University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ON THE ROLE OF POLYHEDRAL ROTATIONS IN MEDIATING ION INSERTION PROCESSES FOR ENERGY STORAGE MATERIALS by Nicholas H. Bashian A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (CHEMISTRY) December 2020 Copyright 2020 Nicholas H. Bashian an effect of the peculiarity of choice ii Acknowledgments As I write this, a pandemic has spread throughout the world upending day to day life and introducing stresses and challenges that I wouldn’t have predicted even a month earlier. My office and laboratory are closed, travel is canceled, and interactions are almost entirely virtual. This time has helped me to realize the great value of friendship and community, of supporting one another, of caring for those closest to us but also those who we barely know. The outbreak of coronavirus has been a strange and scary time, full of uncertainty and confusion but I have been blessed with amazing friends who have continually acted to ensure we all will prevail in these uncertain times. For that, I will be eternally thankful. Throughout my graduate school career, there has been a vast network of col- leagues, friends, and family who have served to support and encourage me, without whom this journey would not have been possible. These people were there for me when I was feeling lost or doubting my abilities, they were there to push me when I needed motivation, they were there to teach me when I needed instruction, and most importantly they were there to remind me that I was never alone during this journey. While it would be impossible to thank everyone, I would like to thank my wonderful colleagues in the Melot group and throughout the chemistry community for their continual support. iii Contents Page Epigraph ii Acknowledgments iii List of Tables vi List of Figures vii Abstract x 1 Introduction 1 1.1 The Energy Storage Crisis . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Overview of Intercalation Batteries . . . . . . . . . . . . . . . . . . 6 1.3 Intercalation Reaction Mechanisms . . . . . . . . . . . . . . . . . . 9 1.4 Anion Intercalation . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.5 Operando Battery Experiments . . . . . . . . . . . . . . . . . . . . 15 1.6 Relevant Crystal Structures . . . . . . . . . . . . . . . . . . . . . . 17 1.7 Dissertation Overview . . . . . . . . . . . . . . . . . . . . . . . . . 18 2 Correlated Polyhedral Rotations in the Absence of Polarons dur- ing Electrochemical Insertion of Lithium in ReO 3 22 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.2 Experimental Details . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 27 2.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 2.5 Supplemental Information . . . . . . . . . . . . . . . . . . . . . . . 38 3 Understanding the Role of Crystallographic Shear on the Electro- chemical Behavior of Niobium Oxyfluorides 43 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 3.2 Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . 46 3.3 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . 49 iv 3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 3.6 Supplemental Information . . . . . . . . . . . . . . . . . . . . . . . 64 4 TransitionMetalMigrationCanFacilitateIonicDiffusioninDefect Garnet Based Intercalation Electrodes 74 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 4.2 Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . 77 4.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 81 4.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 4.5 Supplemental Information . . . . . . . . . . . . . . . . . . . . . . . 94 5 Electrochemical Oxidative Fluorination of an Oxide Perovskite 108 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 5.2 Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . 111 5.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 117 5.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 5.5 Supplemental Information . . . . . . . . . . . . . . . . . . . . . . . 135 References 151 v List of Tables Page 2.1 Calculated lattice parameters of ReO 3 , LiReO 3 and Li 2 ReO 3 . . . . 38 2.2 Rietveld refinement results for ReO 3 . . . . . . . . . . . . . . . . . 38 3.1 Rietveld refinement results for NbO 2 F and Nb 3 O 7 F . . . . . . . . 64 3.2 Vibrational modes of NbO 2 F Raman spectroscopy . . . . . . . . . 65 3.3 Vibrational modes of Nb 3 O 7 F Raman spectroscopy . . . . . . . . . 65 4.1 δ iso and σ iso for yttrium oxide compounds . . . . . . . . . . . . . . 105 4.2 Calculated lattice parameters of Al 2 (MoO 4 ) 3 and Y 2 (MoO 4 ) 3 . . . . 107 5.1 Fluoride diffusion coefficients for F x ReO 3 . . . . . . . . . . . . . . . 139 5.2 Spin-lattice relaxation times of 19 F signals in F 0.2 ReO 3 . . . . . . . 141 5.3 Table of δ calc iso and δ exp iso for fluorine-containing compounds . . . . . . 143 vi List of Figures Page 1.1 Wind and solar electricity generation . . . . . . . . . . . . . . . . . 1 1.2 Global electric vehicle usage . . . . . . . . . . . . . . . . . . . . . . 2 1.3 Battery costs and predicted battery demand . . . . . . . . . . . . . 5 1.4 Lithium-ion battery design . . . . . . . . . . . . . . . . . . . . . . . 7 1.5 Structural distortion of LiFePO 4 upon delithiation . . . . . . . . . . 11 1.6 Fluoride-ion battery design . . . . . . . . . . . . . . . . . . . . . . . 13 1.7 Electrochemical Operando X-ray diffraction cell . . . . . . . . . . . 16 1.8 Various relevant crystal structures . . . . . . . . . . . . . . . . . . . 19 2.1 Crystal structure of ReO 3 . . . . . . . . . . . . . . . . . . . . . . . 23 2.2 Rietveld refinement of ReO 3 X-ray diffraction . . . . . . . . . . . . 28 2.3 Galvanostatic cycling and cyclic voltammetry of ReO 3 . . . . . . . . 29 2.4 Operando X-ray diffraction of Li x ReO 3 . . . . . . . . . . . . . . . . 30 2.5 Structural distortions of Li x ReO 3 . . . . . . . . . . . . . . . . . . . 32 2.6 Extended galvanostatic cycling of ReO 3 . . . . . . . . . . . . . . . . 33 2.7 Electronic structure calculations for ReO 3 , LiReO 3 , and Li 2 ReO 3 . . 34 2.8 X-ray photoelectron spectroscopy of lithiated ReO 3 . . . . . . . . . 36 2.9 Brillouin zone of the Pm3 ¯ m space group . . . . . . . . . . . . . . . 39 2.10 Brillouin zone of the R3c space group . . . . . . . . . . . . . . . . . 39 2.11 Galvanostatic titration cycling of Li x ReO 3 . . . . . . . . . . . . . . 40 2.12 X-ray photoemmision spectroscopy of Li 2 ReO 3 surface layer . . . . 40 2.13 Structural distortions of Li x ReO 3 within primitive unit cell . . . . . 41 2.14 X-ray diffraction patterns of Li 2 ReO 3 upon delithiation . . . . . . . 42 3.1 Crystal structures of NbO 2 Fand Nb 3 O 7 F . . . . . . . . . . . . . . 44 3.2 Rietveld refinement of NbO 2 Fand Nb 3 O 7 FX-ray diffraction . . . . 50 3.3 Cyclic voltammetry and galvanostatic cycling of NbO 2 Fand Nb 3 O 7 F 51 3.4 Operando X-ray diffraction of NbO 2 Fand Nb 3 O 7 F . . . . . . . . . 53 3.5 Operando X-ray absorption spectroscopy of NbO 2 Fand Nb 3 O 7 F . . 57 3.6 Raman spectra of NbO 2 Fand Nb 3 O 7 Fduring lithiation . . . . . . . 60 3.7 Galvanostatic cycling of NbO 2 Fover expanded voltage window . . . 66 vii 3.8 Galvanostatic cycling of Nb 3 O 7 Fover expanded voltage window . . 67 3.9 Additional cyclic voltammetry of NbO 2 F . . . . . . . . . . . . . . . 68 3.10 Additional cyclic voltammetry of Nb 3 O 7 F . . . . . . . . . . . . . . 68 3.11 Heatmap operando X-ray diffraction of NbO 2 F . . . . . . . . . . . . 70 3.12 Operando X-ray diffraction of NbO 2 Fover two cycles . . . . . . . . 70 3.13 Heatmap operando X-ray diffraction of Nb 3 O 7 F . . . . . . . . . . . 71 3.14 Phase changes in Nb 3 O 7 Fduring (de)lithiation . . . . . . . . . . . . 71 3.15 X-ray diffraction of Nb 3 O 7 Fbefore and after lithiation . . . . . . . . 72 3.16 Additional Raman spectroscopy of NbO 2 Fand Nb 3 O 7 F . . . . . . . 72 3.17 Nb K–edge X-ray absorption spectroscopy of NbO 2 Fupon lithiation 73 3.18 Nb K–edge X-ray absorption spectroscopy of Nb 3 O 7 F . . . . . . . . 73 4.1 Crystal structure of A 2 (MoO 4 ) 3 . . . . . . . . . . . . . . . . . . . . 76 4.2 Galvanostatic cycling of Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 . . . . . . . . 82 4.3 Operando X-ray absorption of Li x Y 2 (MoO 4 ) 3 and Li x Al 2 (MoO 4 ) 3 . 83 4.4 OperandoX-raydiffractionandradialdistributionplotsofLi x Y 2 (MoO 4 ) 3 and Li x Al 2 (MoO 4 ) 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 4.5 27 Al NMR spectra of Li x Al 2 (MoO 4 ) 3 . . . . . . . . . . . . . . . . . 89 4.6 Electronic structure calculations for Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 . . 91 4.7 Scheme of structural distortions in Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 . . 92 4.8 X-ray diffraction of Al 2 (MoO 4 ) 3 before and after lithiation . . . . . 94 4.9 Cyclic voltammogram of Al 2 (MoO 4 ) 3 . . . . . . . . . . . . . . . . . 95 4.10 Cyclic voltammogram of Y 2 (MoO 4 ) 3 . . . . . . . . . . . . . . . . . 96 4.11 Y K-edge X-ray absorption spectroscopy of Li x Y 2 (MoO 4 ) 3 . . . . . 97 4.12 Y K-edge radial distribution plot of Li x Y 2 (MoO 4 ) 3 . . . . . . . . . 97 4.13 Mo K-edge X-ray absorption spectroscopy of Li x Y 2 (MoO 4 ) 3 . . . . . 98 4.14 Fits of radial distribution data for Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 . . 100 4.15 27 Al solid-state NMR spectra of Al 2 (MoO 4 ) 3 . . . . . . . . . . . . . 101 4.16 Fitted 27 Al NMR spectrum of Li 5 Al 2 (MoO 4 ) 3 . . . . . . . . . . . . 102 4.17 Al metal region of 27 Al NMR spectra of Li x Al 2 (MoO 4 ) 3 . . . . . . . 103 4.18 7 Li NMR spectra of Li x Al 2 (MoO 4 ) 3 . . . . . . . . . . . . . . . . . . 103 4.19 95 Mo NMR spectra of Al 2 (MoO 4 ) 3 and Y 2 (MoO 4 ) 3 . . . . . . . . . 104 4.20 89 Y NMR spectrum of Y 2 (MoO 4 ) 3 . . . . . . . . . . . . . . . . . . . 104 4.21 δ iso vs. chemical shielding σ iso for yttrium oxide compounds . . . . 104 4.22 Band diagrams of Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 . . . . . . . . . . . . 106 4.23 Brillouin zone for the Pbcn space group . . . . . . . . . . . . . . . . 107 5.1 Scheme demonstrating the potential of dual-ion batteries . . . . . . 109 5.2 Crystal structure of ReO 3 with ionic radii . . . . . . . . . . . . . . 111 5.3 Galvanostatic charge of F x ReO 3 with corresponding in-situ electro- chemical mass spectrometry . . . . . . . . . . . . . . . . . . . . . . 119 5.4 Galvanostatic titration of F x ReO 3 with liquid and solid electrolyte . 121 viii 5.5 Operando X-ray diffraction, transmission electron microscopy, and F K-edge X-ray absorption spectroscopy of F x ReO 3 . . . . . . . . . 124 5.6 Hull diagram and a lattice parameter for FReO 3 . . . . . . . . . . . 127 5.7 Calculated and experimental X-ray absorption spectra of F x ReO 3 . 130 5.8 Heatmap of operando X-ray diffraction of F x ReO 3 . . . . . . . . . . 135 5.9 Entire 2θ range of operando X-ray diffraction of F x ReO 3 . . . . . . 136 5.10 X-ray diffraction before and after application of TBAF to ReO 3 . . 136 5.11 Low angle operando X-ray diffraction of F x ReO 3 . . . . . . . . . . . 137 5.12 Operando X-ray diffraction of F x ReO 3 (0≤x≤ 1 ) . . . . . . . . . 137 5.13 Simulated X-ray diffraction patterns of ReO 3 and FReO 3 . . . . . . 138 5.14 Impedance spectroscopy and diffusion coefficients of F x ReO 3 . . . . 139 5.15 Electrochemical mass spectrometry of CO 2 in F x ReO 3 . . . . . . . . 140 5.16 Solid-state 1D 19 F echo MAS NMR of F 0.2 ReO 3 . . . . . . . . . . . 140 5.17 19 F Hahn-echo MAS NMR saturation-recovery of F 0.2 ReO 3 . . . . . 141 5.18 Plot of δ calc iso vs δ exp iso for fluorine-containing compounds . . . . . . . . 142 5.19 Simulated X-ray emission and absorption for 7-coordinate FReO 3 . 144 5.20 Simulated X-ray emission and absorption for mono-ReO 3 F . . . . . 145 5.21 Simulated X-ray emission and absorption for FReO 3 . . . . . . . . . 145 5.22 Simulated X-ray emission and absorption for tet-ReO 3 F . . . . . . . 146 5.23 Re 4f X-ray photoemmission of F x ReO 3 . . . . . . . . . . . . . . . . 147 5.24 Pair distribution function analysis of ReO 3 and F 0.6 ReO 3 . . . . . . 148 5.25 Simulated pair distribution function of 7-coordinate FReO 3 . . . . 149 5.26 Simulated pair distribution function of mono-ReO 3 F . . . . . . . . 149 5.27 Simulated pair distribution function of tet-ReO 3 F . . . . . . . . . . 150 5.28 Simulated pair distribution function of FReO 3 . . . . . . . . . . . . 150 ix Abstract Understanding the structural transformations that materials undergo during the insertion and deinsertion of ions is crucial for designing high performance inter- calation electrode materials. In this dissertation, a series of studies investigates the role of structural changes on the electrochemical performance of intercalation electrodes and identifies common themes between crystallographic motifs and elec- trochemical behavior. I present a study of the structural distortions of the metallic defect perovskite ReO 3 upon lithiation with the goal of determining whether these distortions are driven by polaronic charge transport (i.e. the electrons and ions moving through the lattice in a coupled way) due to the semiconducting nature of most oxide hosts. I find that the cubic structure of ReO 3 experiences multiple phase changes involving the correlated twisting of rigid octahedral subunits during the insertion of two equivalents of Li-ions; this suggests that phase transforma- tions during alkali ion intercalation are the result of local strains in the lattice and not exclusively due to polaron migration. The cubic perovskite structure was further investigated by examining the isostructural NbO 2 F and its shear derivative Nb 3 O 7 F to probe the role of edge-sharing octahedral planes in suppressing phase changes. These planes modulate structural stability and ionic transport pathways and therefore play an intimate role in the mechanism of ion insertion and cycling x performance of shear compounds. Crystallographic shear led to increased struc- tural stability during Li + (de)intercalation with edge-sharing layers being main- tained, while corner-sharing linkages were seen to degrade rapidly; however, disor- dering in the shear plane stacking introduced by strain during cycling ultimately led to poor capacity retention. In order to reduce lattice strain during cycling, the defect garnets, Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 , with redox active Mo in the tetrahe- dral site were compared. The importance of metal migration during multi-electron redox activity was identified, revealing a competing demand to satisfy bonding requirements and local strains in structures upon alkali intercalation. The local structure of Y 2 (MoO 4 ) 3 was maintained upon lithiation while Al 2 (MoO 4 ) 3 under- went substantial local atomic rearrangements as the stronger ionic character of the bonds in Al 2 (MoO 4 ) 3 allowed Al to migrate from its starting octahedral position to accomodate strain during cycling. However, this mixing was prevented in the more covalent Y 2 (MoO 4 ) 3 which could only accommodate this strain through rota- tional motion of the polyhedral subunits. Lastly, anion intercalation was probed through the electrochemical intercalation of F-ions at room temperature from a liquid electrolyte showing that slightly more than 0.5 equivalents of F − ions can be inserted onto the vacant A-site of the perovskite ReO 3 . This caused migration of Re ions out of the octahedral interstitial and into a newly created tetrahedrally coordinated ReO 3 F site and demonstrated that under the right conditions small anions like fluoride can diffuse through solids as readily as alkali cations, opening new opportunities to oxidatively fluorinate a wide range of new materials. The role of structural change during ion intercalation was characterized with a range of techniques including galvanostatic and potentiometric methods, operando X-ray diffraction and absorption, X-ray photoelectron spectroscopy, nuclear magnetic resonance, and density functional theory calculations. xi Chapter 1 Introduction 1.1 The Energy Storage Crisis 2000 2004 2008 2012 2016 Year 0 20 40 60 80 100 Energy Generation Capacity (GW) Solar Wind Figure 1.1: The use of wind and solar energy has increased rapidly in the United States as costs have fallen, leading to large increases in installed capacity. 1 One might look upon the cobalt mines of the Democratic Repub- lic of the Congo and see a bar- ren wasteland of strip mines, awash with child labor and the devas- tating effects of human exploita- tion. Thecobaltindustryhashada poor track record, with large multi- national mining corporations and foreign government representatives taking advantage of some of the poorest people on earth. 2 Mean- while, Western governments have largely remained on the sidelines. 3 Yet behind this is an economic revolution designed to improve the world for all of humanity by creating a future devoid of the devastating effects of climate change caused by the fossil fuel industry, of freeing the earth from the negative conse- quences of global warming. While the importance of these mines may be difficult to ascertain, upon further consideration one can draw remarkable parallels to the early days of the oil industry of Saudi Arabia, and the greater Middle East energy 1 economy. What once was a desert wasteland was transformed into an economic powerhouse by the ever-present need for the black gold which drove the energy industry of the twentieth century. However, the twenty-first century will represent a new future, one in which the price of oil does not drive swings in the economy but one in which the means of energy storage represents the great engine of a global economy. 4 This is where the dusty mines of one of the poorest countries on earth will prove to be the centerpiece of a future global energy market. 5,6 2012 2014 2016 2018 Year 0 1000 2000 3000 Electric Vehicles Usage (in 1,000s) Figure 1.2: The number of electric vehi- cles in operation globally has risen sharply in recent years and is expected to continue to grow. 7 A recent report by The Inter- governmental Panel on Climate Change says that a reduction in carbon output of forty-five per- cent below 2010 levels by 2030 is needed to avoid catastrophic con- sequences due to global warming. 8 There exists the means to rapidly decarbonize the energy production and transportation sectors, free- ing humanity from the surly bonds of fossil fuels and ushering in a future of cheap, clean, and renew- able energy produced by the likes of wind and solar, allowing continued human development without the impact of climate change. 9 The shift towards clean energy is growing economically feasible, as can be visualized in recent reports by Lazard and the UT Austin Energy Insti- tute, as well as the installed capacity of solar and wind energy, shown in Figure 1.1. 10 Due to advances in solar panel technology and falling installation costs, the 2 levelized cost of wind and solar electricity is now cheaper than coal and natural gas for electricity production in the vast majority of new plant installations. 11 Indeed, a recent study found that renewable sources were cheaper than conventional energy sources in a majority of counties in the United States. 12 In the transportation sec- tor, advances by manufacturers such as Tesla are creating the cars of the future, sleek sporty designs that run entirely on battery power. The automotive industry is increasingly turning towards electric vehicles as the future, evidenced by finan- cial motivation such as a $500 million investments by Ford into Rivian to produce electric trucks and commitments by VW to convert to an entirely electric fleet, leading to a rapid rise in the number of electric vehicles in operation as shown in Figure 1.2. 13,14 The rapid race to clean electricity and transportation, seemingly destined to continue unhindered due to increasing consumer demand and regula- tory pressure, nevertheless hinges on a humble purplish gray rock dug out of the ground in sub-Saharan Africa. 15 In 1991, Sony changed the battery market forever when they released the first lithium ion battery. This battery, worlds ahead of existing technology, relied on the use of a LiCoO 2 cathode, in conjunction with a graphite anode. 16 This cru- cial development provided the high voltages and increased energy densities nec- essary for modern electronics. Fast forward 30 years and the same basic tech- nology is now used for everything from smartphones to the Chevy Volt. While the original LiCoO 2 cathode employed by Sony has been largely replaced by LiNi 1−x−y Mn x Co y O 2 , or NMC as it is known in the industry, there still exists a massive demand for cobalt. 17 In 2018 alone, over 14,000 tons of the metal went to the battery industry, accounting for more than twenty-five percent of the global market. 18,19 While continued improvements, engineering breakthrough, and manu- facturing advances have vastly improved the original lithium ion battery, cells are 3 quickly approaching theoretical capacity. 20 Furthermore, demands for lifespan and durability keep increasing, as well as growing interest in fast charging batteries. 21 Efforts such as the Department of Energy Battery500 research consortium have set ambitious targets of developing cells with a capacity of 500 Wh kg −1 over 1,000 cycles, at a cost below US$100 kWh −1 . 22 Figure 1.3 shows just how rapidly battery pack costs have fallen, yet demand for batteries continues to rise. The adoption of green energy unequivocally depends on commercially viable rechargeable batteries. There is simply no other technology that can provide the scalability, portability and flexibility of lithium-ion batteries. Detractors of solar power often point to the problem of the night. That is, how can one rely on solar panels for electricity when the sun stops shining? The easy answer is lithium- ion batteries – massive banks of them the size of warehouses capable of powering cities for hours. 23 The hard part of that answer is creating batteries that meet an increasingly high performance bar and finding the minerals that go into those batteries. 24 This challenge demands a multi-pronged approach, with an increased focus by governments on securing the necessary minerals in conjunction with con- tinued and focused research on developing advanced battery materials. If governments wish to adopt climate friendly policies, they must begin work immediately to secure sources of the minerals necessary for lithium ion batteries, none of which is more important, yet more scarce, than cobalt. With over fifty percent of mining in 2018, the vast majority of cobalt is sourced from DR Congo and then processed in China. 27 Lithium ion batteries have the potential to be the basis of a clean, renewable energy future, yet without a focus on the necessary minerals they could easily prove to be the insurmountable hurdle to solving climate change. Western governments must focus on developing a presence in this region to ensure a stable supply. Without it, lithium-ion batteries threaten to become 4 prohibitivelyexpensiveorundulyscarce. Thisoft-overlookedtechnologyunderpins the future of the clean energy sector, providing everything from the grid scale storage of electricity produced by wind plants to the power source of engines in electric vehicles. 2010 2013 2016 2019 Year 0 200 400 600 800 1000 Battery Cost (USD kWH �1 ) 2018 2020 2025 2030 0 500 1000 1500 2000 2500 3000 Battery Demand (GWh) Consumer Electronics Energy Storage Electric Vehicles (a) (b) Figure 1.3: In (a), forecasted demand for rechargeable battery capacity over the next ten years is expected to rise, while (b) shows the drop in price for battery packs over the last decade. 25,26 The second part of the solu- tion depends on the discovery of new battery chemistries in order to meet the ever-growing demand placed on rechargeable battery sys- tems. While many advances have been made in recent years which have enabled enhancements in battery performance including increased gravimetric energy den- sity and improved power capabil- ities, current devices do not meet the ever growing demands placed upon batteries. 28 Materials science provides the means to systemat- ically develop and innovate new electrode materials to provide the performance required for a range of applications. 29 By creating a fun- damental understanding of the process by which ionic insertion reactions occur and the effect that these reactions have upon materials, it is possible to relate 5 structural features of inorganic materials to electrochemical properties and formu- late design rules. Creation of these rules allows for rational materials design to be employed in formulating new battery electrodes, while advances in computational methodology and high throughput testing will allow for rapid development. 30 By combining a government and industry focus to secure the necessary raw materials to create the energy systems of the future with simultaneous materials chemistry research to increase the variety of compounds used in rechargeable bat- teries and allow for more efficient mineral usage, the continued shift towards green energy can flourish. In order to usher in the future of clean energy, fundamental researchisrequiredtodevelopthebatteriesthatcanmeetevergreaterperformance metrics and expanding applications. At the same time, government and industry focus is required to secure the necessary raw materials to create the energy systems of the future. Ensuring that the requisite materials, such as cobalt, are responsibly and sustainably sourced will open up vast new markets in countries like DR Congo as battery adaptation continues. 1.2 Overview of Intercalation Batteries The movement of ions through the solid state is of great importance to numerous technological processes and has been the focus of immense study throughout the materials chemistry community. 31 Among the myriad applications enabled by ionic movement, perhaps none is more important than the reversible storage of energy in the modern rechargeable battery. 32 Responsible for enabling the advance of portable electronics, the development of clean transportation networks, and the large scale storage of renewable energy, the humble lithium ion battery has prolif- erated throughout the modern world culminating in the award of a Nobel prize in 6 chemistry to Wittingham, Goodenough, and Yoshino in 2019 for the development of the lithium-ion battery. 33 The general design of a battery cell consists of an anode and cathode separated by an electrically insulating barrier that contains an ionically conducting elec- trolyte. 34 Typically, lithium-ion batteries cathodes consist of a crystalline oxide or polyanionic active material such as LiNi 1−x−y Mn x Co y O 2 or LiFePO 4 blended with conductive carbon and a polymer binder and adhered to a metal foil current collector. 35,36 In commercial cells, the anode is commonly composed of graphitic carbons, again blended with a polymer binder; however, experimental research cells often make use of Li metal anodes in order to simplify cell design. The anode and cathode are separated by a barrier that can be a polymer film or glass fiber pad. This separator is soaked in a liquid electrolyte that is made by dissolving a Li salt, such as LiPF 6 , in ethereal or carbonate solvents. 37 Figure 1.4: Diagram of a general electrochemical cell design for a secondary Li- ion battery based on a LiCoO 2 cathode, a graphite anode, and a polymer separator film soaked with LiPF 6 electrolyte. During operation, electrons flow through an external circuit while Li + ions (yellow spheres) flow between the electrodes. 7 A prototypical commercial lithium ion battery is demonstrated in Figure 1.4 with a LiCoO 2 cathode, a graphite anode, and a polypropylene separator film soaked in an electrolyte of LiPF 6 dissolved in a blend of ethylene carbonate and dimethyl carbonate. 38,39 In order to charge or discharge the battery, known as cycling, Li + ions migrate between the cathode and anode while a commensurate electron flows through an external circuit to complete a redox reaction. To charge the battery, a voltage or current is used to force electrons from the lower energy cathode to the higher energy cathode, while Li + simultaneously moves from the cathode to the anode in order to balance the negative charge of the electron. The reverse of this process occurs during discharge when high energy electrons move through an external circuit from anode to cathode, performing work, while Li + moves back into the cathode to complete the charge balance. 40 The capacity of a battery electrode material, typically measured in units of charge per unit mass or volume, such as mAh g −1 , is determined by the number of ions that can be reversibly removed and reinserted into a host material. This process must occur with a corresponding redox reaction in which a redox active center changes oxidation state and releases or accepts electrons to create a flow of current through an external circuit. The removal of ions followed by the re- insertion of ions defines a complete charge-discharge cycle; using the prototypical LiCoO 2 electrode this can be represented by the following reactions. charge LiCo 3+ O 2 −−→Co 4+ O 2 +Li + +e − discharge Co 4+ O 2 +Li + +e − −−→LiCo 3+ O 2 The cycle life of the battery is determined by the number of these reactions that occur reversibly and can be influenced by a number of factors which are based 8 on the intercalation reactions that occur during battery operation. While a more exhaustive description of the principles of battery design is beyond the scope of this document, the reader is referred to an excellent review of the fundamentals of battery chemistry by Winter and Brodd. 41 1.3 Intercalation Reaction Mechanisms There are multiple mechanisms by which ions can be stored within electrode mate- rials, including alloying reactions such as the formation of LiAl alloys, 42 conversion reactions such as the conversion of Fe 2 O 3 to Fe metal and Li 2 O, 43 and intercalation reactions such as the insertion of Li into ReO 3 . 44 While all three reaction classes are of interest, this manuscript will focus primarily on intercalation based electrode materials and the dynamics of the intercalation reaction. 45 An intercalation reaction describes a process by which a guest species is reversibly inserted into a vacancy within a host material with minimal associ- ated structural change. 46 In battery relevant reactions, the intercalant is a charge carrying ion, such as Li + , which occupies an empty crystallographic site within the host framework. This insertion of a charged species is accompanied with a charge transfer process between the guest species and the host and an associated change in oxidation state of the host species. Crucially, this reaction is reversible, meaning that the inserted guest species can be deintercalated and the host restored to its original state. These reactions can occur by chemical or electrochemical means using either chemical forces or electrical fields to drive the (de)intercalation of guest ions. 47 As these reactions are reversible, host materials may be synthesized with or without a guest ion present and the concentration of guest ions may be varied. 9 As intercalation reactions proceed, redox centers change oxidation state to accommodate guest ions; for example the insertion of Li + into CoO 2 operates on the Co 3+/4+ redox couple. These changes in oxidation state lead to changing bond lengthsandelectrostaticenvironments, whichinturncaninducestructuralchanges in materials. 48 Furthermore, the insertion or removal of charged ions, which must satisfy their own bonding requirements, can also cause structural changes depend- ing on the size and charge of the ion. Additionally, the insertion or removal of electrons, as coupled to guest ions, changes the occupancy of electronic states which changes properties such as conductivity as a function of intercalant con- centration. 49 This can have drastic implications on the performance of electrode materials; for example, the insertion of Li + into the intercalation host PNb 9 O 25 drives a Mott insulator-metal transition. 50 In the pristine state, PNb 9 O 25 is a white, insulating material, however Li + doping increases the electron density on the Nb in the structure which leads to semi-conducting behavior and eventually, at high Li + concentrations, metallic behavior. In addition to electronic changes induced by doping or removing electrons from a material, the movement of guest ions can lead to structural distortions and phase changes in intercalation hosts. These changes have marked effects on the performance of battery electrodes thus understanding and controlling them is of great importance. For example, a large amount of effort was dedicated to deter- mining the mechanism of delithiation in olivine-type LiFePO 4 in order to deduce whether a two-phase or solid solution type phase change was occurring in going from the lithiated to the delithiated phase. 51 As shown in Figure 1.5, the removal of Li from the structure leads to a change in unit cell volume and a small struc- tural rearrangement. 52,53 This lead to the identification of a solid-solution reaction 10 Figure1.5: UponLi + removal, LiFePO 4 contractsbyroughly6.8%withaconcor- dant rotation of polyanionic PO 4 groups. This reaction can occur via a two-phase or solid solution process depending on materials preparation methods; this leads to differing rate capabilities during cycling. type when sufficiently small LiFePO 4 nanoparticles were used, leading to improved high-rate cycling performance. 54 Since the (de)intercalation process can affect both the physical and electronic structure of electrode host materials, understanding and controlling intercalation reactions is extremely important to improving battery metrics and affects many aspects of performance. The degree to which the (de)intercalation reactions can occur reversibly is of key importance for determining the cycle life on an electrode material. For example, the structural changes that occur during intercalation may lead to particle fracturing and loss of electrical contact. Structural changes may also reduce the crystallinity of materials and destroy cystallographic sites for inter- calation, reducing electrode capacity or trapping guest ions within a structure. If the reaction kinetics are sluggish, electrodes are more likely to have poor rate per- formance, limiting the maximum power a battery is capable of supplying. 55 Fur- thermore, the intercalation reaction energy itself determines the operating voltage of an electrode material and therefore influences the energy density of a battery. 56 11 Generally, late transition metals exhibit higher redox voltages than the early tran- sition metals. By developing a fundamental understanding of the crystallographic structure features affecting battery performance, it is possible to create intelligent design rules to produce desired electrode properties. Materials can be optimized to reduce structural distortions that occur upon ion insertion or to maximize the rate at which the reaction can proceed. Chemical tuning can prevent undesired phase changes to improve reversibility and cycle life. Relating the structure and proper- ties of electrode materials allows for the creation of the next generation of battery materials. 1.4 Anion Intercalation In recent years, much focus had been placed on the intercalation of cations, specifi- cally Li + , for battery relevant reactions. However, anions may also be intercalated into solids, and some of the earliest work on intercalation reactions studied the insertion of bulky anionic groups such as SO 2− 4 into graphite. 57 By expanding the chemical toolbox to include both anions and cations, the battery chemist has a much larger set of possible materials to explore as potential electrodes. 58 While the intercalation of anions offers intriguing possibilities, there exists many challenges in electrochemical cell design and a need for better fundamental understanding of the movement of anions in the solid state. Much of the work surrounding batteries in which the charge carrier is an anion have focused on two areas. The first is fluoride shuttle batteries, where fluoride ions reversibly plate and strip from a metal electrode. 59 For example, Reddy and Fichtner demonstrated a prototypical cell with a CuF 2 cathode, a Bi metal anode, 12 and a Ba-doped LaF 3 solid electrolyte. 60 During cycling, the CuF 2 is reversibly reduced to Cu and the Bi is oxidized to BiF 3 . Fluoride ions move through a solid electrolyte, however to increase ionic conductivity, the electrolyte must be heated to around 150 ◦ C which reduces the practical applications of such cells. 61 Similar cells have been demonstrated with various metal and metal fluoride combinations, including CeF 3 , CaF 2 , MgF 2 , Mg, Pb, Zn, and Bi. 62 These batteries, which make use of conversion type electrodes, typically have poor performance including short cycle lives and low practical capacity. The second area of focus is symmetric dual ion batteries where an electrolyte salt is dissociated with simultaneous insertion of either cations or anions into both anode and cathode. Often, these are identical graphite electrodes which serve as intercalation hosts for the reversible uptake of ions. 63 A variety of electrolytes have been studied, although typical compositions consist of an alkali cation, such as Li + or Na + , paired with a bulky polyanionic anion such as ClO − 4 , PF − 6 or AsF − 6 . Figure 1.6: General design of a fluoride ion battery cell with a La 2 CoO 4 F cathode serving as a fluoride source. Fluoride ions, shown in green, move through a Ba- doped LaF 3 solid electrolyte, which requires elevated temperatures to operate. A composite anode of Pb blended with PbF 2 completes the cell. 13 During charge, cations are reductively inserted into the negative electrode while anions are oxidatively inserted into the positive electrode. During discharge, the reverse occurs and ions are removed from both electrodes and recombine in the electrolyte solvent. 64 The high voltages required to oxidatively insert anions into the graphite electrode can lead to a number of electrolyte degradation reactions and deleterious side reactions. 65 Quite recently, several developments have renewed interest in fluoride ion bat- teries. In 2018, Davis et al. demonstrated a room temperature liquid fluoride ion electrolyte with greatly enhanced stability and a large electrochemical win- dow. 66 By making use of a branched quaternary amine fluoride salt in conjunc- tion with a fluoroether solvent, they were able to demonstrate reversible elec- trochemical cycling with good performance. This opened up the possibility of a room-temperature fluoride-ion battery based on a liquid electrolyte. Additionally, Nowroozi et al. reported the intercalation of F − into La 2 CoO 4 electrodes, albeit at elevated temperature, using a solid electrolyte. 67 A schematic of a fluoride ion battery cell employing a solid electrolyte is shown in Figure 1.6. Intercalation reactions are known to offer improved battery performance as less atomic rear- rangement is required during cycling, leading to improved rate performance and better long-term cycle life. As will be explored in chapter five, research is needed to understand the mech- anism of anion intercalation into dense, crystalline solids. As would be expected, this process is very different from cation insertion due to the negative charge and differing size of the intercalated ions. By combining various structural probes, as well as computational methodology it is possible to develop design rules for the process of anion intercalation. These rules clarify how anions are intercalated into solids, as well as the bonding requirements that they must satisfy for stability. 14 These rules allow for the design of potential anion intercalation hosts with targeted performance. In addition to offering an alternative to cation intercalation type electrodes, an exciting proposition is the coupling of cation intercalation with anion intercalation to create multi-ion electrode materials with greatly enhanced capacity. For exam- ple, a cathode could deintercalate Li + ions during oxidative charging. The cathode could then be further oxidized with the insertion of F − ions leading to additional charge storage. Ideas such as these require further research into the mechanisms of anion intercalation in order to properly design electrode materials. 1.5 Operando Battery Experiments Throughout this dissertation, structural changes caused by the (de)intercalation of ions into electrode materials are investigated to better correlate electrochemi- cal behavior with structural evolution. Methods such as X-ray diffraction, X-ray absorption and emission spectroscopy, Raman spectroscopy, and neutron scatter- ing yield valuable insight into changes in electrodes during use and provide atomic level detail of the intercalation process. However, the procedure for preparing electrodes for analysis typically involves disassembling cells, washing electrodes with an organic solvent, and often several days of wait time before measurement. Furthermore, in many cases it can be difficult to remove all trace moisture and oxygen during measurement. Many of the samples themselves are highly reactive or kinetically unstable products prone to degradation. Therefore it is challenging to truly assess if the measured sample is representative of the material within a battery. 68 15 Figure 1.7: An electrochemical cell is equipped with a Be window allowing for the collection of X-ray diffraction patterns during battery operation. In order to avoid this problem, operando measurements can be employed to allow for direct correlation of electrochemical behavior with structural changes. operando measurements make use of specially equipped electrochemical cells with windows to allow for measurement during operation. For example, operando X-ray diffraction may be performed using a cell with a beryllium current collector and window, as shown in Figure 1.7. 69,70 A beryllium disc is placed within a specially designed battery casing and an electrode is placed directly on the disc. The beryl- lium is conductive, allowing for current to flow to the electrode; the beryllium is also transparent to X-rays allowing for diffraction patterns to be collected on the electrode active materials. By slowly discharging the cell over a period of several hours and collecting X-ray diffraction experiments every few minutes, it is possible to create a detailed picture of structural changes in the material which is directly correlated with the electrochemical behavior. Additionally, this method can cap- ture intermediate or reactive phases that would not otherwise be seen by ex situ measurement. 71 16 Similar measurement methods can be used in conjunction with synchrotron X-ray sources, such as those available at the Advanced Photon Source at Argonne NationalLab. Cellsaredesignedwithglassycarbonwindowsdesignedtooperatein transmission mode, such that the X-ray beam passes directly through the cell. Due to the highly focused X-ray beam, a much smaller window can be used for these experiments. Throughout this dissertation, operando measurement methods are employed which have allowed for the detailed visualization of reaction processes and greater materials understanding. These measurement types help to ensure consistent and reliable results and better correlation of electrochemical behavior with structural changes. 1.6 Relevant Crystal Structures While the list of structure types which have been explored for use as battery electrodes is vast, this document will only examine a few structure types, includ- ing the ReO 3 perovskite structure and its derivative shear phases, as well as the defect garnet structure type. All of these structures have some commonality with highly crystalline ordering creating multi-dimensional channels for ion diffusion and vacant crystallographic sites for ion intercalation. As will be demonstrated in chapters two and five, this dissertation made exten- sive use of ReO 3 as a model electrode material. While the high cost of ReO 3 prohibits its use in commercial batteries, it is an excellent model system for study- ing intercalation reactions. ReO 3 is one of the few metallic oxides, allowing for battery electrodes to be prepared without carbon additives. By doing this, it is possible to perform various spectroscopic studies on ex situ samples without interference from carbon additives. ReO 3 crystallizes in a perovskite structure in 17 space groupPm3 ¯ m, with a vacantA-site, assuming the generic perovskite formula of ABO 3 . 72 It is composed of corner-sharing ReO 6 octahedra that form a three- dimensional cubic network with no tilting of the octahedra. The vacant A-site creates a dodecahedral hole that allows for ion insertion. Another A-site vacant perovskite, NbO 2 F is studied in chapter three. This material crystallizes in the same Pm3 ¯ m space group and is composed of corner sharing NbO 4 F 2 octahedra with O and F randomly distributed across a single site. 73 A derivative phase, Nb 3 O 7 F can be formed by the application of crystallo- graphic shear to the NbO 2 F structure. This occurs when a single plane of anions is removed from the structure, leading to a collapse of two layers of corner-sharing octahedra into a plane of edge-sharing octahedra. As such, Nb 3 O 7 F consists of alternating edge-sharing and corner-sharing Nb octahedra with oxygen and fluo- rine randomly distributed. More complex crystallographic structures can be formed by the inclusion of a tetrahedrally coordinated species, as is the case in the defect garnet, A 2 (MoO 4 ) 3 (A = Y, Mo). This structure is formed of a highly symmetric network of corner- sharingMoO 4 tetrahedronandAO 6 octahedrawithvacantdodecahedralsites. The structure can be visualized as a network of octahedral rods wrapped in tetrahedral subunits. 74 1.7 Dissertation Overview This document presents a set of complimentary studies that seek to describe the structural transformations which occur upon the electrochemical intercalation of ions into dense crystalline materials and to relate these transformations to struc- tural features present in the materials. Through systematic variation of chemical 18 Figure 1.8: (a): ReO 3 is a cubic perovskite composed of ReO 6 octahedra with a vacant A-site. As shown in (b), Nb 3 O 7 F can be formed by shearing the ReO 3 structure to create edge-sharing sheets. (c): the complex defect-garnet structure contains corner-sharing MoO 4 tetrahedra and AO 6 (A = Al, Y) octahedra. composition and structure along with concurrent electrochemical studies, it is pos- sible to create design rules for electrode materials which allow for the optimization of properties such as cycle life or rate performance. Through a set of fundamen- tal studies on the properties of electrode materials, greater understanding may be drawn about the systems of interest. Thefirststudypresentedexaminestheroleofpolaronsinmodulatingstructural distortions upon ion insertion by using the model electrode material, ReO 3 . ReO 3 is a metallic oxide that can accommodate up to two units of Li + insertion with corresponding reduction of Re from the +6 to +4 oxidation states. The highly conductive nature of this material prevents the formation of polarons during the intercalation process, however this did not prevent distortions of the structure. As Li + was inserted into the A-site vacancy within the perovskite structure, an octahedral rotation was observed, leading to a compression of the structure. Using operando X-ray diffraction, along with a number of complimentary techniques, it was rationalized that the structural distortions were necessitated by a need for bonding requirements of the Li + to be satisfied. These structural distortions led to poor reversible cycling despite the metallic character of ReO 3 . 19 In the next chapter, the role of crystallographic shear was investigated to deter- mine if it could stabilize the perovskite structure to Li + insertion by using a series of niobium oxyfluoride compounds. NbO 2 F, which crystallizes in the same A-site vacant perovskite structure as ReO 3 , was lithiated and monitored via operando X- ray diffraction and X-ray absorption spectroscopy. It was found that the material distorted in a similar manner as ReO 3 , leading to poor reversibility. Next, the shear derivative, Nb 3 O 7 F, was probed using the same methods. This structure features a collapse of the perovskite structure along one axis, forming edge sharing octahedral units. It was found that the edge sharing units prevented distortions in the layers, although the interlayer perovskite linkages were still free to distort. This led to the formation of stacking faults over extended cycling and ultimately reduced cycling performance. Interestingly, one-dimensional crystallographic shear did not improve electrochemical performance. The fourth chapter of this dissertation explores the effect of shifting redox activity from octahedrally coordinated metal centers to tetrahedrally coordinated metal centers by exploring the defect garnets, Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 . These structures were deliberately chosen with redox inactive elements on the octahedral site, forcing redox activity to occur on the tetrahedrally coordinated molybdenum. It was seen that reduction on the tetrahedral site is accommodated by different mechanisms, depending on the degree of covalency in the bonding in the struc- ture. In the more ionic Al 2 (MoO 4 ) 3 , lithiation is seen to lead to migration of Al from the octahedral site to a newly created tetrahedral site, while Mo migrates to the open octahedral site. However, the more covalent bonding in Y 2 (MoO 4 ) 3 allows the structure to accommodate lithiation through a rocking of the octahedral units. This demonstrates the importance of spectator elements in modulating the behavior of intercalation electrode materials. 20 In the last chapter, the intercalation of anions is investigated through the insertion of fluoride anions into ReO 3 . Due to the differing charge and size of fluoride, a drastically different intercalation process is identified, as compared to the intercalation of Li + into ReO 3 . The highly electronegative fluoride inserts into A-site vacancies, however, in order to satisfy bonding requirements, Re migrates from its octahedral coordination to a newly created tetrahedral site. This mecha- nism is reminiscent of that which was observed when Mo was seen to migrate in Al 2 (MoO 4 ) 3 , again emphasizing the role of metal migration as an alternative to polyhedral rotation as a charge compensation mechanism. This work uses a series of studies to investigate different aspects of structure and bonding in electrode materials in order to better understand the relationship between structure and properties in intercalation reactions. A number of funda- mentaldesignrulesareidentifiedwhichcanbeusedtointelligentlychooseelectrode compositions based on performance metrics. 21 Chapter 2 Correlated Polyhedral Rotations in the Absence of Polarons during Electrochemical Insertion of Lithium in ReO 3 2.1 Introduction The reversible (de)insertion of Li-ions into densely-packed intercalation hosts is a complex process of fundamental importance to rechargeable batteries. As posi- tively charged ions move in and out of a structure, redox-active transition metal centers change their formal oxidation state and, in the process, adjust their bond lengths so as to maintain local charge neutrality. 75,76 These complex structural distortions generate substantial strain in the lattice that manifests itself as large changes to the unit-cell volume during cycling, which can result in cracking of the electrode and delamination from the current collector; ultimately shortening the life of the battery. 77–79 Developing a deeper understanding of these structural transformationsandhowtheyinfluencechargetransportisthereforeacriticalopen question for designing new intercalation hosts. Our group has demonstrated that the presence of rigid subunits within a struc- ture necessitates the ability to undergo highly correlated polyhedral rotations in order to relieve the strain generated when ions are incorporated into the lattice. 75 Contemporaneously, Banerjee and coworkers performed a detailed spectroscopic study on Li intercalation into V 2 O 5 and identified a correlation between changes in the metal’s first coordination shell and charge hopping through the lattice, 22 22 Adapted from Bashian et al. ACS Enery Lett. 2018 3 2513-2519. [doi] 22 suggesting the possibility that the rotations we had reported could be driven by polaronic charge transport. 80,81 Intrigued, we initiated a study of theA-site vacant perovskite ReO 3 , one of only a handful of metallic oxides that are known. The topology of the structure, illustrated in Figure 2.1, contains a perfectly cubic net- workofcorner-sharingoctahedraofReO 6 with90 ◦ and180 ◦ O–Re–OandRe–O–Re bond angles respectively, creating a three-dimensional network of interstitials that Cava, Murphy, and coworkers reported can accommodate up to two Li-ions per formula units (i.e. Li 2 ReO 3 ) when treated with n-BuLi. 82,83 Figure 2.1: Cubic crystal structure of ReO 3 with oxygen and rhenium ions shown in orange and red respectively. The coexistence of intrinsic metallicconductivityandpathways for facile ionic diffusion make ReO 3 a model system for exploring cor- relations between the transport of ions and electrons. While the open framework of ReO 3 has been heav- ily investigated in the past for its negative thermal expansion prop- erties, 84,85 to our knowledge, there have been no studies that examine the role of structural flexibility on the performance of ReO 3 as an intercalation electrode in functioning Li-ion bat- teries. In this chapter, we characterize the electrochemical performance using operando synchrotron X-ray diffraction (XRD), density functional theory (DFT) calculations, and X-ray spectroscopy to better understand the fundamental mech- anism for the (de)insertion of ions into the ReO 3 . Although the metallic character 23 of the starting phase might be expected to minimize lattice deformations result- ing from polaron migration, we observe pronounced rotations of the Re octahedra during the (de)insertion of Li. These tilting modes have a marked effect on the electronic structure of the material, eventually opening a small band gap in the fully lithiated Li 2 ReO 3 end member. These distortions generate sufficient strain within the lattice to severely limit the reversible capacity over extended cycling, clearly demonstrating the importance of local structural distortions such as these for developing high performance intercalation hosts. 2.2 Experimental Details Synthetic Methods. Nanoscale particles of ReO 3 were prepared following the method reported by Chong et al. 86 In a typical preparation, 0.05 mmol of Re 2 O 7 was dissolved in 0.5 ml of methanol within a 10 ml round bottom flask and main- tained at 250 ◦ C for 5 minutes, allowing the methanol to completely evaporate. The reduction of Re 2 O 7 produced ReO 3 , which was deposited as a shiny brick red film on the walls of the round bottom flask. The film was collected and ground via mortar and pestle to yield a deep red powder that exhbited a metallic luster. The powders were dried under vacuum before being introduced to an argon-filled glovebox. Physical Characterization. Operando XRD patterns were collected at the Advanced Photon Source (APS), Argonne National Laboratory using the AMPIX electrochemical cell, following the method detailed by Borkiewicz et al. 87 High res- olution synchrotron powder diffraction data was collected using beamline 17-BM at the Advanced Photon Source (APS), Argonne National Laboratory using an average wavelength of 0.24116Å, with a Perkin-Elmer 2D plate detector. GSAS-II 24 software was used to integrate patterns into the intensity vs 2θ format displayed. 88 The resulting diffraction patterns were refined against the cubic perovskite struc- tureusingtheRietveldmethod 89 asimplementedintheFullProfprogram. 90 Trans- mission Electron Microscopy (TEM) images were collected on a JEOL JEM2100F with an operating voltage of 200 kV. Samples for TEM analysis were prepared by drop-casting a suspension of the powder in ethanol onto a 200 mesh Cu grid coated with a lacy carbon film. X-ray photoeoelectron spectroscopy (XPS) of ReO 3 reference powder and the discharged Li x ReO 3 and Li 2 ReO 3 powder electrodes was performed at the Ana- lytical and Diagnostics Laboratory (ADL), Binghamton University using a Phi VersaProbe 5000 system with a monochromated Al Kα source and a hemispher- ical analyzer. All samples were mounted on Ta foil using conductive tape while inside an Ar glovebox. Using a vacuum suitcase, samples were transported into the XPS chamber without air exposure. The core levels (Li 1s, Re 4f, O 1s, and C 1s) and the valence-band region were measured with a pass energy of 23.5 eV, corresponding to an instrumental resolution of 0.51 eV from analysis of both the Au 4f 7/2 and Fermi edge of the gold foil. Gold foil (in contact with samples dur- ing measurement) was used in binding energy calibration. Measurement time for Li 2 ReO 3 valence band region were increased by a factor of 2 to account for weaker signals due to attenuating surface species and its insulating character. Structure and charge density visualization and analysis were performed using the VESTA 91 software package. Electrochemical Characterization The electrochemical performance of the as-prepared materials was characterized using Swagelok-type cells assembled in an argon-filled glovebox, using Li metal as a combined counter and reference electrode 25 and Whatman GF/D borosilicate glass fiber sheets as the separator. 1M LiPF 6 in ethylene carbonate and dimethylcarbonate (1:1 w/w) was used as the electrolyte (LP40). Powders of ReO 3 were dried under vacuum to remove any residual solvent remaining from synthesis. Cyclic voltammetry was performed with a three-electrode configuration inside anargon-filledgloveboxusingatestcellwithsufficientelectrolytetofullysubmerge the active area of the electrodes. In brief, a suspension of ReO 3 particles was prepared by adding 2mg of powder to 1mL of ethanol and sonicating for one hour. Microelectrodes were then prepared by drop casting 30μL of this suspension onto an oxygen plasma cleaned 1cm×1cm stainless steel strip and drying under vacuum at 110 ◦ C for two hours. Li foil was used for both the counter and reference electrodes. Operando diffraction patterns were collected using an AMPIX electrochemical cell equipped with two glassy carbon windows. Thick film electrodes were pre- pared by blending 10% graphite powder (300 mesh), 10% acetylene black, 20% polytetraflouroethylene (average particle size of 1 μm), and 60% active material, and pressing under a hydrostatic pressure of 0.9 tons. The same glass fiber sepa- rators, metallic counter electrodes, and electrolyte solutions described previously were used during all operando experiments. Computational Methodology Density functional theory calculations were performed using the Vienna ab initio Simulation Package (VASP), with the Pro- jector Augmented Wave (PAW) method used to describe the interactions between core and valence electrons. 92–96 All calculations were spin polarized. Convergence with respect to the plane wave basis set and k-point sampling was tested, with a cut-off energy of 500 eV and k-point grid of Γ-centered 26×26×26 which was found 26 to be sufficient for the 4 atom unit cell of ReO 3 . For LiReO 3 and Li 2 ReO 3 , a cut-off energy of 500 eV and k-point grid of Γ-centered 6×6×6 were found to provide con- vergence. Geometry optimizations were performed using the PBEsol functional, a version of the Perdew Burke and Ernzerhof (PBE) functional revised for solids. 97,98 PBEsol has previously been shown to reproduce the lattice parameters for a broad range of solid-state systems. 99,100 Optimizations were deemed converged when the sum of all forces on each atom totaled less than 10 meVÅ −1 . In order to provide an accurate description of the electronic structure of the materials, the hybrid functional, HSE06, 101 was employed for band structure and density of states calculations. HSE06 combines 75% exchange and 100% of the correlation energies from PBE together with 25% exact Hartree-Fock (HF) exchange at short ranges and has been shown to perform well for a wide range of metals and semiconductors. 102,103 The Brillouin zone for thePm3 ¯ m andR3c space groups, along with the coordinates of the high-symmetry points, are provided in Figures 2.9 and 2.10. 2.3 Results and Discussion Several methods to prepare highly crystalline powders of ReO 3 have been reported. 104,105 The method of Chong et al. was chosen due to its ability to pro- duce fairly monodispersed nanoparticles with an average size around 15nm in a highly repeatable way, as illustrated in the inset of Figure 2.2. 86 The observation of lattice fringes in the TEM images, shown as the inset of Figure 2.2, clearly demonstrates that even very brief heating times were sufficient to yield highly crystalline particles of ReO 3 . This is supported by the results of Rietveld refine- ment of the structure against synchrotron XRD patterns as shown in Figure 2.2, 27 Figure 2.2: Results of the Rietveld refinement of the cubic structure against synchrotron XRD patterns of the nanoscale ReO 3 samples where R Bragg = 2.8%. Experiment Calculation Difference 10 15 20 25 30 2θ(deg) [ λ=0.414584 Å] Intensity (arb. units) which exhibit fairly sharp peaks with only very minor broadening of the reflec- tions, and no signs of any secondary phases. A microstructural analysis of the peak broadening performed within FullProf indicated an average particle size of 20nm, in close agreement with the results of the microscopy. Detailed results of the refinement can be found in Table 2.2. No significant differences were seen when cycling particles with larger sizes, so powders prepared in this way were used in all of the following discussion. These powders were used to prepare film electrodes as described in the Exper- imental Details and their galvanostatic performance was evaluated in standard Swagelok cells. During discharge, three distinct changes in slope can be seen at 2.5V, 2.3V, and 1.3V yet only the first and last feature are distinguishable on reversal, as shown in Figure 2.3(a). The presence of such a polarization implies a kinetic barrier to either electron or ion transport. The metallic character of ReO 3 28 (b) 1.0 1.5 2.0 2.5 3.0 3.5 Voltage vs Li/Li + (V) -0.06 -0.04 -0.02 0.00 0.02 0.04 Current (mA) 2.5 V to 3.6 V 2.0 V to 3.6 V 1.0 V to 3.6 V 0.0 0.5 1.0 1.5 2.0 x in Li x ReO 3 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Voltage vs Li/Li + (V) First Charge/Discharge (a) Figure 2.3: (a) Voltage vs composition of Li x ReO 3 during first cycle. (b) Cyclic voltammetry scans of ReO 3 microelectrodes collected at 1mVs −1 against Li/Li + in a three electrode cell within the glovebox. would seemingly preclude any issues with electronic transport and therefore the polarization is likely related to sluggish Li-ion mobility or an asymmetric inser- tion process. Galvanostatic Intermittent Titration Technique (GITT) experiments were therefore performed to determine how much of the polarization was due to non-equilibrium cycling conditions; however, a significant polarization is still seen even when the cell is allowed to fully relax. (Figure 2.11) The electrochemistry was further characterized using cyclic voltammetry in excesselectrolyteinordertoensure(de)insertionwasnotlimitedbymasstransport effects. Powders of ReO 3 were deposited on strips of steel and characterized in a three-electrode configuration. Figure 2.3 (b) shows the same three Faradaic peaks found in the galvanostatic cycling data between 1.0 and 4.0V. Yet, when sweeping across the entire voltage window, a partially irreversible wave evolves below 1.5V thatcorrespondstotheonsetofelectrolytereductiontoformLi 2 CO 3 onthesurface of the cathode as will be shown more clearly later. 106 29 In order to gain insight into the mechanism for Li-ion intercalation into the per- ovskite framework, operando X-Ray diffraction experiments were performed at the 17-BM line at Argonne National Lab. The use of synchrotron X-rays allows for the collection of high intensity diffraction patterns during cell operation, showing clear correlations between electrochemical behavior and structural changes. Operando methods are important for ensuring the structural changes can be accurately corre- latedwithdistinctelectrochemicalfeatures. Figure2.4displaysdiffractionpatterns that were collected every 15 minutes during a full discharge and charge cycle of Li x ReO 3 (0<x<2.0) at a C/10 rate. Startingondischarge,theBraggreflectionsmaintainarelativelyconstantinten- sity but shift to higher angles for Li x ReO 3 (0<x<0.35), suggesting the cubic symmetry is maintained when small amounts of Li are inserted into the structure and that a solid solution can be obtained over this interval. Subsequently, the 7.5 8.5 9.5 2θ (deg.) ( λ=0.24116 Å) 4.75 5.25 5.75 Intensity (arb. units) Li 1 ReO 3 Li 0.35 ReO 3 Li 0 ReO 3 Li 1 ReO 3 Li 2 ReO 3 Figure 2.4: Operando synchrotron X-ray diffraction patterns of ReO 3 throughout a complete discharge to Li 2 ReO 3 . 30 peaks associated with ReO 3 begin to lose intensity and new reflections that can be indexed to the hexagonal structure of Li x ReO 3 identifed by Cava et al. begin to grow. 83 Counterintuitively, this structural distortion leads to a contraction of the unit cell volume by approximately 5%, which is due to Li + ions being too small to fully fill the vacant A-site in the structure. Figures 2.5(a) and (b) show that the transformation from the highly ordered cubic structure to the hexagonal phase is the result of octahedral tilting modes along the [1 1 1] direction of the cubic cell while Figure 2.13 shows these rotations are accompanied by a contraction along the c-axis of the hexagonal cell. As more Li is inserted, the peaks associated with the hexagonal phase remain essentially unchanged until approximately x = 1.25, at which point new reflec- tions begin to evolve until complete lithiation at x = 2.0. This indicates that little structural rearrangement is required beyond the initial compression of the A-site pocket when only a single Li occupies each pocket. Yet, in order to insert beyond x = 1.0 the A-site must re-expand to allow two Li-ions per pocket, which is clearly seen as the peak at 5.25 ◦ begins to evolve symmetric shoulders at 5.14 ◦ and 5.41 ◦ which continually increase their separation with increasing Li content. Interestingly, the Li x ReO 3 phase never seems to entirely disappear suggesting an incomplete discharge during the operando measurement. Unfortunately, due to the close structural relationship and extensive peak overlap in the patterns, pre- cise Rietveld refinement of individual patterns proved intractable. On charging, changes in the diffraction patterns mirror those in the discharge process albeit over a slightly different compositional range, as shown in Figure 2.14. This can be seen in the peaks at 5.2 ◦ , 8.3 ◦ , and 9.1 ◦ that are visible from 0.5>x>2.0 compared to 2.0<x<1.25 on discharge. The reason for the asymme- try in the structural distortions can be understood by examining Figure 2.5 and 31 Figure 2.5: Illustration of the structural distortion that results in a twisting of the close-packed planes of oxygen in (a) ReO 3 , (b) LiReO 3 , and (c) Li 2 ReO 3 . Each structure is shown along the equivalent direction of the body diagonal within the cubic perovskite structure. 2.13. Migration of Li into the cubic phase of ReO 3 proceeds through large vacan- cies in a highly symmetric close-packed lattice whereas the octahedral rotations that produce the Li 2 ReO 3 structure close these vacancies and block many of the pathways for Li-ion transport. It should, therefore, not be surprising that such a complex rearrangement of atoms to return to cubic symmetry would proceed through a different mechanism that involves a more correlated shuffling of ions across each interstitial. More importantly, these distortions severely limit the long term cycling sta- bility, as shown in Figure 2.6, where only 40% of the initial capacity is retained after 25 cycles when the lower cut-off voltage is set at 1.0V. By setting the cut- off voltage to 2.8V, which avoids these distortions by only inserting lithium into the cubic form of Li x ReO 3 (x < 0.35), the capacity can be stabilized over mul- tiple cycles. This may be expected as structural deformations are well-known to produce particle cracking and dewetting from the current collector that can elec- trically isolate portions of the electrode and reduce the accessible capacity of the cell. 107–109 The severity of the effect is due to local strains as the corner-sharing octahedra rotate, bending the O–Re–O bond angle from 180 ◦ to 145 ◦ in Li x ReO 3 32 and 138 ◦ in Li 2 ReO 3 , corresponding a 23% change at full lithiation. The trans- formation to Li x ReO 3 also results in a 5% contraction of the unit cell, yet further lithiation to Li 2 ReO 3 acts to cancel this contraction by expanding to nearly the same volume as the Li-free starting material in order to accommodate the two Li-ions per A-site. Interestingly, such pronounced structural transformations are somewhat unexpected given that nanoscale powders usually suppress phase trans- formations. 110 0 5 10 15 20 Cycle number 0 50 100 150 200 250 300 Capacity (mA h/g) 2.8 V Cut-Off (Discharge) 2.8 V Cut-Off (Charge) 1.0 V Cut-Off (Discharge) 1.0 V Cut-Off (Charge) Figure 2.6: Charge and discharge capacity for a range of voltage cutoffs showing a severe loss of capacity when intercalating more than 0.35 equivalents of Li-ions into ReO 3 . Having firmly established that the structure of ReO 3 distorts in a manner highly reminiscent to what has been reported for polyanionic intercalation hosts, the effect of these correlated rotations on the electronic structure was explored. As expected, hybrid density func- tional theory calculations indicate ReO 3 is metallic, with three dou- bly degenerate bands (Re 5d-t 2g character),crossingtheFermilevel, with the two Re 5d-e g bands found higher in energy (∼4–6 eV above the Fermi level), in agreement with previous calculations. 111,112 In total, these six bands share a single electron, causing the Fermi level to sit resonant within them. The density of states (DOS) and band structure of for ReO 3 , LiReO 3 , and Li 2 ReO 3 are shown in Figure 2.7 (a)-(f). The orbital projected DOS clearly shows 33 a. b. c. d. e. f. Figure 2.7: (a)-(c) Densities of states and (d)-(f) band structure for ReO 3 , LiReO 3 , and Li 2 ReO 3 , respectively. the edge of the valence band maxima to be entirely composed of Re 5d and O 2p orbitals, with the Re states dominating at the Fermi level. Similar to ReO 3 , the valence band of LiReO 3 is again composed of Re 5d and O 2p orbitals, with the metallic behavior being conserved despite the significant bucklingoftheRe–Obondangles. However, therotationsoftheoctahedraproduce six significantly less disperse bands crossing the Fermi level in ReO 3 to split into two groups: the first consisting of two bands, spanning the energy range−0.6 to 1.0 eV; and the second consisting of four bands, spanning the energy range−2.0 to−0.3 eV. Further distorting the lattice to Li 2 ReO 3 results in a medium band 34 gap semiconductor, with a indirect band gap of 2.38 eV due to the full occupation of these six bands. These calculations indicate that Li x ReO 3 remains metallic for most values ofx up untilx = 2.0, which completely fills the valence band, strongly supporting the fact that the rotational distortions that occur on Li insertion are independent of any kind of polaron migration. In order to experimentally corroborate the calculated DOS and further demon- strate that polaronic effects are not present during Li insertion, X-ray photoe- mission spectroscopy was used to evaluate the electronic structure of the phases. Figure 2.8(a) shows the Li 1s and Re 4f core-regions of pristine and lithiated states of ReO 3 . Pristine ReO 3 is expected to have a Re 6+ oxidation state, with a 4f 7/2 peak centered at 43.1 eV. The photoemission for ReO 3 displays additional peaks associated with surface Re 7+ , since Re 2 O 7 is the most thermodynamically stable phase and exists on the surface as Re 7+ with its main peak at 45.5 eV. 113 Upon lithiation, the Re 4f shifts to lower binding energy and reaches a state indicative of having a Re 4+ oxidation state (main 4f 7/2 peak at 42.2 eV). In addition, the main Li 1s peak becomes present in the fully lithiated state. Due to the low discharge voltages, Li 2 CO 3 and LiF contaminants contribute to this Li 1s signal. These assignments were based on the O 1s and F 1s core levels (see Figure 2.12). Given the fact that we can still see the Re 4f, we can estimate the surface contaminants have a thickness of≤ 3nm. The valence band structure was measured for direct comparisons with the DFT calculations of the distorted Li x ReO 3 phases, with their band structures and den- sity of states plotted in Figure 2.7. Figure 2.8(b) shows the valence band structure of Li x ReO 3 samples. In order to directly compare our experimental results with DFT, we accounted for the orbital cross-sections and experimental broadening. 35 The projected orbital density of states were weighted by the corresponding pho- toionizationcross-section. 114 Figure2.8(c)showstheconvolutedsumoforbitalpro- jections from DFT weighted by their corresponding photoionization cross-section convoluted with a Voigt profile (0.1 eV Lorentzian width, 0.5 eV Gaussian width) to match experimental resolution. Excellent agreement between the measured and calculated electronic structure of ReO 3 is observed; the 3-10 eV is associated with primarily O 2p and the topmost states up to the Fermi level are hybridized Re 5d-O 2p. As ReO 3 discharges, we see two additional features emerge at 4 eV and 9 eV along with increased intensity up to the Fermi-edge, consistent with the DFT of LiReO 3 . Upon dischargingto 2 Li + , we notedifferences between the XPS andDFT due to the aforementioned surface species. Figure 2.8(b) includes reference spectra of Li 2 CO 3 and LiF, which are seen to increase the spectral weight in the 6-10 eV range. These surface species attenuate the Li 2 ReO 3 signal and partially account for the poorer statistics for Li 2 ReO 3 despite the longer data acquisition time taken for this sample. Despite these attenuating surface species, one still observes a very pronounced hybridized Re 5d-O 2p contribution in the XPS for Li 2 ReO 3 (see Figure 2.8: XPS of pristine ReO 3 and lithiated states (Li 0.35 ReO 3 and Li 2 ReO 3 ). (a) Re 4f core-region with vertical line depicting Re oxidation state. (b) Valence band of O 2p and Re 5d hybridized regions with inset of Fermi-level. (c) Weight and broadened density of states from DFT of ReO 3 , LiReO 3 , and Li 2 ReO 3 . 36 the inset of Figure 2.8(b)), consistent with a fully occupied band. The poor pho- toemission signal of Li 2 ReO 3 is consistent with the medium gap semiconducting character determined from the DFT calculations. Combiningalltheseresults, themetalliccharacterofReO 3 appearstobeexcep- tionally robust towards bending of the O–Re–O bond angle. While in the metallic state, electron transport is fully decoupled from the motion of the intercalating ions, which appears to indicate that polaron migration is not driving the corre- lated rotations of the corner-sharing octahedra. 2.4 Conclusions Using complementary electrochemical, spectroscopic, and structural characteriza- tion tools we have presented a detailed study on the mechanism of Li-ion insertion into ReO 3 . While the open framework and metallic character of ReO 3 should, intu- itively, allow it to exhibit exceptionally high performance as a cathode for Li-ion batteries, our results clearly show poor long-term cycling stability related to the highly correlated rotations of the corner-sharing perovskite octahedra. This work highlights the importance of understanding the underlying structural distortions that occur upon guest ion insertion during electrochemical cycling and, further- more, demonstrates that the rotational distortions observed in insulating polyan- ionic intercalation hosts do not appear to be purely caused by polaron migration. These structural distortions severely impede the reversible (de)insertion of Li into ReO 3 and would have to be suppressed in order for similarA-site vacant perovskite hosts to be effective electrode materials. 37 2.5 Supplemental Information Geometry Optimization The lattice parameters of ReO 3 , LiReO 3 , and Li 2 ReO 3 optimised using the PBEsol functional, are provided in Table 2.1. In general, the results show good agreement with experiment (most values within 0.7 %), with the exception of the a parameter of Li 2 ReO 3 , which was found to be slightly underestimated (−1.2 %). Table 2.1: Lattice parameters calculated using the PBEsol functional. Difference versus experiment given in parentheses. For ReO 3 , all cell angles were found to be 90 ◦ . For LiReO 3 and Li 2 ReO 3 , α and β were 90 ◦ and γ was found to be 120 ◦ , as expected based on the R3c space group. Compound a / eV c / eV ReO 3 3.767 (+0.44 %) — LiReO 3 5.092 (−0.14 %) 13.403 (+0.62 %) Li 2 ReO 3 4.911 (−1.22 %) 14.886 (+0.66 %) Rietveld Refinement A Rietveld refinement was performed using the program Fullprof on X-ray diffraction data of ReO 3 . Data was collected at 11-BM, Argonne National Laboratory. Note that the lattice parameter is the only free variable aside from the thermal parameters, which were fixed 1.0 for both Re and O. Re sits at (0,0,0) and O sits at ( 1 2 ,0,0) Table 2.2: Structural parameters calculated using Rietveld refinement of syn- chrotron X-ray diffraction data of pristine ReO 3 . a (Å) R Bragg (%) Size (nm) Strain (%) ReO 3 3.750998(6) 2.8 20.3(1) 42(2) 38 Brillouin Zone Calculations The Brillouin zones for all relevant space groups were visualized to demonstrate high symmetry k-points. L X M Γ k z k x k y Figure 2.9: Brillouin zone for the Pm3 ¯ m (#221) space group, indicating all high-symmetryk-points. The coordinates of the high symmetryk-points are: Γ = (0, 0, 0); R = ( 1 2 , 1 2 , 1 2 ); X = (0, 1 2 , 0); M = ( 1 2 , 1 2 , 0). L Z F Γ k z k x k y Figure 2.10: Brillouin zone for the R3c (#161) space group, indicating all high- symmetry k-points. The coordinates of the high symmetry k-points are: Γ = (0, 0, 0); L = (0, 1 2 , 0); F = (0, 1 2 , 1 2 ); Z = ( 1 2 , 1 2 , 1 2 ). 39 0.0 0.5 1.0 1.5 2.0 x in Li x ReO 3 1.0 1.5 2.0 2.5 3.0 3.5 Voltage vs Li/Li + (V) Figure 2.11: GITT cycling of Li x ReO 3 shows a large polarization in the equilib- rium potentials of Li x ReO 3 . Galvanostatic Intermittent Titration GITT type measurements were per- formed to assess the polarization during cycling of Li x ReO 3 . Even after relax- ation, a large polarization is seen to exist during cycling due to the structural rearrangements which occur upon (de)lithiation. X-ray Photoemission Spectroscopy Figure 2.12: (a) O 1s and (b) F 1s comparing different lithiated states of ReO 3 . Higher binding energy at 532 eV in O 1s core region has been identified as Li 2 CO 3 . Comparing with pristine ReO 3 , discharging to such low voltages results in a large increase in lithium carbonate species on the surface. Additionally, the F 1s core regions shows likely P-O-F and Li x PF y species for the fully discharged state. Due to similarity between binding energies specific species cannot be identified but lie between 686 eV and 688 eV. 40 Structural Distortions Using a primitive unit cell, it is possible to compare the crystal structure for ReO 3 , LiReO 3 , and Li 2 ReO 3 directly. The primitive cells all have the lattice parameters a = b, c and α = β = γ = 90 ◦ where a and c are shown for each phase. Note the contraction along a due to the in-plane rotation of the octhedra and corresponding extension along c. (a) (b) (c) a=5.3038 Å c=12.9918 Å a=5.1154 Å c=13.3573 Å a=4.9845 Å c=14.6342 Å Figure 2.13: Projection of the crystal structure for (a) ReO 3 , (b) LiReO 3 , and (c) Li 2 ReO 3 within a primitive unit cell to facilitate direct comparison. 41 Operando X-ray Diffraction The structure of Li 2 ReO 3 seen to mostly trans- form back to the pristine ReO 3 structure upon delithiation, with a small percent remaining in the LiReO 3 structure. The pattern corresponding to Li 1 ReO 3 was not collected due to a synchrotron outage while the experiment was conducted. 7.5 8.5 9.5 2 θ (deg.) [ λ=0.24116 Å] Li 0 ReO 3 Li 0.8 ReO 3 Li 2 ReO 3 4.75 5.25 5.75 Intensity (arb. units) Figure 2.14: Operando X-ray diffraction patterns of Li 2 ReO 3 through a charge to ReO 3 . 42 Chapter 3 Understanding the Role of Crystallographic Shear on the Electrochemical Behavior of Niobium Oxyfluorides 3.1 Introduction Thestructuraldistortionsoflithiumintercalationhostshavebeenanareaofexten- sive research and are of the utmost importance to the design and operation of electrode materials for use in Li-ion batteries. The intercalation of Li + into a crys- talline metal oxide framework and the associated redox often leads to destructive changes in bond lengths and a rearrangement of crystallographic building blocks, either by the distortion of existing chemical bonds or formation of new bonds, to accommodate inserted cations. 115 This dynamic process has a powerful influence on the electrochemical performance of battery electrode materials, affecting many aspects including capacity retention, polarization, and rate performance. 116,117 Many approaches have attempted to mitigate deleterious consequences of atomic rearrangement, including nanostructuring, architectural modification, and crystal structure engineering. 118–120 Recently, niobium oxides based on perovskite and Wadsley-Roth phases have drawn increased interest for use in energy storage applications, due to a combina- tionofhighionicandelectronicconductivitywhichpermitsgoodrateperformance, even in micron-sized particles. 121–123 First identified in the Ti-Nb-O phase space, Wadsley-Roth phases are formed by the removal of a plane of anions from the 43 43 Adapted from Bashian et al. J. Mat. Chem A 2020 25 12623-12632 [doi] 43 ReO 3 type structure resulting in alternating layers of edge sharing and corner shar- ing octahedral units. 124–129 Oxide shear phases have shown high rate performance in examples such as niobium-tungsten oxides and TiNb 2 O 7 with some structures suppressing phase transformations during cycling, allowing for fast Li diffusion through the material. 130–133 A series of these compounds have been chemically or electrochemically lithiated, with studies suggesting the edge sharing shear planes stabilize the structure. 83,134,135 NiobiumbasedWadsley-Rothshearphasesrepresentarichphasespace, includ- ing Nb 3 O 7 F derived from the perovskite NbO 2 F. 136 These formulas represent a ReO 3 -type perovskite and a shear derivative of the form 3×∞×∞, respectively, with open channels allowing for ion insertion. We note the relatively uncommon Figure 3.1: (a): NbO 2 Fexists in a perovskite structure with corner sharing octa- hedra. (b): Nb 3 O 7 Fcontains layers of edge sharing and corner sharing octahedra formed by the removal of an anionic plane of atoms from the perovskite struc- ture. (c): Representative diagram of the shear relationship between NbO 2 F and Nb 3 O 7 F. 44 one–dimensional structure type represents a shear plane along a single crystallo- graphic axes, as opposed to the more commonm× n×∞ structure which forms two–dimensional m× n blocks. 137 Our group has previously investigated the electrochemical (de)lithiation of the archetypical perovskite, ReO 3 , using operando X-ray diffraction (XRD) measure- ments to characterize a correlated rotation of ReO 6 polyhedra upon lithiation. 44 It was postulated by Cava et al. that a shear structure of ReO 3 would make this twisting motion more difficult due to the inclusion of edge-sharing octahedra. 82 The identification of structural characteristics that control such phase changes is of inherent importance to the design of next generation lithium ion battery elec- trodes. As such, we began a study to determine the effects of shear structuring on the electrochemical performance of niobium oxyfluorides. As both Cava and Permér observed, the formation of crystallographic shear planes affects structural changes caused by Li-insertion reactions into perovskites by limiting the rotational freedom of octahedral units. 83,138,139 Due to the expected differences in ion mobility and structural stability, the electrochemical behavior of these materials is of great interest. Therefore, we began a series of operando XRD and X-ray absorption (XAS) measurements in conjunction with detailed electrochemical characterization and Raman spectroscopy to illustrate the effect of shear planes in the niobium oxyfluorides, NbO 2 Fand Nb 3 O 7 F, identifying multiple regions throughout the (de)lithiation process where the inclusion of shear planes was manifested. 45 3.2 Experimental Methods CAUTION: Anhydrous HF and associated reagents can cause severe chemical burns. Before undertaking any experiments, one should familiarize oneself with the hazards associated with all reagents as well as proper handling and techniques. Fresh tubes of calcium gluconate gel should be available for fast treatment of skin exposure by any reagents. For additional information, please see Segal et al. 140 Synthetic Methods. NbO 2 Fparticles, of several hundred nanometers in diam- eter, were prepared using an adaptation of the method reported by Frevel et al. 73 Large batches of NbO 2 F were prepared by dissolving Nb 2 O 5 , as purchased, in anhydrous HF in a fluorinated ethylene propylene (FEP) tube in a liquid nitrogen bath. The reaction was slowly brought to room temperature while being stirred and the excess HF was removed by vacuum to yield NbO 2 F· HF salt. Due to the hygroscopic nature of the salt, it was stored in an inert atmosphere. The com- plexed HF was removed from the salt by heating under vacuum at 150 ◦ C for a period of 12 hours to yield NbO 2 Fas a pale lavender powder. Nb 3 O 7 Fwas prepared by grinding Nb 2 O 5 with polytetrafluoroethylene (PTFE) in a 1 : 3 molar ratio and pressing into a pellet followed by heating in a quartz glass ampule sealed under vacuum. Heating was performed in a Panasonic 1200 W microwave set to 60% power for a period of 13 minutes, with the glass ampule contained in a bed of charcoal to act as a microwave susceptor. Physical Characterization. Laboratory XRD patterns were collected on a Bruker D8 diffractometer with a Cu K α source (λ 1 = 1.5406Å, λ 2 = 1.5444Å), 46 equipped with a Lynxeye XE-T detector. High resolution synchrotron pow- der diffraction data was collected using beamline 11-BM at the Advanced Pho- ton Source (APS), Argonne National Laboratory using an average wavelength of 0.45784Å. Discrete detectors covering an angular range from -6 to 16 ◦ 2θ were scanned over a 34 ◦ 2θ range, with data points collected every 0.001 ◦ 2θ and scan speed of 0.01 ◦ /s. The resulting diffraction patterns were refined against published structures using the Rietveld method as implemented in the FullProf program. 73,89,141,142 Scanning Electron Microscopy (SEM) was conducted on a FEI Nova Nano 650 FEG SEM. NbO 2 Fwas analyzed at 10 kV; Nb 3 O 7 Fwas analyzed at 5 kV. The particles were uncoated and measurements were made in immersion mode. Electrochemical Characterization. The electrochemical performance of both materials was characterized using Swagelok-type cells assembled in an argon-filled glovebox, using Li metal as a combined counter and reference electrode and What- man GF/D borosilicate glass fiber sheets as the separator. 1M LiPF 6 in ethylene carbonate and dimethylcarbonate (1:1 v/v) was used as the electrolyte (LP30). Thick film electrodes prepared by blending 10% graphite powder (300 mesh), 10% acetylene black, 20% polytetraflouroethylene (average particle size of 1 μm), and 60% active material, and pressing under a hydrostatic pressure of 0.9 tons to yield electrodes of 10-15 mg (total). Given the insulating nature of the active material, the chosen electrode formulation ensured adequate electrical conductivity. Cell components and electrodes of both NbO 2 Fand Nb 3 O 7 Fwere dried under vacuum at 110 ◦ C before assembly 47 Operando Measurements. Operando X-ray Absorption Spectroscopy (XAS) measurements were performed at beamline 12-BM, APS using the AMPIX electro- chemical cell, equipped with a glass fiber separator and a Li foil combined counter and reference electrode. Methods were adopted from Borkiewicz et al. 87 XAS mea- surements were performed in transmission geometry at the Nb K-edge (18.9 KeV), using Nb foil as a reference. Scans were collected in transmission mode over a span of eight minutes. XAS data processing was carried out using the ATHENA software of the package IFEFFIT. 143 The EXAFS component was normalized and converted to wavenumber. The resultant signal in k-space was multiplied by k 3 , Fourier-transformed and left with no phase shift correction. All displayed EXAFS data is shown with no phase shift correction. Ab initio calculations on relevant structure models were done using the code FEFF8.2 with fits being performed using the ARTEMIS software of the IFEFFIT package. 143,144 OperandoXRDmeasurementswereperformedusingaBrukerD8diffractometer with a Cu K α source (λ 1 = 1.5406Å, λ 2 = 1.5444Å), equipped with a Lynxeye XE-T detector. A modified Swagelok cell, with Be window serving as a current collector, allowed for diffraction patterns to be collected during electrochemical cycling. XRD scans were performed in a Bragg-Brentano geometry over a range of 20 ◦ to 50 ◦ 2θ for NbO 2 Fand 15 ◦ to 50 ◦ 2θ for Nb 3 O 7 Fwith a total scan time of 20 minutes. Scans were continuously repeated throughout the duration of electrochemical cycling. The same thick film electrodes, glass fiber separators, LP30 electrolyte, and Li foil counter electrodes previously described were used for cell preparation. Raman Spectroscopy. Electrochemical cells of both NbO 2 F and Nb 3 O 7 F were prepared as described above. Cells were discharged at a rate of C/10 based 48 on one electron per formula unit. Cells were disassembled in an Ar-filled glovebox and the working electrodes were washed with dimethylcarbonate and dried under active vacuum overnight. Raman spectra were collected on a Horiba XploRA One confocal Raman microscope. All spectra were collected with a 532 nm diode laser, a diffraction grating with groove density 2400 g mm −1 , and laser power ranging from 1.6-8 mW. The hole and slit were fixed at 500 and 50 μm, respectively. The laser was focused using a 50× (numerical aperature 0.5) objective, which yielded a spot size of ca. 1.3 μm. An acquisition time of 3 s was used, and 100 spectra were accumulated and averaged. 3.3 Experimental Results Several forms of niobium oxyfluorides are known and present an opportunity to investigate the effects of structural changes on the electrochemical behavior of the redox active niobium, in conjunction with the effects of fluorination. NbO 2 F crystallizes in the prototypical perovskite structure in space group Pm ¯ 3m (#221), with corner sharing NbO 4 F 2 octahedra arranged in an ordered network with Nb– O–Nb bond angles of 180 ◦ , as shown in Figure 3.1(a). 73 The vacant A-site in the structure creates three–dimensional channels for ion intercalation. In order to reduce the probability of residual HF or hydroxyl moieties from remaining in the NbO 2 F, we employed anhydrous HF and in vacuo heatings during synthesis. Based on analysis of synchrotron XRD there is less than one percent impurities present in the sample, however even a minute Nb 4+ impurity could lead to the slight lavender coloring in the sample. 49 5 10 15 20 25 2 �(deg) [ �=0.45786 Å] Intensity (arb. units) obs calc diff (a) R Bragg = 3.82 % 4 8 12 16 20 2 �(deg) [ �=0.45784 Å] Intensity (arb. units) obs calc diff (b) R Bragg = 4.68 % Figure 3.2: Rietveld refinement of synchrotron XRD data of NbO 2 F(a) showed phase pure product while refinement of Nb 3 O 7 F(b) revealed a small impurity of less than 1% NbO 2 F, illustrated by the orange reflections. Both compounds have small particle sizes, as shown in SEM imaging (insets). Nb 3 O 7 F crystallizes in space group Cmmm (#65), as shown in Figure 3.1(b), and is formed by a shearing of the octahedral layers of NbO 2 Fwhich creates alter- nating sheets of edge sharing and corning sharing Nb octahedra in an ABAB stacking pattern. A representative diagram of the shearing motion is displayed in Figure 3.1(c). The edge sharing octahedra are slightly distorted due to repulsion, resulting in a long and a short Nb–O bond along the a-axis, while the four bonds in the b–c plane are found to be equivalent. BoththesymmetryandpurityofNbO 2 F andNb 3 O 7 F wereconfirmedthrough Rietveld refinement of synchrotron XRD data, as shown in Figure 3.2(a) and (b) respectively, with refinement parameters provided in Table 3.1. Given the very similar X-ray cross sections of oxygen and fluorine, no evidence of fluorine ordering can be seen in Rietveld refinements on either compound. As demonstrated in the inset of Figure 3.2(a), SEM imaging of NbO 2 F reveals cubic particles on the order of 100− 200 nm in diameter. Small particles were expected given the low temperature synthetic methods employed. Similarly, the extremely short heating 50 0.0 0.2 0.4 0.6 0.8 1.0 x in Li x NbO 2 F 1.5 2.0 2.5 3.0 V vs.Li/Li + (V) 0 10 20 30 Cycle Number 80 120 160 Capacity (mAh g �1 ) 1.0 1.5 2.0 2.5 3.0 V vs. Li/Li + (V) �100 �50 0 50 I (mA g �1 ) 1.0 V 1.2 V 1.3 V 1.4 V 1.5 V (a) (b) (c) 0.0 0.5 1.0 1.5 x in Li x Nb 3 O 7 F 1.5 2.0 2.5 3.0 V vs.Li/Li + (V) 0 10 20 30 Cycle Number 0 50 100 150 1.0 1.5 2.0 2.5 3.0 V vs. Li/Li + (V) Capacity (mAh g �1 ) � 100 �50 0 50 I (mA g �1 ) 1.0 V 1.2 V 1.3 V 1.4 V 1.5 V (d) (e) (f) Figure 3.3: (a): Cyclic voltammetry with a gradually expanding window illus- trate the effects of structural change on the electrochemical cycling of NbO 2 F. Lower voltage limits of each trace are provided in the legend while all traces used an upper voltage limit of 3.2 V. (b): Galvanostatic cycling of Li x NbO 2 F in the voltage window of 1.5− 2.0 V shows cycling on the Nb 5+/4+ redox couple with good reversibility. (c) Capacity of each cycle shows stable behavior after the ini- tial cycle. In (d), a window opening CV experiment on Nb 3 O 7 F showed that the oxidative peak at 2.5 V is lost as the reductive limit is lowered, with lower volt- ages in the legend. In (e) galvanostatic cycling of Nb 3 O 7 F at a C/10 rate shows substantial irreversible capacity followed by stable solid solution cycling. In (f) the capacity retention is shown over 30 cycles. time used in the synthesis of Nb 3 O 7 Fwas expected to create small particle size and SEM images confirmed particles of 150− 300 nm (Figure 3.2(b) inset). The use of microwave techniques in the synthesis yielded smaller particle sizes than previously reported methods. The (de)lithiation of both NbO 2 F and Nb 3 O 7 F was probed through various methods including galvanostatic cycling and cyclic voltammetry, with several key behavioral differences identified in the following sections. The cycling behavior of NbO 2 F was investigated through cyclic voltammetry (CV) measurements that provided precise identification of redox voltages and details on the intercalation process. A voltage window opening experiment, in which the lower limit of a CV was lowered in a stepwise fashion, was used to identify the coupling of different 51 redox processes in correlation with structural changes. Figure 3.3(a) demonstrates two different regions present, most clearly seen by examining the window from 1.5− 3.2 V in comparison with the window from 1.0− 3.2 V. The first window shows a relatively sharp redox couple centered around 1.9 V; from galvanostatic measurements, it was seen that this region demonstrates better reversibility. This is contrasted by the latter window, in which the redox peaks are seen to become extremely broad with the oxidative wave stretching from 1.3 V to 2.8 V, indicative of disordering introduced in the structure that leads to a range of site energies for (de)intercalation and a poorly defined redox potential. This structural disorder is induced throughout the window opening, as shown by the gradually broadening redox peaks as the voltage window is widened and the 3+ oxidation state of Nb is reached. A CV experiment over a wide voltage window of 1.0− 3.2 V (Figure 3.9) was used to confirm that the broadening was not a function of extended cycling, as a very broad oxidative peak was seen on the first charge cycle. Galvanostatic cycling over a wide voltage window of 1.0− 3.2 showed that it was possible to insert two full units of Li + in NbO 2 F, with corresponding reduc- tion of Nb 5+ to Nb 3+ , however this reaction was found to be mostly irreversible and the cells rapidly lost capacity, as shown in Figure 3.7. However, analysis of the derivative curve (inset of Figure 3.7) indicated multiple reaction redox events occurring with an initial reduction seen between 1.5− 2.0 V, followed by a sub- sequent broad reduction below 1.5 V. In an effort to improve reversible cycling, NbO 2 Fwas cycled with a higher lower voltage cutoff, as shown in Figure 3.3(b). It was found that a cutoff voltage of 1.5 V led to improved cycling stability in these compounds, with a reversible capacity of approximately 90 mAh/g after 30 cycles, as shown in 3.3(c), suggesting that to Nb 3+ is irreversible. 52 In an effort to understand the role of structural shear planes in modifying electrochemical behavior, cyclic voltammetry was performed on Nb 3 O 7 F revealing an evolution in behavior during extended cycling that suggests structural changes were induced due to lithiation. Figure 3.3(a) shows that the redox behavior of Nb 3 O 7 Fchanges as the lower voltage limit is shifted downwards, leading to the loss of an oxidative peak at 2.5 V and the sharpening of a peak at 1.75 V. In the first cycles of the CV, from 1.5− 3.2 V, two distinct oxidation peaks are seen. We note the presence of a distinct peak at 2.5 V, reminiscent of the oxidative peak observed in NbO 2 Fcycled over the same voltage window. However, lowering of the voltage window leads to a loss of intensity and a gradual shifting of the two oxidative peaks, possibly reflecting structural changes in the material. This suggests that structurally distinct lithiation sites in the perovskitic layers of the shear structure are lost upon greater reduction of the structure, possibly due to an inability of the corner sharing Nb octahedra to accommodate changes in bond distance induced by a lowered oxidation state. 25.0 30.0 35.0 40.0 45.0 2 �(deg.) [ �= 1.5406 Å] Intensity (a.u.) 1.0 2.0 3.0 Voltage vs Li/Li + (V) 0 0.5 1.0 1.5 2.0 1.5 x in Li x NbO 2 F 20.0 25.0 30.0 35.0 40.0 45.0 2 �(deg.) [ �= 1.5406 Å] Intensity (a.u.) 1.0 2.0 3.0 Voltage vs Li/Li + (V) 0 1 2 3 2 1 x in Li x Nb 3 O 7 F (b) (a) Figure 3.4: Operando XRD measurements of NbO 2 F(a) throughout a complete discharge/charge cycle reveals the loss of crystallinity induced by lithiation to Li 2 NbO 2 F. The loss of intensity on diffraction peaks is not recovered upon charg- ing. In comparison, the structure of Nb 3 O 7 F(b) is seen to go through multiple changes during lithiation followed by a complete recovery upon charging, albeit with lowered crystallinity. 53 Further evidence of this change in structure is seen in Figure 3.10 which shows multiple CV cycles of Nb 3 O 7 Fover the voltage window 1.0− 3.2 V. The oxida- tive peak at 2.5 V is shifted upwards by 0.25 V on the first cycle, with this larger polarizationreflectingtheincreasedenergynecessarytoremoveLi + fromthestruc- ture. This peak is maintained for only the first two cycles, while the redox couple centered at 1.75 V is seen to cycle with relatively good stability for many cycles, suggesting better structural stability for lithiation at a structurally distinct site We observe that deep discharge to 1.0V leads to a rapid changes in redox behavior with the loss of the higher voltage oxidative peak, likely due to greater structural rearrangement. Galvanostatic cycling of Nb 3 O 7 F from 1.0− 3.2 V shows a large irreversible capacity on the initial cycle associated with the oxidative plateau observed at 2.5 V eventually being replaced with stable solid solution type cycling, which results in a reversible capacity of approximately 85 mAh/g, as shown in Figure 3.8. During the charging process, the voltage trace dips downwards at approximately 2.5V before continuing to rise sharply, implying an overall lowering of the structural energy of the Li x Nb 3 O 7 F after partial delithiation. In an effort to improve the reversibility of cycling Nb 3 O 7 Fa voltage window of 1.5−3.2 V was used, as shown in Figure 3.3(b). Interestingly, we found that the voltage dip observed in 3.8 was also present when a higher voltage limit was used; however, the feature did not reappear on subsequent cycles. Instead multiple charge/discharge cycles led to a smooth, featureless voltage curve with no apparent plateaus. As will be demonstrated in subsequent sections, the drop in voltage observed during the charge curve is associated with structural changes that occur as shear layers in the structure rotate about one another. An analysis of operando XRD patterns provides an excellent visualization of the structural changes induced by 54 lithiation of both NbO 2 F and Nb 3 O 7 F. During the initial lithiation of NbO 2 F, diffraction peaks shift steadily to higher angles as shown in Figure 3.4(a). This peakshiftingobservedislikelyassociatedwitharotationoftheoctahedralsubunits that comprise the structure as the unit cell shrinks during lithiation to stabilize Li + ions in vacancies within the structure. Due to the entirely corner sharing structure in NbO 2 F, contraction can occur only by rotation or distortion of the octahedral units with precedent for octahedral rotation. 44 The upwards peak shift isfollowedbyasubstantialdecreaseinpeakintensityatgreaterdegreesoflithiation as crystallinity and long-range order are lost. The loss of crystallinity is further visualized in a heatmap of diffracted intensity as a function of state of charge in Figure 3.11 As Li + is cycled in NbO 2 F, the bond lengths change to accommodate redox activity which causes changes in bond lengths and a flexing of the unit cell. Sim- ilar behavior has been observed in the isostructural ReO 3 , where the contraction and expansion of the unit cell led to fracturing of particles. 44 While reduction of NbO 2 F by two units of lithium leads to a mostly amorphous material and loss of nearly all crystallinity, cycling of a single Li + with a cutoff voltage of 1.5 V allows for better structural maintenance. As shown in Figure 3.12, during two discharge/charge cycles of NbO 2 F with a lower voltage limit of 1.5 V, diffraction peaks are maintained and a corresponding expansion of the unit cell on delithi- ation is observed, suggesting that the structure can accommodate Nb 5+/4+ redox but accesing Nb 4+/3+ redox leads to an irreversible transformation. During the insertion of one Li + , the main structural peaks of NbO 2 F shift to higher angles indicating a contracting unit cell, while this process is reversed upon deinsertion. The smaller bond distance changes and reduced strain necessitated by the single redox event allows the structure NbO 2 F to be maintained. 55 An interesting contrast is drawn upon examination of the electrochemical and structural evolution of Nb 3 O 7 F during lithiation, in which the presence of shear planes could provide a stabilizing force, allowing for greater structural mainte- nance throughout electrochemical (de)lithiation. In Figure 3.4(b), it is seen that lithiation of Nb 3 O 7 F causes large changes in the diffraction pattern, with an ini- tial expansion of the unit cell in the a-b plane, for example along the (110) peak at 23.5 ◦ . Upon further lithiation, the loss of starting peaks and introduction of additional peaks is observed, but interestingly, the peaks that are retained in the XRD pattern are primarily associated with shear layers. These shear layer peaks, corresponding to reflections such as the (310) or (110) at 26.4 ◦ and 23.5 ◦ respec- tively, are seen to shift and split thus indicating the potential lowering of the space group. This is indeed the case as the peaks are matched from an orthorhombic to triclinic unit cell where the triclinic phase is present at lithium values of x = 1.71 upon discharge to x = 1.96 upon charge (Figures 3.13 and 3.14). This value is con- sistent with data from chemical lithiation by Permér et al., albeit electrochemical lithiation affords a more gentle lithiation process thereby allowing the structure to accommodate more lithium before undergoing a phase transition to a lower symmetry space group. 139 The gradual XRD peak shift on discharge is indicative of a solid solution pro- cess; the charging process occurs through phase growth in which the starting structure is recovered (Figure 3.13 and 3.15). The phase growth occurs in the same region where the oxidative voltage dip was noted in Figure 3.3(b) and 3.7. These differences between discharge and charge suggest that the lithiation process disrupts the perovskitic linkage between shear planes, leading to a loss of peak 56 intensity with diffraction planes containing those layers. Upon charging, the sta- bilized shear layers allow the structure to partially reform, however stacking faults occur in the layers, ultimately affecting electrochemical performance. While XRD measurements provided insight into average structural changes, operando XAS measurements were used to monitor the Nb oxidation state as well aslocalstructuralchangesduringcycling. Alldistancesdiscussedandthedisplayed data are not phase shifted. As shown in Figure 3.17, the Nb K-edge was tracked throughout a complete discharge cycle of NbO 2 F to Li 2 NbO 2 F, which results in the reductionofNb 5+ toNb 3+ . ThiscausesadownshiftintheK-edgeenergythatshifts steadily with lithiation and restructures in shape due to structural rearrangement. The radial distribution function, generated from the Fourier transform of the XAS data, provides a depiction of local structural changes, as shown in Figure 3.5(a). During initial lithiation, the second-shell peak at 3.5 Å is seen to steadily lose intensity while shifting to longer distances. Fits of the EXAFS data show that this 1.0 2.0 3.0 4.0 Radial Distance (Å) | �(R)| (a.u.) Li 2 NbO 2 F Li 1 NbO 2 F NbO 2 F (a) 1.0 2.0 3.0 4.0 Radial Distance (Å) | �(R)| (a.u.) Nb 3 O 7 F Li 1 Nb 3 O 7 F Li 3 Nb 3 O 7 F (b) Figure 3.5: (a): The radial distribution function indicates substantial local rear- rangement around Nb atoms in NbO 2 F, with a loss of a peak at 3.4 Å. (c) The radial distribution function of Nb 3 O 7 F, seen throughout a complete (de)lithiation cycle. Note the loss and eventual partial return of the peak at 3.4Å. 57 peak is primarily composed of Nb–Nb scattering distances between octahedra. As NbO 2 F is lithiated over one unit of Li + , the Nb octahedral linkage is disrupted resulting in a loss of a coherent scattering distance. Further operando XAS measurements were performed on the Nb K-edge in Nb 3 O 7 Fthroughout a complete discharge/charge cycle in order to probe the reac- tion mechanism. As expected, the Nb K-edge is observed to shift to lower energies throughout lithiation, followed by reversal upon delithiation, with this process visualized in Figure 3.18. An incomplete recovery of the K-edge is indicative of the partial irreversibility of the reaction, as Nb is not fully oxidized back to the +5 oxidation state. Additionally, the radial distribution shows changes throughout the cycling of Nb 3 O 7 F, as shown in Figure 3.5(b). As Nb 3 O 7 Fis lithiated, a peak at 3.4 Å is seen to lower in intensity, eventually disappearing completely in the fully discharged material. However, this peak is partially recovered upon charging, as some intensity is seen at 3.4 Å in the fully charged material. This peak cor- responds the Nb-Nb distances in the perovskitic layers of the Nb 3 O 7 Fstructure, which are disrupted during lithiation but may be partially recovered after delitha- tion. Interestingly, this peak does not shift upwards during lithiation as is seen in the peak at 3.4Å in NbO 2 F. To further probe changes to the structure of NbO 2 F and Nb 3 O 7 F upon lithi- ation, ex situ Raman spectroscopy was measured on samples discharged to 1.5 V and 1 V. The spectra for the NbO 2 F compound along with the reduced NbO 2 F are shown in Figure 3.6(a). The full wavenumber range can be found in Figure 3.16. The pristine NbO 2 F spectrum contains several broad vibrational modes that are assigned in Table 3.2. The most notable modes are the intense mode cen- tered at 703 cm −1 and the shoulder at 620 cm −1 , which are assigned to the NbO 6 symmetric stretch and bridging Nb-O-Nb stretch, respectively. 145 Upon reduction 58 to 1.5 V, both modes lose intensity and new, strong modes appear at 597 cm −1 and 816 cm −1 , which suggests a significant change in the crystal symmetry around the NbO 6 octahedra and their linkages. Further evidence for significant symme- try changes is found in the low wavenumber region of the spectrum. The broad modes in pristine NbO 2 F yield to much more defined, sharp modes in the sample discharged to 1.5 V. The new modes are interestingly similar to those in the low wavenumber region of the pristine Nb 3 O 7 F. Upon further reduction to 1 V, the new mode at approx. 597 cm −1 is maintained but the 816 cm −1 disappears again suggesting significant structural changes. The Raman spectrum of NbO 2 F after reduction does not maintain any features found in the spectrum of the pristine material other than a mode at 620 cm −1 related to the corner-sharing connectivity of the octahedra. The loss of the mode at 703 cm −1 coupled with the sharpen- ing of modes at lower wavenumbers shows a significant change in local symmetry upon reduction. The changes are large enough to disturb the long-range order, as observed by the loss in diffraction intensity in the operando XRD. TheRamanspectraofNb 3 O 7 FandthereducedcompoundsareshowninFigure 3.6(b). The modes observed in the pristine material are assigned in Table 3.3. The Raman spectrum of the pristine Nb 3 O 7 F has two modes in the region between 500 and 800 cm −1 , similar to those in NbO 2 F. The mode at approx. 703 cm −1 is similarly attributed to the NbO 6 symmetric stretch. 146 The shoulder at 620 cm −1 is again associated with the bridging Nb-O-Nb stretch, however, the mode is sharper compared to the 620 cm −1 mode in NbO 2 F which could be due to the increased rigidity introduced to the structure by the shear planes in Nb 3 O 7 F. 146 Similarly, the low wavenumber modes are much sharper than those in NbO 2 F indicating less disorder and decreased flexibility due to the restrictions imposed by the added shear plane. Upon reduction to 1.5 V, the modes associated with 59 NbO 6 octahedra and the Nb-O-Nb stretch are largely maintained in addition to the low wavenumber modes. The lack of change between the two vibrational spectra suggest that reduction does little to affect the local symmetry. 200 400 600 800 Raman Shift (cm -1 ) 200 400 600 800 Raman Intensity (a.u.) NbO 2 F Nb 3 O 7 F disch. to 1.5 V disch. to 1 V (a) (b) 620 cm -1 620 cm -1 Figure 3.6: (a): The Raman spectra of pris- tine NbO 2 F as well as NbO 2 F discharged to 1.5 V and 1.0 V, shows the disappearance of several modesfollowedbytheintroductionofnewmodes. (b) The Raman spectra of Nb 3 O 7 F, at various points of discharge, shows better maintenance of the modes seen in the pristine Nb 3 O 7 F. Upon discharging to 1 V, the spectrum maintains a broad mode around 620 cm −1 that is red shifted from the pristine material suggesting that the NbO 6 octahedra are maintained, but the reduced Nb causes a relaxing of the modes. Interestingly, the sharp mode around 620 cm −1 in the pristine and the dis- charged to 1.5 V material either disappears or broadens significantly. As the 620 cm −1 mode is related to the Nb-O-Nb stretch of the corner sharing octahedra, it is unlikely that the mode would disappear and instead we suggest that it is broadening. Furthermore, larger changes are observed in the mode at 703 cm −1 upon discharge of Nb 3 O 7 F to 1V, which is when the triclininc phase is seen to form in operando XRD. We note that there is also a very sharp mode at 128 cm −1 in the Nb 3 O 7 F that blue shifts upon reduction to 1.5 V and red shifts after reduction to 1 V. The assignment of the mode is unknown and so we do not consider this mode further. 60 3.4 Discussion A derivative of NbO 2 F, Nb 3 O 7 Fpresents an interesting comparison on the effects of bonding by introducing a shear plane of edge-sharing NbO 4 F 2 rock salt layers separated by corner sharing octahedra. The crystal structures of NbO 2 Fand Nb 3 O 7 Fwere first characterized by Frevel and Andersson, respectively, whom undertook studies on the synthesis and thermal stability of these com- pounds. 73,142,147,148 It was found that Nb 3 O 7 Fcould be synthesized by the decom- position of NbO 2 F, illustrating the formation of a shear plane by the collapse of a perovskite layer. 148 The close relation between the phases was further examined by Permér and Lundberg, who investigated the chemical lithiation of NbO 2 Fand Nb 3 O 7 Fand their relationship. 138,139,149 The chemical lithiation of NbO 2 Fwas seen to cause a distortion of the cubic structure to the hexagonal phase, LiNbO 3 at higher states of lithiation. 138 We observe greatly improved cycling stability in NbO 2 Fwhen the lower voltage cutoff was limited to 1.5 V preventing the for- mation of Nb 3+ . Given the contraction of the unit cell seen during operando XRD, it is likely that the structure is unable to accommodate the change in bond length necessitated by Nb 3+ , and the large unit cell volume changes result in cracking of particles and electrical isolation that leads to poor reversibility. Therefore, it is advantageous to cycle in a higher voltage regime, avoiding the formation of Nb 3+ . Shear planes have been observed to stabilize perovskite structures by limit- ing distortion, leading to excellent electrochemical cycling stability by resisting cracking during unit cell volume change. The shear plane blocks have been shown to suppress rotational motion. 150 As first observed by Cava et al., the structure of Nb 3 O 7 F does not contain orthogonal shear planes nor the associated struc- tural blocks, such as those found in TiNb 2 O 7 . 83 This was postulated to allow for 61 a twisting motion between sheets to accommodate lithiation; however this flex- ibility ultimately leads to poor reversibility in cycling due to incomplete struc- tural recovery. Further detail can be derived from chemical lithiation experiments, in which Nb 3 O 7 Fwas found to be relatively stable, maintaining symmetry until Li 1.2 Nb 3 O 7 F, followed by a collapse of the structure to hexagonal LiNbO 3 . Chemi- cal delithiation was shown to reform Nb 3 O 7 F, albeit with the inclusion of stacking faults, resulting in greater disorder. 139 A collapse of the corner sharing layers by stacking faults would create a much denser structure, hindering lithium movement and suppressing cycling. The edge sharing layers effectively close a channel in the structure, affecting ion movement through the lattice, while also reducing the crystallographic sites available for lithiation. 83 It is likely that Nb 3 O 7 F forms mainly edge sharing planes after the first cycle, which further decrease Li + channels and result in the loss of capacity observed. Both electrochemical measurements and XRD results show that Nb 3 O 7 Fis nucle- ated after the first charge, likely with an increase in shear layers. The one– dimensional shear planes only interrupt slip planes in a single direction whereas other shear shear structure types interrupt rotation in multiple directions. Additional evidence of rotational disorder is observed in the Raman spectra of Nb 3 O 7 F by tracking the broadening of the mode at 620 cm −1 . A broadening of the mode indicates a broader dispersity of the vibrations, reminiscent of the mode observed in pristine NbO 2 F. Thus, we suggest that the broadening is due to a loss of structural rigidity upon reduction which would suggest that the shear planes are not sufficient to stabilize the structure at 1 V. However, it is important to note that the vibrational spectrum of the reduced Nb 3 O 7 F maintains many of the features in the pristine material with some broadening suggesting that, at least compared to NbO 2 F, the Nb 3 O 7 F structure is more stable during reduction. 62 3.5 Conclusions Through a series of electrochemical and structural studies, we have identified the role of octahedral rotations in destabilizing perovskitic and shear derivative struc- tures towards ion (de)intercalation. In the perovskite NbO 2 F , a contraction of the unit cell, facilitated by twisting of octahedral polyhedra, is induced by reduc- tive Li + insertion eventually, leading to a loss of structural coherence. In order to prevent the destabilization and densification of the material, the Nb 3+ oxidation state must be avoided, as this state causes an unsustainable amount of strain to be accommodated within the structure. Meanwhile, the shear structure Nb 3 O 7 F is seen to be more stable during Li + insertion, as the more rigid edge–sharing shear planes are less susceptible to twist- ing and cracking, leading to the shear planes being maintained during cycling. However, the unique 3×∞×∞ shear structure, where shear planes extend along a single crystallographic direction, allows for movement of the edge sharing sheets as the perovskitic linkages twist and contract during lithiation, similar to what was seen in NbO 2 F. Upon delithiation, mismatch and stacking faults between the shear layers leads to poor cycleability as ion diffusion pathways are closed and Li + sites are lost. Hence, weconcludethatalthoughthestructureofNb 3 O 7 F isslightlystabilized by shear planes, the one-dimensional shear planes are still subject to large degrees of rotational motion. Thus, we suggest that shear planes in orthogonal directions are required to suppress the movement of layers relative to one another. This study provides a valuable design principle for the use of crystallographic shear as a mechanism to stabilize intercalation electrode materials. 63 3.6 Supplemental Information Rietveld Refinement High resolution X-ray diffraction patterns of the as- synthesized NbO 2 F and Nb 3 O 7 F were collected at 11-BM, APS. The resultant patterns were refined against known structures to ensure phase purity of the mate- rials. Relevant results are listed in Table 3.1. Table 3.1: Results of the Rietveld refinement of pristine NbO 2 Fand Nb 3 O 7 Fagainst the synchrotron powder diffraction data. Note that in NbO 2 F , the Nb, O, and F all sit at fixed special positions. Parameter NbO 2 F Nb 3 O 7 F a 3.905(1) 20.676(9) b – 3.834(1) c – 3.927(3) Nb position (0, 0, 0) (0, 0, 0) Nb2 position – (0.183, 0, 0) O position (0.5, 0, 0) 0.5, 0, 0) O2 position – (0, 0, 0.5) O3 position – (0.097, 0, 0) O4 position – (0.706, 0, 0) O5 position – (0.190, 0, 0.5) F position (0.5, 0, 0) (0.5, 0, 0) F2 position – (0, 0, 0.5) F3 position – (0.097, 0, 0) F4 position – (0.706, 0, 0) F5 position – (0.190, 0, 0.5) R Bragg 3.82 4.68 64 Raman Mode Assignments Assignments of Raman modes of spectra dis- played in Figure 3.6. Assignment Raman shift (cm −1 ) Mode description from ref 145,151,152 Literature Measured - 196 193 metal ions inside octahedron - 358 350 Nb(O/F) 6 vibration ν 620 620 Nb-F-Nb stretch or distorted Nb-O-Nb stretch ν s 703 703 Nb-O stretch, Nb-O-Nb stretch ν s 893 870 Nb=O terminal stretch Table 3.2: Vibrational mode assignments of the Raman spectrum of NbO 2 F with descriptions Assignment Raman shift (cm −1 ) Mode description from ref 152,153 Literature Measured - 90 89 NbF 6 -related vibration - 131 128 NbF 6 -related vibration τ 248 254 O=Nb=O twist δ s 299 291 Nb-O-Nb bend - 639 629 Nb-O stretch in corner-sharing octahedra ν s 692 675 NbO 6 symmetric stretch ν s 993 984 Nb=O terminal stretch Table 3.3: Vibrational mode assignments of the Raman spectrum of Nb 3 O 7 F with descriptions 65 Galvanostatic Cycling of NbO 2 F Galvanostatic cycling of NbO 2 F in the voltage range of 1.0−1.5 V was seen to lead to poor reversibility due to irreversible reductiontoNb 3+ , asshowninFigure3.7. Bycalculatingthederivativeofcapacity with respect to voltage, as shown in the inset of Figure 3.7, it is possible to identify two distinct regions of redox with the first region from 1.5− 2.0 V and the second from 1.0− 1.5 V. 0.0 0.5 1.0 1.5 2.0 2.5 3.0 x in Li x NbO 2 F 1.0 1.5 2.0 2.5 3.0 Voltage (V) vs Li/Li + 1.0 2.0 3.0 Voltage (V) vs Li/Li + -8 -6 -4 -2 0 2 d(Q-Q o )/dE (mAh V -1 ) Figure 3.7: Galvanostatic cycling of Li x NbO 2 F in the voltage window of 1.0−3.2 V shows poor reversibility. The inset displays a derivative of capacity with respect to voltage of the first cycle. 66 Galvanostatic Cycling of Nb 3 O 7 F Cycling of Nb 3 O 7 Fwith a voltage range of 1.0−3.2 V shows the differences between the first cycles and subsequent cycles. Figure 3.8 shows two distinct plateaus on discharge, with the insertion of three units of Li + . Upon initial charge, a dip in voltage is seen at 2.5 V. Further cycling shows smooth, featureless curves. 0.0 1.0 2.0 3.0 4.0 x in Li x Nb 3 O 7 F 1.0 1.5 2.0 2.5 3.0 3.5 Voltage (V) vs Li/Li + 0 5 10 Cycle Number 80 120 160 200 Capacity (mAh/g) Figure3.8: GalvanostaticcyclingofLi x Nb 3 O 7 Foverthevoltagerangeof 1.0−3.2 V shows a large irreversible initial capacity followed by cycling with a smooth voltage curve. 67 CyclicVoltammetryofNbO 2 F CyclicvoltammetryofNbO 2 F withavoltage window of 1.0− 3.2 V exhibits a broad oxidative peak due to deep reduction. The broad peaks are seen on both initial and subsequent cycles. 1.0 1.5 2.0 2.5 3.0 Voltage vs Li/Li + (V) -150 -100 -50 0 50 Current (mA/g) Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5 Sweep Rate: 0.1 mV/s Figure 3.9: Cyclic voltammetry of NbO 2 F with a voltage window of 1.0− 3.2 V demonstrates broad oxidative peaks. Cyclic Voltammetry of Nb 3 O 7 F Cyclic voltammetry of Nb 3 O 7 F with a volt- age window of 1.0−3.2 V shows an irreversible peak at 2.5V on the first oxidative cycle. Subsequent cycles show a redox couple centered at 1.75V, which cycles with good reversibility. 1.0 1.5 2.0 2.5 3.0 Voltage vs Li/Li + (V) -60 -40 -20 0 20 40 Current (mA/g) Sweep Rate: 0.1 mV/s Figure 3.10: Cyclic voltammetry of Nb 3 O 7 F with a voltage of 1.0−3.2 V demon- strates two initial oxidative peaks. The peak at 2.5V is lost after the first cycle and replaced with a redox couple at 1.75V. 68 Operando X-ray Diffraction Operando X-ray diffraction (XRD) measure- ments recorded diffracted intensity of both NbO 2 F and Nb 3 O 7 F throughout (de)lithiation in order to track structural changes. In order to better visualize sub- tle variations in intensity, heatmaps were generated in which a colored scale bar represents peak intensity while state of charge is plotted on the Y-axis. Heatmaps corresponding to Figure 5(a) and (b) are shown in Figures 3.11 and 3.13, respec- tively. Upontheintercalationof1.7unitsofLi + , atriclinicphaseisformedinNb 3 O 7 F, asindexedinFigure3.14. Thisphaseismaintainedduringthedischargeandinitial charging until a delithiation of x = 1.9. As shown in Figure 3.15, a large loss of intensity is seen when Nb 3 O 7 F is cycled with the displayed XRD patterns corresponding to the pristine material and the material at the end of a complete discharge/charge cycle. Galvanostatic cycling of NbO 2 F showed good reversibility when the lower volt- age limit was limited to 1.5V, which avoids the formation of Nb 3+ . Operando XRD measurements show that the unit cell contracts and expands during the course of two complete cycles over the higher voltage window (Figure 3.12). In contrast, Figure 5(a) shows a large loss of diffracted intensity when NbO 2 F is lithiated to Li 2 NbO 2 F. 69 Figure 3.11: A heatmap of operando XRD for a complete cycle of Li x NbO 2 F shows a loss of intensity on multiple peaks as long-range crystallinity is lost. Figure 3.12: XRD patterns collected during two complete cycles of NbO 2 F show that peaks shift upwards during discharge and downwards during charge showing the unit cell contracting and expanding. Peak intensity is maintained, showing that the crystallinity of the material is not lost. 70 Figure 3.13: A heatmap of operando XRD for Nb 3 O 7 F demonstrates the differ- ences in behavior during discharge and charge. During discharge, a solid solution type process is observed with steady peak shift throughout lithiation. During charging, growth of Nb 3 O 7 F is observed without the solid solution behavior seen on discharge. 20 30 40 50 Intensity (arb. units) 20 30 40 50 2 �(deg) [ �=1.5406 Å] 1.0 2.0 3.0 Voltage vs Li/Li + (V) 0 1 2 3 2 1 x in Li x Nb 3 O 7 F orthorhombic triclinic Figure 3.14: Indexed unit cells of selectively chosen operando scans show that prior to lithiation, Nb 3 O 7 F is described well in orthorhombic space group Cmmm (bottom). During more complete lithiation (x = 1.7 on discharge to x = 1.9 on charge), the material adopts a triclinic phase (top) before returning to the original orthorhombic phase. 71 20 25 30 35 40 45 2 �(deg.) [ �=1.5406 Å] Intensity (a.u.) Pristine Nb 3 O 7 F Cycled Nb 3 O 7 F Figure 3.15: XRD patterns of pristine and cycled Nb 3 O 7 F show that the original structure is recovered after the first cycle however a large degree of crystallinity is lost. Furthermore, relative peak intensities vary after cycling. Raman Spectroscopy Additional Raman spectra are displayed. 200 400 600 800 10001200140016001800 Raman Shift (cm -1 ) Raman Intensity (a.u.) NbO 2 F Nb 3 O 7 F disch. to 1.5 V disch. to 1 V (a) (b) 620 cm -1 disch. to 1.5 V disch. to 1 V Figure 3.16: The full spectral range of all Raman spectra displayed in Figure 3.6 72 X-ray Absorption Spectroscopy Operando Nb K–edge measurements were used to track the Nb oxidation state as well as local structural changes throughout the (de)lithiation process, as shown in Figures 3.17 and 3.18. 18950 19000 19050 19100 19150 Energy (eV) Normalized ��(E) (a.u.) 18990 19000 19010 19020 Figure 3.17: Nb K–edge X-ray absorption spectroscopy shows the characteristic downshift of the K–edge associated with Nb reduction as NbO 2 F is lithated. 19000 19050 19100 19150 Energy (eV) Normalized ��(E) (a.u.) 19000 19010 19020 19030 19000 19050 19100 19150 Energy (eV) Normalized ��(E) (a.u.) 19000 19010 19020 19030 (a) (b) 18990 19000 19010 19020 Energy (eV) -0.02 -0.01 0.00 0.01 Second Deriv.Normalized ��(E) (a.u.) Pristine Nb 3 O 7 F Lithiated Nb 3 O 7 F Delithiated Nb 3 O 7 F (c) Figure 3.18: The XANES region of the Nb K–edge is seen to shift with redox in Nb 3 O 7 F. The discharge is demonstrated in (a) with an inset displaying the K-edge shift. The pristine material is indicated by circles, while the lithiated material is indicated by squares. The charge region is shown in (b) with circles and squares outlining patterns from the lithiated and delithiated samples respectively. The edge does not shift back completely upon charge, indicative of a partial recovery of the starting material. This is displayed in the second derivative plot in (c), which shows the failure of the delithiated sample to fully match the pristine sample. 73 Chapter 4 Transition Metal Migration Can Facilitate Ionic Diffusion in Defect Garnet Based Intercalation Electrodes 4.1 Introduction Structural transformations triggered by the transport of alkali ions through the solid state can strongly affect the performance of rechargeable Li-ion batter- ies. 117,154 Large changes in the unit-cell volume of active materials leads to degra- dation of electrodes and can eventually fracture the film, creating electrically inac- cessible regions within the cell. 155–157 Similarly, substantial atomic rearrangements can complicate Li-ion diffusion pathways and hinder conductivity. 118 Hence, the performance of intercalation battery electrodes is intimately tied to processes that occur at the atomic length scale. 158 Several charge compensation processes and associated structural changes have been identified which describe a broad range of redox active systems, such as polaron migration and cooperative rotations. The intercalation of Li + into V 2 O 5 is accompanied by simultaneous polaron migration, leading to local structural dis- tortions and commensurate sluggish ion diffusion. 80 Meanwhile, structural anal- ysis shows that the intercalation of Li + into compounds such as ReO 3 is facili- tated by cooperative rotations of polyhedral subunits. 44 Furthermore, studies on Li 2−x Ir 1−y Sn y O 3 show that the substitution of Sn for Ir, and the resulting change in electron count, leads to a different structural change during delithiation, driven 74 74 Adapted from Bashian et al. ACS Enery Lett. 2020 5 1448-1455 [doi] 74 by changes in the energies needed to form anti-site defects to stabilize anionic redox behavior. 159 Thus, by considering the electronic structure of materials, it is possible to improve design rules for accommodating alkali intercalation. To better understand the impact of structural transformations on electrochem- ical performance, we recently investigated the isostructural phases, Fe 2 (MoO 4 ) 3 and Fe 2 (WO 4 ) 3 , which crystallize in the anti-NASICON structure and consists of corner sharing FeO 6 octahedra joined together by tetrahedrally coordinated Mo or W ions. These materials, originally investigated by Manthiram and Goodenough, exhibit highly reversible cycling on the Fe 2+/3+ redox couple and can accommo- date the insertion of up to two full units of lithium or sodium. 160 We demonstrated that this reversibility is facilitated through the cooperative rotation of the rigid tetrahedral subunits which allows Li + or Na + ions to move in and out of the structure while minimizing strain within the framework. 75,76 We later built on the work of Greenblatt, Cava, and Murphy to show that similar octahedral rotations occur in the completely unrelated perovskite structure, which appear to be driven by the need to stabilize Li within interstitial cavities that are too large for ions to become fully coordinated. 44,82,135 The nature of the rotations associated with these rigidsubunits, whicharenecessaryindensely-packedstructures, havewide-ranging impact on multiple facets of electrode performance including Li + migration energy barriers, electrode volume change, and reversibility. 81,161,162 To broaden our understanding to compounds containing redox active elements in tetrahedral, rather than octahedral, coordination environments, we turned our attention to a pair of molybdates with the general composition A 2 (MoO 4 ) 3 where A=Al or Y. While these phases possess homologous compositions to the NASI- CON family mentioned earlier, they instead crystallize in a defect version of the garnet structure where the cubic site is vacant as illustrated in Figure 4.1. 163 The 75 inclusion of redox inactive elements on the octahedral site shifts redox activity to occur on the tetrahedral Mo species. Given the record-setting ionic conductivity achievable in stuffed garnets like Li 7 La 3 Zr 2 O 12 and Li 6 RLa 2 Ta 2 O 12 (whereR=Ca, Sr, or Ba), it would be highly beneficial to elucidate how incorporating redox- active elements into the framework alters the ability of lithium to move through this unique structural topology. 164–166 Figure 4.1: The structure of A 2 (MoO 4 ) 3 in space group Pbcn. Molybdenum tetrahedra and redox-inactive octahedra are shown in green and blue, respectively. The AO 6 octahedra create chains, while vacant dodecahedral sites provide space for lithium intercalation. We used a combination of operando X-ray diffraction and spectroscopictechniques, along with nuclear magnetic reso- nance measurements and com- putational modeling to exam- ine how the structural trans- formations associated with the tetrahedral redox centers in these defect garnets differ from the behavior found for the octahedral sites in the (anti)- NASICON structure. Despite theverysimilarpolyhedralvol- umes and connectivity, we find a stark contrast in the structural evolution during cycling. In particular, both the Al- and Y-based molybdates exhibit a marked loss of long-range crystallinity after the first cycle, but show a significant difference in local structural evolution We find clear evidence that suggests Mo-ions migrate out of tetrahedral and into octahedral positions in the Al-based material during cycling. We rationalize this 76 metal migration as resulting from the increased ionic character compared to the Y-based analogue, necessitated in order to allow the complete reduction of Mo from the 6+ to the 4+ oxidation state. 4.2 Experimental Methods Synthetic Methods Y 2 (MoO 4 ) 3 was synthesized using a modified sol-gel method. Solutions of stoichiometric amounts of Y(NO 3 ) 3 x6H 2 O and Na 2 MoO 4 x 2H 2 O were prepared and stirred together. After centrifugation, a white precipi- tate was isolated from the solution using diethyl ether. The precipitate was then dried under vacuum, ground into a powder, and washed with DI water. The clean powder was calcined in a tube furnace at 900 ◦ C under nitrogen for 5 hours after a 6 hour heating ramp. Upon cooling, this yielded a whitish-gray powder that was ground via mortar and pestle. Al 2 (MoO 4 ) 3 was synthesized by combining stoichiometric amounts of Al 2 O 3 and MoO 3 . Powders were ground together via mortar and pestle and pressed into pellets. The resulting pellets was placed in a crucible and heated for 40 hours at 650 ◦ C, with a 5 hour heating ramp. The compound was subsequently cooled to ambient temperature over a two-hour period and ground to yield a white powder Operando Materials Characterization Operando XRD patterns were col- lected at the Advanced Photon Source (APS), Argonne National Laboratory using the AMPIX electrochemical cell, following the method detailed by Borkiewicz et al. 87 In brief, electrochemical cells equipped with glassy carbon windows were prepared using free standing film electrodes, glass fiber separators soaked in elec- trolytesolution, andLifoilcounterelectrodes. Highresolutionsynchrotronpowder diffractiondata wascollected usingbeamline 17-BM at the APS,Argonne National 77 Laboratory using an average wavelength of 0.24116Å, with a Perkin-Elmer 2D plate detector. XRD collection was performed in transmission geometry through the cell windows. GSAS-II software was used to integrate patterns into the inten- sity vs. 2θ format displayed. 88 Operando X-ray Absorption Spectroscopy (XAS) measurements were performed at beamline 12-BM, APS using the AMPIX electro- chemical cell, equipped with a glass fiber separator and a Li foil combined counter and reference electrode. Scans were collected in transmission mode over a span of 19 minutes with a constant interval of 120 minutes between scans. XAS data proc- cessing was carried out using the ATHENA software of the package IFEFFIT. 143 The EXAFS component was normalized and converted to wavenumber. The resul- tant signal in k-space was multiplied with a k 3 , Fourier-transformed and left with no phase shift correction. All displayed EXAFS data is shown with no phase shift correction. Ab initio calculations on relevant structure models were done using the code FEFF8.2 with fits being performed using the ARTEMIS software of the IFEFFIT package. 143,144 Solid State NMR Measurements Solid-state nuclear magnetic resonance (NMR) spectroscopy was performed on a Bruker Advance III spectrometer at a magnetic field strength of 16.4 T corresponding to Larmor frequencies (ν 0 ) for 7 Li (I = 3/2), 27 Al (I = 5/2), 89 Y (I = 1/2), and 95 Mo (I = 5/2) of 272.1 MHz 182.4 MHz, 34.3 MHz, and 45.6 MHz, respectively. 7 Li and 27 Al, spectra were collected with a 3.2 mm HXY Bruker probe under magic angle spinning (MAS) at a rate of 20 kHz. 89 Y and 95 Mo spectra were recorded in a 4.0 mm HXY Bruker probe at ambient temperature and an MAS rate of 12.5 kHz. All samples were packed in ZrO 2 rotors with a Kel-F cap; lithiated samples of 8–18 mg were center-packed in the rotor between poly(tetrafluoroethylene) ribbon. Single-pulse measurements 78 were performed on all samples; rotor-synchronized Hahn echo spectra were mea- sured alongside the single-pulse spectra for 7 Li and 27 Al. Excitation pulses of π/6 were used for 27 Al and 95 Mo to ensure homogeneous excitation of sites with vari- able nuclear quadrupolar coupling constants. Spectra were referenced as follows: 7 Li—1.0 M LiCl(aq.) at 0 ppm; 27 Al—AlF 3 at –15 ppm; 89 Y—Y 2 O 3 at 273 and 314 ppm; 95 Mo—1.0 M Na 2 MoO 4 (aq.) at 0 ppm. Recycle delays were 15 s for 7 Li, 25 s for 27 Al, 400 s for 89 Y, and 10 s for 95 Mo. The T 1 relaxation of 7 Li and 27 Al were checked to ensure the spectra were quantitative (recycle delay≥ 5 T 1 ). T 1 -filtered 7 Li measurements were performed with a recycle delay of 0.1 s. First principles calculations of the NMR shift and quadrupolar tensors were performed in the density functional theory-based CASTEP code with the Perdew Burke and Ernzerhof (PBE) exchange-correlation functional. 98,167 Geometry opti- mization of the starting structures was performed by relaxing the (i) atomic posi- tions or (ii) atomic positions and lattice parameters until the maximum force was less than 0.01 eV·Å −1 . 168,169 Full shift and quadrupolar tensor calculations were performed on the unrelaxed structure and both relaxed structures. Geometry opti- mization and NMR calculations used Vanderbilt ultrasoft pseudopotentials calcu- lated “on-the-fly” in CASTEP with a plane wave basis set, plane wave energy cut-off of 700 eV, and Brillouin zone sampling with a Monkhorst–Pack grid of k-points finer than 0.04 × 2π Å −1 . 170,171 Core electrons for NMR calculations were reconstructed with the gauge-including projector-augmented wave (GIPAW) method implemented in the code. Electrochemical Characterization The electrochemical performance of the as-prepared materials was characterized using Swagelok-type cells assembled in an 79 argon-filled glovebox, using Li metal as a combined counter and reference elec- trode and Whatman GF/D borosilicate glass fiber sheets as the separator. 1M LiPF 6 inethylenecarbonateanddimethylcarbonate(1:1w/w)wasusedastheelec- trolyte (LP30). Powders of active material were dried under vacuum to remove any residual moisture. Thick film electrodes were prepared by blending 10% graphite powder (300 mesh), 10% acetylene black, 20% polytetrafluoroethylene (average particle size of 1 μm), and 60% active material, and pressed under a hydrostatic pressure of 0.9 tons, with a typical electrode mass of 20 mg. All cell components and electrodes were dried under vacuum at 110 ◦ C for 1 hour before assembly. Galvanostatic cycling was typically performed from 1 V - 3.5 V at a current corre- sponding to a C/20 rate, while cyclic voltammetry was collected with a sweep rate of 0.1 mV/s. Operando XRD patterns and operando XAS spectra were collected using an AMPIX electrochemical cell equipped with two glassy carbon windows. The same glass fiber separators, metallic counter electrodes, and electrolyte solu- tions described previously were used during all operando experiments. Computational Methodology First-principles calculations were performed within the framework of Density Functional Theory using the Vienna ab initio Simulation Package (VASP). 92,94,95,172 Interactions between core and valence elec- trons were described using the Projector Augmented Wave method. Convergence with respect to the plane wave basis set and k-point sampling were tested, with a cut-offenergyof400 eV andk-pointmeshesof Γ-centred 1×2×1and 1×2×2found to be sufficient for Al 2 (MoO 4 ) 3 and Y 2 (MoO 4 ) 3 , respectively. Geometry optimisa- tions were performed using the PBEsol functional, 97 a version of the Perdew Burke and Ernzerhof (PBE) functional 98 revised for solids. PBEsol has previously been shown to reproduce the lattice parameters for a broad range of oxide systems. 44,173 80 Further computational details on the accuracy of the structural optimizations are given in Figures 4.23, 4.22, and Table 4.2. Density of states and band structure diagrams were plotted using the sumo package. 174 In order to correctly describe the electronic structure of the materials, the hybridfunctional, HSE06, 101 wasemployedforbandstructureanddensityofstates calculations. HSE06 combines 75% exchange and 100% of the correlation energies from PBE together with 25% exact Hartree-Fock (HF) exchange at short ranges and has been shown to perform well for oxide materials. 44 4.3 Results and Discussion Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 were prepared using the techniques described in the ExperimentalDetailssection. AmodifiedBellcoremethodwaschosenforelectrode preparation during electrochemical testing, which provided compatibility with the operando characterization methods employed. 175 Owing to their insulating nature, active materials were combined with a blend of carbon sources to ensure good particle coverage and adequate electrical conductivity. The chosen method of elec- trochemical testing allowed for ready comparison between operando characteriza- tion techniques and electrochemistry, creating a more accurate depiction of the behavior of these systems. The electrochemical properties of Al 2 (MoO 4 ) 3 were first evaluated using both cyclic voltammetry and galvanostatic measurements. As seen in Figure 4.2, the first discharge process consists of a relatively long voltage plateau until three for- mula units of Li + have been inserted, at which point a gradually sloping region begins around 1.85V. This sloped region persists up to an additional three equiva- lents of Li + until the cutoff of 1.0 V. On reversal, the voltage curve is seen to follow 81 (a) 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 x in Li x Y 2 (MoO 4 ) 3 1.0 1.5 2.0 2.5 3.0 3.5 Voltage vs Li/Li + (V) 0 5 10 Cycle Number 0 100 200 300 Capacity (mAh/g) (b) 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 x in Li x Al 2 (MoO 4 ) 3 1.0 1.5 2.0 2.5 3.0 3.5 Voltage vs Li/Li + (V) 0 5 10 Cycle Number 0 100 200 300 Capacity (mAh/g) Figure 4.2: The cycling behavior of (a) Y 2 (MoO 4 ) 3 and (b) Al 2 (MoO 4 ) 3 shows initial irreversibility followed by smooth cycling. The last cycles in both materials are outlined in black to illustrate polarization differences. The insets demonstrates the capacity fade associated with irreversibility upon cycling. the same solid-solution type profile, yet despite all efforts to optimize the prepa- ration of the electrodes it was never possible to remove all six of the intercalated lithium. More typically, a maximum of three lithium could be reversibly cycled between1.0Vto3.5V., butasseenintheinsetofFigure4.2(b)thiscapacityslowly fades over subsequent cycles. This lost capacity is correlated with a significant loss of diffracted intensity before and after cycling that can be seen in Figure 4.8 and will be discussed in greater detail later. Within the garnet crystal structure, the pseudo-octahedral position can accommodate up to six lithium while the empty cubic position can theoretically contain an additional three. 176 Given the size of Li, however, it seems unlikely that it would sit on the cubic site, which more typically contains large ions like those of the alkaline earth metals or lanthanides. 177 The cyclic voltammogram of Li x Al 2 (MoO 4 ) 3 mirrors the behavior seen in the galvanostatic cycling (Figure 4.9) with the initial reductive wave reflecting the irreversibility of the first cycle. The rectangular character of the CV loop suggests asignificantcapacitivecontributiontothechargestorage, whereboththeoxidative 82 and reductive peaks are significantly broadened, extending from 1.0V to nearly 3.0V,whichcanbeattributedtothesolid-solutioncharacterofthesecondinsertion process. When taken together, the presence of a single reversible Faradaic feature and the intercalation of six equivalents of lithium per formula unit appears to point towards a reduction of all three Mo ions in the structure from the 6+ to the 4+ oxidation state. Given the insertion of six formula units of Li + , along with the associated multi-electron redox, one can expect a large degree of strain to be introduced, necessitating structural change. Alternatively, Al 3+ ionsmaybeextrudingoutofthestructureandsubsequently platingonthesurfaceoftheelectrodeparticlesatmorereducingpotentials, leaving an amorphous form of MoO 3 that continues to cycle. 178 To evaluate this possibil- ity, we turned our attention to the isostructural Y 2 (MoO 4 ) 3 which substitutes (a) (b) 19900 20000 20100 20200 Energy (eV) 0.3 0.5 0.8 1.0 1.3 Normalized ��(E) (a.u.) 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 x in Li x Al 2 (MoO 4 ) 3 1.0 2.0 3.0 Voltage vs Li/Li + (V) 20000 20025 20050 0.5 0.8 1.0 1.3 Al 2 (MoO 4 ) 3 Y 2 (MoO 4 ) 3 Figure 4.3: Operando XAS measurements were used to track the Mo K-edge in (a) Li x Y 2 (MoO 4 ) 3 and (b) Li x Al 2 (MoO 4 ) 3 during lithiation. The XAS spectra are displayed in the top panels, while the corresponding electrochemical data is displayed on the bottom with color change indicating states of lithiation. The electrochemical reduction of Mo 6+ to Mo 4+ was correlated to a shift in the Mo K-edge to lower energy. A stronger shift and and restructuring occurred in the Mo K-edge of Al 2 (MoO 4 ) 3 , indicative of greater change in the Mo local environment, facilitated by metal migration. 83 extremely stable Y ions for the redox promiscuous Al. As seen in Figures 4.2(a) and 4.10, the insertion of Li into Y 2 (MoO 4 ) 3 exhibits behavior extremely simi- lar to what was observed for Al 2 (MoO 4 ) 3 , which suggests the Faradaic reactions most likely involve electronic states derived from the tetrahedral Mo ions. Yet, to remove all doubt, X-ray absorption spectroscopy (XAS) experiments were used to monitor the K-edge of the Y and Mo in a series of ex-situ samples at different states of lithiation. As expected, the onset of absorption for Y in Y 2 (MoO 4 ) 3 was shifted to lower energy relative to the elemental metal at 17.1keV, consistent with the trivalent oxidation state of the material. More importantly, the X-ray absorp- tion near edge spectroscopy (XANES) region for Y remains essentially constant during lithiation Figure 4.11) reflecting an effectively constant oxidation state for Y while the higher energy EXAFS region clearly shows that the local octahedral coordination environment also remains unchanged (Figure 4.12). In contrast to the Y edge, Figure 4.13 shows EXAFS measurements made on the Mo K-edge of the same ex-situ samples used for the Y K-edge measurement that indicate the characteristic shift associated with the reduction of Mo as a func- tion of lithiation. The Mo K-edge was subsequently monitored during lithiation using an operando technique, which reveals a steady shift from higher to lower energy and evolution of the Mo K-edge in Y 2 (MoO 4 ) 3 , due to the reduction of Mo 6+ to Mo 4+ during lithiation (Figure 4.3(a)). A loss in the pre-edge feature at 20 keV is also seen during lithiation, which is typically only allowed in tetrahe- drally coordinated Mo compounds, as it is symmetry forbidden for octahedrally coordinated compounds. 179 Hence, the loss of intensity in this feature represents a distortion of the starting geometry away from ideal tetrahedral geometry as would be expected as the Mo 6+ is reduced and the bond lengths expand to satisfy local bond valence requirements. 84 While both the Mo K-edge and pre-edge features change in Al 2 (MoO 4 ) 3 , there is a far greater restructuring of the Mo K-edge during lithiation of Al 2 (MoO 4 ) 3 , with a significantly more intense and well-defined peak evolving on reduction, as shown in Figure 4.3(b). This suggested a greater degree of structural evolution in theAl-basedmaterialthatwasexploredusing operando X-raydiffraction(XRD)to directly correlate the electrochemical behavior with the evolution of the long-range average structure. Figure 4.4(a) shows that during the lithiation of Y 2 (MoO 4 ) 3 , no new peak formation is observed but instead a significant loss of intensity is noted throughout the collection process, indicating a decrease in the crystallinity of the active material. While the absence of any new peaks rules out the formation of any new crystalline phases, it cannot preclude the possibility of amorphous products that would be invisible to average scattering techniques. A loss of long range crystalline order is consistent with the results seen in EXAFS, in which local order is maintained but the order of farther shells is reduced during the discharge process for Y 2 (MoO 4 ) 3 . Similar measurements were conducted on the Al 2 (MoO 4 ) 3 system, in which the material was seen to lose crystallinity during the lithiation process. It is seen that Al 2 (MoO 4 ) 3 decomposes more rapidly, becoming mostly amorphous. Additionally, there is no evidence of new peak formation at higher states of charge, which sug- gests that crystalline phases are not formed during the lithiation process. These changes are demonstrated in Figure 4.4(b) which shows the operando XRD data collected during the lithiation of Al 2 (MoO 4 ) 3 . Due to the loss of crystallinty during lithiation, the radial distributions from the Fourier transform of the Mo K-edge EXAFS were examined for information about the short-range structural transformations. 143 Despite exhibiting very sim- ilar electrochemical behavior, the operando EXAFS measurements clearly show 85 that local the coordination environment of the Mo in Al 2 (MoO 4 ) 3 and Y 2 (MoO 4 ) 3 changes in markedly different ways as demonstrated by the radial distribution plots of the Mo K-edge. As seen in Figure 4.4(c), the first coordination shell of Li x Y 2 (MoO 4 ) 3 , ranging from 0Å to 1.6Å and highlighted with red, maintains a relatively constant intensity at shorter radial distances while longer distances 0 1 2 3 4 Radial Distance (Å) | χ(R)| (A -3 ) x = 6.0 x = 3.0 x = 0.0 -3 0 1 2 3 4 Radial Distance (Å) | χ(R)| (A ) x = 6.0 x = 3.0 x = 0.0 (a) (b) (c) (d) Figure 4.4: Heatmaps of peak intensities during electrochemical cycling demon- strates the loss of diffracted intensity throughout the cycling process of Y 2 (MoO 4 ) 3 (a) and Al 2 (MoO 4 ) 3 (b). No new peak formation is observed in either compound, however Al 2 (MoO 4 ) 3 is seen to lose peak intensity more rapidly than Y 2 (MoO 4 ) 3 . The regions of zero intensity (purple) are due to beam outages during data col- lection. In (c), the radial distribution plot in Li x Y 2 (MoO 4 ) 3 obtained from Mo K-edge spectroscopy demonstrating a lack of local structural changes as a function of state of charge. In contrast, in (d), the radial distribution plot in Li x Al 2 (MoO 4 ) 3 exhibits substantial change upon lithiation due to structural rearrangement. The spacing between traces in (c) and (d) represents approximately 0.375 Li + inserted with key values labeled for clarity. 86 exhibit a loss of definition. Maintenance of the peaks and relative intensity in the red region is indicative of the MoO 4 tetrahedra remaining intact throughout lithiation. In contrast, the second shell, associated with the Mo-Y distances and highlighted in blue, shows the merger of several peaks into a single broad feature; this lost intensity can be attributed to a loss of long-range order on lithiation, as tetrahedra undergo short-range rotational disorder and a multitude of scattering paths cancel out. The higher shells exhibit a similar loss of definition in the peaks and significant broadening as the material loses long range order. Unlike the Y- based phase, the local structure of Li x Al 2 (MoO 4 ) 3 undergoes far more significant structural distortions as lithium is incorporated. In particular, the first shell of the radial distribution function appears to change completely with a significant shift of the peaks to higher scattering lengths, as seen in Figure 4.4(d). This suggests the tetrahedral Mo coordination environment is lost as lithiation proceeds for the Al-based phase while the evolution of peaks at longer paths appears to suggest a new Mo environment begins to form. In order to better understand this contrasting behaviors, fits to the radial distribution functions were performed on EXAFS data for the pristine and fully lithiated endpoints of both Al 2 (MoO 4 ) 3 and Y 2 (MoO 4 ) 3 . Fits to the pristine end- members agreed well with the ideal defect garnet starting structure, as shown in Figures 4.14(a) and 4.14(c) respectively. While the lithiated Y 2 (MoO 4 ) 3 , in Figure 4.14(b) shows very few changes, it is clear from Figure 4.14(d) that the lithiated form of Al 2 (MoO 4 ) 3 undergoes a significant distortion that, after analyzing the fits, appearstorepresentatransitionfromtetrahedralMoO 4 tositesmorecharacteristic of octahedral coordination environments, which necessitates a more exaggerated elongation of the Mo–O bonds. 87 Indeed, by adding an octahedral Mo environment, as could be created if Al and Mo were to switch sites within the structure, it was possible to fit the main peaks of the fully lithiated Li 6 Al 2 (MoO 4 ) 3 material, as shown in Figure 4.14(d). As no new crystalline phases are seen in the operando XRD, this new site may correspond to a highly disordered derivative of the parent phase where the Mo has migrated out of the tetrahedra and into an environment like the “stuffed” position found in the lithium rich garnet solid electrolytes like Li 7 La 3 Zr 2 O 12 . 180 As this octahedral site is positioned directly between two faces of neighboring tetrahedra, diffusion of Mo-ions could be achieved with minimal structural rearrangement The possibility of Mo-ion diffusion is also consistent with the relatively small polarization seen during the galvanostatic cycling data, which is not typically characteristic of a conversion mechanism into an amorphous product. While examination of the Al K-edge would be highly beneficial for discriminat- ing between conversion reactions and Mo-ion diffusion, the low X-ray absorption energy of 1.56 keV precludes the collection of useful EXAFS data. Instead, the local coordination of Al was studied using solid-state NMR spectroscopy, which is shown in Figure 4.5. It can be seen that Al 2 (MoO 4 ) 3 exhibits several overlapping resonances between –9 and –14 ppm that agree well with previous measurements (Figure 4.15). 181,182 DFT-based calculations on the Al-based material predict a shift range of 4 ppm and nuclear quadrupolar coupling constants (C Q ) of 1.2± 0.6 MHz, in agreement with the observed spectrum. Additionally, a broad resonance at 16.0 ppm was seen, which previous 27 Al NMR investigations of Al 2 (MoO 4 ) 3 by Kunath-Fandrei et al. also observed. 181,183 We attribute this signal to a six– coordinate AlO 6 environment, possibly formed by defects in Al 2 (MoO 4 ) 3 in which Al occupies the “stuffed" octahedral position between adjacent Mo tetrahedra. 88 80 40 0 60 20 –20 δ 27 Al (ppm) 1 st charge 2 nd dis. 1 st discharge Li1.0 Li3.0 Li5.0 Li6.5 Li5.5 Li4.5 Li3.0 Li4.5 Li5.5 Al 2 (MoO 4 ) 3 80 40 60 δ 27 Al (ppm) Figure4.5: 27 AlNMRspectraofLi x Al 2 (MoO 4 ) 3 duringthefirstdischarge/charge cycles show the resonances from Al 2 (MoO 4 ) 3 between -9 and -14 ppm; the initial defects assigned to AlO 6 and the Li x Al 2 (MoO 4 ) 3 signal that forms on discharge and is assigned to octahedral Al moving to the “stuffed" position at 16 ppm; a second Al environment from Li x Al 2 (MoO 4 ) 3 that is assigned to a four–coordinate site in the garnet at ca. 70 ppm. Expanded views of the region from 30 to 90 ppm are provided in the insets to the left. Crucially, this peak, along with the other structural peaks are seen to modulate during cycling. As lithium is inserted into the structure, the 27 Al NMR signal for Al 2 (MoO 4 ) 3 , seen in Figure 4.5, irreversibly broadens. The negatively shifted resonance almost disappears as lithiation increases to Li 3 Al 2 (MoO 4 ) 3 ; it is replaced by a resonance centered at 14.8 ppm. This new resonance may indicate Al movement within the structure, possibly to the “stuffed” octahedral garnet site. This is consistent with the probable octahedral defect at 16.0 ppm in the pristine structure. We note the smaller ionic radii of Al 3+ in comparison to that of Y 3+ would allow for more facile ionic migration. The original signal from Al 2 (MoO 4 ) 3 is completely absent 89 at the nominal compositions of Li 5 Al 2 (MoO 4 ) 3 and Li 6.5 Al 2 (MoO 4 ) 3 and a broad new resonance appears around 70 ppm. The 27 Al shift of this signal is consistent with four–coordinate aluminum. Furthermore, the shift (75± 5 ppm) and C Q (6± 1 MHz) are entirely consistent with aluminum in the four–coordinate garnet site (Figure 4.16). 184 The large quadrupolar coupling constant of this environment indicates highly distorted aluminum coordination, which is consistent with the diffraction showing an increase in local structural disorder on lithium insertion. If a conversion reaction were occurring, the most likely product would involve the deposition of Al nanoparticles. Al metal resonates at 1640 ppm, giving rise to a sharp signal under MAS; examination of this region in Li x Al 2 (MoO 4 ) 3 showed only spinning sideband intensity from the satellite transitions of the new lithiated phase but no new signals corresponding to the formation of Al metal are visible in Figure 4.16 and Figure 4.17. 185,186 Furthermore, the normalized 27 Al NMR signal in both the fully discharged and charged material is essentially constant, indicating that Al is not removed from the structure. On charging, little change is observed for Li 5.5 Al 2 (MoO 4 ) 3 or Li 4.5 Al 2 (MoO 4 ) 3 . At the top of charge, upon removal of 3.5 Li to Li 3 Al 2 (MoO 4 ) 3 , the broad peak at ca. 70 ppm shifts toward lower frequency while the signal at 14.8 ppm is unchanged, as seen in Figure 4.5. The negatively shifted 27 Al resonances of pristine Al 2 (MoO 4 ) 3 do not reappear after charging, indicating the irreversibility of the structural transformation. No apparent changes occur on the second discharge. From this evolution of the 27 Al NMR spectra, it is clear that the aluminum local environment is irreversibly altered from the Al 2 (MoO 4 ) 3 host during the charging process, though no conversion to Al metal nanoparticles is observed. In parallel, the insertion of Li-ions was followed with 7 Li NMR, shown in Figure 4.18, but the broad signal centered at –1 ppm neither changedpositionnorlineshapefrominitiallithiationthroughfulldischarge, charge, 90 Figure 4.6: Computed density of states for both Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 . Note the introduction of Yd-Op states at the bottom of the conduction band. or second discharge. 89 Y and 95 Mo spectra of the host structures are also shown in Figures 4.19 and 4.20. In order to evaluate the character of the bonding in the two phases and look for a potential origin for difference in the local structural changes, Hybrid Density Functional Theory (DFT) was employed. Both the density of states (DOS) and band dispersions were calculated, as shown in Figure 4.6 and Figure 4.22 respec- tively. The DOS for both materials exhibits large band gaps on the order of 5eV with the top of the valence band dominated by Op states while the bottom of the conduction band for both phases contain unoccupied hybrid Mod-Op states. The most significant difference between the two materials is the presence of Yd-Op states at the bottom of the conduction band, which result from the greater orbital overlap between the larger Y 3+ ions and their oxygen ligands. This increased covalent character of the Y-based phase likely explains the retention of the local structure during cycling given the Y-ions would be far less mobile due to the strong 91 bonds. This can be contrasted against the more ionic character of the bonds in Al 2 (MoO 4 ) 3 , which allows Al to rearrange during cycling. Figure 4.7: Differences in bonding between Y 2 (MoO 4 ) 3 (a) and Al 2 (MoO 4 ) 3 (b) lead to alter- nate structural distortions upon ion intercalation. Taking these results as a whole, the defect garnet struc- tureexhibitsaremarkableabil- ity to incorporate six equiva- lents of Li per formula, which corresponds to the complete reduction of Mo 6+ to Mo 4+ and two electrons per tran- sition metal. Unfortunately, it appears that the number of Li that can be reversibly (de)inserted is restricted to three, or one per Mo site, which may be related to the significant loss of crystallinity during the initial intercalation. Despite this loss of long-range order, the operando EXAFS measurements clearly show the increased covalent character of the Y-based phase preserves the local coordination environment of the polyhedral subunits. In contrast, the predominantly ionic char- acter of the octahedral sites in Al 2 (MoO 4 ) 3 results in a redistribution of Al within the structure during the first cycle, which appears to decrease the polarization of the voltage profiles during (de)insertion of Li. We note the similarities with the metal migration proposed by Radin et al. as a charge compensation mecha- nism during the delithiation of lithium-rich manganese oxides, wherein manganese migrates between tetrahedral and octahedral sites. 187 The intercalation of multiple alkali units, along with changing oxidation states necessitates structural changes to 92 accommodate strain, however electronic structure differences lead to multiple pos- sible behaviors. The different mechanisms observed in the A 2 (MoO 4 ) 3 (A= Y,Mo) system are further illustrated in Figure 4.7, which demonstrates the difference in structural change upon lithiation. 4.4 Conclusions In summary, the defect garnet polymorph of theA 2 (MoO 4 ) 3 family has been stud- ied as a platform for understanding the structural distortions that occur during the two electron reduction of tetrahedrally coordinated Mo 6+ ions. We have found that the increased ionic character of Al on the A site allows Mo to migrate out of its tetrahedral site and mix into the otherwise vacant “stuffed” octahedral position. This is most likely driven by the difficulty associated with distorting the tetrahe- dral position in the structure in order to accommodate the lower oxidation state of the tetravalent Mo in the fully intercalated end member. It further highlights the importance of using a diverse set of local structural probes to study systems that lose long-range coherence in their periodic structure during cycling as a way to characterize and understand the complex atomic rearrangements that accompany ionic transport. 188 This work provides critical insight into one of the many ways that extended solids transform during the intercalation of alkali ions like lithium and may help point the way towards accessing the full reversible capacity. 93 4.5 Supplemental Information Change of Intensity in XRD. As shown in Figure 4.8, the intensity of diffracted peaks is greatly reduced upon lithiation of Al 2 (MoO 4 ) 3 . Conversely, the diffuse background increases, indicating an increase in incoherent scattering from amorphous regions. Similar behavior is observed in Y 2 (MoO 4 ) 3 where peaks uniformly loose intensity. 2.0 4.0 6.0 8.0 10.0 12.0 2 θ (deg.) [ λ=0.24116 Å] Intensity (a.u.) Initial (Pre-lithiation) Final (Post-lithiation) Figure 4.8: Upon lithiation, an increase in the diffuse background and a decrease in peak intensity is observed in Li x Al 2 (MoO 4 ) 3 which is associated with the for- mation of amorphous products. 94 Cyclic Voltammetry of Li x Al 2 (MoO 4 ) 3 . Figure 4.9 demonstrates the large irreversible capacity associated with the initial lithiation of Li x Al 2 (MoO 4 ) 3 . A substantial decrease in the charge passed is observed between the first and second reductive cycles, while the broad featureless peaks are maintained. 1.0 1.5 2.0 2.5 3.0 3.5 Voltage vs Li/Li + (V) -0.8 -0.6 -0.4 -0.2 0.0 0.2 Current (mA) Cycle 1 Cycle 2 Cycle 3 Figure 4.9: Cyclic voltammogram of Al 2 (MoO 4 ) 3 with 3 cycles against Li metal in a two electrode Swagelok cell. The first cycle differs substantially from the following cycles. 95 Cyclic Voltammetry of Li x Y 2 (MoO 4 ) 3 . Li x Y 2 (MoO 4 ) 3 has more Faradaic behavior as seen in Figure 4.10 which shows the sharper peaks present in a cyclic voltammogram. The intercalation based charge storage, and associated polyhedral rotations result in sharper features and less broad peaks, when compared to Figure 4.9. 1.0 1.5 2.0 2.5 3.0 3.5 Voltage (V) vs Li/Li + -1.6 -1.2 -0.8 -0.4 0 0.4 Current (mA) Cycle 1 Cycle 2 Cycle 3 Figure 4.10: The Y 2 (MoO 4 ) 3 cyclic voltammogram shows Faradaic peaks asso- ciated with lithation. The first cycle shows a larger area due to irreversibility. Y K-edge XAS Measurements. The Y K-edge was measured on a series of samples of Li x Y 2 (MoO 4 ) 3 and the radial distribution was calculated from the Fourier transform of the k-space data, in Figure 4.11 and 4.12 respectively. In the radial distribution function, peaks in higher shells are seen to change as the structure is rearranged during ion insertion. The displayed data is not phase shifted. 96 17000 17200 17400 17600 17800 Energy (eV) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Normalized χμ(E) (a.u.) Y 2 (MoO 4 ) 3 Li 1 Y 2 (MoO 4 ) 3 Li 2 Y 2 (MoO 4 ) 3 Li 3 Y 2 (MoO 4 ) 3 17030 17040 17050 17060 17070 0.6 0.8 1.0 1.2 Figure 4.11: XAFS measurements on the Y K-edge in ex-situ samples of Li x Y 2 (MoO 4 ) 3 show a constant K-edge energy despite different states of charge. 0.0 2.0 4.0 6.0 Radial Distance (Å) | χ(R)| (Å -3 ) Y 2 (MoO 4 ) 3 Li 1 Y 2 (MoO 4 ) 3 Li 2 Y 2 (MoO 4 ) 3 Li 3 Y 2 (MoO 4 ) 3 Figure 4.12: The radial distribution function as obtained from the Fourier trans- form of Y K-edge data collected on Y 2 (MoO 4 ) 3 . The peak at 1.8 Å represents the Y-O distance within the YO 6 octahedra and remains constant at various states of charge. 97 Comparison of Mo K-edge to Y K-edge XAS Measurements. The Y K- edgewasmeasuredforaseriesofY 2 (MoO 4 ) 3 samplesatvariousstatesoflithiation, which demonstrated that the Y K-edge did not shift due to the redox inactivity of Y in this structure. For comparison, the Mo K-edge was measured on the same set of samples, after the Y K-edge measurements. As shown in figure 4.13, the Mo K-edge showed a substantial shift and restructuring, analogous to that which was observed in the operando measurements performed on Y 2 (MoO 4 ) 3 . Operando measurements were not performed on the Y K-edge due to beam time constraints. 20000 20020 20040 20060 20080 20100 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Norm. χμ(E) (a.u.) Y 2 (MoO 4 ) 3 Li 1 Y 2 (MoO 4 ) 3 Li 2 Y 2 (MoO 4 ) 3 Li 3 Y 2 (MoO 4 ) 3 20000 20020 20040 20060 20080 20100 Energy (eV) 0.00 0.02 0.04 0.06 Norm. dx μ/dE Figure 4.13: The Mo K-edge in Li x Y 2 (MoO 4 ) 3 shows a shift to lower energies corresponding to the reduction of Mo 6+ during discharge. This change matches the results seen in operando measurements on the same system. 98 EXAFS Radial Distribution Fits. Radial distribution fits were performed on Mo K-edge EXAFS data of the pristine and fully lithiated endpoints of both Al 2 (MoO 4 ) 3 and Y 2 (MoO 4 ) 3 . Fits of the pristine Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 agreed well with their starting structure, as shown in Figure 4.14(a) and 4.14(c) respectively. It was seen that the initial Al 2 (MoO 4 ) 3 structure decayed upon lithi- ation, while new scattering lengths increased in intensity throughout discharge. As the EXAFS data is compromised mainly of the first atomic shell about the Mo atoms, this change in signal represents a marked rearrangement in the struc- ture. This rearrangement represents a transition from the initial tetrahedral MoO 4 environment to an octahedral geometry (MoO 6 ) about the scattering Mo atoms, which results in an elongation of the Mo-O bonds. It is clear that the rearrange- ment induced by lithiation of Al 2 (MoO 4 ) 3 leads to the formation of new species, as Mo migrates from the tetrahedral site. In comparison, the lithiation of Y 2 (MoO 4 ) 3 leads to more nuanced changes, especially in the first shell region centered around 1.25Å asshowninFigure4.4(c). ItwasfoundthatEXAFSradialdistributiondata of the fully lithiated Li 6 Y 2 (MoO 4 ) 3 structure, while more disordered, fits well with the structure of the pristine material, as shown in Figure 4.14(b). In Y 2 (MoO 4 ) 3 , the Mo local environment is better maintained throughout the lithiation process. 99 1.0 2.0 3.0 4.0 Radial Distribution (Å) 0.0 5.0 10.0 15.0 20.0 | �(R)| (Å -4 ) Measured Fit Window 1.0 2.0 3.0 4.0 Radial Distribution (Å) 0.0 5.0 10.0 15.0 | �(R)| (Å -4 ) Measured Fit Window 1.0 2.0 3.0 4.0 Radial Distribution (Å) 0.0 5.0 10.0 15.0 20.0 | �(R)| (Å -4 ) Measured Fit Window 1.0 2.0 3.0 4.0 Radial Distribution (Å) 0.0 2.0 4.0 6.0 | �(R)| (Å -4 ) Measured Fit Window (a) (b) (c) (d) Figure 4.14: Fits of EXAFS radial distribution data for both Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 in their pristine (a, c) and fully lithiated states (b, d) using the models described in the text. 100 Solid State NMR Measurements. Solid state NMR measurements of the 27 Al, 7 Li, 95 Mo, and 89 Y nuclei were used to provide information on local coordi- nation environments and structural changes in Al 2 (MoO 4 ) 3 and Y 2 (MoO 4 ) 3 . To better characterize the structure, and out of interest in building our understanding of solid-state inorganic NMR, 89 Y and 95 Mo NMR spectra of the host structures were measured. 89 Y suffers from long relaxation times and thus it was not possible to record spectra on the relatively small amount of lithiated sample in a way that would have been analogous to 27 Al in Figure 4.5. In addition to the low receptivity of 95 Mo, it is generally not possible to measure NMR spectra directly on a redox active transition metal outside of its diamagnetic state(s) due to extremely rapid paramagnetic relaxation. 100 0 –100 δ 27 Al (ppm) 40 20 0 –20 –40 * * * * * * * * * * * * 15 kHz MAS 20 kHz MAS 20 kHz MAS, isotropic Figure 4.15: 27 Al solid-state NMR spectra of Al 2 (MoO 4 ) 3 at different MAS speeds. Spinning sidebands move with MAS speed and are denoted with asterisks. Spectra were recorded at 16.4 T. The 95 Mo spectra of Al 2 (MoO 4 ) 3 and Y 2 (MoO 4 ) 3 and 89 Y spectrum of Y 2 (MoO 4 ) 3 are expected to show six distinct Mo envi- ronments of equal population in Al 2 (MoO 4 ) 3 and two Mo environments in a 2:1 ratio as well as a single Y environment in Y 2 (MoO 4 ) 3 . The 95 Mo spectrum of Al 2 (MoO 4 ) 3 is consistent with the structure model in P2 1 /a if the signals at –280 and –325 ppm are comprised of two overlapping quadrupolar resonances. The calculated shift range 101 of 7 ppm and C Q values of 0.7–1.8 MHz are consistent with the spectrum. On the other hand, the 95 Mo and 89 Y spectra of Y 2 (MoO 4 ) 3 suggest that multiple polymorphs or phases may be present because there are multiple 89 Y and 95 Mo signals. DFT calculations of the 89 Y chemical shielding Y 2 (MoO 4 ) 3 were calibrated into a chemical shift by linear fitting of the shielding/shift relationship for 20 different yttrium environments in 12 yttrium-containing oxides, see Figure 4.21 and Table 4.1. The calculations predict an 89 Y chemical shift of –38.1 ppm, in reasonable agreement with the strongest 89 Y resonance in Figure 4.20 at –52 ppm. 2000 1000 0 –1000 –2000 δ 27 Al (ppm) Al site 2 Al site 1 Experimental Figure 4.16: Fitted 27 Al NMR spectrum of Li 5 Al 2 (MoO 4 ) 3 at 16.4 T and 20 kHz MAS showing deconvolution into two sites. Site 1, at 16.0± 0.5 ppm, has a C Q of 2.4(1) MHz and η of 0.15(5) while site 2, at 75± 5 ppm, is broad with a large C Q of 6(1) Mhz. 102 δ 27 Al (ppm) 1700 1600 1500 Li 5.5 Al 2 (MoO 4 ) 3 2 nd dis. Li 4.5 Al 2 (MoO 4 ) 3 2 nd dis. Li 3.0 Al 2 (MoO 4 ) 3 1 st ch. Li 4.5 Al 2 (MoO 4 ) 3 1 st ch. Li 5.5 Al 2 (MoO 4 ) 3 1 st ch. Li 6.5 Al 2 (MoO 4 ) 3 1 st ch. Li 5.0 Al 2 (MoO 4 ) 3 1 st dis. Li 3.0 Al 2 (MoO 4 ) 3 1 st dis. Li 1.0 Al 2 (MoO 4 ) 3 1 st dis. Al 2 (MoO 4 ) 3 fit Al 2 (MoO 4 ) 3 experimental Figure 4.17: Al metal region of the 27 Al NMR spectra of Li x Al 2 (MoO 4 ) 3 during the first 1.5 discharge/charge cycles. The observed signals are spinning sidebands arising from the satellite transitions of the isotropic resonances of Al 2 (MoO 4 ) 3 , the AlO 6 -like defect, and Li x Al 2 (MoO 4 ) 3 centered at –12 ppm, 16 ppm, and 14 ppm, respectively. The purple highlighted region denotes where the signal from Al metal would appear. Note that the intensity of these peaks represents < 0.5% of the total intensity (compare to Figure 4.5) and thus not even a subpercent Al metal signal is observed. Spectra were recorded at 16.4 T and 20 kHz MAS. 50 0 –50 * δ 7 Li (ppm) * Li 1.0 Al 2 (MoO 4 ) 3 Li 3.0 Al 2 (MoO 4 ) 3 Li 5.0 Al 2 (MoO 4 ) 3 Li 6.5 Al 2 (MoO 4 ) 3 Li 5.5 Al 2 (MoO 4 ) 3 Li 4.5 Al 2 (MoO 4 ) 3 Li 3.0 Al 2 (MoO 4 ) 3 Li 4.5 Al 2 (MoO 4 ) 3 Li 5.5 Al 2 (MoO 4 ) 3 2 nd discharge 1 st charge 2 nd discharge 1 st charge 1 st charge 1 st discharge 1 st discharge 1 st discharge 1 st discharge Figure 4.18: 7 Li NMR spectra of Li x Al 2 (MoO 4 ) 3 during the first 1.5 dis- charge/charge cycles. No changes in shift or lineshape were observed as a function of state-of-charge nor as a function of T 1 -filtering experiments with a short recycle delay (latter not shown). Spectra were recorded at 16.4 T and 20 kHz MAS. 103 –150 –200 –250 –300 –350 δ 95 Mo (ppm) Al 2 (MoO 4 ) 3 Y 2 (MoO 4 ) 3 Figure 4.19: 95 Mo spectra of Al 2 (MoO 4 ) 3 (blue) and Y 2 (MoO 4 ) 3 (black) at 16.4 T and 12.5 kHz MAS. 0 500 1000 –500 –1000 –52 ppm 12 ppm 97 ppm δ 89 Y (ppm) Figure 4.20: 89 Y spectrum of Y 2 (MoO 4 ) 3 at 16.4 T and 12.5 kHz MAS. Figure 4.21: Experimental isotropic chemical shift (δ iso ) vs. calculated isotropic chemical shielding (σ iso ) for 20 different 89 Y sites in the 12 yttrium oxide com- pounds listed in Table 4.1. 104 Compound Space Group ICSD No. Site σ iso (ppm) δ iso (ppm) Y 2 O 3 Ia ¯ 3 66243 1 2331.55 314 2 2377.17 273 Y(OH) 3 P6 3 /m 200098 1 2535.03 66 α-Y 2 Si 2 O 7 P ¯ 1 164148 1 2496.64 132.9 2 2543.70 95.1 3 2459.61 170.8 4 2587.46 37.7 β-Y 2 Si 2 O 7 C2/m 281312 1 2427.69 207.3 δ-Y 2 Si 2 O 7 Pnam 33721 1 2514.29 121.1 Y 2 Sn 2 O 7 Fd ¯ 3mZ 74706 1 2485.04 150 Y 2 Ti 2 O 7 Fd ¯ 3mZ 14242 1 2624.15 65 Y 4 Al 2 O 9 P2 1 /c 51076 1 2436.30 195 2 2410.92 231 3 2446.47 184 4 2426.03 216 YAlO 3 Pnma 4115 1 2384.18 215 χ-Y 2 SiO 5 I2/a 28021 1 2401.61 237.1 2 2471.41 149.5 Y 3 Al 5 O 12 Ia ¯ 3d 20090 1 2388.06 222 YScO 3 Pbnm 237285 1 2362.06 263.1 Table 4.1: Table of 89 Y computed chemical shielding vs. experimental chemical shift for a series of yttrium oxide crystals. 105 Computed Band Diagrams of Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 . Density functional theory was used to calculate the band diagrams of Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 as shown in Figure 4.22. Both materials possess large band gaps on the order of 5 eV, with flat bands suggestive of low electron mobility. The intro- duction of Y 4d states in Y 2 (MoO 4 ) 3 increases the covalency of the material in comparison to Al 2 (MoO 4 ) 3 . Figure 4.22: The calculated band diagrams of both Y 2 (MoO 4 ) 3 and Al 2 (MoO 4 ) 3 . 106 Geometry Optimization and Brillouin Zones. Structural optimisations were deemed converged when the sum of all forces on each atom totalled less than 10 meVÅ −1 . The lattice parameters of Al 2 (MoO 4 ) 3 and Y 2 (MoO 4 ) 3 optimised using the PBEsol functional are provided in Table 4.2. In general, the results show good agreement with experiment, with all values within 1 % of the experimentally determined parameters. Compound a / Å b / Å c / Å α / ◦ β / ◦ γ / ◦ Al 2 (MoO 4 ) 3 15.472 (+0.60) 9.090 (+0.49) 17.959 (+0.40) 90 (0) 125.3 (−0.04) 90 (0) Y 2 (MoO 4 ) 3 14.001 (+0.96) 9.989 (+0.55) 10.103 (+0.82) 90 (0) 90 (0) 90 (0) Table 4.2: Lattice parameters for Al 2 (MoO 4 ) 3 and Y 2 (MoO 4 ) 3 calculated using the PBEsol functional. Difference versus experiment in percent given in parenthe- ses k z k y k x Z R X Γ U S Y T Figure 4.23: Brillouin zone for the Pbcn space group, indicating all high- symmetry k-points. The coordinates of the high symmetry k-points are: Γ = (0, 0, 0); Y = ( 1 2 , 0, 0); X = (0, 1 2 , 0); Z = (0, 0, 1 2 ); U = (0, 1 2 , 1 2 ); T = ( 1 2 , 0, 1 2 ); S = ( 1 2 , 1 2 , 0); R = (0, 1 2 , 1 2 ) . 107 Chapter 5 Electrochemical Oxidative Fluorination of an Oxide Perovskite 5.1 Introduction The ability to topotactically control the insertion of ions into crystalline hosts is critical to a range of technologies, including energy storage, electrochromic dis- plays, superconductivity, and catalysis. 17,189,190 While research on cationic inter- calation chemistry has flourished, 28 anionic intercalation has largely been over- looked 191 with most work in the area focused on bulky polyanionic groups inserted into layered materials. 192 Given their substantially larger radius and negative charge polarity, anions require drastically different conditions from cations to pro- mote fast ionic mobility. 46,193 The earliest work on anion insertion chemistry was performed by Schafhaeutl and then built upon by Rüdorff and Hofmann when they showed that SO 2− 4 could intercalate between the sheets of graphite in the presence of a chemical oxidant. 57,194 Electrochemical intercalation into crystalline graphite was not achieved until the early 1980’s using slow cyclic voltammetry to drive the ionic diffusion of species such as ClO − 4 , SO 2− 4 , and BF − 4 . 195,196 Mallouk and Bartlett later described the chemical insertion of fluoride in graphite using HF, identifying the formation of biflouride C 12 HF 2 . 197 Carlin then built on this work to develop a symmetric graphite cell that intercalated bulky ions like imidazolium at the cathode and AlCl − 4 at the anode to create one of the first embodiments of a dual-ion battery. 198 108 Cation Intercalation Reductive Insertion Oxidative Deinsertion Anion Intercalation Oxidative Insertion Reductive Deinsertion Figure 5.1: Combining anion and cation insertion processes could break gravi- metric capacity limits by allowing for multi-electron redox processes. While the intercalation chemistry of bulky polyanionic groups have received some attention, 65,199 the electrochemistry of smaller species like fluoride has been mostlylimitedtoconversion-basedsystemswhereareversiblechemicaltransforma- tion is used. 200–202 A critical challenge facing these systems is that they rely almost exclusively on solid-state fluoride conductors, like Ba-doped LaF 3 , 61,203,204 sand- wiched between metal and metal-fluoride electrodes. 62,205 Fichtner and coworkers, for example, have demonstrated a F-ion battery consisting of a CuF 2 cathode cycled against a film of La metal which results in the reduction of the cathode to Cu metal and a conversion to LaF 3 on the anode side. 60 More recently, Clemens and coworkers have leveraged all solid-state cells, showing some success with inter- calating fluoride into La 2 CoO 4 , though the poor conductivity of the LaF 3 solid electrolyte required operating cells at 170 ◦ C. 67,206 This was very recently followed by the work of Banerjee et al. who demonstrated chemical insertion and removal of fluoride from FeSb 2 O 4 at room temperature using chemical redox methods. 207 109 Thehighoperatingtemperatureofthesesolidsprecludestheiruseinanypracti- cal devices, which would require a fluoride electrolyte with high mobility at room- temperature. Christe and others reported one of the earliest advances towards such an electrolyte when they reported that quaternary ammonium salts can sta- bilize free fluoride in THF. 208,209 More recently, Davis et al. showed that salts of N,N,N-trimethyl-N-neopentylammonium fluoride (Np1F) dissolved in fluoroether solvents could facilitate the stable cycling of CuF 2 @LaF 3 core-shell nanoparticles as conversion cathodes. 66 This seminal work represented some of the first tangible evidence that electrochemical energy storage can leverage anions like fluoride in similar ways to lithium. Drawing inspiration from this extensive work in the literature, we sought to explore the fundamental structural requirements that facilitate fluoride ion mobil- ity in the solid state. Our recent work with ReO 3 , in which we studied the fun- damental structural distortions that occur during lithiation, 44 offered a natural starting material for these studies given that the A-site vacancy within perovskite provides an obvious interstitial where the fluoride could intercalate. Demonstrated in Figure ??, dual cation and anion insertion could provide an exciting break- through to increase capacity limits. Furthermore, ReO 3 is intrinsically metallic, which removes the need for carbon coating and drastically simplifies the interpre- tation of spectroscopic data. In this contribution, we demonstrate the first observation of oxidative interca- lation of fluoride into an oxide host from a liquid fluoride electrolyte at room tem- perature. We find that the negative charge and larger ionic radii of F − demand a vastly different mechanism for intercalation compared to what was observed during Li insertion. Using complementary spectroscopic and structural tools, combined 110 Figure 5.2: (a) The structure of ReO 3 with orange oxygen spheres and maroon rhenium-centered octahedra, shown along [001]. (b) The structure or ReO 3 is displayed along the [111] using the ionic radii of Re and O, emphasizing the tight packing of anions. withoperando electrochemicalcharacterization, weshowthatfluoride-ionsincorpo- rate onto the A-site of ReO 3 during cycling, which creates a highly unstable phase that rapidly decomposes. While this decomposition prevents reversible cycling, it also provides insight into a completely new mechanism for fluoride intercalation under relatively mild and tunable electrochemical conditions and provides insight into how changing the host composition could stabilize F-ions on the A-site. 5.2 Experimental Methods Materials Synthesis. Nanoparticles of ReO 3 were prepared using a method previously employed by our group. 44,86 In a typical preparation, a 0.1 M solution was prepared by dissolving Re 2 O 7 in methanol, and then evaporating on a glass plate that was heated at 250 ◦ C for 5 minutes in a convection oven, until the methanol had completely evaporated. The reduction of Re 2 O 7 produced ReO 3 , which was deposited as a brick red film on the glass plate. The film was removed from the plate and ground via mortar and pestle to yield a deep red metallic 111 powder. The as-prepared samples were dried under vacuum at 110 ◦ C before electrochemical cell preparation. Materials Characterization. Laboratory X-ray powder diffraction (XRPD) patterns were collected on a Bruker D8 Advance diffractometer with a Cu K α source, equipped with a LynxEye XE-T detector. Neutron total scattering mea- surements of ex situ samples was performed at the NOMAD beamline at the Spal- lation Neutron Source at Oak Ridge National Laboratory. 210 Data were normalized against a vanadium rod, the background was subtracted and the total structure factor were transformed into pair distribution function (PDF) data using specific IDL codes developed for the NOMAD scattering instrument with a Q range of 0-25 Å and a pseudo-voigt polynomial H correction. Operando synchrotron X-ray powder diffraction (SXRD) patterns were col- lected at the Advanced Photon Source, Argonne National Laboratory using the AMPIX electrochemical cell, following the method detailed by Borkiewicz et al. 87 SXRD collection was performed in transmission geometry through the cell win- dows,usingamonochromatedX-raybeamwith(λ=0.25463Å)andaPerkinElmer 2D Plate Detector. GSAS-II software was used to integrate patterns into the inten- sity vs. 2θ format displayed. 88 sXASandXESmeasurementswereperformedattheiRIXSendstationatbeam- line 8.0.1 at the Advanced Light Source. Absorption spectra were collected in total electron yield (TEY) and total fluorescence yield (TFY) modes. A TiO 2 reference was used for excitation energy calibration of the O K-edges The emission energies was calibrated using F 1s and O 1s core level energies for F K- and O K-edges 112 respectively. F 1s and O 1s core levels were obtained by X-ray Photoelectron Spec- troscopy (XPS) measurements performed using a Phi VersaProbe 5000 system at the Analytical and Diagnostics Laboratory (ADL), Binghamton University. Transmission Electron Microscopy (TEM) images were collected on a JEOL JEM2100F with an operating voltage of 200 kV. Samples for TEM analysis were prepared by drop-casting a suspension of the powder in ethanol onto a 200 mesh Cu grid coated with a lacey carbon film followed by drying under vacuum. Ex situ samples for all measurements were prepared by collecting pure F x ReO 3 electrodes from disassembled electrochemical cells cycled to the desired state of charge and washing the material in dimethyl carbonate in an argon glovebox. Electrochemical Characterization. Several methods were developed for preparing fluoride ion batteries with both liquid and solid electrolytes to enable different aspects of testing. All cell components and electrodes were dried under vacuum at 110 ◦ C for 1 hour before assembly and cell assembly was performed in an argon glovebox. Stainless steel Swagelok cells were employed for electrochemi- cal testing and were equipped with borosilicate glass fiber pads (Whatman GF/D) soaked with electrolyte solution. 1M Tetrabutylammonium Fluoride in tetrahy- drofuran (TBAF) was used as the liquid electrolyte. Thick film electrodes were prepared by blending 10 wt% graphite powder (300 mesh), 10 wt% acetylene black, 20 wt% polytetrafluroethylene (average particle size of 1 μm), and 60 wt% active material, and pressed under a hydrostatic pressure of 0.9 tons. The electrodes had a typical mass loading of 4.5 mg cm 2 . Film electrodes of pure ReO 3 as the active material were also prepared by depositing onto Ni foil disks with a typical mass loading of 4.5 mg cm 2 . 113 Solid state cells were prepared with a film of Ba-doped LaF 3 . Tysonite- type barium-doped lanthanum fluoride (LBF) electrolyte of composition La 0.95 Ba 0.05 F 2.95 was prepared by a co-precipitation method described elsewhere, and used as the solid electrolyte. 211 In a typical synthesis, stoichiometric amounts of La(NO 3 ) 3 •6H 2 O and Ba(NO 3 ) 2 were dissolved in de-ionized (DI) water followed by the dropwise addition of an aqueous solution of NH 4 F. The precipitate was collected by centrifugation, washed with DI water and dried at 80 ◦ C overnight. The resulting powder was then mixed with 5 wt% polyvinylidene fluoride (PVDF) binder in n-methyl-2-pyrrolidone (NMP) solvent, cast on an aluminum foil, and dried at 110 ◦ C under vacuum. The layer was then peeled off to create a free- standing 100 micrometer thick film of LBF electrolyte. Similarly, an active elec- trode was prepared by mixing 30 wt% ReO 3 active material, 50 wt% LBF elec- trolyte, 10wt%PVDFand10wt%SuperPcarbonwithaminimalamountofNMP toformaslurrywhichwascastonaluminumfoilanddriedat110 ◦ Cundervacuum. A copper fluoride electrode was cast onto copper foil using the above method and contained 70 wt% anhydrous CuF 2 , 20 wt% SuperP carbon and 10 wt% PVDF. All-solid-state fluoride ion battery cells were assembled by pressing ReO 3 elec- trodes, LBF electrolyte layer and CuF 2 counter electrodes together into 10 mm diameter pellets using a hydraulic press at a pressure of 1300 kg cm −2 for 30 min- utes. The pellet was inserted into a Swagelok cell inside an argon-filled glovebox, and electrochemical testing was performed at 150 ◦ C. Electrochemical impedance spectroscopy (EIS) measurements were carried out using a VersaStat MC electrochemical workstation over the frequency range of 1 MHz to 0.1 Hz with a sinusoidal excitation voltage of 10 mV. The measurements were done under open circuit condition at various levels of fluoride intercalation. 114 Operando synchrotron XRD patterns were collected using an AMPIX electro- chemical cell equipped with two glassy carbon windows. 87 The same glass fiber separators, metallic counter electrodes, and electrolyte solutions described previ- ously were used during all operando experiments. Differential Electrochemical Mass Spectrometry. The custom-built differ- ential electrochemical mass spectrometry (DEMS) system and the cell geometry used is described in detail in previous publications. 212,213 In brief, electrochemical half cells of modified Swagelok design were assembled in an argon filled glovebox using a ReO 3 working electrode (80 wt % ReO 3 , 10 wt % Super P, 10 wt % PTFE) pressed onto a stainless steel mesh current collector, a TBAF soaked glass fiber separator, and a Cu metal counter and reference electrode. The electrochemi- cal cells were then attached to the DEMS system such that a head of positive argon pressure (1.2 bar) was maintained. During the measurements, argon gas pulses periodically swept the accumulated gases to a mass spectrometer chamber for identification of H 2 , O 2 , and F 2 . Solid-State 19 F NMR Spectroscopy Solid-state 19 F MAS-NMR spectra were acquired on a Bruker AVANCE- III Ultrashield Plus 800 NMR spectrometer using a narrow-bore 18.8 T superconducting magnet and operating at a Larmor fre- quency of 752.980 MHz for 19 F. Experiments were conducted using a 1.3 mm double-resonance H/F–X magic-angle spinning (MAS) probehead. For the solid- state NMR measurements, the conductive F x ReO 3 powders were mixed with KBr powder in a roughly 1:5 ratio by mass to reduce effects of eddy currents on the spinning samples. KBr also served as an internal temperature probe for accurate determination of the sample temperature under the different measurement condi- tions. 214,215 One-dimensional (1D) 19 F Hahn-echo MAS spectra were acquired at 115 23 kHz MAS unless otherwise specified, at 298 K, and using a 90 ◦ -τ-180 ◦ -τ pulse sequence with rotor-synchronized τ delay times of one rotor period and 90 ◦ radio frequency(rf)pulses of3.5μs witha pulsepower of40 W. The 19 F spin-lattice (T 1 ) relaxation times were measured by using a saturation–recovery pulse sequence with a Hahn-echo detection (Figure 5.17, Supporting Information). Isotropic 19 F NMR chemical shifts were referenced to CFCl 3 using PTFE (-122 ppm) as a secondary standard. Computational Thermodynamic Stability and Prediction of 19 F NMR Chemical Shifts The CASM software package 216–218 was used to explore the phase stability of topotactic F insertion into perovskite ReO 3 by enumerating dif- ferent F-vacancy orderings over the empty A sites of ReO 3 . The energies of these configurations were calculated with DFT-PBE using the VASP plane-wave code and 219–222 the generalized gradient approximation of Perdew, Burke and Ernzerhof (GGA-PBE) was used for the correlation and exchange potentials with a 17× 17× 17 K-point mesh was used for all structures (i.e. ReO 3 and ReO 3 F). Additionally, several configurations that allowed for a redistribution of Re or mixing between O and intercalated fluoride ions were considered. 19 F NMR shielding tensors for various structural models were calculated using the fully periodic Gauge Including Projector Augmented Wave method (GIPAW) 170,171,223 following a similar approach to Griffin and coworkers. 224 In brief, a linear relationship between the calculated chemical shifts for several refer- ence compounds were correlated to their experimentally observed values and fitted to establish relationship: δ exp iso = –k(δ calc iso –δ ref ) wherek = 0.68 andδ ref = 160 from the fits as shown in S.I. Table 5.3 and S.I. Figure 5.18. This expression was then 116 usedtopredicttheexperimentallyobservedchemicalshiftofthemodelsconsidered in this work. Computational Simulations of X-ray Spectroscopy Density functional theory (DFT) calculations to simulate the X-ray emission spectra were per- formed using the WIEN2k 225,226 software package, which uses a full potential and linearized-augmented planewaves with local orbitals (LAPW + lo) to self- consistently solve the Kohn-Sham equations. Structures from the geometric relax- ations in VASP were simulated without further optimization. 227 The O K-edge of various intercalated forms were calculated using a planewave cutoff parameter, RKMAX, was chosen to be 6.5 and the cutoff between core and valence state set as -8.0 Ry. The effect of the core hole in simulation was characterize by performing partial core hole calculations, in which the unit cell is used instead of super cell to include the interaction between neighboring core holes. In the partial core hole approach, the occupancy of thecore levels was reduced, andthe missing charge was added as a uniform background charge to the unit cell to avoid re-normalization problems. 228,229 For ReO 3 F, the O K-edge calculations were performed six times due to the existence of six oxygen atoms at different sites, after which the average of the six individual contributions according to the multiplicity of each oxygen atom was calculated to obtain the theoretical O K-edge spectrum. 5.3 Results and Discussion ReO 3 was chosen as a model electrode compound to investigate the fundamentals of fluoride intercalation for several reasons, including the high degree of covalency inbonding, theexistenceofvacanciestoaccommodateions, anditsinherentmetal- lic character. ReO 3 crystallizes in the Pm ¯ 3m space group (#221) and possesses a 117 perovskite-like structure composed of highly symmetric corner sharing ReO 6 octa- hedra in a perfectly cubic arrangement with an empty A-site that creates large three dimensional channels for ionic diffusion 72 as shown in Figure 5.2(a). While cation insertion has been demonstrated previously, the close-packed oxygen net- work [Figure 5.2(b)] makes anion transport pathways less obvious. The synthesis of ReO 3 , described in the Experimental Methods section of the Supporting Information, yields highly uniform and crystalline nanoparticles between 20 and 50 nanometers in diameter. ReO 3 is one of the few metallic oxides, exhibiting a conductivity on the order of 10 −5 S/cm, 230 making it an excel- lent model battery electrode because it is possible to prepare test cells without conductive carbon additives, which would complicate many spectroscopic mea- surements. Additionally, using test electrodes comprised entirely of ReO 3 removes any potential side reactions related to carbon or binder additives, an important consideration in new cell design. In an attempt to insert fluoride into ReO 3 , F-ion half cells were assembled using a Cu metal combined counter and reference electrode, and an organic liq- uid fluoride electrolyte, as described in the S.I. Experimental Methods section. Glass fiber separators soaked in an electrolyte of tetra-n-butylammonium fluoride (TBAF) dissolved in tetrahydrofuran (THF) served as the source of free fluoride ions between the working and counter electrode. TBAF was chosen for its relative stability and wide electrochemical window as well as a reasonable fluoride-ion con- ductivity. As shown in Figure 5.3(a), the oxidation reaction produces a smooth voltage curve with an initial plateau region followed by a smooth voltage rise until approximately F 0.6 ReO 3 . While the use of Cu metal provides a consistent reference potential, it cannot be further reduced which raised questions about the nature of the counter reaction occurring at the cathode. To better understand this counter 118 0.0 2.0 4.0 6.0 8.0 10.0 Time (hours) Ion Current (arb. units) Mass of 2 AU (H 2 ) Mass of 32 AU (O 2 ) Mass of 34 AU (F 2 ) (b) 0.0 0.2 0.4 0.6 0.8 x in F x ReO 3 0.0 0.5 1.0 1.5 2.0 Voltage vs Cu QRE (V) C/10 Charging (a) Decomp. Cell at Rest Figure 5.3: (a): Galvanostatic charge curve of a fluoride-ion half cell during mass spectrometry measurements with a ReO 3 working electrode and Cu metal counter and reference electrode. (b): Differential electrochemical mass spectrometry iden- tified H 2 evolution in fluoride ion battery half cells, however no O 2 or F 2 generation was seen. reaction, Differential Electrochemical Mass Spectrometry (DEMS) measurements were employed to monitor any evolution of gas in situ during the cycling at both the anode and cathode. Several gases were monitored during oxidation, including O 2 , F 2 , and CO 2 (shown in S.I. Figure 5.15), but as seen in Figure 5.3(b), the only gas that was observed to evolve corresponded to H 2 , which persisted through- out charging. Therefore, the reductive counter reaction is found to be associated with a hydrogen evolution reaction (HER) from an attack of the THF or TBAF salt. The lack of oxygen release from ReO 3 throughout the measurement is notable given that the quasi-reference potentials observed scale to be far in excess of 4V vs. Li + /Li. This suggests there is unlikely to be very much decomposition of the ReO 3 particles through the release of lattice oxygen in the manner previously reported for other metal oxides. 231,232 This high voltage stability is most likely associated with the covalent character of the material – the strong hybridization between the metal and oxygen orbitals make it very difficult to strip oxygen out of the lattice. 119 To validate the liquid-phase electrochemistry, all-solid state cells were prepared using a Ba-doped LaF 3 electrolyte to eliminate solvent decomposition and only allow for redox reactions involving the transport of fluoride-ions. Given the slow fluoride conductivity of the solid electrolyte, it was necessary to heat the cells to 150 ◦ C during cycling in order to facilitate facile ion transport in the solid electrolyte. As the H 2 evolution reaction observed in the liquid electrolyte half cells in Figure 5.3(b) is not possible in the solid state, a CuF 2 counter electrode was chosen as a reductive fluoride source. A similar voltage profile was observed during charging of the solid state cells as that of the liquid cells, however, the sluggish nature of F-ion transport in the solid state led to a high cell resistance in the solid state cells which were tested. Therefore Galvanostatic Intermittent Titration (GITT) measurements were used to examine the equilibrium voltage of the reaction. GITT measurements were also performed on cells with a liquid electrolyte allowing for direct comparison to the electrochemistry performed in solid state cells. As shown in Figure 5.4, a similar voltage relaxation profile was observed in both solid state and liquid cells, suggesting a very similar oxidative reaction at the ReO 3 electrode for both cell designs. Electrochemical Impedance Spectroscopy (EIS) studies on the solid electrolyte cell were also used to monitor variations in the charge transfer and mass transfer resistances at different levels of oxidation. The Nyquist plot in S.I. Figure 5.14(a) shows a depressed semicircle in the high frequency region characteristic of charge transfer resistance (R ct ) in parallel with the interfacial double layer capacitance (C dl ). The values of R ct do not change significantly until x = 0.5 and a small rise of 2–3 Ω is noticed thereafter. The increase in R ct occurs concurrently with the transition from the initial sloping voltage to a higher voltage plateau. In the low-frequency region of the impedance spectrum, the impedance rises steadily with decreasing frequency 120 with a phase angle of approximately 45 ◦ suggesting a diffusion-limited process. Considering the diffusion length and the thickness of the sample, this section of the impedance spectrum can be modeled with the semi-infinite linear diffusion boundary conditions, as described in the S.I. Impedance Spectroscopy section. A decrease inD is observed with increasing state of charge, as shown in S.I. Table 5.1, suggesting that changes to the material during charging hinder fluoride transport. 0.50 1.00 1.50 Voltage vs Cu QRE 0.0 0.2 0.4 0.6 x in F x ReO 3 0.25 0.50 0.75 Voltage vs CuF 2 /Cu Liquid Electrolyte Solid Electrolyte (a) (b) Figure 5.4: Galvanostatic intermittent titration measurements in (a) liquid elec- trolyte cells prepared with TBAF electrolyte show similar open circuit voltage traces when compared to (b) voltage traces from all solid state cells. Having validated that the oxidative currents were indeed associated with a reaction at the ReO 3 electrode, and were not sim- ply electrolyte decomposition, we turned to developing a mechanistic understanding of how fluoride intercalation could occur within the structure of ReO 3 . Operando X-ray Diffraction (XRD) patterns were collected during oxidation until an equivalent current for one fluoride per formula unit had been passed. As seen in Figures 5.5(a) and 5.5(b), the (111) reflection continuously shifts very slightly to higher angles, indicating a small contraction of the unit cell during charging but no new peaks evolve nor do any of the existing peaks split to suggest changes in the cubic symmetry of the lattice. Beyond a nominal fluoride content of roughly 0.6, the intensity of diffraction patterns begin to significantly decrease, indicating the onset of decomposition as illustrated in S.I. Figure 5.12. We note, however, 121 that immediately after exposure to the TBAF electrolyte, and prior to cycling, several low intensity peaks appear in data obtained at the synchrotron (further demonstrated in S.I. Figure 5.10). These peaks cannot be seen with laboratory X-ray sources and do not change position or intensity at any point during the cycling (see S.I. Figure 5.11), and are therefore believed to be associated with an electrochemically inert contaminant associated with the electrolyte solution. While the extremely minor changes seen in the diffraction patterns initially seem puzzling, a careful look at the cubic structure shows that placing fluoride ions onto the vacant A-site would not break any of the symmetry elements within the parent perovskite (see S.I. Figure 5.13), which is consistent with the absence of any new diffraction peaks. In parallel, ex situ electron microscopy and electron energy loss spectroscopy (EELS) mapping were used to monitor the particle morphology and distribution of fluoride in the electrode after charging. As shown in Figure 5.5(c), the particles maintain their cubic morphology with signs of crystallinity suggesting that the ReO 3 does not dissolve or directly extrude any secondary phases during oxidation. EELS measurements, shown in Figure 5.5(d) show an unevenfluoridedistributionthroughouttheparticleswithfluoriderichregionsnear the surfaces. Nevertheless, there is no evidence that the morphology of the ReO 3 nanoparticlesaltersduringcycling, whichrulesoutadissolution-precipitationstyle reaction. Furthermore the fluoride persists on the particles even after washing the ex situ samples thoroughly to remove any residual electrolyte. Ex situ X-ray Emission Spectroscopy (XES) was also collected on similarly washed samples to more directly track the fluoride content of the oxidized elec- trodes. The greater penetration depth allows for sensitivity to bulk fluoride rather than simply probing species isolated to the surface. By comparing the intensity of the spectra from the O 2p orbitals to the F 1s orbitals, relative changes in the 122 bulk fluoride content are clearly seen. As shown in Figure 5.5(e), the intensity of the peak at approximately 677 eV, associated with the F 1s orbitals from bulk flu- oride species, grows with increasing state of charge while that of the O 2p peak at 531 eV remains constant. As demonstrated by the DEMS measurements in Figure 5.3(b), there is no reason to believe that O 2 is released from the lattice during oxidation; thus the variation in the F peak corresponds an increase in the fluo- ride content throughout the ReO 3 electrode during charge. In contrast, the states associated with the lattice oxygen of ReO 3 stay relatively constant, but do show some restructuring due to changes in the chemical environment of the materials, which will be discussed in greater detail later. Taking these results together, it seems clear that the fluoride reacts with the ReO 3 , but it is not immediately clear whether the fluoride intercalates into the structure or if there is a more nuanced transformation that is missed by the operando X-ray diffraction experiments. In order to to assess the thermodynamics of various reaction mechanisms on the insertion of F into ReO 3 , first-principles density functional theory (DFT) cal- culations were performed. We first explored topotactic insertion by enumerating different F-vacancy orderings over the A-sites of ReO 3 . Figure 5.6(a) shows the calculated formation energies of these orderings at a variety of concentrations using the perovskite forms of ReO 3 and fully fluorinated FReO 3 as the reference states. The formation energies are very slightly negative indicating that F-vacancy order- ing over the A sites might be expected at low temperatures, while a solid solution will emerge at elevated temperatures provided the ReO 3 perovskite host remains intact. An alternative to topotactic insertion would be a reconstructive or conversion reactionmechanism, 233 wherebyacompletelynewcrystalstructureemergesduring 123 Intensity (arb. units) F K-edge O K-edge 528 532 536 670 680 690 540 Emission Energy (eV) 5.3 5.4 5.5 5.6 5.7 2 θ (deg.) ( λ=0.25463 Å) Intensity (arb. units) (1 1 1) (a) 5.3 5.4 5.5 5.6 5.7 2 θ (deg.) ( λ=0.25463 Å) Intensity (arb. units) (1 1 1) F 0.25 ReO 3 F 0.50 ReO 3 F 0.75 ReO 3 F 1.00 ReO 3 (b) (d) (e) (c) F 0.0 ReO 3 F 0.25 ReO 3 F 0.50 ReO 3 F 0.67 ReO 3 Fluorination Figure 5.5: (a): Operando synchrotron X-ray powder diffraction captured at var- ious states of charge during the fluorination of ReO 3 . A loss of diffracted intensity is observed during charging. (b): The (111) reflection is highlighted to demon- strate the slight shift to higher 2θ values in conjunction with a loss of diffracted intensity at higher states of charge. (c): TEM images of fluorinated ReO 3 show the cubic particle morphology is maintained while (d) EELS spectra show a F–rich shell on the particles. (e): X-ray emission spectroscopy of F x ReO 3 at various states of charge shows an increase in F signal due to fluoride intercalation. oxidation. A survey of the Inorganic Crystal Structure Database 234 shows the exis- tence of a monoclinic form of ReO 3 F where the structure consists of edge-sharing chains of octahedrally coordinated ReO 4 F 2 units with two fluoride ions sitting along common edges of the polyhedra (see S.I. Figure 5.20). 235 In the following, we distinguish the intercalated phase from this one by writing the compositions as F x ReO 3 or mono-ReO 3 F respectively. The energy of this compound is predicted to be more than 2 eV lower per formula unit than that of the fully intercalated perovskite form of F x ReO 3 , which points to the existence of an enormous thermo- dynamic driving force for a reconstructive transformation. Given that the crystal structure of the mono-ReO 3 F form is very different from that of the perovskite form of ReO 3 , it may not be kinetically accessible at room temperature. We therefore decided to explore other hypothetical structures that 124 may be more readily accessed during F − insertion into ReO 3 . One such structure can be obtained by inserting F in the vacant A-sites, followed by a coordinated migration of Re cations from their octahedral sites to a newly created tetrahedral interstitial, referred to as F x ReO 3 –1F (see structure in Figure 5.7). This hopping of the cations was considered for several different compositions of fluoride but was repeatedly found to be higher in energy than when there is no migration of Re out of the octahedral sites. Similarly, geometric relaxations of the structures also indicated that the forces on the atoms can be fully minimized by distorting Re out of the center of the octahedra towards one of the triangular faces of the octahedra to create a highly distorted seven coordinate environment, referred to as seven-coordinate ReO 3 F (see S.I. Figure 5.19). This too was found to be far less favorable than F-ions sitting in the empty cages of the cubic form of ReO 3 . Interestingly, however, we find that when all of the A-sites are filled by fluoride and all of the Re ions simultaneously migrate into the tetrahedral interstitials, as illustrated in S.I. Figure 5.22 this structure is significantly lower in energy than the structurewithfluorideonlyontheA-site, consideredasintermediatecompositions. This phase, referred to as tet-FReO 3 , is best viewed as a molecular crystal with covalently bonded ReO 3 F tetrahedra that are loosely held together by weaker van der Waals forces. In this work, we have considered the polar form of the structure where the apical F-ions on each tetrahedron point in the same direction, but, in reality, the tetrahedra are likely to be far more disordered, which we have not considered here at all. These calculations clearly suggest that if fluoride-intercalated ReO 3 were actu- ally to be obtained, there would be a very strong driving force for it to transform into either the mono-ReO 3 polymorph reported in the materials project or, while not as likely, the form with all of the Re migrated to tetrahedral interstitals [S.I. 125 Figure 5.22]. At small values of x, i.e. less than F 0.5 ReO 3 , the energy to form the phase with Re moved into the newly formed tetrahedral interstitial is actually higher than that of simply placing F-ions on the A-site. This suggests that a reconstructive reaction could occur through a two-phase mechanism, and, indeed, an application of the common tangent construction to all the formation energies of structures that are kinetically accessible predicts a two- phase coexistence between ReO 3 and tet-FReO 3 as shown by the dashed gray line in Figure 5.6(a). This partitioning into two-phases could be expected to result in a core-shell geometry, in which the new tet-FReO 3 phase forms on the surface and grows inward, consuming the original ReO 3 core, which appears consistent with the EELS maps but, in principle, should have been observable in the operando diffraction patterns if the shell were fully crystalline. However, given that the mono-ReO 3 F is more stable than even tet-FReO 3 , it is difficult to know exactly what phase the F-rich shell seen in EELS is composed of. For this, we turned to structural tools that were more sensitive to the local environments within the material such as neutron total scattering to generate Pair Distribution Function (PDF). Starting with the top panel of S.I. Figure 5.24, the ex situ neutron total scattering experiments for the pristine ReO 3 are clearly well-described by the long-range average cubic structure whereas the significantly changed scattering from a washed sample of F 0.6 ReO 3 is shown in the bottom portion of the same figure. The first two peaks at 1.88 Å and 2.67 Å in the pristine phasecan be assigned tothe Re–Oand O–O nearestneighbors, which after oxidation, evolve into a complex mixture of peaks that makes precisely modeling the pattern exceptionally challenging. To aid in deconvoluting the contributions at each distance, the bottom panel of S.I. Figure 5.24 shows the calculated PDF adapted from the tet-FReO 3 model considered earlier, as described in the Neutron 126 (b) 0.0 0.1 0.2 0.3 0.4 0.5 x in F x ReO 3 3.755 3.765 3.775 3.785 3.795 a Lattice Parameter FReO 3 0.0 0.2 0.4 0.6 0.8 1.0 x in F x ReO 3 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 Formation Energy (eV/F x ReO 3 ) FReO 3 Mono-ReO 3 F Tet-ReO 3 F 0.0 0.4 0.8 -0.06 -0.04 -0.02 0.00 (a) Figure 5.6: (a): The computed Hull diagram indicates that tetrahedral Re is energetically favorable upon fluoride intercalation while the inset shows a small energetic minima for the insertion of F to the A-site. (b): The calculated a lattice parameter for FReO 3 indicates a unit cell contraction, in agreement with operando XRD measurements. Pair Distribution Analysis section of the Supporting Information. For example, the first shell of distances at 1.88 Å splits into two distinct contributions that can be attributed to three short Re–O distances and one slightly longer Re–F bond within the newly created tetrahedra in the model. Furthermore, the peak at 3.75 Å, associated with Re-Re distances is seen to split as Re migrates from its starting position and the O–O peak (2.67 Å) evolves into multiple peaks as new O–O and O–F environments are created. We also note that contributions from 127 short C–H and C–C distance had to be included in order to account for all of the changes in the oxidized sample. These organic species are believed to be associated with a decomposition product created during the electrolyte decomposition at the counter electrode discussed earlier. While the new distances in the PDF might initially appear to support the migration of Re into the tetrahedral interstitials, the predicted lattice parameters for such a hop suggest the unit cell should expand during such a transformation. In contrast, the operando diffraction data showed a contraction of the lattice, which would seem to imply that if such a hop were to occur it would need to be in an uncorrelated fashion that exhibits no long-range periodic order to the new tetrahedral sites. Alternatively, this may also suggest that a Re environment resembling the tetrahedral or seven coordinate models described previously might exist at the surface of the particles rather than as a completely separate new phase (see S.I. Figures 5.25-5.28). To clarify the ambiguity of the PDF data, solid-state 19 F MAS NMR was used asacomplementaryprobeofthelocalcompositionandstructureoftheintercalated fluoride ions. While the PDF results yield insights on locally-averaged structural distortions within the bulk material, 19 F MAS NMR allows for the direct detec- tion and resolution of signals from fluoride species in different local environments. As shown in Figure 5.16 of the Supporting Information, the 1D 19 F MAS NMR spectrum for bulk F 0.2 ReO 3 , produced as a result of oxidative insertion during par- tial charge, exhibits a complicated distribution of 19 F intensity over the frequency range of -100 to -200 ppm. Interestingly, several distinct and relatively narrow (∼5 ppm, full-width-half-maximum) 19 F signals are observed at isotropic chemical shifts of -136, -141, -150, -170, -174, and -189 ppm, which manifest fluorine species 128 in well-defined local environments. The signal at -136 ppm is furthermore asso- ciated with a 19 F spin-lattice relaxation time, T 1 , of 4.0 s (Table 5.2, Supporting Information), which is consistent with fluoride in a diamagnetic perovskite envi- ronment. 236,237 Interestingly, all of the other signals correspond to much shorter 19 F T 1 values of 0.3-0.4 s (Table 5.2), consistent with fluoride environments that are influenced by conducting or donor electrons. 236–238 DFT calculations of several structural models selected from the phase diagram for fluoride intercalation into ReO 3 predict an isotropic 19 F chemical shift of -141 ppm for mono-ReO 3 F, close to the values of several of the measured signals. By comparison, the DFT calcu- lations predict the isotropic chemical shift of a model structure with 7-coordinate distorted-octahedra to be near -73 ppm and of tet-FReO 3 to be near -11 ppm, neither of which are experimentally observed (Figure 5.16). The multiple sig- nals clearly show that the fluoride species in the electrode are inhomogeneously distributed in diverse local environments and corroborate the challenges to the analyses of the PDF data. In a final attempt to reconcile the experimental results with computational models, X-ray spectroscopy was performed on the oxygen and fluorine K-edges for pristine and oxidized samples and compared with simulation based on the various structural permutations discussed previously. The O K-edge emission and absorp- tion spectra reflect the bulk occupied and unoccupied O 2p partial density of states near the Fermi level. The oxygen K-edge, shown in Figure 5.7(c), evolves signifi- cantly after cycling, which clearly indicates a change in the electronic structure of ReO 3 during oxidation. While the pristine material (blue) shows a large absorp- tion peak at 0-2 eV associated with O 2p – Re 5d bonding orbitals, these states are suppressed upon charging (pink), reflecting electron density being extracted from the material. The emergence of new peak at 9 eV indicates a change to the local 129 oxygen environment during oxidation. The presence of F − near very electrophilic Re 6+ is expected to draw electron density from the O 2p states to higher binding energies. As shown in Figure 5.7(b) and (c), DFT calculations were used to model the expected O K-edge absorption and emission in both ReO 3 and FReO 3 , using the structural models previously described, where Re migrates to one of the tetrahe- dral sites in FReO 3 . Additional models are provided in S.I. Figures 5.19 – 5.22. Curiously, the only model that captures the shape and energy shifts of the mea- sured O 2p partial density of states is where a small number tetrahedral sites are created, rather than the phase that was predicted to be more thermodynamically favored i.e. where all of the Re jump to the tetrahedra (see S.I. Figure 5.22) The relative change in intensity between X-ray absorption peaks at -2 eV and -9 eV, representing a change in O bonding environments as ReO 3 is fluorinated, is matched by the DFT calculations, as is the shape and energy shift of the main Oxygen K-edge XAS ReO 3 F 0.6 ReO 3 XES DFT Oxygen 2p ReO 3 Standard F x ReO 3 - 1F Energy (eV) -14 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 Energy (eV) Fluorine K-edge FReO 3 Standard F x ReO 3 A-site F x ReO 3 X-site F x ReO 3 - 1F -5 0 5 10 15 20 Intensity (arb. units) F 0.6 ReO 3 A-site F x ReO 3 - 1F ReO 3 X-site (a) (b) (c) Figure 5.7: (a): Density functional theory calculations of the O and F K-edges were performed for different structural models of FReO 3 . (b) calculated X-ray emission spectra of the F K-edge are shown in black while the measured spectra of ReO 3 is shown in orange and F 0.6 ReO 3 is shown in blue. The energy shift of the measured spectra agrees well with the model of Re migration to the tetrahedral site. (c) The calculated and measured O K-edge X-ray emission and absorption spectra agree for both pristine and fluorinated ReO 3 using the proposed structural model. 130 emission features observed. The fluorescent nature of O K-edge XES, suggests that the spectral changes from the electrochemical reaction must not be limited to a surface reaction, but instead some change to the bulk ReO 3 electronic structure must also occur. A closer look at the geometrically relaxed model containing a small number of tetrahedral Re shows a slight off-centering within the octahedral cages and some rotational disorder that likely explains the changes seen in the spectra. Calculations of the fluorine K-edge for each model are shown as black traces in Figure 5.7(b). As seen, the measured XES spectra of F 0.6 ReO 3 also agrees well with the model where Re migrates to stabilize the intercalated F-ions. While challenges with alignment of the F K-edge led to broadening of the measured signal and prevented distinct peaks from being resolved, the strong agreement between the calculated and measured energy shift further supports the presence of some tetrahedral species within the oxidized electrode. Finally, HAXPES measurements were performed on the Re 4f, O 2p, and F 1s orbitals of pristine and fluorinated ReO 3 and used to track changes in oxidation state of the various species during oxidative charging. As mentioned, the highly covalent bonding of ReO 3 delocalizes electrons over both oxygen and Re states, which leads to partial changes in the Re oxidation state during charging. This can be seen in S.I. Figure 5.23, where the Re 4f photoemission was measured during various states of charge. The doublet peak at 43 and 46 eV are diagnostic of Re 6+ , while a similar doublet at 46 and 48 eV is attributed to Re 7+ . During oxidative charging, the formation of a shoulder at 48 eV is observed as Re is partially oxidized. However, owing to the strong covalency, no evidence for a fully ionic description of Re 7+ species was detected. 131 Taking all of the experimental and computational results together, there is clear evidence that F − ions have incorporated into ReO 3 during oxidative charging, but the precise mechanism remains ambiguous. Operando X-ray diffraction data shows a clear lattice contraction that agrees closely with the trends predicted for the models containing intercalated F-ions on the A-site of the ReO 3 cage. Yet, this A-site model is predicted to be unstable compared to less densely packed polymorphs, and the ex situ materials characterization suggests the fluoride in the oxidized materials is most likely associated with those more stable tetrahedral tet- ReO 3 F species or a decomposition to the structurally distinct mono-ReO 3 F phase. This leads us to conclude that this discrepancy is likely due to changes in the material once it is removed from the electrochemical cell. Note that the operando diffraction experiments are collected from start to finish in approximately five hours with each pattern being collected in a matter of seconds. Hence, the most likely explanation for the discrepancy between the various characterization tools is that F-ions intercalate onto the A-site of the perovskite host, but quickly begin to decompose to the more thermodynamically favored phases over time. It is curious that best agreement between the experimental data and the com- putation is found for the model containing a dilute number of Re that have jumped totetrahedralsites. Wesuggesttwopossibilitiestoexplainthisobservation. Either a shell of tetrahedrally coordinated Re forms as a decomposition product on the surface of the particles and the model we have considered is effectively capturing important interfacial effects; or, the correlated migration of Re into the tetrahe- dra environments is kinetically challenging and eventually results in some of the tetrahedral species trapped within the octahedral network even though they are thermodynamically disfavored. It is also possible that the final decomposition product contains some aspects of the dilute tetrahedral Re model, but the model 132 fails to precisely capture all of the structural nuances. While we have gone to great lengths to evaluate as many possible forms of the intercalated structure, it is very likely that there are a vast number of alternate ways the perovskite lattice may be able to distort through octahedral tilting or cation migration. We therefore believe there is ample evidence to conclude that F-ions have incorporated into ReO 3 dur- ing oxidation despite being unable to definitively assign a unique structural model to the resulting material. Further exploration using the thermodynamic modeling with a greatly expanded phase space along with additional operando local probes would be helpful in this regard, but falls outside the scope of the present study. While the intercalation of fluoride ions into structures as dense as perovskite may initially seems too energetically demanding, several studies have shown the possibility to intercalate fluoride perovskite derivatives using aggressive chemical oxidants. 239–242 Other work has investigated the formation of interstitial fluoride species within similar Sr 3 Ru 2 O 7 species, in which layers of fluoride are formed and result in correlated rotations of the RuO 6 octahedra. 243 Crucially, fluorination of the parent oxide was performed at a low reaction temperature of 220 to 300 ◦ C, demonstrating the relative ease of fluoride intercalation into an oxide. It is also known that halide ions are very mobile within the perovskite struc- tures, with studies of CH 3 NH 3 PbI 3 showing this migration occurs easily under an applied bias. 244,245 These reactions are typically limited by halide diffusion through the structure via interstitials or vacancies but can proceed quite quickly. In fact, Walsh and Stranks have clearly shown that both the cationic and anionic sublat- tices in perovksites are mobile and the activation barrier for defect migration is relatively low, which is in good agreement with what we observe here. 246 133 5.4 Conclusions In summary, we have presented, for the first time, the oxidative fluorination of the dense oxide ReO 3 using electrochemical cells and a liquid electrolyte at room tem- perature. We find that electrons are removed from the strongly hybridized Re-O orbitals at the Fermi level resulting in partial oxidation of both Re and O, with charge compensation occurring via fluoride intercalation into the vacant perovskite A-site. The intercalation of fluoride results in complex structural changes as some Re appear to migrates from the octahedral sites to newly formed tetrahedral sites, as evidenced by changes in the XAS and XES spectra, SXRD patterns, and neu- tron PDF scattering. While DFT calculations clearly shows that the intercalated form of FReO 3 is strongly disfavored by thermodynamics, more exhaustive explo- rations of the phase space are necessary to fully understand whether more complex tilting/alterations of the initial octahedral network may service to stabilize these phases. Fluoride intercalation at ambient conditions represents a paradigm shift compared to traditional Li-ion battery technologies and offers a window into a broad new avenue for developing methods for higher storage capacity if cationic and anionic intercalation can be successfully and sequentially achieved within sin- gle phase materials. 134 5.5 Supplemental Information Operando X-rayDiffraction Operando synchrotronX-raydiffractionwasused in order to track average structural changes during the charging of F x ReO 3 , reveal- ing a very small shift to higher angles of all peaks associated with ReO 3 . Through- out the charging process, no new peak formation is observed, as shown in Figure 5.8 and 5.9 due to the localized structural changes. However, a series of addi- tional low angle peaks are seen (Figure 5.11), which are attributed to an SEI like layer which is formed from the decomposition of TBAF electrolyte. These peaks appear upon the addition of TBAF electrolyte, as shown in Figure 5.10 prior to electrochemical testing. Figure 5.8: A heatmap of the operando synchrotron X-ray diffraction collected during the oxidative fluorination of ReO 3 displays diffracted intensity as a function of color. No new peaks are seen to form throughout the electrochemical charging process and reflections associated with cubic ReO 3 are maintained. We note a reduction in peak intensity beginning at higher states of charge. 135 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 2 � (deg.) [ �=0.25463 Å] Intensity (arb. units) Fluorination Figure5.9: Operando synchrotronX-raydiffractionshowsnonewpeakformation throughout the charging process of F x ReO 3 , however we note the formation of a series of low angle peaks due to electrolyte decomposition. The spacing between each pattern is approximately 0.025 equivalent units of fluoride. 2.0 4.0 6.0 8.0 10.0 2 θ (deg.) [ λ=0.25463 Å] Intensity (arb. units) Before TBAF Electrolyte After TBAF Electrolyte Figure 5.10: Upon the application of TBAF to ReO 3 electrodes, a series of addi- tional XRD peaks are seen to form, associated with the decomposition of TBAF electrolyte. The XRD pattern shown in blue was collected on a ReO 3 electrode which was placed in a fully assembled AMPIX cell without any electrolyte. Once the electrolyte was added, a pattern (pink) was again collected prior to electro- chemical measurement which showed the formation of a series of low-angle diffrac- tion peaks. 136 1.6 2.0 2.4 2.8 3.2 2 � (deg.) [�=0.25463 Å] Intensity (arb. units) F 0.0 ReO 3 F 0.25 ReO 3 F 0.50 ReO 3 F 0.67 ReO 3 Fluorination Figure 5.11: A series of low angle peaks form due to electrolyte decomposition in the operando X-ray diffraction experiments. None of the peaks observed at low angles are associated with cubic ReO 3 . These peaks persist throughout the electrochemical charging process but do not vary in intensity or location. The spacing between each pattern is approximately 0.025 equivalent units of fluoride. 3.6 3.8 4.0 Intensity (arb. units) 5.4 5.6 2 � (deg.) [ �=0.25463 Å] 6.6 6.8 7.0 (1 0 0) (1 1 0) (1 1 1) Fluorination Figure 5.12: Throughout the initial oxidative fluorination process, diffraction peaks associated with ReO 3 steadily shift to higher angles, indicating a contraction of the unit cell. At high states of charge this abruptly reverses and the peaks of ReO 3 are seen to shift to lower angles, suggesting a decomposition of the material back to the original phase. 137 2.0 4.0 6.0 8.0 10.0 2 � (deg.) [�=0.25463 Å] Intensity (arb. units) ReO 3 FReO 3 Figure 5.13: Comparison of simulated X-ray diffraction patterns of ReO 3 and FReO 3 in which F sits in the central A-site vacancy (see inset). The insertion of fluoride on the A-site does not produce any additional diffraction peaks, merely a slight contraction of the unit cell. ImpedanceSpectroscopy Theslopeoftherealpartofimpedancevsω, plotted in S.I. Figure 5.14(b), gives the Warburg factor (σ) that is inversely related to the diffusion coefficient (D), by the following equation where R is the gas constant, T is the absolute temperature, A is the surface area of the electrode, F is the Faraday constant, and C is the molar concentration of F − ions in the active material. D = R 2 T 2 2A 2 n 2 F 4 C 2 σ 2 (5.1) The calculated value of D at x = 0.2 is 1.75× 10 −16 cm 2 s −1 which is in close agreement with the reported values of 3× 10 −15 cm 2 s −1 for fluoride ion diffusion in oxyfluoride materials. 247 The D values drop to 2.5× 10 −18 cm 2 s −1 and 5.7× 10 −19 cm 2 s −1 at x = 0.4 and x = 0.5 respectively, as shown in S.I. Table 5.1, indicative of change in the mechanism for the oxidative process around this point, and is consistent between the liquid and solid-state cells. 138 1.0 1.5 2.0 2.5 3.0 3.5 � �0.5 (s �0.5 ) 100 200 300 400 500 Z' (k �) F 0.2 ReO 3 F 0.4 ReO 3 F 0.5 ReO 3 (b) 0 100 200 300 400 500 Z' (k �) 100 200 300 400 500 -Z'' (k �) F 0.2 ReO 3 F 0.4 ReO 3 F 0.5 ReO 3 0 4 8 12 16 4 8 12 16 (a) Figure 5.14: (a): Electrochemical impedance spectroscopy on all solid state cells demonstratevariationsinthechargetransferresistanceatdifferentstatesofcharge. As seen in the inset, the resistance rises slightly during fluorination. (b): Diffusion coefficients at different state of charge were calculated from the slope of a plot of ω −0.5 vs Z 0 . x in F x ReO 3 Diffusion Coefficient (D) cm 2 s −1 0.2 1.75× 10 −16 0.4 2.5× 10 −18 0.5 5.7× 10 −19 Table 5.1: Calculated fluoride diffusion coefficients for various states of charge of F x ReO 3 . 139 0.0 2.0 4.0 6.0 8.0 10.0 Time (hours) 0.0 0.5 1.0 1.5 2.0 Voltage vs Cu/Cu 2+ (V) Ion Current (a.u.) Electrochemistry Mass of 2 AU (H 2 ) Mass of 32 AU (O 2 ) Mass of 34 AU (F 2 ) Mass of 44 AU (CO 2 ) Cell at Rest C/10 Charging Figure 5.15: Differential electrochemical mass spectrometry identified H 2 evolu- tion in fluoride ion battery half cells as well as CO 2 formation during the initial charge curve, due to electrolyte decomposition. Differential Electrochemical Mass Spectrometry Experimental 19 F NMR Measurements Additional 19 F NMR measurements were performed. [ppm] 0 -100 - 200 - 300 -136 -141 -150 -189 -170 -174 19 F Figure 5.16: Solid-state 1D 19 F echo MAS NMR spectrum acquired at 18.8 T, 298 K, 23 kHz MAS, with a recycle delay of 2 s for F 0.2 ReO 3 diluted 1:5 by mass with KBr. Partially resolved signals are observed at isotropic 19 F chemical shifts of -136, -141, -150, -170, -174, and -189 ppm (dotted grey lines). 140 Figure 5.17: Solid-state 19 F Hahn-echo MAS NMR saturation-recovery data for F 0.2 ReO 3 . Normalizedintegrated 19 Fsignalintensityisplottedasfunctionsoftheτ delay time for signals with isotropic chemical shifts of (a) -136 ppm, corresponding to a spin-lattice relaxation timeT 1 of 4.0 s, and (b) -150 ppm, corresponding to a T 1 value of 0.3 s; the latter is representative of the other signals at -141, -170, -174, and -189. The solid-state 19 F Hahn-echo NMR spectra were acquired for F 0.2 ReO 3 diluted 1:5 by mass with KBr, and acquired at 18.8 T, 23 kHz, and 298 K. Chemical Shift (ppm) Relaxation Time (sec) -136 4.0 -141 0.3 -150 0.3 -170 0.4 -174 0.4 -189 0.3 Table 5.2: Spin-lattice relaxation times associated with the different resolved 19 F signals of F 0.2 ReO 3 , measured at 18.8 T, 23 kHz MAS, and 298 K from solid-state 19 F Hahn-echo saturation recovery MAS NMR spectra CalculatedNMRShifts 19 FNMRshieldingtensorsforvariousstructuralmod- els were calculated using the fully periodic Gauge Including Projector Augmented Wave method (GIPAW) 223 following a similar approach to Griffin and cowork- ers. 224 In brief, a linear relationship between the calculated chemical shifts for several reference compounds were correlated to their experimentally observed val- ues and fitted to establish relationship: δ exp iso = –k(δ calc iso –δ ref ) where k = 0.68 and 141 δ ref = -160 from the fits. This expression was then used to predict the scale the calculated chemical shift of the models considered in this work to compare more directly with the experiments. -50 0 50 100 150 200 250 � iso calc (ppm) -200 -150 -100 -50 0 50 � iso exp (ppm) R 2 = 99.5% � iso exp = 0.68 � iso calc -160 Figure 5.18: A plot of isotropic shieldings, δ calc iso vs δ exp iso for several fluorine- containing compounds was fit and the expression was used to scale calculated 19 F chemical shifts for comparison with experimental data. 142 Compounds site δ exp iso (ppm) σ calc iso (ppm) δ calc,scaled iso (ppm) AlF 3 a -172.8 -26.1 -177.3 CaF 2 a -107 92.6 -96.6 CdF 2 a -192.1 -38.2 -185.6 HgF 2 a -196.4 -44.0 -189.4 Hg 2 F 2 a -95.8 105.4 -87.9 KF-2H 2 O a -133 30.4 -138.9 LaF 3 a -23.5 216.8 -12.1 b 24.9 252.2 11.9 c 17.1 252.2 11.9 Na 5 Al 3 F 14 a -190 -48.0 -192.2 b -166 -15.7 -170.2 c -182 -44.8 -190.0 SrF 2 a -84.1 118.0 -79.3 KF a -130 36.0 -135.0 LiF a -204 -4.8 -162.8 NaF a -221 -83.4 -216.2 PbF 2 a -20.5 190.9 -29.7 b -57.7 190.9 -29.7 Table 5.3: Calculated 19 F isotropic shieldings, δ calc iso , scaled calculated isotropic chemical shifts, δ calc,scaled iso , and experimental 19 F chemical shifts, δ exp iso , for several fluorine-containing compounds. Experimental values taken from Griffen et al. and the references therein. 224 143 X-ray Emission and X-ray Absorption Spectroscopy In order to charac- terize changes to the oxygen and fluoride environments, XES and XAS measure- ments were conducted on the F and O K-edges. This allowed for identification of changes in the band structure and the local environment. DFT calculations were then performed on a number of candidate crystal structures in order to simulate XES and XAS spectra for comparison to the experimental results. As shown in main text Figure 6, the experimental data was found to be well fit by a model in which Re migrated from an octahedral to a tetrahedral environment to bond to the inserted fluoride ion in the so-called F x ReO 3 - 1F structure. A series of other structures were also considered and the predicted O K-edge XES and XAS structures are shown below. The experimental results are shown on the bottom while the calculated results are shown on the top panel for each given structural model. Figure 5.19: Experimental data (red) compared to simulated X-ray emission and absorption patterns for seven-coordinate FReO 3 formed when Re migrates towards the face of an ReO 6 octahedra to form seven coordinate ReO 6 F unit. 144 Figure 5.20: Experimental data (red) compared to simulated X-ray emission and absorption patterns for mono-ReO 3 F, the structure identified by edge sharing octahedral units. Figure 5.21: Experimental data (red) compared to simulated X-ray emission and absorption patterns for FReO 3 in which F occupies the perovskite A site and O moves to the X site. 145 Figure 5.22: Experimental data (red) compared to simulated X-ray emission and absorption patterns for tet-ReO 3 F model in which all Re atoms migrate to a tetrahedral configuration. 146 Hard X-ray Photoelectron Spectroscopy HAXPES measurements were used to track the Re oxidation state in samples of F x ReO 3 at various states of charge. Figure5.23: TheRe4fX-rayphotoemissionwasusedinordertotrackchangesin the Re oxidation state during fluorination. A doublet at 43 and 46 eV is indicative of Re 6+ , while a doublet at 46 and 48 eV is characteristic of Re 7+ . Charging is seen to result in the growth of the peak at 48 eV, indicating partial oxidation of Re in F x ReO 3 . Neutron Pair Distribution Function Analysis Several different structural motifs were explored to account for the changes observed in the neutron pair distribution function (NPDF) upon fluoride insertion. As can be seen in Fig. 5.24, the data for pristine ReO 3 is in excellent agreement with the structural model. The pattern for F 0.6 ReO 3 is significantly more complicated, and attempts to fit the data have been unsuccessful. Instead, we have calculated the patterns for different structural models and compared them to the raw data. In this example, the atom–atom pair interactions were calculated from a hypothetical cell generated bymakinga 2× 2× 2supercellofReO 3 , insertingfluoridetohalfoftheperovskite A-sites (approx. F 0.6 ), and translating half of the rhenium atoms from octahedral sitestowardtheinsertedfluorides, thusgeneratingtetrahedralReO 3 Fcoordination sites. As can be seen, this correctly models the most significant features of the 147 Re–Re Re–O Re–F O–F O–O C–C C–H G(r) Pristine ReO 3 F 0.6 ReO 3 r (Å) 1.0 2.0 3.0 4.0 5.0 Figure 5.24: Fit of NPDF data for pristine ReO 3 and comparisons of different calculated atom–atom pair interactions with F 0.6 ReO 3 . NPDF—such as the splitting of the first Re–X coordination sphere (1.87 Å to 1.73 and 1.90 Å)—but does not adequately capture the splitting observed in the X–X nearest neighbor distances around 2.47 and 2.68 Å. Figures 5.25–5.28 show several different structural motifs explored to try to explain the structural changes that occurred upon fluoride insertion. Several of them can account for some of the features observed in the data, but none of them can completely account for what is observed. 148 G(r) F 0.6 ReO 3 7-coordinate FReO 3 r (Å) 1 2 3 4 5 Figure 5.25: Comparison of observed data (black dots) for F 0.6 ReO 3 versus the seven-coordinate FReO 3 formed when Re migrates towards the face of an ReO 6 octahedra. G(r) F 0.6 ReO 3 Mono-FReO 3 r (Å) 1 2 3 4 5 Figure 5.26: Comparison of observed data (black dots) for F 0.6 ReO 3 versus the mono-ReO 3 F model. 149 G(r) F 0.6 ReO 3 Tet-FReO 3 r (Å) 1 2 3 4 5 Figure 5.27: Comparison of observed data (black dots) for F 0.6 ReO 3 versus the tetrahedral only ReO 3 F model. G(r) F 0.6 ReO 3 FReO 3 r (Å) 1 2 3 4 5 Figure 5.28: Comparison of observed data (black dots) for F 0.6 ReO 3 versus the FReO 3 model (with fluoride at the perovskite A-site). 150 References [1] US Department of Energy, 2018 Renewable Energy Data Book; 2018. [2] Nkulu, C. B. L.; Casas, L.; Haufroid, V.; Putter, T. D.; Saenen, N. D.; Kayembe-Kitenge, T.; Obadia, P. M.; Mukoma, D. K. W.; Ilunga, J.-M. L.; Nawrot, T. S.; Numbi, O. L.; Smolders, E.; Nemery, B. Nat. Sustain. 2018, 1, 495–504. [3] Tsurukawa, N.; Prakash, S.; Manhart, A. Social impacts of artisanal cobalt mining in Katanga, Democratic Republic of Congo; 2011. [4] Tkaczyk, A. H.; Bartl, A.; Amato, A.; Lapkovskis, V.; Petranikova, M. J. Phys. D: Appl. Phys. 2018, 51, 203001. [5] British Geological Survey, Study on the review of the list of critical raw materials: critical raw materials factsheets; 2017; OCLC: 1111145995. [6] British Geological Survey, Study on the review of the list of critical raw materials: final report.; 2017; OCLC: 1111119694. [7] International Energy Agency, Global EV Outlook 2019; 2019. [8] Masson-Delmotte, V. et al. Global Warming of 1.5 ◦ C. An IPCC Special Report on the impacts of global warming of 1.5 ◦ C above pre-industrial lev- els and related global greenhouse gas emission pathways, in the context of strengthening the global response to the threat of climate change, sustainable development, and efforts to eradicate poverty; 2018. [9] InternationalRenewableEnergyAgency, Global Renewables Outlook: Energy transformation 2050; 2020. [10] Lazard’s Levelized Cost of Energy Analysis - Version 13.0; 2019. [11] Tyson, M.; Bloch, C. Breakthrough Batteries: Powering the Era of Clean Electrification; 2019. 151 [12] Rhodes, J. D.; King, C.; Gulen, G.; Olmstead, S. M.; Dyer, J. S.; Heb- ner, R. E.; Beach, F. C.; Edgar, T. F.; Webber, M. E. Energy Policy 2017, 102, 491–499. [13] Ford, Rivian Announces $500 Million Investment From Ford Partnership To Deliver All-New Ford Battery Electric Vehicle; 2019. [14] Miller, J. Volkswagen increases investment in electric vehicles; 2019. [15] Lewis, M. Wells, Wires, and Wheels... EROCI and the tough road ahead for oil; 2019. [16] Nishi, Y. The Chemical Record 2001, 1, 406–413. [17] Whittingham, M. S. Chem. Rev. 2004, 104, 4271–4302. [18] Jaffe, S. Joule 2017, 1, 225–228. [19] Mineral commodity summaries 2020; US Geological Survey, 2020. [20] Whittingham, M. S. Chem. Rev. 2014, 114, 11414–11443. [21] Larcher, D.; Tarascon, J.-M. Nat. Chem. 2014, 7, 19–29. [22] Liu, J. et al. Nat. Energy 2019, 4, 180–186. [23] Whittingham, M. S. MRS Bulletin 2008, 33, 411–419. [24] Tarascon, J.-M.; Armand, M. Nature 2001, 414, 359–367. [25] Bloomberg, Lithium-ion Battery Pack Costs Worldwide; 2018. [26] McKinsey, A Vision for a Sustainable Battery Value Chain in 2030; 2019. [27] Elshkaki, A.; Shen, L. Energy 2019, 180, 903–917. [28] Manthiram, A. Nat. Commun. 2020, 11, 1550. [29] Dunn, B.; Kamath, H.; Tarascon, J.-M. Science 2011, 334, 928–935. [30] Wang, C. et al. Science 2020, 368, 521–526. [31] Whittingham, M. S. Science 1976, 192, 1126–1127. [32] Goodenough, J. B. Acc. Chem. Res. 2012, 46, 1053–1061. [33] Kamat, P. V. ACS Energy Lett. 2019, 4, 2757–2759. [34] Volta, A. Philos. Trans. Roy. Soc. London 1800, 90, 403–431. 152 [35] Delmas, C.; Ménétrier, M.; Croguennec, L.; Saadoune, I.; Rougier, A.; Pouil- lerie, C.; Prado, G.; Grüne, M.; Fournès, L. Electrochim. Acta 1999, 45, 243–253. [36] Yang, S.; Zavalij, P. Y.; Whittingham, M. S. Electrochem. Commun. 2001, 3, 505–508. [37] Yoshino, A.; Sanechika, K.; Nakajima, T. Secondary battery. 1987; US Patent 4,668,595. [38] Mizushima, K.; Jones, P. C.; Wiseman, P. J.; Goodenough, J. B. Mater. Res. Bull. 1980, 15, 783 – 789. [39] Brandt, K. Solid State Ionics 1994, 69, 173–183. [40] Whittingham, M. S.; Yoshino, A. Lithium-Ion Batteries 2019, [41] Winter, M.; Brodd, R. J. Chem. Rev. 2004, 104, 4245–4270. [42] Dey, A. N. J. Electrochem. Soc. 1971, 118, 1547. [43] Cabana, J.; Monconduit, L.; Larcher, D.; Palacín, M. R. Adv. Mater. 2010, 22, E170–E192. [44] 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. [45] Ellis, B. L.; Lee, K. T.; Nazar, L. F. Chem. Mater. 2010, 22, 691–714. [46] Whittingham M. S., J. A. J. Intercalation chemistry; Academic Press, 1982. [47] Rouxel, J. Intercalated Layered Materials; Springer Netherlands, 1979; pp 201–250. [48] Chebiam, R. V.; Prado, F.; Manthiram, A. Chem. Mater. 2001, 13, 2951– 2957. [49] Montoro, L. A. Electrochem. Solid-State Lett. 1999, 3, 410. [50] 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.; der Ven, A. V.; Seshadri, R.; Dunn, B. S. Chem. Mater. 2020, 32, 4553–4563. [51] Padhi, A. K. J. Electrochem. Soc. 1997, 144, 1188. 153 [52] Andersson, A.; Kalska, B.; Häggström, L.; Thomas, J. O. Solid State Ion. 2000, 130, 41–52. [53] Roberts, M.; Biendicho, J. J.; Hull, S.; Beran, P.; Gustafsson, T.; Svens- son, G.; Edström, K. J. Power Sources 2013, 226, 249–255. [54] Liu, H.; Strobridge, F. C.; Borkiewicz, O. J.; Wiaderek, K. M.; Chap- man, K. W.; Chupas, P. J.; Grey, C. P. Science 2014, 344, 1252817–1252817. [55] Braun, P. V.; Cho, J.; Pikul, J. H.; King, W. P.; Zhang, H. Curr. Opin. Solid State Mater. Sci. 2012, 16, 186–198. [56] Whittingham, M. S. J. Electroanal. Chem. Interfacial Electrochem. 1981, 118, 229–239. [57] Rüdorff, W.; Hofmann, U. Z. Anorg. Allg. Chem. 1938, 238, 1–50. [58] Gao, P.; Reddy, M. A.; Mu, X.; Diemant, T.; Zhang, L.; Zhao-Karger, Z.; Chakravadhanula, V. S. K.; Clemens, O.; Behm, R. J.; Fichtner, M. Angew. Chem. Int. Ed. 2016, 55, 4285–4290. [59] Reddy, M. A.; Fichtner, M. J. Mater. Chem. 2011, 21, 17059. [60] Thieu, D. T.; Fawey, M. H.; Bhatia, H.; Diemant, T.; Chakravadhanula, V. S. K.; Behm, R. J.; Kübel, C.; Fichtner, M. Adv. Funct. Mater. 2017, 27, 1701051. [61] Schoonman, J.; Oversluizen, G.; Wapenaar, K. Solid State Ion. 1980, 1, 211–221. [62] Rongeat,C.; Reddy,M.A.; Diemant,T.; Behm,R.J.; Fichtner,M. J. Mater. Chem. A 2014, 2, 20861–20872. [63] Jiang, X.; Luo, L.; Zhong, F.; Feng, X.; Chen, W.; Ai, X.; Yang, H.; Cao, Y. ChemElectroChem 2019, 6, 2615–2629. [64] Rodríguez-Pérez, I. A.; Ji, X. ACS Energy Lett. 2017, 2, 1762–1770. [65] Seel, J. A.; Dahn, J. R. J. Electrochem. Soc. 2000, 147, 892. [66] Davis, V. K. et al. Science 2018, 362, 1144–1148. [67] Nowroozi, M. A.; Ivlev, S.; Rohrer, J.; Clemens, O. J. Mater. Chem.A2018, 6, 4658–4669. [68] Wang, F.; Yu, H.-C.; Chen, M.-H.; Wu, L.; Pereira, N.; Thornton, K.; der Ven, A. V.; Zhu, Y.; Amatucci, G. G.; Graetz, J. Nat. Commun. 2012, 3. 154 [69] Leriche, J. B.; Hamelet, S.; Shu, J.; Morcrette, M.; Masquelier, C.; Ouvrard, G.; Zerrouki, M.; Soudan, P.; Belin, S.; Elkaïm, E.; Baudelet, F. J. Electrochem. Soc. 2010, 157, A606. [70] Morcrette, M.; Chabre, Y.; Vaughan, G.; Amatucci, G.; Leriche, J.-B.; Patoux, S.; Masquelier, C.; Tarascon, J.-M. Electrochim. Acta 2002, 47, 3137–3149. [71] Grey, C. P.; Tarascon, J. M. Nat. Mater. 2016, 16, 45–56. [72] Meisel, K. Z. Anorg. Allg. Chem. 1932, 207, 121–128. [73] Frevel, L. K.; Rinn, H. W. Acta Crystallogr. 1956, 9, 626–627. [74] O-Keeffe, M.; Andersson, S. Acta Crystallog.A 1977, 33, 914–923. [75] Zhou, S.; Barim, G.; Morgan, B. J.; Melot, B. C.; Brutchey, R. L. Chem. Mater. 2016, 28, 4492–4500. [76] Barim, G.; Cottingham, P.; Zhou, S.; Melot, B. C.; Brutchey, R. L. ACS Appl. Mater. Interfaces 2017, 9, 10813–10819. [77] Liu, Q.; Liu, Y.; Yang, F.; He, H.; Xiao, X.; Ren, Y.; Lu, W.; Stach, E.; Xie, J. ACS Appl. Mater. Interfaces 2018, 10, 4622–4629. [78] Bareño, J.; Shkrob, I. A.; Gilbert, J. A.; Klett, M.; Abraham, D. P. J. Phys. Chem. C 2017, 121, 20640–20649. [79] Feng, Y.; Dou, S.; Wei, Y.; Zhang, Y.; Song, X.; Li, X.; Battaglia, V. S. ACS Omega 2017, 2, 8075–8085. [80] Jesus, L. R. D.; Horrocks, G. A.; Liang, Y.; Parija, A.; Jaye, C.; Wangoh, L.; Wang, J.; Fischer, D. A.; Piper, L. F. J.; Prendergast, D.; Banerjee, S. Nat. Commun. 2016, 7, 12022. [81] Jesus, L. R. D.; Andrews, J. L.; Parija, A.; Banerjee, S. ACS Energy Lett. 2018, 3, 915–931. [82] Cava, R.; Santoro, A.; Murphy, D.; Zahurak, S.; Roth, R. J. Solid State Chem. 1982, 42, 251–262. [83] Cava, R.; Santoro, A.; Murphy, D.; Zahurak, S.; Roth, R. J. Solid State Chem. 1983, 50, 121–128. [84] Rodriguez, E. E.; Llobet, A.; Proffen, T.; Melot, B. C.; Seshadri, R.; Little- wood, P. B.; Cheetham, A. K. J. Appl. Phys. 2009, 105, 114901. 155 [85] Chatterji, T.; Henry, P. F.; Mittal, R.; Chaplot, S. L. Phys. Rev. B 2008, 78, 134105. [86] Chong, Y. Y.; Fan, W. Y. J. Colloid Interface Sci. 2013, 397, 18 – 23. [87] Borkiewicz, O. J.; Shyam, B.; Wiaderek, K. M.; Kurtz, C.; Chupas, P. J.; Chapman, K. W. J. Appl. Crystallogr. 45, 1261–1269. [88] Toby, B. H.; Von Dreele, R. B. J. Appl. Crystallogr. 2013, 46, 544–549. [89] Rietveld, H. M. J. Appl. Crystallogr. 1969, 2, 65. [90] Rodríquez-Carvajal, J.; Roisnel, T. Mater. Sci. Forum 2004, 443, 123–126. [91] Momma, K.; Izumi, F. J. Appl. Crystallogr. 2008, 41, 653–658. [92] Kresse, G.; Hafner, J. Phys. Rev. B 1993, 47, 558–561. [93] Kresse, G.; Hafner, J. Phys. Rev. B 1994, 49, 14251–14271. [94] Kresse, G.; Furthmüller, J. Phys. Rev. B 1996, 54, 11169–11186. [95] Kresse, G.; Furthmüller, J. Comput. Mater. Sci. 1996, 6, 15. [96] Kresse, G.; Joubert, D. Phys. Rev. B 1999, 59, 1758–1775. [97] Perdew, J. P.; Ruzsinszky, A.; Csonka, G. I.; Vydrov, O. a.; Scuseria, G. E.; Constantin, L. a.; Zhou, X.; Burke, K. Phys. Rev. Lett. 2008, 100, 136406. [98] Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865– 3868. [99] Travis, W.; Knapp, C. E.; Savory, C. N.; Ganose, A. M.; Kafourou, P.; Song, X.; Sharif, Z.; Cockcroft, J. K.; Scanlon, D. O.; Bronstein, H.; Pal- grave, R. G. Inorg. Chem. 2016, 55, 3393–3400. [100] Ganose, A. M.; Savory, C. N.; Scanlon, D. O. J. Mater. Chem.A 2017, 5, 7845–7853. [101] Krukau, A. V.; Vydrov, O. A.; Izmaylov, A. F.; Scuseria, G. E. J. Chem. Phys. 2006, 125, 224106. [102] Quackenbush, N. F. et al. Phys. Rev. B 2016, 94, 085105. [103] Savory, C. N.; Ganose, A. M.; Scanlon, D. O. Chem. Mater. 2017, 29, 5156– 5167. [104] Biswas, K.; Rao, C. N. R. J. Phys. Chem. B 2006, 110, 842–845. 156 [105] Nechamkin, H.; Kurtz, A. N.; Hiskey, C. F. J. Am. Chem. Soc. 1951, 73, 2828–2831. [106] Quackenbush, N. F.; Wangoh, L.; Scanlon, D. O.; Zhang, R.; Chung, Y.; Chen, Z.; Wen, B.; Lin, Y.; Woicik, J. C.; Chernova, N. A.; Ong, S. P.; Whittingham, M.; Piper, L. F. J. Chem. Mater. 2015, 27, 8211–8219. [107] Wang, H.; Jang, Y.; Huang, B.; Sadoway, D. R.; Chiang, Y. J. Electrochem. Soc. 1999, 146, 473–480. [108] Dokko, K.; Nishizawa, M.; Horikoshi, S.; Itoh, T.; Mohamedi, M.; Uchida, I. Electrochem. Solid-State Lett. 2000, 3, 125–127. [109] 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. [110] Kim, H.-S.; Cook, J. B.; Tolbert, S. H.; Dunn, B. J. Electrochem. Soc. 2015, 162, A5083–A5090. [111] Mattheiss, L. Phys. Rev. 1969, 181, 987. [112] Cora, F.; Stachiotti, M.; Catlow, C.; Rodriguez, C. J. Phys. Chem. B 1997, 101, 3945–3952. [113] Greiner, M. T.; Rocha, T. C.; Johnson, B.; Klyushin, A.; Knop-Gericke, A.; Schlagl, R. Z. Phys. Chem. 2014, 228, 521–541. [114] Scofield, J. H. T. Theoretical Photoionization Cross Sections from 1 to 1500 keV; 1973. [115] Whittingham, S. M. Intercalation Chemistry; Academic Press, 1982. [116] Sadoway, D. R.; Mayes, A. M. MRS Bulletin 2002, 27, 590–596. [117] Julien, C.; Mauger, A.; Zaghib, K.; Groult, H. Inorganics 2014, 2, 132–154. [118] Kang, K.; Meng, Y. S.; Bréger, J.; Grey, C. P.; Ceder, G. Science 2006, 311, 977–980. [119] Brezesinski, T.; Wang, J.; Polleux, J.; Dunn, B.; Tolbert, S. H. J. Am. Chem. Soc. 2009, 131, 1802–1809. [120] Brezesinski, T.; Wang, J.; Tolbert, S. H.; Dunn, B. Nat. Mater. 2010, 9, 146–151. [121] Cheng, Q.; Liang, J.; Zhu, Y.; Si, L.; Guo, C.; Qian, Y. J. Mater. Chem. A 2014, 2, 17258–17262. 157 [122] Griffith, K. J.; Wiaderek, K. M.; Cibin, G.; Marbella, L. E.; Grey, C. P. Nature 2018, 559, 556–563. [123] Griffith, K. J.; Forse, A. C.; Griffin, J. M.; Grey, C. P. J. Am. Chem. Soc. 2016, 138, 8888–8899. [124] Wadsley, A. D. Acta Crystallogr. 1961, 14, 664–670. [125] Wadsley, A. D. Acta Crystallogr. 1961, 14, 660–664. [126] Roth, R. S.; Wadsley, A. D. Acta Crystallogr. 1965, 18, 724–730. [127] Anderson, J. S. Intercalation Chemistry; Elsevier, 1982; pp 503–538. [128] Bursill, L. A.; Hyde, B. G. Philos. Mag. 1971, 23, 3–15. [129] Bursill, L. A.; Hyde, B. G.; Philp, D. K. Philos. Mag. 1971, 23, 1501–1513. [130] Griffith, K. J.; Seymour, I. D.; Hope, M. A.; Butala, M. M.; Lamon- tagne, L. K.; Preefer, M. B.; Koçer, C. P.; Henkelman, G.; Morris, A. J.; Cliffe, M. J.; Dutton, S. E.; Grey, C. P. J. Am. Chem. Soc. 2019, 141, 16706–16725. [131] Koçer, C. P.; Griffith, K. J.; Grey, C. P.; Morris, A. J. J. Am. Chem. Soc. 2019, 141, 15121–15134. [132] Kim, J. W.; Augustyn, V.; Dunn, B. Adv. Energy Mater. 2011, 2, 141–148. [133] Augustyn, V.; Come, J.; Lowe, M. A.; Kim, J. W.; Taberna, P.-L.; Tol- bert, S. H.; Abruña, H. D.; Simon, P.; Dunn, B. Nat. Mater. 2013, 12, 518–522. [134] Murphy, D. W.; Christian, P. A. Science 1979, 205, 651–656. [135] Murphy, D.; Greenblatt, M.; Cava, R.; Zahurak, S. Solid State Ionics 1981, 5, 327–329. [136] Cordier, S.; Roisnel, T.; Poulain, M. J. Solid State Chem. 2004, 177, 3119– 3126. [137] Andersson, S. Bull. la Soc. Fr. Mineral. Cristallogr. 1967, 90, 522–527. [138] Permér, L.; Lundberg, M. J. Less Common Met. 1989, 156, 145–159. [139] Permér, L. J. Solid State Chem. 1992, 97, 105–114. [140] Segal, E. B. Chem. Health Saf. 2000, 7, 18–23. 158 [141] Rodriguez-Carvajal, J. IUCR Newsl. 2001, 26, 1. [142] Andersson, S.; Åström, A.; Weedon, B. C. L. Acta Chem. Scand. 1964, 18, 2233–2236. [143] Ravel, B.; Newville, M. J. Synchrotron Radiat. 2005, 12, 537–541. [144] Ankudinov, A. L.; Ravel, B.; Rehr, J. J.; Conradson, S. D. Phys. Rev. B 1998, 58, 7565–7576. [145] Ullah, H.; Guerin, K.; Bonnet, P. Eur. J. Inorg. Chem.2019, 2019, 230–236. [146] Idrees, F.; Cao, C.; Butt, F. K.; Tahir, M.; Tanveer, M.; Aslam, I.; Ali, Z.; Mahmood, T.; Hou, J. CrystEngComm 2013, 15, 8146. [147] Andersson, S.; Åsbrink, S.; Reistad, T.; Sparrow, D. R. Acta Chem. Scand. 1964, 18, 2339–2344. [148] Andersson, S.; Åström, A.; Issidorides, C.; Kallner, A. Acta Chem. Scand. 1965, 19, 2136–2138. [149] Permér, L.; Lundberg, M. J. Solid State Chem. 1989, 81, 21–29. [150] Yu, H.; Zhang, J.; Zheng, R.; Liu, T.; Peng, N.; Yuan, Y.; Liu, Y.; Shu, J.; Wang, Z.-B. Mater. Chem. Front. 2020, 1. [151] Zhu, H.; Zheng, Z.; Gao, X.; Huang, Y.; Yan, Z.; Zou, J.; Yin, H.; Zou, Q.; Kable, S. H.; Zhao, J.; Xi, Y.; Martens, W. N.; Frost, R. L. J. Am. Chem. Soc. 2006, 128, 2373–2384. [152] Barner, J. H. V.; Christensen, E.; Bjerrum, N. J.; Gilbert, B. Inorg. Chem. 1991, 30, 561–566. [153] Gross, U.; Rüdiger, S.; Kemnitz, E.; Brzezinka, K.-W.; Mukhopadhyay, S.; Bailey, C.; Wander, A.; Harrison, N. J. Phys. Chem. A 2007, 111, 5813– 5819. [154] Lin, F.; Markus, I. M.; Nordlund, D.; Weng, T.-C.; Asta, M. D.; Xin, H. L.; Doeff, M. M. Nat. Commun. 2014, 5. [155] Zhou, G.; Wang, D.-W.; Li, F.; Zhang, L.; Li, N.; Wu, Z.-S.; Wen, L.; Lu, G. Q. M.; Cheng, H.-M. Chem. Mater. 2010, 22, 5306–5313. [156] Sun, Y.; Wang, C.; Xue, Y.; Zhang, Q.; Mendes, R. G.; Chen, L.; Zhang, T.; Gemming, T.; Rümmeli, M. H.; Ai, X.; Fu, L. ACS Appl. Mater. Interfaces 2016, 8, 9185–9193. 159 [157] Zhao, Y.; Li, X.; Yan, B.; Xiong, D.; Li, D.; Lawes, S.; Sun, X. Adv. Energy Mater. 2016, 6, 1502175. [158] Poizot, P.; Laruelle, S.; Grugeon, S.; Dupont, L.; Tarascon, J.-M. Nature 2000, 407, 496–499. [159] Hong, J. et al. Nat. Mater. 2019, 18, 256–265. [160] Manthiram, A.; Goodenough, J. B. J. Solid State Chem. 1987, 71, 349–360. [161] Armstrong, A. R.; Kuganathan, N.; Islam, M. S.; Bruce, P. G. J. Am. Chem. Soc. 2011, 133, 13031–13035. [162] Fisher, C. A. J.; Prieto, V. M. H.; Islam, M. S. Chem. Mater. 2008, 20, 5907–5915. [163] Bruce, P. G., Ed. Solid state electrochemistry; Cambridge University Press, 1994. [164] Murugan, R.; Thangadurai, V.; Weppner, W. Angew. Chem. Int. Ed. 2007, 46, 7778–7781. [165] Thangadurai, V.; Weppner, W. J. Am. Ceram. Soc. 2005, 88, 411–418. [166] Ahmad, M. M.; Al-Jaafari, A. J. Mater. Sci.: Mater. Electron. 2015, 26, 8136–8142. [167] Clark, S. J.; Segall, M. D.; Pickard, C. J.; Hasnip, P. J.; Probert, M. I. J.; Refson, K.; Payne, M. C. Z. Kristallogr. Cryst. Mater. 2005, 220. [168] Marinkovic, B.; Jardim, P.; de Avillez, R.; Rizzo, F. Solid State Sci. 2005, 7, 1377–1383. [169] Harrison, W.; Cheetham, A.; Faber, J. J. Solid State Chem. 1988, 76, 328– 333. [170] Vanderbilt, D. Phys. Rev. B 1990, 41, 7892–7895. [171] Monkhorst, H. J.; Pack, J. D. Phys. Rev. B 1976, 13, 5188–5192. [172] Kresse, G.; Hafner, J. Phys. Rev. B 1994, 49, 14251–14269. [173] Regoutz, A. et al. Phys. Rev. Materials 2019, 3, 025001. [174] Ganose, A. M.; Jackson, A. J.; Scanlon, D. O. J. Open Source Softw. 2018, 3, 717. 160 [175] Tarascon, J.-M.; Gozdz, A.; Schmutz, C.; Shokoohi, F.; Warren, P. Solid State Ion. 1996, 86-88, 49–54. [176] Morgan, B. J. R. Soc. Open Sci. 2017, 4, 170824. [177] Geller, S. Z. Kristallogr. Cryst. Mater. 1987, 125, 1–47. [178] Schulz, B.; Andersen, H. L.; Bahri, O. K. A.; Johannessen, B.; Liu, J.; Primig, S.; Sharma, N. CrystEngComm 2018, 20, 1352–1360. [179] Yamamoto, T. X-Ray Spectrom. 2008, 37, 572–584. [180] Awaka, J.; Kijima, N.; Hayakawa, H.; Akimoto, J. J. Solid State Chem. 2009, 182, 2046–2052. [181] Kunath-Fandrei, G.; Bastow, T.; Jaeger, C.; Smith, M. Chem. Phys. Lett. 1995, 234, 431–436. [182] Haddix, G. W.; Narayana, M.; Tang, S. C.; Wu, Y. J. Phys. Chem. 1993, 97, 4624–4627. [183] Düvel, A.; Romanova, E.; Sharifi, M.; Freude, D.; Wark, M.; Heitjans, P.; Wilkening, M. J. Phys. Chem. C 2011, 115, 22770–22780. [184] Dupree, R.; Lewis, M. H.; Smith, M. E. J. Appl. Crystallogr. 1988, 21, 109–116. [185] Andrew, E.; Hinshaw, W.; Tiffen, R. J. Magn. Reson. 1974, 15, 191–195. [186] Wiench, J.; Balema, V.; Pecharsky, V.; Pruski, M. J. Solid State Chem. 2004, 177, 648–653. [187] Radin, M. D.; Vinckeviciute, J.; Seshadri, R.; der Ven, A. V. Nat. Energy 2019, 4, 639–646. [188] Boebinger, M. G.; Lewis, J. A.; Sandoval, S. E.; McDowell, M. T. ACS Energy Lett. 2020, 5, 335–345. [189] Mortimer, R. J. Chem. Soc. Rev. 1997, 26, 147. [190] Rives, V.; del Arco, M.; Martín, C. Appl. Clay Sci. 2014, 88-89, 239–269. [191] Nazri, G.; Pistoia, G. Lithium Batteries: Science and Technology; Springer US, 2008. [192] Gamble, F. R.; Osiecki, J. H.; Cais, M.; Pisharody, R.; DiSalvo, F. J.; Geballe, T. H. Science 1971, 174, 493–497. 161 [193] 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. [194] Schafhaeutl, C. J. Prakt. Chem. 1840, 21, 129–157. [195] Beck, F.; Junge, H.; Krohn, H. Electrochim. Acta 1981, 26, 799–809. [196] Beltrop, K.; Meister, P.; Klein, S.; Heckmann, A.; Grünebaum, M.; Wiemhöfer, H.-D.; Winter, M.; Placke, T. Electrochim. Acta 2016, 209, 44–55. [197] Mallouk, T.; Bartlett, N. J. Chem. Soc., Chem. Commun. 1983, 103. [198] Carlin, R. T. J. Electrochem. Soc. 1994, 141, L73. [199] Deunf, É.; Jiménez, P.; Guyomard, D.; Dolhem, F.; Poizot, P. Electrochem. Commun. 2016, 72, 64–68. [200] Schoonman, J. J. Electrochem. Soc. 1976, 123, 1772–1775. [201] Gschwind, F.; Rodriguez-Garcia, G.; Sandbeck, D. J. S.; Gross, A.; Weil, M.; Fichtner, M.; Hörmann, N. J. Fluorine Chem. 2016, 182, 76–90. [202] 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. [203] Roos, A.; Aalders, A.; Schoonman, J.; Arts, A.; de Wijn, H. Solid State Ion. 1983, 9-10, 571–574. [204] Breuer, S.; Gombotz, M.; Pregartner, V.; Hanzu, I.; Wilkening, M. Energy Storage Mater. 2019, 16, 481–490. [205] A. Reddy, M.; Fichtner, M. J. Mater. Chem. 2011, 21, 17059. [206] Nowroozi, M.; Clemens, O. ACS Appl. Energy Mater. 2018, 1, 6626–6637. [207] Zaheer, W.; Andrews, J. L.; Parija, A.; Hyler, F. P.; Jaye, C.; Weiland, C.; Yu, Y.-S.; Shapiro, D. A.; Fischer, D. A.; Guo, J.; Velázquez, J. M.; Baner- jee, S. ACS Energy Lett. 2020, 2520–2526. [208] Christe, K. O.; Wilson, W. W.; Wilson, R. D.; Bau, R.; Feng, J. A. J. Am. Chem. Soc. 1990, 112, 7619–7625. [209] Sharma, R. K.; Fry, J. L. J. Org. Chem. 1983, 48, 2112–2114. 162 [210] Neuefeind, J.; Feygenson, M.; Carruth, J.; Hoffmann, R.; Chipley, K. K. Nucl. Instrum. Methods Phys. Res., Sect. B 2012, 287, 68–75. [211] Rongeat, C.; Reddy, M. A.; Witter, R.; Fichtner, M. ACS Appl. Mater. Interfaces 2014, 6, 2103–2110. [212] McCloskey, B. D.; Valery, A.; Luntz, A. C.; Gowda, S. R.; Wallraff, G. M.; Garcia, J. M.; Mori, T.; Krupp, L. E. J. Phys. Chem. C 2013, 4, 2989–2993. [213] McCloskey, B. D.; Scheffler, R.; Speidel, A.; Bethune, D. S.; Shelby, R. M.; Luntz, A. C. J. Am. Chem. Soc. 2011, 133, 18038–18041. [214] Yesinowski, J. P.; Ladouceur, H. D.; Purdy, A. P.; Miller, J. B. J. Chem. Phys. 2010, 133, 234509. [215] Thurber, K. R.; Tycko, R. J. Magn. Reson. 2009, 196, 84–87. [216] Thomas, J. C.; Van der Ven, A. Phys. Rev. B 2013, 88, 214111. [217] Puchala, B.; Van der Ven, A. Phys. Rev. B 2013, 88, 094108. [218] Van der Ven, A.; Thomas, J. C.; Puchala, B.; Natarajan, A. R. Annu. Rev. Mater. Res. 2018, 48, 27–55. [219] Kresse, G.; Hafner, J. Phys. Rev. B 1993, 47, 558. [220] Kresse, G.; Hafner, J. Phys. Rev. B 1994, 49, 14251. [221] Kresse, G.; Furthmüller, J. Comput. Mater. Sci. 1996, 6, 15–50. [222] Kresse, G.; Furthmüller, J. Phys. Rev. B 1996, 54, 11169. [223] Pickard, C. J.; Mauri, F. Phys. Rev. B 2001, 63, 24. [224] Griffin, J. M.; Yates, J. R.; Berry, A. J.; Wimperis, S.; Ashbrook, S. E. J. Am. Chem. Soc. 2010, 132, 15651–15660. [225] Schwarz, K.; Blaha, P. Comp. Mater. Sci. 2003, 28, 259–273. [226] Blaha, P.; Schwarz, K.; Tran, F.; Laskowski, R.; Madsen, G. K. H.; Marks, L. D. J. Chem. Phys. 2020, 152, 074101. [227] McCulloch, D.; Lau, D.; Nicholls, R.; Perkins, J. Micron 2012, 43, 43–48. [228] Hébert, C.; Luitz, J.; Schattschneider, P. Micron 2003, 34, 219–225. [229] Luitz, J.; Maier, M.; Hébert, C.; Schattschneider, P.; Blaha, P.; Schwarz, K.; Jouffrey, B. Eur. Phys. J. B 2001, 21, 363–367. 163 [230] Ferretti, A.; Rogers, D. B.; Goodenough, J. B. J. Phys. Chem. Solids 1965, 26, 2007–2011. [231] Tran, N.; Croguennec, L.; Menetrier, M.; Weill, F.; Biensan, P.; Jordy, C.; Delmas, C. Chem. Mater. 2008, 20, 4815–4825. [232] Armstrong, A. R.; Holzapfel, M.; Novák, P.; Johnson, C. S.; Kang, S.-H.; Thackeray, M. M.; Bruce, P. G. J. Am. Chem. Soc. 2006, 128, 8694–8698. [233] Van der Ven, A.; Deng, Z.; Banerjee, S.; Ong, S. P. Chem. Rev. 2020, 120, 6977–7019. [234] Zagorac, D.; Müller, H.; Ruehl, S.; Zagorac, J.; Rehme, S. J. Appl. Crystal- logr. 2019, 52, 5. [235] Supeł, J.; Marx, R.; Seppelt, K. Z. Anorg. Allg. Chem. 2005, 631, 2979– 2986. [236] Ruiz-Preciado, M. A.; Kubicki, D. J.; Hofstetter, A.; McGovern, L.; Futscher, M. H.; Ummadisingu, A.; Gershoni-Poranne, R.; Zakeerud- din, S. M.; Ehrler, B.; Emsley, L.; Milić, J. V.; Grätzel, M. J. Am. Chem. Soc. 2020, 142, 1645–1654. [237] Cho, S. H.; Ghosh, S.; Berkson, Z. J.; Hachtel, J. A.; Shi, J.; Zhao, X.; Reimnitz, L. C.; Dahlman, C. J.; Ho, Y.; Yang, A.; Liu, Y.; Idrobo, J.-C.; Chmelka, B. F.; Milliron, D. J. Chem. Mater. 2019, 31, 2661–2676. [238] Yesinowski, J. P.; Berkson, Z. J.; Cadars, S.; Purdy, A. P.; Chmelka, B. F. Phys. Rev. B 2017, 95, 23. [239] Greaves, C.; Francesconi, M. G. Curr. Opin. Solid State Mater. Sci. 1998, 3, 132–136. [240] Greaves, C.; Kissick, J. L.; Francesconi, M. G.; Aikens, L. D.; Gillie, L. J. J. Mater. Chem. 1999, 9, 111–116. [241] McCabe, E. E.; Greaves, C. J. Fluorine Chem. 2007, 128, 448–458. [242] Case, G. S.; Hector, A. L.; Levason, W.; Needs, R. L.; Thomas, M. F.; Weller, M. T. J. Mater. Chem. 1999, 9, 2821–2827. [243] Li, R. K.; Greaves, C. Phys. Rev. B 2000, 62, 3811–3815. [244] Pellet, N.; Teuscher, J.; Maier, J.; Grätzel, M. Chem. Mater. 2015, 27, 2181–2188. 164 [245] Elmelund, T.; Scheidt, R. A.; Seger, B.; Kamat, P. V. ACS Energy Lett. 2019, 4, 1961–1969. [246] Walsh, A.; Stranks, S. D. ACS Energy Lett. 2018, 3, 1983–1990. [247] Katayama, T.; Chikamatsu, A.; Hirose, Y.; Takagi, R.; Kamisaka, H.; Fuku- mura, T.; Hasegawa, T. J. Mater. Chem. C 2014, 2, 5350–5356. 165

Abstract (if available)

Abstract Understanding the structural transformations that materials undergo during the insertion and deinsertion of ions is crucial for designing high performance intercalation electrode materials. In this dissertation, a series of studies investigates the role of structural changes on the electrochemical performance of intercalation electrodes and identifies common themes between crystallographic motifs and electrochemical behavior. I present a study of the structural distortions of the metallic defect perovskite ReO₃ upon lithiation with the goal of determining whether these distortions are driven by polaronic charge transport (i.e. the electrons and ions moving through the lattice in a coupled way) due to the semiconducting nature of most oxide hosts. I find that the cubic structure of ReO₃ experiences multiple phase changes involving the correlated twisting of rigid octahedral subunits during the insertion of two equivalents of Li-ions

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

Understanding intercalation-driven structural transformations in 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

PDF

Small organic molecules in all-organic redox flow batteries for grid-scale energy storage

PDF

The effects of crystal structure and composition on the magnetism of garnets & phyllosilicates

PDF

Design and characterization of flow batteries for large-scale energy storage

PDF

Nanomaterials for energy storage devices and electronic/optoelectronic devices

PDF

Two-dimensional metal dithiolene metal-organic frameworks as conductive materials for solar energy conversion

PDF

Controlling structure and photophysical properties of bi triple perovskite through substitutional series Cs₃Bi₂Br₍₉₋ₓ₎Iₓ

PDF

Modification of electrode materials for lithium ion batteries

PDF

First steps of solar energy conversion; primary charge generation processes in condensed phase coupled chromophores

PDF

Advancing lithium batteries and related electrochemical technologies for a sustainable future

PDF

Studies on iron-chloride redox flow battery for large scale energy storage

PDF

Nanostructure design of sulfur cathodes and lithium metal anodes for lithium-ion batteries

PDF

Vibrational sum frequency generation studies of electrode interfaces

PDF

Selective fluoroalkylation methods and synthesis of water-soluble organic molecules for organic redox flow batteries

PDF

Understanding the genetics, evolutionary history, and biomechanics of the mammalian penis bone

PDF

Electrochemical investigations and imaging tools for understanding extracellular electron transfer in phylogenetically diverse bacteria

Asset Metadata

Creator Bashian, Nicholas H. (author)

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

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Chemistry

Publication Date 12/04/2020

Defense Date 09/30/2020

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

Tag battery,energy storage,intercalation,lithium-ion,materials chemistry,OAI-PMH Harvest,operando,X-ray absorption,X-ray diffraction

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Melot, Brent C. (committee chair), Narayan, Sri R. (committee member), Ravichandran, Jayakanth (committee member)

Creator Email nbashian@usc.edu,nickbashian@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-402304

Unique identifier UC11666597

Identifier etd-BashianNic-9180.pdf (filename),usctheses-c89-402304 (legacy record id)

Legacy Identifier etd-BashianNic-9180.pdf

Dmrecord 402304

Document Type Dissertation

Rights Bashian, Nicholas H.

Type texts

Source 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 a...

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