa, S.; Miyazaki, S.; Koizumi, M. J. Solid State Chem. 1986, 62, 35–39.
(54) Amatucci, G.; Tarascon, J.-M.; Klein, L. C. J. Electrochem. Soc. 1996, 143, 1114–
1123.
(55) Masquelier, C.; Croguennec, L. Chem. Rev. 2013, 113, 6552–6591.
(56) Nyt´en, A.; Abouimrane, A.; Armand, M.; Gustafsson, T.; Thomas, J. O. Electrochem.
Commun. 2005, 7, 156–160.
(57) Zhou, F.; Cococcioni, M.; Kang, K.; Ceder, G. Electrochem. Commun. 2004, 6, 1144–
1148.
(58) Kang, B.; Ceder, G. Nature 2009, 458, 190–193.
(59) Andersson, A. S.; Kalska, B.; H¨aggstr¨om, L.; Thomas, J. O. Solid State Ionics 2000,
130, 41–52.
(60) Murphy, D. W.; Greenblatt, M.; Cava, R. J.; Zahurak, S. M. Solid State Ion. 1981,
5, 327–329.
(61) Bashian, N. H.; Zhou, S.; Zuba, M.; Ganose, A. M.; Stiles, J. W.; Ee, A.; Ashby, D. S.;
Scanlon, D. O.; Piper, L. F. J.; Dunn, B.; Melot, B. C. ACS Energy Lett. 2018, 3,
2513–2519.
(62) Cava, R.; Santoro, A.; Murphy, D.; Zahurak, S.; Roth, R. J. Solid State Chem. 1982,
42, 251–262.
(63) Wang, H.; Jang, Y.-I.; Huang, B.; Sadoway, D. R.; Chiang, Y.-M. J. Electrochem.
Soc. 1999, 146, 473–480.
(64) Dokko, K. Electrochem. Solid-State Lett. 1999, 3, 125.
(65) Jang, Y.-I.; Huang, B.; Wang, H.; Sadoway, D. R.; Ceder, G.; Chiang, Y.-M.; Liu, H.;
Tamura, H. J. Electrochem. Soc. 1999, 146, 862–868.
176

References
(66) Bashian, N. H.; Preefer, M. B.; Milam-Guerrero, J.; Zak, J. J.; Sendi, C.; Ahsan,
S. A.; Vincent, R. C.; Haiges, R.; See, K. A.; Seshadri, R.; Melot, B. C. J. Mater.
Chem. A 2020, 8, 12623–12632.
(67) Patterson, A. R.; Elizalde-Segovia, R.; Wyckoff, K. E.; Zohar, A.; Ding, P. P.; Turner,
W. M.; Poeppelmeier, K. R.; Narayan, S. R.; Cl´ement, R. J.; Seshadri, R.; Griffith,
K. J. Chemistry of Materials 2023, 35, 6364–6373.
(68) Wadsley, A. D. Acta Cryst. 1961, 14, 660–664.
(69) Wadsley, A. D. Acta Cryst. 1961, 14, 664–670.
(70) Roth, R. S.; Wadsley, A. D. Acta Cryst. 1965, 19, 26–32.
(71) Roth, R. S.; Wadsley, A. D. Acta Cryst. 1965, 19, 32–38.
(72) Roth, R. S.; Wadsley, A. D. Acta Cryst. 1965, 19, 38–42.
(73) Roth, R. S.; Wadsley, A. D. Acta Cryst. 1965, 19, 42–47.
(74) Bursill, L. A.; Hyde, B. G. Philos. Mag. 1971, 23, 3–15.
(75) Bursill, L. A.; Hyde, B. G.; Philp, D. K. Philos. Mag. 1971, 23, 1501–1513.
(76) Anderson, J. In Intercalation Chemistry, Whittingham, M. S., Jacobson, A. J., Eds.,
1982, pp 503–538.
(77) Dreele, R. B. V.; Cheetham, A. K.; Anderson, J. S. Proceedings of the Royal Society
of London. A. Mathematical and Physical Sciences 1974, 338, 311–326.
(78) Ko¸cer, C. P.; Griffith, K. J.; Grey, C. P.; Morris, A. J. J. Am. Chem. Soc. 2019, 141,
15121–15134.
(79) Griffith, K. J.; Harada, Y.; Egusa, S.; Ribas, R. M.; Monteiro, R. S.; Von Dreele,
R. B.; Cheetham, A. K.; Cava, R. J.; Grey, C. P.; Goodenough, J. B. Chem. Mater.
2021, 33, 4–18.
(80) Andersson, S. Bulletin de Min´eralogie 1967, 90, 522–527.
(81) Cordier, S.; Roisnel, T.; Poulain, M. J. Solid State Chem. 2004, 177, 3119–3126.
(82) Bohnke, C.; Bohnke, O.; Fourquet, J. L. Mol. Cryst. Liq. Cryst. Sci. Technol., Sect.
A 1998, 311, 23–29.
177

References
(83) Bohnke, C; Fourquet, J.; Randrianantoandro, N; Brousse, T; Crosnier, O J. Solid
State Electrochem. 2001, 5, 1–7.
(84) Reddy, M.; Madhavi, S.; Rao, G. S.; Chowdari, B. J. Power Sources 2006, 162,
1312–1321.
(85) Saritha, D Mechanics, Materials Science & Engineering Journal 2018, 14, 1–6.
(86) Lundberg, M.; Sundberg, M.; Magn´eli, A. Journal of Solid State Chemistry 1982,
44, 32–40.
(87) Krumeich, F. Crystals 2021, 11, DOI: 10.3390/cryst11121514.
(88) Krumeich, F. Chemistry of Materials 2022, 34, 911–934.
(89) R¨udorff, W; Hofmann, U. Zeitschrift f¨ur anorganische und allgemeine Chemie 1938,
238, 1–50.
(90) Xiao, A. W.; Galatolo, G.; Pasta, M. Joule 2021, 5, 2542–4785.
(91) Davis, V. K. et al. Science 2018, 362, 1144–1148.
(92) Hou, X.; Zhang, Z.; Shen, K.; Cheng, S.; He, Q.; Shi, Y.; Denis, Y.; Su, C.-y.; Li,
L.-J.; Chen, F. Journal of The Electrochemical Society 2019, 166, A2419.
(93) Anji Reddy, M.; Fichtner, M. J. Mater. Chem. 2011, 21, 17059–17062.
(94) Rongeat, C.; Anji Reddy, M.; Diemant, T.; Behm, R. J.; Fichtner, M. J. Mater.
Chem. A 2014, 2, 20861–20872.
(95) Gschwind, F.; Rodriguez-Garcia, G.; Sandbeck, D.; Gross, A.; Weil, M.; Fichtner,
M.; H¨ormann, N. J. Fluor. Chem. 2016, 182, 76–90.
(96) Grenier, A.; Porras-Gutierrez, A.-G.; Groult, H.; Beyer, K. A.; Borkiewicz, O. J.;
Chapman, K. W.; Dambournet, D. J. Mater. Chem. A 2017, 5, 15700–15705.
(97) Nowroozi, M. A.; Wissel, K.; Rohrer, J.; Munnangi, A. R.; Clemens, O. Chem. Mater.
2017, 29, 3441–3453.
(98) Nowroozi, M. A.; Ivlev, S.; Rohrer, J.; Clemens, O. J. Mater. Chem. A 2018, 6,
4658–4669.
(99) Schoonman, J.; Oversluizen, G.; Wapenaar, K. Solid State Ion. 1980, 1, 211–221.
178

References
(100) Roos, A.; Aalders, A.; Schoonman, J.; Arts, A.; de Wijn, H. Solid State Ion. 1983,
9-10, 571–574.
(101) Breuer, S.; Gombotz, M.; Pregartner, V.; Hanzu, I.; Wilkening, M. Energy Storage
Mater. 2019, 16, 481–490.
(102) Mohammad, I.; Witter, R.; Fichtner, M.; Anji Reddy, M. ACS Appl. Energy Mater.
2018, 1, 4766–4775.
(103) Liu, L.; Yang, L.; Liu, M.; Li, X.; Shao, D.; Luo, K.; Wang, X.; Luo, Z. J. Alloys
Compd. 2020, 819, 152983.
(104) Nowroozi, M. A.; Mohammad, I.; Molaiyan, P.; Wissel, K.; Munnangi, A. R.;
Clemens, O. J. Mater. Chem. A 2021, 9, 5980–6012.
(105) Galatolo, G.; Alshangiti, O.; Di Mino, C.; Matthews, G.; Xiao, A. W.; Rees, G. J.;
Schart, M.; Chart, Y. A.; Olbrich, L. F.; Pasta, M. ACS Energy Letters 2024, 9,
85–92.
(106) Zou, P.; Wang, C.; He, Y.; Xin, H. L.; Lin, R. Nano Letters 2024, 24, 5429–5435.
(107) Fu, Z.; Yang, X.; Tian, Y.; Hu, X.; Wang, Y.; Lin, L.; Kang, F.; Wang, G.; Li, B.;
Zhou, D. Energy Storage Materials 2024, 70, 103533.
(108) Bashian, N. H. et al. Chem. Mater. 2021, 33, 5757–5768.
(109) Andrews, J. L. et al. ACS Energy Letters 2022, 7, 2340–2348.
(110) West, W. C.; Whitacre, J. F.; Leifer, N.; Greenbaum, S.; Smart, M.; Bugga, R.;
Blanco, M.; Narayanan, S. R. J. Electrochem. Soc. 2007, 154, A929.
(111) Borkiewicz, O. J.; Shyam, B.; Wiaderek, K. M.; Kurtz, C.; Chupas, P. J.; Chapman,
K. W. Journal of Applied Crystallography 2012, 45, 1261–1269.
(112) Goodenough, J. B.; Park, K.-S. Journal of the American Chemical Society 2013,
135, 1167–1176.
(113) Baltruschat, H. Journal of the American Society for Mass Spectrometry 2004, 15,
1693–1706.
(114) Bruckenstein, S.; Gadde, R. R. Journal of the American Chemical Society 1971, 93,
793–794.
(115) Berkes, B. B.; Jozwiuk, A.; Vraˇcar, M.; Sommer, H.; Brezesinski, T.; Janek, J. Analytical Chemistry 2015, 87, 5878–5883.
179

References
(116) Subramanian, M.; Aravamudan, G.; Subba Rao, G. Prog. Solid State Chem. 1983,
15, 55–143.
(117) Momma, K.; Izumi, F. J. Appl. Crystallogr. 2008, 41, 653–658.
(118) Pannetier, J. J. Phys. Chem. Solids 1973, 34, 583–589.
(119) Nedjar, R.; Borel, M.; Hervieu, M.; Raveau, B. Materials Research Bulletin 1988,
23, 91–97.
(120) Kurtz, W.; Geller, R.; Dachs, H.; Convert, P. Solid State Commun. 1976, 18, 1479–
1483.
(121) Wilde, P.; Catlow, C. Solid State Ion. 1998, 112, 173–183.
(122) van Dijk, M.; Burggraaf, A.; Cormack, A.; Catlow, C. Solid State Ion. 1985, 17,
159–167.
(123) Sayed, F. N.; Mandal, B. P.; Jain, D.; Pillai, C. G. N. S.; Tyagi, A. K. J. Eur. Ceram.
Soc. 2012, 32, 3221–3228.
(124) W¨ohler, F Ann. Phys 1824, 2, 350–358.
(125) Dickens, P. G.; Whittingham, M. S. Q. Rev. Chem. Soc. 1968, 22, 30–44.
(126) Wissel, K.; Schoch, R.; Vogel, T.; Donzelli, M.; Matveeva, G.; Kolb, U.; Bauer, M.;
Slater, P. R.; Clemens, O. Chemistry of Materials 2021, 33, 499–512.
(127) Gurusinghe, N. N.; de la Figuera, J.; Marco, J. F.; Thomas, M. F.; Berry, F. J.;
Greaves, C. Materials Research Bulletin 2013, 48, 3537–3544.
(128) Gurusinghe, N. N. M.; Fones, J. C.; Marco, J. F.; Berry, F. J.; Greaves, C. Dalton
Trans. 2014, 43, 2038–2043.
(129) Slater, P. Journal of Fluorine Chemistry 2002, 117, 43–45.
(130) Berry, F. J.; Ren, X.; Heap, R.; Slater, P.; Thomas, M. F. Solid State Communications
2005, 134, 621–624.
(131) Berry, F. J.; Heap, R; Helgason, O.; Moore, E. A.; Shim, S; Slater, P. R.; Thomas,
M. F. Journal of Physics: Condensed Matter 2008, 20, 215207.
(132) Clemens, O.; Haberkorn, R.; Slater, P. R.; Beck, H. P. Solid State Sciences 2010, 12,
1455–1463.
180

References
(133) Berry, F. J.; Coomer, F. C.; Hancock, C.; Orn Helgason; Moore, E. A.; Slater, P. R.; ¨
Wright, A. J.; Thomas, M. F. Journal of Solid State Chemistry 2011, 184, 1361–
1366.
(134) Hancock, C. A.; Herranz, T.; Marco, J. F.; Berry, F. J.; Slater, P. R. Journal of Solid
State Chemistry 2012, 186, 195–203.
(135) Wissel, K.; Heldt, J.; Groszewicz, P. B.; Dasgupta, S.; Breitzke, H.; Donzelli, M.;
Waidha, A. I.; Fortes, A. D.; Rohrer, J.; Slater, P. R.; Buntkowsky, G.; Clemens, O.
Inorganic Chemistry 2018, 57, 6549–6560.
(136) Wissel, K.; Vogel, T.; Dasgupta, S.; Fortes, A. D.; Slater, P. R.; Clemens, O. Inorganic
Chemistry 2020, 59, 1153–1163.
(137) Sophronov, V. L.; Kalaev, M. E.; Makaseev, Y. N.; Sachkov, V. I.; Verkhoturova,
V. V. IOP Conference Series: Materials Science and Engineering 2016, 110, 012069.
(138) Flynn, S.; Zhang, C.; Griffith, K. J.; Shen, J.; Wolverton, C.; Dravid, V. P.; Poeppelmeier, K. R. Inorganic Chemistry 2021, 60, 4463–4474.
(139) Whittingham, M. S. Chem. Rev. 2004, 104, 4271–4302.
(140) Manthiram, A. Nat. Commun. 2020, 11, 1550.
(141) Thieu, D. T.; Fawey, M. H.; Bhatia, H.; Diemant, T.; Chakravadhanula, V. S. K.;
Behm, R. J.; K¨ubel, C.; Fichtner, M. Adv. Funct. Mater. 2017, 27, 1701051.
(142) Christe, K. O.; Wilson, W. W.; Wilson, R. D.; Bau, R.; Feng, J. A. J. Am. Chem.
Soc. 1990, 112, 7619–7625.
(143) Kurtz, W.; Roth, S. Physica B+C 1977, 86-88, 715–716.
(144) Segal, E. B. Chem. Health Saf. 2000, 7, 18–23.
(145) Coelho, A. A. J. Appl. Crystallogr. 2018, 51, 210–218.
(146) Peterson, M.; Hunt, C.; Wang, Z.; Heinrich, S. E.; Wu, G.; M´enard, G. Dalton Trans.
2020, 49, 16268–16277.
(147) Wang, Z.; Johnson, S. I.; Wu, G.; M´enard, G. Inorg. Chem. 2021, 60, 8242–8251.
(148) Ravel, B.; Newville, M. J. Synchrotron Radiat. 2005, 12, 537–541.
(149) Picone, P.; Saegusa, N.; Srivastava, R.; Wong, C.; Morrish, A.; Wanklyn, B. Solid
State Commun. 1982, 44, 279–283.
181

References
(150) Nowroozi, M. A.; Clemens, O. ACS Appl. Energy Mater. 2018, 1, 6626–6637.
(151) Nowroozi, M. A.; Wissel, K.; Donzelli, M.; Hosseinpourkahvaz, N.; Plana-Ruiz, S.;
Kolb, U.; Schoch, R.; Bauer, M.; Malik, A. M.; Rohrer, J., et al. Commun. Mater.
2020, 1, 1–16.
(152) Grenier, A.; Porras-Gutierrez, A. G.; Body, M.; Legein, C.; Chr´etien, F.; RaymundoPi˜nero, E.; Doll´e, M.; Groult, H.; Dambournet, D. J. Phys. Chem. C 2017, 121,
24962–24970.
(153) Banks, E.; Deluca, J.; Berkooz, O. J. Solid State Chem. 1973, 6, 569–573.
(154) Menil, F. J. Phys. Chem. Solids. 1985, 46, 763–789.
(155) Ati, M.; Sougrati, M.-T.; Rousse, G.; Recham, N.; Doublet, M.-L.; Jumas, J.-C.;
Tarascon, J.-M. Chem. Mater. 2012, 24, 1472–1485.
(156) Wyckoff, K. E.; Robertson, D. D.; Preefer, M. B.; Teicher, S. M. L.; Bienz, J.;
Kautzsch, L.; Mates, T. E.; Cooley, J. A.; Tolbert, S. H.; Seshadri, R. Chemistry
of Materials 2020, 32, 9415–9424.
(157) Craig, D. C.; Stephenson, N. C. Acta Crystallographica Section B 1969, 25, 2071–
2083.
(158) M. Montemayor, S.; Alvarez Mendez, A.; Mart´ınez-de la Cruz, A.; F. Fuentes, A.;
M. Torres-Mart´ınez, L. J. Mater. Chem. 1998, 8, 2777–2781.
(159) De la Cruz, A.; Ram´ırez, I.; Gonz´alez, L. Materials Research Bulletin 2003, 38, 525–
531.
(160) Yan, L.; Cheng, X.; Yu, H.; Zhu, H.; Liu, T.; Zheng, R.; Zhang, R.; Shui, M.; Shu, J.
Energy Storage Materials 2018, 14, 159–168.
(161) Levin, E. E.; Grebenkemper, J. H.; Pollock, T. M.; Seshadri, R. Chemistry of Materials 2019, 31, 7151–7159.
(162) Gengenbach, T. R.; Major, G. H.; Linford, M. R.; Easton, C. D. Journal of Vacuum
Science & Technology A 2021, 39.
(163) Vakarin, S.; Baraboshkin, A.; Kaliev, K.; Zyrianov, V. Journal of Crystal Growth
1995, 151, 121–126.
(164) Augustyn, V.; Come, J.; Lowe, M. A.; Kim, J. W.; Taberna, P.-L.; Tolbert, S. H.;
Abru˜na, H. D.; Simon, P.; Dunn, B. Nat. Mater. 2013, 12, 518–522.
182

References
(165) Yao, W.; Zhu, H.; Wang, M.; Li, P.; Liu, P.; Zou, P.; Nie, A.; Wang, G.; Kang, F.;
Yang, C. Chemistry of Materials 2022, 34, 388–398.
(166) Chen, H.; Adams, S. IUCrJ 2017, 4, 614–625.
(167) Chen, H.; Wong, L. L.; Adams, S. Acta Crystallographica Section B 2019, 75, 18–33.
(168) Wong, L. L.; Phuah, K. C.; Dai, R.; Chen, H.; Chew, W. S.; Adams, S. Chemistry of
Materials 2021, 33, 625–641.
(169) Brown, I. D. Chemical Reviews 2009, 109, 6858–6919.
(170) Aroyo, M. I.; Perez-Mato, J. M.; Capillas, C.; Kroumova, E.; Ivantchev, S.;
Madariaga, G.; Kirov, A.; Wondratschek, H. Zeitschrift f¨ur KristallographieCrystalline Materials 2006, 221, 15–27.
(171) Chen, Z.; Fan, T.; Zhang, Q.; He, J.; Fan, H.; Sun, Y.; Yi, X.; Li, J. Journal of Colloid
and Interface Science 2019, 536, 105–111.
(172) Todera¸s, M; Filip, S; Ardelean, I Journal of optoelectronics and advanced materials
2006, 3, 1121–1123.
(173) Ghorbani-Moghadam, T.; Kompany, A.; Bagheri-Mohagheghi, M.; Abrishami, M. E.
Journal of Magnetism and Magnetic Materials 2018, 465, 768–774.
(174) Nakamoto, K., Infrared and Raman Spectra of Inorganic and Coordination Compounds: Theory and applications in inorganic chemistry; Wiley: 2008.
(175) Jacobs, J.; Marques, M. A. L.; Wang, H.-C.; Dieterich, E.; Ebbinghaus, S. G. Inorganic Chemistry 2021, 60, 13646–13657.
(176) Jacobs, J.; Wang, H.-C.; Marques, M. A. L.; Xu, K.; Schmedt auf der G¨unne, J.;
Ebbinghaus, S. G. Inorganic Chemistry 2024, 63, 6075–6081.
(177) Chong, Y. Y.; Fan, W. Y. Journal of colloid and interface science 2013, 397, 18–23.
(178) Meisel, K. Zeitschrift f¨ur anorganische und allgemeine Chemie 1932, 207, 121–128.
183

Abstract (if available)

Abstract Identifying and understanding the structural deformations induced by the (de)intercalation and diffusion of ions is critical for the development of improved energy storage materials. This dissertation is comprised of a collection of research studies that investigate both cationic and anionic intercalation-driven structural transformations in host crystal structures, toward the goal of understanding their impact on overall battery performance.

First, I present a study on the electrochemical (de)fluorination of CsMnFeF6 from a liquid electrolyte to investigate the potential of the cubic pyrochlore crystal structure as a fluoride intercalation host. Up to one fluoride ion can be reversibly (de)inserted into mechanochemically synthesized CsMnFeF6 for multiple cycles at room temperature. The structural transformation during cycling begins as a subtle, but clearly reversible, expansion and contraction of the CsMnFeF6 cubic lattice on fluoride insertion and removal, respectively. As cycling progresses, a topotactic transformation of CsMnFeF6 from the original defect pyrochlore structure into a related, orthorhombic polytype is observed. While the distorted material continues to reversibly cycle fluoride ions for multiple cycles, significant capacity fade is observed soon after, raising the question of whether the structural distortion or other cell components (i.e., electrolyte stability) negatively impact the cycling performance.

This is followed by an investigation into the structural evolution of tetragonal tungsten bronze-type Nb8W9O47 during lithium (de)intercalation to identify the origin of fast Li cycling in bronze and bronze-derived materials. Operando X-ray diffraction studies reveal that the space group is unchanged and the unit cell parameters vary anisotropically during discharge. Although higher capacities are obtained by cycling to lower voltages, this incurs significantly greater unit cell volume expansion and negatively impacts the reversibility. Furthermore, the variation in refined transition metal positions indicates charge compensation occurs through off-centering of the metals within their octahedra, suggesting that rigid crystal structures, like that of Nb8W9O47, exhibit such impressive cycling capabilities because they undergo displacive rather than rotational deformations during (de)lithiation.

I then revisit fluoride (de)intercalation with a study on the chemical fluorination of Ruddlesden-Popper BaLa2Fe2O7, focused on the impact of the fluorinating agent identity and the fluorination conditions on the structural evolution of the oxide and product distribution. By investigating the structural distortions induced by chemical fluorination, this study offers insight into the changes in structure that might be observed when BaLa2Fe2O7 is electrochemically fluorinated. Two oxyfluoride series were synthesized by reacting BaLa2Fe2O7 with either polyvinylidene fluoride (PVDF) or ammonium bifluoride (NH4HF2). A symmetry lowering from the pristine oxide’s tetragonal unit cell to an orthorhombic unit cell is observed at low levels of fluorinating agent, however, the original tetragonal symmetry is regained with increasing amounts of fluorine. Both fluorinating agents induce the same structural transformation, but NH4HF2 produces lower purity samples and larger amounts of NH4HF2 are required to impart the same structural changes as PVDF, underscoring the importance of optimized synthetic conditions when leveraging chemical fluorination techniques to investigate fluoride intercalation mechanisms.

Finally, I describe the design and testing of an F-ion compatible continuous flow differential electrochemical mass spectrometry (DEMS) cell and cell-to-mass spectrometer interface. Developing operando techniques, like DEMS, is crucial for the advanced characterization of electrochemical reactions in F-ion batteries, especially for those hypothesized to be limiting the advancement of these systems. The electrochemical fluorination of ReO3 from a liquid electrolyte is first used as a “control” reaction for cell testing and adaptation because it consistently results in oxidative H2 evolution due to electrolyte degradation. Then, the finalized DEMS set-up is used to investigate the reversible, electrochemical (de)fluorination of CsMnFeF6, which is found to similarly exhibit H2 evolution on oxidation, indicative of electrolyte instability in the employed voltage window.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Understanding the structure-property relationship in electrode materials for electrochemical energy storage

PDF

On the role of polyhedral rotations in mediating ion insertion processes for energy storage materials

PDF

Understanding the role of electrode design in determining the electrochemical performance of high-energy/high-power lithium-ion and lithium-sulfur batteries

PDF

Solution phase synthesis routes to functional nanomaterials for energy storage

PDF

Electrochemical pathways for sustainable energy storage and energy conversion

Asset Metadata

Creator Andrews, Jessica Lou Ann (author)

Core Title Understanding intercalation-driven structural transformations in energy storage materials

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Chemistry

Degree Conferral Date 2024-12

Publication Date 09/26/2024

Defense Date 08/29/2024

Publisher Los Angeles, California (original), University of Southern California (original), University of Southern California. Libraries (digital)

Tag anionic intercalation,battery materials,bronze,cationic intercalation,differential electrochemical mass spectrometry,fluoride-ion battery,lithium-ion battery,OAI-PMH Harvest,pyrochlore,Rietveld refinement,Ruddlesden-Popper,X-ray diffraction

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Melot, Brent (committee chair), Brutchey, Richard (committee member), Shao, Yu-Tsun (committee member)

Creator Email andrewsjessica59@gmail.com,jla95903@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC11399B9YU

Unique identifier UC11399B9YU

Identifier etd-AndrewsJes-13549.pdf (filename)

Legacy Identifier etd-AndrewsJes-13549

Document Type Dissertation

Format theses (aat)

Rights Andrews, Jessica Lou Ann

Internet Media Type application/pdf

Type texts

Source 20240926-usctheses-batch-1214 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu