Used nuclear fuel (UNF) discharged from nuclear reactors contains many elements and associated isotopes of the periodic table produced from neutron absorption, fission, activation, and corrosion. Some isotopes have significant recycle values due to reasons such as fuel fabrication (e.g., 235U), proliferation risks (e.g., 249Pu), medical uses (e.g., 99Mo, 153Gd), nuclear battery production (e.g., 237Np), and waste disposal (e.g., 90Sr, 137Cs, 241Am). The elements and associated isotopes are differentiated by their chemical properties and behavior to facilitate bulk separations, which lays the foundation to reprocess UNF. Singling out individual elements is ideal and yet needs tremendous effort, so the concessions are separating elements with similar properties first followed by further purification if necessary. The most common tactic is separating U and Pu from other metal elements first by the Plutonium Uranium Redox EXtraction (PUREX) process, followed by their purification from impurities (e.g., Np) and each other. Then trivalent lanthanides and actinides are separated from the rest metal elements, followed again by their respective purification by various processes, for instance, the exemplary Trivalent Actinide Lanthanide Separation with Phosphorus-Reagent Extraction from Aqueous Komplexes (TALSPEAK) process.

Aqueous reprocessing has been the reference for more than 60 years and it relies on separation by liquid-liquid extraction (LLE) that is widely used on an industrial scale. A separating agent (either mass or energy) is needed in separation by either the mechanic force (e.g., filtration, centrifugation) or the differences in mass transfer. In the separation and purification of U and Pu by LLE, oxidants and reductants (e.g. Fe(NH2SO3)2, U4+, N2H4, or NH2OH) are added to adjust the oxidation state of U and Pu. The most noteworthy impurity metal, Np, needs additional reductant, such as (iso-)butyraldehyde, hydroxamic acid, and (di-)hydroxyurea. Because the redox agents sometimes are unable to control the final oxidation states of the U/Np/Pu trio species accurately, resulting in the same oxidation states with extractability and a failure in separation. Many repeated redox steps for purifying U and Pu complicate the PUREX process. When the oxidation states are not easily regulated by redox agents, which poses significant challenges in separations, such as separations of trivalent lanthanides (Ln3+) from trivalent actinides (An3+) by the TALSPEAK process, aqueous hold-back chelators (HBAs) are required. Ideal HBAs have different electrostatic and steric interactions with Ln3+ and An3+, and are thus able to discern the nuance in size between the two types of metal ions. While the HBAs change the thermodynamic equilibrium of the metal ions, they should have minimal negative impacts on the kinetics and mass transport of the ions and also have robust applicability in different chemical environments (e.g., over a wide range of acidities). The most widely applied HBAs, the polyaminocarboxylate HBAs, have a long-term bottleneck in achieving the desired properties, which limits their applications. Therefore, it is essential and ideal to search for an agent that acts efficiently as both a redox regulator and an HBA to overcome all snags mentioned above.

An octadentate hydroxypyridinone (HOPO) chelator, 3,4,3-LI(1,2-HOPO) (abbreviated as 343HOPO), has been used in this work as both the oxidation state controller and the HBA in the aqueous phase. The introductions of UNF compositions, LLE, and research goals and approaches are covered in Chapter 1. UNF contains most elements in the periodic table and recovery of elements of interest entails different types of separation processes. LLE stands out due to many reasons and is presented briefly. In LLE, aqueous HBAs are often used when extractants alone do not separate well, so Chapter 2 summarizes and compares HBAs used in various hydrometallurgical processes. Of all the categories of HBAs, one ligand, 343HOPO, is of particular interest and is studied in the following three chapters. Chapter 3 and Chapter 4, illustrate the application of 343HOPO in separating the U/Np/Pu trio and the Ln3+/An3+ duo, respectively. These two groups of elements are usually prioritized and recovered with great efforts in the UNF hydroprocessing. Both thermodynamics and kinetics of the ligand’s redox and complexation properties are detailed and compared with other ligand counterparts whenever applicable. The reduction and stabilization of Np by 343HOPO appeared very fast and meanwhile separations of Np from U and of Pu from U were both enhanced, enabling a potential simplified version of the PUREX process. 343HOPO’s application in assisting the separation of Ln3+ from An3+ ions eliminates the need for a pH buffer, making the separation process more compatible with the highly acidic upstream steps. Since 343HOPO is used to chelate radioactive isotopes, the radiation robustness of this ligand needs investigation. Its resistance to different types of radiolytic degradation is illustrated in Chapter 5. Again, thermodynamics and kinetics of radiolytic degradation are elaborated. The structural changes and radiolytic influences on separation performance are investigated. 343HOPO is resistant to radiation up to ~10 kGy without the protection of scavenging species and without compromising the distributions of metal ions, illustrating the potentiality in scaled-up operations. A variety of spectrophotometric, electrochemical, and radiometric techniques have been utilized in Chapter 3 to Chapter 5. Finally, Chapter 6 discusses three techniques of radiochemical separations in classroom settings by accentuating the fundamental thermodynamic concepts and on the proposed experiments that offer students hands-on experience of separating radioactive materials. This is beneficial to the nuclear chemistry and radiochemistry programs in higher education. Parts of Chapter 3 and Chapter 4 have been published1,2 and this dissertation integrates all work by linking them with more information.

Computer vision and image understanding is the problem of interpreting images by locating, recognizing objects, attributes and other higher level features in an image. In this thesis, I seek to tackle this broad problem using deep learning techniques. More specifically, I build deep neural network based models to solve two specific problems to understand images in a high level: album wise image understanding with event-specific image importance score, and description generation for an image.

I first focus on the understanding of a collection of images in an event album. In an event album, some images are more important or interesting to save or present than others, and I show that with an event-specific image importance property, we can learn the interestingness of an image given an album, and the performance of the model generated importance score is very close to human preference. I build a siamese network that can predict image importance score given the event type of that image, using novel objective function and learning scheme. Next, to make the process fully automated, I propose an iterative updating procedure for event type and image importance score prediction, that can simultaneously decide the event type of the album and the importance score of every image. It consists of a Convolutional Neural Network that recognizes the event type, a Long-Short Term Memory (LSTM) that uses sequential information for event type recognition, and a siamese network that predicts image importance score.

Furthermore, not just limited to describing an image with a score or by a classified type, I seek the possibility to describe it with a phrase or sentence. I propose a coarse-to-fine LSTM based method that decomposes the original image description into a skeleton sentence and its notable attributes, and demonstrate that in this way the language model can generate better descriptions, with the capability to generate image descriptions that better accommodates user preference.