3D Vat Photopolymerization of Microarchitected Magnetic Metal Alloys for Chemotherapy Capture Filters Citation Shaker, Sammy (2026) 3D Vat Photopolymerization of Microarchitected Magnetic Metal Alloys for Chemotherapy Capture Filters. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/gc1t-5869. https://resolver.caltech.edu/CaltechTHESIS:09132025-054711849 Abstract Primary liver cancer constitutes an object of concern for communities across the globe. With the majority of new diagnoses of the most common subtype of primary liver cancer, hepatocellular carcinoma, inoperable at diagnosis, the standard of care revolves around the use of liver-directed therapies to treat tumors. The most popular therapy is termed transarterial chemoembolization, wherein the unique anatomy of the liver is exploited to deliver chemotherapy and embolic agents selectively to a large tumor, leading to tumor cell death and improved survival for patients. However, the chemotherapeutics used in this procedure have toxic effects on organs outside of the liver, and are as such dose-restricted on the basis of these side effects. In order to increase the amount of drug used and thus increase the chance of tumor cell death, chemotherapy capture devices are necessary. While some materials have been developed for this application, these devices regularly suffer from such restrictions as passive capture mechanisms necessitating the design of devices with poor hemodynamics or the use of immunogenic materials such as heteroDNA. Magnetic nanoparticles conjugated to chemotherapeutics as well as magnetic nanoparticle chemotherapy capture agents, delivered with the chemotherapeutics, present a potential way out of this conundrum, but the application of these materials requires the design of magnetic capture devices with favorable hemodynamic and magnetic properties. Architected magnetic metal devices can potentially provide the sought after solution to these difficulties, but techniques to such devices with high spatial resolution are lacking. An additive manufacturing technique that can provide high spatial resolution utilizes vat photopolymerization in tandem with thermal processing to produce well-resolved metal lattices that can provide for this ongoing need. This thesis applies this technique to the synthesis of magnetic lattices for particle capture. Lattices are synthesized in iron, nickel-iron, and copper-nickel-iron compositions and characterized structurally and magnetically. Simulations of particle capture in these lattices under various conditions are performed. In addition, attempts at particle capture are described and methods of characterization of particle capture are discussed. This technique is also explored with regards to the synthesis of iron-nickel and iron-cobalt lattices and the characterization of the resultant products is discussed and evaluated. Finally, a modification to this technique is used to generate metal-carbon microcomposites with unusual magnetic properties when compared with their counterparts synthesized using the unmodified procedure. Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: materials chemistry; three-dimensional printing; nanoparticles Degree Grantor: California Institute of Technology Division: Biology and Biological Engineering Major Option: Bioengineering Thesis Availability: Public (worldwide access) Research Advisor(s): Greer, Julia R. Thesis Committee: Shapiro, Mikhail G. (chair) Kornfield, Julia A. Guttman, Mitchell Greer, Julia R. Defense Date: 17 June 2024 Funders: Funding Agency Grant Number National Institute of General Medical Sciences (NIGMS) T32 GM008042 National Institute of General Medical Sciences (NIGMS) T32 GM152342 Record Number: CaltechTHESIS:09132025-054711849 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:09132025-054711849 DOI: 10.7907/gc1t-5869 ORCID: Author ORCID Shaker, Sammy 0000-0003-1751-4908 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 17678 Collection: CaltechTHESIS Deposited By: Sammy Shaker Deposited On: 29 Sep 2025 19:26 Last Modified: 07 Oct 2025 21:05 Thesis Files PDF (Thesis) - Final Version See Usage Policy. 307MB Repository Staff Only: item control page

3D Vat Photopolymerization of Microarchitected Magnetic Metal Alloys for Chemotherapy Capture Filters Thesis by Sammy Shaker In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CALIFORNIA INSTITUTE OF TECHNOLOGY Pasadena, California 2026 Defended June 17th , 2024 ii © 2026 Sammy Shaker ORCID: 0000-0003-1751-4908 All rights reserved except where otherwise noted iii ACKNOWLEDGEMENTS This work represents the culmination of my time at Caltech, and I would like to recognize all of those who shaped any part of my time here. That begins with my advisor, Prof. Julia R. Greer, who consistently pushed me to explore the scientific world independently and carve out my own section of it. I would not have been exposed to the world of materials science in the same way had I been advised by anyone else. I would like to thank the other members of my committee, including Prof. Mikhail Shapiro, Prof. Julia Kornfield, and Prof. Mitchell Guttman for their insistence on academic rigor and their willingness to share their time with me to improve the quality of my own academic pursuits. Their willingness to discuss projects that fit imperfectly into their own interests set an excellent example of the intellectual flexibility and adaptability so celebrated at Caltech. The path I took to working at Caltech began with my undergraduate work in the Stein lab. I owe Prof. Andreas Stein a great debt for taking me into his group and inspiring me to pursue scientific research in general and materials chemistry in particular. Stephen Rudisill was my first graduate student mentor and still the person with the greatest impact on my attitude towards scientific problem solving, and his support throughout my doctorate was invaluable. Prof. David Clark and Prof. Duane Nykamp guided my inquiries into mathematics research, and while I ended up taking a different direction in my research interests I was able to apply the mathematics training I received from them throughout my graduate work. My time with Sir Anthony Cheetham at Cambridge instilled in me the ability to believe in my own work and apply myself diligently to a problem, even after early failures. Yue Wu was a critical part of learning to attack a scientific problem from different directions, instead of rushing forward with the same approach repeatedly. The graduate research experience generally centers around the research group. During my time in the Greer group, I had the pleasure of interacting with world-class colleagues many of whom are now independent academic and industrial scientists as well as employees at top consulting firms across the globe. Among these colleagues were Andrey Vyatskikh and Daryl Yee, who were my first introduction to work in the group that I could understand as a prior synthetic chemist. I would later become more acquainted with the other senior graduate students in the Greer group, iv including Amylynn Chen, Rebecca Gallivan, Kai Narita, and Max Saccone, who maintained the standard for excellence that the Greer group has had through their doctoral work. I am especially indebted to Daryl Yee, Max Saccone and Kai Narita for their work on the HIAM process (more on this later) which would captivate me through the end of my doctorate and beyond. The senior students whose work would help shape the group for most of my time in graduate school were Yuchun Sun, Weiting Deng, and Andrew Friedman, and I always enjoyed their update presentations and the chance to ask questions and learn more about aspects of their work with which I was not familiar. In addition to learning from the work of others in the group and training on equipment with their supervision, I had the opportunity to collaborate with several excellent researchers during my time in the Greer group. Widianto Moestopo, one of my early officemates, introduced me to the world of micromechanics, and while he didn’t manage to recruit me as a mechanics researcher I had the chance to incorporate some postprocessing techniques to alter mechanical properties. We also, not incidentally, became closer friends through that work and beyond. Wenxin Zhang and I had the chance to observe some unusual microstructures and discover an interesting set of materials with unique structure-property relationships, while also trying some spicy sauces together. A chance exposure to a characterization technique that I had previously made use of allowed Wenyuan Chen and me to run a long experiment and get to know each other better. Outside of the Greer group, I had the pleasure of characterizing some air-sensitive compounds with interesting water absorption properties with Zachery Iton of the See group. Working in the Greer group required a lot of logistical and intellectual support from fellow graduate students. Kevin Nakahara provided both of these by making sure orders were placed, reimbursements were not too delayed, and that our office was always available for naps, rants, or serious discussions as needed. Fernando Villafuerte was always available for a discussion about anything in the world, whether related to the lab or not, and a late evening in the lab wasn’t the same after he left. Thomas Tran took my late-night “alerts" about the electron microscopes in stride while also showing me how to polish samples and when to let time solve some problems. Seneca Velling’s unending zest for life, wide and deep fund of knowledge, and seemingly unlimited energy helped motivate me through late nights and beyond. I owe a great deal to Ethan Klein, with whom I spent a lot of the last two years of my time at Caltech and whose attitude towards building research infrastructure and v blazing his own trail in the research world is an example I continue to try to emulate. Cyrus Fiori had a convoluted path in joining the Greer group, but his extensive training across multiple disciplines and ability to quickly gain mastery of different fields made our research conversations especially fruitful while also lending some levity to our troubles in the lab as an aspect of the absurdity of life itself. Joel Chacko joined the lab in time to supply enough Dr. Pepper to edge this thesis across the line, and in the meantime listen to enough monologues that I was able to get back to work without too much delay. Seola Lee and I had passed through the same chaotic first year with the tumult of the COVID-19 pandemic disrupting any chance at a normal graduate school experience without a before-and-after aspect to it, and I enjoyed watching her projects branch out and develop along different paths. Yingjin Wang joined the lab in time to be saddled with the responsibility of training people on the always-dying-but-never-quite-dead Ember 3D printer, a responsibility for which I struggled to find an appropriately conscientious person. The other MD-PhD students with whom I shared the lab deserve a special mention. Nikhil Linaval was my first junior student and a stalwart of the late nights in the lab; I hope we have more time to work together in the future. Akash Dhawan followed soon after him, and his dedication to the daytime work schedule provided a good example of healthy experimental planning in the face of our sleep cycle-ruining experimental schedules. A brief mention to Blade Olson, who came to our lab to compress some samples and found that he preferred the wet lab to the cold lab. These three convinced me that USC wasn’t so bad a place, at least while we were as far away from UCLA as Caltech was. I’d like to highlight our European contingent in the Greer group, Matias Kagias and Ben Spöttling. Matias formed the early morning team of the group and his deep knowledge of and extensive experience with X-ray physics and optics helped prompt many fruitful discussions; specifically, I appreciated his thorough training on the laser interference lithography setup and his help in submitting proposals to the ESRF for X-ray experiments that regretfully were not to be. Ben came to Caltech to explore another research university at a time I was exploring different research projects and trying to make headway in one, and we spent a great deal of time learning more about Caltech in the process. I am grateful for his companionship at that time and since, even if he did end up as a consultant. I had the pleasure of mentoring several excellent undergraduate and graduate researchers. Robin McDonald was my first SURF student and set the standard for vi SURFs during her time in the Greer group. Faiza Shabibi followed her one summer later and matched my highest expectations for work ethic just out of her freshman year. Andrii Syrota applied his physics knowledge to tackling simulations that we tailored to the experimental work I had started in the group, with interesting and insightful results using a lot of computing power. Finally, Mete Bayrak was my final student in the Greer group, and I felt that he surpassed my teaching very early on. I look forward to following the illustrious scientific careers that they’ve already embarked upon. The facilities and equipment teams at Caltech make sure that everything runs smoothly and that publically listed equipment is actually available and prepared for use. I thank Dr. Nathan Dalleska for his help in purchasing an attachment to the ICP-MS and spending many long hours troubleshooting it with me. Dr. Sonjong Hwang ran many solid-state NMR samples I prepared and was always patient with sample holder returns. Dr. Bruce Brunschwig helped keep the M-Probe XPS alive and revive it when it crashed with my sample still loaded. Dr. Michael Takase kept the X-ray crystallography facility running and the powder diffractometers usable for much of the work in this thesis. Dr. Paul Oyala allowed me to use an EPR magnet for an out-of-box application in the form of magnetization of lattices. Finally, Dr. Daniel Silevitch contacted me when I wrote my name on a board on the other side of campus to let me know that the instrument I was looking for was soon to arrive, and spent an inordinate amount of time keeping that same instrument up and running throughout my time at Caltech. The ChemoFilter consortium connected me with brilliant researchers bridging the gap between basic and clinical research. Prof. Steven Hetts introduced our group to the idea of capturing chemotherapy via catheter methods. Prof. Vitaliy Rayz provided support for our computational endeavors and was always available to go over calculations or simulation results. Ms. Teri Moore helped coordinate my trip to UCSF as well as arrange the consortium meetings and grant applications. Prof. Michael Schulz assisted with some early explorations into wider-ranging applications of the ChemoFilter idea. Colin Yee aided in the design of experimental apparatuses and provided input regarding methods of characterizing nanoparticle solutions. I learned an enormous amount while interacting with each member of the consortium. My friends in the Muslim Student Association made the second half of my doctorate far more lively than the first. I would never have explored as far as I had vii and tried as many things as I did without their presence, and while it wasn’t always smooth sailing they helped keep me from straying away from my community due to neglect and the pressures of the work at Caltech. Yousuf Abo Rahama, Mohammed Al-Dajani, Mohammad Arbab, Eray Atay, Sina Booeshaghi, Aidan Fenwick, Mohammad Muntasir Hassan, Noor Ibrahim, S. Süleyman Kahraman, A. Yusuf Kavranoğlu, Dženan Lapandić, Basel Mostafa, Omar Mehio, Mohammadamin Panahandeh, Dristy Parveg, Abdulqayyum Vlahakis, Hesham Zaini, and Humza Zubair; together we made sure Friday congregations were always available and tea always plentiful in the late evenings. I look forward to further discussions of everything and anything as well as seeing their own work progress at Caltech and beyond. I made many friends in medical school that I lost touch with, but those that I kept in contact with helped me push through some of the obstacles I encountered in my time at Caltech, reminding me that that there was still more work to go at the other end. Matiar Jafari had experienced the whole process and gave me excellent advice that I tried to implement at Caltech and back at UCLA. Mohanad Alazzeh, Usman Alam, Ayman Teeham Ullah, and Dylan Dean were the core of my circle of friends in my preclinical years, and they inspired me to keep going so that I could bring together the research and clinical sides of my degree program in my future career. Richard Paucek and I spent many Caltech afternoons discussing our work and finding some overlapping aspects, and his intellectual leadership in his research work helped me plod on to the end of my degree. I was born and raised in Minnesota and I have missed it since I first came to Southern California. I was grateful, then, to find a longtime friend from Minnesota here at Caltech in the form of Andrew Ylitalo. He reminded me of good times back home, and I got the chance to reconnect with someone whose work I had always admired. In addition, I appreciated the efforts of my friends who visited me while I was here and brought me glad tidings of back home, including Abdirahman Inshar, Faheem Jabir, Salah al Din Deban, Abderrahman Day, Omar Adam, Muhammad Uluyol, Jaylani Hussein, and Mahdi Yusuf. A special acknowledgement goes out to my friend Faiz Jabir, who came to see me here and spent plenty of time in conversation with me when I needed to vent and be reminded of what was going on back home. Some friends I made during my time in England managed to stay in contact with me despite the efforts of our competing schedules and differing timezones. Mohamed Ainte was never without a good story during his time in England, the Netherlands viii and Somalia. Osaid Ather kept up his streak of confounding me with a new, bizarre story each time we talked. Ahmed Komber made me want to rush back to medical school to become a Gibraltarian physician, though his subsequent descriptions of the facilities on the Rock tempered that enthusiasm. I hope to return and see how much they and other things have changed since I was last there. My final acknowledgments are to my family. I was blessed to be born to the parents I had, who dedicated themselves to my siblings and me and went above and beyond to assure us the opportunity to strive for excellence no matter what we pursued. My father encouraged me to apply myself to research like he had during his doctorate. My mother reminded me to ensure my faith grounded my work and not to shirk it so that I may not lose out on the rewards of both. Doubly blessed I was that I grew up with many siblings, so that I could learn to be myself while always having someone to hang out and share my time with. They never ceased to challenge me intellectually as well as provide perspective when I blew things out of proportion. My older sister Salma has a special place among them. She was always available to listen to my concerns and made sure that I knew that I always had a home to return to no matter what happened, while also keeping me accountable if I was acting or complaining unreasonably. I wish her and her husband Abdelhafid El Akri the utmost happiness. My older brother Tamer helped put things in perspective and made time to help me organize my thoughts and plan out my time when things seemed to stagnate. My younger brother Rami set for me a great example of diligence and perseverance in the pursuit of his goals, while providing some much needed levity from time to time. My older sister Zainah encouraged me throughout my studies to pursue excellence. I hope that my accomplishments are taken as a reflection of their influence on me, and my failings as a reflection of my falling short of the standard they set. ix ABSTRACT Primary liver cancer constitutes an object of concern for communities across the globe. With the majority of new diagnoses of the most common subtype of primary liver cancer, hepatocellular carcinoma, inoperable at diagnosis, the standard of care revolves around the use of liver-directed therapies to treat tumors. The most popular therapy is termed transarterial chemoembolization, wherein the unique anatomy of the liver is exploited to deliver chemotherapy and embolic agents selectively to a large tumor, leading to tumor cell death and improved survival for patients. The benefits of this procedure are blunted by the fact that the chemotherapeutics used in this procedure have toxic effects on organs outside of the liver, and as such are doserestricted by the extent of these off-target effects. In order to increase the amount of drug used and therefore increase the chance of tumor cell death, chemotherapy capture devices are necessary. While some materials have been developed for this application, these devices regularly suffer from such restrictions as passive capture mechanisms necessitating the design of devices with poor hemodynamics or the use of immunogenic materials such as heteroDNA. Magnetic nanoparticles conjugated to chemotherapeutics as well as magnetic nanoparticle chemotherapy capture agents, delivered with the chemotherapeutics, present a potential way out of this conundrum, but the application of these materials requires the design of magnetic capture devices with favorable hemodynamic and magnetic properties. Architected magnetic metal devices can potentially provide the sought after solution to these difficulties, but techniques to such devices with high spatial resolution are lacking. An additive manufacturing technique that can provide high spatial resolution utilizes vat photopolymerization in tandem with thermal processing to produce well-resolved metal lattices that can provide for this ongoing need. This thesis applies this technique to the synthesis of magnetic lattices for particle capture. Lattices are synthesized in iron, nickel-iron, and copper-nickel-iron compositions and characterized structurally and magnetically. Simulations of particle capture in these lattices under various conditions are performed. In addition, attempts at particle capture are described and methods of characterization of particle capture are discussed. This technique is also explored with regards to the synthesis of iron-nickel and iron-cobalt lattices and the characterization of the resultant products is discussed and evaluated. Finally, a modification to this technique is used x to generate metal-carbon microcomposites with unusual magnetic properties when compared with their counterparts synthesized using the unmodified procedure. xi TABLE OF CONTENTS Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi List of Illustrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxii Chapter I: Introduction and Literature Review . . . . . . . . . . . . . . . . . 1 1.1 Liver Cancer and Liver-Directed Therapy . . . . . . . . . . . . . . . 1 1.2 State-of-the-art in Chemotherapy Capture . . . . . . . . . . . . . . . 4 1.3 Additive Manufacturing of Metals . . . . . . . . . . . . . . . . . . . 7 1.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Chapter II: Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.1 Hydrogel Infusion Additive Manufacturing . . . . . . . . . . . . . . 12 2.2 X-ray Diffraction Phase Identification . . . . . . . . . . . . . . . . . 15 2.3 Energy Dispersive X-ray Spectroscopy Characterization . . . . . . . 19 2.4 Vibrating Sample Magnetometry . . . . . . . . . . . . . . . . . . . 21 2.5 Inductively-Coupled Plasma Mass Spectrometry . . . . . . . . . . . 23 2.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Chapter III: Additive Manufacturing of Iron and Iron-Alloy Lattices for Magnetic Nanoparticle Capture . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.1 Introduction to Transarterial Chemoembolization . . . . . . . . . . . 26 3.2 Magnetic Lattice Synthesis and Characterization . . . . . . . . . . . 29 3.3 In vitro Filtration Testing System . . . . . . . . . . . . . . . . . . . 38 3.4 COMSOL Multiphysics Capture Simulations . . . . . . . . . . . . . 42 3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Chapter IV: Exploring the Hydrogel Infusion Additive Manufacturing Parameter Space via Additively Manufactured Iron-Nickel and Iron-Cobalt . . . 54 4.1 Motivation for Iron-Nickel and Iron-Cobalt Compositions . . . . . . 54 4.2 Iron-Nickel Alloy Composition Tuning . . . . . . . . . . . . . . . . 55 4.3 Iron-Cobalt Alloy Exploration . . . . . . . . . . . . . . . . . . . . . 60 4.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 Chapter V: Enhancing Hydrogel Infusion Additive Manufacturing: Vacuum Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 5.1 Motivation for Vacuum Processing . . . . . . . . . . . . . . . . . . 68 5.2 Vacuum Processing Method . . . . . . . . . . . . . . . . . . . . . . 70 5.3 Iron vs. Copper-Nickel-Iron Vacuum HIAM . . . . . . . . . . . . . 70 5.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 Chapter VI: Summary and Outlook . . . . . . . . . . . . . . . . . . . . . . . 78 6.1 Transarterial Catheterization and Additive Manufacturing . . . . . . 78 6.2 Magnetic Materials Additive Manufacturing . . . . . . . . . . . . . 80 xii Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 Appendix A: ImageJ Measurements . . . . . . . . . . . . . . . . . . . . . . 90 A.1 Fe » Ni . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 Appendix B: EDS Measurements . . . . . . . . . . . . . . . . . . . . . . . . 93 B.1 Chapter 3 System: Ni-Fe . . . . . . . . . . . . . . . . . . . . . . . . 93 B.2 Chapter 3 System: Cu-Ni-Fe . . . . . . . . . . . . . . . . . . . . . . 99 B.3 Chapter 4 System: FeNi . . . . . . . . . . . . . . . . . . . . . . . . 104 B.4 Chapter 4 System: FeCo . . . . . . . . . . . . . . . . . . . . . . . . 106 Appendix C: Magnetic Nanoparticle Preparation . . . . . . . . . . . . . . . . 112 Appendix D: Vacuum HIAM Nanocrystalline Size and EDS Measurement . . 113 Appendix E: Magnetic Property Calculations . . . . . . . . . . . . . . . . . 115 Appendix F: COMSOL Simulation Parameters . . . . . . . . . . . . . . . . . 116 xiii LIST OF ILLUSTRATIONS Number Page 1.1 Overview of the dual blood supply to the liver. The proper hepatic artery here is labeled the hepatic artery; the common hepatic artery, of which this vessel is a branch, additionally supplies other parts of the gastrointestinal tract. Figure adapted with permission from [16]. . 3 1.2 Curves demonstrating the decrease in colony forming ability for tumor cells derived from testicular and bladder cancer cell lines with increasing dose of doxorubicin. Figure adapted with permission from [27]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.1 Overview of the hydrogel infusion additive manufacturing method used to generate metal samples from organogel blanks. Figure adapted with permission from [49]. . . . . . . . . . . . . . . . . . . 13 2.2 A schematic of honeycomb lattices modeled for idealized chemotherapy capture by Maani et al. Note the twisted, perforated structure on the top right was chosen for these studies as the optimum between capture efficacy and hemodynamic stability. Figure adapted with permission from [32]. . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.3 A schematic of the profiles used to generate samples in the oxidation step. (a) The sample is ramped at 1 deg/min to 100◦ C, 0.25 deg/min to 500◦ C, 2 deg/min to 700◦ C, held for 3 h, then cooled at 2 deg/min to 20◦ C. (b) The sample is ramped at 0.25 deg/min to 700◦ C, held for 3 h, then cooled at 2 deg/min to 20◦ C. . . . . . . . . . . . . . . . 15 2.4 An illustration of X-ray photons diffracting off of a pair of planes in a square lattice, where 𝑑 is the spacing between the planes and 𝜃 is the angle at which the photons contact the plane. The extra distance traveled by the 𝑆2 photon is equivalent to the sum of 𝐴𝐵 and 𝐵𝐶. Figure adapted with permission from [51]. . . . . . . . . . . . . . . 16 2.5 A depiction of the Bragg-Brentano configuration for powder X-ray diffraction. The X-ray source at left is rotated through an angle 𝜃 while the detector at right is locked to a position 2𝜃 from the undisturbed path of the X-ray beam. Radiation filters are mounted in the detector assembly. Figure adapted with permission from [52]. . . . . . . . . . 18 xiv 2.6 2.7 3.1 3.2 3.3 3.4 3.5 An illustration of the atomic process underlying EDS. Note that the precipitating excitation can occur as a result of the impingement of radiation or particles upon the atom; the latter process governs EDS as performed in an electron microscope as in this study. Figure adapted with permission from [56]. . . . . . . . . . . . . . . . . . . . . . . . The illustration of the VSM apparatus as invented by S. Foner in 1956. In this system, vibration is accomplished by mounting the sample (label 5) to a sample holding rod (label 3) attached to a loudspeaker transducer (label 1). A magnetic field is applied by an electromagnet (label 8) and the output signal is measured from the sample by one set of pickup coils (label 7) and another set of coils (label 6) are used to measure a reference sample response (label 4). Figure adapted with permission from [58]. . . . . . . . . . . . . . . An illustration depicting the fundamentals of the TACE procedure. The doxorubicin formulation includes a suspending oil and potentially drug-eluting beads or gelation agents for embolization. Figure adapted with permission from [65]. . . . . . . . . . . . . . . . . . . This diagram uses the iron-dominant sample to outline the HIAM process from exchanged hydrogel gel to swollen metal-ion hydrogel and subsequently the calcined and reduced samples. The scanning electron microscopy image provides a higher resolution image of the sample features. . . . . . . . . . . . . . . . . . . . . . . . . . . . . A demonstration of the transformation of the (a) swollen samples to (b) calcined oxides and finally (c) reduced metal samples. . . . . . . Gross structural images of nickel-iron and copper-nickel-iron lattices. (a) A photograph of the nickel-iron sample and the (b) copper-nickeliron sample as well as a scanning electron microscopy image of the overall structure and an inset highlighting one pore for each sample. . Powder diffraction patterns of the reduced (a) iron-dominant (b) nickel-iron and (c) copper-nickel-iron samples along with relevant simulated patterns for reference materials, as calculated from CIFs obtained from the following references via the ICSD: [70–72]. . . . . 20 22 28 30 31 32 33 xv 3.6 Cross-sectional studies of an iron lattice. (a) An EDS map of the focused ion beam milled node of the iron-dominant sample shown in Fig.3.2, along with the maps corresponding to the iron (red) and oxygen (blue) elements (b) Scanning electron microscopy image and phase map electron backscatter diffraction of the iron-dominant sample. 34 3.7 A summary diagram of the three categories of samples discussed in this sample along with visualizations of the beam and pore sizes and the powder diffraction patterns corresponding to the three systems. The simulated patterns are calculated from CIFs obtained from the following references via the ICSD: [71, 72]. . . . . . . . . . . . . . . 35 3.8 Visualizations of the ten regions from which EDS maps were collected for the (a) nickel-iron and (b) copper-nickel-iron systems, along with the average composition obtained from this collected data. See Appendix B for more information. . . . . . . . . . . . . . . . . . . . 36 3.9 The computed nickel-iron ratio in the (a) central (001) and (b) peripheral (010) locations of the nickel-iron system and the (c) central (001) and (d) peripheral (010) locations of the copper-nickel-iron system. . 37 3.10 Vibrating sample magnetometry curves for the iron-dominant, nickeliron, and copper-nickel-iron systems including an inset focusing on the center of the curves and illustrating the coercivity and remnance of the systems, along with a table demonstrating the saturation and remnant magnetization as well as the maximum energy product and coercivity of the depicted sample systems. . . . . . . . . . . . . . . . 38 3.11 Experimental particle capture system. (a) A schematic demonstrating the essential parts of the particle capture testing system along with a (b) photograph of the real-life system. . . . . . . . . . . . . . . . . . 39 3.12 Images of post-nanoparticle-capture iron-dominant samples along with EDS maps demonstrating a difference in iron counts between the rounded particles on the surface of the sample and the bulk sample, along with the associated EDS spectrum demonstrating the presence of solely iron and oxygen in the imaged area. . . . . . . . . . . . . . 40 xvi 3.13 Plots depicting the measured nanoparticle concentration in digested aliquots taken after running a magnetite nanoparticle solution of 0.2 mg/mL concentration through metal lattices in the capture setup. In (a) nanoparticle solutions of concentration 0.05-0.20 mg/mL were diluted, digested, and run in the system with the resultant 56 Fe signals recorded versus undiluted concentrations along with a linear fit to the data with the displayed equation and 𝑅 2 , while in (b) solutions of initial concentration 0.20 mg/mL that had been passed through the experimental setup using three magnetized and three nonmagnetized lattices were diluted, digested, and run and the resultant 56 Fe signals plotted as the blue dots. Red triangles display the average signal recorded for each lattice condition which is displayed numerically along with the sample standard deviation of the signals recorded. . . 3.14 The mesh underlying the capture simulations performed here including the (a) visualization of the sphere of air containing the vessel and the (b) twisted, perforated honeycomb geometry of the single capture lattice as well as the (c) triple capture device. With thanks to Andrii Syrota for generating these structures. . . . . . . . . . . . . . . . . . 3.15 A map of the magnetic flux density within and surrounding a magnetic lattice with an overall magnetic flux density of 1.45 T. The traced lines demonstrate the general orientation of the magnetic field generated by the lattice in the free volume surrounding the lattice. The scale at the right demonstrates the intensity of the flux density in Tesla at each point in the lattice and in the free volume surrounding the lattice. With thanks to Andrii Syrota for generating this plot. . . . . . . . . . 3.16 This plot shows the radially averaged development of the flow as it passes into the twisted, perforated honeycomb magnetic lattice. A curve with label Distance = −𝑑 represents the radially averaged velocity profile of the flow a distance 𝑑 from the major face of the lattice closest to the origin of the flow. At the right, a figure demonstrates the axis 𝑍 which shows how negative distance is related to the direction of the inflow 𝑢 from the proximal magnetic lattice face at 𝑍 = 0. The flow breaks down from the smooth parabolic state it begins with to a jagged series of jets passing into the honeycomb pores as it encounters resistance from the beams of the lattice. With thanks to Andrii Syrota for generating this plot. . . . . . . . . . . . . . . . . . . . . . 41 43 44 45 xvii 3.17 Simulation results displaying the number of particles captured along the axis of a single lattice orthogonal to the main honeycomb face with a velocity of 1.6 m/s and an assigned magnetic flux density of 1.45 T (blue) with capture of 50.6% of the particles and alternatively an assigned magnetic flux density of 0.028 T (red) with capture of 30.4% of the particles. With thanks to Andrii Syrota for generating this plot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.18 Simulation results displaying the number of particles captured as a function of the depth of penetration 𝑧 into the lattice, where 𝑧 = 0 denotes the location of the major face of the lattice that the particleladen flow initially impinges upon. The different curves correspond to simulation data in systems with inflow velocities ranging from 0.01 to 1 m/s and assigned magnetic flux densities ranging from 0.01 to 1 T. The data for the capture efficiency for these nine lattice conditions is in the inset and reproduced in a larger form in Table 3.1. With thanks to Andrii Syrota for generating this plot. . . . . . . . . . . . . 3.19 Capture in a triple lattice device. (a) Simulation results demonstrated the number of captured particles as a function of the distance along the axis of a triple lattice device as seen in (b). These simulations were run using an inflow velocity of 1.6 m/s for a triple lattice device in which each lattice was assigned a magnetic flux density of 1.45 T (blue) demonstrating capture of 77.8% of particles, and triple lattice device in which each lattice was assigned a magnetic flux density of 0.028 T (red) demonstrating capture of 53.4% of particles. With thanks to Andrii Syrota for generating this plot. . . . . . . . . . . . . 4.1 A plot demonstrating the difference between the concentration of nickel in the swell-in solution versus the nickel content of the reduced metal sample as determined by EDS in the nickel-iron system targeting the Ni80 Fe20 composition. The dashed line represents systems in which the swell-in solution and final composition are identical 46 48 50 56 xviii 4.2 4.3 4.4 4.5 4.6 4.7 4.8 5.1 The normalized powder diffraction patterns corresponding to the oxide phases derived from the nickel-iron samples synthesized to target the Ni80 Fe20 along with simulated patterns derived from the following references obtained via the ICSD:[85, 86]. Note that the compositions in the pattern labels refer to the metal molar percentages in the final determination and not the stoichiometry of any given compound. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The powder diffraction patterns corresponding to the metal phases derived from the nickel-iron samples synthesized to target the Ni80 Fe20 along with simulated patterns derived from the following references obtained via the ICSD:[71, 72]. . . . . . . . . . . . . . . . . . . . . A plot demonstrating the difference between the concentration of cobalt in the swell-in solution versus the cobalt content of the reduced metal sample as determined by EDS across a range of swell-in solution concentrations. The dashed line represents systems in which the swell-in solution and final composition are identical. . . . . . . . The powder diffraction patterns corresponding to the oxide phases derived from the cobalt-iron samples synthesized in a range of compositions along with simulated patterns derived from the following references obtained via the ICSD:[91–93]. . . . . . . . . . . . . . . . The powder diffraction patterns corresponding to the metal phases derived from the cobalt-iron samples synthesized in a range of compositions along with simulated patterns derived from the following references obtained via the ICSD:[72, 95, 96]. . . . . . . . . . . . . . A plot of the hysteresis loops for the synthesized Fe-Co samples as collected by VSM. An inset is provided at the right for the area of interest with regards to magnetic characterization. . . . . . . . . . . Plots of the three magnetic properties discussed in Table 4.1 displaying the (a) remnant magnetization, (b) coercivity, and (c) maximum energy product versus the cobalt content of the metal as determined via EDS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A schematic illustrating the synthesis of lattices via the vacuum HIAM processing method with reference to the iron vacuum HIAM sample. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 59 61 62 63 64 66 70 xix 5.2 5.3 5.4 5.5 5.6 A.1 A.2 A.3 B.1 A scanning electron image of the iron vacuum HIAM sample along with a powder diffraction pattern from a representative sample and a simulation derived from a CIF file obtained from the following reference via the ICSD:[72]. . . . . . . . . . . . . . . . . . . . . . . 71 A demonstration of the preparation of the Fe sample for TEM characterization. (a) An inset illustrating the approximate orientation of the lamella with regards to the bulk sample of vacuum HIAM iron (b) A depiction of the lamella obtained from the sample and the identified regions of the sample (c) Images of the specific regions of the lamella illustrated in part (b) along with (d) the fast Fourier transform of the nanocrystalline phases of the lamella to obtain pseudo-spacings. . . . 72 A scanning electron image of the copper-nickel-iron vacuum HIAM sample along with a powder diffraction pattern from a representative sample and a simulation derived from a CIF file obtained from the following reference via the ICSD:[70]. . . . . . . . . . . . . . . . . . 74 A demonstration of the preparation of the Cu-Ni-Fe sample for TEM characterization. (a) An inset illustrating the approximate orientation of the lamella with regards to the bulk sample of vacuum HIAM copper-nickel-iron. (b) An image of the large crystallites noted on the lamella and a fast Fourier transform of the image to generate pseudospacings as well as (c) an image of the fine nanocrystalline matrix and its fast Fourier transform to obtain the same pseudo-spacings of interest. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Vibrating sample measurements demonstrating the hysteresis loop for (a) traditional HIAM synthesized iron and copper-nickel-iron and (b) vacuum HIAM synthesized iron and copper-nickel-iron composites. 76 Histograms depicting the size of the beams and pores in the irondominant samples discussed in Chapter 3. For the beams, n=250, and for the pores, n=70. . . . . . . . . . . . . . . . . . . . . . . . . 91 Histograms depicting the size of the beams and pores in the nickeliron samples discussed in Chapter 3. For the beams, n=244, and for the pores, n=74. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 Histograms depicting the size of the beams and pores in the coppernickel-iron samples discussed in Chapter 3. For the beams, n=236, and for the pores, n=72. . . . . . . . . . . . . . . . . . . . . . . . . 92 EDS of Ni-Fe site 001 from Fig. 3.8. . . . . . . . . . . . . . . . . . 93 xx B.2 B.3 B.4 B.5 B.6 B.7 B.8 B.9 B.10 B.11 B.12 B.13 B.14 B.15 B.16 B.17 B.18 B.19 B.20 C.1 EDS of Ni-Fe site 002 from Fig. 3.8. . . . . . . . . . . . . . . . . . 94 EDS of Ni-Fe site 003 from Fig. 3.8. . . . . . . . . . . . . . . . . . 94 EDS of Ni-Fe site 004 from Fig. 3.8. . . . . . . . . . . . . . . . . . 95 EDS of Ni-Fe site 005 from Fig. 3.8. . . . . . . . . . . . . . . . . . 95 EDS of Ni-Fe site 006 from Fig. 3.8. . . . . . . . . . . . . . . . . . 96 EDS of Ni-Fe site 007 from Fig. 3.8. . . . . . . . . . . . . . . . . . 96 EDS of Ni-Fe site 008 from Fig. 3.8. . . . . . . . . . . . . . . . . . 97 EDS of Ni-Fe site 009 from Fig. 3.8. . . . . . . . . . . . . . . . . . 97 EDS of Ni-Fe site 010 from Fig. 3.8. . . . . . . . . . . . . . . . . . 98 EDS of Cu-Ni-Fe site 001 from Fig. 3.8. . . . . . . . . . . . . . . . 99 EDS of Cu-Ni-Fe site 002 from Fig. 3.8. . . . . . . . . . . . . . . . 99 EDS of Cu-Ni-Fe site 003 from Fig. 3.8. . . . . . . . . . . . . . . . 100 EDS of Cu-Ni-Fe site 004 from Fig. 3.8. . . . . . . . . . . . . . . . 100 EDS of Cu-Ni-Fe site 005 from Fig. 3.8. . . . . . . . . . . . . . . . 101 EDS of Cu-Ni-Fe site 006 from Fig. 3.8. . . . . . . . . . . . . . . . 101 EDS of Cu-Ni-Fe site 007 from Fig. 3.8. . . . . . . . . . . . . . . . 102 EDS of Cu-Ni-Fe site 008 from Fig. 3.8. . . . . . . . . . . . . . . . 102 EDS of Cu-Ni-Fe site 009 from Fig. 3.8. . . . . . . . . . . . . . . . 103 EDS of Cu-Ni-Fe site 010 from Fig. 3.8. . . . . . . . . . . . . . . . 103 A schematic illustrating the preparation of magnetic nanoparticles for capture testing experimentation. Iron(II,III) oxide nanoparticles, as purchased from Sigma-Aldrich, were suspended in water to which citric acid was added and subsequently heated at 65◦ C for one hour. The particles were allowed to settle and the supernatant removed, at which point the nanoparticles were washed with ethanol twice and subsequently dried in air. Nanoparticle solutions for capture testing were synthesized by combining these treated nanoparticles with phosphate buffered saline in the concentration of interest; in this case, 5 milligrams per milliliter of solution, the equivalent of a 3.58 mM solution of functionalized nanoparticles in phosphate-buffered saline. The citrate coating procedure was inspired by [130]. . . . . . 112 D.1 Vacuum HIAM liftout sample preparation for TEM, including (a) an indication of the region characterized by EDS in section (b), where the iron, carbon, and oxygen distribution are shown. Note the contrast of the particles in part (a) is the opposite of that seen in the lamella in Fig. 5.3. Data courtesy of Wenxin Zhang. . . . . . . . . . . . . . 113 xxi D.2 A schematic illustrating the identified crystallites in each layer along with circles outlining grains to be used to estimate grain size in (a) the superficial nanocrystalline and (b) deep nanocrystalline layers of the iron vacuum HIAM system and (c) the fine nanocrystalline layer of the copper-nickel-iron vacuum HIAM system. . . . . . . . . . . . 114 xxii LIST OF TABLES Number Page 2.1 3D Printing Parameters for BL5050 Resin on the Autodesk Ember. These parameters are generally identical to those found in [49], without an offset between the first two layers. . . . . . . . . . . . . . . . 15 3.1 Simulated capture efficiencies in systems with flow velocities ranging from 0.01 to 1 m/s and assigned magnetic flux densities ranging from 0.01 to 1 T. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.1 Magnetic properties of the synthesized iron-cobalt samples, including the remnant magnetization, coercivity, and maximum energy product. 65 A.1 Shrinkage results for the iron-dominant samples discussed in Chapter 3 Sections 2 and 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 A.2 Shrinkage percentages with respect to the preceding phase of the system for the iron-dominant samples discussed in Chapter 3 Sections 2 and 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 B.1 Iron and nickel content used to determine average composition of the Ni-Fe sample in Chapter 3 Section 2.1. . . . . . . . . . . . . . . . . 98 B.2 Copper, nickel, and iron content used to determine average composition of the sample in Chapter 3 Section 2.1. . . . . . . . . . . . . . . 104 B.3 Nickel mole percentage in the swelling solution versus the average EDS-measured mole percentage. . . . . . . . . . . . . . . . . . . . . 104 B.4 Iron and nickel content determining the composition of the Ni80 Fe20 swell-in sample. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 B.5 Iron and nickel content determining the composition of the Ni84 Fe16 swell-in sample. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 B.6 Iron and nickel content used to determinine the composition of the Ni87 Fe13 swell-in sample. . . . . . . . . . . . . . . . . . . . . . . . 105 B.7 Iron and nickel content used to determinine the composition of the Ni90 Fe10 swell-in sample. . . . . . . . . . . . . . . . . . . . . . . . 105 B.8 Iron and nickel content used to determinine the composition of the Ni93 Fe7 swell-in sample. . . . . . . . . . . . . . . . . . . . . . . . . 105 B.9 Iron and nickel content used to determinine the composition of the Ni88 Fe12 swell-in sample. . . . . . . . . . . . . . . . . . . . . . . . 105 B.10 Iron and nickel content used to determinine the composition of the Ni89 Fe11 swell-in sample. . . . . . . . . . . . . . . . . . . . . . . . 106 xxiii B.11 B.12 B.13 B.14 B.15 B.16 B.17 B.18 B.19 B.20 B.21 B.22 D.1 F.1 Iron and nickel content used to determinine the composition of the Ni90.4 Fe9.6 swell-in sample. . . . . . . . . . . . . . . . . . . . . . . 106 Iron and nickel content used to determinine the composition of the other Ni90 Fe10 swell-in sample. . . . . . . . . . . . . . . . . . . . . 106 Cobalt and iron mole percentages in the swell-in samples synthesized to characterize the iron-cobalt HIAM system. . . . . . . . . . . . . . 106 Cobalt and iron content used to determine average composition of the Co90 Fe10 swollen sample. . . . . . . . . . . . . . . . . . . . . . 107 Cobalt and iron content used to determine average composition of the Co80 Fe20 swollen sample. . . . . . . . . . . . . . . . . . . . . . 107 Cobalt and iron content used to determine average composition of the Co70 Fe30 swollen sample. . . . . . . . . . . . . . . . . . . . . . 108 Cobalt and iron content used to determine average composition of the Co60 Fe40 swollen sample. . . . . . . . . . . . . . . . . . . . . . 108 Cobalt and iron content used to determine average composition of the Co50 Fe50 swollen sample. . . . . . . . . . . . . . . . . . . . . . 109 Cobalt and iron content used to determine average composition of the Co40 Fe60 swollen sample. . . . . . . . . . . . . . . . . . . . . . 109 Cobalt and iron content used to determine average composition of the Co30 Fe70 swollen sample. . . . . . . . . . . . . . . . . . . . . . 110 Cobalt and iron content used to determine average composition of the Co20 Fe80 swollen sample. . . . . . . . . . . . . . . . . . . . . . 110 Cobalt and iron content used to determine average composition of the Co10 Fe90 swollen sample. . . . . . . . . . . . . . . . . . . . . . 111 Grain diameters as computed from the area of the circle inscribing several nanoparticles for each of the nanocrystalline regimes discussed in Chapter 5. . . . . . . . . . . . . . . . . . . . . . . . . . . 114 Parameters used to compute the capture efficiency of magnetic lattices as discussed in Chapter 3. Collected and determined by S. Shaker and A. Syrota. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 1 Chapter 1 INTRODUCTION AND LITERATURE REVIEW 1.1 Liver Cancer and Liver-Directed Therapy Primary liver cancer, malignancy that arises from the tissue of the liver and gallbladder, is a current and growing public health concern in the twenty-first century.1–7 These conditions are among the ten most frequent diagnoses of cancer and among the five most frequent causes of cancer death globally.8,9 The most common type of primary liver cancer is hepatocellular carcinoma (HCC) which makes up anywhere between 75 % and 90 % of diagnoses of primary liver cancer.9 Hepatocellular carcinoma occurs as a result of various predisposing factors; among the most common are the chronic infections of the viral hepatitides, hepatitis C (HCV) and hepatitis B (HBV), non-alcoholic fatty liver disease (NAFLD), alcoholic liver disease (ALD), aflatoxin and other toxin exposures, as well as genetic conditions (alpha1-antitrypsin, Wilson disease, etc.). Due to ongoing efforts regarding vaccination, needle safety and blood transfusion screening, HCV and HBV infections are decreasing in importance as causes of HCC, although they remain the major causes of HCC in East Asia. Instead, due to heavy alcohol use and increasing rates of obesity, NAFLD and ALD are responsible for a greater portion of conditions resulting in HCC, as well as an increasingly greater portion of HCC cases. This is especially true in most regions of the world outside Southeast and East Asia. As such, while the composition of causes differs by region under consideration, HCC is a serious concern in all parts of the world. Part of the concern surrounding HCC, as can be imagined from the discrepancy between the relative rank of primary liver cancer as a diagnosis and a cause of cancer death, relates to the aggressive clinical course and existence of relatively few efficacious options for treatment in late stage disease. Early stage disease can be treated by ablation of the tumor, resection of the tumor, resection of a part of the liver, or liver transplantation; all of these treatment modalities are potentially curative and some modalities provide greater than 50 % five-year survival in qualifying patient cohorts. While these numbers portend a much less dire outcome for sufferers of HCC, most hepatocellular carcinomas are diagnosed at intermediate or advanced stages, including tumors diagnosed in the context of hepatic fibrosis 2 or complete cirrhosis, and as such cannot be treated with these curative therapies. Instead, HCC at these stages is treated using systemic chemotherapeutics and immunotherapies as well as liver-directed therapies. Systemic chemotherapeutics and immunotherapies in the context of HCC resemble the treatments familiar to other malignancies; specifically, sorafenib, cabozantinib, regorafenib, and lenvanitinib as multikinase inhibitors, bevacizumab and ramucirumab as VEGF inhibitors, durvalumab, nivolumab, pembrolizumab, and atezolizumab as anti-PD-L1 immunotherapies, and tremelimumab and ipilimumab as anti-CTLA-4 immunotherapies.10–12 The limited therapeutic benefit versus the significant risk of off-target side effects associated with these use of these therapies render them secondary options to liverdirected or locoregional therapies. Notably, patients with advanced HCC exhibit hepatic decompensation in addition to their primary tumor, and the degree to which decompensation has advanced dictates the tolerability of systemic treatments in these patients as indiscriminate liver damage can be a side effect of these therapies.9,12 As such, directing therapy locally can thread the needle between curative therapy and difficult to tolerate noncurative systemic therapies. Liver-directed therapies for hepatocellular carcinoma center around techniques employing one or more of the following modalities: thermal ablation, localized chemotherapy, radiation, or embolization.12–14 These modalities target different susceptibilities that cancer cells exhibit: to hot or cold temperatures, chemotherapeutic drugs, X-ray and small particle irradiation, and starvation of vital nutrients. While usually not curative, these techniques can often serve to bridge patients until curative therapies or feasible or help “down-stage” patients who may then qualify for curative therapies.9 Some widely used techniques include radiofrequency ablation, cryoablation, percutaneous ethanol or acetic acid injection, stereotactic body radiation therapy, proton radiation therapy, transarterial chemoembolization, and radioembolization.15 Chemical (ethanol, acetic acid) or thermal ablation techniques can cure early stage HCC with limited damage to healthy liver tissue but are used at early stages for patients who cannot tolerate transplantation or resection, and as such have limited applicability to the majority of HCC cases diagnosed in the intermediate or advanced stage.9,13,15 Radiotherapy in the context of HCC is challenging due to the potential damage to healthy liver tissue resulting in exacerbation of underlying liver disease in these patients; in order to address this challenge, targeted radiation therapy has been developed so that lethal doses can be applied to small tumors while limiting damage to nonmalignant tissue. External radiation techniques, such as stereotactic body radiation, that can deliver high local doses are under investigation 3 as first-line therapies for intermediate stage HCC, and can be combined with other liver-directed therapies to improve outcomes. In the case of radioembolization and transarterial chemoembolization, among other therapies, these embolotherapeutic techniques have found extensive use in treating intermediate stage HCC. Embolotherapy for hepatocellular carcinoma revolves around the unique anatomy of the blood supply of the liver. The liver processes metabolites transported to it from the intestines via the portal vein, which carries deoxygenated blood, while the oxygen supply of the liver is mostly derived from the hepatic artery, a branch of the abdominal aorta (see Fig. 1.1).15 Hepatocellular carcinomas, like other solid malignancies, obtain oxygen from the arterial supply of the liver, while the rest of the liver tissue can be supplied via the remnant oxygen in the portal vein.14 Figure 1.1: Overview of the dual blood supply to the liver. The proper hepatic artery here is labeled the hepatic artery; the common hepatic artery, of which this vessel is a branch, additionally supplies other parts of the gastrointestinal tract. Figure adapted with permission from [16]. The hepatic artery can be used to introduce drugs, radiation, and embolic agents to treat intermediate stage HCC with limited effect to the healthy tissue of the liver 4 crucial to improving liver function after treatment is completed. The initial approach to exploiting this anatomic arrangement involved simple embolization of branches of the hepatic artery using inert particles introduced via catheter and were introduced to the medical literature by Japanese medical researchers in the 1980s.17–19 Further studies confirmed the plausibility of embolotherapy to treat intermediate stage HCC, with added variations such as the introduction of chemotherapy, oil to suspend chemotherapy, and drug-eluting particles along with the embolic agents.20–22 With the publication of a systematic review comparing these techniques in 2003 in the journal Hepatology, the use of embolotherapy combined with chemotherapy and a viscous oil as a carrier agent, termed transarterial chemoembolization (TACE), would become the first-line treatment for unresectable intermediate stage HCC, a status it retains to this day.8,23 Much of the innovation since the first explications of the TACE procedure involve modifying the chemotherapy cocktail as well as supplementing the TACE procedure with additional systemic chemotherapeutics, ablation procedures, radiotherapy, and surgical resection. While the benefits of the TACE procedure are undeniable, its applicability still suffers due to the side effect profile of the procedure, such as general symptoms of illness (nausea, malaise, fever, fatigue, etc.) as well as resistance to the supplied drug and difficulty tolerating the off-target effects of the chemotherapeutics used in the procedure due to leakage into the systemic circulation. This is reflected in studies of TACE that demonstrate its superiority to supportive care with respect to patient survival but can seen limited use in some patient populations that have difficulty tolerating the effects of the procedure.15 Decreasing the spread of chemotherapeutics outside of the target organ represents a rich avenue of research with regards to increasing the use of the TACE procedure in sensitive patient populations. 1.2 State-of-the-art in Chemotherapy Capture Chemotherapeutics can show toxicity at both the target and off-target organs of the cancer under treatment; they may be limited in application due to toxicity to either type of organ. When toxicity is more significant in off-target organs, drug doses are limited by damage that is not “required” in order to treat the malignancy at hand. This decoupling of therapeutic effect and adverse effect implies that increasing local chemotherapeutic doses while decreasing systemic doses is a viable avenue to improving patient outcomes-this rationale underlies the previously described TACE procedure, but this procedure is an ideal target for this sort of intervention because of the choice of chemotherapy. Doxorubicin, an anthracycline antitumor agent, is 5 a drug of choice for TACE procedures, and while it exhibits limited hepatotoxicity it is also well known for causing permanent cardiac injury in the form of dilated cardiomyopathy.24,25 The incidence of this condition increases with total dose above 500 mg/m2 ; at the same time, increasing doses of doxorubicin have been shown to increase tumor cell death as in Fig. 1.2.26,27 Figure 1.2: Curves demonstrating the decrease in colony forming ability for tumor cells derived from testicular and bladder cancer cell lines with increasing dose of doxorubicin. Figure adapted with permission from [27]. As seen in Fig. 1.1, the hepatic veins carry deoxygenated blood back to the heart. The placement of a device in the hepatic veins provides an ideal opportunity to absorb remnant doxorubicin before it reaches the heart which decouples the intrahepatic dose from the cardiac dose of doxorubicin. Previous attempts to capture doxorubicin have taken advantage of the mechanism of action of doxorubicin as a chemotherapeutic as well as the attraction of doxorubicin to various charged species in vivo. Oh et al. in 2019 demonstrated the formation of a doxorubicin sequestration device composed of a 3D printed scaffold coated with a block copolymer with polystyrenesulfonate ((CH2 CH(C6 H4 SO−3 ))𝑛 ) blocks to rapidly absorb doxorubicin by electrostatic interaction with these groups; doxorubicin has a positive charge at blood pH.28,29 Doxorubicin intercalates between DNA strands in order to interrupt DNA replication and subsequently halt the mul- 6 tiplication of cancer cells.30 This mechanism appears agnostic to the origin of the DNA in question, and so affixing DNA of any origin on an inert intravenous device provides one mechanism for selective capture of doxorubicin from the bloodstream during the TACE procedure and its subsequent removal at the end of the procedure. This approach was adopted by Yee et al. in 2021, where 3D-printed structures were coated with DNA and demonstrated to pick up doxorubicin from human serum and phosphate buffered saline.31 The limitations in these systems can be best appreciated by the fact that they rely on a passive mechanism to bring doxorubicin and the capture material in contact; namely, the flow of blood around the device in question. The hemodynamics of the hepatic venous system and the inferior vena cava require capture devices that do not pose large obstacles to the flow of blood; the pressure of the inferior vena cava is approximately 10 mmHg. A pressure drop between 5-7 mmHg is likely the maximum that can be tolerated to avoid stagnation and thrombosis.32 This implies that passive capture mechanisms, which will rely on increased surface areas, can be an obstacle to maintaining stable hemodynamics in these vessels, especially in the context of such low pressure systems. More hemodynamically favorable methods would capture large amounts of chemotherapy via active sequestration methods. In this context, work by Blumenfeld et al. and Krishnamoorthy et al. demonstrate the functionalization of magnetite nanoparticles with DNA and aldehyde moieties, respectively, to accomplish chemotherapy capture in a massive expanded surface area using particles that incorporate into the flow as opposed to static objects obstructing the venous blood flow.33,34 This approach can be taken further by binding chemotherapy directly to magnetite nanoparticles can be taken advantage of by synthesizing conjugated doxorubicin-magnetite nanoparticles, that can act on the target organ and subsequently diffuse out of the cell to be captured by magnetic devices, as demonstrated in work by Kondapavulur et al. in 2016 and Mabray et al. in 2016.35,36 This approach can then be incorporated with magnetic capture devices, which employ an active mechanism for capture, by limiting the surface area and tortuosity required to effect capture and subsequently improving the hemodynamic properties of such a capture device. Previous studies employed magnetic devices consisting of small or large magnets strung along a wire, and showed efficacious capture of these particles from in vitro flow systems.35,36 In order to advance the capture of these particles, methods of manufacturing complex, architected magnetic materials could be applied to attain capture devices with optimized hemodynamic properties and active capture mechanisms. 7 1.3 Additive Manufacturing of Metals Traditional manufacturing methods can be lumped into two categories: subtractive and molding processes. In subtractive manufacturing, raw materials are processed into a shape encompassing the desired architecture of the final product, and material is then removed to attain the intended final design. Molding processes produce a negative of the desired final architecture, which is then used to shape a precursor material into the final configuration of interest. Both of these methods, while suitable for the manufacture of simple parts, have significant limitations when it comes to manufacturing parts with complex architectures. Molding processes are difficult to design when overhangs and small holes are present in the geometry of interest, and subtractive processes are wasteful when materials may be rare or expensive to obtain in bulk. In the midst of the limitations of these traditional manufacturing methods, parts with complex architectures are of increasing interest for applications in medical engineering, battery materials, and mechanical devices. Processes termed additive manufacturing (AM) have been developed to overcome the limitations of traditional manufacturing with regards to the manufacture of parts with complex architectures. Additive manufacturing processes generate parts by adding material selectively such that the final architecture is generated without the need for removal of excess material or removal from a mold. In the ASTM International Standard Terminology for Additive Manufacturing Technologies, additive manufacturing processes are categorized as follow (the following block quote consisting of a list of bullet points is set off from the rest of the text as a reproduction of material from the ASTM standard consistent with the APA citation style): • binder jetting, n—an additive manufacturing process in which a liquid bonding agent is selectively deposited to join powder materials. • directed energy deposition, n—an additive manufacturing process in which focused thermal energy is used to fuse materials by melting as they are being deposited. DISCUSSION —“Focused thermal energy” means that an energy source (e.g., laser, electron beam, or plasma arc) is focused to melt the materials being deposited. • material extrusion, n—an additive manufacturing process in which material is selectively dispensed through a nozzle or orifice. 8 • material jetting, n—an additive manufacturing process in which droplets of build material are selectively deposited. DISCUSSION —Example materials include photopolymer and wax. • powder bed fusion, n—an additive manufacturing process in which thermal energy selectively fuses regions of a powder bed. • sheet lamination, n—an additive manufacturing process in which sheets of material are bonded to form an object. • vat photopolymerization, n—an additive manufacturing process in which liquid photopolymer in a vat is selectively cured by light-activated polymerization.37 Many additive manufacturing processes are termed three-dimensional (3D) printing due to the resemblance between the selective placement of material in three dimensions and the selective placement of ink in two dimensions by a printer. With regards to capture devices, additive manufacturing techniques resulting in metal parts are of special interest. In this regard, three approaches are traditionally of greatest historical and practical significance: binder jetting (BJ), laser powder bed fusion (LPBF) and direct energy deposition (DED).38 Binder jetting and laser powder bed fusion will be discussed in more detail as well as a third approach, vat photopolymerization, that has gradually found greater use in metal additive manufacturing with the development of postprocessing methods. Binder jetting refers to the use of a binder, often a polymeric material or the like, to combine metal powders into three-dimensional shapes without sintering or melting.38 In this technique, a feedstock material is bound together using a liquid binder to create an “intermediate” or green body structure of interest. In the case of metal systems, the feedstock consists of a metal powder of the final composition of interest. This material is distributed over the build surface of a moveable stage, and a printhead is then used to add liquid binder to form a layer of the geometry of interest. The stage is then moved away from the printhead and fresh feedstock material is placed on top of it; the next layer is formed in the same manner on top of the initial layer, and the process continues until the “intermediate” part is fully formed. In order to increase density and obtain a part of pure composition, the binder is then removed using thermal processing at a relatively low temperature (Li et al. cites a range of 175-450◦ C) to form what is termed a “brown body.”38 This material 9 often has a density below 70 % of metal parts in the total volume of the structure, and so it is often sintered to increase the part density. This involves heating at higher (> 450◦ C) temperatures for prolonged periods of time to encourage diffusion across grain boundaries and the closure of extant voids and pores. Particle size, feedstock packing density, and composition dictate sintering time and temperature. This process can yield a part with a density greater than 90 %, although highly dependent on the thermal profile and composition of the part of interest. Laser powder bed fusion refers to the use of a laser source to fuse metal particles into a three-dimensional monolith. First introduced by Deckard and Beaman in the mid-1980s, in the context of metal manufacturing this technique begins with a feedstock composed of metal particles of the final phase of interest.39 This material is steadily placed at the top of a loading cartridge, and a roller is used to evenly distribute the feedstock over the build surface as necessary. A moveable stage, to which the sample is attached, underlies this build surface, and a laser is used to fuse particles into the shape of one of the slices of the structure of interest. This stage is then moved downward, new material is distributed over the top of the freshly fused layer, and the process is repeated until the final structure is built. The layers are attached because the melted layer is deeper than the fresh feedstock layer alone, so the previously printed layer partially melts and fuses with the newest layer formed.40 This cycle repeats until the monolith is fully formed. Once formed and removed from the build surface, parts are often thermally treated, as in binder-jetting, to further densify the part. With regards to both of these techniques, a brief assessment of their strengths and drawbacks is in order. Both techniques generate parts in a range of compositions.38 Binder jetting suffers from comparatively low density, as the removal of the binder and subsequent sintering process can be prohibitively time and energy-consumptive when materials of interest have low diffusivities at high temperatures. Laser powder bed fusion, on the other hand, can produce parts with very high (99.9 %) densities without much postprocessing, as the feedstock material is directly fused together and the power of the laser involved in the process can be modulated to avoid the formation and trapping of bubbles in the fusion process.41 This advantage conceals a less objective benefit in the form of the microstructure produced by the laser fusion process; the laser process forms a melt pool in which particles are combined into a liquid mass that cools as the laser moves away. The rapid cooling process introduces unusual microstructural features, including fine grained and anisotropic 10 structures.42 Because binder jetting does not necessitate the formation of melt pools to fuse nearby particles into a single monolith, it often retains a microstructure similar to the feedstock material used in the binder jetting process, allowing for more equilibrium microstructures in the final part.43 Major limitations shared by both of these techniques include the spatial resolution of both of these methods, often limited to the laser spot size and particle shape of the feedstock, as well as the requirement of optimizing printing parameters anew for each composition of interest. These difficulties can be mitigated by instead pursuing vat photopolymerization techniques coupled with part postprocessing to form metal parts. Vat photopolymerization relies on the use of light to initiate polymerization of a resin solution, leading to the formation of a network and subsequent phase separation resulting in the 3D part of interest. This technique can be applied using laser light sources, often termed stereolithography, projectors, as in digital light processing, and lamps or light emitting diodes as light sources, among other sources.44 Stereolithography proceeds by rastering the laser in the resin in a given pattern so as to cross link the resin in a given configuration and form the structure of interest. In the projection or lamp based techniques, two-dimensional slices of the structure of interest are projected into the resin tank for given amounts of time to form the structure layer by layer.45 Both of these techniques include a monomer to form the bulk of the polymeric structure, a crosslinking agent including multiple moieties involved in the polymerization process to connect disparate chains, a photoinitiator to transduce the light signal to the chemical initiation of polymerization, and a photoblocker to prevent off-target polymerization due to scattered or diffuse light from the polymerization source.46 Because of the use of light sources, parts produced by stereolithography can have layer thicknesses around 10 𝜇m and in-plane resolution on the order of 25 𝜇m as printed. Vat photopolymerization methods historically focus on the formation of polymer parts, as the polymerization process almost invariably requires an organic or aqueous resin to proceed, but the combination of nanoparticle, complex, and ion loading with postprocessing techniques has allowed the extension of photopolymerization methods to the synthesis of oxide and metal structures, with even further improved spatial resolution due to part shrinkage.47,48 Metal additive manufacturing has and can been performed via vat photopolymerization.49 The superior spatial resolution provides one example of the advantage of vat photopolymerization over binder jetting or laser powder bed fusion for metal additive manufacturing, but the other involves the customizability of the composition of interest.44 While some previous methods for the synthesis of metal oxide and 11 metal parts using this technique required the reformulation of the resin and reoptimization per composition of interest, recent work in vat photopolymerization has demonstrated the use of polymer “blanks” to generate metal parts of a wide range of compositions using a single resin composition. This technique, the hydrogel infusion additive manufacturing method, takes advantage of the amphipathic properties of a given polymer to allow for organic resins during printing and hydrogel blanks during metal ion incorporation.49 1.4 Summary In sum, primary liver cancers, the majority of which are cases of hepatocellular carcinomas, represent a widespread and lethal cause of cancer. As hepatocellular carcinoma is characterized by an aggressive clinical course in tandem with frequent late diagnosis, curative therapies such as liver transplant and hepatic resection have limited applicability in most cases. Liver-directed therapies are the mainstays of treatment in cases diagnosed at a later stage, and in particular transarterial chemoembolization represents a standard treatment of intermediate stage hepatocellular carcinoma. In order to improve the side effect profile of this procedure, chemotherapeutic capture devices are under active investigation. Examples of such devices include DNA-functionalized lattices, ionic resin devices, and surface functionalized (sf)-magnetite nanoparticles. In the case of sf-magnetite nanoparticles, intravenous magnetic materials are needed to capture remnant nanoparticles, and due to the hemodynamic properties of the hepatic venous circulation, magnetic metal devices with highly architected structures are necessary. Additive manufacturing techniques are used to efficiently produce metal structures with the needed architecture, and while binder jetting and laser powder bed fusion represent well-known methods of metal additive manufacturing, photopolymerization methods incorporating postprocessing techniques offer flexibility in choice of composition as well as high resolution difficult to attain using other methods. The development of highly architected magnetic metal filters with favorable hemodynamics that are facile to manufacture represents a significant approach to improved treatment of primary liver cancers. 12 Chapter 2 METHODS The subsequent work discussed in Chapters 3 through 5 makes use of the following techniques to generate and characterize samples. The background and application of the techniques involved are discussed as follows with further elaboration as necessary; notably, chapter-specific content is discussed only where appropriate. In brief, samples are produced by following the hydrogel infusion additive manufacturing method. Phase and composition characterization are carried out using X-ray diffraction and energy dispersive X-ray spectroscopy, and magnetic properties are characterized by vibrating sample magnetometry. Magnetic nanoparticle capture assessment was attempted via inductively-coupled plasma mass spectroscopy, and COMSOL Multiphysics was used to estimate the capture of magnetic nanoparticles. 2.1 Hydrogel Infusion Additive Manufacturing The hydrogel infusion additive manufacturing method combines the architectural flexibility of vat photopolymerization with the compositional range of combustion synthesis. The method, as described by Saccone et al., relies on the use of an amphipathic polymer to physically separate the choice of final composition from the choice of architecture by printing an organogel “blank” structure and converting it to a hydrogel gel as a postprocessing step.49 The hydrogel is then submerged in a metal ion solution at an elevated temperature to promote diffusion, at which point the now-swollen gel is calcined in air to produce an oxide phase. The oxide is reduced using hydrogen to form a structure. The feature size of the final part produced is reduced as expected from the dehydration and combustion of the polymer. A schematic illustrating this process is found in Fig. 2.1. 13 Figure 2.1: Overview of the hydrogel infusion additive manufacturing method used to generate metal samples from organogel blanks. Figure adapted with permission from [49]. Elaboration of the specific chemical species used in the synthetic process and the choice of article is discussed in the sections to follow. Materials The following chemical species were used in the syntheses throughout this study: polyethylene glycol diacrylate (PEGda, molecular weights 575 and 700), N,Ndimethylformamide (DMF), Michler’s ketone, Irgacure 379, Sudan I, water, nickel(II) nitrate tetrahydrate, copper(II) nitrate trihydrate, copper(II) nitrate hemipentahydrate, iron(III) nitrate nonahydrate, cobalt(II) nitrate hexahydrate. Synthesis A photoactive precursor solution was synthesized by combining 7 mL of DMF with g of Michler’s ketone, g of Irgacure 379, and g of Sudan I. The 3D printing resin was synthesized by combining 35 mL of PEGda 575 or 700 and 28 mL of DMF with the photoactive precursor solution. This resin is termed BL5050 due to the approximately 50-50 volume ratio of DMF and PEGda in the mixture. BL5050 was used, in concert with the Autodesk Ember Digital Light Processing 3D printer, to print organogel blanks of the capture filter architecture of interest; in the work carried out in Chapters 3-5, the architecture of interest was a twisted honeycomb lattice with perforations. This structure, depicted in Fig. 2.2, was chosen on the basis of previous computational work demonstrating efficacious mixing of fluid in the structure with a limited drop in pressure across the length of the lattice to 14 optimize both the capture efficacy and hemodynamics of the lattice.32 Figure 2.2: A schematic of honeycomb lattices modeled for idealized chemotherapy capture by Maani et al. Note the twisted, perforated structure on the top right was chosen for these studies as the optimum between capture efficacy and hemodynamic stability. Figure adapted with permission from [32]. The parameters used to print the lattices are contained in Table 2.1. Once printed, lattices were soaked in DMF and subsequently water for two hours each and heated at 70◦ C to remove unpolymerized resin and form hydrogels. This step was performed in two ways; with or without replacement of the exchange media at the 1 h mark. Afterwards, the hydrogel lattices were soaked in 2M aqueous solutions for 48 h (unless otherwise noted) at 70◦ C. Swollen gels were blotted and subsequently heated in a single zone tube furnace under a 50 standard cubic centimeter/min flow of compressed air with pumping vacuum using one of the two heating profiles seen in Fig. 2.3. These calcined gels were subsequently heated under a 150 standard cubic centimeter/min flow of forming gas (95% N2 -5% H2 ) with pumping vacuum at 3◦ C/min to 900◦ C with a hold of 6 hours before ramping down to room temperature at 3◦ C/min. 15 Parameter Layer Thickness First Layer Exposure Time Model Layer Exposure Time Value 50 𝜇m 4-7 s 4-7 s Table 2.1: 3D Printing Parameters for BL5050 Resin on the Autodesk Ember. These parameters are generally identical to those found in [49], without an offset between the first two layers. Figure 2.3: A schematic of the profiles used to generate samples in the oxidation step. (a) The sample is ramped at 1 deg/min to 100◦ C, 0.25 deg/min to 500◦ C, 2 deg/min to 700◦ C, held for 3 h, then cooled at 2 deg/min to 20◦ C. (b) The sample is ramped at 0.25 deg/min to 700◦ C, held for 3 h, then cooled at 2 deg/min to 20◦ C. 2.2 X-ray Diffraction Phase Identification Overview and Historical Development X-ray techniques have been used to identify the crystalline structure of matter for well over one hundred years. The impingement of X-ray radiation on matter can result in many processes; one of the most important of these properties involves the scattering of X-rays from the electronic structure of the material. While this process can result in diffuse or incoherent scattering of X-rays with the generated data containing little useful information, regularity in atomic ordering leads to patterns of constructive and destructive interference between scattered X-ray photons that can be used to back-out the atomic ordering that resulted in the observed X-ray scattering. To make this more explicit, if a beam of X-rays is aimed at a single crystal, the resulting points of constructive interference resulting from the scattered X-ray beam represent a two-dimensional projection of the Fourier transform of the electronic structure of the crystal. If this projection is then reconstituted as a 16 three-dimensional object in reciprocal space, this represents the Fourier transform of the electronic structure from real space to reciprocal space. From this Fourier transform, the shape of the electronic structure can be uniquely identified and the atomic structure reduplicated on this basis.50 In practice, small atoms can be more difficult to place, isoelectronic species can muddle the characterization process, and the lack of perfect crystallinity limits the precision of the reduplicated atomic structure. Polycrystalline samples cannot be characterized in an identical fashion as the presence of multiple differentially orientated crystallites in a sample leads to the presence of reduplicated points in the reciprocal lattice due to the physical misorientation between adjacent crystallites, rather than the pattern of constructive and destructive interference that defines the projection in the single crystal case. Polycrystalline samples are still far more common than monocrystalline materials and still exhibit crystallinity within each crystallite domain. X-ray diffraction from these materials should still yield usable results. By altering the method of data collection, some of this information can be extracted. Observe a pair of parallel planes in a two-dimensional square lattice in Fig. 2.4. Figure 2.4: An illustration of X-ray photons diffracting off of a pair of planes in a square lattice, where 𝑑 is the spacing between the planes and 𝜃 is the angle at which the photons contact the plane. The extra distance traveled by the 𝑆2 photon is equivalent to the sum of 𝐴𝐵 and 𝐵𝐶. Figure adapted with permission from [51]. As a pair of X-rays impinge on the sample at an angle 𝜃, we can imagine the two photons impacting at adjacent positions in two parallel atomic planes separated by a distance 𝑑, with one of the photons traveling an additional distance 2𝑑 sin 𝜃 which 17 will introduce a difference in phase between the two photons of the same magnitude. In order to allow for completely constructive interference between the two photons, we must have that this difference in phase is an integer multiple 𝑛 of the wavelength of the radiation, say 𝜆. Putting this together geometrically, we find that: 𝑛𝜆 = 2𝑑 sin 𝜃 (2.1) also known as the Bragg condition for diffraction.50 With this relation we now have the basis for X-ray characterization of polycrystalline samples. Beginning with monochromatic radiation, the X-ray source can be manipulated such that the beam impinges on the sample at a range of angles 𝜃. At the same time, a detector can be rotated such that it aligns with an angle 2𝜃 with respect to the path of the impinging radiation. This is known as the Bragg-Brentano configuration for powder X-ray diffraction studies, as illustrated in Fig. 2.5. 18 Figure 2.5: A depiction of the Bragg-Brentano configuration for powder X-ray diffraction. The X-ray source at left is rotated through an angle 𝜃 while the detector at right is locked to a position 2𝜃 from the undisturbed path of the X-ray beam. Radiation filters are mounted in the detector assembly. Figure adapted with permission from [52]. A signal is detected only if the Bragg condition is satisfied at the current value of 𝜃; this is conditioned on the set of spacings {𝑑 ℎ𝑘𝑙 } of the planes of the sample. The resultant collection of signal intensities forms a one-dimensional pattern depicting the relationship between the measured intensity and the angle of impingement 𝜃 is generated. The set of spacings between parallel planes, termed interplanar spacings, is generally unique to a given crystal, and as such the pattern of measured intensities and angle of impingement resulting from a monochromatic X-ray source can be used to uniquely identify the structure of the material via reference to previously collected or computed patterns. This method relies on the polycrystallinity of the sample such that every plane with a unique interplanar spacing is assured to be parallel to the 19 surface of the characterized sample in a robust portion of the constituent crystallites. Phase confirmation is carried out using this method by comparing collected patterns with computed patterns from materials of known phase composition appropriate to the predicted phase composition of the sample of interest.53 Instrumentation and Experimental Specifications X-ray diffraction was carried out using a Rigaku SmartLab Multipurpose X-ray Diffractometer equipped with a copper K𝛼 radiation source of wavelength 1.54 nm and nickel K𝛽 filters. Samples were variably characterized with or without a scattered radiation protector, and in monolithic form for metal alloy samples and powder or monolithic form for oxide or microcomposite samples. Collection was performed with X-ray fluorescence suppression via the HyPix 3000 high-energyresolution detector. Samples were compared to powder diffraction patterns computed from crystallographic information files contained in the Inorganic Crystal Structure Database.54 2.3 Energy Dispersive X-ray Spectroscopy Characterization Overview Phase identification of a material via X-ray diffraction provides information about the average crystallographic composition of a given material. In addition to this information, the elemental composition of a given material is necessary to inform the understanding of the material properties of interest including the predicted and measured magnetic properties. In other methods of manufacturing, elemental composition can be more easily ascertained by precise measurement of the constituents of the alloy of interest or the concentration of the ionic species in the solutions of interest that are then used to form the final product. In contrast, the HIAM method produces swollen gels with compositions different that of the initial swelling solution, and as such the produced metals need to be characterized so as to determine the composition of the final sample. Energy dispersive X-ray spectroscopy (EDS) can be used to characterize the composition of these metal parts.55 This technique relies on a side-process to the electron-matter interaction underlying scanning electron microscopy (SEM); electrons can impinge on atoms so as to dislodge tightly bound core electrons from the sample of interest. This process excites the atom of interest into an unstable high-energy configuration, which is stabilized by the replacement of the inner shell electron with an outer shell electron along with the concomitant release of X-ray radiation. This process is sketched out in the schematic in Fig. 2.6. 20 Figure 2.6: An illustration of the atomic process underlying EDS. Note that the precipitating excitation can occur as a result of the impingement of radiation or particles upon the atom; the latter process governs EDS as performed in an electron microscope as in this study. Figure adapted with permission from [56]. Because the energy of these orbitals is determined partially by the atomic number of the atom itself, this information can be catalogued, and the measured X-ray radiation released from a sample bombarded with electrons can then be cross-referenced with this catalogue to identify the source atom. By measuring the intensity of the emitted radiation at each wavelength of interest, the atomic composition of the sample can be determined, which provides insight into the atomic composition of the metallic samples. In addition to this insight, by focusing the impinging electron beam at a given point of the sample at a time, a map of the distribution of elements in the sample can be made, which allows for qualitative identification of elemental segregation.57 EDS is less insightful as to the absolute amount of low atomic number elements in the sample of interest, which in this case concerns the common contaminants carbon and oxygen, but by taking the ratio of the metal atoms to each other we can use this as a benchmark for the metal atom ratio in the sample as a whole. This technique is applied throughout the work that follows to characterize the atomic composition of samples as well as the presence or lack thereof of obvious elemental segregation in the sample. 21 Instrumentation and Experimental Specifications EDS maps were collected using the Bruker Quantax XFlash 6|60 detector as installed onto a Versa 3D DualBeam scanning electron microscope. Maps were collected for 300 seconds at an accelerating voltage of 20 kV; this was chosen as it was twice the magnitude of the energy of the characteristic radiation of the elements of interest–namely copper (8.040 kV), nickel (7.471 kV), iron (6.398 kV), and cobalt (6.924 kV)–rounded to the nearest 10 kV. 2.4 Vibrating Sample Magnetometry Overview and Experimental Setup Classical characterization of the magnetic properties of a material involves the measurement of the magnetism induced by the sample while imposing a magnetic field of varying strength onto the sample itself. This is a challenging requirement for various reasons; specifically, the ability to measure the magnetism due to the sample itself versus the imposed magnetic field of the measurement instrument. To overcome this difficulty, the vibrating sample magnetometer was invented at the MIT Lincoln Laboratory in 1956 by S. Foner.58 This instrument makes use of Lenz’s law (see Eqn. 2.2) to measure the magnetism due to the sample undergoing characterization in the simultaneous presence of an applied magnetic field–this was accomplished by placing the sample on a rod and oscillating the sample through a pickup coil in the middle of a magnetic field produced by a surrounding electromagnet.59 𝜀=− 𝑑Φ𝐵 𝑑𝑡 (2.2) The applied magnetic field magnetizes the sample, but as it is held steady through the timeframe of the measurement itself, it does not change the magnetic flux through the pickup coil and does not contribute to the electromotive force produced in the coil. On the other hand, the physical motion of the magnetized sample through the coil produces a magnetic flux and subsequently an electric current in the pickup coil. A reference coil removes the potential effect of the sample rod from the measurement. Refer to Figure 2.7 for a schematic of the instrument setup. 22 Figure 2.7: The illustration of the VSM apparatus as invented by S. Foner in 1956. In this system, vibration is accomplished by mounting the sample (label 5) to a sample holding rod (label 3) attached to a loudspeaker transducer (label 1). A magnetic field is applied by an electromagnet (label 8) and the output signal is measured from the sample by one set of pickup coils (label 7) and another set of coils (label 6) are used to measure a reference sample response (label 4). Figure adapted with permission from [58]. The general approach to measurement involves applying a powerful magnetic field to the sample of interest in order to magnetize and erase any previous magnetic memory from past exposure to magnetic or electric fields. As measurements are taken on the magnetism induced by the sample, the applied field is decreased stepwise until it is of equivalent magnitude in the opposite direction, at which point it is increased to the initial applied field strength and direction. In a ferromagnetic material, this procedure traces a loop known as a hysteresis loop when the measured magnetic moment is plotted versus the applied field, as the retained magnetic ordering at zero 23 applied field results in a nonzero measured magnetic moment known as the remnant magnetization, and the resistance to demagnetization results in a nonzero value of the applied field at which measured magnetic moment drops to zero known as the coercivity. These properties help determine the suitability of a material for use as a permanent magnet, and are benchmarks for the performance of a magnetic material. Instrumentation and Experimental Specification Hysteresis loops were measured at 293 K using a Quantum Design Physical Property Measurement System. Samples were placed under a 5000 oersted magnetic field, at which point the field was swept to –5000 oersteds and returned to 5000 oersteds. In some cases, the field was swept to –5000 oersteds once more to assure complete closure of the loop due to sample movement during initial demagnetization. Coercivity was computed as the average of the values of the field between which magnetic moment changed sign in each sweep of the loop. Measured magnetic moments were converted to magnetic flux density values using the following expression: 𝐵= 4𝜋𝜇𝐷 𝑚 (2.3) where 𝜇 is the measured moment output from the instrument, 𝑚 is the sample mass, and 𝐷 is the density of the sample. 𝐷 is estimated for samples of multiple components by using the molar volume per component weighted by the atomic ratio as determined by EDS. 2.5 Inductively-Coupled Plasma Mass Spectrometry Overview Characterization of nanoparticle concentrations in solution requires identifying the number of particles in the solution, the size of the particles, and identifying the species present in the nanoparticle. Inductively coupled mass spectrometry allows for this characterization by utilizing a plasma to vaporize nanoparticles and characterizing the resulting particle gas by mass spectrometry.60 The technique begins with an appropriately dilute solution of nanoparticles that are introduced into the flow via a nebulizer that converts the solution to droplets suspended in a carrier gas. The nanoparticle-laden aerosol is then carried into a device which surrounds the nanoparticle aerosol with argon gas used to form the plasma, at which point the nanoparticle aerosol is dehydrated and decomposed into individual ions of the constitutive atoms. In the case of nanoparticles, the usage of a narrow-diameter tube to carry the aerosol helps assure vaporization of the nanoparticle into its constituents 24 atoms. This highly charged stream of particles then passes through a series of fine ports to separate the species under analysis from the plasma species. The remainder of the instrument comprises a mass spectrometer that then identifies species by their mass-to-charge ratio.61 With an adequately dilute solution, nanoparticles can be individually studied in the mass spectrometer. Using the frequency of nanoparticle events per window as determined by the output of the spectrometer, the number concentration of the inputting solution can be backed out. The signal of the mass spectrometer can also be used to determine the absolute amount of material analyzed within a given window, yielding a mass estimate of the initial nanoparticle, and spherically modeling an individual nanoparticle yields a size estimate of the nanoparticles in the aerosol. Nanoparticles can also be digested to yield mass concentrations without associated nanoparticle size and count distribution analysis as the instrument measures single ion interactions in the mass spectrometry. Instrumentation and Experimental Specification Samples were run on an Agilent 8800 Triple Quadrupole ICP-MS with a 1.5 mm fused quartz torch and helium as a carrier gas. Reference samples used to calibrate the instrument included an ionic blank made with Type II water and 1% HNO3 /1% HCl, an ionic standard made with Type II water with 1% HCl and gold and iron nitrate reference material at 1 ppb concentration, and nanoComposix bare gold nanoparticles of diameter 30 and 60 nm. 2.6 Summary The preceding discussion outlined the synthesis and characterization methods used to generate and understand the morphology, composition, and magnetic properties of the metal structures discussed in Chapters 3-5, along with understanding the capture of nanoparticles by the resultant lattices. The hydrogel infusion additive manufacturing method is applied to a geometry optimized for capture and hemodynamics in a previous study by Maani et al to generate metal lattices in a twisted honeycomb geometry.32,49 Powder X-ray diffraction was used to characterize the structural composition of the metal lattices by determining the constituent crystal systems in a given sample, while energy dispersive X-ray spectroscopy was used to characterize the molar composition of a given lattice by determining the ratio of the metal atoms of interest in the sample. Vibrating sample magnetometry was applied to synthesized lattices to generate hysteresis loops from which properties such as the magnetic remnance, saturation magnetization, coercivity, and maximum energy 25 product could be derived. Finally, inductively-coupled plasma mass spectrometry was used to characterize the concentration of nanoparticles in solutions exposed to the metal lattices of interest. The results of these processes and techniques to generate metal lattices for chemotherapy capture filters and the results of characterization are explored in depth with regards to iron, iron-nickel, and copper-nickel-iron lattices in the chapter to follow. 26 Chapter 3 ADDITIVE MANUFACTURING OF IRON AND IRON-ALLOY LATTICES FOR MAGNETIC NANOPARTICLE CAPTURE The author thanks Colin Yee of the UCSF Interventional Radiology Research Lab for assistance in designing the filtration testing system as well as Andrii Syrota for assistance with computational studies and plotting of simulation results and Dr. Nathan Dalleska for assistance with nanoparticle ICP-MS. This chapter describes the synthesis and characterization of iron and iron-alloy lattices for capture of magnetic nanoparticles conjugated with or bound to chemotherapeutics used in the TACE procedure. The bulk characteristics of the lattices are described and analyzed with reference to figures of merit from the literature and attempts to characterize the performance of the lattices are discussed, along with simulation studies of the efficacy of nanoparticle capture by the lattices under a range of conditions. 3.1 Introduction to Transarterial Chemoembolization As described in Chapter 1 Section 1, primary liver cancers are often diagnosed at an inoperable stage, and as such liver-directed therapies are the standard treatment option in these cases. Transarterial chemoembolization is of special importance in this regard as it is a standard-of-care for intermediate stage hepatocellular carcinoma per meta-analyses of the literature and the only such treatment to have a survival benefit. This procedure targets the blood supply of the tumor to provide high local doses of chemotherapy while starving the tumor of oxygen. The following discussion touches on the development of this technique and its potential sequelae when used to treat HCC. Catheter Methods for Targeted Therapy Catheter methods are applied to target treatment to specific organs as opposed to therapies requiring systemic infusions or invasive operations. Methods such as vascular stenting, embolization, and local radiotherapy represent the utility and superiority of these methods when compared to systemic or invasive alternative approaches. Percutaneous coronary intervention, or the stenting of stenosed coronary arteries, is the standard of care for acute myocardial infarctions and provide a substantial 27 survival benefit over such alternatives as thrombolytic therapy, which is associated with serious bleeding events.62 Embolization of cerebral aneurysms can prevent devastating strokes while avoiding some of the risk involved in cranial surgery whether from damage to the tissues of the brain or infection.63 Finally, local radiotherapies in the form of catheter administration of radioisotopes to a tumor of interest provide a significant survival benefit and the potential for cure when malignancies are small whilst avoiding much of the systemic side effects introduced by external irradiation of bodily tissues.15,64 While just a sample of the catheter-based therapies utilized in modern methods, these techniques demonstrate the potential benefits stemming from their use instead of a systemic or invasive procedure. Transarterial Chemoembolization Transarterial chemoembolization (TACE) was introduced into the world of HCC treatment due to an inability to utilize curative therapies such as radioablation or surgical resection and the inadequacy of systemic chemotherapy in controlling intermediate or advanced stage disease.15,23 The roots of the procedure were found in past knowledge of the redundancy of the liver blood supply and the use of embolization to obliterate portions of the vasculature to stem hemorrhage or neutralize aneurysms. While the initial treatments involved simple embolization of branches of the hepatic artery leading to the tumor, the introduction of chemotherapeutics and carrier oils increased survival significantly which led to the adoption of the combination of embolic and chemotherapeutic agents as well as carrier substances.22,23 The TACE procedure managed to confine both the hypoxic effects of embolization as well as the cytotoxic effects of the administered chemotherapy to the liver, preventing much of the adverse effects of systemic chemotherapy. An illustration of the procedure can be seen in Fig. 3.1. 28 Figure 3.1: An illustration depicting the fundamentals of the TACE procedure. The doxorubicin formulation includes a suspending oil and potentially drug-eluting beads or gelation agents for embolization. Figure adapted with permission from [65]. 29 Systemic Side Effects The release of remnant chemotherapeutics into the systemic circulation poses a continuing challenge for the TACE procedure and its expanded applicability. With existing knowledge of dose-dependent activity of chemotherapeutics, as well as dose-dependent off-target toxicity, attempts to increase significantly the local dose of chemotherapy would require the prevention of the escape of drugs from the liver. Chemotherapeutics used for the TACE procedure include platinum-based agents such as carboplatin or cisplatin as well as anthracyclines such as doxorubicin.15 Platinum-based agents are significantly nephro- and ototoxic, and anthracyclines are associated with cardiotoxicity, with worsened effects correlated to higher systemic doses. In addition to these focused toxicities, the escape of drugs from the liver in the TACE procedure manifests in increases in nausea and vomiting as well as damage to the bone marrow resulting in hematological abnormalities.27,66 This potential for chemotherapy escape from the liver in the TACE procedure both worsens the side effect profile of the procedure itself while also limiting the amount of drug used in the procedure due to the dose-dependent nature of the toxicity. 3.2 Magnetic Lattice Synthesis and Characterization Structural and X-ray Diffraction Characterization The lattices discussed herein were synthesized following the HIAM process discussed in Chapter 2 Section 1, using a 24 h swell-in time and profile (a) in Fig. 2.3. The iron sample utilized a swell-in solution consisting of 99.5 mol% iron and 0.5 mol% nickel, the nickel-iron sample a swell-in solution consisting of 80 mol% nickel and 20 mol% iron, and the copper-nickel-iron sample a swell-in solution consisting of 60 mol% copper, 20 mol% nickel, 20 mol% iron. Powder diffraction was performed on flattened metallic monoliths and magnetic properties testing was performed on bisections of flattened metallic monoliths. The first set of samples synthesized via the HIAM procedure to explore the possibilities of magnetic particle capture were the lattices of the iron system. In this system, a small amount of nickel was added as a sintering aid after initial exploration demonstrated a poorly sintered microstructure for the iron lattices.67 With the incorporation of 0.5 mol% nickel to the swell-in solution, the iron lattices were synthesized as seen in Fig. 3.2. 30 Figure 3.2: This diagram uses the iron-dominant sample to outline the HIAM process from exchanged hydrogel gel to swollen metal-ion hydrogel and subsequently the calcined and reduced samples. The scanning electron microscopy image provides a higher resolution image of the sample features. By tracking the size of the major face of the twisted honeycomb architecture of the swollen polymer gel, estimates of the shrinkage of the lattices from the swollen gel phase to the final metal product were made using FIJI. The images used to measure this change are depicted in Fig. 3.3 as follows: 31 Figure 3.3: A demonstration of the transformation of the (a) swollen samples to (b) calcined oxides and finally (c) reduced metal samples. 32 As seen in Appendix A.1, the lattices shrink from an average of 14.96 mm in diameter for swollen gels to 3.22 mm in diameter for metal honeycombs, approximately 21.70 % from start to finish. This means that printed gels need to be scaled up by approximately five times the desired final part diameter. The nickel-iron and copper-nickel-iron systems, chosen for their, respectively, low and high coercivity as an expression of tunability of magnetic properties in the HIAM process, are displayed in Fig. 3.4.68,69 Figure 3.4: Gross structural images of nickel-iron and copper-nickel-iron lattices. (a) A photograph of the nickel-iron sample and the (b) copper-nickel-iron sample as well as a scanning electron microscopy image of the overall structure and an inset highlighting one pore for each sample. 33 The size and distribution of the pore and beam sizes in the iron, nickel-iron, and copper-nickel-iron systems were measured using FIJI and can be found in Appendix A.1. In order to assess the phase composition of the as-synthesized lattice, powder diffraction was performed as shown in Fig. 3.5: Figure 3.5: Powder diffraction patterns of the reduced (a) iron-dominant (b) nickeliron and (c) copper-nickel-iron samples along with relevant simulated patterns for reference materials, as calculated from CIFs obtained from the following references via the ICSD: [70–72]. Iron at room temperature is expected to display a body-centered cubic crystal structure, while nickel-iron and copper-nickel-iron compositions somewhat comparable to the ideal 80:20 ratio for permalloy and 60:20:20 ratio for copper-nickel-iron should both exhibit face-centered cubic crystal structures.68,69 All three of these expectations hold in this case, as revealed by the comparison to simulated patterns from the scientific literature in Fig. 3.5. The intensity pattern of the iron sample is peculiar, which instigated an attempted exploration of the internal microstructure of the as-synthesized sample to assess the likelihood of a retained oxide phase internal to an outer metal layer, the results of which are displayed in Fig. 3.6. 34 Figure 3.6: Cross-sectional studies of an iron lattice. (a) An EDS map of the focused ion beam milled node of the iron-dominant sample shown in Fig.3.2, along with the maps corresponding to the iron (red) and oxygen (blue) elements (b) Scanning electron microscopy image and phase map electron backscatter diffraction of the iron-dominant sample. The first step in these explorations involved the milling of a cross-sectional window into the node of the iron lattice. From there, EDS was performed and the oxygen and iron maps were compared in order to identify any potential enrichment of oxygen inside of a beam, which could indicate the presence of a trapped iron oxide. Because these results, see in Fig.3.2 part (a), were fairly inconclusive, and the unreliability of quantitative estimates for the content of low atomic number elements in EDS is a well-known phenomenon, this microstructural question was subsequently attacked via backscatter diffraction methods. In part (b), a polished sample was imaged with a scanning electron microscope equipped with a detector for backscattered electrons which allowed for diffractive characterization of the sample in a local environment and at a scale unattainable by traditional diffraction methods.73 A phase map of the internal surface of the beam was generated, demonstrating the presence of body-centered cubic iron, hematite, and magnetite and compared to the scanning electron image, at which point a correlation was observed between the presence of differences in charging between the area assigned to iron and that assigned to magnetite. The fact that the magnetite regions were completely surrounded by iron indicated that these were the suspected trapped oxides. The regions assigned to hematite were dismissed as instrumental artifacts considering the low conductivity of hematite versus either iron or magnetite.74 The hydrogen reduction of hematite to iron proceeds stepwise through a magnetite intermediate, which lent further support to the identification of these regions as magnetite inclusions.75 A summary of the microstructural characterization of the three phases synthesized, along with the size 35 of the beams and pores in each phase, can be seen in Fig. 3.7. Figure 3.7: A summary diagram of the three categories of samples discussed in this sample along with visualizations of the beam and pore sizes and the powder diffraction patterns corresponding to the three systems. The simulated patterns are calculated from CIFs obtained from the following references via the ICSD: [71, 72]. Energy Dispersive X-ray Spectroscopy Composition Analysis In order to understand the relationship between the composition of the swell-in solution and the final metal part for metal alloys in these two compositions, EDS was performed at ten sites distributed throughout the nickel-iron and copper-nickeliron honeycombs, and the metal composition at all sites were averaged to yield an estimate for the overall metal composition of the sample as in Fig. 3.8. 36 Figure 3.8: Visualizations of the ten regions from which EDS maps were collected for the (a) nickel-iron and (b) copper-nickel-iron systems, along with the average composition obtained from this collected data. See Appendix B for more information. These results from EDS exhibit enrichment of iron in both the nickel-iron and coppernickel-iron samples, which could indicate increased affinity between iron and the polymer chains of the gel. This differential affinity would necessitate increased quantities of non-iron metals in the swell-in solution in order to synthesize samples with the potential to behave similarly to permalloy and CuNiFe, respectively.76 In order to assess if the location of the EDS map affected the measured composition, measurements from the peripheral (010) and central (001) sites were compared in both the nickel-iron and copper-nickel-iron systems, as described in Fig. 3.9. 37 Figure 3.9: The computed nickel-iron ratio in the (a) central (001) and (b) peripheral (010) locations of the nickel-iron system and the (c) central (001) and (d) peripheral (010) locations of the copper-nickel-iron system. The very close match between the computed compositions in both of these regions provides evidence as to the lack of difference between measurement at the periphery of the structure versus the center. Given that the shortest diffusion path into the polymer gel is from the surface inward as opposed to a path from node to node, and that all beams are bathed in the swell-in solution throughout the swell-in process, it is unlikely that there would be large differences due solely to the location of a beam or node in the greater gel monolith. Magnetic Property Comparison Magnetic properties of the as-synthesized lattices were assessed by recording a magnetic hysteresis loop for each sample using a vibrating sample magnetometer as discussed in Chapter 2. These hysteresis loops were converted from units of magnetic moment to magnetic flux density by utilizing the EDS determined compositions for the nickel-iron and copper-nickel-iron phases to estimate the sample density in order to apply Equation 2.2. The remnance, coercivity, and maximum energy product were subsequently determined as described in Appendix E, and Fig. 3.10 depicts both these hysteresis loops as well as the magnetic properties of interest: 38 Figure 3.10: Vibrating sample magnetometry curves for the iron-dominant, nickeliron, and copper-nickel-iron systems including an inset focusing on the center of the curves and illustrating the coercivity and remnance of the systems, along with a table demonstrating the saturation and remnant magnetization as well as the maximum energy product and coercivity of the depicted sample systems. The coercivity of all three samples is low, with the copper-nickel-iron sample having a coercivity approximately twice that of the other compositions, potentially owing to off-target compositions as well as impurities and a lack of postprocessing of as-synthesized magnets; soft iron magnets often have coercivities near 1 Oe.77,78 Copper-nickel-iron magnets of the proper composition can have coercivities in the hundreds of oersteds.68 Remnant magnetization of the samples is not unusual for magnets of these compositions, although not particularly high, and the maximum energy product values, with magnitudes in the low kG-Oe, are clearly inadequate for most permanent magnet applications; neodymium-iron-boron magnets can have energy products in the MG-Oe range.79 3.3 In vitro Filtration Testing System Device Description Before assembling the capture device, the metal lattices were prepared as “assynthesized" or “magnetized" lattices as appropriate. In the “magnetized" lattices, the samples were placed in a 1.5 Tesla magnetic field generated by an electromagnet for five minutes, at which point they were removed and the device was assembled as soon as possible. The capture device setup involves heat-shrinking a metal lattice 39 in between two pieces of catheter sheath tubing. This device is then attached to a plastic connection joint with a valve that is then secured via heat-shrink wrap to a syringe loaded with magnetic nanoparticles suspended in phosphate-buffered saline (see Appendix C for details). A syringe pump is then used to pass the solution through the device at a constant volumetric flow as seen in Fig. 3.11. Figure 3.11: Experimental particle capture system. (a) A schematic demonstrating the essential parts of the particle capture testing system along with a (b) photograph of the real-life system. Aliquots of the magnetic nanoparticle suspension were taken from the solution before loading into the syringe and after passing through the testing system. In order to compare the concentration of nanoparticles as well as characterize the distribution of nanoparticle sizes in these aliquots, assessment by inductively coupled plasma mass spectrometry, as described in Chapter 2 Section 5, was attempted. Exploration of Particle Capture by Inductively Coupled Plasma Mass Spectrometry Before attempting to characterize nanoparticle solutions via inductively coupled plasma mass spectrometry, an attempt was made to ascertain the possibility of nanoparticle capture in metal lattices entirely. An as synthesized metal lattice was 40 placed in the setup described in Fig. 3.11, and was subsequently imaged via scanning electron microscopy after the magnetic nanoparticle suspension was passed through it. The results of this imaging can be seen in Fig. 3.12. Suggestive Evidence of Particle Sequestration Figure 3.12: Images of post-nanoparticle-capture iron-dominant samples along with EDS maps demonstrating a difference in iron counts between the rounded particles on the surface of the sample and the bulk sample, along with the associated EDS spectrum demonstrating the presence of solely iron and oxygen in the imaged area. In these images, the presence of several small, roughly spherical particles were noted on the surface of the lattice. When EDS was performed on these particles, iron and oxygen were the only detected species, and the particle demonstrated a qualitative enrichment in oxygen over iron. This was taken as potential evidence of particle sequestration by the lattice, although several questions remained regarding this particle. Its size was not appropriate when compared to the expected 50-100 nm diameter of the initial magnetite nanoparticles from which the nanoparticle suspension was synthesized, although the ferromagnetic nature of these particles often leads to rapid aggregation in solution and the free surfaces of the lattice might have provided a useful stimulus to ensure such aggregation. While these images were taken as support for the idea of nanoparticle capture by these lattices, in order to quantify the capture or lack thereof of magnetic nanoparticles by these lattices, inductively coupled plasma mass spectrometry was attempted on acid digested nanoparticles. The results of digested solutions passed through magnetized and nonmagnetized lattices can be seen in Fig. 3.13, along with a calibration curve for digested nanoparticles measured in the same run: 41 Figure 3.13: Plots depicting the measured nanoparticle concentration in digested aliquots taken after running a magnetite nanoparticle solution of 0.2 mg/mL concentration through metal lattices in the capture setup. In (a) nanoparticle solutions of concentration 0.05-0.20 mg/mL were diluted, digested, and run in the system with the resultant 56 Fe signals recorded versus undiluted concentrations along with a linear fit to the data with the displayed equation and 𝑅 2 , while in (b) solutions of initial concentration 0.20 mg/mL that had been passed through the experimental setup using three magnetized and three nonmagnetized lattices were diluted, digested, and run and the resultant 56 Fe signals plotted as the blue dots. Red triangles display the average signal recorded for each lattice condition which is displayed numerically along with the sample standard deviation of the signals recorded. Plot (a) in Fig. 3.13 demonstrates a calibration curve of the nanoparticle solutions and the 56 Fe signal recorded by ICPMS, but the 0.2 mg/mL solution tested had an unexpectedly low value, resulting in a suboptimal R2 value of 0.72 and giving a wide range for which measured signals may be concordant with a given nanoparticle solution concentration. The dilution range of our samples was such that we would otherwise expect a linear fit to the recorded data in this concentration range, and as such this decreased signal may have been due to poor sampling of the nanoparticle mixture before dilution and digestion. In plot (b) in Fig. 3.13, we see measured signals from nanoparticle solutions that were flowed through magnetized and nonmagnetized Fe»Ni lattices in the experimental setup shown in Fig. 3.11. We note that the average measured signal in both systems is below that of a nanoparticle solution of concentration 0.2 mg/mL, whether compared via the measured signal or the expected signal according to our calibration curve. Notably, no magnetized lattice solution had a signal post-flow greater than that of the measured 0.2 mg/mL 42 nanoparticle solution. This provides stronger absolute evidence of nanoparticle capture in the magnetized lattices. In the nonmagnetized lattices, on the other hand, the average signal is much closer to the measured signal of the 0.2 mg/mL nanoparticle solution, which weakens evidence of nanoparticle capture in these lattices. With less strict conditions, we can apply the linear fit to the data in Fig. 3.13 plot (a) to the measured signals for the magnetized and nonmagnetized capture lattices, to obtain estimated average concentrations of the post-capture nanoparticle solutions of 0.10 mg/mL for magnetized lattices, demonstrating a capture efficiency of 49 ± 16 %, and of 0.14 mg/mL for nonmagnetized lattices, demonstrating a capture efficiency of 31 ± 10 %. One can appreciate the difference between these capture efficiencies while also noticing the significant overlap between both magnetized and nonmagnetized data, which indicates that much of the capture accomplished by these lattices is due to the flow dynamics in the system as opposed to lattice magnetism. This would indicate that the low remnant magnetization of the lattices may be too small to accomplish effective active capture, or that the low coercivity of the samples results in rapid decay of permanent magnetism due to exposure to various weak magnetic fields from unshielded electronics or the magnetic field of the Earth. 3.4 COMSOL Multiphysics Capture Simulations Modeling the particle capture process in these lattices was performed using the COMSOL Multiphysics software package with the AC/DC Module. The constitutive equations for the modeling process were as follows: 𝜌(u · ∇)u = ∇ · [−𝑝I + 𝜇[∇u + (∇u)𝑇 ]] + F ∇(𝜌u) = 0 B = 𝜇0 𝜇𝑟 H + B𝑟 ∇·B=0 H = −∇𝑉𝑚 𝜏𝑝 𝑑q =u+ (F𝐷 + F𝑚𝑎 𝑝 ) 𝑑𝑡 𝑚𝑝 F𝑚𝑎 𝑝 = 2𝜋𝑟 3𝑝 𝜇0 𝜇𝑟 ∇H2 F𝐷 = 1 𝑚 𝑝 (u − v) 𝜏𝑝 (3.1) (3.2) (3.3) (3.4) (3.5) (3.6) (3.7) (3.8) where equations (3.1) and (3.2) are the Navier-Stokes equations and the continuity equations, equations (3.3), (3.4) and (3.5) represent the magnetostatic equations, and equations (3.6), (3.7), and (3.8) represent the equation of motion for a particle and 43 the form of the magnetophoretic and drag forces acting on a ferromagnetic particle. By analogy with a previous study, gravitational forces and the force of inertia are neglected due to their small magnitude in the context of micrometer scale magnetite particles, and so the motion of the particles in the fluid is governed entirely by the magnetophoretic and drag forces acting upon it.80 The system is then modeled as a twisted, perforated honeycomb in the middle of a tube of water of equal diameter to the honeycomb, suspended within a ball of air. The mesh and twisted honeycomb geometry can be seen in Fig. 3.14. Figure 3.14: The mesh underlying the capture simulations performed here including the (a) visualization of the sphere of air containing the vessel and the (b) twisted, perforated honeycomb geometry of the single capture lattice as well as the (c) triple capture device. With thanks to Andrii Syrota for generating these structures. In order to simulate capture using lattices comparable to those synthesized in Fig.3.2, the size of the lattice was made comparable to experimental lattices with a radius of 1.66 mm and a length of 1.91 mm, and the simulated vessel was given a radius of 1.72 mm and a length of 17 mm surrounding this lattice. An inflow velocity comparable to that found in the hepatic circulation was assigned at 1.6 m/s unless specified otherwise. The fluid density and dynamic viscosity were set equal to that of water at 0.997 g/cm3 and 10−3 Pa·s. Finally, magnetic flux densities of 0.01 to 1.45 T were used in simulations as needed to approximate the magnetic flux density in the iron (Fe»Ni) lattices for the former values and more permanent magnetic lattices for the latter values. Capture efficiency in these simulations was calculated 44 as the number of particles retained in the lattice, 𝑁 𝐿 , divided by the number of particles released in the flow, 𝑁𝑇 , as follows: CE = 𝑁𝐿 𝑁𝑇 (3.9) The magnetic field lines around a lattice magnetized in a direction parallel to the axis orthogonal to the major face of a single honeycomb lattice are depicted in Fig. 3.15. Figure 3.15: A map of the magnetic flux density within and surrounding a magnetic lattice with an overall magnetic flux density of 1.45 T. The traced lines demonstrate the general orientation of the magnetic field generated by the lattice in the free volume surrounding the lattice. The scale at the right demonstrates the intensity of the flux density in Tesla at each point in the lattice and in the free volume surrounding the lattice. With thanks to Andrii Syrota for generating this plot. This visualization of the magnetic field lines resultant from the magnetization of the lattice demonstrate the effect of perforations on the intensity of the magnetic field near the surface of the lattice; namely, these perforations correspond to increased intensity of the magnetic field. As the particle solution washes over the lattice, the particles are acted upon with more force at these regions, which should encourage 45 additional capture of the particles in question at these regions. This would imply a more stepwise pattern for particle capture in these systems, as the perforations are more evenly spaced along the lattice and align in the same plane with other perforations. Before this can be assessed, an understanding of the development of the flow with a velocity of 1 m/s as it encounters the magnetic lattice and progresses through it should be developed, and a plot depicting this development is shown in the following figure: Figure 3.16: This plot shows the radially averaged development of the flow as it passes into the twisted, perforated honeycomb magnetic lattice. A curve with label Distance = −𝑑 represents the radially averaged velocity profile of the flow a distance 𝑑 from the major face of the lattice closest to the origin of the flow. At the right, a figure demonstrates the axis 𝑍 which shows how negative distance is related to the direction of the inflow 𝑢 from the proximal magnetic lattice face at 𝑍 = 0. The flow breaks down from the smooth parabolic state it begins with to a jagged series of jets passing into the honeycomb pores as it encounters resistance from the beams of the lattice. With thanks to Andrii Syrota for generating this plot. The flow profile is abruptly interrupted as the flow impacts the lattice, and flows at fairly uniform velocities in each local region without transforming into a welldeveloped flow once in contact with the lattice. This may negatively impact capture of particles as the lack of a parabolic fluid velocity in each channel means a higher speed of the fluid at the walls of that channel and a more limited opportunity for the magnetic field to alter the path of the particles and allow for capture from the flow; a more developed flow, with slower fluid flow at the walls of the channels, 46 would likely allow for greater particle capture as the force required to accelerate particles into the wall of the lattices before the fluid exits the lattice would be lower. With the depictions of the magnetic field lines in the fluid in Fig.3.15 and the development of the flow as it nears the entrance of the lattice in Fig.3.16, the results for particle capture simulation can be adequately discussed in context. A depiction of the distribution of captured particles along the length of one filter with two sets of experimental parameters can be seen in Fig. 3.17: Figure 3.17: Simulation results displaying the number of particles captured along the axis of a single lattice orthogonal to the main honeycomb face with a velocity of 1.6 m/s and an assigned magnetic flux density of 1.45 T (blue) with capture of 50.6% of the particles and alternatively an assigned magnetic flux density of 0.028 T (red) with capture of 30.4% of the particles. With thanks to Andrii Syrota for generating this plot. In both of these systems, a significant portion of particles are captured when the flow comes in contact with the surface of the lattice. From there, the lattice with the stronger magnetic field shows a greater tendency to capture particles at every other position in the lattice, likely as a consequence of the greater magnetic force 47 imparted upon the particles by the magnetic lattices. In both systems one also notes the apparent step-wise pattern of particle capture past the initial surface capture. This is likely a result of the increased magnetic field strength at the perforations of the lattices in question, as observed in Fig. 3.15. These results appear to demonstrate that much of the increased impact of a stronger magnet in the capture of magnetic nanoparticles from a flow with a given velocity arises from the increased ability to redirect the path of nanoparticles suspended in the flow as it travels through the lattice past the initial surface in contact with the flow. If the magnetization of a lattice is too low to appreciably alter the flow the particles internal to the lattice, it will likely yield similar results to a nonmagnetized or control lattice. From here, the flow rate of the particle containing fluid as well as the magnetization of the lattices were varied to better sketch out the relationship between these two variables and the capture efficiency of the lattice as a whole. A plot of these results is depicted in Fig. 3.18 as follows: 48 Figure 3.18: Simulation results displaying the number of particles captured as a function of the depth of penetration 𝑧 into the lattice, where 𝑧 = 0 denotes the location of the major face of the lattice that the particle-laden flow initially impinges upon. The different curves correspond to simulation data in systems with inflow velocities ranging from 0.01 to 1 m/s and assigned magnetic flux densities ranging from 0.01 to 1 T. The data for the capture efficiency for these nine lattice conditions is in the inset and reproduced in a larger form in Table 3.1. With thanks to Andrii Syrota for generating this plot. B (T)/𝑣 (m/s) 0.01 0.01 21.4 % 0.1 48.6 % 1 98.6 % 0.1 18.6 % 30.8 % 70.8 % 1 27.2 % 27.4 % 51.2 % Table 3.1: Simulated capture efficiencies in systems with flow velocities ranging from 0.01 to 1 m/s and assigned magnetic flux densities ranging from 0.01 to 1 T. In these results, a couple of features are readily available. In general, increasing the magnitude of the magnetic flux density of the magnetic lattice at hand increases the capture efficiency of the lattice over all measured speeds. This result corresponds 49 well with the intuition regarding the difference between capture of nanoparticles by mechanical versus magnetic means; as the magnetic flux density of the magnet is increased, the greater the force that is applied to the magnetic nanoparticles, and the more particles are retained by the lattice as a result, regardless of the velocity of the flow entering the magnetic lattice. At the same time, it would be expected that a longer transit time in the lattice would allow for more opportunities for the applied magnetic force on a particle to direct that particle to the lattice walls before exiting the lattice entirely which should increase the capture efficiency. This trend holds strongly for the twisted honeycomb lattice, as increasing flow velocity leads to decreased capture efficiency, except in the case of the transition from 𝐵 = 0.01 T, 𝑣 = 0.1 m/s to 𝐵 = 0.01 T, 𝑣 = 1 m/s. In this case, an increase in the capture efficiency is realized. An explanation for this phenomenon can be found in the mechanics of capture with a low magnetic flux density magnet. One could imagine that capture in this case is dominated by flow features-namely, fluid crashes into the lattice walls and particles adhere as a consequence, as opposed to being drawn to the walls of the lattice by the imposed magnetic force on a given particle. In this case, the fluid contacting the initial surface of the magnetic lattice, namely, its major face, carries all of the particles in it to crash into this major face and adhere to the lattice, whereas the slower flows have enough time to adjust around the surface of the structure and prevent immediate capture of the particles on this surface. This argument is strengthened by the flow profile shown in Fig. 3.16, where the 1 m/s profile is abruptly disturbed when in contact with the surface of the lattice, carrying particles are into the lattice surfaces and potentially increasing capture. This trend is present with the lowest magnetic flux density of the three values used in these calculations as the stronger magnets allow for more efficient capture of the particles coming in contact with or near the initial surface of the lattice at all the velocities of interest which overwhelms the capture effect due to the flow dynamics of the lattice. The ideal capture device, as discussed in previous work by Maani et al, is composed of a three-filter setup.32 In order to give an assessment of the capture efficiency of such a setup, a simulation was run using a three-filter setup design under two sets of experimental parameters. A depiction of the distribution of captured particles under both sets of parameters can be seen in Fig. 3.19: 50 Figure 3.19: Capture in a triple lattice device. (a) Simulation results demonstrated the number of captured particles as a function of the distance along the axis of a triple lattice device as seen in (b). These simulations were run using an inflow velocity of 1.6 m/s for a triple lattice device in which each lattice was assigned a magnetic flux density of 1.45 T (blue) demonstrating capture of 77.8% of particles, and triple lattice device in which each lattice was assigned a magnetic flux density of 0.028 T (red) demonstrating capture of 53.4% of particles. With thanks to Andrii Syrota for generating this plot. Analysis of COMSOL Simulation and Experimental Capture Results These simulation results indicated an expected capture efficiency of 30.4 % under conditions similar to those used in the experimental system, including the magnetic properties of the lattice under investigation and the flow rate in the experimental setup. The capture results estimated by digestion of nanoparticles in nitric acid solu- 51 tion before ICP-MS analysis demonstrated capture efficiencies of approximately 49 ± 16 % and 31 ± 10 % for magnetized and nonmagnetized iron lattices respectively, but the spread in capture values for the three trials of each type of lattice demonstrated considerable overlap between the two sets of measurements. This overlap may indicate a mechanism of capture dominated by the fluid dynamics of the system, as opposed to capture mediated by the inherent magnetism of the lattices. The simulation results utilize realistic values of magnetic flux density and fluid velocity that compare well with the experimental setup detailed in Section 3 of this chapter. The assumption of Gaussian distribution of the particles in the flow is well justified by comparison to literature detailing nanoparticle distributions in flow systems. When particles traveling through a developed flow impact the closest major face of the lattice they likely have limited impact on the flow proximal to the major face because of their small size and relatively low concentration. While in the experimental setup nanoparticles are not distributed in one batch over the crosssectional area of the proximal end of the catheter sheath tubing then pushed through the sheath, their travel in the flow through the tubing is reasonably approximated by that distribution, and their interaction with the major face of the lattice in an additive fashion as the solution passes through the tubing is a reasonable approximation. This linear interaction implies that the simulated capture efficiencies provide reasonable expectations for the capture efficiency in the simulated lattices. One caveat to this use of the simulated capture efficiencies is that they are likely underestimates of truly attainable capture efficiencies due to nanoparticle aggregation. As nanoparticles in this flow still retain a tendency to attract while traveling through the flow, it is likely that larger aggregates will form by the time the solution reaches the proximal major face of the lattice. At that point, it is more likely that a given particle will be captured as its increased size increases the probability that it will physically encounter a portion of the lattice as well as run into areas of greater magnetic flux density at the edges of lattice walls and pores that will exert greater attractive force on the particles and increase their probability of capture. This relationship between particle size and probability of capture through increased magnetophoretic force is consistent with other simulations of capture efficiency in magnetophoresis in the literature.[80] In previous work regarding magnetophoresis of magnetic nanoparticles using magnets strung along a catheter wire, simulated capture efficiencies using permanent, rare earth magnet magnetic flux densities, which are roughly Tesla scale, were significantly lower than the results of our studies for magnetic systems with significantly lower magnetic flux densities, reaching a maximum of 24.6 % versus the 30.4 % 52 demonstrated for low magnetic flux density in our work.[35] This difference is likely due to differences in the simulation parameters of the systems at hand; whereas the neodymium magnet system was simulated as cylindrical magnets suspended in a much larger cylinder, the magnetic lattices in our simulations occlude the venous flow apart from the lattice pores. This architecture forces more direct contact between the whole flow profile and the lattice, and capture is easier when particles are brought close to the lattice by the flow itself. Consistent with this explanation, we show that the simulated capture efficiency in systems with assigned magnetic flux densities comparable to neodymium magnets approaches 50.6 % as in Fig. 3.17. In the experimental capture results attained in this work, the capture efficiencies of both magnetized and nonmagnetized lattices did not vary significantly from that predicted by the particle capture simulations performed here, 3.5 Summary In the preceding pages, three metal lattice compositions, in the form of iron, nickeliron, and copper-nickel-iron, were synthesized and characterized by powder X-ray diffraction and EDS to determine the phase and elemental composition of each system. The structure of the iron dominant system was further studied via electron backscatter diffraction to determine the presence or absence of retained iron oxides; the presence of magnetite was demonstrated and the connection to the mechanism of hydrogen reduction of hematite was elaborated upon. The magnetic properties of each metal lattice, including the remnant magnetization, coercivity, and maximum energy product were determined, and the comparison to values from soft iron and neodymium magnets demonstrated the soft nature of the magnetism in these samples as well as the relatively poor quality of the synthesized magnets. The designed capture testing system was described and the difficulties involved with the characterization of the nanoparticle concentration in the capture testing aliquots using ICP-MS was discussed and interpreted in the context of acid digestion of magnetic nanoparticle solution as opposed to direct measurement of nanoparticles in the system, yielding capture efficiency estimates of 49 ± 16 % and 31 ± 10 % in the magnetized and nonmagnetized systems, respectively. Finally, computational studies regarding the capture of nanoparticles by twisted, perforated magnetic lattices showed the possibility of capturing 30% of magnetic nanoparticles using lattices with magnetic properties and of macroscopic size and shape consistent with the iron lattices found in Fig.3.2. These results were compared to the ICP-MS capture values from acid digestion of nanoparticles and discrepancies between ex- 53 perimental and computational values were discussed. These results could be taken further forward by determining a reliable method for direct nanoparticle concentration characterization via the ICP-MS nanoparticle characterization method. In addition, the computational results could be more extensively probed and compared to experimental values by synthesizing lattices in rare earth magnet compositions and changing the flow rate used in the experimental studies of the nanoparticle solutions to analyze a larger portion of the magnetic properties-flow velocity-particle capture manifold.81 54 Chapter 4 EXPLORING THE HYDROGEL INFUSION ADDITIVE MANUFACTURING PARAMETER SPACE VIA ADDITIVELY MANUFACTURED IRON-NICKEL AND IRON-COBALT The author thanks Robin McDonald and Faiza Shabibi for their assistance in sample synthesis and data generation. This chapter describes the synthesis and characterization of iron-nickel and ironcobalt lattices to expand the range of existing metal alloys produced by the HIAM process and further catalog the idiosyncrasies and potential mechanisms behind these idiosyncrasies to further the applicability of the HIAM process to magnetic and nonmagnetic materials. The phases of the synthesized oxides and metals are discussed along with the relationship between the swell-in solution composition and the composition of the produced metal sample as characterized by EDS. The magnetic properties of the iron-cobalt system are measured and compared to the properties observed for lattices in Chapter 3. 4.1 Motivation for Iron-Nickel and Iron-Cobalt Compositions The synthesis of ferromagnetic metal alloys demonstrated in Chapter 3 focused on the synthesis of iron alloy materials, as iron phases were previously unknown with regards to the HIAM process. The choice of alloys was governed by the interest in modifying the coercivity of the base iron material, and subsequently modifying the maximum energy product in order to produce magnets of varying properties. In tandem with this effort, some consideration was given to alternative goals; namely, targeting certain compositions of ferromagnetic alloys, as well as introducing additional ferromagnetic metals. In this regard, two phases of interest stood out: iron-nickel and iron-cobalt alloys. Iron-nickel alloys are an appealing target for two major reasons: the biocompatibility of nickel and iron metals as noted in numerous implants and stent devices, and the magnetic properties of these alloys. Alloys of this type, more specifically in the 80 wt% nickel and 20 wt% iron composition, are known as permalloys and demonstrate high permeability, relating to the ease of change of magnetization of the material; as such, this is inversely proportional to coercivity.69 Because of their 55 extremely low coercivity, these compositions are especially interesting for use in temporary magnetic materials for particle capture. Due to the known difference between the composition of the swelling solution and the composition of the final metal sample, in order to attain the composition of interest a range of compositions near the desired ratio needs must be synthesized and examined.49 In doing so, the relationship between swell-in solution composition and final phase ratio for a binary system containing elements of different valence. On the other hand, iron-cobalt alloys are of interest due to the hardening effect of cobalt incorporation into iron alloys. In Chapter 3, the synthesized metal samples were of uniformly low maximum energy product, signaling their less-than-desirable remnant magnetizations as well as coercivities. These materials would not suffice for permanent magnet devices, and would require regular remagnetization. The solution would clearly entail increasing the coercivity and subsequently the energy product of these materials, and cobalt has been demonstrated to increase magnetic hardness due to the magnetocrystalline anisotropy of the cobalt atom imparted by the underlying spin-coupling active in cobalt atoms.82,83 In the case of the additively manufactured samples in Chapter 3, the synthesized parts demonstrated coercivities on the lower end of the expected range. Attempting to harden the iron magnets by the introduction of cobalt would allow for an exploration of the phase space of the iron-cobalt system and iron-cobalt oxide system as produced by the HIAM procedure versus the expected phases from thermodynamic studies. Additionally, the relationship between swell-in solution and final composition could be mapped out for yet another binary metal system with mixed valence in the solution phase. 4.2 Iron-Nickel Alloy Composition Tuning N.B. In this section, samples were printed using the twisted, perforated honeycomb geometry and swollen for 48 h and calcined underneath profile (b) in Fig. 2.3. X-ray diffraction samples were obtained on oxide and metal monoliths. The attempt to obtain an iron-nickel alloy of composition near Ni80 Fe20 consisted of cycles of HIAM synthesis starting with compositions near the desired target composition and adjusting the relative ratio of the metal salts in the swell-in solution in accordance with the procedure noted in [49]. The full table of tested compositions and the resultant final composition is available in Table B.3 in Appendix B, along with the measured ratios of the nickel and iron in each sample as determined by EDS 56 in Tables B.4 through B.12. Fig. 4.1 depicts the mole percentage of nickel in the metal part as determined by EDS versus the mole percentage of nickel in the swell-in solution. Figure 4.1: A plot demonstrating the difference between the concentration of nickel in the swell-in solution versus the nickel content of the reduced metal sample as determined by EDS in the nickel-iron system targeting the Ni80 Fe20 composition. The dashed line represents systems in which the swell-in solution and final composition are identical The plot in Fig.4.1 clearly indicates that nickel is generally underenriched in the final metal part versus its concentration in the swell-in solution, barring one tested composition. This is not an unexpected finding as a similar phenomenon was at play in Chapter 3, but it hints at the underlying mechanism of the swell-in process. Different metal ions with differing charges, acidities, and electronegativities will demonstrate different behavior in interacting with the same polymer system; this was demonstrated in [49] with the example of copper and nickel, similarly sized atoms and ions that have the same charge in their nitrate salts, +2, which resulted in minor but appreciable differences in composition between the swell-in solution and the metal composition. In the case of iron and nickel, the precursor salts used in this study provide iron(III) and nickel(II) cations, and this adds another variable into the mix. Given the PEGda backbone contains oxygen atoms with free electron 57 pairs, it can be argued that this interaction plays some role in the differential affinity of metal ions to the hydrogel polymer blanks. With the knowledge that the oxygen atoms in the PEGda backbone have been described as acting as hard bases, harder acids should demonstrate a greater affinity to the backbone than softer acids, and the iron(III) cation should represent a harder acid than the nickel(ii) cation.76,84 As such, this may drive the uptake of one ion preferentially during the swell-in process, leading to an imbalance between the swell-in solution concentration and the composition of the final metal part. At the very least, the understanding that nickel underenriches in the sample versus the swell-in solution allows for the design of solutions such that compositions of interest can be attained more rapidly. Moving on from the question of nickel underenrichment in the final sample, the crystallographic characterization of the oxide and metal parts produced from the compositions shown in Tab. B.3 will allow for an understanding of the normality of the phases generated during the HIAM procedure. The oxides will lead the discussion, as depicted in Fig. 4.2. 58 Figure 4.2: The normalized powder diffraction patterns corresponding to the oxide phases derived from the nickel-iron samples synthesized to target the Ni80 Fe20 along with simulated patterns derived from the following references obtained via the ICSD:[85, 86]. Note that the compositions in the pattern labels refer to the metal molar percentages in the final determination and not the stoichiometry of any given compound. Note that all patterns in Fig. 4.2 demonstrate a high background at low two-theta due to air scattering and collection without a scattered radiation protector. In this system, the powder diffraction patterns contain varying contributions from two types of phases: a nickel oxide-like phase and a magnetite-like phase (also termed a spinel).87 In materials such as the maroon colored Ni79 Fe21 oxide composition, the significant contribution of the nickel oxide-like phase to the system is clear, while another Ni79 Fe21 oxide composition in light blue seems dominated by a spinel phase. While mixtures of these two phases are expected in this range of nickel containing oxides, from Ni75 Fe25 to Ni91 Fe9 , from samples heated to a maximum of 700◦ C in air, the discrepancy between phases assigned such close compositions by EDS is 59 potentially a cause for concern. Complications could have arisen via poor sampling of the metal structure during the EDS determination of the metal composition in the final metal sample. The exploration of phase confirmation in these systems, while intriguing in the oxide case due to some potentially incommensurate results, would be incomplete without a discussion of the resultant metal phases. These metal phases are depicted in Fig. 4.3. Figure 4.3: The powder diffraction patterns corresponding to the metal phases derived from the nickel-iron samples synthesized to target the Ni80 Fe20 along with simulated patterns derived from the following references obtained via the ICSD:[71, 72]. The metal compositions characterized by powder diffraction in Fig.4.3 all demonstrate a strong match to that expected from a face-centered cubic crystal structure which nickel takes on as a pure metal. Given the aforementioned range of compositions for these metal samples at 900◦ C, the iron-nickel phase diagram predict the formation of a single face-centered cubic phase containing a mixture of nickel and iron.88 At the same time, this exploration of the phase composition of these 60 systems is potentially incomplete. Phases of nickel and iron mixtures can exhibit order within the framework of a face centered cubic cell, such that certain sites are no longer degenerate as atoms preferentially sort into one category or another; this can affect the magnetic properties of the resulting alloy.89 An example of such a phase in the iron-nickel system is tetrataenite, a mineral found in meteorites due to the slow cooling required to allow for ordering of nickel and iron atoms along alternating (002) planes. While the presence of tetrataenite or other ordered ironnickel phases in the samples discussed thus far is unlikely due to issues with the required composition and the method of preparation, the rarity of these phases and the tendency of this ordering to “hide” from simple crystallographic analysis merits consideration with regards to advanced characterization of binary metal alloys.90 4.3 Iron-Cobalt Alloy Exploration In this section, samples were printed using the twisted, perforated honeycomb geometry and swollen for 48 h and calcined underneath profile (a) in Fig. 2.3. X-ray diffraction samples were obtained on oxide powders and flattened metal monoliths. VSM hysteresis loops were measured using twisted, perforated honeycombs bisected orthogonal to the major face. Exploring the phase space of iron-cobalt alloys synthesized via the HIAM process involved generating samples using a range of swell-in solutions and characterizing the result products crystallographically as well as magnetically, whilst also examining the relationship between the composition of the swell-in solution and the resultant metal composition after the completion of the HIAM process. The table of iron-cobalt swell-in compositions tested can be seen in Appendix B.4 Table B.13. The composition of the resultant metal parts from the HIAM process, as determined by EDS and detailed in Appendix B.4, are seen plotted against these swell-in solution compositions in Fig. 4.4. 61 Figure 4.4: A plot demonstrating the difference between the concentration of cobalt in the swell-in solution versus the cobalt content of the reduced metal sample as determined by EDS across a range of swell-in solution concentrations. The dashed line represents systems in which the swell-in solution and final composition are identical. This plot demonstrates the overenrichment of cobalt over iron in all solutions composed of a mixture of the two metal salts. This overenrichment presents a quandary for the simple story of affinity between hard acids and bases leading to enrichment of a given metal ion in the swollen gel versus the swelling solution discussed in Section 2 of this chapter, as the cobalt(ii) cation has been characterized as lying between the hard and soft dichotomy just as the nickel(ii) cation has, and yet here iron is underenriched in the final metal compared to cobalt in all relevant compositions.84 While the hard acid-hard base perspective might be a useful model when generally thinking about the products of chemical processes and the interaction between chemical moieties, it is well known that these characterizations are qualitative rather than quantitative. PEGda has been demonstrated to solubilize cobalt(II) nitrates more effectively than chromium(III) nitrates, despite the generally harder nature of the chromium(III) ion and similar sizes between the two ions.76 The relative 62 overenrichment of cobalt in the iron-cobalt HIAM system demonstrates an expanded awareness of the complicated interaction between the metal ion species and the polymer gel in the HIAM process. The next avenues of exploration in the iron-cobalt HIAM system involve the crystallographic characterization of the oxide and metal resultant parts in the process. The powder diffraction patterns corresponding to the synthesized oxides are displayed in Fig. 4.5. Figure 4.5: The powder diffraction patterns corresponding to the oxide phases derived from the cobalt-iron samples synthesized in a range of compositions along with simulated patterns derived from the following references obtained via the ICSD:[91–93]. As expected from phase diagrams of the iron oxide-cobalt oxide system at 700◦ C, the pattern corresponding to the pure iron oxide sample matches the hematite structure, while the pattern corresponding to the pure cobalt oxide sample matches the cobalt spinel structure (Co3 O4 ).94 In between these two extremes, the expected phase composition is a mixture of hematite and spinel crystal structures as long as the cobalt content is below 35 mol%, which is apparent in the Co14 Fe86 and Co26 Fe74 63 oxide systems in Fig.4.5. With increasing cobalt content, we expect to see the formation of a single spinel phase to a mixture of spinel phases as the miscibility gap is approached in the phase diagram, and the Co36 Fe64 oxide system corresponds to the expected single spinel phase followed by a mixture of two spinel phases of differing lattice parameter in the Co50 Fe50 , Co59 Fe41 , Co67 Fe33 , and Co75 Fe25 oxide systems. Finally, the Co85 Fe15 , and Co91 Fe9 oxide systems are composed of a single spinel phase. Moving on from the limited differences between the expected phases of the ironcobalt oxide systems and the experimental patterns, the metal samples are next with regards to understanding the products of the reduction of these oxides versus the known phases otherwise attained in this system as described by the iron-cobalt phase diagram. Powder diffraction patterns for these metal samples are depicted in Fig. 4.6. Figure 4.6: The powder diffraction patterns corresponding to the metal phases derived from the cobalt-iron samples synthesized in a range of compositions along with simulated patterns derived from the following references obtained via the ICSD:[72, 95, 96]. 64 When compared to the expected phases from the iron-cobalt phase diagram at 900◦ C, a few features of the patterns displayed in Fig.4.6 stand out.97 In the high to moderate iron content phases, the body-centered cubic phase is present, including the Fe through to the Co75 Fe25 , a feature readily predictable from the literature. Below this content of iron to the pure Co system, the patterns of the remaining system display nearly entirely face-centered cubic phases without mixing, except in the case of the pure Co system in which the most intense reflection from the hexagonal cobalt phase is present. At lower temperatures, the hexagonal cobalt phase may readily form, and as such it is not a particularly unexpected phase in a pure sample of cobalt as in the pattern under discussion. The diffraction results of the metal samples generally align with predictions from the iron-cobalt phase diagram. The magnetic properties of cobalt with respect to iron constituted the primary motivation behind the synthesis and characterization of iron-cobalt mixtures via the HIAM procedure; hardening the resultant magnets and increasing the maximum energy product would likely enable greater nanoparticle capture as described in Chapter 3. The magnetic hysteresis loops measured using these samples are visible in Fig. 4.7. Figure 4.7: A plot of the hysteresis loops for the synthesized Fe-Co samples as collected by VSM. An inset is provided at the right for the area of interest with regards to magnetic characterization. The remnant magnetizations, seen here as the points at which the hysteresis loops cross the zero magnetization line, in these systems are generally larger than those found in the samples displayed in Fig. 3.10. This could be a consequence of the lack of sintering aids such as nickel and metals with a high homologous temperature 65 such as copper in these compositions; this could lead to smaller grain sizes and subsequently higher magnetocrystalline anisotropy, increasing the remnant magnetization of the samples. In order to compare the values of interest to each other, the remnant magnetization, coercivity, and maximum energy product were computed from the collected hysteresis loops (as described in Appendix E), and displayed in Fig. 4.1. Sample ID Co Co91.5 Fe8.5 Co84.8 Fe15.2 Co74.9 Fe25.1 Co67.2 Fe33.8 Co58.6 Fe41.4 Co50 Fe50 Co35.9 Fe64.1 Co26.4 Fe73.6 Co13.9 Fe86.1 Fe 𝐵𝑟 (Gauss) 912.4 1791 720.8 587.9 663.4 684.3 714.2 425.3 668.9 237.5 656.8 𝐻𝑐 (Oe) 45.78 12.35 30.00 29.77 30.11 24.46 33.71 25.52 32.09 26.07 32.06 −(𝐵𝐻)𝑚𝑎𝑥 (kG-Oe) 9.484 3.735 5.879 8.414 4.071 6.071 4.524 1.962 4.617 5.005 8.906 Table 4.1: Magnetic properties of the synthesized iron-cobalt samples, including the remnant magnetization, coercivity, and maximum energy product. The coercivities found for these materials do not stand out with respect to the samples discussed in Chapter 3. The maximum value, of 45.78 Oe, is comparable to the coercivity of the copper-nickel-iron system at 46.79 Oe, and the other iron-cobalt materials maintain coercivities higher by at most 10 Oe than the iron and nickel-iron systems featured in Fig. 3.10. The remnant magnetizations of these samples are generally higher than the previously described samples, barring a few compositions, and the maximum energy products compare well with those of the iron, nickel-iron, and copper-nickel-iron systems seen in Chapter 3. These are likely the result of the high magnet moment of the cobalt atom compared to the copper atoms in the copper-nickel-iron system, which compensates somewhat for the lower coercivity of some of these compositions. With regards to trends with respect to cobalt and iron content in these samples, the three parameters displayed in Table 4.1 were plotted versus the cobalt mole percentage of the sample as determined by EDS, and these plots are displayed in Fig. 4.8. 66 Figure 4.8: Plots of the three magnetic properties discussed in Table 4.1 displaying the (a) remnant magnetization, (b) coercivity, and (c) maximum energy product versus the cobalt content of the metal as determined via EDS. There do not appear to be any discernible patterns with regards to the relationship between cobalt content and the remnant magnetization, coercivity, and maximum energy product in the as-synthesized samples. Magnetic materials are often thermally processed in order to improve the properties of the as-synthesized materials as this can allow for increased alignment of grains with high magnetocrystalline anisotropy which improves the remnance and thereby the magnetic energy product of the as-synthesized magnets. Kinetics of crystal growth in the initial synthesis process can prevent this sort of alignment from occurring.78 This lack of postprocessing may underlie the underwhelming properties of the iron-cobalt magnetic materials described here. The increased remnance and maximum energy product of the iron-cobalt magnets described here is still an interesting development that does follow expectations with regards to introducing cobalt into the iron HIAM system. 4.4 Summary The iron-nickel and iron-cobalt systems discussed in the preceding pages expand the known features of metals synthesized by the HIAM process. In the case of the iron-nickel sample, the underenrichment of nickel with respect to iron was reinforced and connections to hard acid-hard base features of the constituents of the swell-in solutions and the polymer gel were discussed. The phases exhibited by the as synthesized samples were compared and an apparent contradiction to the known phase behavior in this system was discussed and rationalized. In the ironcobalt system, the products of the HIAM process were confirmed as conforming to expected phase behavior given the composition of the samples and the processing temperature for the oxidation and reduction steps. The magnetic properties of the iron-cobalt system were measured and the improved remnance and maximum energy 67 product in many of the synthesized compositions was remarked upon, along with the underwhelming differences in coercivity between the compositions in this system and those discussed in Chapter 3. The apparent lack of a visible trend between the cobalt composition and the magnetic properties was observed. Moving forward, the use of postprocessing techniques to improve the magnetic properties of the samples should be at the forefront of any future exploration of both the iron-cobalt system and other magnetic metal systems synthesized by the HIAM procedure.78 Additionally, the microstructure of these samples should be further investigated with transmission electron microscopy to connect the global picture painted by the powder diffraction results with the local environment of the constituent materials in order to explain the observed magnetic properties of the as-synthesized samples. Finally, the affinity between the polymer backbone of the blank gels and the metal ions in the swell-in solution should be further elucidated to allow for facile programming of the final metal composition via the synthesis of a swell-in solution of a given composition; this may be attainable via the use of isothermal titration calorimetry, a technique that allows for the exploration of the thermodynamics of the interactions between polymer chains and other moieties in solution.98 The exploration of mixed metal atom systems can be generalized to metal-carbon composites by modifying the experimental process, which is discussed in the chapter to follow. 68 Chapter 5 ENHANCING HYDROGEL INFUSION ADDITIVE MANUFACTURING: VACUUM PROCESSING The author would like to thank Wenxin Zhang for TEM lamellae synthesis and collection of TEM data. This chapter describes the development of an alternative HIAM process that utilizes gel decomposition to produce metal-carbon composites demonstrating intriguing microstructural features and magnetic properties. The synthetic method is discussed and products of the method are briefly compared to those synthesized by the traditional HIAM process. The coercivity of the samples are compared to those generated by the traditional HIAM method and correlation to the microstructural features elucidated by transmission electron microscopy using knowledge of the magnetic-microstructural interplay is attempted. 5.1 Motivation for Vacuum Processing Modifying the HIAM procedure explicated in Chapter 2 allows for the expansion of the potential phase space of materials produced by the HIAM procedure. The production of steel lattices via additive manufacturing provides one such phase in which the traditional HIAM procedure is less useful. Steel production generally follows the Bessemer method, in which smelted iron is combined with a source of carbon, often coke, under high-temperature oxygen-free conditions, to form iron phases with high carbon contents such as slag or pig iron.99 These phases are then mildly oxidized to remove carbon and form steel in a variety of compositions. As seen in Chapter 3, the retained carbon content in the iron materials synthesized by the HIAM methods do not interfere with or modify the phase composition of the iron phase-a BCC phase is synthesized with, upon inspection, a similar lattice parameter to pure iron, as opposed to the more complex crystal structures of high carbon iron phases. The swollen blank placed into the furnace has all of the elements of interest to form iron in the form of the swollen iron nitrate and the poly(ethylene glycol) diacrylate backbone. Decomposing the swollen gels in the absence of oxygen may lead to different interactions between these two gel components resulting in products incorporating iron and carbon in different proportions than those produced by the traditional method. 69 In addition to the formation of different single phases with regards to the traditional HIAM procedure, the use of the HIAM procedure to form composite materials is also limited. Using materials that do not alloy, the traditional HIAM procedure can lead to phase separation and the formation of composite alloys. The formation of such materials as carbides, in contrast, is prevented by the oxidation step of the traditional HIAM procedure which removes the organic phase present in the swollen gel. Due to the atomic level mixing between polymer and metal ion in the swollen gel, preventing this elimination of the organic phase could be an opportunity for the formation of micro- or nano-composite structures. Works in the literature have shown the utility of applying such procedures to form metal oxide-carbon nanocomposites with various applications, including lithium vanadate/carbon composites for battery anodes and titanium carbide-carbon nanotube composites for the generation of carbide-derived carbon used in supercapacitors.100,101 The second of these two examples illustrates another aspect of the expanded utility of foregoing a high-temperature oxidation step; the composite material formed allows for the formation of unique carbidederived carbon products by first creating a microstructured metal carbide-carbon composite, which would be difficult to otherwise generate from the HIAM process as titanium dioxide is an exceedingly stable phase under reducing conditions.100 This type of novel composite synthesis is another motivation for the modification of the HIAM procedure. Architected carbon phases are another category of material that the traditional HIAM process cannot produce due to the initial oxidation step. While metals produced by the HIAM process retained carbon even after the oxidation step, the majority of the incorporated carbon is removed in this initial step and carbon phases as produced in pyrolytic processes are generally unavailable.49 Architected carbon materials have been made by a variety of methods, including inverse opal templating, for use as sensor and battery materials. While these materials have been used in a wide range of applications, inverse opal methods can only access a narrow range of final architectures.102 The use of 3D printed polymers in the case of the HIAM and other 3D printing procedures overcomes this difficulty, but modification to the thermal treatment procedure is necessary; in this case, the replacement of the oxidation step with a pyrolysis step, whether under argon or nitrogen atmospheres or vacuum.103 At the same time, the HIAM procedure has the potential to produce phases of carbon that normally require high temperature processing through catalytic graphitization. In this process, the formation of metal carbides at high temperatures (≈ 900◦ C) and the subsequent decomposition of the material at lower temperatures yields a 70 graphitic carbon phase whereas generally temperatures in excess of 2000 ◦ C are required to form graphitic carbon from pyrolysis of organic precursors.104,105 As such, the HIAM process, if suitably modified, could allow for the flexible formation of architected carbon phases, including the formation of graphitic carbon at lower temperatures than traditional pyrolysis techniques. 5.2 Vacuum Processing Method The vacuum processing method differs from the traditional HIAM method in the choice of atmosphere under which the initial thermal treatment step was performed. The polymer blanks were printed and swollen for 48 h at 70 ◦ C as described in Chapter 2 Section 1; in these explorations, the chosen solutions included a solution of composition 99.5 mol% iron(III) nitrate-0.5 mol% nickel(II) nitrate, designated the “Iron” system as in Chapter 3 Section 2, and a solution of composition 60 mol% copper(II) nitrate-20 mol% nickel(II) nitrate-20 mol% iron(III) nitrate, the “Copper-Nickel-Iron” system. A depiction of the vacuum modification of the HIAM procedure can be seen in Fig. 5.1, with specific reference to the “Iron” system. Figure 5.1: A schematic illustrating the synthesis of lattices via the vacuum HIAM processing method with reference to the iron vacuum HIAM sample. 5.3 Iron vs. Copper-Nickel-Iron Vacuum HIAM The initial structural characterization of the iron vacuum HIAM system can be seen in Fig. 5.2. The lattice is visibly larger than the traditional Fe HIAM lattice seen in Fig. 3.2, and large aggregates are noticeably present in the structure; these large aggregates, viewed in the macroscopic image in Fig. 5.1, show a metallic 71 luster. This larger size is predictable considering the lack of combustion of the polymer in the decomposition process, and the formation of metallic aggregates is also unsurprising considering the presence of nickel as a sintering agent as well as the limited solubility of carbon and iron in the other element, with the formation of cementite implicated at high carbon contents.106,107 The powder diffraction pattern of the sample shows no evidence of cementite presence as the experimental pattern is consistent with a body-centered cubic iron phase. Figure 5.2: A scanning electron image of the iron vacuum HIAM sample along with a powder diffraction pattern from a representative sample and a simulation derived from a CIF file obtained from the following reference via the ICSD:[72]. At the low angles of the diffraction pattern in Fig.5.2, there is the noticeable presence of a broad reflection around 25 degrees two-theta. This likely corresponds to the disordered carbon presence in the sample, which has been seen in the relevant literature.108 While the powder diffraction pattern can provide information about the long-range structure of this sample, it has limited ability to probe or characterize the short-range order of the system. Transmission electron microscopy allows for the exploration of the microstructure of a small portion of the system under consideration, with the assumption that local insights can be extrapolated beyond the individual area(s) probed. The results of such examination are found in Fig. 5.3. 72 Figure 5.3: A demonstration of the preparation of the Fe sample for TEM characterization. (a) An inset illustrating the approximate orientation of the lamella with regards to the bulk sample of vacuum HIAM iron (b) A depiction of the lamella obtained from the sample and the identified regions of the sample (c) Images of the specific regions of the lamella illustrated in part (b) along with (d) the fast Fourier transform of the nanocrystalline phases of the lamella to obtain pseudo-spacings. In this lamella, we observe the formation of three partially distinct layers, labeled characteristically in Fig.5.3 as the large crystallite layer, the superficial nanocrystallite layer, and the deep nanocrystallite layer. In the large crystallite layer, large crystallites (near micrometer diameter) are visible, and the presence of iron and carbon is confirmed by EDS (see Appendix Fig. D.1). Deep to this layer, the first nanocrystalline layer is found; at higher magnification, it can be seen that this layer consists of nanocrystallites with sizes on the order of 10 to 20 nm, as demonstrated in Fig. D.2 and Table D.1. When a Fourier transform of this image is performed, the 73 characteristic spacings between layers in the original image can be compared with known d-spacings for body-centered cubic iron, demonstrating that the crystallites match what might be expected for nanocrystalline iron. Moving on to the deepest layer of the lamella, another nanocrystalline layer can be observed; in this layer, the nanocrystallites appear smaller still, with diameters around 5 nm or less. Again, performing a Fourier transform of the high resolution image of this region demonstrates the presence of characteristic spacings corresponding to the d-spacings expected of body-centered cubic iron, which provides evidence for the nanocrystalline iron nature of this region. Even if we can readily identify the crystalline spacings of iron in this system, in all three of these regions we note the widespread presence of carbon as determined by EDS. These results suggest the formation of an iron-carbon nanocomposite via the vacuum HIAM procedure. In order to more broadly assess the outcomes of the vacuum HIAM procedure, the decision was made to pursue the synthesis of a system with a face-centered cubic structure and different magnetic properties. In this case, the copper-nickel-iron system discussed in Chapter 3 appeared to provide a useful contrast, as the synthesized parts had a face-centered cubic symmetry and showed increased coercivity with respect to the iron system synthesized under the traditional HIAM procedure. A gel swollen with the copper-nickel-iron solution processed under the vacuum HIAM condition to yield a nanocomposite, with some bulk structural characterization displayed in Fig. 5.4. 74 Figure 5.4: A scanning electron image of the copper-nickel-iron vacuum HIAM sample along with a powder diffraction pattern from a representative sample and a simulation derived from a CIF file obtained from the following reference via the ICSD:[70]. The copper-nickel-iron vacuum HIAM system demonstrated greatly increased fragility when compared to the iron vacuum HIAM system, as monoliths rarely survived transfer to scanning electron microscopy mounting stubs, but at low magnification the surface of the surviving portion did not appear different from that of the iron system. The powder diffraction pattern demonstrated one significant difference; while the pattern seemed to match well with the prediction for a face-centered cubic lattice as previously seen in Chapter 3, the reflections appeared much broader than those of the simulated pattern or the experimental pattern seen in Fig. 3.5. Reflection broadening, while a potentially complex topic in powder X-ray crystallography, is often a consequence of a decrease in the size of crystallites to the submicron regime, a phenomenon described by the Scherrer equation, or inhomogeneous strain on crystals, which can be characterized using the Williamson-Hull method.109 Without further microstructural information, confirming the cause of this broadening would be difficult. Considering the wealth of structural information that was attained by transmission electron microscopy of an iron system lamella, the characterization of a lamella from the copper-nickel-iron system was carried out, yielding the results seen in Fig. 5.5. 75 Figure 5.5: A demonstration of the preparation of the Cu-Ni-Fe sample for TEM characterization. (a) An inset illustrating the approximate orientation of the lamella with regards to the bulk sample of vacuum HIAM copper-nickel-iron. (b) An image of the large crystallites noted on the lamella and a fast Fourier transform of the image to generate pseudo-spacings as well as (c) an image of the fine nanocrystalline matrix and its fast Fourier transform to obtain the same pseudo-spacings of interest. The structure of the bulk fragment shows a surface layer of of small, semispherical droplets. Visualizing the lamella in the transmission electron microscope, the presence of two somewhat distinct regions are detectable, although with less obvious depth resolution compared to the iron system. The lamella is composed of large crystallites and a fine nanocrystalline layer; analyzing the larger crystallites, with sizes ranging from hundreds of nanometers to microns, the Fourier transform of the electron micrograph shows the presence of characteristic spacings comparable to the d-spacings expected from face-centered cubic copper, with differences likely 76 attributable to the presence of iron and nickel in the sample. Characterization of the fine nanocrystallite material surrounding these large crystallites reveals the presence of nanocrystallites with diameters around 10 nm, and the Fourier transform of the micrograph demonstrates characteristic spacings in line with face-centered cubic copper. Given the size of these crystallites, it is not inconceivable that both sizes of crystallites could contribute to the reflection broadening seen in Fig. 5.4.110 Given the additional presence of carbon in both of these samples, as well as the complex microstructures present in both lamellae, the next step taken in the characterization of these vacuum HIAM products was the measurement of their magnetic properties. As such, vibrating sample magnetometry was performed on powders of both samples, and a comparison of the obtained hysteresis loops between the samples and with traditionally manufactured HIAM samples can be seen in Fig. 5.6. Figure 5.6: Vibrating sample measurements demonstrating the hysteresis loop for (a) traditional HIAM synthesized iron and copper-nickel-iron and (b) vacuum HIAM synthesized iron and copper-nickel-iron composites. The iron vacuum HIAM sample demonstrates an increased coercivity of 64.2 Oe in comparison to both copper-nickel-iron samples as well as the iron traditional HIAM sample at 12.0 Oe, while the copper-nickel-iron vacuum HIAM sample demonstrates a lower coercivity of 17.7 Oe in comparison to the copper-nickel-iron traditional HIAM sample with a coercivity of 47.0 Oe. With regards to the increased magnetic hardness of the iron system undergoing vacuum HIAM versus traditional HIAM, a potential explanation can be found in the presence of the superficial nanocrystalline layer. Decreases in grain size can inhibit the motion of domain walls which increases the difficulty involved in changing the magnetization of a polycrystalline sample consequently increasing the coercivity of the sample.111,112 77 The superficial nanocrystalline layer in the iron sample contains grains small enough to increase the coercivity beyond that found in systems with large (1-10 micrometer) grains, while also being large enough to prevent significant magnetic softening due to comparable domain wall thickness and grain sizes. The smaller size of the fine nanocrystalline layer in the copper-nickel-iron vacuum HIAM system may explain the decrease in coercivity when compared with both the copper-nickel-iron traditional HIAM system as well as the iron vacuum HIAM system, as bulk grain copper-nickel-iron may be more magnetically hard than grains on the order of 10 nm in size. The powder X-ray diffraction pattern associated with the copper-nickel-iron vacuum HIAM system may provide further evidence for this hypothesis; the greater broadening of the reflections may itself reflect the greater content of nanocrystalline material in the sample overall as compared to the iron vacuum HIAM sample or the copper-nickel-iron traditional HIAM sample. 5.4 Summary The vacuum HIAM technique demonstrated herein extends the domain of materials attainable with the traditional HIAM technique to metal-carbon nanocomposites. In these nanocomposites, spatial segregation with respect to crystallite size was demonstrated in the iron and copper-nickel-iron systems using transmission electron microscopy. Along with the synthesis of microstructurally interesting samples, the measured coercivity of the samples demonstrated an inversion of the behavior shown in samples synthesized via the traditional HIAM process. This difference was tentatively attributed to the observed differences in nanocrystallite size between the two samples. In order to explore this phenomenon more thoroughly, two experimental courses seem intuitive. The first involves the use of total scattering methods coupled with computational chemistry techniques such as density functional theory to get a grasp on local microstructure throughout the sample, rather than in a single lamella, to demonstrate the presence or lack thereof as well as the nature of differences in nanocrystallite size distribution between the two systems.113 The second involves the application of the technique to metals that cannot be reduced under forming gas by using the decomposition of the polymer in the reaction as a substrate for carbothermic reduction and the generation of metal carbide-carbon nanocomposites or pure metal carbide samples, which cannot be synthesized via the traditional HIAM approach.114 78 Chapter 6 SUMMARY AND OUTLOOK In the preceding chapters, iron and iron-alloy lattices were synthesized via the traditional HIAM procedure. The phase composition, elemental composition, and magnetic properties of these lattices were explored, and attempts were made to capture nanoparticles and quantify the concentration of the uncaptured nanoparticles in aqueous solutions using ICP-MS. Computational studies regarding the capture of particles using iron lattices with properties comparable to those described in Chapter 3 were additionally performed, and demonstrated the possibility of nanoparticle capture even with the middling magnetic properties of the as synthesized lattices. Then, the iron-nickel and iron-cobalt phases synthesized by the HIAM procedure were further explored by EDS, powder X-ray diffraction, and in the case of the iron-cobalt system, vibrating sample magnetometry. Finally, an innovation upon the traditional HIAM procedure, in the form of the vacuum HIAM procedure, was described and the application of the technique to the iron and copper-nickeliron systems was characterized by transmission electron microscopy and the effect on magnetic properties was revealed and discussed. In addition to the directions discussed in the summaries of these preceding chapters, two topics will form the focus of the rest of this chapter: the future of transarterial catheterization techniques and additive manufacturing, and the path forward for magnetic materials additive manufacturing. 6.1 Transarterial Catheterization and Additive Manufacturing The primary application the filters discussed in Chapter 3 were designed for involved the sequestration of doxorubicin during the TACE procedure. With regards to the future of chemotherapy sequestration, a number of interesting advances have come to light. Su et al. have recently demonstrated the use of nucleobases embedded in a polymer network to capture doxorubicin by a mechanism similar to that discussed in the work of Yee et al. in Chapter 1 Section 2, but with several features that warrant greater investigation.31,115 This system utilizes free nucleobases in the mixture of monomer and dimer units used to form the polymer backbone, rather than chemically bound DNA to a scaffold structure. This relieves the concern around heteroDNA leakage into the bloodstream during capture while introducing an intriguing aspect 79 into the materials of chemotherapy capture. At least some of the doxorubicin capture capacity in this material is attributed to molecular imprinting of the nucleobases onto the polymer during the polymerization process. The components of the polymer network can be easily incorporated into a vat photopolymerization system, rendering the adaptation of this system to additive manufacturing potentially straightforward. In addition to the nucleobase-embedded polymer application, the concept of in vivo pharmaceutical capture can be readily applied to systems revolving around problematic metabolites and pharmaceuticals. Two of these problems represent targets that both require selective sequestration of molecules in vivo and are suitable for solutions involving additive manufacturing: ammonia sequestration to prevent hepatic encephalopathy and aminoglycoside capture to mitigate oto- and nephrotoxicity. With regards to hepatic encephalopathy, this condition develops in patients with advanced liver disease such that ammonia and ammonium delivered to the liver via the portal vein cannot be processed into nontoxic metabolites or bypass the liver due to portal vein hypertension and increased collateral circulation. In both of these cases, these metabolites entry the systemic circulation and, most deleteriously, the cerebral circulation, where they can pass through the blood-brain barrier to enter the central nervous system and cause mild to severe dysfunction, including confusion, personality changes, tremors (“liver flap”), cerebral edema, and coma.116 Even subclinical forms of hepatic encephalopathy are associated with significant decreases in quality of life. Treatments for hepatic encephalopathy include compounds that increase intestinal motility to prevent the buildup and absorption of ammonia in the intestines as well as diet modification to limit protein metabolism and subsequent ammonia production.116 Additively manufactured devices could be used as indwelling ammonia sequestration devices by drawing on work showing the affinity of ammonium ions for sulfate functional groups, including the use of such systems as ion exchange materials.117 Three-dimensionally printable polymers, such as poly(styrene sulfonate) or polymethacrylates, could serve as a substrate for such devices, and the need for suitable hemodynamics implies a role for architecture in such devices.117,118 While this application involves increasing as much as possible the capture of ammonium from the bloodstream, the capture of aminoglycosides relies on the temporal course of drug concentration in the bloodstream as a whole. Aminoglycosides constitute a class of potent antibiotics that find application in a variety of conditions.119 While they are used prolifically in modern medicine, the side effect profile of these drugs limits their applicability; aminoglycosides have 80 significant oto- and nephrotoxicity, and these side effects lead to limitations on the supplied dose of aminoglycosides as well as the ability to use some aminoglycosides as intravenous antibiotics, such as neomycin.120,121 The toxic side effects of these drugs can be mitigated by providing them in such a manner that the peak concentration is high enough to accomplish the antibiotic effect of the aminoglycosides while maintaining low concentrations once the dose of the drug falls to its lowest level; the concentration of the drug before the dose is administered, termed the trough concentration, is related to the occurrence of side effects from these drugs.122,123 In this case, a device could be used to absorb the drug intravenously with a capture rate slow enough such that the sudden administration of a large quantity of the aminoglycoside of interest would cause a spike in the blood concentration of the drug, but would subsequently scavenge the drug such that the blood concentration of the drug would subsequently crater, allowing for a high peak concentration to efficiently treat the infection of interest but a low trough concentration such that these toxicities are mitigated.123 The mechanism of action of these drugs involves binding to RNA in the ribosomes of prokaryotes. RNA in turn represents a handle that can be used to design a capture device for aminoglycosides.119 Using 3D printing to form scaffolds with favorable hemodynamic properties incorporating RNA as a surface coating or a component of the resin, these devices could be used to disaggregate the anti-bacterial properties of these drugs from their side effects, leading to wider use of existing used and unused aminoglycosides. 6.2 Magnetic Materials Additive Manufacturing Moving on from the realm of medical problems soluble by the application of capture technologies and devices designed by additive manufacturing, the next question of interest involves the additive manufacturing of magnetic materials. As demonstrated in Chapter 3 and Chapter 5, the HIAM and vacuum HIAM processes demonstrate the ability to generate soft magnetic materials with different coercivities and maximum energy products, while having little ability to tune magnetic properties within a given composition. A way in which to improve upon these materials would be to generate hard magnetic materials, specifically permanent magnetic materials, using additive manufacturing technologies. Several well-known compositions of permanent magnets include the alnico (Al-Ni-Co), neodymium (Nd-Fe-B), and samarium cobalt (Sm-Co) magnets.124 These compositions are difficult to attain in the HIAM process as each one contains a metal whose oxide is difficult to reduce under hydrogen gas; respectively, these metals are aluminum, neodymium, and 81 samarium. One method developed by Baker et al. used to generate samarium-cobalt magnetic particles utilized the metallothermic reduction of samarium cobalt oxide via calcium metal in a forming gas environment; the subsequent calcium oxidesamarium cobalt mixture was separated using degassed water to produce highly coercive samarium cobalt nanoparticles.125 The challenge in this regard would be to extend this method to architectured magnets without sacrificing the structural integrity of the architected monoliths. This approach would combine the utility and flexibility of the HIAM procedure with the hardness and energy product values of rare earth magnets. Other approaches used to additively manufacture permanent magnets include bonded magnets formed by extrusion-type methods and laser powder bed fusion.124 With regards to bonded magnets, wherein magnetic material is held together with a nonmagnetic binding material, difficulties continue to arise because of the lack of alignment between the magnetic particles of interest, leading to even lower maximum energy products than expected from the fraction of magnetic material in the sample.124 Laser powder bed fusion can be used to generate neodymium rare earth magnets with properties comparable to bonded magnets, while samarium cobalt rare earth magnets with favorable magnetic properties are more difficult to generate due to the need for multiple annealing steps to increase coercivity in the as-printed parts.126,127 By optimizing printing parameters and improving magnetic particle alignment in bonded magnets, the properties of these additively manufactured magnets can be further improved. At the same time, in addition to the general interest displayed in the studies discussed in this thesis with architected magnetic materials, there exists an intriguing direction in which to proceed regarding architected magnetism within architected magnets; e.g., additively manufacturing a magnet while imposing a selected orientation upon the magnetic domains in that region which introduces an additional avenue for local manipulation of the magnetic properties of the material.124 In addition to the pursuit of permanent magnetic materials via additive manufacturing methods, the field of magnetic material additive manufacturing has pushed forward the field of responsive material manufacturing using magnetism as a trigger.46 In these systems, the incorporation of a magnetoresponsive material, such as a dispersion of metal nanoparticles, into the bulk phase allows for manipulation of the synthesized part by the imposition of a magnetic field or by exposure to a permanent magnet. In these systems, vat photopolymerization stands out as an exemplary 82 technique, because the motivation behind the synthesis of these materials is not the formation of the hardest or highest quality magnet possible, but rather the responsiveness to an imposed magnetic field, so the incorporation of a flexible binding agent can only aid in this responsiveness while remaining in one piece. Responses such as curling, gripping, and deflection under an applied field were found in various photopolymerized composites of polymer and magnetic nanoparticles.46 Changing the loading of the magnetic material modifies the degree to which these responses occur under an identical magnetic stimulus. In addition to the responsiveness to stimuli, the incorporation of magnetic particles into these polymeric structures can improve the mechanical properties of these materials, wherein the alignment of magnetic particles can be shown to increase the modulus of the material under consideration.46 These materials will find growing application in the fields of soft robotics and micro-actuators, as robotic limbs and remote actuation mechanisms can be synthesized in a variety of architectures that magnify the applied stimulus or allow for complex operations to be carried out with a simple set of stimuli.128,129 83 BIBLIOGRAPHY (1) Toh, M. R.; Wong, E. Y. T.; Wong, S. H.; Ng, A. W. T.; Loo, L.-H.; Chow, P. K.-H.; Ngeow, J. Gastroenterology. 2023, 164, 766–782. (2) Ding, J.; Wen, Z. BMC Cancer. 2021, 21, 1157. (3) Choi, S.; Kim, B. K.; Yon, D. K.; Lee, S. W.; Lee, H. G.; Chang, H. H.; Park, S.; Koyanagi, A.; Jacob, L.; Dragioti, E.; Radua, J.; Shin, J. I.; Kim, S. U.; Smith, L. Clin Mol Hepatol. 2023, 29, 433–452. (4) Younossi, Z. M.; Wong, G.; Anstee, Q. M.; Henry, L. Clin Gastroenterol Hepatol. 2023, 21, 1978–1991. (5) Dasgupta, P.; Henshaw, C.; Youlden, D. R.; Clark, P. J.; Aitken, J. F.; Baade, P. D. Front Oncol. 2020, 10, 171. (6) Oh, J. H.; Jun, D. W. Clin Mol Hepatol. 2023, 29, 355–357. (7) Singal, A. G.; Kanwal, F.; Llovet, J. M. Nat Rev Clin Oncol. 2023, 20, 864– 884. (8) Ganesan, P.; Kulik, L. M. Clin Liver Dis. 2023, 27, 85–102. (9) Suresh, D.; Srinivas, A. N.; Prashant, A.; Harikumar, K. B.; Kumar, D. P. Clin Exp Med. 2023, 23, 1901–1916. (10) Wei, H.; Dong, C.; Li, X. J Clin Transl Hepatol. 2024, 12, 389–405. (11) Leowattana, W.; Leowattana, T.; Leowattana, P. World J Gastroenterol. 2023, 29, 1551–1568. (12) Yang, J. D.; Heimbach, J. K. BMJ. 2020, m3544. (13) O’Leary, C.; Mahler, M.; Soulen, M. C. Chin Clin Oncol. 2021, 10, 8–8. (14) Viveiros, P.; Riaz, A.; Lewandowski, R. J.; Mahalingam, D. Cancers. 2019, 11, 1085. (15) Dalzell, C. G.; Taylor, A. C.; White, S. B. Cancers. 2023, 15, 5749. (16) UK, C. R. Diagram showing the two lobes of the liver and its blood supply. https://commons.wikimedia.org/wiki/File:Diagram_showing_ the_two_lobes_of_the_liver_and_its_blood_supply_CRUK_376. svg (accessed 05/21/2024). (17) Yamada, R.; Sato, M.; Kawabata, M.; Nakatsuka, H.; Nakamura, K.; Takashima, S. Radiology. 1983, 148, 397–401. (18) Nakamura, H.; Tanaka, T.; Hori, S.; Yoshioka, H.; Kuroda, C.; Okamura, J.; Sakurai, M. Radiology. 1983, 147, 401–405. 84 (19) Kato, T.; Nemoto, R.; Mori, H.; Takahashi, M.; Tamakawa, Y.; Harada, M. JAMA. 1981, 245, 1123–1127. (20) Lin, D.-Y.; Liaw, Y.-F.; Lee, T.-Y.; Lai, C.-M. Gastroenterology. 1988, 94, 453–456. (21) Kobayashi, H.; Hidaka, H.; Kajiya, Y.; Tanoue, P.; Inoue, H.; Ikeda, K.; Nakajo, M.; Shinohara, S. Acta Radiol Diagn. 1986, 27, 139–147. (22) Yang, C.-F.; Ho, Y.-Z.; Chang, J.-M.; Chiang, R.-H.; Lai, K.-H.; Lee, S.-D.; Tsai, Y.-T.; Lui, W.-Y.; Liu, T.-J.; Chen, G.-H. Cancer Chemother Pharmacol. 1989, 23, S26–S28. (23) Llovet, J. M.; Bruix, J. Hepatology 2003, 37, 429–442. (24) Thorn, C. F.; Oshiro, C.; Marsh, S.; Hernandez-Boussard, T.; McLeod, H.; Klein, T. E.; Altman, R. B. Pharmacogenet Genomics. 2011, 21, 440. (25) Takemura, G.; Fujiwara, H. Prog Cardiovasc Dis. 2007, 49, 330–352. (26) Lefrak, E. A.; Piťha, J.; Rosenheim, S.; Gottlieb, J. A. Cancer. 1973, 32, 302–314. (27) Walker, M. C.; Parris, C. N.; Masters, J. R. W. J Natl Cancer Inst. 1987, 79, 213–216. (28) Oh, H. J.; Aboian, M. S.; Yi, M. Y. J.; Maslyn, J. A.; Loo, W. S.; Jiang, X.; Parkinson, D. Y.; Wilson, M. W.; Moore, T.; Yee, C. R.; Robbins, G. R.; Barth, F. M.; DeSimone, J. M.; Hetts, S. W.; Balsara, N. P. ACS Cent Sci. 2019, 5, 419–427. (29) Liu, B.; Jin, Z.; Chen, H.; Liang, L.; Li, Y.; Wang, G.; Zhang, J.; Xu, T. Regen Biomater. 2021, 8, rbab043. (30) Pérez-Arnaiz, C.; Busto, N.; Leal, J. M.; García, B. J Phys Chem B. 2014, 118, 1288–1295. (31) Yee, D. W.; Hetts, S. W.; Greer, J. R. ACS Appl Mater Interfaces. 2021, 13, 41424–41434. (32) Maani, N.; Diorio, T. C.; Hetts, S. W.; Rayz, V. L. Biomech Model Mechanobiol. 2020, 19, 1865–1877. (33) Krishnamoorthy, S.; Grubbs, R. H. ACS Omega. 2020, 5, 29121–29126. (34) Blumenfeld, C. M.; Schulz, M. D.; Aboian, M. S.; Wilson, M. W.; Moore, T.; Hetts, S. W.; Grubbs, R. H. Nat Commun. 2018, 9, 2870. (35) Kondapavulur, S.; Cote, A. M.; Neumann, K. D.; Jordan, C. D.; McCoy, D.; Mabray, M. C.; Liu, D.; Sze, C.-H.; Gautam, A.; VanBrocklin, H. F.; Wilson, M.; Hetts, S. W. Biomed Microdevices. 2016, 18, 109. (36) Mabray, M. C.; Lillaney, P.; Sze, C.-H.; Losey, A. D.; Yang, J.; Kondapavulur, S.; Liu, D.; Saeed, M.; Patel, A.; Cooke, D.; Jun, Y.-W.; El-Sayed, I.; Wilson, M.; Hetts, S. W. J Vasc Interv Radiol. 2016, 27, 426–432.e1. 85 (37) F42 Committee Terminology for Additive Manufacturing Technologies, http://www.astm.org/cgi- bin/resolver.cgi?F2792- 12A (accessed 05/27/2024). (38) Li, M.; Du, W.; Elwany, A.; Pei, Z.; Ma, C. J Manuf Sci Eng Trans ASME. 2020, 142, 090801. (39) Yadroitsev, I.; Yadroitsava, I.; Plessis, A. D.; MacDonald, E., Fundamentals of Laser Powder Bed Fusion of Metals; Elsevier: 2021. (40) Beaman, J. J.; Bourell, D. L.; Seepersad, C. C.; Kovar, D. J Manuf Sci Eng Trans ASME. 2020, 142, DOI: 10.1115/1.4048193. (41) Narasimharaju, S. R.; Zeng, W.; See, T. L.; Zhu, Z.; Scott, P.; Jiang, X.; Lou, S. J Manuf Process. 2022, 75, 375–414. (42) Weaver, J. S.; Rosenthal, I. Understanding anisotropic tensile properties of laser powder bed fusion additive metals : A detailed review of select examples; tech. rep. NIST AMS 100-44; Gaithersburg, MD: National Institute of Standards and Technology (U.S.), 2021, NIST AMS 100–44. (43) Yegyan Kumar, A.; Bai, Y.; Eklund, A.; Williams, C. B. Addit Manuf. 2018, 24, 115–124. (44) Al Rashid, A.; Ahmed, W.; Khalid, M. Y.; Koc, M. Addit Manuf. 2021, 47, 102279. (45) Pagac, M.; Hajnys, J.; Ma, Q.-P.; Jancar, L.; Jansa, J.; Stefek, P.; Mesicek, J. Polymers. 2021, 13, 598. (46) Shah, M.; Ullah, A.; Azher, K.; Ur Rehman, A.; Juan, W.; Aktürk, N.; Sami Tüfekci, C.; U. Salamci, M. RSC Adv. 2023, 13, 1456–1496. (47) Yee, D. W.; Lifson, M. L.; Edwards, B. W.; Greer, J. R. Adv Mater. 2019, 31, 1901345. (48) De Camargo, I. L.; Morais, M. M.; Fortulan, C. A.; Branciforti, M. C. Ceram Int. 2021, 47, 11906–11921. (49) Saccone, M. A.; Gallivan, R. A.; Narita, K.; Yee, D. W.; Greer, J. R. Nature. 2022, 612, 685–690. (50) Ladd, M. F. C.; Palmer, R. A.; Palmer, R. A., Structure determination by X-ray crystallography; Springer: 1977; Vol. 233. (51) Freundchen Deutsch: Veranschaulichung der Bragg’schen Beugung am Kristallgitter.English: Illustration of the bragg diffraction. https : / / commons . wikimedia.org/wiki/File:Bragg_XRD.svg (accessed 05/20/2024). (52) DrBoStefanov English: This is a drawing of a XRD diffractometer (modelled after an image of Siemens D5000) in grazing incidence x-ray diffraction configuration. (Drawn for my PhD Thesis). https://commons.wikimedia. org/wiki/File:XRD_diffractometer.svg (accessed 05/20/2024). 86 (53) Kaduk, J. A.; Billinge, S. J. L.; Dinnebier, R. E.; Henderson, N.; Madsen, I.; Černý, R.; Leoni, M.; Lutterotti, L.; Thakral, S.; Chateigner, D. Nat Rev Methods Primers. 2021, 1, 77. (54) Inorganic Crystal Structure Database – ICSD | FIZ Karlsruhe https:// www.fiz - karlsruhe.de/ en/produkte- und- dienstleistungen / inorganic-crystal-structure-database-icsd. (55) What is EDS/EDX? - Nanoanalysis https : / / nano . oxinst . com / campaigns/what-is-eds/edx (accessed 06/04/2024). (56) Muso Principle of EDX https : / / commons . wikimedia . org / wiki / File:EDX-scheme.svg (accessed 05/20/2024). (57) Element mapping https://serc.carleton.edu/research_education/ geochemsheets/elementmapping.html (accessed 05/20/2024). (58) Foner, S. Rev Sci Instrum. 1956, 27, 548. (59) 13.3: Lenz’s Law https : / / phys . libretexts . org / Bookshelves / University_Physics/University_Physics_(OpenStax)/Book%3A_ University_Physics_II_- _Thermodynamics_Electricity_and_ Magnetism_(OpenStax)/13%3A_Electromagnetic_Induction/13. 03%3A_Lenz’s_Law (accessed 06/04/2024). (60) Inductively Coupled Plasma Mass Spectrometry, Single Quadrupole ICPMS | Agilent https://www.agilent.com/en/product/atomic-spectroscopy/ inductively- coupled- plasma- mass- spectrometry- icp- ms (accessed 05/20/2024). (61) A Beginner’s Guide to ICP-MS, Mass Spectrometry basics | Agilent https: //www.agilent.com/en/product/atomic-spectroscopy/inductivelycoupled-plasma-mass-spectrometry-icp-ms/what-is-icp-msicp-ms-faqs (accessed 05/20/2024). (62) Bundhun, P. K.; Janoo, G.; Chen, M.-H. Medicine. 2016, 95, e3877. (63) Taha, M. M.; Nakahara, I.; Higashi, T.; Iwamuro, Y.; Iwaasa, M.; Watanabe, Y.; Tsunetoshi, K.; Munemitsu, T. Surg Neurol. 2006, 66, 277–284. (64) Timmer, F. E.; Geboers, B.; Nieuwenhuizen, S.; Schouten, E. A.; Dijkstra, M.; de Vries, J. J.; van den Tol, M. P.; de Gruijl, T. D.; Scheffer, H. J.; Meijerink, M. R. Curr Oncol Rep. 2021, 23, 68. (65) Kuruvilla, S. P.; Tiruchinapally, G.; Crouch, A. C.; ElSayed, M. E. H.; Greve, J. M. PLOS ONE. 2017, 12, e0181944. (66) Oun, R.; E. Moussa, Y.; J. Wheate, N. Dalton Trans. 2018, 47, 6645–6653. (67) Hwang, K.; Shiau, M. Metall Mater Trans B. 1996, 27, 203–211. (68) Cooter, I.; Mundy, R. J Res Natl Bur Stand. 1957, 59, 379. 87 (69) Arnold, H.; Elmen, G. Bell Syst Tech J. 1923, 2, 101–111. (70) Suh, I.-K.; Ohta, H.; Waseda, Y. J Mater Sci. 1988, 23, 757–760. (71) Jette, E. R.; Foote, F. J Chem Phys. 1935, 3, 605–616. (72) Owen, E.; Yates, E. Lond Edinb Dubl Phil Mag. 1933, 15, 472–488. (73) Rollett, A. D.; Lee, S.-B.; Campman, R.; Rohrer, G. Annu Rev Mater Res. 2007, 37, 627–658. (74) Tannhauser, D. S. J Phys Chem Solids. 1962, 23, 25–34. (75) Lin, H.-Y.; Chen, Y.-W.; Li, C. Thermochim Acta. 2003, 400, 61–67. (76) Elliott, B. J.; Scranton, A. B.; Cameron, J. H.; Bowman, C. N. Chem Mater. 2000, 12, 633–642. (77) Herzer, G. Acta Mater. 2013, 61, 718–734. (78) Brown, D.; Ma, B.-M.; Chen, Z. J Magn Magn Mater. 2002, 248, 432–440. (79) Croat, J. J.; Herbst, J. F.; Lee, R. W.; Pinkerton, F. E. Appl Phys Lett. 1984, 44, 148–149. (80) Heo, W.-S.; Hetts, S. W.; Rayz, V. L. Int J Eng Sci. 2023, 193, 103946. (81) Arı, B.; Can, S. Z.; Bakırdere, S. Talanta. 2020, 209, 120503. (82) Weiss, P. Trans Faraday Soc. 1912, 8, 149–156. (83) Mohapatra, J.; Xing, M.; Elkins, J.; Liu, J. P. J Alloys Compd. 2020, 824, 153874. (84) Pearson, R. G. Science. 1966, 151, 172–177. (85) Fleet, M. E. Acta Crystallogr B. 1981, 37, 917–920. (86) Sasaki, S.; Fujino, K.; Takéuchi, Y. Proc Jpn Acad Series B. 1979, 55, 43– 48. (87) Raghavan, V. JPED. 2010, 31, 369–371. (88) Brocchi, E.; Moura, F.; Avalo, O.; Navarro, R.; Souza, R. European Metallurgical Conference, EMC 2013 2013, 397–408. (89) Howald, R. A. Metall Mater Trans A. 2003, 34, 1759–1769. (90) Clarke, R. S.; Scott, E. R. D. Am Mineral. 1980, 65, 624–630. (91) Picard, J. P.; Baud, G.; Besse, J. P.; Chevalier, R. J Less-Common Met. 1980, 75, 99–104. (92) Wyckoff, R. W. G.; Crittenden, E. D. Z Kristallogr Cryst Mater. 1926, 63, 144–147. (93) Gökhan Ünlü, C.; Burak Kaynar, M.; Şimşek, T.; Tekgül, A.; Kalkan, B.; Özcan, Ş. J Alloys Compd. 2019, 784, 1198–1204. 88 (94) Block, T.; Knoblauch, N.; Schmücker, M. Thermochim Acta. 2014, 577, 25–32. (95) Dmitriev, D. S.; Khristyuk, N. A.; Popkov, V. I. J Alloys Compd. 2020, 849, 156625. (96) Zhang, H.; Xia, W.; Shen, H.; Guo, W.; Liang, Z.; Zhang, K.; Wu, Y.; Zhu, B.; Zou, R. Angew Chem Int Ed. 2020, 59, 1871–1877. (97) Ohnuma, I.; Enoki, H.; Ikeda, O.; Kainuma, R.; Ohtani, H.; Sundman, B.; Ishida, K. Acta Mater. 2002, 50, 379–393. (98) Archer, W. R.; Schulz, M. D. Soft Matter. 2020, 16, 8760–8774. (99) Fruehan, R. J. 1998. (100) Pohl, M.; Kurig, H.; Tallo, I.; Jänes, A.; Lust, E. Chem Eng J. 2017, 320, 576–587. (101) Thauer, E.; Zakharova, G. S.; Wegener, S. A.; Zhu, Q.; Klingeler, R. J Alloys Compd. 2021, 853, 157364. (102) Lee, K. T.; Lytle, J. C.; Ergang, N. S.; Oh, S. M.; Stein, A. Adv Funct Mater. 2005, 15, 547–556. (103) Eggeler, Y. M.; Chan, K. C.; Sun, Q.; Lantada, A. D.; Mager, D.; Schwaiger, R.; Gumbsch, P.; Schröder, R.; Wenzel, W.; Korvink, J. G.; Islam, M. Adv Funct Mater. 2024, 34, 2302068. (104) Lu, A.-H.; Li, W.-C.; Salabas, E.-L.; Spliethoff, B.; Schüth, F. Chem Mater. 2006, 18, 2086–2094. (105) Sevilla, M.; Fuertes, A. B. Carbon. 2006, 44, 468–474. (106) Singer, J.; Anolick, E. S. J Appl Phys. 1959, 30, S193–S194. (107) Chipman, J. Metall Trans. 1972, 3, 55–64. (108) Li, K.; Liu, Q.; Cheng, H.; Hu, M.; Zhang, S. Spectrochim Acta A. 2021, 249, 119286. (109) Crystallite Size and Strain http://pd.chem.ucl.ac.uk/pdnn/peaks/ size.htm (accessed 06/03/2024). (110) Holder, C. F.; Schaak, R. E. ACS Nano. 2019, 13, 7359–7365. (111) Herzer, G. IEEE Trans Magn. 1990, 26, 1397–1402. (112) Kumar, D.; Narayan, J.; Kvit, A. V.; Sharma, A. K.; Sankar, J. J Magn and Magn Mater. 2001, 232, 161–167. (113) Billinge, S., Local structure from total scattering and atomic pair distribution function (PDF) analysis, 2008. (114) Anderson, R. N.; Parlee, N. A. D. J Vac Sci Technol. 1976, 13, 526–529. 89 (115) A. Su, G.; J. Wadsworth, O.; Suzanne Muller, H.; R. Archer, W.; W. Hetts, S.; D. Schulz, M. J Mater Chem B. 2023, 11, 8449–8455. (116) Häussinger, D.; Dhiman, R. K.; Felipo, V.; Görg, B.; Jalan, R.; Kircheis, G.; Merli, M.; Montagnese, S.; Romero-Gomez, M.; Schnitzler, A.; TaylorRobinson, S. D.; Vilstrup, H. Nat Rev Dis Primers. 2022, 8, 1–22. (117) Boyd, G. E.; Larson, Q. V. J Am Chem Soc. 1967, 89, 6038–6042. (118) Damtie, M. M.; Volpin, F.; Yao, M.; Tijing, L. D.; Hailemariam, R. H.; Bao, T.; Park, K.-D.; Shon, H. K.; Choi, J.-S. Environ. Eng. Res. 2021, 26. (119) Aradi, K.; Di Giorgio, A.; Duca, M. Chem Eur J. 2020, 26, 12273–12309. (120) Chen, Y.; Huang, W.-G.; Zha, D.-J.; Qiu, J.-H.; Wang, J.-L.; Sha, S.-H.; Schacht, J. Hear Res. 2007, 226, 178–182. (121) Rosenberg, C. R.; Fang, X.; Allison, K. R. PLOS ONE. 2020, 15, e0237948. (122) Guo, X.; Nzerue, C. Cleve Clin J Med. 2002, 69, 289–290. (123) Peaks and Troughs: What Are They, and More | Osmosis https://www. osmosis.org/answers/peaks-and-troughs (accessed 06/02/2024). (124) Coey, J. M. D. Engineering. 2020, 6, 119–131. (125) E. Baker, S.; A. Baker, A.; A. Orme, C.; A. Worthington, M.; T. Li, T.; M. Sedillo, E.; Dudoff, J.; I. Lee, J. R.; D. Kuntz, J.; K. McCall, S. RSC Adv. 2022, 12, 3721–3728. (126) Bittner, F.; Thielsch, J.; Drossel, W.-G. Prog Addit Manuf. 2020, 5, 3–9. (127) Chen, Y.; Xiong, C.; Li, Y. Metals. 2024, 14, 446. (128) Yap, Y. L.; Sing, S. L.; Yeong, W. Y. Rapid Prototyp J. 2020, 26, 1345–1361. (129) Zhao, W.; Wang, Z.; Zhang, J.; Wang, X.; Xu, Y.; Ding, N.; Peng, Z. Adv Mater Technol. 2021, 6, 2001218. (130) Dheyab, M. A.; Aziz, A. A.; Jameel, M. S.; Noqta, O. A.; Khaniabadi, P. M.; Mehrdel, B. Sci Rep. 2020, 10, 10793. (131) Online onversions, calculations and formulas https://www.aqua-calc. com/ (accessed 06/04/2024). 90 Appendix A IMAGEJ MEASUREMENTS A.1 Fe » Ni Swollen (mm) Calcined (mm) Reduced (mm) 14.750 7.28 2.99 14.939 7.42 3.41 14.777 7.61 3.25 14.932 7.6 15.003 7.61 15.351 8.48 Average ± Standard Deviation 14.96 ± 0.22 7.67 ± 0.42 3.22 ± 0.21 Table A.1: Shrinkage results for the iron-dominant samples discussed in Chapter 3 Sections 2 and 3. S to C (%) 49.36 49.67 51.50 50.90 50.72 55.24 Average ± Standard Deviation 51.23 ± 2.12 C to R (%) 41.07 45.96 42.71 S to R (%) 20.27 22.83 21.99 43.25 ± 2.49 21.70 ± 1.30 Table A.2: Shrinkage percentages with respect to the preceding phase of the system for the iron-dominant samples discussed in Chapter 3 Sections 2 and 3. 91 Figure A.1: Histograms depicting the size of the beams and pores in the irondominant samples discussed in Chapter 3. For the beams, n=250, and for the pores, n=70. Figure A.2: Histograms depicting the size of the beams and pores in the nickel-iron samples discussed in Chapter 3. For the beams, n=244, and for the pores, n=74. 92 Figure A.3: Histograms depicting the size of the beams and pores in the coppernickel-iron samples discussed in Chapter 3. For the beams, n=236, and for the pores, n=72. 93 Appendix B EDS MEASUREMENTS This appendix describes and documents the measurements derived from the EDS detector regarding the metal content in various samples of interest. Of note, for the samples discussed in Chapter 3, the single element maps, scanning electron micrographs of the region from which the sample was taken, and the table depicting the atomic composition of the sample are included for each site to demonstrate the use of the technique. The remaining samples include just the metal mole percentages as determined by EDS in those The metal ratio was determined by dividing the mole percentage of the metal of interest by the sum of the mole percentages of all metals of interest in the sample. B.1 Chapter 3 System: Ni-Fe Figure B.1: EDS of Ni-Fe site 001 from Fig. 3.8. 94 Figure B.2: EDS of Ni-Fe site 002 from Fig. 3.8. Figure B.3: EDS of Ni-Fe site 003 from Fig. 3.8. 95 Figure B.4: EDS of Ni-Fe site 004 from Fig. 3.8. Figure B.5: EDS of Ni-Fe site 005 from Fig. 3.8. 96 Figure B.6: EDS of Ni-Fe site 006 from Fig. 3.8. Figure B.7: EDS of Ni-Fe site 007 from Fig. 3.8. 97 Figure B.8: EDS of Ni-Fe site 008 from Fig. 3.8. Figure B.9: EDS of Ni-Fe site 009 from Fig. 3.8. 98 Figure B.10: EDS of Ni-Fe site 010 from Fig. 3.8. Location ID 001 002 003 004 005 006 007 008 009 010 Average ± Standard Deviation Nickel Mol% Iron Mol% 68.87 31.13 65.19 34.81 66.45 33.55 66.14 33.86 65.31 34.69 69.24 30.76 66.71 33.29 69.59 30.41 71.77 28.23 68.87 31.13 67.81 ± 2.16 32.19 ± 2.16 Table B.1: Iron and nickel content used to determine average composition of the Ni-Fe sample in Chapter 3 Section 2.1. 99 B.2 Chapter 3 System: Cu-Ni-Fe Figure B.11: EDS of Cu-Ni-Fe site 001 from Fig. 3.8. Figure B.12: EDS of Cu-Ni-Fe site 002 from Fig. 3.8. 100 Figure B.13: EDS of Cu-Ni-Fe site 003 from Fig. 3.8. Figure B.14: EDS of Cu-Ni-Fe site 004 from Fig. 3.8. 101 Figure B.15: EDS of Cu-Ni-Fe site 005 from Fig. 3.8. Figure B.16: EDS of Cu-Ni-Fe site 006 from Fig. 3.8. 102 Figure B.17: EDS of Cu-Ni-Fe site 007 from Fig. 3.8. Figure B.18: EDS of Cu-Ni-Fe site 008 from Fig. 3.8. 103 Figure B.19: EDS of Cu-Ni-Fe site 009 from Fig. 3.8. Figure B.20: EDS of Cu-Ni-Fe site 010 from Fig. 3.8. 104 Location ID 001 002 003 004 005 006 007 008 009 010 Average ± Standard Deviation Copper Mol% Nickel Mol% Iron Mol% 33.99 23.05 42.96 39.15 23.20 37.65 33.23 22.98 43.79 34.41 22.27 43.32 37.13 24.18 38.69 39.16 22.22 38.62 36.34 23.53 40.13 33.39 22.72 43.89 35.03 22.40 42.57 32.42 22.96 44.62 35.43 ± 2.42 22.96 ± 0.60 41.62 ± 2.58 Table B.2: Copper, nickel, and iron content used to determine average composition of the sample in Chapter 3 Section 2.1. B.3 Chapter 4 System: FeNi Swell-In Nickel Mol% 80 84 87 90 93 88 89 90.4 90 EDS Nickel Mol% 70.01 70.81 77.58 91.06 86.67 74.92 80.51 79.48 78.66 Table B.3: Nickel mole percentage in the swelling solution versus the average EDSmeasured mole percentage. Location ID 001 Nickel Mol% 70.01 Iron Mol% 29.99 Table B.4: Iron and nickel content determining the composition of the Ni80 Fe20 swell-in sample. Location ID 001 Nickel Mol% 70.81 Iron Mol% 29.19 Table B.5: Iron and nickel content determining the composition of the Ni84 Fe16 swell-in sample. 105 Location ID 001 002 003 004 Average ± Standard Deviation Nickel Mol% Iron Mol% 80.34 19.66 78.53 21.47 72.08 27.92 79.37 20.63 77.58 ± 3.74 22.42 ± 3.74 Table B.6: Iron and nickel content used to determinine the composition of the Ni87 Fe13 swell-in sample. Location ID 001 002 003 Average ± Standard Deviation Nickel Mol% 90.72 90.05 92.40 91.06 ± 1.21 Iron Mol% 9.28 9.95 7.60 8.94 ± 1.21 Table B.7: Iron and nickel content used to determinine the composition of the Ni90 Fe10 swell-in sample. Location ID 001 002 003 Average ± Standard Deviation Nickel Mol% Iron Mol% 84.30 15.79 87.58 12.42 88.12 11.88 86.67 ± 2.06 13.33 ± 2.06 Table B.8: Iron and nickel content used to determinine the composition of the Ni93 Fe7 swell-in sample. Location ID 001 002 003 Average ± Standard Deviation Nickel Mol% Iron Mol% 73.95 26.05 74.82 25.18 75.98 24.02 74.92 ± 1.02 25.08 ± 1.02 Table B.9: Iron and nickel content used to determinine the composition of the Ni88 Fe12 swell-in sample. 106 Location ID 001 002 003 Average ± Standard Deviation Nickel Mol% Iron Mol% 80.55 19.45 79.64 20.36 81.33 18.67 80.51 ± 0.85 19.49 ± 0.85 Table B.10: Iron and nickel content used to determinine the composition of the Ni89 Fe11 swell-in sample. Location ID 001 002 003 Average ± Standard Deviation Nickel Mol% Iron Mol% 80.35 19.65 79.11 20.89 78.98 21.02 79.48 ± 7.55 20.52 ± 7.55 Table B.11: Iron and nickel content used to determinine the composition of the Ni90.4 Fe9.6 swell-in sample. Location ID 001 002 003 Average ± Standard Deviation Nickel Mol% Iron Mol% 78.74 21.26 78.60 21.40 78.65 21.35 78.66 ± 0.07 21.34 ± 0.07 Table B.12: Iron and nickel content used to determinine the composition of the other Ni90 Fe10 swell-in sample. B.4 Chapter 4 System: FeCo Cobalt Mol% 100 90 80 70 60 50 40 30 20 10 0 Iron Mol% 0 10 20 30 40 50 60 70 80 90 100 Table B.13: Cobalt and iron mole percentages in the swell-in samples synthesized to characterize the iron-cobalt HIAM system. 107 Location ID 001 002 003 004 005 006 007 008 009 010 Average ± Standard Deviation Cobalt Mol% 91.50 91.22 91.23 90.87 91.11 91.39 91.43 91.09 91.43 91.54 91.28 ± 0.21 Iron Mol% 8.50 8.78 8.77 9.13 8.89 8.61 8.57 8.91 8.57 8.46 8.72 ± 0.21 Table B.14: Cobalt and iron content used to determine average composition of the Co90 Fe10 swollen sample. Location ID 001 002 003 004 005 006 007 008 009 010 Average ± Standard Deviation Cobalt Mol% Iron Mol% 84.80 15.20 84.85 15.15 84.58 15.42 84.22 15.78 85.09 14.91 84.54 15.46 84.40 15.60 84.35 15.65 84.79 15.21 85.09 14.91 84.67 ± 0.30 15.33 ± 0.30 Table B.15: Cobalt and iron content used to determine average composition of the Co80 Fe20 swollen sample. 108 Location ID 001 002 003 004 005 006 007 008 009 010 Average ± Standard Deviation Cobalt Mol% Iron Mol% 74.89 25.11 75.23 24.77 74.20 25.80 75.28 24.72 74.45 25.55 75.13 24.87 75.17 24.83 75.33 24.67 74.34 25.66 75.31 24.69 74.93 ± 0.44 25.07 ± 0.44 Table B.16: Cobalt and iron content used to determine average composition of the Co70 Fe30 swollen sample. Location ID 001 002 003 004 005 006 007 008 009 010 Average ± Standard Deviation Cobalt Mol% Iron Mol% 67.32 32.68 66.93 33.07 67.19 32.81 66.97 33.03 67.48 32.52 67.23 32.77 66.88 33.12 66.51 33.49 67.26 32.74 67.78 32.22 67.16 ± 0.35 32.84 ± 0.35 Table B.17: Cobalt and iron content used to determine average composition of the Co60 Fe40 swollen sample. 109 Location ID 001 002 003 004 005 006 007 008 009 010 Average ± Standard Deviation Cobalt Mol% Iron Mol% 58.20 41.80 58.23 41.77 57.76 42.24 59.11 40.89 58.46 41.54 59.11 40.89 58.65 41.35 57.97 42.03 59.06 40.94 59.66 40.34 58.62 ± 0.60 41.38 ± 0.60 Table B.18: Cobalt and iron content used to determine average composition of the Co50 Fe50 swollen sample. Location ID 001 002 003 004 005 006 007 008 009 010 Average ± Standard Deviation Cobalt Mol% Iron Mol% 51.11 48.89 50.38 49.62 48.68 51.32 51.58 48.42 49.52 50.48 49.62 50.38 49.36 50.64 48.77 51.23 49.65 50.35 51.59 48.41 50.02 ± 1.08 49.97 ± 1.08 Table B.19: Cobalt and iron content used to determine average composition of the Co40 Fe60 swollen sample. 110 Location ID 001 002 003 004 005 006 007 008 009 010 Average ± Standard Deviation Cobalt Mol% Iron Mol% 35.89 64.11 35.32 64.68 36.39 63.61 36.95 63.05 35.57 64.43 35.73 64.27 36.21 63.79 34.07 65.93 36.23 63.77 36.52 63.48 35.89 ± 0.80 64.11 ± 0.80 Table B.20: Cobalt and iron content used to determine average composition of the Co30 Fe70 swollen sample. Location ID 001 002 003 004 005 006 007 008 009 010 Average ± Standard Deviation Cobalt Mol% Iron Mol% 26.32 73.68 25.68 74.32 26.88 73.12 27.07 72.93 26.12 73.88 26.45 73.55 27.69 72.31 26.47 73.53 25.83 74.17 25.16 74.84 26.37 ± 0.73 73.63 ± 0.73 Table B.21: Cobalt and iron content used to determine average composition of the Co20 Fe80 swollen sample. 111 Location ID 001 002 003 004 005 006 007 008 009 010 Average ± Standard Deviation Cobalt Mol% Iron Mol% 14.32 85.68 14.18 85.82 13.37 86.63 13.55 86.45 14.28 85.72 13.80 86.20 14.18 85.82 13.77 86.23 14.87 85.13 13.15 86.85 13.95 ± 0.52 86.05 ± 0.52 Table B.22: Cobalt and iron content used to determine average composition of the Co10 Fe90 swollen sample. 112 Appendix C MAGNETIC NANOPARTICLE PREPARATION Figure C.1: A schematic illustrating the preparation of magnetic nanoparticles for capture testing experimentation. Iron(II,III) oxide nanoparticles, as purchased from Sigma-Aldrich, were suspended in water to which citric acid was added and subsequently heated at 65◦ C for one hour. The particles were allowed to settle and the supernatant removed, at which point the nanoparticles were washed with ethanol twice and subsequently dried in air. Nanoparticle solutions for capture testing were synthesized by combining these treated nanoparticles with phosphate buffered saline in the concentration of interest; in this case, 5 milligrams per milliliter of solution, the equivalent of a 3.58 mM solution of functionalized nanoparticles in phosphatebuffered saline. The citrate coating procedure was inspired by [130]. 113 Appendix D VACUUM HIAM NANOCRYSTALLINE SIZE AND EDS MEASUREMENT Figure D.1: Vacuum HIAM liftout sample preparation for TEM, including (a) an indication of the region characterized by EDS in section (b), where the iron, carbon, and oxygen distribution are shown. Note the contrast of the particles in part (a) is the opposite of that seen in the lamella in Fig. 5.3. Data courtesy of Wenxin Zhang. 114 Figure D.2: A schematic illustrating the identified crystallites in each layer along with circles outlining grains to be used to estimate grain size in (a) the superficial nanocrystalline and (b) deep nanocrystalline layers of the iron vacuum HIAM system and (c) the fine nanocrystalline layer of the copper-nickel-iron vacuum HIAM system. Superficial Fe (nm) Deep Fe (nm) Fine CuNiFe (nm) 22.43 3.35 9.94 18.82 3.76 10.97 18.22 3.71 9.94 19.13 3.31 9.60 13.01 4.02 10.24 15.31 2.92 9.40 21.51 2.88 9.70 15.09 4.31 7.98 32.84 4.38 7.68 19.90 3.51 13.04 19.83 3.31 10.04 17.22 4.73 8.12 17.30 4.22 9.40 35.05 4.01 9.40 24.41 3.61 10.58 25.72 8.71 13.01 9.60 14.62 10.68 14.77 9.90 22.50 8.91 Average ± Standard Deviation 20.03 ± 5.98 3.73 ± 0.54 9.69 ± 1.18 Table D.1: Grain diameters as computed from the area of the circle inscribing several nanoparticles for each of the nanocrystalline regimes discussed in Chapter 5. 115 Appendix E MAGNETIC PROPERTY CALCULATIONS Coercivity was computed from the collected magnetic data by averaging the magnetization at the point of the lowest magnitude positive magnetic moment with the magnetization at the point of the lowest magnitude negative magnetic moment during demagnetization and remagnetization, and taking the average of the absolute value of these two values. The recorded magnetic data was offset by such an amount that the coercivity, when measured by the newly offset curve, was of the same magnitude in both the negative and positive magnetization directions. Remnance was then calculated by taking the positive intercept of the magnetic hysteresis loop with the magnet moment/magnetic flux density axis, and the maximum energy product was calculated by finding the maximum magnitude value of the product of the magnetization and magnetic flux density at any point on the magnetic hysteresis loop where the magnetization and the magnetic flux density were of opposite sign. The density of complex samples was estimated by calculating a molar volume based on a linear combination of molar volumes weighted by the mole percentage of each element in the material; i.e„ for a combination of metals A and B with 𝑀 𝐴 and 𝑀𝐵 molar masses, 𝑣 𝐴 and 𝑣 𝐵 molar volumes, and 𝑚 𝐴 and 𝑚 𝐵 mole percentages in the mixture, the density 𝐷 was taken to be: 𝐷= 𝑚 𝐴 ∗ 𝑀 𝐴 + 𝑚 𝐵 ∗ 𝑀𝐵 . 𝑚 𝐴 ∗ 𝑣 𝐴 + 𝑚𝐵 ∗ 𝑣𝐵 Molar volumes were obtained from [131]. (E.1) 116 Appendix F COMSOL SIMULATION PARAMETERS Parameter 𝜌 𝜇 u𝑖𝑛 𝑝 𝑜𝑢𝑡 𝜃 𝐵𝑟 𝜇𝑟 𝜇𝑟,𝑝 𝜌𝑝 𝑑𝑝 𝐿𝑣 𝑅𝑣 𝑅𝑠 𝐿𝑚 𝑅𝑚 𝐻𝑤𝑎𝑙𝑙 𝑇 𝑤𝑖𝑠𝑡 𝐴𝑛𝑔𝑙𝑒 𝐻𝑠𝑖𝑑𝑒 𝑃𝑒𝑟 𝑓𝑙 𝑃𝑒𝑟 𝑓𝑎 Description Fluid density Fluid dynamic viscosity Inflow velocity Outflow pressure Magnetization Angle vs. Central Axis Remnant Magnetization Relative magnetic permeability Relative permeability of magnetite Magnetite density Particle diameter Surrounding Tube Length Surrounding Tube Radius Air Sphere Radius Magnetic Lattice Length Magnetic Lattice Radius Lattice Beam Thickness Angle Between Adjacent Lattice Mag. Lattice Pore Radius Perforations Along Lattice Area of Lattice Perforations Value 0.997 g/cm3 10−3 Pa·s 1.6 m/s 0 Pa35 0 0.028 or 1.45 T 2.88 7.36 ∗ 106 5.15 g/cm335 1 𝜇m35 17 mm 1.72 mm 10 mm 1.91 mm 1.66 mm 0.06 mm 0 0.16 mm 10 0.1244x0.1244 mm2 Table F.1: Parameters used to compute the capture efficiency of magnetic lattices as discussed in Chapter 3. Collected and determined by S. Shaker and A. Syrota.