University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content UNDERSTANDING THE STRUCTURE-PROPERTY RELATIONSHIP IN ELECTRODE MATERIALS FOR ELECTROCHEMICAL ENERGY STORAGE by Shiliang Zhou 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) May 2017 Copyright 2017 Shiliang Zhou Dedication Dedicated to my family. ii Acknowledgments The journey started when I first arrived at the LAX airport in a clear summer night of 2011, and ends when I am due to go back to China in the spring of 2017. Walking inches close towards the finish line of this journey, I have nothing but endless appreciation to all the wonderful people I have met along the way. First and foremost, I want to thank my advisor, Professor Brent Melot. Brent’s tremendous passion and curiosity towards scientific research is what kept me constantly motivated and inspired in the past 5 years. And his immense knowl- edge and excellent management skills made sure that I was always on the right track. Moreover, ever since I joined the group, Brent has been a great friend. His kindness and easy accessibility has helped me at every corner of my life at USC. He not only walked me through each aspect of graduate school, whether it is designing an experiment or writing a paper, but also allowed me to learn almost everything at my own pace, which I will always be grateful for. I owe a great deal to all the past and present members of the Melot group. I want to thank Wolfgang, not only for the right attitude and time management skills he showed me as a graduate student, but also for the great soccer games and banters we shared together. Thanks to Abbey, Kelsey, Erica, Joanna, Justin and Kyle, for taking care of me and, taking care of each other whenever in need. All the days that we spent together, helping each other out of a research problem iii or a plumbing duty, destressing at the beach, and telling light-hearted jokes, have all been invaluable to me. The research I have done would be impossible without the help from my col- laborators, all of whom have been exceptionally smart, skillful and professional. Professor David Scanlon, Professor Benjamin Morgan, Dr. Katherine Page, Dr. Moulay Sougrati, and Dr. Jue Liu have all offered extraordinary help. I also want to thank all the faculty and staff members at the USC Chemistry Depart- ment, whose brilliant and diligent work is what made possible for me to enjoy this wonderful journey as a chemistry graduate student. The Richard Brutchey Group, the Sri Narayan Group and the Mark Thompson Group offered precious help in allowing me to share their lab spaces and equipment when our lab was still under construction and continued to assist me throughout the years with their expertise and open minds. Outside research, I have explored an eventful and joyful life. This of course would be impossible without all the friends that I have met in the US. The friends I met during first year of graduate school have always been special to me. They helped me navigate through the periods of cultural shock, adaption, experimen- tation, and immersion. These friends include all my classmates, especially Lily, Kim, Lena, Sean, Piyush, Mayro, Xu, Yinchu, Alvin, John, Derek, Narin, Hang, Atanu, Subodh and many more. They also include Alexa, Ben, Xin, Chao, Zoey, and Linfei. As an international graduate student, I have also been fortunate to have many compatriot friends with me throughout graduate school. I want to thank Jun- song, Weisheng, Meng, Maoqi, Long, Guangtong, Lunce, Yongjian, and Kang, for the countless hours that we spend together, playing basketball, video games, and celebrating each other’s birthday. Special thanks to Xing, Bo, Dan, Yangyang, iv and Yanan for being the friends that I can always turn to whenever I needed help. Traveling, especially road trips, has been my favorite way of exploring this for- eign land, not just for all the beautiful parks and cities that the trips are heading into, but also for all the precious moments that I have shared with friends on the road. I have shared road trips with Sijia, Jue, Junchong, Vicky, Zhaoyang, Xin, Haipeng, Henry, Youyu, and many of the friends that I have already mentioned. Ever since I was 12 years old, I had to live near school but away from home. As my academic quest extended further and further, so was my distance from home, from dozens of miles in a different city to hundreds of miles in a different province, and to thousands of miles in a different continent. But the constant support and endless love from my family has never been distant. In fact, they became closer and closer. They will feel weird if I say thanks to them. And I know I can never pay back what I am indebted to them. The only thing I can hope for, is many more days and years that I can share with them in the rest of my life. v Curriculum Vitae Education University of Southern California, Los Angeles, CA 2011-2017 Ph.D., Chemistry University of Science and Technology of China, Hefei, China 2007-2011 B.S., Materials Science and Engineering Publications 10. S. Zhou, Erica S. Howard, Jue Liu, Kyle Nolan, Sankarganesh Krishnamoor- thy, Moulay T. Sougrati, Geo Rangel, G. K. Surya Prakash, Katharine Page, Brent C. Melot Leveraging Complex Cation Distributions in Iron-based Clays for Elec- trochemical Energy Storage (submitted for publication) 9. M. E. Marisa, S. Zhou, B. C. Melot, G. F. Peaslee, J. R. Neilson Paracrys- talline Disorder from Phosphate Ion Orientation and Substitution in Synthetic Bone Mineral Inorg. Chem. 55 (2016) , 12290-12298 [doi] 8. S. Zhou, G. Barim, B. J. Morgan, B. C. Melot, and R. L. Brutchey Influ- ence of Rotational Distortions on Li + - and Na + -Intercalation in Anti-NASICON Fe 2 (MoO 4 ) 3 Chem. Mater. 28 (2016), 4492-4500 [doi] 7. M. M. Butala, K. R. Danks, M. A. Lumley, S. Zhou, B. C. Melot, and R. Seshadri MnO Conversion in Li-Ion Batteries: In Situ Studies and the Role of Mesostruc- turing ACS Applied Mater. Interfaces 8 (2016), 6496-6503 [doi] vi 6. B. Lopez-Bermudez, W. G. Zeier, S. Zhou, D. O. Scanlon, B. J. Morgan, and B. C. Melot Lithium-Ion Conductivity in Li 6 Y(BO 3 ) 3 : A Thermally and Electrochem- ically Robust Solid Electrolyte J. Mater. Chem. A 4 (2016), 4972-6979 [doi] 5. S. Zhou, W. G. Zeier, M. C. Kemei, M. T. Sougrati, M. Mecklenburg, B. C. Melot Hydrothermal Preparation and Magnetic Properties of NaFeSi 2 O 6 : Nanowires vs Bulk Samples Inorg. Chem. 53 (2014), 12396-12401 [doi] 4. K. K. Bass, R. E. McAnally, S. Zhou, P. I. Djurovich, M. Thompson, B. C. Melot Influence of Moisture on the Preparation, Crystal Structure, and Photo- physical Properties of Organohalide Perovskites Chem. Commun. 50 (2014), 15819-15822 [doi] 3. S. Zhou, G. King, D. O. Scanlon, M. T. Sougrati, B. C. Melot Low Temperature Preparation and Electrochemical Properties of LiFeSi 2 O 6 J. Electrochem. Soc. 161 (2014), A1642-A1647 [doi] 2. W. G. Zeier, S. Zhou, B. Lopez-Bermudez, K. L. Page, B. C. Melot Dependence of the Li-Ion Conductivity and Activation Energies on the Crystal Structure and Ionic Radii in Li 6 MLa 2 Ta 2 O 12 ACS Appl. Mater. Interfaces 6 (2014), 10900-10907 [doi] 1. S. P. Culver, F. A. Rabuffetti, S. Zhou, M. Mecklenburg, Y. Song, B. C. Melot, R. L. Brutchey Low-Temperature Synthesis of AMoO 4 (A= Ca, Sr, Ba) Scheelite Nanocrystals Chem. Mater. 25 (2013), 4129-4134 [doi] vii Contents Dedication ii Acknowledgments iii Curriculum Vitae vi List of Tables x List of Figures xi Abstract xvi 1 Introduction 1 1.1 The Need for Energy Storage . . . . . . . . . . . . . . . . . . . . 1 1.2 Challenges in Developing Rechargeable Batteries for Large-Scale Energy Storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.3 Research on Intercalation Electrode Materials . . . . . . . . . . . 5 2 Low Temperature Preparation and Electrochemical Properties of Pyrox- ene LiFeSi 2 O 6 10 2.1 Experimental Details . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . 15 2.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3 Preparation and Magnetic Properties of NaFeSi 2 O 6 : Nanowires vs Bulk Samples 26 3.1 Experimental Details . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.1.1 Synthetic Methods . . . . . . . . . . . . . . . . . . . . . . 28 3.1.2 Physical Characterization . . . . . . . . . . . . . . . . . . 29 3.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . 31 3.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 viii 4 Leveraging Complex Cation Distributions in Iron-based Clays for Elec- trochemical Energy Storage 44 4.1 Experimental Details . . . . . . . . . . . . . . . . . . . . . . . . . 45 4.1.1 Synthetic Methods . . . . . . . . . . . . . . . . . . . . . . 45 4.1.2 Physical Characterization . . . . . . . . . . . . . . . . . . 46 4.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . 48 4.2.1 Effect of reaction conditions . . . . . . . . . . . . . . . . . 53 4.2.2 Electrochemistry . . . . . . . . . . . . . . . . . . . . . . . 58 4.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 5 Influence of Rotational Distortions on Li + - and Na + -Intercalation in Anti-NASICON Fe 2 (MoO 4 ) 3 63 5.1 Experimental Details . . . . . . . . . . . . . . . . . . . . . . . . . 65 5.1.1 Synthesis of Fe 2 (MoO 4 ) 3 . . . . . . . . . . . . . . . . . . . 65 5.1.2 Chemical insertion . . . . . . . . . . . . . . . . . . . . . . 65 5.1.3 Structural Characterization . . . . . . . . . . . . . . . . . 66 5.1.4 X-ray photoelectron and Raman spectroscopy . . . . . . . 67 5.1.5 Electrochemical measurements . . . . . . . . . . . . . . . 68 5.1.6 Computational . . . . . . . . . . . . . . . . . . . . . . . . 68 5.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 5.2.1 Li + insertion . . . . . . . . . . . . . . . . . . . . . . . . . 71 5.2.2 Na + insertion . . . . . . . . . . . . . . . . . . . . . . . . . 76 5.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 5.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 6 Influence of local distortion on Li + - and Na + -intercalation in ReO 3 88 6.1 Experimental Details . . . . . . . . . . . . . . . . . . . . . . . . . 90 6.1.1 Electrochemical Testing . . . . . . . . . . . . . . . . . . . 91 6.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . 92 6.2.1 Li + Intercalation . . . . . . . . . . . . . . . . . . . . . . . 94 6.2.2 Na + Intercalation . . . . . . . . . . . . . . . . . . . . . . . 98 6.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 Reference List 105 ix List of Tables 2.1 Unit cell parameters from the Rietveld refinement of the sintered LiFeSi 2 O 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.2 Atomic positions and atomic displacement parameters from the Rietveld refinement of the sintered LiFeSi 2 O 6 . . . . . . . . . . . 17 4.1 Distribution of Fe environments of phyllosilicates prepared with increasing KOH concentration . . . . . . . . . . . . . . . . . . . . 55 4.2 Distribution of Fe environments of phyllosilicates prepared with different PVP additions . . . . . . . . . . . . . . . . . . . . . . . . 56 x List of Figures 1.1 Illustration of a lithium ion battery . . . . . . . . . . . . . . . . . 2 1.2 Structures of representative intercalation electrode materials . . . 6 2.1 Crystal structure of LiFeSi 2 O 6 . . . . . . . . . . . . . . . . . . . . 11 2.2 Results of the Rietveld refinements of as-prepared LiFeSi 2 O 6 . . . 15 2.3 Mössbauer spectrum of the as prepared LiFeSi 2 O 6 . . . . . . . . . 18 2.4 Electrochemical cycling of LiFeSi 2 O 6 . . . . . . . . . . . . . . . . 19 2.5 The derivative of the voltage-composition curve of sintered LiFeSi 2 O 6 cycled at C/200. The derivative shows redox peaks around 2 V. The inset shows the capacity retention curve for the batter- ies cycled at C/50. After the initial three cycles, the capacity is retained more than 97% with 25 cycles. . . . . . . . . . . . . . . 20 2.6 Electrochemical performance of LiFeSi 2 O 6 clacinated at 700 ◦ C and 900 ◦ C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.7 Electrochemical cycling of LiFeSi 2 O 6 . . . . . . . . . . . . . . . . 22 2.8 Illustration of the crystal structure of LiFeSi 2 O 6 with atoms repre- sented as ellipsoids . . . . . . . . . . . . . . . . . . . . . . . . . . 23 xi 3.1 Crystal structure of C 2/c NaFeSi 2 O 6 . The SiO 4 tetrahedra are grey and the FeO 6 octahedra are green. The sodium and oxygen atoms are shown as yellow and orange spheres respectively. (a) shows the close packing Si 2 O 6 layered structure with sodium and iron ions located in the octahedral sites. (b) shows the zig-zag edge- sharing FeO 6 chains connected by the zig-zag corner-sharing SiO 4 chains through the oxygen ions on their edges. . . . . . . . . . . 28 3.2 Synchrotron XRD patterns of products from hydrothermal prepa- ration at temperatures ranging from 160 ◦ C to 220 ◦ C. A minimum temperature of 180 ◦ C is required to obtain crystalline NaFeSi 2 O 6 . 32 3.3 (a) SEM image of the as-prepared NaFeSi 2 O 6 , which is shown to consist of bundles of nanowires. (c) TEM image of the as-prepared sample reveals its crystallinity. . . . . . . . . . . . . . . . . . . . . 34 3.4 Rietveld refinement of as-prepared and sintered NaFeSi 2 O 6 against synchrotron X-ray diffraction patterns obtained on the 11-BM beam- line at Argonne National Laboratory. χ 2 values of 2.42 and 2.28 for the refinements on the as-prepared sample and the sintered sample were obtained respectively. . . . . . . . . . . . . . . . . . 35 3.5 Comparison of the room temperature Mössbauer patterns for the as-prepared and sintered phases. Best fits to the experimental data are obtained with two doublets for the as-prepared sample powder and a single doublet for the sintered one. . . . . . . . . . . . . . . 36 3.6 Comparison of the temperature-dependent magnetic susceptibilty for the as-prepared and sintered phases. . . . . . . . . . . . . . . 37 xii 3.7 Comparison of the specific heat capacity (a) and change in entropy associated with the magnetic ordering transition (b) for the as- prepared and sintered phases. . . . . . . . . . . . . . . . . . . . . 40 3.8 Comparison of the constant temperature magnetization loops for the as-prepared and sintered NaFeSi 2 O 6 at 2 K. The magnetization of the as-prepared NaFeSi 2 O 6 nanowires shows a nearly linear response to the applied magnetic field, while two distinct field- induced features are observed in the sintered sample. . . . . . . . 42 4.1 Crystal structure of 2:1 phyllosilicates . . . . . . . . . . . . . . . 45 4.2 SEM image and synchrotron XRD of as-prepared sample . . . . . 49 4.3 Profile refinement of the local structure of KFe 2.75 Si 3.25 O 10 (OH) 2 . 51 4.4 Refinement results of KFe 2.75 Si 3.25 O 10 (OH) 2 using the Ritveld aver- age structure method and the numerical 30-layer supercell model 52 4.5 XRD patterns of Fe-phyllosilicates prepared with different KOH concentrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 4.6 SEM images of samples prepared with different KOH concentrations 55 4.7 Mössbauer spectrum of the as prepared sample with K:Fe ratio of 10:1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 4.8 Electrochemical performance of as-prepared sample . . . . . . . . 58 4.9 Cyclic Voltammetry of the KFeSiO-based cell . . . . . . . . . . . . 59 5.1 Crystal structure of Fe 2 (MoO 4 ) 3 . . . . . . . . . . . . . . . . . . . 70 5.2 Ritveld refinements of pristine Fe 2 (MoO 4 ) 3 . . . . . . . . . . . . . 71 5.3 Galvanostatic cycling and in-situ XRD of Fe 2 (MoO 4 ) 3 against Li + insertion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 xiii 5.4 Amplimodes analysis between pristine Fe 2 (MoO 4 ) 3 and lithiated Li 2 Fe 2 (MoO 4 ) 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 5.5 Galvanostatic cycling and in-situ XRD of Fe 2 (MoO 4 ) 3 against Na + insertion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 5.6 Lattice evolution along Na + insertion into Fe 2 (MoO 4 ) 3 . . . . . . 78 5.7 Total scattering data for pristine, Li + -inserted, and Na + -inserted Fe 2 (MoO 4 ) 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 6.1 The structure of ReO 3 with a space group of P m ¯ 3m with viewing direction of (1 0 0) (a) and (1 1 1) (b). Oxygen and rhenium atoms are shown in orange and brown spheres, respectively. The ReO 6 octahedra share corners with each other to form a three- dimensional defect-perovskite network. . . . . . . . . . . . . . . 89 6.2 TEM micrograph of as-prepared ReO 3 nanoparticles. . . . . . . . 93 6.3 Refinement against Laboratory XRD patterns of bulk ReO 3 sam- ples. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 6.4 Cyclic voltammetry scans of ReO 3 microelectrodes against Li/Li + in a three electrode cell inside the glovebox. . . . . . . . . . . . 95 6.5 Two-dimensional presentation of in-situ XRD patterns collected during a full discharging and charging cycle using ReO 3 powder as the working electrode in a swagelok cell. The y-axis corresponds to the equivalent amount of Li (de)inserted in ReO 3 , calculated from the charge delivered by the potentiostat. The color indicates the instensity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 6.6 Capacity retention of Li + -intercalation in ReO 3 with lower cut-off voltage set at 2.8 V and 0.5 V. The higher cut-off voltage is set at 4.0 V for both occasions. . . . . . . . . . . . . . . . . . . . . . . . 97 xiv 6.7 Structure of LiReO 3 viewed from the (0 0 1) direction (a) and a zoomed-in illustration of the local coordination of Li in LiReO 3 (b). Structure representation constructed with reported structural data by Cava et. al[1] using the VESTA program.[2] . . . . . . . . . . 98 6.8 (a)Cyclic voltammetry scans of ReO 3 microelectrodes against Na/Na + at a rate of 1 mV/s in a three electrode cell inside the glove- box.(b)Capacity retention of the same electrodes cycled between different voltage cut off at a rate of 1 mV/s. The higher voltage cut off is set at 3.5 V for both occasions. . . . . . . . . . . . . . . 100 6.9 (a) CV scans of ReO 3 microelectrodes between 2.4 V and 3.5 V against Na/Na + between 1 mV/s and 100 mV/s. (b) Plot and lin- ear fitting of log(peak current) against log(sweep rate). The linear fit (R 2 = 0.9886) results in a slope of 0.5086 with a standard error of 0.0222. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 6.10 Insitu XRD patterns of ReO 3 cycled against Na/Na + in a swagelok cell with powder ReO 3 as working electrode. . . . . . . . . . . . . 102 xv Abstract The development of sustainable, robust and large-scale energy storage is crit- ical for renewable energy sources to assume a major role in our energy supply scheme. Among the pool of technologies proposed for large-scale energy stor- age, rechargeable Li- and Na-ion batteries offer substantial advantages in terms of mobility, efficiency, and power density. However, for such technologies to be used in a grid scale, dramatic reduction in the cost is needed for them to become viable. And although Li-ion batteries have already enjoyed certain degree of suc- cess in electric vehicles, they still face challenges such as cost, and charging rate before they can become a dominant player in the automotive market. The work presented here aims to explore a large family of mineral materi- als, silicates, for their potential application in electrochemical storage and other functional applications. With a wide range of characterization tools, we set out to not only understand the electrochemical performance of these materials, but also find out how their functional properties are related to their structures. We then took a similar approach in two oxide phases, Fe 2 (MoO 4 ) 3 and ReO 3 , to further understand how the structural changes in these materials correlate with their elctrochemical properties. First we found that LiFeSi 2 O 6 undergoes a reversible electrochemical reaction against Li/Li + centered around 2 V with capacities near 60% of the theoretical xvi maximum. We employ high resolution synchrotron X-ray and neutron diffraction to characterize the structure and correlate the rigid connectivity with the very slow kinetics of diffusion. We also use computational tools to understand the origin of the low potential compared with other Fe-based electrodes. We then use similar low-temperature techniques to prepare high-purity NaFeSi 2 O 6 and an iron-based muscovite phase with unique microstructures. Interestingly, we found that the NaFeSi 2 O 6 nanowires prepared by our low-temperature tech- nique exhibit substantial difference in temperature- and field-dependent mag- netic properties. The differences can likely be attributed to the reduced particle size and increased number of spins on the surface of the nanowires. In the case of our synthetic iron muscovite, a reversible capacity equal to 0.4 mole Li per formula unit was obtained. Cyclic voltammetry analysis showed significant con- tribution of the electrochemical capacity come from a surface controlled process. This is probably due to the limited amount of sites available for Li + intercalation. In our study of the electrochemical properties of Fe 2 (MoO 4 ) 3 , significant dif- ferences in the structural and electrochemical properties during the intercalation of Li + and Na + ions were observed. To understand the origin of this behav- ior, we use a combination of in-situ X-ray and high-resolution neutron diffrac- tion, total scattering, electrochemical measurements, density functional theory calculations, and symmetry-mode analysis. We find that for Li + -intercalation, which proceeds via a two-phase monoclinic-to-orthorhombic (Pbcn) phase tran- sition, the host lattice undergoes a concerted rotation of rigid polyhedral subunits driven by strong interactions with the Li+ ions, leading to an ordered lithium arrangement. Na + -intercalation, which proceeds via a two-stage solid solution insertion into the monoclinic structure, similarly produces rotations of the lattice polyhedral subunits. However, using a combination of total neutron scattering xvii data and density-functional theory calculations, we find that while these rota- tional distortions upon Na+ intercalation are fundamentally the same as for Li+ intercalation, they result in a far less coherent final structure, with this difference attributed to the substantial difference between the ionic radii of the two alkali metals. Finally, we explored the electrochemical Li + - and Na + -intercalation into defect- perovskite type ReO 3 . The polhedral rotational distortions that we found dictates the mechanism of the guest ion intercalation in Fe 2 (MoO 4 ) 3 also seems to play an important role in ReO 3 . Theoretical two Li per formula interlcation in ReO 3 is possible with multiple stages during the electrochemical cycling. However, due to the huge mismatch between the sizes of Li + ions and the interstitial void, and the enormous twist of the host lattice required to accommodate the inserted Li + , Li + -intercalation shows poor reversibility. On the other hand, reversibility of Na + -intercalation is greatly improved if it is limited to the solid solution pro- cess on the first stage. The closer fit between Na + and the interstitial space in ReO 3 , which reduces the twist of host framework and the local distortion, likely explains the enhanced reversibility. xviii Chapter 1 Introduction 1.1 The Need for Energy Storage Advancements in the transition of our society’s energy dependence from fos- sil fuels to renewable sources have been constant and promising over the last decade.[3, 4, 5, 6] The levelised cost of solar cells and wind power have become much more cost-competitive and the capacity of electricity generated from renewable sources in the US have increased by 91% from 2005 to 2015[4, 6] However, the electricity generated by solar and wind powers still accounts for only 5% of the world’s total.[7, 6] Moreover, with the world population pro- jected to surpass 11 billion by the end of the 21st century and the emergence of increasingly urgent consequences of heavy exploitation of fossil fuels, efforts to increase the share of renewable sources in the world energy landscape still need to be revved up.[3, 8, 9] Besides the cost associated with harvesting renew- able energy and integrating them into our current distribution systems, there is also mounting needs for sustainable, highly efficient and robust energy storage technologies.[10, 11] First of all, most renewable energy sources, such as solar and wind, are inter- mittent in nature.[12, 10, 13] While our consumption of energy is everlasting, renewable energy production varies throughout the year, and the day. Even in Los Angeles, California, the average number of sunny days is only 284. Sec- ondly, the demand for energy fluctuates significantly throughout the day.[14, 9] 1 Thus it is critical to deploy sustainable, robust and large-scale energy storage if renewable energy sources were to become viable and dominant. 1.2 Challenges in Developing Rechargeable Batter- ies for Large-Scale Energy Storage A number of technologies have been proposed as large-scale energy storage solutions, owing to their respective advantages.[15, 10, 16] Among them, Li- and Na-ion batteries offer substantial advantages since they can be located nearly anywhere, have high round-trip Coulombic efficiencies, and are easily scaled to meet a wide variety of applications.[17, 18, 15] Figure 1.1: Illustration of a lithium ion battery. Lithium ions are inserted into the cathode during discharge and de-inserted during charge. Figure constructed based on a similar figure by Goodenough et. al.[19] Li-ion batteries store energy in the form of chemical potential and reversible charge/discharge is realized through mobile Li + ions that move between the cathode and anode through the internal pathway while electrons move in the same direction through the external circuit. For example, a schematic illustration 2 of a lithium-ion battery is given in Figure 1.1. During discharge, because of the chemical potential difference between the cathode and anode, electrons conduct work through the external circuit while flowing from the anode to the cathode. A net flow of Li + ions with same amount of charge also move from the anode and cathode through the electrolyte to maintain charge balance. During charge, the flow of electrons and Li + ions are both reversed under the drive of external bias. Ever since its first commercialization by SONY in 1990, lithium-ion batteries have essentially underpinned the great success of portable electronics in the last two decades.[17, 19] Its relatively high energy and power density as well as a long cycle life have made the use of cell phones and laptops practical. However, mounting challenges are still facing the development of lithium-ion batteries for them to be a viable solution in energy storage and competitive in the current market. First of all, the cost of lithium-ion batteries still needs to be lowered sig- nificantly for grid- and automotive-scale energy storage. The United States Advanced Battery Consortium (USABC) is targeting battery cost of $125/kWh for PEV battery by 2022, while most battery packs currently manufactured for electric vehicles, such as those by Nissan, General Motors, BYD, and Tesla, cost around $200/kWh, 50% higher than the target.[20] And in stationary energy storage systems, where cost-competitiveness is a far more relevant parame- ter than that in portable electronics and electric vehicles, an average cost of goods sold (COGS) of 100 $100/kWh is targeted for any technology to be truly disruptive.[16, 21, 22] Depending on the system, the electrode materials can account for 90% of the weight of a battery cell and as much as 50% of the cost.[23, 24] Thus there is a strong incentive to address the high cost of electrode 3 materials, by driving down the cost of raw materials and materials processing, or by developing more cost-effective materials. Our efforts to address this challenge by exploring iron silicate based materials as cathodes are shown in the following chapters 1, 2, and 3.[25, 26] It should also be noted that over concerns of the cost Li-based raw materi- als and its geopolitical complications, there is a growing interest in developing sodium-ion batteries.[27] In sodium-ion batteries, Na + ions replace Li + ions as the internal charge carrier. Promises of sodium-ion batteries are driven by the fact that cost of raw materials for Na is significantly lower than that of Li, and the two share similar chemical characteristics. Although a wide variety of studies have shown that it is hardly the case that materials and technologies developed for lithium-ion batteries can be translated into sodium-ion batteries, advance- ments in this field have proven to be quite promising.[28, 29] For example, bat- tery packs based on aλ-MnO 2 cathode, an aqueous electrolyte and carbon based composite anode have been commercialized for stationary energy storage.[22] Secondly, the charging rate and energy density are still limiting wider appli- cation of lithium-ion batteries in electric vehicles. Because of the strong inter- action between the mobile ions and the host framework, the charge transfer at high currents usually become diffusion-limited.[19] While the loss of capacity due to diffusion limitations is reversible, i.e., the capacity can be recovered when the cell is charged at lower rates again, the changes resulting from potential side reactions and decomposition are not reversible.[19, 30] Batteries in most plug- in hybrid electric vehicles take a few hours to be fully charged. And although electric vehicle (EV) manufacturers have recently developed fast-charging tech- niques, such as the "superchargers" in the Tesla S and BMW i3 models, to reach a modest rate of 30 mins to charge to 80%, EV owners are usually advised to use 4 such techniques sparingly as they can result in faster aging of the battery pack. Abundant battery systems with ultra fast charging and discharging capability have been reported literature, their translation from small-scale academic lab to large-scale manufacturing workshops consistently prove to be challenging.[31] Given that many of the current limitations are intrinsic to the materials used as positive and negative electrodes there has been a surge of interest in finding new phases which can undergo reversible electrochemical reactions. 1.3 Research on Intercalation Electrode Materials The research on intercalation compounds for potential use as electrode mate- rials for rechargeable batteries dates back to 1970s or even earlier.[32, 33] Lay- ered transition metal chalcogenides are among the earliest phases studied.[32, 34] Although LiTiS 2 , the lightest phase in these chalcogenides, was demonstrated with high specific density and long cycle life, it enjoyed little commercial success due to a relatively low operating voltage.[35, 36] Large commercial success of lithium-ion batteries was not realized until Goodenough et. al demonstrated reversible Li + -intercalation in LiCoO 2 at a much higher voltage, and SONY com- mercialized the system combining it with a carbon anode.[37, 38, 39] In exploring if a material has potential for commercialization and success, there are a few apparent requirements. In addition to being cost-effective and environmentally friendly, it should contain a reox couple, usually based on a transition metal ion, at a suitable voltage. Considering practical viability and the potential window of available electrolytes, the voltage should be close to 4 V against Li/Li + for lithium-ion batteries. Moreover, it should be a good ionic 5 conductor and preferably a good electronic conductor, to allow for facile inser- tion and de-insertion of guest ions and minimization of requirement for inactive materials. It is these requirements that govern the design principles for interca- lation electrode materials. Figure 1.2: Structures of representative commercialized cathode materials for lithium-ion batteries, LiCoO 2 (a), LiFePO 4 (b), and Li 4 Ti 5 O 12 (c). CoO 6 octahedra are shown light purple, FeO 6 in brown, and TiO 6 in light blue. Li and O are shown in yellow and orange spheres, respectively. Figure constructed using the VESTA program based on reported structural information.[40, 41, 42, 2] Structures of several representative commercialized electrode materials for lithium-ion batteries are shown in Figure 1.2. It is apparent that a common feature for these structures is the planes and channels of Li + ions formed by the interstitial spaces in a usually closely pack lattice. Although in the case of LiCoO 2 only half of the theoretical capacity can be achieved since it under- goes irreversible phase change if more than half of the Li is removed.[37, 36] The combination of facile ionic diffusion pathway and robust host framework underpins their excellent performance in Li + -ion intercalation. Since the suc- cess of LiCoO 2 , a wide variety of oxides with redox centers of Mn, Fe, Co, Ni and a mix of two or more of them have been studied for intercalation electrodes.[43, 44, 45, 46, 47, 48, 49] Tremendous progresses have been made in the optimization of these systems through engineering of the size, interface, and 6 microstructure of these phases. Improved reversibility, cycling life, and energy density are usually achieved through shortened the diffusion distance for the mobile ions, enhanced contact between the electrode and the electrolyt, as well as a more robust microstructure.[50, 51, 52, 53, 54, 55, 56, 57, 58] The discovery by Goodenough and coworkers[59, 60] that electrically insu- lating phases like LiFePO 4 could be carbon-coated to increase their electrical con- ductivity and so they could be used as alkali-ion intercalation hosts heralded a new age of exploration into polyanionic compounds as alternatives to oxides and chalcognides. Polyanionic materials are considered to be safer than oxides because they do not release oxygen on decomposition, which can exacerbate thermal runaway when cells fail.[61, 62] Another huge advantage of polyan- ionic materials is the extraordinary variety in composition that can be achieved through substitutions. Through mechanisms such as the inductive effects, these substitutions can have profound effects on the electrochemical performance, which provides an excellent toolbox for designing materials for next-generation electrochemical energy storage.[63] Since that initial work, several other compositions based on the tetrahedral phosphate group have been explored including pyrophosphate- (Li 2 MP 2 O 7 ) [64, 65, 66] and fluorophosphate-based phases (LiMPO 4 F). Compounds containing the trigonal planar borate groups (LiMBO 3 ) [67, 68, 69, 70, 71] have also garnered a significant amount of attention due to the reduced formula weight which pushes the theoretical capacity to over 200 mA·h·g −1 . Even more elabo- rate compositions with multiple polyanionic groups like the carbono-phosphates, Li 3 MCO 3 PO 4 ,[72] have been shown to exhibit highly reversible electrochemical behavior. More recently, Tarascon and coworkers in France and the group of Nazar in Canada have characterized a large number of bisulfate- (Li 2 M(SO 4 ) 2 ) 7 and fluorosulfate- (LiMSO 4 F) based compositions.[73, 74, 75, 76, 77, 78, 79, 80] These materials proved particularly interesting as an example of how the induc- tive effect of strongly electronegative groups like SO 4 can drive the redox poten- tial of many of these phases up to 3.6 V and, in the case of the triplite polymorph, all the way to 3.9 V. [75] In these SO 4 -based systems, the goal was primarily to enhance the oper- ating potential while minimizing the molecular weight as a way to increase the energy and power density of the cells. However, as mentioned earlier, in order to develop energy storage technologies that can easily scale to meet global energy demands, a paradigm shift towards a focus on materials based on Earth- abundant reagents free from geopolitical complications are needed.[21] And as polyanionic cathodes gain in utility, it is crucial to identify the relevant struc- tural distortions that occur across multiple polyanionic compounds in order to elucidate guidelines for the design of next-generation electrodes. Following this logic, a significant amount of work has focused on composi- tions like the polymorphs of Li 2 FeSiO 4 which have been shown to demonstrate reversible (de)insertion of Li-ions in the range of 2.8-3.1V.[81, 82, 83, 84, 85, 86, 87] While at first glance these phases offer an ideal way to reduce the cost - Fe is the most abundant transition metal present in the Earth’s crust and Si can be found on any beach around the world - there are significant problems asso- ciated with these phases.[88] Aside from being unstable in the presence of air and moisture, the higher potential phases are also known to convert to the lower potential polymorph after extended cycling.[85] Building on these reports, we explored the electrochemical energy storage in the pyroxene LiFeSi 2 O 6 , NaFeSi 2 O 6 , and iron based clay silicates. With a wide range of characterization tools, we were able to link its electrochemical 8 performance to their respective structural features. More specifically, we demon- strated the effects of structural rigidity and polyhedral distortions in these sili- cates on their electrochemical performances. We later explored these effects in anti-NASICON type Fe 2 (MoO 4 ) 3 and defect-perovskite type ReO 3 . By comparing their elctrochemical peformance against Li + - and Na + -intercalation, we found out that how the host lattice accommodate the guests through polyhedral rota- tional distortions have profound effects on their electrochemical performance as well. These results indicate that polyhedral rotational distortions may play a far-reaching role in electrode materials rechargeable batteries and the insights gained through these studies can provide guidelines for the understanding and optimization of existing and new electrode materials. 9 Chapter 2 Low Temperature Preparation and Electrochemical Properties of Pyroxene LiFeSi 2 O 6 LiFeSi 2 O 6 , whose structure is shown in Figure 2.1, belongs to the fam- ily of pyroxene minerals which have the general formula AMSi 2 O 6 .[89, 90] Early studies on LiFeSi 2 O 6 studied the stuctural, magnetic, and spectroscopic properties.[91, 92, 93, 94, 95, 96, 97, 98, 99, 100] More recent work has been driven by the discovery of multiferroic coupling in the pyroxene family of mate- rials. [89, 101, 102, 103, 90, 104] Yet considering the presence of large channels throughout the structure, it is surprising that the electrochemical properties of LiFeSi 2 O 6 have never been investigated. Here, we use a combination of electrochemical, structural, and computa- tional tools to demonstrate that, despite its extremely poor electrical conductiv- ity, LiFeSi 2 O 6 is capable of reversibly (de)inserting as large as 78 mA·h·g −1 (60%) of the theoretical capacity of 125 mA·h·g −1 around a potential of 2.0 V. Exten- sive details on the structure of LiFeSi 2 O 6 have been previously reported, [105, 89, 102] but to our knowledge this is the first time those structural parameters have been correlated with the electrochemical performance. In the course of this investigation, we have also developed a novel route to prepare LiFeSi 2 O 6 using 10 a hydrothermally-assisted route which can be used to enable the rapid prepara- tion of large quantities of high purity material. This is of particular interest since LiFeSi 2 O 6 is traditionally prepared using ceramic approaches which can take sev- eral days of extended heating at temperatures above 900 ◦ C. 2.1 Experimental Details Synthetic Methods In a typical preparation, 0.02 mol Li 2 CO 3 , 0.01 mol SiO 2 (fumed silica, 0.2- 0.3μm average particle size), and 0.004 mol FeCl 3 ·6H 2 O were combined in 15 mL of deionized water. The resultant red slurry was transferred into a 23 mL Teflon-lined stainless steel autoclave, sealed and maintained at 200 ◦ C for 12-16 hours. After cooling to room temperature on the bench top, the resulting Figure 2.1: Illustration of the crystal structure of LiFeSi 2 O 6 . The topology of the compounds contains edge-sharing chains of FeO 6 octahedra (green) which share oxygen with chains of corner-sharing tetrahedral SiO 4 groups (grey). Li ions (yellow spheres) sit in the void space which exists between the chains. 11 product, consisting of a dark green and nearly amorphous powder, was collected by vacuum filtration and washed with distilled water. This gel was subsequently heated to 900 ◦ C for 6 hours to yield a light green powder which consisted pri- marily of LiFeSi 2 O 6 with some minor reflections corresponding to polymorphs of SiO 2 . Washing the sintered powder in a 1 M solution of NaOH in a Teflon- lined autoclave for 1 hour at 200 ◦ C was found to preferentially dissolve the SiO 2 impurities with no deleterious effect to the crystallinity or purity of the title phase. Adjusting the time and temperature of the post-hydrothermal annealing was found to provide a facile way to control the particle size of the material. Structural Characterization Laboratory X-ray diffraction patterns were collected on a Bruker D8 diffrac- tometer with a Co Kα source (λ 1 = 1.78897 Å, λ 2 = 1.79285 Å), equipped with a Lynxeye detector. The collection process was kept the same for different sam- ples, as using a 0.6 mm slit, with a step size 0.02 degrees and a scan rate of one second per step in the 2θ range from 10 to 90 degrees. High resolution synchrotron powder diffraction data were collected using beamline 11-BM at the Advanced Photon Source (APS), Argonne National Lab- oratory using an average wavelength of 0.413724 Å. Discrete detectors covering an angular range from -6 to 16 ◦ 2θ were scanned over a 34 ◦ 2θ range, with data points collected every 0.001 ◦ 2θ and a scan speed of 0.01 ◦ s −1 . The result- ing diffraction patterns were refined using the Rietveld method [106] with the FullProf program, starting with the atomic coordinates determined by Clark et al. [91] Neutron scattering data were collected at 298 K on polycrystalline powders loaded in a vanadium can using the time-of-flight HIPD instrument at the Lujan 12 Center, Los Alamos Neutron Science Center, Los Alamos National Laboratory. The resulting data were refined using the General Structure Analysis System (GSAS).[107] 57 Fe Mössbauer spectra were collected in the transmission geometry with a source of 57 Co in Rhodium metal. During the measurements, both the source and the absorber were kept at ambient temperature (294 K). The spectrome- ter was operated with a triangular velocity waveform. The velocity scale was calibrated with the magnetically split sextet spectrum of a high-purity α-Fe foil as the reference absorber. The absorbers were made by mixing 40 mg of the compound with 80 mg of boron nitride. The spectra of the measured samples were fitted to an appropriate combination of Lorentzian profiles representing quadrupole doublets by least-squares methods. In this way, spectral parameters such as quadrupole splitting (QS), isomer shift (IS), and relative resonance areas of the different spectral components were determined. Isomer shifts are given relative toα-Fe metal. Electrochemical Measurements The electrochemical performance of LiFeSi 2 O 6 was characterized using Swagelok-type cells. The positive electrodes were prepared by ball-milling (Spex R 8000) the title compound with 30 wt% of Super P carbon for 20 min- utes under Ar atmosphere. Cells were assembled in an argon-filled glove box with a Li-metal disk as the negative electrode. Two Whatman R GF/D borosili- cate glass fiber sheets were used as the separator. LiPF 6 was initially tried as the electrolyte, but there appeared to be reactions with the active material, as might be expected for a fluorinated solution in contact with a Si-containing phase. As a 13 result, 1 M LiClO 4 in ethylene carbonate and dimethylcarbonate (1:1 w/w) was used for all measurements reported here. Computational Methodology All density functional theory (DFT) calculations were performed using the VASP package,[108, 109] in which the valence electronic states are expanded within a plane-wave basis. The valence–core interaction is described using the projector augmented wave (PAW) approach,[110] and cores of [He] for oxygen, [Ne] for silicon, [Ar] for iron and [He] for lithium were used. The Perdew-Burke- Ernzerhof generalized gradient approximation (GGA) exchange-correlation func- tional revised for solids was used,[111] with the addition of a rotationally invari- ant +U term as proposed by Dudarev.[112] Standard density functional calcu- lations routinely fail to accurately describe the electronic structure of strongly correlated materials,[113, 114, 115] due to the self-interaction error inherent to such functionals. This effect is most acute for strongly localized orbitals such as transition metald or rare earthf states. The inclusion of a +U correction allows this deficiency to be addressed without an excessive increase in the computa- tional cost of the calculations. In the case of LiFeSi 2 O 6 , the unknown delithiated phase FeSi 2 O 6 , and the unknown lithium intercalated phase Li 2 FeSi 2 O 6 , we use +U values in the range of 2.0− 10.0 eV on the Fe d states, which brackets the range of U-vales that have been previously used in the literature for ferric silicates. A cutoff of 750 eV was used for the all calculations, with the Brillouin zone sampled using a Γ- centered 2× 4× 6 Monkhorst Pack grid for bulk LiFeSi 2 O 6 . All calculations were spin polarized and were deemed to have converged when the forces on all the 14 atoms were less than 0.01 eV Å −1 . Structure and charge density visualization and analysis were performed using VESTA.[2] Figure 2.2: Results of the Rietveld refinements of as-prepared LiFeSi 2 O 6 against synchrotron XRD pattern obtained on the 11-BM beamline at Argonne National Laboratory (upper panel) and the neutron diffraction pattern obtained on the HIPD at Los Alamos National Laboratory (lower panel). For simplicity, we only show the results from the refinement against the 153 degree backscattering bank, but all six banks were refined simultaneously. 2.2 Results and Discussion The hydrothermal preparation discussed in the experimental section was used to prepare the LiFeSi 2 O 6 characterized in the following. A precise control of the pH in the starting solution was critical to obtaining a completely phase-pure product. Any deviation of the starting pH from 8 produced a brown rather than homogeneous dark green gel, which upon firing gave LiFeO 2 as an impurity. The as-prepared gel was analyzed using a Simultaneous Thermal Analyzer (STA-449 Jupiter, Netzsch) and it was found that heating above 600 ◦ C was required to 15 crystallize the LiFeSi 2 O 6 phase. Temperatures of 900 ◦ C produced the most crys- talline product in the form of a light green powder while samples fired at 700 ◦ C took on a pale yellow coloration probably due to the reduced particle size. After washing with NaOH, as described in the experimental section, both products were completely phase pure as determined from laboratory and synchrotron X- ray diffraction. Figure 2.2 shows the results of Rietveld refinement of the structure against the synchrotron X-ray diffraction (upper panel) and neutron diffraction (lower panel) data on the material heated at 900 ◦ C. Detailed structural parameters including cell parameters, atomic positions, and anisotropic atomic displace- ment parameters are given in Table 2.1 and Table 2.2. Neither diffraction pattern shows any evidence of impurity peaks, confirming the extremely high purity of the samples which can be obtained with this new preparation. The Mössbauer spectrum of the as-prepared LiFeSi 2 O 6 is shown in Figure 2.3. The spectrum is well-fit using a single doublet having the same Mössbauer parameters as previously reported by Dollase and Baum[116, 93] (IS = 0.38 mm·s −1 and QS = 0.30 mm·s −1 ) and is indicative of a single iron (III) site with octahedral symmetry. Table 2.1: Resulting unit cell parameters from the Rietveld refinement of the sin- tered LiFeSi 2 O 6 against synchrotron X-ray powder diffraction data (left column, R Bragg = 2.57%) and neutron diffraction data (right column, theR wp was deter- mined to be 1.08% for all banks refining simultaneously). Both diffraction data are obtained at room temperature. Space group C 2/c C 2/c a (Å) 9.668420(12) 9.65781(12) b (Å) 8.666983(11) 8.65770(12) c (Å) 5.293612(7) 5.28781(7) β (deg) 110.14001(11) 110.1306(8) V (Å 3 ) 416.4602(9) 415.127(10) 16 Table 2.2: Resulting atomic positions and atomic displacement parameters (isotropic parameters for 11-BM and anisotropic parameters for HIPD) from the Rietveld refinement of the sintered LiFeSi 2 O 6 against synchrotron X-ray powder diffraction data (left six columns of the top panel) and neutron diffraction data (right three columns of the top panel and the bottom panel). Both sets of data were collected at room temperature. Synchrotron X-ray diffraction Neutron diffraction atom Wyckoffx y z U iso (Å 2 ) x y z Li 4e 0.00000 0.2594(7) 0.25000 0.0132(20) 0.00000 0.26425(30) 0.25000 Fe 4e 0.00000 0.89826(7) 0.25000 0.0051(2) 0.00000 0.89902(7) 0.25000 Si 8f 0.29590(8) 0.08966(10) 0.26467(14) 0.0053(2) 0.29632(9) 0.08916(11) 0.26515(20) O1 8f 0.11483(14) 0.08513(17) 0.1502(3) 0.0012(4) 0.11596(7) 0.08348(7) 0.14730(13) O2 8f 0.36722(16) 0.25829(16) 0.3254(3) 0.0052(5) 0.36465(8) 0.25799(7) 0.32216(14) O3 8f 0.35483(17) -0.00118(16) 0.0553(3) 0.0037(5) 0.35465(8) -0.00062(9) 0.05458(15) atom Wyckoff U 11 U 22 U 33 U 12 U 23 U 13 Li 4e 0.0418(26) 0.0063(15) 0.0131(16) 0.0 0.0 0.0118(18) Fe 4e 0.01517(32) 0.01274(35) 0.01034(29) 0.0 0.0 0.00604(25) Si 8f 0.0067(5) 0.0155(4) 0.0114(4) -0.0032(4) -0.0008(4) 0.00274(32) O1 8f 0.00796(31) 0.0131(4) 0.01022(35) -0.00489(32) -0.00302(32) 0.00048(24) O2 8f 0.0193(4) 0.0114(4) 0.0203(4) -0.00148(28) -0.0035(4) 0.01124(33) O3 8f 0.0179(4) 0.0227(4) 0.00888(32 0.00499(31) -0.00080(28) 0.00793(33) 17 This is an important observation that reduces the likelihood that the uncon- ventional synthetic method employed produced an off-stoichiometric product. Figure 2.4 and Figure 2.5 illustrate the electrochemical performance of LiFeSi 2 O 6 . The assembled Swagelok cells were cycled between 1.5 V and 3.8 V since cut-off voltages above or below these limits appeared to result in irre- versible reactions which ultimately degraded the performance of the cells. Cycling the phases prepared at 900 ◦ C at a rate of C/50 (1 Li per formula unit over 50 hrs) yielded reversible capacities around 48 mA·h·g −1 , corresponding to roughly 40% of the theoretical capacity (125 mA·g·h −1 ). The derivative curve, shown in Figure 2.5, clearly shows a redox process with a potential centered around 2 V versus Li + /Li 0 , but a slight degradation can be seen in the curve over several cycles. This may be an an indication that the process is slightly irre- versible or that a proper passivating layer has not been formed on the surface of the particles and there is reactivity with the electrolyte. Regardless, the inset of Figure 2.5 shows that more than 97% of the capacity could be retained after 30 cycles which indicates an effectively reversible process. The fairly low capacity compared to the theoretical maximum appears to be a reflection that the process Figure 2.3: Room temperature Mössbauer spectrum of the as prepared LiFeSi 2 O 6 . The experimental data is well-fit with one sharp Fe 3+ doublet. 18 Figure 2.4: Voltage-capacity curves of sintered LiFeSi2O6 cycled at C/50 (upper panel) and C/200 (lower panel). A recyclable capacity around 78mA·h· gâ ´ LŠ1, corresponding to more than 60% of its theoretical value, is reached when LiFeSi2O6 is cycled at a rate of C/200. of Li-ion (de)insertion is significantly hindered by either the electrical conduc- tivity of the particles or the process of ionic diffusion. Several types of carbon and variations on ball-milling time were tested to optimize the performance, but little improvement was found between preparations. To explore the possibility that the rate performance was hindered by the kinet- ics of diffusion, cells were also cycled at much slower rates of C/200. This performance, illustrated in the lower panel of Figure 2.4, showed a substan- tial increase in the reversible capacity corresponding to more than 60% of the theoretical value. To further investigate the influence of kinetics in the elec- troechemical cycling, smaller particles of the title phase were prepared by heat- ing the hydrothermally prepared precursor at 700 ◦ C, which produced signifi- cantly broader diffraction peaks with much lower intensity as shown in Figure2.6 reflecting the reduced long-range periodicity in the particles. 19 Figure 2.5: The derivative of the voltage-composition curve of sintered LiFeSi 2 O 6 cycled at C/200. The derivative shows redox peaks around 2 V. The inset shows the capacity retention curve for the batteries cycled at C/50. After the initial three cycles, the capacity is retained more than 97% with 25 cycles. Given the smaller particle size, Ketjen Black carbon was used with these par- ticles to increase the interfacial contact between the particles. Electrochemi- cal impedance measurements were employed to confirm no significant change in the electrical conductivity between the two samples prepared at higher and lower temperatures. While the capacity at each cycling rate was not found to improve significantly, the polarization between the charge and discharge curves was reduced by several hundred mV as illustrated in Figure2.7. To understand the reason for the poor kinetics of diffusion in the pyroxene structure, the resulting anisotropic atomic displacement parameters from the diffraction experiments were examined more closely. Figure 2.8 shows the struc- ture of LiFeSi 2 O 6 with the atoms illustrated by the atomic displacement parame- ters from the Rietveld refinement results against HIPD data (Table 2.1 and 2.2). Firstly, the cigar-like shape of the Li-ion displacements around their ideal position reinforces the idea that Li + is mobile in the structure. However, the major axis of 20 20 40 60 80 2 theta (deg) ( λ = 1.788970 Å) Intensity (arb. unit) LFS900 LFS700 Figure 2.6: Voltage-composition curve of LiFeSi2O6 calcinated at 700 ◦ C (LFS700) and 900 ◦ C (LFS900) cycling at C/50 and C/200. A much smaller polar- ization is obtained for LFS700 compared to LFS900, which should be attributed to the decrease in kinetic hindering with smaller particle size. the Li ellipsoid indicates that Li ions prefer to move through the void space along the diagonal axes enclosed by the corner-sharing FeO 6 octahedra and SiO 4 tetra- hedra [Figure 2.8 (b)], instead of the channels alongc-axis enclosed by the FeO 6 and SiO 4 polyhedra chains [Figure 2.8 (a)]. It is important to note, however, that even if the removal of Li from the parent structure were actually possible without collapsing the structure it would require oxidation of Fe 3+ to Fe 4+ , a reaction which is unlikely to occur within the window of stability for standard electrolytes. Perhaps more significantly, the O3 oxygen which sits at the corner of the shared SiO 4 tetrahedra along the length of the Si 2 O 6 chains takes on a flat- tened saucer shape. Such a flat atomic displacement ellipsoid is indicative of the fact that the Si 2 O 6 chains are extremely rigid with very little flex to accommodate changes in the unit cell volume. Considering that these chains are oriented along the channels through which Li ions are expected to move, this likely explains the 21 Figure 2.7: Voltage-capacity curves of sintered LiFeSi2O6 cycled at C/50 (upper panel) and C/200 (lower panel). A recyclable capacity around 78mA·h· gâ ´ LŠ1, corresponding to more than 60% of its theoretical value, is reached when LiFeSi2O6 is cycled at a rate of C/200. strong kinetic hindrance of Li-ion diffusion in LiFeSi 2 O 6 . However, as shown by our preliminary electrochemial tests, slow rates and significantly reduced parti- cle size are capable of increasing the reversible capacity of the title phase. Thus, it is possible that with further optimization, such as a systematic study on the electrode-electrolyte interface and preparation of thin-film based electrodes, the full capacity of LiFeSi 2 O 6 becomes accessible. The final outstanding question regarding the electrochemical performance of the title phase, is why the Fe 3+ to Fe 2+ redox couple occurs at such a low voltage. It is well-known that the inductive effect of polyanionic groups can be exploited to increase the open circuit voltage of positive electrodes. However, since the electronegativity of Si and Fe are very close, the inductive effect in silicate-based phases is not expected to increase the potential very much. Still, 22 even the lowest potential obtained from the polymorphs of Li 2 FeSiO 4 is found to occur at 2.8 V. [85] To gain insight into this reduced potential, density functional theory calcula- tions were performed. To calculate the open circuit voltage, we use the standard thermodynamic approach as developed by Ceder and co-workers.[117] We have considered two plausible Li positions: (i) Li sitting between the Fe and Si chains [(0.7,0.25,0.5), Wyckcoff 4c], and (ii) Li sitting in the center of the channel [(0,0.5,0), Wyckcoff 4b]. The calculated voltage lie between 0.7 V and 1.5 V using a range ofU values from 2 eV-10 eV, but but there is clearly a large discrep- ancy between the experimentally observed voltages, and the voltages obtained Figure 2.8: Illustration of the crystal structure of LiFeSi 2 O 6 . Atoms are rep- resented as ellipsoids based on the atomic displacement parameters with 95% probability from Rietveld refinement against Bank 1 of the HIPD data (Table 2.1 and SI Table 1). 23 for reasonable values of U for Fe-based electrodes. This discrepancy could reflect the highly localized nature of the Fe 3d states in LiFeSi 2 O 6 , and high resolu- tion photoelectron spectroscopy are planned to examine this. Alternatively, and perhaps more likely, while the two Li sites considered seem plausible based on Bond Valence Sum analysis of the relaxed structure, they might not reflect the absolute minimum structure for Li 2 FeSi 2 O 6 . In-situ X-ray diffraction experiments were attempted using a laboratory diffractometer; however, the changes in the diffraction patters were too subtle to accurately determine the structure of the lithiated phase. A global search for the Li 2 FeSi 2 O 6 is currently underway, but is beyond the scope of the present study. 2.3 Conclusion In summary, we have presented a rapid and facile route to the prepara- tion of the Earth-abundant pyroxene mineral LiFeSi 2 O 6 with very high purity. Using a comprehensive suite of characterization tools we have demonstrated that LiFeSi 2 O 6 undergoes a reversible electrochemical reaction centered around 2 V. Results from computational characterization further confirm the electrochemical activity of LiFeSi 2 O 6 and show that the reversible redox processes of LiFeSi 2 O 6 are strongly hindered by the rigid framework of the host. An investigation of the atomic displacement parameters based on the neutron diffraction data relates the kinetic hindering to the structural rigidity of LiFeSi 2 O 6 . We have demon- strated that controlling the particle size of the as-prepared LiFeSi 2 O 6 significantly improved its electrochemical performance. Our comprehensive study shows the significance of close examination of structural properties in understanding the 24 electrochemical performance of these materials and provides constructive infor- mation on the electrochemical performance of the clinopyroxene minerals. It should be noted, however, that the 2 V potential of LiFeSi 2 O 6 is typically too high for use as an anode yet still too low for use as a cathode. So while the discovery of reversible electrochemical performance in a rock-forming mineral is funda- mentally interesting, the feasibility of commercial application of the title phase is unlikely because of the small theoretical capacity and low operating potential. 25 Chapter 3 Preparation and Magnetic Properties of NaFeSi 2 O 6 : Nanowires vs Bulk Samples Single-phase materials that simultaneously possess two or more of the pri- mary ferroic properties (ferroelectricity, ferromagnetism, and ferroelasticity), known as multiferroics, have attracted a significant amount of attention due to the coupling that exists between these order parameters. [118] Recently, Jodlauk et al. reported the observation of magnetoelectric coupling in the naturally-occurring minerals NaFeSi 2 O 6 . [89] These measurements were con- ducted on mineral samples, which present some challenges due to compositional inhomogeneities. The single crystals studied by Jodlauk et al. were found to have an average composition determined from electron microprobe analysis of Na 1.04 Fe 0.83 Ca 0.04 Mn 0.02 Al 0.01 Ti 0.08 Si 2 O 6 . While this off-stoichiometry makes an accurate characterization of the magnetic properties difficult, the presence of Mn impurities is guaranteed to alter the nature of the bulk magnetic properties. The use of mineral samples, instead of phase-pure synthetic samples, is jus- tified by the extreme challenge of preparing phase-pure NaFeSi 2 O 6 . Traditional synthetic routes involve quenching a molten glass with the correct ratio of Na 2 O- Fe 2 O 3 -SiO 2 to yield a polycrystalline powder that is then annealed to improve 26 crystallinity and site-order.[102] This process can produce nominally pure sam- ples; however, the phase diagram reported by Bowen and coworkers [119] shows that NaFeSi 2 O 6 melts incongruently above 990 ◦ C which results in the creation of small impurities of Fe 2 O 3 if not cooled quickly enough. More recently, Red- hammer et al. reported a method for growing small phase-pure single crystals from fluxes of molten molybdate and vanadate salts.[103] Low temperature neu- tron diffraction measurements on these single crystals confirmed the existence of complex non-collinear magnetic order in the homogeneous samples and this is proposed by the authors to be the origin of the magnetoelectric coupling.[103] Here we present a route for the preparation of phase-pure polycrystalline NaFeSi 2 O 6 at low temperatures and pressures. Through a simple digestion of small particles of silica in a strongly basic media in the presence of a soluble iron salt, we were able to prepare hundreds of milligrams of NaFeSi 2 O 6 at a time. Electron microscopy showed the powders prepared in this way adopt a wire- like morphology, having widths on the order of a few nanometers. Considering the interest in the multiferroic nature of NaFeSi 2 O 6 , we set out to compare the magnetic properties of these nanowires with the bulk properties. Herein, we show high resolution synchrotron X-ray diffraction, scanning and transmission electron microscopy, Mössbauer spectroscopy, as well as temperature- and field- dependent magnetization and dielectric measurements which demonstrate the phase which adopts a nanowire morphology has distinctly different magnetic properties. We attribute the difference in properties to the increased number of iron sites at the surface which alters the nature of the long-range magnetic order at the transition temperature. 27 Figure 3.1: Crystal structure of C 2/c NaFeSi 2 O 6 . The SiO 4 tetrahedra are grey and the FeO 6 octahedra are green. The sodium and oxygen atoms are shown as yellow and orange spheres respectively. (a) shows the close packing Si 2 O 6 layered structure with sodium and iron ions located in the octahedral sites. (b) shows the zig-zag edge-sharing FeO 6 chains connected by the zig-zag corner- sharing SiO 4 chains through the oxygen ions on their edges. 3.1 Experimental Details 3.1.1 Synthetic Methods In a typical hydrothermal synthesis, 0.02 mol NaOH (≥ 98%, anhydrous pel- lets, Sigma-Aldrich), 0.01 mol SiO 2 (fumed silica, 0.2-0.3 μm average particle size, Sigma-Aldrich), and 0.004 mol Fe(NO 3 ) 3 ·9H 2 O (≥ 98%, Sigma-Aldrich) were dissolved in 15 mL deionized water by the order as indicated. Then the mixture was transferred into a Teflon-lined stainless steel autoclave, sealed and maintained at temperatures ranging from 160 ◦ C to 220 ◦ C for 14 hours. After cooling in air to room temperature, the resulting product was collected and cleaned by filtration with distilled water and ethanol. 28 In order to prepare bulk samples for the comparison of magnetic properties, the as-prepared powder was pressed into a pellet and heated to 950 ◦ C at a rate of 5 ◦ C/min, held for 40 hours, and allowed to cool to room temperature naturally. The bottom of the crucible was lined with powder of the same composition as the pellet to avoid any reactions with alumina. 3.1.2 Physical Characterization High resolution synchrotron powder diffraction data were collected using beamline 11-BM at the Advanced Photon Source (APS), Argonne National Lab- oratory using an average wavelength of 0.413961 Å. Discrete detectors cover- ing an angular range from -6 to 16 ◦ 2θ are scanned over a 34 ◦ 2θ range, with data points collected every 0.001 ◦ 2θ and scan speed of 0.01 ◦ s −1 . The result- ing diffraction patterns were refined using the Rietveld method with the FullProf program,[106, 120] starting with the atomic coordinates determined by Clark et al.[91] 57 Fe Mössbauer spectra were measured in the transmission geometry with a source of 57 Co in rhodium metal. During the measurements, both the source and the absorber were kept at ambient temperature (294 K). The spectrometer was operated with a triangular velocity waveform. The velocity scale was calibrated with the magnetically split sextet spectrum of a high-purity α-Fe foil as the ref- erence absorber. The absorbers were made by mixing 20 mg of the compound with 80 mg of boron nitride. The spectra of the measured samples were fitted to appropriate combination of Lorentzian profiles representing quadrupole doublets by least-squares methods. In this way, spectral parameters such as, quadrupole splitting (Δ), isomer shift (δ) and relative resonance areas of the different spec- tral components were determined. Isomer shifts are given relative toα-Fe metal. 29 Simultaneous Thermogravimetric Analysis (TGA) and Differential Scanning Calorimetry (DSC) measurements were carried in the temperature range of 30 ◦ C-1100 ◦ C in air using a Simultaneous Thermal Analyser (STA-449 Jupiter, Netzsch). Scanning Electron Microscopy (SEM) was performed on a JEOL JSM-7001 microscope (JEOL Ltd). Transmission Electron Microscopy (TEM) images were obtained with a JEOL JEM-2100F (JEOL Ltd) operating at 200 kV. Specimens for TEM were prepared by dispersing the as-prepared sample in ethanol and sonicated for 20 minutes, and then deposited on a 200 mesh Cu grid coated with a Lacey carbon film (Ted Pella, Inc). Temperature- and field-dependent magnetic susceptibility and specific heat measurements were collected using a 14 T Quantum Design Dynacool Physical Properties Measurement System. Samples for specific heat measurements were made by mixing the title phase with equal parts by mass of silver powder and pressing into a bar in order to improve thermal coupling to the stage. The specific heat of Ag was measured separately and subtracted. Capacitance measurements were collected at various frequencies as a func- tion of field and temperature using an Andeen-Hagerling AH2700A capacitance bridge. A 9 T Quantum Design Dynacool Physical Properties Measurement Sys- tem was used to provide the variable temperature and field environment for capacitance measurements. Prior to measurement, the as-prepared powder was densified using spark plasma sintering with a final density that was 89% of the theoretical maximum. The resulting pellet had a diameter of 10.63 mm and a height of 1.191 mm and was subsequently coated with conducting epoxy to elec- trically connect the sample to the capacitance probe. Shielded coaxial cables 30 linked the clamped pellet to the capacitance bridge to restrict sample movement during the measurement. 3.2 Results and Discussion NaFeSi 2 O 6 , commonly known as the mineral aegirine, crystallizes in theC2/c space group. The structure, as illustrated in Figure 4.1, contains chains of the edge-sharing FeO 6 octahedra which zig-zag along the c-axis. The FeO 6 chains are connected by the SiO 4 tetrahedra, which form zig-zag corner-sharing chains along thec-axis [Figure 3.1 (b)], through sharing the oxygen ions on their edges. Alternatively, the structure can be considered to be formed from close packed Si 2 O 6 chains, with Na and Fe ions occupying the octahedral sites [Figure 3.1(a)]. The interstices, enclosed by the sodium and oxygen atoms, are connected along thec-axis and form tunnels with dimensions of 3 Å×4.4 Å. NaFeSi 2 O 6 is typically prepared using traditional ceramic reactions or fluxes to grow single crystals, which inevitably require temperatures above 1000 ◦ C, and sometimes high pressure environments. [93, 121, 96, 122, 103] A relatively low temperature process was reported by Redhammer et al., in which gels were prepared starting from Na 2 CO 3 , Fe(NO 3 ) 3 ·9H 2 O, and TEOS with oxygen fugac- ity controlled by the pressure vessels, and subsequently heated at temperatures around 700 ◦ C.[123] A similar preparation, reported earlier by Rusakov et al., involved the evaporation of a mixture of NaNO 3 , Fe(NO 3 ) 3 , and silicoethyl ether in ethanol and nitric acid to obtain a gel, which was subsequently annealed at 500 ◦ C and reacted with NaOH solution in Teflon capsules to attain the crystal- lization of NaFeSi 2 O 6 .[121] 31 Previous studies of NaFeSi 2 O 6 have primarily focused on its structural,[124, 125, 91, 123, 96, 122, 126, 127, 128] spectroscopic,[97, 129] and mineralogical properties.[130, 131, 132] However, the discovery of multiferroic coupling in the pyroxene NaFeSi 2 O 6 by Jodlauketal. led to intensive studies on the magnetic and dielectric properties of NaFeSi 2 O 6 and similar structures.[89, 133, 103, 134] Figure 3.2: Synchrotron XRD patterns of products from hydrothermal prepara- tion at temperatures ranging from 160 ◦ C to 220 ◦ C. A minimum temperature of 180 ◦ C is required to obtain crystalline NaFeSi 2 O 6 . The diffraction patterns for powder prepared using the hydrothermal reac- tions described in the experimental section are illustrated in Figure 3.2. Reac- tions were carried out from 160 ◦ C to 220 ◦ C with samples collected at 160 ◦ C and 170 ◦ C yielding a red powder while all others were a pale yellow-green. The low temperature samples were found to contain a small amount of hematite mixed with an amorphous phase indicating a minimum temperature of 180 ◦ C is required to crystallize NaFeSi 2 O 6 . All peaks in the patterns for samples prepared from 180 ◦ C up were found to be fully indexed to the structure reported by Clark 32 et al.[91] Increasing the temperature to 200 ◦ C was found to increase the sharp- ness of the diffraction peaks which is likely associated with an increase in particle size. No significant difference in crystallinity or morphology was found between samples prepared at 200 ◦ C and 220 ◦ C. Hereinafter, we refer to the hydrother- mal samples prepared at 220 ◦ C as the as-prepared sample when compared to sintered samples. It is important to note the difference in color between the two temperature regimes. Usually Fe(III) compounds are light yellow to an orangish red. As a point of comparison to the hydrothermal samples, we attempted to prepare NaFeSi 2 O 6 using a traditional ceramic reaction between oxides and carbonates. Regardless of reaction times or temperatures a small percentage of hematite impurity was always found and resulted in a completely red product as illus- trated in Supporting Info (SI) Figure S 1. Previous studies focusing on the solid solution of NaFeSi 2 O 6 with other minerals, and the synthetic NaFeSi 2 O 6 stud- ied in previous works have reported red coloration. [135, 119] The hydrother- mally prepared powder is yellowish green which darkens to an olive green on sintering, which is in good agreement with the coloration of single crystals of NaFeSi 2 O 6 .[136, 96] The morphology of the as-prepared sample was studied by Scanning and Transmission Microscopy, which shows the as-prepared powder consists of NaFeSi 2 O 6 nanowires that cluster together(Figure 3.3). The spacing of the lattice fringes indicates that the nanowires predominantly grow along the c-axis. The wires appear to grow from a common origin and are not simply agglomerated together, so post-synthetic sonication was unable to separate the bundles. As this is the first nanoscale preparation of NaFeSi 2 O 6 , a comparison with the bulk physical properties is of great interest. Given the difficulties associated 33 Figure 3.3: (a) SEM image of the as-prepared NaFeSi 2 O 6 , which is shown to con- sist of bundles of nanowires. (c) TEM image of the as-prepared sample reveals its crystallinity. with the conventional preparation as previously mentioned, bulk NaFeSi 2 O 6 was prepared by sintering the as-prepared sample by pressing the powder into a pellet and firing at 950 ◦ C for 40 hours. The diffraction pattern of the sintered sample, illustrated in Figure 3.4, clearly shows significantly sharper peaks which reflects the substantially increased particle size on firing at elevated temperatures. 34 Figure 3.4: Rietveld refinement of as-prepared and sintered NaFeSi 2 O 6 against synchrotron X-ray diffraction patterns obtained on the 11-BM beamline at Argonne National Laboratory. χ 2 values of 2.42 and 2.28 for the refinements on the as-prepared sample and the sintered sample were obtained respectively. The nanoscale morphology presents some difficulty in refinement of the diffraction pattern. Nevertheless, high-quality fits were achieved for both the as- prepared nanowires prepared at 220 ◦ C and the sintered sample against the high- resolution diffraction data from 11-BM synchrotron facility. However, attempts to refine the atomic displacement parameters (ADP) of the oxygen atoms were tried without any success, which is probably due to the low scattering form fac- tor of oxygen and statistical error. Thus for the consistency and legitimacy of the refinements, the ADPs of all oxygen atoms are locked at a reasonable value (B iso = 0.3 Å 2 ). The refinement results are shown in Figure 3.4. Detailed struc- tural parameters including cell parameters, atomic positions, and atomic dis- placement parameters are given in SI Table S 1 and Table S 2. The Mössbauer spectra of the two powders obtained at room temperature are given in Figure 3.5. The first observation is that for both powders all the iron is 35 Figure 3.5: Comparison of the room temperature Mössbauer patterns for the as- prepared and sintered phases. Best fits to the experimental data are obtained with two doublets for the as-prepared sample powder and a single doublet for the sintered one. in the Fe 3+ state and no trace of Fe 2+ is detected. The best fit to the experimental data is obtained using a single doublet (red) for the sintered powder and two doublets (red and blue) for the as-prepared one. This implies that there is only a single type of iron environment with little to no disorder or defects in the sintered powder. In contrast, the need for two doublets as well as the broadening of the spectrum of the as prepared powder indicates that there is a distribution of envi- ronments in the as-prepared NaFeSi 2 O 6 . Possible origins for the additional sites could be the presence of an impurity phase (13%, Fe 3+ ) or a sufficient number of defects that a second signal can be seen. Given that no impurity is seen in the synchrotron diffraction data and that the sintered phase, which is made from the as-prepared material, shows no indication of a secondary phases, it is unlikely that an impurity phase is present. The fitted Mössbauer parameters are similar to those reported by Schmidbauer and Kunzmann[129] (note these authors worked at 84 K which is why they found higher isomer shift). For the sintered sample, it 36 is extremely well crystallized NaFeSi 2 O 6 . No trace of Fe 2+ is detected. We thus attribute the additional Fe 3+ environment in the as-prepared NaFeSi 2 O 6 to the fact that its surface is hydroxide terminated. Such a finding is fairly reasonable considering the preparation is carried out in aqueous media and is supported by the presence of hydroxyls stretches in the Raman patterns as well as by a small mass loss in the thermal gravimetric analysis (see SI Figures S 2 and S 3). The hydroxide terminated surface would also explain that the quadrupole splitting is in close agreement with what is found in hydroxysulfate-based phases.[137] Figure 3.6: Comparison of the temperature-dependent magnetic susceptibilty for the as-prepared and sintered phases. Figures 3.6 (a) shows the temperature-dependent magnetic susceptibility of the as-prepared nanowires in comparison with the phase-pure sintered powders collected from 2 to 300K in a field of 10 kOe. At first glance, the sintered phase appears to order substantially higher than the as-prepared phase. However, a 37 closer examination of dχ/dT for the two measurements reveals a single max- imum in each sample at 6 K and 7 K for the as-prepared and sintered phases respectively so there is very little difference in the ordering temperature between the two samples (SI Figure S 6). Regardless, there is a clear change in the nature of the ordering since the sintered phase demonstrates the characteristic cusp of an antiferromagnet whereas the as-prepared powder shows a slight increase in the susceptibility on ordering. We note that our samples show no evidence for 8 K transition reported in the natural samples of Jodlauk et al.[89] To gain more insight into the nature of these differences, the high temper- ature region (100 K to 400 K) of both measurements was fitted to the Curie- Weiss equation, C/(T− Θ CW ) with Curie-Weiss temperatures (effective param- agnetic moments) of−33 K (5.47μ B ) and−39 K (5.87μ B ) being found for the as-prepared and sintered phases respectively. A four percent weight loss found in the TGA analysis of the as-prepared sample was taken into account when normal- izing the effective moments for the nanoscale phase. The theoretical spin-only effective moment for a d 5 , S=5/2, ion in an octahedral crystal field is 5.92μ B which can be obtained using the relationship μ S = 2 q S(S + 1). The high tem- perature data was also fitted using two modifications to the Curie-Weiss equa- tion which incorporated a temperature independent correction,χ 0 , and a term to account for non-interacting spins,C/T. A diamagnetic contribution is commonly applied to account for the background from the plastic sample holder while the non-interacting term has previously been used to model substituted materials where two regions of spins exist. [138] In this instance, the non-interacting term is intended to model disordered spins on the surface of the nanowires. While the addition of the temperature independent term, which typically yielded χ 0 around -4×10 −4 , resulted in a slight improvement of the fit, the non-interacting 38 term did not appear to improve the quality of the analysis. Thus, despite the increased surface area and the presence of hydroxyl groups in the as-prepared materials, it appears that all of the spins in the materials still experience the same mean field interaction strength. It should also be noted that our results for the sintered sample agree well with the bulk sample studied by Redhammer et al., in which temperature-dependent magnetic susceptibility and neutron diffrac- tion of a NaFeSi 2 O 6 sample diluted with small amount of SiO 2 and Fe 2 O 3 were reported.[103] Normalizing the inverse susceptiblity data to the fitted Curie-Weiss param- eters allows a more direct comparison of the temperature-dependent behavior of the two materials. Deviations from ideal Curie-Weiss behavior in such a plot reflect the development of short-range correlations which are antiferromagnetic in the upward direction and ferromagnetic in the negative direction. [139] As seen in Figure 3.6 (b), both materials are well described by the Curie-Weiss law until very near theoretical Curie-Weiss ordering temperature (T/|Θ CW |=1 in this plot), at which point antiferromagnetic correlations begin to develop as expected from the negative sign of the Curie-Weiss theta. This comparison indicates that, despite the jump in the susceptibility on ordering, the overall sign of any short range correlations in the as-prepared phase are antiferromagnetic and there is no evidence for ferro- or ferri-magnetic behavior as a function of temperature. Figure 3.7 (a) shows the temperature-dependent specific heat of the title phases. The horizontal axis was extrapolated to 0 K by fitting data in the region of 1.8 K to 5 K with a three-parameter polynomial. Substantial differences can be seen between the two materials, with a sharp lambda-like anomaly that is typically associated with the development of long-range order clearly seen in the sintered material yet absent from the as-prepared powder. Subtracting the 39 Figure 3.7: Comparison of the specific heat capacity (a) and change in entropy associated with the magnetic ordering transition (b) for the as-prepared and sintered phases. lattice contribution and integrating C p /T yields a change in entropy due to mag- netic ordering of 14.2 J·mol −1 ·K −1 for the sintered sample and 10.2 J·mol −1 ·K −1 for the as-prepared sample. The former value is very close to the value of 14.9 J·mol −1 ·K −1 predicted by the Boltzmann equation (ΔS = R ln(2S + 1), S = 5/2).[140] Lower spin entropy in the as-prepared nanowire sample could be an overestimation of the lattice contribution or the hydroxyls disrupting the spins on the surface of the nanowires. Both values are notably different from those reported in previous literature, where a value of 11.2 J·mol −1 ·K −1 for a synthetic sample and 13.9 J·mol −1 ·K −1 for a natural sample,[102] as well as 10.5 J·mol −1 ·K −1 for a synthetic sample were reported.[141] While the differences in entropy values may simply due to inaccuracies asso- ciated with the estimation of the lattice contribution to the specific heat and other artifacts, a comparison between the overall shapes of the curves should reveal more information about the nature of the materials. C mag /T, where C mag 40 is the magnetic contribution of the specific heat and T is temperature, is plotted against T in the onset of Figure 3.7 (a). The broad feature of the as-prepared sam- ple indicates less order in the nanowires, which has been shown in Figure 3.6. The curve for the sintered sample is highlighted by an ordering peak sitting on a broad hump, which looks very close to that of the synthetic sample reported by Baker et al. and differs significantly from their natural sample.[102] Accord- ing to Fisher, the heat capacity of an antiferromagnet can be determined from the magnetic susceptibility by using the equation C (T) = A(∂(χT )/∂T).[142] A comparison between the Fisher heat capacity (SI Figure S 7) and specific heat curve of the sintered material reveals that the broad hump is probably responsi- ble for the spin entropy, while the leading feature could be associated with the 8 K transition that we do not detect in temperature dependent susceptibility mea- surements (Figure 3.6 and SI Figure S 6). The origin of these transitions has been discussed in previous literature,[102, 89] and an accurate probe would require more advanced experiments such as low-temperature neutron scattering, which are beyond the scope of this report. Figure 3.8 (a) shows the constant temperature magnetization loops at 2 K. The as-prepared material shows a nearly linear response to the application of strong magnetic field, as would be expected for an antiferromagnetically ordered material; however, there is clearly a slight curvature. In contrast, the sin- tered material shows two distinct field-induced features in the magnetization at±50 kOe and±90 kOe as emphasized in Figure 3.8 (b). These field-induced features may be related to the reorientation of the ferrolectric polarization from theb-axis to thec-axis that was reported in the work of Jodlauk. [89] In contrast to what was found there, our measurements show that the field-induced change in the dielectric properties does not occur until significantly higher fields than 41 Figure 3.8: Comparison of the constant temperature magnetization loops for the as-prepared and sintered NaFeSi 2 O 6 at 2 K. The magnetization of the as-prepared NaFeSi 2 O 6 nanowires shows a nearly linear response to the applied magnetic field, while two distinct field-induced features are observed in the sintered sam- ple. previously observed in the site-mixed geological samples. Constant temperature loops of sintered NaFeSi 2 O 6 at temperatures from 3 K to 8 K can be found in SI Figure S 8(a). The derivative curves, shown in Figure S 8(b), indicate only one of these field-induced features persists at temperatures from 3 K to 7 K and both of them are absent at temperatures above 7 K. In order to examine whether this field-induced feature was really corre- lated with changes in the electric polarization, magnetocapacitance measure- ments were collected on a densified pellet of the sintered phase. Since capac- itance measurements require a highly dense pellet, it was not possible to mea- sure the capacitance of the as-prepared nanowires for comparison. Field- and temperature-dependent capacitance measurements of NaFeSi 2 O 6 are illustrated in the SI Figures S 4 and S 5 respectively. We observe a jump in the dielectric 42 constant at 7 K, compared to a value of 6 K paraelectric to ferroelectric tran- sition reported by Jodlauk etal.[89] While the ferroelectric transition is field- dependent according to the report of Jodlauketal., our results show a constant transition temperature (T FE ) in varying fields (SI Figure S 5). Above T FE , the capacitance is field-independent [SI Figure S 4(a)], however, significant enhance- ment of the capacitance is observed in what Jodlauk etal. identified as a ferro- electric phase when 60 kOe≤H≤80 kOe. However, the use of powder samples in our measurements restricts our results to average bulk properties. Fully resolv- ing the orientation-dependent magnetic and dielectric properties would require the preparation of high-purity single crystals. 3.3 Conclusion We present a single-step route for the preparation of phase-pure nanowires of NaFeSi 2 O 6 at low temperatures and pressures. Considering the interest in the multiferroic nature of NaFeSi 2 O 6 , we set out to compare the magnetic properties of these nanowires with sintered bulk samples. Herein, we show high resolution synchrotron X-ray diffraction, scanning and transmission electron microscopy, Mössbauer spectroscopy, as well as temperature- and field-dependent magneti- zation measurements which demonstrate the phase that adopts a nanowire mor- phology has distinctly different magnetic properties from the bulk sample. We attribute the difference in properties to the increased number of iron sites at the surface which alters the nature of the long-range magnetic order at the transition temperature. 43 Chapter 4 Leveraging Complex Cation Distributions in Iron-based Clays for Electrochemical Energy Storage In this work, we sought to identify a more open silicate framework that would be capable of faster rate performance while remaining stable within the voltage window of an electrochemical cell. We first identified the phyllosilicate, or sheet- silicate, family of minerals as having the greatest potential because of the well- segregated layers of transition metal and silicate groups, which is reminiscent of the ordered rock-salt phases like LiCoO 2 (Figure 4.1). In the following, we describe a highly tunable set of reaction conditions where the morphology, structure, and atomic distributions can be altered at the syn- thetic level. We further demonstrate that this class of materials is capable of electrochemically cycling on the Fe 3+ to Fe 2+ redox couple through a pseudoca- pacitive, rather than intercalative, mechanism. 44 4.1 Experimental Details 4.1.1 Synthetic Methods In a typical preparation, 0.020 mol KOH, 0.004 mol SiO 2 (fumed silica, 0.2- 0.3μm average particle size), 0.002 mol Fe(NO 3 ) 3 ·9H 2 O (K/Fe = 10/1), and 1.00 g polyvinylpyrrolidone (PVP) (average molecular weight = 10,000 g/mol) were combined in 15 mL of deionized water. The resultant red slurry was trans- ferred into a 23 mL Teflon-lined stainless steel autoclave, sealed and maintained at 220 ◦ C for 16-18 hours. Once the autoclave was cooled, a dark green pow- der was collected by vacuum filtration, washed with distilled water, and dried at 110 ◦ C for 30 mins. When varying amount of one precursor was chosen to study Figure 4.1: (a) Illustration of a unit cell of 2:1 phyllosilicates. The topology of the compounds contains corner-sharing tetrahedral SiO 4 layers (grey, b) sand- wiching edge-sharing layers of FeO 6 octahedra. K + ions (yellow spheres) sit in the interlayers that exist between the repeating sandwiched layers. There are two metal sites in the octahedral layer, a “honeycomb site” and a “stuffed site”, shown in green and brown, respectively, in (c) and (d). Black spheres represent hydrogen atoms. 45 its effect on the preparation, loading of all other precursors were kept the same as stated above. 4.1.2 Physical Characterization Elemental analysis was performed in-house using a Thermo Scientific iCAP 7000 inductively coupled plasma-optical emission spectroscometer(ICP-OES). Roughly 2.0 mg of sample powder was digested with 10% HNO 3 and 2% HF in a polyethylene volumetric flask, with each sample being measured three times, and the presented values representing the average over all measurements. Laboratory X-ray diffraction patterns were collected on a Bruker D8 diffrac- tometer with a Co Kα source (λ 1 = 1.78897 Å, λ 2 = 1.79285 Å), equipped with a Lynxeye detector. High resolution synchrotron powder diffraction data were collected using beamline 11-BM at the Advanced Photon Source (APS), Argonne National Laboratory using an average wavelength of 0.413682 Å. Discrete detec- tors covering an angular range from -6 to 16 ◦ 2θ were scanned over a 34 ◦ 2θ range, with data points collected every 0.001 ◦ 2θ at a scan speed of 0.01 ◦ s −1 . Neutron pair distribution function (PDF) data were collected on the NOMAD beamline at the Spallation Neutron Source at Oak Ridge National Laboratory at 60 Hz setting. [143] Le Bail fits and Rietveld refinement of the structure were carried out using the TOPAS software suite (version 6) using the fundamental parameter approach.[144] Anisotropic broadening of the Bragg reflections due to strain was modeled using the Stephen’s model.[145] A numerical method was used to 46 characterize the honeycomb stacking faults in the synthesized silicates with a 30- layer supercell. The local structure was refined using either least-square refine- ments or simulated annealing of the initial Rietveld model using total scattering neutron data. Structures and charge density were visualized using VESTA.[2] 57 Fe Mössbauer spectra were collected in the transmission geometry with a source of 57 Co in Rhodium metal. During the measurements, both the source and the absorber were kept at ambient temperature (294 K). The spectrome- ter was operated with a triangular velocity waveform. The velocity scale was calibrated with the magnetically split sextet spectrum of a high-purity α-Fe foil as the reference absorber. The absorbers were made by mixing 40 mg of the compound with 80 mg of boron nitride. The spectra of the measured samples were fitted to an appropriate combination of Lorentzian profiles representing quadrupole doublets by least-squares methods. In this way, spectral parameters such as quadrupole splitting (QS), isomer shift (IS), and relative resonance areas of the different spectral components were determined. Isomer shifts are given relative toα-Fe metal. The electrochemical performance of the materials was characterized using Swagelok-type cells. Prior to cell-assembly, the as-prepared sample was first heated at 600 ◦ C for 2 hours in air in order to remove any organic residue from the PVP and ensure efficient electrical conductivity. The active materials were then ball-milled (Spex R 8000) with 30 wt% of Super P carbon for 10 min- utes under Ar atmosphere. Cells were assembled in an argon-filled glove box with a Li-metal disk as the negative electrode. Two Whatman R GF/D borosil- icate glass fiber sheets were used as the separator. 1 M lithium 4,5-dicyano-2- trifluoromethyl-imidazolide (LiTDI) in ethylene carbonate and dimethylcarbon- ate (1:1 v/v) was used as the electrolyte. 2 wt% of vinylene carbonate (VC) was 47 also added to facilitate the formation of a more robust passivating layer on the cathode. 4.2 Results and Discussion Phyllosilicate clays are hugely abundant, making up roughly ten percent of the Earth’s crust. [88] Previous studies on the naturally occurring minerals have focused primarily on their structural, morphological or spectroscopic character- istics, [146, 147, 148, 149, 150, 151, 152] with their physical catalytic proper- ties drawing the most attention from the petrochemical industry for use as oil drilling mud and the catalytic cracking of hydrocarbons.[153, 154, 155, 156, 157, 158, 159, 160] The electrochemical activity of transition metal containg clays has been explored in the past; however, much of the work was focused on electron transfer between intercalated species and structural ions in miner- alogical samples.[161, 162, 163, 164, 165, 166, 167, 168] There are also scat- tered reports on the electrochemistry of synthetic phyllosilicates,[169, 170] but the lack of crystallinity prevented the development of more detailed correlations between structural features like stacking faults and cation disorder.[168] The structure of a prototypical 2:1 trioctahedral phyllosilicate is illustrated in Figure 4.1. Each sheet of the clay is composed of a transition metal layer sand- wiched by two layers of corner-sharing SiO 4 tetrahedra that form the complex polyanionic group shown in Figure 4.1 (b). Within the octahedral layer there are two metal sites: one forming a hexagonal network, which we will refer to as the “honeycomb site” (green in Figure 4.1), while the other sits in the interstitial of the hexagonal net that we will refer to as the “stuffed site" (brown in Figure 4.1). The honeycomb site is distinguished locally by hydroxyl ions that coordinate in 48 Figure 4.2: (a) SEM image of as-prepared sample and (b) results of a LeBail fit of as-prepared sample using a C2/m space group against synchrotron XRD pattern obtained on the 11-BM beamline at Argonne National Laboratory. cis positions whereas the stuffed site has hydroxyls located in a trans configu- ration. K + ions located between the layers bind the layers tightly together and compensate the net negative charge on each sheet. The precise composition of the as-prepared material was determined using a combination of ICP-OES and CHNS elemental analysis, which resulted in a nom- inal stoichiometry of KFe 2.75 Si 3.25 O 10 (OH) 2 . This composition suggests that the as-prepared samples are a synthetic form of iron muscovite exhibiting a nominal cation distribution of KFe 2 (Si 3.25 Fe 0.75 )O 10 (OH) 2 , where most of the iron is found 49 within the transition metal layer, but a small fraction also substitutes for some of the silicon in the tetrahedral sites.[171] Figure 4.2(a) shows a typical SEM image of the as-prepared samples where it can be seen that most of the particles adopt a plate-like shape with a thickness around 100 nm and a diameter of 1μm. This kind of morphology presents unique challenges in that the resulting X-ray diffraction pattern exhibits a significant degree of anisotropic broadening of the peaks. This is best seen in the insets of Figure 4.2(b), which demonstrate that the (00l) peaks, corresponding to the direction along which the sheets are stacked, are significantly sharper than the intralayer reflections. Given the complexities associated with the morphology and its impact on the scattering data, we set out to develop structural models that could accurately fit the resulting patterns. For the purpose of our modeling, we defined a single repeat slab as consisting of one transition metal layer, the two adjacent silicon tetrahedra layers, and one layer of K-ions. The various ways these slabs can be stacked with respect to each other results in a number of space groups,[172, 173, 174, 175, 176] but the three most common are P3 1 12, C2/c and C2/m. Le Bail fits using the synchrotron X-ray diffraction data,[177] described in detail in the Supporting Information, showed that the pattern is best described using the monoclinic unit cell C2/m, with cell parameters of 5.342 Å, 9.227 Å, 10.270 Å, and 100.864 ◦ . P3 1 12 could not describe the small monoclinic distortion and C2/c results in extra reflections. To determine the exact distribution of Fe across each site, small-box least- square refinements were performed on total scattering neutron data, the results of which can be found in Figure 4.3 and Supporting Info Table S 1. From these refinements, it is clear that Fe predominantly occupies the honeycomb site while 50 3 6 9 12 r (Å) -3 -2 -1 0 1 2 3 G(r) (Å -2 ) Observed Calculated Difference Figure 4.3: Least-square refinement of the local structure of KFe 2.75 Si 3.25 O 10 (OH) 2 using neutron PDF data obtained at the NOMAD beamline at the Oak Ridge National Laboratory (R wp = 14.65%). Locally, the octahedral Fe and vacancy sites are found to be well-ordered though about 5% Fe are refined to be randomly distributed on the vacancy site. the stuffed position remains mostly vacant except for a roughly 5% random occu- pancy by Fe. This is fully consistent with Mössbauer data that will be discussed in greater detail later(Table 4.2), and confirms that the materials we obtain are analogues of iron muscovite. After confirming that Fe was well-ordered within the ab-plane, we next sought to describe the coherence between these sheets. Our first approach was to estimate the degree of stacking faults by introducing Fe and vacancy anti-site disordering during Rietveld refinement. This anti-site mixing does not truly indicate site mixing but instead reflects the degree of stacking disorder between the honeycomb layers along the stacking direction.[178, 179] The best fit using this method, shown in Figure 4.4(a), indicated a stacking fault con- centration of approximately 20%, suggesting that the coherence of the stacking sequence is interrupted every 5 layers on average. Structural parameters from 51 Figure 4.4: (a) Rietveld refinement of average structure of KFe 2.75 Si 3.25 O 10 (OH) 2 using synchrotron XRD data. The saw tooth shaped diffraction peak (or Warren peak) at about 5.2 ◦ indicates the existence of diffuse scattering signal induced by stacking faults in c* direction. The Fe occupancy on the octahedral Fe 2+ and vacancy sites were allowed to vary during the refinement to account for the honeycomb stacking disorder along c-axis direction. The degree of site mix- ing reflects the degree of honeycomb stacking disorder in c-axis direction. (b) Refinement of KFe 2.75 Si 3.25 O 10 (OH) 2 based on synchrotron data using the numer- ical 30-layer supercell with lateral translation vector model. The inserts show the fits to the major 020m or 100p (m indicates miller indices in monoclinic con- figuration and p indicate miller indices in primary trigonal configuration) diffuse scattering peak and other related reflections. this refinement are listed in Supporting Info Table S 2. While reasonable fits could be achieved using the method above, obvious intensity mismatches were 52 still observable for many reflections, especially for the strong diffuse scattering (or Warren) peak at about 5.2 ◦ .[180] It was therefore clear that a larger supercell along the stacking direction, which could account for the aperiodic stacking of the sheets, was needed to effectively describe the structure. Therefore, a super- cell with a c-axis translation periodicity of 30 layers (∼30 nm) was constructed within TOPAS. The three lateral translation vectors were initially randomly dis- tributed among these 30 layers, and it should be noted that the in-plane compo- nents of these vectors were allowed to vary from ideal values in order to account for the monoclinic distortion. As seen in Figure 4.4(b), a dramatic improvement could be achieved and almost all diffuse scattering intensity successfully modeled in this way. This method yielded approximately a 25% stacking fault concentra- tion based on the counting scheme proposed by Liu et al,[179] which is slightly larger than what was obtained from the Rietveld method, but is expected to be a more accurate reflection of the degree of disorder in the compound. 4.2.1 Effect of reaction conditions Having developed a better understanding of the structure, the effects of vary- ing the synthetic conditions were then explored, beginning with the influence of KOH concentration. The laboratory diffraction patterns for different loadings of KOH are shown in Figure 4.5, where the amount of Fe(NO 3 ) 3 , silica, and PVP were held constant at the values discussed in the Experimental Section while the molar ratio of KOH to Fe was adjusted from 2.5:1 to 10:1. As more KOH was added to the reaction mixture, the diffracted intensity increased and was correlated with a sharpening of the peaks, indicating the crystallinity of the final product was significantly improved. This suggests that the reaction proceeds through a dissolution-precipitation process, with higher concentrations of KOH 53 20 40 60 80 2 θ ( λ = 1.78897 Å) Intensity (arb. units) 8 10 12 2 θ Intensity Increasing KOH Figure 4.5: Laboratory X-ray diffraction patterns on hydrothermally prepared Fe-phyllosilicates with different KOH concentrations. The bottom, black curve corresponds to sample prepared with a KOH to Fe ratio of 2.5:1, or a KOH con- centration of 0.33 M. The blue, yellow and green curves correspond to KOH to Fe ratios of 5:1, 7.5:1 and 10:1, respectively. dissolving the SiO 2 more efficiently and improving the quality of the resulting crystalites. [181, 182] This dissolution-precipitation mechanism was further supported by Fig- ure 4.6, which shows SEM images of samples prepared from various KOH concen- trations. For the smallest amount of KOH, particles resembling the morphology of fumed silica are clearly seen, supporting the notion that the solution is not basic enough to fully dissolve all of the precursors. As the KOH to Fe ratio reaches 5:1, heavily agglomerated plate-like crystallites can be found, but only after the con- centration of base exceeds 7.5:1 do well-defined particles begin to appear. Most likely, low concentration of KOH prevents the correct Fe:Si ratio from existing in solution and results in off-stoichiometric particles with poor crystallinity. 54 Figure 4.6: SEM images of phyllosilicate samples prepared with different KOH concentrations. (a), (b), (c), and (d) correspond to samples prepared with KOH to Fe ratio of 2.5:1, 5:1, 7.5:1, and 10:1, respectively. Table 4.1: Distribution of Fe environments of phyllosilicates prepared with increasing KOH concentration. %Oct and Tet are short for octahedral and tetra- hedral environments, respectively. The number ahead of Oct and Tet indicates the valence state of the element. And the number in the bracket indicates the type of environment. K:Fe Environment % IS QS LW (mol/mol) mm/s mm/s mm/s 2.5:1 3Oct(1) 88 0.346 0.66 0.44 3Oct(2) 12 0.431 1.28 0.39 5.0:1 2Oct(1) 8 1.100 0.97 0.49 3Oct(1) 39 0.430 0.77 0.30 3Tet(1) 54 0.271 0.43 0.46 7.5:1 3Oct(1) 87 0.332 0.58 0.45 3Tet(1) 13 0.165 0.36 0.35 10.0:1 2Oc(1) 4 1.150 2.39 0.46 2Oct(2) 5 1.050 0.78 0.30 3Oct(1) 76 0.348 0.56 0.30 3Tet(1) 15 0.172 0.36 0.33 55 Table 4.2: Distribution of Fe environments of phyllosilicates prepared with dif- ferent PVP additions. PVP (g) Fe 3+ Fe 3+ Fe 3+ Fe 2+ Oct,% Tet, % total, % Oct, % 0 72 22 94 5 0.1 81 19 100 0 0.4 75 15 90 10 0.7 79 16 95 5 1.0 82 14 96 4 Mössbauer spectroscopy was also used to examine how the local environment of Fe was affected by different concentrations of KOH. A representative Möss- bauer spectrum is shown in Figure 4.7, with the resulting fits to several samples prepared with different KOH concentrations given in Table 4.1. As mentioned earlier, the iron appears to be distributed between two different octahedral envi- ronments as well as on the tetrahedral sites. Indeed, we see some signature of Fe 3+ filling the tetrahedral site for KOH to Fe ratios exceeding 5:1 as listed in Table 4.1. Given that the (001) reflection (Figure 4.5) and the disk-like mor- phology (Figure 4.6) do not evolve until KOH to Fe ratio is greater than 5:1, the substitution of Fe 3+ onto the tetrahedral site appears to be correlated with an increase in coherence between the (00l) planes. It is interesting to note that despite using a purely trivalent Fe(NO 3 ) 3 ·9H 2 O precursor, samples prepared in the presence of PVP often contained Fe 2+ between 5-10%. PVP has frequently been used in hydrothermal reactions as a tem- plating agent to control the shape of particle growth, but is also known to have a slight reducing character.[183, 184] Indeed, NMR spectra of the solu- tion after hydrothermal treatment showed the presence of typical aromatic sp 2 carbon, which is probably due to the oxidation of PVP (SI Figure S 6). Vary- ing the PVP content was also found to produce a continuous change of color 56 Figure 4.7: Room temperature Mössbauer spectrum of the as prepared sample with K:Fe ratio of 10:1. in the resulting powder from brown, when no PVP was added, to dark green, when 1 g PVP was added, despite nearly identical powder XRD and SEM images for all of the products. (Supporting Info Figures S 7 and S 8) More interest- ing, the Mössbauer spectra of samples prepared with varying amounts of PVP show that increasing PVP content appears correlated with a decreasing concen- tration of Fe 3+ on the tetrahedral site, staying around 15 % when more than 0.4 g of PVP was used (Table 4.2). This change in the concentration of tetra- hedral iron likely explains the variation in colors of the samples as there may be some kind of charge transfer process between the octahedral and tetrahe- dral iron sites. Combining the results from elemental analysis and Mössbauer spectroscopy, the composition of the as-prepared KFeSiO-20 sample, which was used for all of the electrochemical characterization, was confirmed to be K(Fe 3+ 1.75 Fe 2+ 0.125 ) h (Fe 2+ 0.125 ) s (Si 3.25 Fe 3+ 0.75 ) t O 10 (OH) 2 , withh ands indicating the octahe- dral honeycomb and stuffed sites respectively while thet denotes the tetrahedral position. 57 0 0.4 0.8 1.2 x-in-Li x KFe 2.75 Si 3.25 O 11 1 2 3 4 Voltage-(V-vs-Li/Li + ) 1.5 2 2.5 3 3.5 4 Voltage-(V-vs-Li/Li + ) -dx/dV-(V -1 ) (a) (b) Figure 4.8: Electrochemical performance of as-prepared sample. The top panel shows the voltage-composition curve with a recyclable capacity equivalent to 0.4 Li per formula unit at a rate of C/20. The bottom panel shows the derivative curve with redox peaks lying around 2.5 V. 4.2.2 Electrochemistry Having firmly established the structure and composition of our synthetic mus- covite, we sought to characterize the electrochemical properties. Prior to test- ing, all samples were pre-treated by heating at 600 ◦ C for two hours in order to 58 0.1 mV/s 0.2 mV/s 0.5 mV/s -1 -0.5 0 0.5 1 log(r) (log(mV/s) -2 -1.5 -1 log(I) (log(mA)) 1.5 2 2.5 3 3.5 E we (V vs Li/Li + ) -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 I (mA) 1 mV/s 4 mV/s 10 mV/s (a) (b) Figure 4.9: (a) Cyclic Voltammetry of the cell at sweep rates between 0.1 mV· s −1 and 10 mV· s −1 . (b) The relationship between the peak current and sweep rate during continuous cyclic voltammetry experiments. remove any interlayer water and potential residual products from PVP decom- position that could interfere with the electrochemical performance.[185] Impor- tantly, no obvious change to the diffraction patterns was observed following this heat treatment(SI Figure S 9). Given that fluorinated electrolytes, such as LiPF 6 , are known to chemically attack silicate electrodes, [186, 25] LiClO 4 in ethylene 59 carbonate and dimethylcarbonate (1 M, 1:1 w/w) was used during the initial characterizations. Yet in these cells, the system showed very little reversibility, which is believed to be related to the difficulty in creating stable solid-electrolyte interfaces (SEIs) on the surface of silicates.[186] (Supporting Info Figure S 10) Niedzicki et al. have demonstrated imidazole-derived salts, like lithium 4,5- dicyano-2-(trifluoromethyl)imidazole (LiTDI), exhibit a high degree of stability over a fairly large voltage window.[187] Of the electrolytes tested, 1 M LiTDI in ethylene carbonate and dimethylcar- bonate (1:1 v/v) provided the best performance. Figure 4.8 shows the electro- chemical performance between 1.5 V and 4.0 V after ball-milling with 30 wt% carbon. Cycling below 1.5 V was found to result in an irreversible conversion reaction that decomposed the active material, and ultimately degraded the per- formance of the cells. A reversible capacity of 40% of the theoretical capac- ity (58 mA·g·h −1 ) could be obtained when cycling at a rate of C/20. Derivative curves of the voltage-composition profile show that the reversible process centers around 2.5 V against Li + /Li (Figure 4.8(b)). While this potential is lower than what is seen for Fe 3+ /Fe 2+ in other polyanionic electrodes, it lies just between the 2.8 V found in Li 2 FeSiO 4 and 2.0 V reported for LiFeSi 2 O 6 . [82, 25] Considering the plate-like morphology, it is not surprising that, despite the pronounced Faradaic peak, a significant degree of capacitive contribution to the overall reversible capacity is seen in Figure 4.8. The individual contributions from the diffusion limited process (intercalating) and a surface controlled pro- cess (capacitative) are best differentiated by examining the relationship between peak current and sweep rate in cyclic voltammetry (CV) analysis.[188, 189] Fig- ure 4.9(a) shows CV scans collected at sweep rates between 0.1 mV· s −1 and 10 mV· s −1 . The scans feature the same redox peaks centered around 2.5 V 60 against Li/Li + that were seen in the derivative curve of the galvanostatic cycling, with the peaks broadening with increasing sweep rates. Figure 4.9 (b) shows a plot between the log(peak current) against log(sweep rate), which shows a clear deviation from a linear fit between scan rates slower than 1 mV·s −1 and those faster than 1 mV·s −1 . Instead, two linear fits were used, which resulted in a slope of 0.83 (R 2 = 0.9936) for the slower region and a slope of 0.62 (R 2 = 0.9961) for the faster region. Both values are between 0.5, expected for solid-state diffusion limited process, and 1, expected for surface con- trolled electrochemical process. This confirms that intercalation based redox pro- cess takes place when the sample is cycled against Li, although a certain amount of surface based process also contributes to the overall reversible capacity. The change of slope from 0.83 to 0.62 is similar to what is seen in T-Nb 2 O 5 and other systems, and is attributed to sources such as a change of ohmic contribution due to active material resistance, or solid-electrolyte interphase resistance.[190, 188] The observation of surface-dominated electrochemistry intially comes as something of a surprise. The layered topology of the phyllosilicate structure is closely related to intercalation hosts like LiCoO 2 , with the principle distinction being that the inter-layer cation is potassium. It is well known in the geochem- istry literature that potassium-based clays are far more difficult to exchange com- pared to Na- or Ca-containing analogues.[191, 192] As we have seen through our structural analysis, potassium significantly increases the crystallinity and the coherence between the planes, despite the complex stacking sequence. In hindsight, it then comes as no surprise that the electrochemistry of the synthetic iron muscovite we have studied is primarily restricted to the surface of the particles. Removal of the K from the interlayer space would result in the delamination of the structure and require an oxidation of iron to the tetravalent 61 oxidation state. Given that this Fe 3+ /Fe 4+ redox couple often lays well outside the window of electrolyte stability,[193] electrochemical insertion of lithium is far more easily achieved. Intercalation most likely results in Li insertion into the empty stuffed position. This helps to rationalize the limited accessible capacity we observe since Li is unlikely to diffuse easily through the interlayer space that is fully occupied by potassium. 4.3 Conclusion In summary, we have presented a systematic study of the effect of prepara- tion conditions on the structure, composition, and morphology of synthetic mus- covites. Using a combination of spectroscopic and structural characterization tools, we were able to determine precisely the composition of a highly crystalline sample with diverse Fe site distribution. We have also prepared an electrode with this highly crystalline sample and found that a reversible capacity equivalent to 0.4 Li + per formula unit can be obtained. Although the reversible capacity is relatively low against other systems that have been commercialized in portable electronics, the extremely-low cost and future improvement in reversibility may make these Fe-phyllosilicates competitive in large-scale energy storage applica- tions. More importantly, this study demonstrates how preparation conditions can be used to control the structural and morphological properties of the synthetic phyllosilicates, which can facilitate the improvement of their performance in a range of applications, such as heterogeneous electrocatalysis. 62 Chapter 5 Influence of Rotational Distortions on Li + - and Na + -Intercalation in Anti-NASICON Fe 2 (MoO 4 ) 3 NASICON related structures, with general formula MM’(XO 4 ) 3 , were among the first polyanionic compounds studied as hosts for intercalation electrodes.[61] The structures feature large interstitial spaces formed by three-dimensionally interconnected polyhedra. Moreover, the versatility of these structures allow polyanion substitution to tune the position of the redox couples.[194, 195] Anti- NASICON Fe 2 (MoO 4 ) 3 is a potential candidate for both rechargeable lithium and sodium ion batteries, owing to the earth abundant nature of Fe and the high reversibility of Fe 2 (MoO 4 ) 3 for both Li + - and Na + -intercalation.[196, 197, 198, 199] It is known that Fe 2 (MoO 4 ) 3 crystallizes into a monoclinic (P2 1 /c) structure, which consists of FeO 6 octahedra and MoO 4 tetrahedra interconnected through corner sharing oxygen atoms (Figure 5.1). The basic motif of the structure can be described as a ’lantern unit,’ which consists of three MoO 4 tetrahedra connected to two FeO 6 octahedra (Figure 5.1b). The lantern units stack in an antiparallel fashion along the (2b + c) direction in the anti-NASICON structure, while in the NASICON structure they are stacked in a parallel fashion along the c-axis.[61] Although Li + and Na + (de)insertion in Fe 2 (MoO 4 ) 3 have been previously studied, the mechanism of alkali guest ion insertion and extraction for this phase 63 is still ambiguous and contradictory in the literature. For example, Nadiri et al. reported a single-phase structural change in monoclinic Fe 2 (MoO 4 ) 3 during Na + (de)insertion in the composition ranges of 0.3 6 x 6 1.0 and 1.10 6 x 6 1.60 for Na x Fe 2 (MoO 4 ) 3 .[197] On the other hand, Bruce et al. demonstrated a mostly two-phase process for Na + insertion in monoclinic Fe 2 (MoO 4 ) 3 with an expan- sion in the unit cell.[198] Recent studies by Yue et al. reported a two-phase structural change for Li + (de)insertion, while a single-phase process for Na + - intercalation in orthorhombic Fe 2 (MoO 4 ) 3 was observed.[200] Moreover, up to this point, the mechanism through which alkali guest ions intercalate into the electrode, or how the structural evolutions are correlated with electrochemical cycling, have not been reported. Herein, we use a combination of structural and electrochemical tools, along with symmetry-mode analysis, to investigate the relationship between elec- trochemical properties and structural evolutions in monoclinic anti-NASICON Fe 2 (MoO 4 ) 3 upon Li + and Na + (de)insertion. We reveal the key underlying dif- ference in Li + and Na + insertion into anti-NASICON Fe 2 (MoO 4 ) 3 and the effect of the intrinsic properties of Li + and Na + on the insertion process. Based on our detailed structural and symmetry-mode analysis, we propose a polyhedral rota- tional distortion mechanism for the intercalation process, which may find wider application in polyanionic electrode materials. 64 5.1 Experimental Details 5.1.1 Synthesis of Fe 2 (MoO 4 ) 3 Fe 2 (MoO 4 ) 3 was synthesized by a solid-state precipitation method.[201] (NH 4 ) 6 Mo 7 O 24 ·4H 2 O (1.0 mmol) and Fe(NO 3 ) 3 ·9H 2 O (4.65 mmol) were dis- solved separately in 100 mL and 50 mL of H 2 O, respectively. 0.4 mL of NH 3 ·H 2 O was added to the molybdate solution to establish basic conditions. The Fe(NO 3 ) 3 solution was then slowly added to the molybdate solution with stirring, resulting in a yellow precipitate. The reaction mixture was stirred overnight. The result- ing yellow precipitate was collected and washed with ethanol by sonicating for 30 min, and isolated by centrifugation (6000 rpm for 15 min). This washing procedure was repeated twice. The resulting Fe 2 (MoO 4 ) 3 powder was dried in a vacuum oven at 60 ◦ C for 5 h. The dried Fe 2 (MoO 4 ) 3 was then ground with a mortar and pestle and annealed at 400 ◦ C for 6 h in air. A yellowish-green powder was obtained. 5.1.2 Chemical insertion Li 2 Fe 2 (MoO 4 ) 3 was prepared by stirring 1.0 g Fe 2 (MoO 4 ) 3 in fivefold stoichio- metric excess of LiI (dissolved in dry acetonitrile) for 2 weeks. For Na + insertion, a sodium napthalenide solution was prepared by dissolving 3.0 g of Na metal and 1.5 g of naphthalene in 75 mL of dry THF. Na 2 Fe 2 (MoO 4 ) 3 was prepared by mixing 1.0 g Fe 2 (MoO 4 ) 3 in 25 mL of the resulting sodium napthalenide solution for a week. 65 5.1.3 Structural Characterization Laboratory powder X-ray diffraction (XRD) data was collected on a Bruker D8 diffractometer with a CoK α source (λ 1 = 1.78897 Å,λ 2 = 1.79285 Å), equipped with a LynxEye detector. The collection process was kept the same for different samples using a 0.6 mm slit with a step size 0.02 degrees and a total collection time of 1 h in the 2θ range from 10 to 60 degrees. A Swagelok-type electro- chemical cell with a beryllium disk as current collector for the working electrode was used for in-situ XRD as well as for diffraction data collection over chem- ically lithiated and sodiated samples.[202, 203] Laboratory XRD data were all collected in reflectance. High-resolution synchrotron powder diffraction data were collected at room temperature using the 11-BM beamline at the Advanced Photon Source (APS), Argonne National Laboratory using an average wavelength of 0.413965 Å. The data were collected in the 2θ range of -6.5 to 28.0 ◦ with a step size of 0.001 ◦ and 0.1 s spent on each step. Neutron diffraction and total scattering data were collected at room temperature on polycrystalline powders loaded in a vanadium can using the time-of-flight POWGEN instrument (BL-11A) and NOMAD instru- ment (BL-1B), respectively, at the Spallation Neutron Source, Oak Ridge National Laboratory. POWGEN data were collected with 16 h spent on bank 3 and 4 h on bank 4. Bank 3 uses frame No.1.5 at 60 Hz with a Q range of 1.1688 Å −1 to 15.1700 Å −1 and bank 4 uses frame No.1.75 at 60 Hz with a Q range of 0.6820 Å −1 to 5.6923 Å −1 . Data from all detectors with angles from 20 to 150 ◦ were combined into one histogram for each bank. NOMAD data were collected in the scattering angle range of 3 to 175 ◦ with 40 min spent on each sample. Data from all detectors were reduced into one PDF pattern after the background 66 contribution was subtracted. The resulting data were analyzed using the General Structure Analysis System (GSAS) and PDFgui.[204, 205] Symmetry-mode analysis was carried out using the AMPLIMODES program of the Bilbao Crystallographic Server.[206] The inputs of the program include structural parameters of a high-symmetry and a low-symmetry structure. Results from Rietveld refinements against synchrotron X-ray and neutron diffraction data on pristine and Li + -inserted Fe 2 (MoO 4 ) 3 were used as low-symmetry and high- symmetry phases, respectively. 5.1.4 X-ray photoelectron and Raman spectroscopy The high-resolution X-ray photoelectron spectroscopy (XPS) spectrum was acquired using a Kratos Axis Ultra X-ray photoelectron spectrometer with the analyzer lens in hybrid mode. A monochromatic aluminum anode with an oper- ating current of 6 mA and voltage of 10 kV was used with a step size of 0.1 eV, a pass energy of 20 eV, and a pressure range between 1x10 −8 torr to 3x10 −8 torr. The binding energy was referenced to the C 1s core level at 284 eV. Raman spectra were recorded in the 200-1200 cm −1 range using a Horiba Xplora Raman microscope (Horiba Scientific). Laser irradiation of 532 nm wave- length was employed as the excitation source and the power at the sample level was 50 mW. All spectra were recorded under ambient conditions, a quartz cell was used for the measurement of lithiated and sodiated samples. Intercalated samples were loaded into the quartz cell in an Ar-filled glove box. 67 5.1.5 Electrochemical measurements Electrochemical measurements were carried out using Swagelok-type cells.18 The positive electrodes were prepared by grinding Fe 2 (MoO 4 ) 3 and carbon Super P (CSP) for 30 min in an Ar-filled glove box. A slurry was prepared by adding the resulting mixture into polyvinylidenefluoride (PVDF) dissolved in N-methyl- 2-pyrrolidone (NMP). The slurry, comprised of 76.4 wt% Fe 2 (MoO 4 ) 3 , 8.5 wt% CSP, and 15.1 wt% PVDF, was cast onto an aluminum foil and dried in a vacuum oven at 110 ◦ C overnight. Electrodes were peeled off the Al foil and punched into 1.0 cm diameter circles. Li or Na metal disks were used as the negative electrodes. Whatman GF/D borosilicate glass fiber sheets were used as the separator and soaked in elec- trolyte solution of either 1 M LiPF6 in ethylene carbonate (EC), propylene car- bonate (PC), and dimethylcarbonate (DMC) (45:45:10 by weight) or 1 M NaClO 4 in EC and DMC (1:1 by weight). 5.1.6 Computational Density functional theory (DFT) calculations were performed using the plane- wave code VASP.[207, 208] Interactions between core and valence electrons were described with the PAW method, with cores of [Kr] for Mo, [Ar] for Fe, [He] for O, and [Ne] for Na.[209] All calculations used the PBEsol exchangeâ ˘ A¸ S- correlation functional.[210] To describe the strongly correlated Fe 3d electrons, we applied a Hubbard-type "+U" correction of Ud = 4.3 eV, using the rotation- ally invariant approach of Dudarev et al.[112] This value of Ud = 4.3 eV was chosen for consistency with previous calculations of orthorhombic Li + - and Na + - inserted Fe 2 (MoO 4 ) 3 by Yue et al.15 who, in turn, selected this value from earlier 68 calculations on Li x FePO 4 .[211] A planewave cutoff of 550 eV was used, and all calculations were spin-polarized. High spin Fe was assumed, extrapolating from calculations of the parent monoclinic Fe 2 (MoO 4 ) 3 phase that predicted a high- spin anti-ferromagnetic solution. All structures were geometry-optimized with no assumed symmetry by performing a series of constant volume structural relax- ations, and fitting the resultant data to the Murnaghan equation of state. Indi- vidual geometry optimizations were deemed converged when all atomic forces fell below 0.01 eV Å −1 . Orthorhombic Li 2 Fe 2 (MoO 4 ) 3 was modeled using a single unit cell (76 atoms), with 2 x 2 x 2 Monkhorst-Pack k-point sampling, using our Rietveld-refined structure as a starting configuration. Because we were unable to directly refine the monoclinic Na 2 Fe 2 (MoO 4 ) 3 structure, an approximate struc- ture used to initialize the DFT geometry optimization was constructed by project- ing the optimized Li 2 Fe 2 (MoO 4 ) 3 structure onto a (1 x 2 x 1)-expanded mono- clinic cell (152 atoms). The structural relaxation of the Na + -intercalated phase used a 2 x 1 x 2 Monkhorst-Pack k-point grid. 5.2 Results Pristine Fe 2 (MoO 4 ) 3 was prepared by a precipitation method following the approach of Peng et al.[201] After annealing at 400 ◦ C for 6 h, the phase-pure powder possessed a nanorod-like morphology (Supporting Information, Figure S1). Figure 5.2 shows the results of the Rietveld refinement of the monoclinic P2 1 /c structure against the synchrotron X-ray (Figure 5.2a) and neutron diffrac- tion (Figure 5.2b) patterns of the as-prepared Fe 2 (MoO 4 ) 3 electrode material. 69 Figure 5.1: Illustration of (a) the unit cell of pristine, monoclinic Fe 2 (MoO 4 ) 3 and (b) a "lantern unit" that consists of three MoO 4 tetrahedra connecting two FeO 6 octahedra. The diffraction patterns were refined simultaneously with structural parame- ters starting from the model of Chen et al.[212] Detailed structural parame- ters including atomic positions, cell parameters, and anisotropic displacement parameters are given in the Supporting Information (Table S1). Both diffrac- tion patterns could be fully indexed to lattice planes from the expected mono- clinic phase without any evidence of impurity peaks, confirming that the parent Fe 2 (MoO 4 ) 3 utilized throughout this study was phase pure. Raman spectroscopy showed that the as-prepared Fe 2 (MoO 4 ) 3 possessed bands characteristic of the monoclinic structure at 988, 967, 930 cm −1 (symmetric stretching modes of ter- minal Mo=O bonds in three distinct MoO 4 tetrahedra); 817, 776 cm −1 (asym- metric stretching modes of MoO 4 units); and 356 cm −1 (MoO 4 bending mode) (Supporting Information, Figure S2).[213] X-ray photoelectron spectra (XPS) of 70 the Fe2p and Mo3d regions suggest there is only one chemical environment for both Fe and Mo in the pristine, as-prepared Fe 2 (MoO 4 ) 3 by the presence of char- acteristic Fe3+ 2p3/2 and Mo6+ 3d5/2 peaks with binding energies of 711.3 eV and 232.5 eV, respectively (Supporting Information, Figure S3).[214, 215] Figure 5.2: Rietveld refinements of pristine, monoclinic Fe 2 (MoO 4 ) 3 against (a) synchrotron XRD pattern obtained at the 11-BM beamline at Argonne National Laboratory and (b) neutron diffraction pattern obtained at the POWGEN beam- line at Oak Ridge National Laboratory. The weighted profile R-factor (Rwp) was determined to be 6.01% for all banks being refined simultaneously and weighed equally. Absorption correction was carried out using absorption function 0 in GSAS. Only the diffraction pattern from bank 3 at POWGEN is given here. 5.2.1 Li + insertion The first ten galvanostatic electrochemical cycles of Fe 2 (MoO 4 ) 3 against Li/Li + at a current rate of C/10 are given in Figure 5.3a. The initial specific capacity of 92 mA·h·g −1 agrees well with the theoretical capacity of intercalation 71 of two Li + ions per formula unit (90 mA·h·g −1 ), with the slight excess capac- ity likely associated with the formation of a passivating solid electrolyte inter- face layer.[216] Without rigorous optimization of the cell assembly, the batteries showed a high capacity retention of around 90% of the initial capacity after 25 cycles (Supporting Information, Figure S4). The derivative of the galvanostatic cycling (inset of Figure 5.3a) shows two peaks centered at 3.0 V and 3.05 V dur- ing reduction and oxidation, respectively. Such a small polarization is reflective of the highly reversible intercalation of Li + into the framework of Fe 2 (MoO 4 ) 3 . The voltagecomposition curve exhibits a single plateau over a wide range of lithium content during Li + (de)insertion, which indicates the intercalation pro- cess predominantly occurs through a two-phase process where a lithium-rich phase is created directly rather than a solidsolution process, since there is a con- tinuous change in the lithium content throughout each particle.[196] The coexistence of two phases was further demonstrated using in-situ X-ray diffraction, taken continuously during electrochemical cycling, and is shown as a heat map in Figure 5.3b with the y-axis corresponding to the equivalents of inserted Li + . During the discharging process, new reflections at 24.5 ◦ , along with several others at 36 ◦ and 38.5 ◦ , clearly appear (Supporting Information, Figures S5 and S6). The intensity of other reflections (e.g., at 23.5 ◦ , 36.8 ◦ , and 39.8 ◦ ) gradually decreases until finally disappearing as the Li + content increases. As shown in the Supporting Information in Figure S6, all reflections of the fully elec- trochemically lithiated phase, except those contributed by the beryllium cell used for the in-situ measurements, can be indexed to the orthorhombic Li 2 Fe 2 (MoO 4 ) 3 phase originally reported by Torardi et al.[217] During the charging process, changes in the intensities of reflections were observed exactly in the opposite way, indicating excellent structural reversibility 72 Figure 5.3: (a) Galvanostatic electrochemical cycling of Fe 2 (MoO 4 ) 3 against Li + insertion and its derivative (shown as inset). (b) 2D pattern based on the in-situ XRD of Li + insertion into Fe 2 (MoO 4 ) 3 . upon Li + (de)insertion. Positions of the reflections do not vary with Li + inser- tion in the pristine Fe 2 (MoO 4 ) 3 , which excludes the possibility of a solid solu- tion process. These changes in the intensities of reflections are illustrative of the evolution of the initial monoclinic Fe 2 (MoO 4 ) 3 structure into orthorhombic Li 2 Fe 2 (MoO 4 ) 3 with the increasing Li + content, and the coexistence of pristine and lithiated structures in various ratios depending on the Li + concentration. 73 Thus, a two-phase Li + (de)insertion process in the anti-NASICON Fe 2 (MoO 4 ) 3 was confirmed for the first time through in-situ XRD. The crystal structure of orthorhombic Li 2 Fe 2 (MoO 4 ) 3 (Supporting Informa- tion, Figure S7) shows an identical structural topology to the parent monoclinic phase with regards to polyhedral connectivity, so the transformation is purely displacive in nature. While the in-situ XRD experiments provide some insight into the mechanism for these displacements, the patterns do not have sufficient resolution or intensity to precisely refine changes in the atomic positions. There- fore, Fe 2 (MoO 4 ) 3 was chemically lithiated as discussed in the experimental sec- tion. The results of the Rietveld refinement of the inserted structure against synchrotron X-ray and neutron diffraction data is shown in the Supporting Infor- mation in Figure S8 with detailed structural parameters listed in Table S2. To describe the transition between the monoclinic and orthorhombic structures in terms of the reversible symmetry-allowed distortions, the AMPLIMODES pro- gram at the Bilbao Crystallographic Server[206] was used to identify the trans- formation matrix (given in Supporting Information, Table S3) for the transition from the low-symmetry (P2 1 /c) pristine structure to the higher-symmetry (Pbcn) lithiated structure. The vectors corresponding to the direction of the displace- ments for each element were calculated and are illustrated in Figure 5.4. A com- plete list of the vectors is given in the Supporting Information, Tables S5 and S6. The most prominent displacement seen from these vectors are those associ- ated with oxygen atoms. The vectors on Fe and Mo atoms are extremely small, indicating that the center of mass for the rigid polyhedral subunits remains fixed as they rotate to enlarge the unit cell volume. It is interesting to note, as shown in Figure 5.4b, that Li + ions are inserted between two neighboring lantern units 74 Figure 5.4: Amplimodes analysis between pristine Fe 2 (MoO 4 ) 3 and lithiated Li 2 Fe 2 (MoO 4 ) 3 , with positions of Fe, Mo, and O in pristine Fe 2 (MoO 4 ) 3 and posi- tion of Li in Li 2 Fe 2 (MoO 4 ) 3 converted into the same reference structure setting. The transformation vectors are plotted on each atom in (a) and filtered by ampli- tude of 0.4 Å in (b) for clarity. FeO 6 octahedra are shown in green, and MoO 4 tetrahedra are shown in gray. Li + ions are left out for amplimodes analysis, but are drawn in the reference structure for clarity. and, as a result, there are significant displacement of the oxygen atoms toward the guest ions, presumably a consequence of local bonding or electrostatic inter- actions. This suggests that the ability for the framework to distort in such a way that Li-O interactions can form and break in a reversible fashion is an important consideration in the design of efficient intercalation hosts. 75 5.2.2 Na + insertion Figure 5.5a shows the galvanostatic cycling of Fe 2 (MoO 4 ) 3 against Na/Na + . An initial capacity of 79 mA·h·g −1 was found, which corresponds to the inser- tion of 1.7 Na + per Fe 2 (MoO 4 ) 3 , which is comparably smaller than that observed with Li + insertion (Supporting Information, Figure S4). The capacity was found to drop to 87% of the initial capacity after 20 cycles, showing a similar capac- ity retention to that of the Li/Fe 2 (MoO 4 ) 3 cell. The most pronounced difference from Li + insertion is that there are two distinct slope regions during the Na + - intercalation process instead of a single plateau. This is signified by two peaks (centered at ca. 2.7 V and 2.58 V) in the derivative curve given as the inset of Fig- ure 5.5a. The reproducibility of these peaks with respect to cell potential during subsequent cycles demonstrates that the Na + de(insertion) process is also highly reversible, and the electrochemical profile is intrinsic to the Na + -intercalation process. The in-situ XRD experiments during Na + intercalation showed substantial differences compared to that of Li + insertion (Figure 5.5b). During discharge, a constant shift in the Bragg reflections toward lower angles was observed with all the peaks returning to their original positions after fully charging. This shift corresponds to the expansion of the unit cell volume in order to accommodate the guest alkali cations during the insertion process, but given that no new reflections evolve during cycling, this indicates that a solid solution mechanism is at play, in contrast to the biphasic process described for lithium. While the laboratory XRD patterns did not provide enough intensity to allow for a full refinement of the atomic positions, it was possible to track changes in the unit cell using Le Bail fits to the inserted patterns. Relative changes in these parameters are plotted in Figure 5.6 with a complete list of the refined lattice 76 Figure 5.5: (a) Galvanostatic electrochemical cycling of Fe 2 (MoO 4 ) 3 against Na + insertion and its derivative (shown as inset). There are two slope regions during both insertion and deinsertion. The turn from one region to the other corre- sponds to 0.8 Na + per formula being (de)inserted. (b) 2D pattern based on the in-situ XRD of Na + insertion into Fe 2 (MoO 4 ) 3 . parameters given in the Supporting Information, Table S7. A volume change of 6% was found for insertion of 1.5 Na + ions per formula unit, with the dominant change in the unit cell corresponding to a roughly 4% elongation of the b-axis. While the magnitude of the change along the a- and c-axes is less pronounced, the trend seems to suggest the underlying insertion mechanism. The c-axis does 77 not significantly change during the initial stages of Na + insertion, while the a- axis exhibits a continuous extension. After the insertion of 0.8 Na + per formula unit, the trends are reversed (i.e., the change in a-axis flattens out while the c- axis shows a continuous increase). It is also important to recognize that the unit cell parameters of the Na + intercalated monoclinic phase and the orthorhom- bic structure identified for Li 2 Fe 2 (MoO 4 ) 3 suggests that the parent framework of Fe 2 (MoO 4 ) 3 actually rotates in the same way regardless of Li + or Na + insertion. The primary difference between the two space group choices depends heavily on the position where the guest ions end up sitting. Within this context, the pro- nounced elongation of the b-axis is probably the result of the larger Na + ions relative to that of Li + and that the size of the intercalant plays a crucial role in determining how the structure can distort. Figure 5.6: Lattice evolution along Na + insertion into Fe 2 (MoO 4 ) 3 from Le Bail fitting of insitu X-ray diffraction patterns. Given the complexity of the monoclinic structure, which contained 34 unique atomic positions and over 150 refinable parameters, it proved impossible to directly refine the structure of Na + -inserted Fe 2 (MoO 4 ) 3 by starting from the 78 model of the pristine structure. Therefore, we turned to density functional theory calculations to obtain a predicted structure for monoclinic Na 2 Fe 2 (MoO 4 ) 3 , which was used for further Rietveld refinement. We first calculated an optimized struc- ture for orthorhombic Li 2 Fe 2 (MoO 4 ) 3 , starting from the experimentally deter- mined geometry. This gave excellent agreement with the Rietveld refined geom- etry (e.g., lattice parameters agree to within 0.5%), which demonstrates that our PBEsol+U calculations are able to well describe the structure for this high- symmetry lithium-ordered phase. The atomic positions from the DFT-optimized orthorhombic Li 2 Fe 2 (MoO 4 ) 3 were then projected onto a monoclinic cell as a starting structure for the Na 2 Fe 2 (MoO 4 ) 3 structure optimization. The calculated Li + -inserted and Na + -inserted structures from the simulations are included in the Supporting Information, Tables S8 and S9. Despite the extremely good agree- ment between the DFT cell for the lithiated compound and its experimentally determined structure, there is a striking difference between the increase of the unit cell volume for Na + -insertion predicted by DFT (16.0%) and that found experimentally (6.0%). Notwithstanding this discrepancy, the DFT-optimized Na 2 Fe 2 (MoO 4 ) 3 structure generates a close match for the peak position and inten- sities in the neutron diffraction data (Supporting Information, Figure S9). Using the DFT-optimized structure for subsequent Rietveld refinement, however, pro- duces unphysically large atomic displacement parameters for the sodium atoms. These large displacement parameters indicate that the intensity contributed to the pattern by Na + is negligible, and suggests a more complex distribution of sodium ions in the experimental sample. Refining the optimized structure again after removing sodium atoms produces a fit nearly identical to the refinement with the sodium included (Supporting Information, Figure S10). The resulting 79 framework is illustrated in the Supporting Information, Figure S11 with the full structural description listed in Table S10. There are two possible explanations for the lack of scattered intensity associ- ated with the sodium atom positions. Either the Na + ions are disordered through the lattice, or they adopt some kind of long wavelength superstructure that is not immediately obvious in the powder diffraction experiment. Considered along- side the erroneous predicted volume increase for the DFT-optimized structure, these results suggest that the calculated volume expansion is a consequence of an (artificial) ordering of Na + within the polyanion framework. The use of a (1 x 2 x 1) periodic cell for the DFT calculations imposes an artificial symmetry on the Na + ions, and this may block a fully disordered sodium configuration, instead predicting an ordered Na + -inserted structure. The larger size of Na + ver- sus Li + , however, means this Na + ordering can only be accommodated within the polyanion framework through a significant volume expansion. This conceptual model predicts that the smaller volume expansion and apparent Na + disorder observed in the experimental refined structure are coupled. We propose that in the experimental system the lattice strain produced by intercalating large Na + ions could be accommodated by a disordering of these ions, rather than the oth- erwise necessary volume expansion predicted in the higher symmetry DFT calcu- lations. In contrast, smaller Li + ions are able to occupy an ordered arrangement of sites upon intercalation, without imposing a large volume change, leading to highly consistent DFT-predicted and experimental structures. The inability of our DFT calculations to predict structures in close agreement with those obtained 80 from experiment for Na + -inserted Fe 2 (MoO 4 ) 3 indicates the limitations of stan- dard periodic zero-temperature calculations for modeling disordered intercala- tion phases, and suggests alternative approaches that consider thermally dis- ordered systems, such as large-scale molecular dynamics, may be necessary to obtain a detailed atomic-scale description of these disordered ions. Figure 5.7: Experimental total scattering data for pristine, Li + -inserted, and Na + - inserted Fe 2 (MoO 4 ) 3 . To further characterize the nature of the insertion process on the local struc- ture of Fe 2 (MoO 4 ) 3 , total neutron scattering data was collected on the pristine and chemically inserted samples (Figure 5.7). The peak observed at 1.76 Å, which corresponds to the Mo-O interatomic distance within the MoO 4 tetrahedra, does not change at all after Li + /Na + insertion. This is consistent with the more covalent and therefore more rigid nature of the MoO 4 tetrahedra. In contrast, there is a very significant elongation of the Fe-O bonds that should be expected as the oxidation state of Fe changes from 3+ to 2+ upon insertion of the alkali 81 ions. It is also interesting to point out that that the long range correlations, espe- cially those at long r values > 3 Å, are significantly more dampened in the Na + inserted phase than that of the pristine or Li + -inserted phases. This reflects the local disorder that results as a consequence of the interactions with the complex distribution of Na + through the lattice. Finally, Raman spectra of Li + - and Na + -inserted Fe 2 (MoO 4 ) 3 were collected to determine if signatures of the disorder could be identified (Supporting Infor- mation, Figure S2). Both the lithiated and sodiated samples show Raman bands belonging to the MoO 4 units that appear to be either weak or unresolved. The strongest asymmetric stretching band of the MoO 4 units in the Raman spectrum of the pristine material appears to be very broad and weak in the spectra of lithiated and sodiated samples. However, the weaker bending and symmetric stretching modes remained unresolved. In addition, no shift in the positions of Raman bands was observed. This broadening and loss of intensity may be attributed to the redistribution of electron density in the Mo-O bonds through the intercalation of alkali guest ions into the pristine material, which may also be affected by the reduction of Fe3+ ions to Fe2+.[218, 219] Therefore, the effective force constants and polarizability derivatives may be varied as the elec- tron density is redistributed, resulting in the observed broadening and intensity loss in the Raman bands. The major difference was observed in the asymmetric stretching band of MoO 4 tetrahedra. The intensity of the strongest asymmetric stretching mode at 776 cm −1 was found to be lower than of the second asymmet- ric stretching mode at 817 cm −1 in the intercalated samples. This suggests that strong interactions between alkali ions and the polyhedral oxygen ions results in changed intensities for stretching motions in the MoO 4 units, while the frame- work remains largely the same. 82 With a description of the Li + - and Na + -inserted phases of Fe 2 (MoO 4 ) 3 , a com- parison between the changes to the structural frameworks should provide insight into the mechanism for intercalation. A representative portion of the refined framework with lithium determined by Li 2 Fe 2 (MoO 4 ) 3 refinement is illustrated in the Supporting Information, Figure S12. To understand how the polyhedral rotations are affected by the alkali ions, a set of dihedral angles are calculated from the structural models. The center atoms of two connected polyhedra form two planes with one oxygen atom from each of the polyhedra in these dihe- dral angles, which thus measures the relative rotational displacement between the two polyhedra. For example, the tweaking between the FeO 6 octahedra and MoO 4 tetrahedra can be measured by the dihedral angles of such as O2-Mo2- Fe1-O5. A complete list of the dihedral angles between these two polyhedra is given in the Supporting Information, Table S11. It can be seen there is a difference in every dihedral angle between the Na + -inserted and Li + -inserted Fe 2 (MoO 4 ) 3 , and a difference smaller than 2 ◦ can be considered as systematic residuals. There are more significant differences, such as the 3 ◦ difference in the O2-Mo2-Fe1-O5 and O2-Mo2-Fe1-O6 angles. From our symmetry mode analy- sis, the transformation from the pristine to the inserted phase is accompanied by oxygen atoms moving towards the inserted ion (Li + ), possibly driven by a strong electrostatic Li-O interaction. Thus, as O5 on the bottom left corner in Figure S12 moves towards the alkali ion, driving the FeO 6 to rotate in a way that essentially results in O5 and O6 moving away from O2 in the MoO 4 tetrahedra on the top left corner, and because the Li-O bond is shorter than the Na-O bond, the dihedral angle of O2-Mo2-Fe1-O5 is larger and that of O2-Mo2-Fe1-O6 is smaller in Li + - inserted Fe 2 (MoO 4 ) 3 than Na + -inserted phase. This again highlights that while a striking difference has been observed in the electrochemical cycling curves, 83 as well as structural phase changes between the Li + - and Na + -intercalation in Fe 2 (MoO 4 ) 3 , the fundamental mechanism for Li + - and Na + -intercalation is effec- tively the same but structural differences arise because of the differences in ionic radii. 5.3 Discussion We have investigated the insertion mechanisms of two different alkali guest ions into the Fe 2 (MoO 4 ) 3 framework and found that the ability for the frame- work to distort via cooperative rotations may be important for facilitating inter- calation. This has been confirmed by the symmetry-mode analysis of pristine and Li + -inserted Fe 2 (MoO 4 ) 3 , and through careful analysis of the total scattering data on both Li + - and Na + -inserted samples. Similar observations have been found in LiFeSO4F cathode materials, and the lack of connection between these two structure types implies that such a mechanism may have wider applications in understanding polyanionic electrode materials.[220, 221, 137] It should also be noted that the distortion through rigid and strongly covalently bonded polyhedra restricts the way the host structure can distort, and thus offers better longevity than oxide materials, which usually suffer from drop in performance due to severe structural changes during cycling.[63, 222] While the mechanism of polyhedral rotational distortion we have proposed has not been widely discussed with respect to intercalation electrodes, it has been extensively explored in the superionic conductor literature.[? 223] This is typically found in studies of two competing mechanisms: (i) a "paddle 84 wheel" mechanism, which suggests a strong correlation between the trans- port of cations and the rotation of polyhedral subunits, and (ii) a "percola- tion" mechanism, which claims independence between conduction pathway and polyhedral rotations.[224, 225] For example, through powder neutron diffrac- tion and reverse Monte Carlo (RMC) modeling, Karlsson et al. revealed that both types of mechanisms are present in high-temperature forms of Li2SO4 and LiNaSO4.[224, 223] These rotational distortions may suggest a more generalized mechanism for alkali ion intercalation in polyanionic materials, but more importantly, this work has revealed a difference in the behavior between Li + - and Na + -intercalation into the same host, which suggest that differences in the intrinsic nature of Li + and Na + guest ions may result in a modification of such a mechanism. We observed a two-stage, solid solution Na + insertion process as compared to a single, two- phase Li + insertion process, which is similar to what has been found in olivine- type FePO4 and explained by stronger interactions of Na + ions as compared to Li + ions with the host structure by Moreau et al.44 Our results also indicate that the two redox peaks in Fe 2 (MoO 4 ) 3 do not appear to correspond to the complete filling of two distinct Na + positions, but rather that the distribution of Na + ions is much more nuanced. There are two possible explanations for the lack of scat- tered Bragg intensity. First is that the Na + ions are disordered through the lat- tice. This disorder could be the result of two discrete crystallographic positions, each having a distinct chemical potential for the alkali ion, but as the frame- work rotates and distorts to accommodate more ions the Na + ions already in the framework are pushed off their ideal position. Second is that the Na + ions adopt a long wavelength superstructure that cannot be seen in the powder diffraction experiment. A more careful investigation into intermediate compositions may 85 provide more valuable information in this regard, but a comparison between the total neutron scattering data on Li + - and Na + -inserted Fe 2 (MoO 4 ) 3 shows that peaks at high r values are less intense in Na + -inserted sample than those in Li + - inserted sample, which reinforces the notion that the rotational distortions are less coherent following Na + than Li + insertion. Both explanations could result from the larger ionic size of Na + than Li + , and thus a steric hindrance for the rotational distortions to allow the Na + ions to find an ideal position. Finally, we note that the highly insulating character of these materials would require that the charge transport through the lattice would have to be mediated through some kind of polaronic hopping of the electrons. It could be possible that the structural distortions reported here are related to this mechanism of charge hopping, and further work using Xray absorption spectroscopy may be useful in elucidating this aspect of the transport properties. 5.4 Conclusion In summary, we have presented a detailed structural characterization of the process by which Li + and Na + ions intercalate into Fe 2 (MoO 4 ) 3 , and how these structural transformations are related to the correlated rotations of rigid polyan- ionc subunits. The combination of electrochemical cycling and in-situ powder XRD confirmed a two-phase process for Li + -intercalation, compared to a two- stage solid solution insertion process for Na + -intercalation, which corresponds to Na + filling two different sites. Furthermore, based on our Rietveld refinements, symmetry-mode analysis and examination of neutron total scattering data, we proposed a concerted polyhedra rotational distortion mechanism for the alkali guest ion intercalation into the anti-NASICON Fe 2 (MoO 4 ) 3 host framework. It is 86 also shown that during Li + -insertion, Li + ions fill defined positions that allow the transformation from the pristine monoclinic phase to the lithiated orthorhombic structure. However, Na + insertion occurs in a more complex manner that appears to result from the larger size of Na + compared to Li + . Such a mechanism, as well as the application of symmetry-mode analysis along with structural and electro- chemical characterization, may be applied to other polyanionic material systems. Insights gained through these analyses may facilitate the discovery of new inter- calation materials and the improvement of existing ones. 87 Chapter 6 Influence of local distortion on Li + - and Na + -intercalation in ReO 3 The intermittent nature of renewable energy, such as solar and wind, requires that cost-effective methods be developed to store the energy produced during peak supply and deliver during peak demand periods. Moreover, the fluctuating demand of energy during the day requires that the stored energy be delivered in a highly efficient manner. A number of energy storage technologies, such as Li- and Na-ion batteries, redox flow batteries, and capacitors, have all gained significant amount of attention owing to their respective advantages. The excellent energy density of rechargeable Li- and Na-ion batteries underpin their great success in portable electronics. However, wider applications of rechargeable batteries in large scale energy storage such as grid and electric vehicles still face challenges in cost and long-term reliability. Redox flow batteries and capacitors are superior in terms of either cost or rate capability, but each facing their own limitations as well. Center to each of these electrochemical systems are the electrode materials and the associated energy storage processes. In recent years, tremendous efforts have been carried out to take advantage of different materials and redox pro- cesses in order to design a next-generation energy storage system that is sustain- able, safe, efficient and cost-effective. Such efforts include utilizing intercalation systems of multi-valent cations and combining intercaltion of different cations. It 88 has also been shown that by taking advantage of surface-controlled redox when particl size of the electrode material is sufficiently small, traditional intercalation compounds can also exhibit superior rate capabilities. Such systems, coined as pseudocapacitors, are capable of charging and discharging rates level with those of capacitors while maintaining energy density comparable to those of Li-ion batteries, have great promise in energy storage areas when power and energy density are both demanding, such as electric vehicles. Figure 6.1: The structure of ReO 3 with a space group of P m ¯ 3m with viewing direction of (1 0 0) (a) and (1 1 1) (b). Oxygen and rhenium atoms are shown in orange and brown spheres, respectively. The ReO 6 octahedra share corners with each other to form a three-dimensional defect-perovskite network. In an effort to discover similar systems that works at higher potentials, we turn to a less studied oxide material, ReO 3 . The structure of ReO 3 , shown 89 in Figure 6.1, features corner-sharing ReO 6 tetrahedra that resembles a per- ovskite structure. The empty A site of the perovskite structure serves as large, three-dimensional interstitial for intercalation of guest ions. Previous chemi- cal insertion studies have shown that ReO 3 are capable of accommodating two Li + ions per formula unit with both LiReO 3 and Li 2 ReO 3 take on a hexagonal structure.[226, 1] However, there is so far no full reports of electrochemical studies to understand the reversibility of the Li-interclation process in ReO 3 , and especially how the intercalation process is correlated to the local distortion of polyhedral groups. Our previous studies have shown that polyhedral rota- tional distortions plays an important role in the intercalation process of both polyanionic and oxide electrode materials. It has also been shown that intercal- tion of Na + ions can be drastically different from intercaltion of Li + ions in the same system.(cite) Here we present an electrochemical study of Li + - and Na + - intercalation into ReO 3 and how the electrochemical properties are correlated with its structure. Our results show that local coordination and strain has signif- icant effects on the electrochemical performance Li + - and Na + -intercalation in ReO 3 . 6.1 Experimental Details Synthetic Methods. ReO 3 nanoparticles were prepared with a method reported by Chong et. al.[227] 0.0242 g Re 2 O 7 was added into 0.5 mL methanol in a 10 mL round bottom flask and stirred inside an Argon-filled glovebox until the Re 2 O 7 was completely dissolved. The solution was kept in an oven at 250 ◦ C for 5 mins. A shinny, thin layer of ReO 3 was formed on the bottom of the flask following the evaporation of methanol. Then the sample is scraped off the flask 90 and ready for physical characterizations. Powder samples used for electrochemi- cal cycling are heated at 200 ◦ C under vacuum for at least two hours before being transferred directly into the glovebox. Physical Characterization. Laboratory X-ray diffraction patterns were col- lected on a Bruker D8 diffractometer with a Co Kα source (λ 1 = 1.78897 Å, λ 2 = 1.79285 Å), equipped with a Lynxeye detector. High resolution synchrotron powder diffraction data were collected using beamline 11-BM at the Advanced Photon Source (APS), Argonne National Laboratory using an average wavelength of 0.413682 Å. Discrete detectors covering an angular range from -6 to 16 ◦ 2θ were scanned over a 34 ◦ 2θ range, with data points collected every 0.001 ◦ 2θ and a scan speed of 0.01 ◦ s −1 . Structure and charge density visualization and analysis were performed using VESTA.[2] 6.1.1 Electrochemical Testing Microelectrodes are primarily used for characterizing the redox processes. These electrodes are prepared by drop casting 30μL ReO 3 suspension, prepared by adding 2 mg ReO 3 in 1 mL ethanol and sonicating for an hour, onto an oxy- gen plasma cleaned 1 cm x 1 cm stainless steel shim, and dried in vacuum at 110 ◦ C for two hours. Cyclic voltammetry (CV) scans are conducted with a three- electrode configuration using a three-neck flask inside an Argon-filled glovebox. To study Li + -intercalation, polished Li metal strips are used as both counter elec- trode and reference electrode. And 1 M LiPF 6 in EC:DMC (1:1 by volume) is used as electrolyte. To study Na + -intercalation, Na metal strips are used in place of Li. 91 And 1 M NaPF 6 in EC:DMC (1:1 by volume) is used as electrolyte. Around 10 mL electrolyte is used in a 25 mL flask for sufficient coverage of the electrodes. Thick film electrodes are prepared by mixing and grinding ReO 3 , PVDF, and carbon super P at the ratio of 8:1:1. Suitable amount of N-Methyl-2-pyrrolidone (NMP) is added to the mixture to form a slurry. The slurry was casted onto an aluminum foil and flattened to a film with a thickness of 40μm using a doctor blade. The slurry was dried in air for 4 hours and in a vacuum oven at 100 ◦ C overnight. Galvanostatic cycling of the thick film electrodes was carried out using Swagelok-type cells. Cells were assembled in an argon-filled glove box with a Li-metal disk as the negative electrode. Two Whatman R GF/D borosilicate glass fiber sheets were used as the separator. In-situ XRD was carried out using powder ReO 3 as working electrode inside a larger Swagelok cell with a Beryllium metal disk, with a thickness of 0.4 mm and a diameter of 30 mm, to serve as current collect for the working electrode while being mostly transparent to X-ray upto 60 ◦ for our Co radiation. 6.2 Results and Discussion The structure of ReO 3 is shown in Figure 6.1. The cubic ReO 3 has a space group of P m ¯ 3m and features an A-site deficient perovskite structure with corner- sharing ReO 6 octahedra forming a three-dimensional network. A few methods have been reported to prepare crystalline ReO 3 samples.[228, 229] We chose the method reported by Chong et. al for its simplicity and promise to produce monodispersed nanoparticles.[227] Brick-red colored film of ReO 3 nanoparticles are formed on the wall of a 10-mL round bottom flask after the solution of Re 2 O 7 92 Figure 6.2: TEM micrograph of as-prepared ReO 3 nanoparticles. in methanol is heated in a convection oven at 250 ◦ C for 5 minutes. It is our experience that the longer the solution is sit between being prepared and being treated, the larger the resultant particle size. When the methanol solution is heated as soon as Re 2 O 7 is fully dissolved, highly monodispersed nanoparticles of around 5 nm in diameter are produced as shown in the TEM micrographs in Figure 6.2(a). Lattice fringes observed in HRTEM, shown in Figure 6.2(b), indicates that even with such short reaction time and small particle size, the resultant nanoparticles are highly crystalline. Figure 6.3 shows a Lebail fit of a laboratory XRD pattern using a starting structure reported by Jørgensen et. 93 al.[230] The XRD reflections confirms the high crystalinity of the the sample and no apparent sign of any secondary phases. 20 40 60 80 2 θ (λ =f0.413682fÅ)f Intensityf(arb.funit) Observation Calculation Difference Figure 6.3: Refinement against Laboratory XRD patterns of bulk ReO 3 samples. 6.2.1 Li + Intercalation To explore the redox processes involved in Li + intercalation into ReO 3 , cyclic- voltammetry (CV) scans of ReO 3 microelectrodes in a three electrode system are performed. It can be seen from Figure 6.4 that drastically different features are obtained when different cut-off voltages are used. When the lower voltage is set at 2.5 V against Li/Li + , a pair of symmetrical redox peaks are seen. When the cut-off voltage is lowered to 2 V, an additional redox peak at roughly 2.3 V is seen during discharging. On charging the additional peak becomes less significant and the higher-potential peak becomes broader. Yet another additional pair of redox peaks are observed when the lower limit of the cut-off voltage is extended to 1 V. As a complement to the CV scans, galvanostatic cycling using ReO 3 powder in a two-electrode swagelok cell is also conducted. Similar features of redox peaks at 2.8 V, 2.3 V, and 1.3 V can be found. (Supporting Info Figure S1 and S2) It 94 is important to note that the completion of the peaks at 2.8 V, 2.3 V and 1.3 V corresponds to 0.3 Li + , 1.0 Li + , and 2.0 Li + inserted per formula unit of ReO 3 . Figure 6.4: Cyclic voltammetry scans of ReO 3 microelectrodes against Li/Li + in a three electrode cell inside the glovebox. To fully understand mechanism involved in each redox process, in-situ XRD is carried out in our laboratory X-ray diffractometer. A two-dimensional presen- tation of the XRD patterns collected every 30 minutes during a full discharge (to 1 V) and charge (to 4 V) cycle is given in Figure 6.5. During the initial dis- charge, when less than 0.35 Li + is inserted, there is a continuous shift of all the ReO 3 reflections. All the peaks, most significantly observed at 27.5 ◦ , 40 ◦ , and 57 ◦ , shifted to higher angles without any significant intensity loss. No new reflections is observed during this period. It can thus be concluded that a solid solution process involving a cubic Li x ReO 3 phase is responsible for this insertion period and is correlated with the redox peaks centered around 3 V when the ReO 3 is cycled between 2.5 V and 4.0 V. When more Li + is inserted, no further peak shift is observed. There is a gradual loss of intensity at the ReO 3 peaks, most obviously evidenced at the peak at 40 ◦ , which is completely gone after 1.2 95 Figure 6.5: Two-dimensional presentation of in-situ XRD patterns collected dur- ing a full discharging and charging cycle using ReO 3 powder as the working electrode in a swagelok cell. The y-axis corresponds to the equivalent amount of Li (de)inserted in ReO 3 , calculated from the charge delivered by the potentiostat. The color indicates the instensity. equivalent amount of Li + is inserted. Due to limitations of the instrument reso- lution and reflections from the insitu cell itself, changes elsewhere is not easily observable. However, towards the end of the discharge, appearance of a new peak and intensity growth at 42 ◦ is observed. The pre-existing intensity comes from the empty cell. (SI Figure S3) When the cycling is reversed, so is evolution of the XRD patterns. The disappearance of the peak at 42 ◦ spans longer range than its appearance. The re-appearance and recovery of the peak at 40 ◦ does not occur until the amount of Li + per formula is reduced to 0.6. The fact that the charge and discharge process does not exactly mirror each other is best demon- strated by the asymmetric shape of the redox peaks during discharge and charge in the derivative curve of the galvanostatic cycing data. (SI Figure S2) Although galvanostatic cycling of ReO 3 shows that it is able to reversibly cycle 2 Li + ions at the initial cycles, it seems to suffer low long term cycling stability. 96 Figure 6.6: Capacity retention of Li + -intercalation in ReO 3 with lower cut-off voltage set at 2.8 V and 0.5 V. The higher cut-off voltage is set at 4.0 V for both occasions. As it’s shown in Figure 6.6, only 50% of the initial capacity is retained after 45 cycles when the lower cut-off voltage is set at 0.5 V. Even when the lower cut- off voltage is set at 2.8 V, which means that intercalation is limited to only the higher-potential solid solution process, significant capacity is lost after 50 cycles. (Figure 6.6) The long term cycling stability of ReO 3 against Li/Li + probably origi- nates from the structural changes associated with ReO 3 during Li + -intercalation. As Cava et. al pointed out, large degree of twist of the framework is required to for the favored coordination of Li due to the large mismatch between the sizes of the interstitial void and Li + ion. As demonstrated in Figure 6.1(b) and Figure 6.7(a), the phase change from ReO 3 to LiReO 3 does not require break- ing any bonds but a rotation of ReO 3 along the (1 1 1) direction.[1] According to reported structures of LiReO 3 and Li 2 ReO 3 , a full insertion of two Li + ions involves a 5% unit cell volume expansion and same amount of shrinkage. More- over,even with such twist, the local coordination of Li in the inserted LiReO 3 and Li 2 ReO 3 still involve significant amount of distortion. Bond length distor- tion index of LiO 6 octahedra is 0.095 in LiReO 3 .[231] In Li 2 ReO 3 the repulsion 97 Figure 6.7: Structure of LiReO 3 viewed from the (0 0 1) direction (a) and a zoomed-in illustration of the local coordination of Li in LiReO 3 (b). Structure representation constructed with reported structural data by Cava et. al[1] using the VESTA program.[2] between the two Li + ions pushed their polyhedra to become less distorted. But the distortion indices are still as high as 0.084 and 0.047. To put it into per- spective, such index is 0.00477 for Re in LiReO 3 and 0.020 for Li in LiFePO 4 , [1, 42] Such high distortion likely explains the low crystalinity of the Li-inserted structure in our in-situ experiments and the poor cycling stability of ReO 3 . 6.2.2 Na + Intercalation Having established the mechanism of Li + intercalation into ReO 3 and that its poor long-term cycling stability probably originates from the local strain of the Li- inserted structure, we turn into the intercalation of Na + into ReO 3 . The apparent 98 difference in ionic radii between Li + and Na + may result in different intercalation mechanism and provide insights into how optimum cycling performance can be achieved. Figure 6.8(a) shows the CV scans of a ReO 3 microelectrode cycling against Na/Na + with different cut-off voltages. Similar to what has been found in Li + intercalation, there are several redox peaks on discharge, centered around 2.7 V, 2.35 V, and 0.9 V. Based on the redox potential of Na/Na + and Li/Li + , cell voltages against Na/Na + should be 0.3 V lower than that agsint Li/Li + for the same processes. It then seems reasonable to speculate that the redox peaks at 2.7 V, 2.35 V and 0.9 V in Na + intercalation are related to the same processes associated with the peaks at 3.0 V, 2.65 V and 1.25 V in Li + intercalation. However, it should be noted that unlike the asymmetry in shapes of the reduc- tion and oxidation peaks in Li + intercalation, the two peaks at higher poten- tials in Na + intercalation are quite symmetrical, indicating a full reverse of the redox processes. Through integration of the peak areas during the CV scans, the capacity retention of Na + -intercalation in ReO 3 with different cut-off voltages is shown in Figure 6.8(b). The theoretical capacity of ReO 3 based on Re 6+ /Re 4+ two-electron reduction is 228 mA·h/g. The extra capacity achieved during the initial discharge when the lower cut-off voltage is set at 0.3 V is probably asso- ciated with further reduction of Re through conversion reactions, which likely explains its extremely poor capacity retention. When only the higher redox pro- cess is involved by cutting off the voltage at 2.4 V, no significant capacity loss is observed at all after 30 cycles, demonstrating superior reversibility of Na + - intercalation than Li + in the low concentration region. To explore the rate capa- bility of this highly reversible redox process, CV scans of the microelectrodes at a wide range of rates are carried out and shown in Figure 6.9(a). As it is shown 99 Figure 6.8: (a)Cyclic voltammetry scans of ReO 3 microelectrodes against Na/Na + at a rate of 1 mV/s in a three electrode cell inside the glovebox.(b)Capacity reten- tion of the same electrodes cycled between different voltage cut off at a rate of 1 mV/s. The higher voltage cut off is set at 3.5 V for both occasions. by a range of previous studies, capacitive, i.e. surface process, exhibit differ- ent relationship between peak current and sweep rates during CV scans from diffusion-limited process. The current (i) scales with sweep rate (v) in capacitive process but scaless with v 1/2 in processes that are limited by solid-state diffusion. Log(peak current) and log(sweep rate) are plotted in Figure 6.9(b). A linear fit with a slop of 0.51 indicates that the redox process under study is almost exclu- sively limited by solid-state diffusion, although the average particle size is as small as 4 nm. This is probably due to the strong interaction between Na + ions 100 Figure 6.9: (a) CV scans of ReO 3 microelectrodes between 2.4 V and 3.5 V against Na/Na + between 1 mV/s and 100 mV/s. (b) Plot and linear fitting of log(peak current) against log(sweep rate). The linear fit (R 2 = 0.9886) results in a slope of 0.5086 with a standard error of 0.0222. and the host structure. To further understand the kinetic parameters of the Na + - insertion process, GITT experiments on a film electrode in a swagelok cell were carried out. The voltage-composition plot of the experiment, in which the work- ing electrode is titrated with 30 mins of constant current that is equivalent to C/10, followed by 2 hours of relaxation, is plotted in SI Figure S5. By combining the relaxation peaks throughout the GITT experiment, a thermodynamic voltage- composition curve is obtained, shown as the red curve in SI Figure S5. The ther- modynamic voltage-composition curve still features a polarization between the charging and discharging as high as 1.5 V. Such high polarization may result 101 from a highly insulating phase produced during insertion. The equation derived by Weppner and Huggins is used to extract the chemical diffusion coefficient of Na + in Na x ReO 3 when x is less than 0.3.[232] With a contact area of 1.13 cm 2 , a constant current of 5.9 x 10 −5 A and other parameters available from the plot, a diffusion coefficient of 7.2 x 10 −11 cm 2 /s is obtained. Figure 6.10: Insitu XRD patterns of ReO 3 cycled against Na/Na + in a swagelok cell with powder ReO 3 as working electrode. The in-situ XRD patterns of ReO 3 against Na/Na + are plotted in Figure 6.10. There is a continuous decrease of intensity of the ReO 3 reflections during dis- charging, and a continuous increase during charging. The (1 0 0) reflection at 27.6 ◦ is plotted at the beginning and end of each half cycle for the first two cycles is shown in SI Figure S4. A full recovery of the intensities after each cycle is achieved, demonstrating the excellent reversibility of the Na + -intercalation when the irreversible process observed at 0.7 V is avoided due to a high over- potential. The absence of new reflections during Na + -intercalation may due to the limit of our instrument resolution, or the Na + -insertion results in an amor- phous phase. However, unlike what occurs during Li + -intercalation, there is no any observable peak shift at any stage of the Na + -intercalation. As discussed earlier, the reduction peak at 3 V is associated with the initial solid solution 102 process of Li + -insertion, which resulted in continuous shift of peaks to higher angles. Since CV scans show same series of peaks at same potentials, consider- ing the shift from Li/Li + to Na/Na + , the redox process at 2.7 V is expected to be associated with solid solution of Na + -insertion. It is unlikely that the absence of peak shift is due to instrument resolution, since the effect is well captured during Na + -intercalation. Then a plausible explanation is that the size of Na + ions match well with the interstitial in ReO 3 , which means no apparent unit cell shrinkage/expansion. This also likely explains the better reversibility of Na + - intercalation than Li + -intercalation. Efforts to determine the structure of the Na + -inserted phases is currently underway. 6.3 Conclusion Through a variety of electrochemical and structural characterization tech- niques, we have presented a detailed study of the electrochemical cycling of ReO 3 against the intercalation of Li + and Na + ions. Although it has been shown that an interclation of two Li + ions perform formula unit of ReO 3 is possible, our results show such process suffer from poor long-term cycling stability. Combing our insitu XRD results and previous chemical insertion studies, we were able to propose a mechanism for the Li + -intercalation process and relate its poor reversibility to the local distortion of Li + ions. Building upon our understand- ing on the Li + -intercalation process, we we studied the Na + -intercalation into ReO 3 . Na + -intercalation into ReO 3 showed superior reversibility when it is lim- ited to the redox process at a potential higher than 2.4 V against Na + /Na. The improved reversibility is probably due to the reduced local distortion of the guest ion in the inserted structure. This study demonstrates the significance of local 103 rotational distortion on the performance of guest ion intercalation during elec- trochemical cycling. The understanding of the structure-property relationship in ReO 3 paves way for the study and optimization of electrochemical cycling of a wide range of phases. 104 Reference List [1] R.J. Cava, A. Santoro, D.W. Murphy, S.M. Zahurak, and R.S. Roth. The structures of the lithium inserted metal oxides li0.2reo3 and li0.36wo3. Journal of Solid State Chemistry, 50(1):121–128, 1983. [2] K. Momma and F. Izumi. VESTA: a three-dimensional visualization system for electronic and structural analysis. J. Appl. Cryst., 41:653–658, 2008. [3] DESA. World population prospects: The 2015 revision. Technical report, UN Department of Economic and Soical Affairs, 2015. [4] USDOE. 2015 renewable energy data book. Technical report, U. S. Depart- ment of Energy, 2015. [5] EIA. Monthly energy review. Technical report, Energy Information Admin- istration, 2016. [6] LBNL. 2015 wind tehnologies market report. Technical report, Lawrence Berkeley Naitonal Laboratory, 2015. [7] REN21. 2015 rrenewable global status report. Technical report, Renew- able Energy Policy Network for the 21st Century, 2015. [8] Susan Solomon, Gian-Kasper Plattner, Reto Knutti, and Pierre Friedling- stein. Irreversible climate change due to carbon dioxide emissions. Pro- ceedings of the National Academy of Sciences, 106(6):1704–1709, 2009. [9] M. Asif and T. Muneer. Energy supply, its demand and security issues for developed and emerging economies. Renewable and Sustainable Energy Reviews, 11(7):1388–1413, September 2007. [10] J. P. Barton and D. G. Infield. Energy storage and its use with intermittent renewable energy. IEEE Transactions on Energy Conversion, 19(2):441– 448, 2004. [11] Henrik Lund. Renewable energy strategies for sustainable development. Energy, 32(6):912–919, June 2007. 105 [12] John A. Turner. A realizable renewable energy future. Science, 285(5428):687–, July 1999. [13] H Lund and E Münster. Management of surplus electricity- production from a fluctuating renewable-energy source. Applied Energy, 76(1Ã´ câ ´ Cˇ nâ ˘ AIJ3):65–74, September 2003. [14] A. J. Roscoe and G. Ault. Supporting high penetrations of renewable gen- eration via implementation of real-time electricity pricing and demand response. IET Renewable Power Generation, 4(4):369–382, 2010. [15] Anya Castillo and Dennice F. Gayme. Grid-scale energy storage applica- tions in renewable energy integration: A survey. Energy Conversion and Management, 87:885–894, November 2014. [16] Imre Gyuk, Mark Johnson, John Vetrano, Kevin Lynn, William Parks, Rachna Handa, Landis Kannberg, Sean Hearne, Karen Waldrip, and Ralph Braccio. Grid energy storage. Technical report, U.S. Department of Energy, 2013. [17] J-M. Tarascon and M. Armand. Issues and challenges facing rechargeable lithium batteries. Nature, 414:359–367, 2001. [18] Joseph F. DeCarolis and David W. Keith. The economics of large-scale wind power in a carbon constrained world. Energy Policy, 34(4):395–410, March 2006. [19] J. B. Goodenough and Y. Kim. Challenges for rechargeable li batteries. Chem. Mater., 22:587–603, 2010. [20] USABC. Technology transition case study. Technical report, United States Advanced Battery Consortium, 2015. [21] D. Howell. Annual progress report for energy storage R&D. Technical report, US Department of Energy, 2014. [22] J.F. Whitacre, T. Wiley, S. Shanbhag, Y. Wenzhuo, A. Mohamed, S.E. Chun, E. Weber, D. Blackwood, E. Lynch-Bell, J. Gulakowski, C. Smith, and D. Humphreys. An aqueous electrolyte, sodium ion functional, large format energy storage device for stationary applications. Journal of Power Sources, 213:255–264, September 2012. [23] USABC. Development of active materials for cells used in automotive traction applications. Technical report, United States Advanced Battery Consortium, 2015. 106 [24] Peter Faguy. Overview of the doe advanced battery r&d program. Techni- cal report, U. S. Department of Energy, 2015. [25] Shiliang Zhou, Graham King, David O. Scanlon, Moulay T. Sougrati, and Brent C. Melot. Low temperature preparation and electrochemical prop- erties of lifesi2o6. Journal of The Electrochemical Society, 161(10):A1642– A1647, January 2014. [26] Shiliang Zhou, Wolfgang G. Zeier, Moureen C. Kemei, Moulay T. Sougrati, Matthew Mecklenburg, and Brent C. Melot. Hydrothermal preparation and magnetic properties of nafesi2o6: Nanowires vs bulk samples. Inorg. Chem., 53(23):12396–12401, December 2014. [27] Jean-Marie Tarascon. Is lithium the new gold? Nat Chem, 2(6):510–510, June 2010. [28] K. T. Lee, T. N. Ramesh, F. Nan, G. Botton, and L. F. Nazar. Topochemical synthesis of sodium metal phosphate olivines for sodium-ion batteries. Chem. Mater., 23:3593–3600, 2011. [29] Sung-Wook Kim, Dong-Hwa Seo, Xiaohua Ma, Gerbrand Ceder, and Kisuk Kang. Electrode materials for rechargeable sodium-ion batteries: Potential alternatives to current lithium-ion batteries. Adv. Energy Mater., 2(7):710– 721, 2012. [30] Tsuyoshi Sasaki, Yoshio Ukyo, and Petr NovÃ ˇ CÂ ˛ ak. Memory effect in a lithium-ion battery. Nat Mater, 12(6):569–575, June 2013. [31] Byoungwoo Kang and Gerbrand Ceder. Battery materials for ultrafast charging and discharging. Nature, 458(7235):190–3, March 2009. [32] Donald W. Murphy and Forrest A. Trumbore. The chemistry of tis3 and nbse3 cathodes. Journal of The Electrochemical Society, 123(7):960–964, 1976. [33] Naoki Nitta, Feixiang Wu, Jung Tae Lee, and Gleb Yushin. Li-ion bat- tery materials: present and future. Materials Today, 18(5):252–264, June 2015. [34] Cyrus Umrigar, D. E. Ellis, Ding-Sheng Wang, H. Krakauer, and M. Poster- nak. Band structure, intercalation, and interlayer interactions of transition-metal dichalcogenides: Tis 2 and litis 2 . Phys. Rev. B, 26(9):4935– 4950, November 1982. [35] M. S. Whittingham. Electrical energy storage and intercalation chemistry. Science, 192:1126–1127, 1976. 107 [36] M.S. Whittingham. Lithium batteries and cathode materials. 104(10):4271–4301, 2004. [37] K. Mizushima, P. C. Jones, P. J. Wiseman, and J. B. Goodenough. Li x CoO 2 (0<x<-1): A new cathode material for batteries of high energy density. Mater. Res. Bull., 15:783 – 789, 1980. [38] T. Nagaura and K. Tozawa. Lithium ion rechargeable battery. Progress in Batteries & Solar Cells, 9:209–217, 1990. [39] Kazunori Ozawa. Lithium-ion rechargeable batteries with licoo2 and car- bon electrodes: the licoo2/c system. Solid State Ionics, 69(3):212–221, 1994. [40] M. Holzapfel, C. Haak, and A. Ott. Lithium-ion conductors of the system lico 1−x fe x o 2 , preparation and structural investigation. Journal of Solid State Chemistry, 156(2):470–479, 2001. [41] Qianyu Zhang, Michael G. Verde, Joon Kyo Seo, Xi Li, and Y. Shirley Meng. Structural and electrochemical properties of gd-doped li4ti5o12 as anode material with improved rate capability for lithium-ion batteries. Journal of Power Sources, 280:355–362, April 2015. [42] O. V. Yakubovich, E. L. Belokoneva, V. G. Tsirelson, and V. S. Urusov. Elec- tron density distribution in synthetic triphylite lifepo4. Moscow University Geology Bulletin, 45(2):93–99, 1990. [43] D. Zhang, B. N. Popov, Y. M. Podrazhansky, P. Arora, and R. E. White. Cobalt doped chromium oxides as cathode materials for secondary lithium batteries. Journal of Power Sources, 83(1-2):121–127, October 1999. [44] C. G. Chen, Y. P. Liu, and L. Li. Development of vanadium oxide in lithium ion batteries. Journal of Inorganic Materials, 19(6):1225–1230, November 2004. [45] X. P. Fang, X. Lu, X. W. Guo, Y. Mao, Y. S. Hu, J. Z. Wang, Z. X. Wang, F. Wu, H. K. Liu, and L. Q. Chen. Electrode reactions of manganese oxides for secondary lithium batteries. Electrochemistry Communications, 12(11):1520–1523, November 2010. [46] H. Kobayashi, M. Hibino, Y. Ogasawara, K. Yamaguchi, T. Kudo, S. Okuoka, K. Yonehara, H. Ono, Y. Sumida, M. Oshima, and N. Mizuno. Improved performance of co-doped li2o cathodes for lithium-peroxide batteries using licoo2 as a dopant source. Journal of Power Sources, 306:567–572, February 2016. 108 [47] M. Dolle, J. Hollingsworth, T. J. Richardson, and M. M. Doeff. Investiga- tion of layered intergrowth lixmymn1-yo2+z (m=ni, co, al) compounds as positive electrodes for li-ion batteries. Solid State Ionics, 175(1-4):225– 228, November 2004. [48] J. R. Liu, M. Wang, D. C. Yin, and W. D. Huang. Low-temperature syn- thesis of liv3o8 as cathode material for rechargeable lithium-ion batteries. Journal of Inorganic Materials, 17(3):617–620, May 2002. [49] Y. L. Ding, B. M. Goh, H. Zhang, K. P. Loh, and L. Lu. Single-crystalline nanotubes of spinel lithium nickel manganese oxide with lithium titanate anode for high-rate lithium ion batteries. Journal of Power Sources, 236:1– 9, August 2013. [50] R. Alcantara, P. Lavela, J. L. Tirado, E. Zhecheva, and R. Stoyaneva. Recent advances in the study of layered lithium transition metal oxides and their application as intercalation electrodes. Journal of Solid State Electrochemistry, 3(3):121–134, April 1999. [51] T. F. Yi, Y. R. Zhu, X. D. Zhu, J. Shu, C. B. Yue, and A. N. Zhou. A review of recent developments in the surface modification of limn2o4 as cathode material of power lithium-ion battery. Ionics, 15(6):779–784, December 2009. [52] H. Liu, Y. P. Wu, E. Rahm, R. Holze, and H. Q. Wu. Cathode materials for lithium ion batteries prepared by sol-gel methods. Journal of Solid State Electrochemistry, 8(7):450–466, June 2004. [53] F. F. C. Bazito and R. M. Torresi. Cathodes for lithium ion batteries: The benefits of using nanostructured materials. Journal of the Brazilian Chem- ical Society, 17(4):627–642, July 2006. [54] J. W. Fergus. Recent developments in cathode materials for lithium ion batteries. Journal of Power Sources, 195(4):939–954, February 2010. [55] M. Hu, X. L. Pang, and Z. Zhou. Recent progress in high-voltage lithium ion batteries. Journal of Power Sources, 237:229–242, September 2013. [56] S. T. Myung, K. Amine, and Y. K. Sun. Nanostructured cathode materials for rechargeable lithium batteries. Journal of Power Sources, 283:219–236, June 2015. [57] W. Liu, P. Oh, X. Liu, M. J. Lee, W. Cho, S. Chae, Y. Kim, and J. Cho. Nickel- rich layered lithium transition-metal oxide for high-energy lithium-ion batteries. Angewandte Chemie-international Edition, 54(15):4440–4457, April 2015. 109 [58] Z. Z. Pei, Z. G. Li, and X. L. Zheng. Porous materials for lithium-ion batteries. Journal of Nanoscience and Nanotechnology, 16(9):9028–9049, September 2016. [59] A. K. Padhi, K. S. Nanjundaswamy, and J. B. Goodenough. Phospho- olivines as positive-electrode materials for rechargeable lithium batteries. J. Electrochem. Soc., 144(4):1188–1194, April 1997. [60] N. Ravet, Y. Chouinard, J. F. Magnan, S. Besner, M. Gauthier, and M. Armand. Electroactivity of natural and synthetic triphylite. J. Power Sources, 97–98:503–507, 2001. [61] Christian Masquelier and Laurence Croguennec. Polyanionic (phosphates, silicates, sulfates) frameworks as electrode materials for rechargeable li (or na) batteries. Chem. Rev., 113(8):6552–6591, August 2013. [62] Vinodkumar Etacheri, Rotem Marom, Ran Elazari, Gregory Salitra, and Doron Aurbach. Challenges in the development of advanced li-ion batter- ies: a review. Energy Environ. Sci., 4(9):3243–3262, 2011. [63] B. C. Melot and J.-M. Tarascon. Design and preparation of materials for advanced electrochemical storage. Acc. Chem. Res., 46:1226–1238, 2013. [64] L. Adam, A. Guesdon, and B. Raveau. A new lithium manganese phos- phate with an original tunnel structure in the a2mp2o7 family. J. Solid State Chem., 181(11):3110–3115, November 2008. [65] H. Zhou, S. Upreti, N. A. Chernova, G. Hautier, G. Ceder, and M. S. Whit- tingham. Iron and manganese pyrophosphates as cathodes for lithium-ion batteries. Chem. Mater., 23(2):293–300, December 2010. [66] H. Kim, S. Lee, Y.-U. Park, H. Kim, J. Kim, S. Jeon, and K. Kang. Neutron and x-ray diffraction study of pyrophosphate-based li 2 mp 2 o 7 (m = fe, co) for lithium rechargeable battery electrodes. Chem. Mater., 23(17):3930– 3937, August 2011. [67] V. Legagneur, Y. An, A. Mosbah, R. Portal, A. Le Gal La Salle, A. Verbaere, D. Guyomard, and Y. Piffard. Limbo 3 (m=mn, fe, co):: synthesis, crystal structure and lithium deinsertion/insertion properties. Solid State Ionics, 139(1é ´ Lˇ e?):37–46, January 2001. [68] Y.Z. Dong, Y.M. Zhao, P. Fu, H. Zhou, and X.M. Hou. Phase relations of li 2 o-feo-b 2 o 3 ternary system and electrochemical properties of lifebo 3 compound. J. Alloys Compd., 461(1?):585–590, August 2008. 110 [69] A. Yamada, N. Iwane, S. Nishimura, Y. Koyama, and I. Tanaka. Synthesis and electrochemistry of monoclinic li(mnxfe1-x)bo3: a combined experi- mental and computational study. J. Mater. Chem., 21(29):10690–10696, 2011. [70] S. Afyon, D. Kundu, F. Krumeich, and R. Nesper. Nano limnbo 3 , a high-capacity cathode material for li-ion batteries. J. Power Sources, 224(0):145–151, February 2013. [71] S. Li, L. Xu, G. Li, M. Wang, and Y. Zhai. In-situ controllable synthesis and performance investigation of carbon-coated monoclinic and hexago- nal limnbo 3 composites as cathode materials in lithium-ion batteries. J. Power Sources, 236(0):54–60, August 2013. [72] H. Chen, G. Hautier, A. Jain, C. Moore, B. Kang, R. Doe, L. Wu, Y. Zhu, Y. Tang, and G. Ceder. Carbonophosphates: A new family of cathode materials for li-ion batteries identified computationally. Chem. Mater., 24(11):2009–2016, May 2012. [73] P. Barpanda, N. Recham, J-N. Chotard, K. Djellab, W. Walker, M. Armand, and J-M. Tarascon. Structure and electrochemical properties of novel mixed li(fe1-xmx)so4f (m = co, ni, mn) phases fabricated by low temper- ature ionothermal synthesis. J. Mater. Chem., 20(9):1659–1668, 2010. [74] R. Tripathi, T. N. Ramesh, B.L. Ellis, and L. F. Nazar. Scalable synthesis of tavorite lifeso4f and nafeso4f cathode materials. Angew. Chem. Int. Ed., 49(46):8738–8742, November 2010. [75] P. Barpanda, M. Ati, B. C. Melot, G. Rousse, J-N. Chotard, M-L. Doublet, M. T. Sougrati, S. A. Corr, J-C. Jumas, and J-M. Tarascon. A 3.90 v iron- based fluorosulphate material for lithium-ion batteries crystallizing in the triplite structure. Nat. Mater., 10(10):772–779, October 2011. [76] M. Ati, M. Sathiya, S. Boulineau, M. Reynaud, A. Abakumov, G. Rousse, B. C. Melot, G. Van Tendeloo, and J-M. Tarascon. Understanding and promoting the rapid preparation of the triplite-phase of lifeso4f for use as a large-potential fe cathode. J. Am. Chem. Soc., 134(44):18380–18387, October 2012. [77] R. Tripathi, G. Popov, B. L. Ellis, A. Huq, and L. F. Nazar. Lithium metal flu- orosulfate polymorphs as positive electrodes for li-ion batteries: synthetic strategies and effect of cation ordering. Energy Environ. Sci., 5(3):6238– 6246, 2012. 111 [78] M. Reynaud, M. Ati, B. C. Melot, M. T. Sougrati, G. Rousse, J-N. Chotard, and J-M. Tarascon. Li 2 fe(so 4 ) 2 as a 3.83 v positive electrode material. Electrochem. Commun., 21(0):77–80, July 2012. [79] R. Tripathi, G. Popov, X. Sun, D. H. Ryan, and L. F. Nazar. Ultra-rapid microwave synthesis of triplite lifeso 4 f. J. Mater. Chem. A, 1(9):2990– 2994, 2013. [80] G. Rousse and J. M. Tarascon. Sulfate-based polyanionic compounds for li-ion batteries: Synthesis, crystal chemistry, and electrochemistry aspects. Chem. Mater., 26:394–406, 2014. [81] R. Dominko, M. Bele, M. Gaberšˇ cek, A. Meden, M. Remškar, and J. Jamnik. Structure and electrochemical performance of li 2 mnsio 4 and li 2 fesio 4 as potential li-battery cathode materials. Electrochem. Commun., 8(2):217–222, February 2006. [82] K. Zaghib, A. Ait Salah, N. Ravet, A. Mauger, F. Gendron, and C.M. Julien. Structural, magnetic and electrochemical properties of lithium iron orthosilicate. J. Power Sources, 160(2):1381–1386, October 2006. [83] A. Nytén, S. Kamali, L. Haggstrom, T. Gustafsson, and J. O. Thomas. The lithium extraction/insertion mechanism in li 2 fesio 4 . J. Mater. Chem., 16(23):2266–2272, 2006. [84] P. Larsson, R. Ahuja, A. Nytén, and J. O. Thomas. An ab initio study of the li-ion battery cathode material li 2 fesio 4 . Electrochem. Commun., 8(5):797– 800, May 2006. [85] C. Sirisopanaporn, C. Masquelier, P. G. Bruce, A. R. Armstrong, and R. Dominko. Dependence of Li 2 FeSiO 4 electrochemistry on structure. J. Am. Chem. Soc., 133:1263, 2011. [86] M. S. Islam, R. Dominko, C. Masquelier, C. Sirisopanaporn, A. R. Arm- strong, and P. G. Bruce. Silicate cathodes for lithium batteries: alterna- tives to phosphates? J. Mater. Chem., 21(27):9811–9818, 2011. [87] A. R. Armstrong, N. Kuganathan, M. S. Islam, and P. G. Bruce. Structure and lithium transport pathways in li2fesio4 cathodes for lithium batteries. J. Am. Chem. Soc., 133(33):13031–13035, July 2011. [88] S. R. Taylor. Abundance of chemical elements in the continental crust: a new table. Geochim. Cosmochim. Acta, 28:1273, 1964. 112 [89] S. Jodlauk, P. Becker, J. A. Mydosh, D. I. Khomskii, T. Lorenz, S. V. Streltsov, D. C. Hezel, and L. Bohatý. Pyroxenes: a new class of multi- ferroics. J. Phys.: Condens. Matter, 19(43):432201, October 2007. [90] A. Siegel, P. Piekarz, and K. Parlinski. Structural phase transition in lifesi 2 o 6 from ab initio calculations. J. Phys.: Condens. Matter, 24(19):195401, May 2012. [91] J. R. Clark, D. E. Appleman, and J. J. Paplike. Crystal-chemical characteri- zation of clinopyroxenes based on eight new structure refinements. Miner. Soc. Am. S., 2:31–50, 1969. [92] M. Behruzi, T. Hahn, C. T. Prewitt, and K. Baldwin. Low-temperature and high-temperature crystal-structures of lifege2o6, lifesi2o6 and licrsi2o6. Acta Crystallogr., Sect. A, 40:C247–C247, 1984. [93] E. Baum, W. Treutmann, M. Behruzi, W. Lottermoser, and G. Amthauer. Structural and magnetic-properties of the clinopyroxenes NaFeSi 2 O 6 and LiFeSi 2 O 6 . Z. Kristallogr., 183:273–284, 1988. [94] W. Lottermoser, G. Redhammer, K. Forcher, G. Amthauer, W. Paulus, G. Andre, and W. Treutmann. Single crystal mossbauer and neutron pow- der diffraction measurements on the synthetic clinopyroxene li-acmite lifesi2o6. Z. Kristallogr., 213(2):101–107, 1998. [95] G. J. Redhammer, G. Roth, W. Paulus, G. André, W. Lottermoser, G. Amthauer, W. Treutmann, and B. Koppelhuber-Bitschnau. The crys- tal and magnetic structure of li-aegirine life3+si2o6: a temperature- dependent study. Phys. Chem. Miner., 28(5):337–346, 2001. [96] G. J. Redhammer and G. Roth. Structural variations in the aegirine solid- solution series (na,li)fesi 2 o 6 at 298 k and 80 k. Z. Kristallogr., 217(2):63– 72, 2002. [97] M. Zhang, G. J. Redhammer, E. K. H. Salje, and M. Mookherjee. Lifesi 2 o 6 and nafesi 2 o 6 at low temperatures: an infrared spectroscopic study. Phys. Chem. Miner., 29(9):609–616, October 2002. [98] G. J. Redhammer and G. Roth. Structural changes upon the temperature dependent c2/c -> p2(1)/c phase transition in lime3+si2o6 clinopyrox- enes, me = cr, ga, fe, v, sc and in. Z. Kristallogr., 219(10):585–605, 2004. [99] G.Diego Gatta, TizianaBoffa Ballaran, and Gianluca Iezzi. High-pressure x-ray and raman study of a ferrian magnesian spodumene. Phys. Chem. Miner., 32(2):132–139, 2005. 113 [100] C. J. S. Pommier, R. T. Downs, M. Stimpfl, G. J. Redhammer, and M. B. Denton. Raman and x-ray investigations of lifesi2o6 pyroxene under pres- sure. J. Raman Spectrosc., 36(9):864–871, 2005. [101] G. J. Redhammer, G. Roth, W. Treutmann, M. Hoelzel, W. Paulus, G. André, C. Pietzonka, and G. Amthauer. The magnetic structure of clinopyroxene-type lifege2o6 and revised data on multiferroic lifesi2o6. J. Solid State Chem., 182(9):2374 – 2384, 2009. [102] P. J. Baker, H. J. Lewtas, S. J. Blundell, T. Lancaster, I. Franke, W. Hayes, F. L. Pratt, L. Bohaty, and P. Becker. Muon-spin relaxation and heat capacity measurements on the magnetoelectric and multiferroic pyrox- enes lifesi 2 o 6 and nafesi 2 o 6 . Phys. Rev. B, 81(21):214403, June 2010. [103] G. J. Redhammer, A. Senyshyn, M. Meven, G. Roth, S. Prinz, A. Pach- ler, C. Tippelt, G.and Pietzonka, W. Treutmann, M. Hoelzel, B. Pedersen, and G. Amthauer. Nuclear and incommensurate magnetic structure of nafege2o6 between 5 k and 298 k and new data on multiferroic nafesi 2 o 6 . Phys. Chem. Miner., 38(2):139–157, February 2011. [104] C. Lee, J. Kang, J. Hong, J. H. Shim, and M.-H. Whangbo. Analy- sis of the difference between the pyroxenes lifesi2o6 and lifege2o6 in their spin order, spin orientation, and ferrotoroidal order. Chem. Mater., 26(4):1745–1750, 2014. [105] G. Amthauer, W. Lottermoser, G. Redhammer, and G. Tippelt. Mossbauer studies of selected synthetic silicates. Hyperfine Interact., 113(4):219– 248, 1998. [106] H. M. Rietveld. A profile refinement method for nuclear and magnetic structures. J. Appl. Cryst., 2:65, 1969. [107] A. C. Larson and R. B. Von Dreele. General structure analysis system (gsas). Los Alamos National Laboratory Report LAUR 86-748 (2000), 86:748, 1994. [108] G. Kresse and J. Hafner. Ab initio molecular-dynamics simulation of the liquid-metal–amorphous-semiconductor transition in germanium. Phys. Rev. B, 49:14251–14271, 1994. [109] G. Kresse and J. Furthmuller. Efficiency of Ab-Initio Total Energy Calcula- tions for Metals and Semiconductors Using a Plane-Wave Basis Set. Comp. Mater. Sci., 6:15–50, 1996. 114 [110] P. E. Blöchl. Projector augmented-wave method. Phys. Rev. B, 50:17953, 1994. [111] J. P. Perdew, K. Burke, and M. Ernzerhof. Generalized gradient approxi- mation made simple. Phys. Rev. Lett., 77:3865–3868, 1996. [112] S. L. Dudarev, G. A. Botton, S. Y. Savrasov, C. J. Humphreys, and A. P. Sutton. Electron-energy-loss spectra and the structural stability of nickel oxide: an lsda+u study. Phys. Rev. B, 57:1505–1509, Jan 1998. [113] J. P. Allen, D. O. Scanlon, and G. W. Watson. Electronic Structure of Mixed-Valence Silver Oxide AgO from Hybrid Density-Functional Theory. Phys. Rev. B, 81:161103(R), 2010. [114] D. O. Scanlon, B. J. Morgan, and G. W. Watson. The origin of the enhanced oxygen storage capacity of Ce 1−x (Pd/Pt) x O 2 . Phys. Chem. Chem. Phys., 13:4279–4284, 2011. [115] D. O. Scanlon and G. W. Watson. Understanding the p-type defect chem- istry of CuCrO 2 . J. Mater. Chem., 21:3655–3663, 2011. [116] W. A. DOLLASE and W. I. GUSTAFSON. Fe-57 mossbauer spectral-analysis of the sodic clinopyroxenes. Am. Mineral., 67(3-4):311–327, 1982. [117] G. Ceder, M.K. Aydinol, and A.F. Kohan. Application of first-principles calculations to the design of rechargeable li-batteries. Comp. Mater. Sci., 8(1é ´ Lˇ eï£¡):161–169, May 1997. [118] W. Eerenstein, N. D. Mathur, and J. F. Scott. Multiferroic and magneto- electric materials. Nature, 442(7104):759–765, August 2006. [119] NL Bowen, JF Schairer, and HWV Willems. The ternary system: Na 2 sio 3 - fe 2 o 3 -sio 2 . Am. J. Sci, 20:405–455, 1930. [120] J. Rodríguez-Carvajal. Recent advances in magnetic structure determina- tion by neutron powder diffraction. Physica B, 192:55–69, 1993. [121] V. S. Rusakov, O. E. Yagova, N. A. Kuz’mina, and A. M. Bychkov. Mossbauer study of the low-temperature hydrothermal crystallization of acmite, nafesi 2 o 6 . Geokhimiya, (4):404–409, April 1999. [122] G. J. Redhammer, G. Amthauer, G. Roth, G. Tippelt, and W. Lottermoser. Single-crystal x-ray diffraction and temperature dependent fe-57 moss- bauer spectroscopy on the hedenbergite-aegirine (ca,na)(fe 2+ ,fe 3+ )si 2 o 6 solid solution. Am. Mineral., 91(8-9):1271–1292, August 2006. 115 [123] G. J. Redhammer, G. Amthauer, W. Lottermoser, and W. Treutmann. Eur. J. Mineral., 12:105–, 2000. [124] J. Nolan and A. D. Edgar. Mineral. Mag., 33:625–, 1963. [125] J. Nolan. Mineral. Mag., 37:216–, 1969. [126] Robert T. Downs and A. K. Singh. Analysis of deviatoric stress from non- hydrostatic pressure on a single crystal in a diamond anvil cell: The case of monoclinic aegirine, nafesi 2 o 6 . J. Phys. Chem. Solids, 67(9- 10):Consortium Mat Property Res Earth Sci; High Pressure Collaborat AccessEOLEOLTeam; Florida Int Univ, Coll Engn, September 2006. [127] Coralie Weigel, Laurent Cormier, Laurence Galoisy, Georges Calas, Daniel Bowron, and Brigitte Beuneu. Determination of fe 3+ sites in a nafesi 2 o 6 glass by neutron diffraction with isotopic substitution coupled with numerical simulation. Applied Physics Letters, 89(14):141911, October 2006. [128] G. J. Redhammer, G. Tippelt, G. Amthauer, and G. Roth. Structural and fe-57 mossbauer spectroscopic characterization of the synthetic nafesi 2 o 6 (aegirine) - camgsi 2 o 6 (diopside) solid solution series. Z. Kristallogr., 227(7):396–410, 2012. [129] E. Schmidbauer and T. Kunzmann. Electrical conductivity, thermopower and fe-57 mossbauer spectroscopy of aegirine (nafesi 2 o 6 ). Physics and Chemistry of Minerals, 31(2):102–114, March 2004. [130] E. S. Grew, D. I. Belakovskiy, M. E. Fleet, M. G. Yates, J. J. Mcgee, and N. Marquez. Eur. J. Mineral., 5:971–, 1993. [131] P. L. Mcswiggen, G. B. Morey, and J. M. Cleland. Can. Mineral., 32:589–, 1994. [132] Tore Prestvik and Calvin G. Barnes. A new occurrence of aegirine in nor- way. Norwegian Journal of Geology, 87(4):451–456, 2007. [133] Bruno Mettout, Pierre Toledano, and Manfred Fiebig. Symmetry replica- tion and toroidic effects in the multiferroic pyroxene nafesi 2 o 6 . Physical Review B, 81(21):214417, June 2010. [134] Ingyu Kim, Byung-Gu Jeon, Deepak Patil, Seema Patil, Gwilherm Nenert, and Kee Hoon Kim. Observation of multiferroic properties in pyroxene nafege 2 o 6 . Journal of Physics-condensed Matter, 24(30):306001, August 2012. 116 [135] O. BALLET, J. M. D. COEY, G. FILLION, A. GHOSE, A. HEWAT, and J. R. REGNARD. Magnetic order in acmite - nafesi2o6. Physics and Chemistry of Minerals, 16(7):672–677, 1989. [136] C. Frondel and J. Ito. Am. Mineral., 51:1406–, 1966. [137] M. Ati, M. Sathiya, S. Boulineau, M. Reynaud, A. Abakumov, G. Rousse, B. Melot, G. Van Tendeloo, and J.-M. Tarascon. Understanding and pro- moting the rapid preparation of the triplite-phase of lifeso4f for use as a large-potential fe cathode. J. Am. Chem. Soc., 134:18380–, 2012. [138] G. Lawes, A. B. Harris, T. Kimura, N. Rogado, R. J. Cava, A. Aharony, O. Entin-Wohlman, T. Yildirim, M. Kenzelmann, C. Broholm, and A. P. Ramirez. Magnetically driven ferroelectric order in Ni 3 V 2 O 8 . Phys. Rev. Lett., 95:087205, 2005. [139] B. C. Melot, J. E. Drewes, R. Seshadri, E. M. Stoudenmire, and A. P. Ramirez. Magnetic phase evolution in the spinel compounds Zn 1−x Co x Cr 2 O 4 . J. Phys.: Condens. Matter, 21:216007, 2009. [140] B. C. Melot, B. Paden, R. Seshadri, E. Suard, G. Nénert, A. Dixit, and G. Lawes. Magnetic structure and susceptibility of CoSe 2 O 5 : An antifer- romagnetic chain compound. Phys. Rev. B, 82:014411, 2010. [141] H. C. Ko, M. J. Ferrante, and J. M. Stuve. Proceedings of the Seventh Symposium on Thermophysical Properties. 1977. [142] M. E. Fisher. Proc. R. Soc. London Ser. A, 254:66–, 1960. [143] JÃ ˇ CÂ˝ urg Neuefeind, Mikhail Feygenson, John Carruth, Ron Hoffmann, and Kenneth K. Chipley. The nanoscale ordered materials diffractome- ter nomad at the spallation neutron source sns. Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, 287:68–75, September 2012. [144] R. W. Cheary and A. Coelho. A fundamental parameters approach to x- ray line-profile fitting. Journal of Applied Crystallography, 25(2):109–121, 1992. [145] Stephens and Peter. Phenomenological model of anisotropic peak broadening in powder diffraction. Journal of Applied Crystallography, 32(2):281–289, 1999. [146] A. Baumgartner, C. Butterhof, S. Koch, R. Mariychuk, and J. Breu. Melt synthesis and characterization of synthetic mn-rich tainiolite. Clays and Clay Minerals, 57(2):271–277, April 2009. 117 [147] M. F. Brigatti and S. Guggenheim. Mica crystal chemistry and the influ- ence of pressure, temperature, and solid solution on atomistic models. Micas: Crystal Chemistry and Metamorphic Petrology, 46:1–97, 2002. [148] R. Mariychuk, A. Baumgartner, F. E. Wagner, A. Lerf, A. Dubbe, R. Moos, and J. Breu. Synthesis, structure, and electric conductivity of ferrous tain- iolite and its oxidative conversion into coarse-grained swellable smectite. Chemistry of Materials, 19(22):5377–5387, October 2007. [149] G. J. Redhammer, G. Amthauer, W. Lottermoser, M. Bernroider, G. Tippelt, and G. Roth. X-ray powder diffraction and fe-57-mossbauer spectroscopy of synthetic trioctahedral micas K[me-3](tsi3)o-10(oh)(2), me = ni2+, mg2+, co2+, fe2+; t = al3+, fe3+. Mineralogy and Petrology, 85(1- 2):89–115, November 2005. [150] Q. Williams, E. Knittle, H. P. Scott, and Z. X. Liu. The high-pressure behav- ior of micas: Vibrational spectra of muscovite, biotite, and phlogopite to 30 gpa. American Mineralogist, 97(1):241–252, January 2012. [151] B. Wunder and S. Melzer. Interlayer vacancy characterization of synthetic phlogopitic micas by ir spectroscopy. European Journal of Mineralogy, 14(6):1129–1138, November 2002. [152] A. P. Zhukhlistov. Crystal structure of celadonite from the electron diffrac- tion data. Crystallography Reports, 50(6):902–906, November 2005. [153] W. T. REICHLE. Catalytic reactions by thermally activated, synthetic, anionic clay-minerals. Journal of Catalysis, 94(2):547–557, 1985. [154] Noboru Oyama and Fred C. Anson. Catalysis of the electroreduction of hydrogen peroxide by montmorillonite clay coatings on graphite elec- trodes. Journal of Electroanalytical Chemistry and Interfacial Electrochem- istry, 199(2):467–470, 1986. [155] P. Laszlo. Heterogeneous catalysis of organic reactions. Journal of Physical Organic Chemistry, 11(5):356–361, May 1998. [156] Z. Ding, J. T. Kloprogge, R. L. Frost, G. Q. Lu, and H. Y. Zhu. Porous clays and pillared clays-based catalysts. part 2: A review of the catalytic and molecular sieve applications. Journal of Porous Materials, 8(4):273–293, 2001. [157] J. P. Ferris. Mineral catalysis and prebiotic synthesis: Montmorillonite- catalyzed formation of rna. Elements, 1(3):145–149, June 2005. 118 [158] K. G. Bhattacharyya and S. Sen Gupta. Adsorption of a few heavy met- als on natural and modified kaolinite and montmorillonite: A review. Advances In Colloid and Interface Science, 140(2):114–131, August 2008. [159] M. D. Subramaniam and I. H. Kim. Clays as dietary supplements for swine: A review. Journal of Animal Science and Biotechnology, 6:38, August 2015. [160] K. Shimizu and A. Satsuma. Toward a rational control of solid acid catal- ysis for green synthesis and biomass conversion. Energy & Environmental Science, 4(9):3140–3153, September 2011. [161] Pushpito K. Ghosh and Allen J. Bard. Clay-modified electrodes. J. Am. Chem. Soc., 105(17):5691–5693, August 1983. [162] Walter E. Rudzinski and Allen J. Bard. Clay modified electrodes. Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, 199(2):323– 340, 1986. [163] Randal D. King, Daniel G. Nocera, and Thomas J. Pinnavaia. On the nature of electroactive sites in clay-modified electrodes. Journal of Electroanalyt- ical Chemistry and Interfacial Electrochemistry, 236(1):43–53, 1987. [164] Michael T. Carter and Allen J. Bard. Clay modified electrodes. Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, 229(1):191– 214, 1987. [165] Gilles Villemure and Alien J. Bard. Clay modified electrodes. Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, 282(1):107– 121, 1991. [166] Christopher A. Gorski, Michael Aeschbacher, Daniela Soltermann, Andreas Voegelin, Bart Baeyens, Maria Marques Fernandes, Thomas B. Hofstetter, and Michael Sander. Redox properties of structural fe in clay minerals. 1. electrochemical quantification of electron-donating and -accepting capacities of smectites. Environ. Sci. Technol., 46(17):9360– 9368, September 2012. [167] Christopher A. Gorski, Laura KlÃ ˇ CÂijpfel, Andreas Voegelin, Michael Sander, and Thomas B. Hofstetter. Redox properties of structural fe in clay minerals. 2. electrochemical and spectroscopic characterization of electron transfer irreversibility in ferruginous smectite, swa-1. Environ. Sci. Technol., 46(17):9369–9377, September 2012. [168] Christopher A. Gorski, Laura E. KlÃ ˇ CÂijpfel, Andreas Voegelin, Michael Sander, and Thomas B. Hofstetter. Redox properties of structural fe in clay 119 minerals: 3. relationships between smectite redox and structural proper- ties. Environ. Sci. Technol., 47(23):13477–13485, December 2013. [169] Jun Xiao and Gilles Villemure. Preparation, characterization and electro- chemistry of synthetic copper clays. Clays Clay Miner., 46:195–203, 1998. [170] Bradley Scott, Jeff Crouse, Maria Correia, Laisheng Sun, and Gilles Ville- mure. Preparation of well-ordered hexagonal particles of synthetic cobalt smectites containing electrochemically active cobalt(ii) sites. Applied Clay Science, 48(1Ã´ câ ´ Cˇ nâ ˘ AIJ2):46–54, March 2010. [171] G. Donnay, N. Morimoto, and H. Takeda. Trioctahedral one-layer micas. i. crystal structure of a synthetic iron mica. Acta Crystallographica, 17(11):1369–1373, 1964. [172] S. B. Hendricks and M. E. Jefferson. Polymorphism of the micas with optical measurements. Amer. Min., 24:729, 1939. [173] J. v. Smith and H. S. Yoder. Experimental and theoretical studies of the mica polymorphs. Mineral Mag., 31:209, 1956. [174] M. ROSS, H. TAKEDA, and D. R. WONES. Mica polytypes - systematic description and identification. Science, 151(3707):191–+, 1966. [175] H. F. XU and D. R. VEBLEN. Periodic and nonperiodic stacking in biotite from the bingham canyon porphyry copper-deposit, utah. Clays and Clay Minerals, 43(2):159–173, April 1995. [176] Y. F. Wang and H. F. Xu. Geochemical chaos: Periodic and nonperiodic growth of mixed-layer phyllosilicates. Geochimica Et Cosmochimica Acta, 70(8):1995–2005, April 2006. [177] Armel Le Bail. Whole powder pattern decomposition methods and appli- cations: A retrospection. 20(4):316–326–, 2005. [178] Rebecca W. Smaha, John H. Roudebush, Jake T. Herb, Elizabeth M. Seibel, Jason W. Krizan, Gary M. Fox, Qingzhen Huang, Craig B. Arnold, and Robert J. Cava. Tuning sodium ion conductivity in the layered honeycomb oxide na3Ã´ câ ´ Cˇ nâ ˘ AIJxsn2Ã´ câ ´ Cˇ nâ ˘ AIJxsbxnao6. Inorg. Chem., 54(16):7985–7991, August 2015. [179] Jue Liu, Liang Yin, Lijun Wu, Jianming Bai, Seong-Min Bak, Xiqian Yu, Yimei Zhu, Xiao-Qing Yang, and Peter G. Khalifah. Quantification of hon- eycomb number-type stacking faults: Application to na3ni2bio6 cathodes for na-ion batteries. Inorg. Chem., 55(17):8478–8492, September 2016. 120 [180] C. R. Houska and B. E. Warren. X-ray study of the graphitization of carbon black. Journal of Applied Physics, 25(12):1503–1509, December 1954. [181] Huarui Xu and Lian Gao. New evidence of a dissolution-precipitation mechanism in hydrothermal synthesis of barium titanate powders. Mate- rials Letters, 57(2):490–494, December 2002. [182] Xue Qin, Xiaohui Wang, Huimin Xiang, Jie Xie, Jingjing Li, and Yanchun Zhou. Mechanism for hydrothermal synthesis of lifepo4 platelets as cath- ode material for lithium-ion batteries. J. Phys. Chem. C, 114(39):16806– 16812, October 2010. [183] J. H. Zhang, H. Y. Liu, Z. L. Wang, N. B. Ming, Z. R. Li, and A. S. Biris. Polyvinylpyrrolidone-directed crystallization of zno with tunable morphology and bandgap. Advanced Functional Materials, 17(18):3897– 3905, December 2007. [184] Y. J. Zhang and J. J. Lu. A mild and efficient biomimetic synthe- sis of rodlike hydroxyapatite particles with a high aspect ratio using polyvinylpyrrolidone as capping agent. Crystal Growth & Design, 8(7):2101–2107, July 2008. [185] T. Dudek, J. Srodon, D. D. Eberl, F. Elsass, and P. Uhlik. Thickness dis- tribution of illite crystals in shales. i: X-ray diffraction vs. high-resolution transmission electron microscopy measurements. Clays and Clay Minerals, 50(5):562–577, October 2002. [186] D. Ensling, M. Stjerndahl, A. Nyten, T. Gustafsson, and J. O. Thomas. A comparative xps surface study of li2fesio4/c cycled with litfsi- and lipf6- based electrolytes. Journal of Materials Chemistry, 19(1):82–88, 2009. [187] L. Niedzicki, G. Z. Zukowska, M. Bukowska, P. Szczecinski, S. Grugeon, S. Laruelle, M. Armand, S. Panero, B. Scrosati, M. Marcinek, and W. Wiec- zorek. New type of imidazole based salts designed specifically for lithium ion batteries. Electrochimica Acta, 55(4):1450–1454, January 2010. [188] Veronica Augustyn, Jérémy Come, Michael A. Lowe, Jong Woung Kim, Pierre-Louis Taberna, Sarah H. Tolbert, Hécctor D. Abruña, Patrice Simon, and Bruce Dunn. High-rate electrochemical energy storage through li+ intercalation pseudocapacitance. Nat Mater, 12(6):518–522, June 2013. [189] Kirstin Brezesinski, John Wang, Jan Haetge, Christian Reitz, Sven O. Stein- mueller, Sarah H. Tolbert, Bernd M. Smarsly, Bruce Dunn, and Torsten Brezesinski. Pseudocapacitive contributions to charge storage in highly ordered mesoporous group v transition metal oxides with iso-oriented 121 layered nanocrystalline domains. J. Am. Chem. Soc., 132(20):6982–6990, May 2010. [190] Myounggu Park, Xiangchun Zhang, Myoungdo Chung, Gregory B. Less, and Ann Marie Sastry. A review of conduction phenomena in li-ion bat- teries. Journal of Power Sources, 195(24):7904–7929, December 2010. [191] Yutaka Morikawa. Catalysis by metal ions intercalated in layer lattice silicates. volume Volume 39, pages 303–327. Academic Press, 1993. [192] Matthias Stöter, Sabine Rosenfeldt, and Josef Breu. Tunable exfoliation of synthetic clays. Annu. Rev. Mater. Res., 45(1):129–151, July 2015. [193] M. Armand and M. E. A. Y. de Dompablo. Benefits of n for o substitution in polyoxoanionic electrode materials: a first principles investigation of the electrochemical properties of li2fesio4-yny (y=0, 0.5, 1). Journal of Materials Chemistry, 21(27):10026–10034, 2011. [194] A. K. Padhi, V. Manivannan, and J. B. Goodenough. Tuning the position of the redox couples in materials with nasicon structure by anionic substi- tution. J. Electrochem. Soc., 145:1518–, 1998. [195] J. Shirakawa, M. Nakayama, M. Wakihara, and Y. Uchimoto. Changes in electronic structure upon li insertion reaction of monoclinic li3fe2(po4)3. J. Phys. Chem. B, 110:17743–, 2006. [196] A. Manthiram and J. B. Goodenough. Lithium insertion into fe2(mo4)3 frameworks: Comparison of m= w with m= mo. J. Solid State Chem., 71:349–, 1987. [197] A. Nadiri, C. Delmas, R. Salmon, and P. Hagenmuller. Chemical and electrochemical alkali metal intercalation in the 3d-framework of fe2(moo4)3. Chem. Informationsdienst, 16:537–, 1985. [198] P. G. Bruce and G. Miln. Sodium intercalation into the defect garnets fe2(moo4)3 and fe2(wo4)3. J. Solid State Chem., 89:162–, 1990. [199] Q. Sun, Q.-Q. Ren, and Z.-W. Fu. Nasicon-type fe2(moo4)3 thin film as cathode for rechargeable sodium ion battery. Electrochem. Commun., 23:145–, 2012. [200] J.-L. Yue, Y.-N. Zhou, S.-Q. Shi, Z. Shadike, X.-Q. Huang, J. Luo, Z.-Z. Yang, H. Li, L. Gu, X.-Q. Yang, and Z.-W. Fu. Discrete li-occupation versus pseudo-continuous na-occupation and their relationship with structural change behaviors in fe2(moo4)3. Sci. Rep., 5:8810–, 2015. 122 [201] C. Peng, L. Gao, S. Yang, and J. Sun. A general precipitation strategy for large-scale synthesis of molybdate nanostructures. Chem. Commun., 43:5601–, 2008. [202] M. Morcrette, Y. Chabre, G. Vaughan, G. Amatucci, J.-B. Leriche, S. Patoux, C. Masquelier, and J. M. Tarascon. In situ x-ray diffraction techniques as a powerful tool to study battery electrode materials. Elec- trochim. Acta, 47:3137–, 2002. [203] N. A. Caãs, K. Hirose, B. Pascucci, N. Wagner, K. A. Friedrich, and R. Hies- gen. Investigations of lithiumÃ´ câ ´ Cˇ nâ ˘ AIJsulfur batteries using electro- chemical impedance spectroscopy. Electrochim. Acta, 97:42–, 2013. [204] A. C. Larson and R. B. V. Dreele. General Structure Analysis System (GSAS); Los Alamos National Laboratory Report LAUR 86Ã´ câ ´ Cˇ nâ ˘ AIJ748. 2004. [205] C. L. Farrow, P. Juhas, J. W. Liu, D. Bryndin, E. S. BoÃ ˇ EÂ¿in, J. Bloch, T. Proffen, and S. J. L. Billinge. Pdffit2 and pdfgui: Computer pro- grams for studying nanostructure in crystals. J. Phys.: Condens. Matter, 19:335219–, 2007. [206] D. Orobengoa, C. Capillas, M. I. Aroyo, and J. M. Perez-Mato. Amplimodes: Symmetry-mode analysis on the bilbao crystallographic server. J. Appl. Crystallogr., 42:820–, 2009. [207] G. Kresse and J. Hafner. Norm-conserving and ultrasoft pseudopotentials for first-row and transition elements. J. Phys.: Condens. Matter, 6:8245–, 1994. [208] G. Kresse and J. Furthmüller. Efficient iterative schemes for ab initio total- energy calculations using a plane-wave basis set. Phys. Rev. B, 54:11169– 11186, 1996. [209] G. Kresse and D. Joubert. From ultrasoft pseudopotentials to the projector augmented-wave method. Phys. Rev. B, 59:1758–1775, 1999. [210] J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vydrov, G. E. Scuseria, L. A. Constantin, X. Zhou, and K. Burke. Restoring the density-gradient expan- sion for exchange in solids and surfaces. Phys. Rev. Lett., 100:136406–, 2008. [211] F. Zhou, C. A. Marianetti, M. Cococcioni, D. Morgan, and G. Ceder. Phase separation in li xfepo4 induced by correlation effects. Phys. Rev. B: Con- dens. Matter Mater. Phys., 69:201101–, 2004. 123 [212] H. Chen. The crystal structure and twinning behavior of ferric molybdate, fe2(moo4)3. Mater. Res. Bull., 14:1583–, 1979. [213] H. Tian, I. E. Wachs, and L. E. Briand. Comparison of uv and visible raman spectroscopy of bulk metal molybdate and metal canadate catalysts. J. Phys. Chem. B, 109:23491–, 2005. [214] N. S. McIntyre and D. G. Zetaruk. X-ray photoelectron spectroscopic stud- ies of iron oxides. Anal. Chem., 49:1521–, 1977. [215] A. Cimino and B. A. De Angelis. The application of x-ray photoelectron spectroscopy to the study of molybdenum oxides and supported molybde- num oxide catalysts. J. Catal., 36:11–, 1975. [216] P. Lu, C. Li, E. W. Schneider, and S. J. Harris. Chemistry, impedance, and morphology evolution in solid electrolyte interphase films during forma- tion in lithium ion batteries. J. Phys. Chem. C, 118:896–, 2014. [217] C. C. Torardi and E. Prince. Structure of the lithium insertion compound li2fe2(moo4)3 from neutron powder diffraction data. Mater. Res. Bull., 21:719–, 1986. [218] C. M. Burba and R. Frech. Raman and ftir spectroscopic study of lixfepo4(0 Ã´ cÂl’Â¡ xÃ´ cÂl’Â¡ 1). J. Electrochem. Soc., 151:A1032–, 2004. [219] B. C. Melot, D. O. Scanlon, M. Reynaud, G. Rousse, J.-N. Chotard, M. Henry, and J.-M. Tarascon. Chemical and structural indicators for large redox potentials in fe-based positive electrode materials. ACS Appl. Mater. Interfaces, 6:10832–, 2014. [220] B. L. Ellis, T. N. Ramesh, L. J. M. Davis, G. R. Goward, and L. F. Nazar. Structure and electrochemistry of two-electron redox couples in lithium metal fluorophosphates based on the tavorite structure. Chem. Mater., 23:5138–, 2011. [221] R. Tripathi, G. R. Gardiner, M. S. Islam, and L. F. Nazar. Alkali-ion conduc- tion paths in lifeso4f and nafeso4f tavorite-type cathode materials. Chem. Mater., 23:2278–, 2011. [222] S.-Y. Chung, J. T. Bloking, and Y.-M. Chiang. Electronically conductive phospho-olivines as lithium storage electrodes. Nat. Mater., 1:123–128, 2002. [223] D. A. Keen, M. G. Tucker, and M. T. Dove. Reverse monte carlo modelling of crystalline disorder. J. Phys.: Condens. Matter, 17:S15–, 2005. 124 [224] L. Karlsson. Mechanisms of ionic conduction in li2so4 and linaso4: Paddle wheel or percolation? Solid State Ionics, 76:301–, 1995. [225] D. A. Keen. Disordering phenomena in superionic conductors. J. Phys.: Condens. Matter, 14:R819–, 2002. [226] R.J. Cava, A. Santoro, D.W. Murphy, S. Zahurak, and R.S. Roth. The structures of lithium-inserted metal oxides: Lireo3 and li2reo3. Journal of Solid State Chemistry, 42(3):251–262, 1982. [227] Yuan Yi Chong and Wai Yip Fan. Facile synthesis of single crystalline rhenium (vi) trioxide nanocubes with high catalytic efficiency for pho- todegradation of methyl orange. Journal of Colloid and Interface Science, 397:18–23, May 2013. [228] Kanishka Biswas and C. N. R. Rao. Metallic reo3 nanoparticles. J. Phys. Chem. B, 110(2):842–845, January 2006. [229] H. Nechamkin, A. N. Kurtz, and C. F. Hiskey. A method for the preparation of rhenium (vi) oxide1. J. Am. Chem. Soc., 73(6):2828–2831, June 1951. [230] J.-E. Jørgensen, J. D. Jorgensen, B. Batlogg, J. P. Remeika, and J. D. Axe. Order parameter and critical exponent for the pressure-induced phase transitions in reo 3 . Phys. Rev. B, 33(7):4793–4798, April 1986. [231] Wemer H Baur. Bond length variation and distorted coordination polyhe- dra in inorganic crystals. Trans. Am. Crystallogr. Assoc., 6:129–155, 1970. [232] W. Weppner and R. A. Huggins. Determination of the kinetic parame- ters of mixed-conducting electrodes and application to the system li3sb. Journal of The Electrochemical Society, 124(10):1569–1578, 1977. 125

Abstract (if available)

Abstract The development of sustainable, robust and large-scale energy storage is critical for renewable energy sources to assume a major role in our energy supply scheme. Among the pool of technologies proposed for large-scale energy storage, rechargeable Li- and Na-ion batteries offer substantial advantages in terms of mobility, efficiency, and power density. However, for such technologies to be used in a grid scale, dramatic reduction in the cost is needed for them to become viable. And although Li-ion batteries have already enjoyed certain degree of success in electric vehicles, they still face challenges such as cost, and charging rate before they can become a dominant player in the automotive market. ❧ The work presented here aims to explore a large family of mineral materials, silicates, for their potential application in electrochemical storage and other functional applications. With a wide range of characterization tools, we set out to not only understand the electrochemical performance of these materials, but also find out how their functional properties are related to their structures. We then took a similar approach in two oxide phases, Fe₂(MoO₄)₃ and ReO₃, to further understand how the structural changes in these materials correlate with their elctrochemical properties. ❧ First we found that LiFeSi₂O₆ undergoes a reversible electrochemical reaction against Li/Li⁺ centered around 2 V with capacities near 60% of the theoretical maximum. We employ high resolution synchrotron X-ray and neutron diffraction to characterize the structure and correlate the rigid connectivity with the very slow kinetics of diffusion. We also use computational tools to understand the origin of the low potential compared with other Fe-based electrodes. ❧ We then use similar low-temperature techniques to prepare high-purity NaFeSi₂O₆ and an iron-based muscovite phase with unique microstructures. Interestingly, we found that the NaFeSi₂O₆ nanowires prepared by our low-temperature technique exhibit substantial difference in temperature- and field-dependent magnetic properties. The differences can likely be attributed to the reduced particle size and increased number of spins on the surface of the nanowires. In the case of our synthetic iron muscovite, a reversible capacity equal to 0.4 mole Li per formula unit was obtained. Cyclic voltammetry analysis showed significant contribution of the electrochemical capacity come from a surface controlled process. This is probably due to the limited amount of sites available for Li⁺ intercalation. ❧ In our study of the electrochemical properties of Fe₂(MoO₄)₃, significant differences in the structural and electrochemical properties during the intercalation of Li⁺ and Na⁺ ions were observed. To understand the origin of this behavior, we use a combination of in-situ X-ray and high-resolution neutron diffraction, total scattering, electrochemical measurements, density functional theory calculations, and symmetry-mode analysis. We find that for Li⁺-intercalation, which proceeds via a two-phase monoclinic-to-orthorhombic (Pbcn) phase transition, the host lattice undergoes a concerted rotation of rigid polyhedral subunits driven by strong interactions with the Li⁺ ions, leading to an ordered lithium arrangement. Na⁺-intercalation, which proceeds via a two-stage solid solution insertion into the monoclinic structure, similarly produces rotations of the lattice polyhedral subunits. However, using a combination of total neutron scattering data and density-functional theory calculations, we find that while these rotational distortions upon Na⁺ intercalation are fundamentally the same as for Li⁺-intercalation, they result in a far less coherent final structure, with this difference attributed to the substantial difference between the ionic radii of the two alkali metals. ❧ Finally, we explored the electrochemical Li⁺- and Na⁺-intercalation into defect-perovskite type ReO₃. The polhedral rotational distortions that we found dictates the mechanism of the guest ion intercalation in Fe₂(MoO₄)₃ also seems to play an important role in ReO₃. Theoretical two Li per formula intercalation in ReO₃ is possible with multiple stages during the electrochemical cycling. However, due to the huge mismatch between the sizes of Li⁺ ions and the interstitial void, and the enormous twist of the host lattice required to accommodate the inserted Li⁺, Li⁺-intercalation shows poor reversibility. On the other hand, reversibility of Na⁺-intercalation is greatly improved if it is limited to the solid solution process on the first stage. The closer fit between Na⁺ and the interstitial space in ReO₃, which reduces the twist of host framework and the local distortion, likely explains the enhanced reversibility.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

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

Modification of electrode materials for lithium ion batteries

PDF

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

PDF

Exploring new frontiers in catalysis: correlating crystal chemistry and activity in layered silicates

PDF

Advancing lithium batteries and related electrochemical technologies for a sustainable future

PDF

Understanding the relationship between crystal chemistry and physical properties in magnetic garnets

PDF

Effects of cationic species on the structure and properties of hybrid organic-inorganic materials

PDF

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

PDF

Electrocatalysts for direct liquid-feed fuel cells and advanced electrolytes for lithium-ion batteries

PDF

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

PDF

The chemistry of polymorphism in semiconductor nanocrystals

PDF

Simple complexes: synthesis and photophysical studies of luminescent, monovalent, 2-coordinate carbene-coinage metal complexes and higher coordination geometries

PDF

Properties and applications of dipyrrin-based molecules in organic photovoltaics

PDF

Electrochemical studies of outward and inward extracellular electron transfer by microorganisms from diverse environments

PDF

Influence of polymer structure on electronic properties and performance as donor materials in bulk heterojunction solar cells

PDF

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

Asset Metadata

Creator Zhou, Shiliang (author)

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

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Chemistry

Degree Conferral Date 2017-05

Publication Date 04/20/2017

Defense Date 03/14/2017

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

Tag electrochemistry,electrode,energy storage,lithium-ion batteries,OAI-PMH Harvest,structure-property

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Melot, Brent C. (committee chair), Nutt, Steven (committee member), Thompson, Mark E. (committee member)

Creator Email shilianz@usc.edu,zhoushiliang@yahoo.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC11255616

Unique identifier UC11255616

Identifier etd-ZhouShilia-5260.pdf (filename)

Legacy Identifier etd-ZhouShilia-5260

Dmrecord 364207

Document Type Dissertation

Format theses (aat)

Rights Zhou, Shiliang

Internet Media Type application/pdf

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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

Repository Name University of Southern California Digital Library

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

Repository Email cisadmin@lib.usc.edu