M. Taheri-Moudavi, B. Webler, J. Kitchin, A.I. Torres
Carnegie Mellon University,
United States
Keywords: rare earth alloys, machine learning, density functional theory, process design, additive manufacturing
Summary:
The aim of this contribution is to present the advances of two interconnected projects carried out between Carnegie Mellon University and the Naval Nuclear Laboratory to exploit REEs in the design of next-generation additively manufactured (AM) structural alloys for high-temperature applications. These multidisciplinary projects include the full design cycle, from designing processes for the separation of different REE mixes from ores to the computer-based AI-enhanced design of alloys that can achieve desired mechanical properties using the mentioned REE mixes as feedstocks, to actual alloy manufacturing and, finally, the characterization and mechanical testing of the alloys. Figure 1 shows the proposed concept. The long-term goal is to develop a novel multi-modal, multi-field generative AI framework that can predict alloy properties and manufacturing cost, which will allow for a more efficient exploration of the alloy design space and reduce the time for the design. Such a framework will discover the non-linear roles of the elements on various properties. For example, different ore-to-REE mix processes will be designed and economically and environmentally evaluated. Currently, REEs ores are processed until a 99.9% purity on each individual REE is obtained. However, this high-purity state-of-the-art standard requires substantial processing steps, which leads to an increase in manufacturing costs and environmental burdens. We hypothesize that some of these processing steps may not be required as REEs and accompanying metals are later recombined during the alloy manufacturing step. We aim to correlate processing steps with REE mix composition, manufacturing costs, and environmental impacts. Alloy composition - alloy property correlations will be first established using Density Functional Theory (DFT) and Calculation of Phase Diagrams (CALPHAD) methodologies. Successful alloy candidates will be prepared using additive manufacturing; those that can be proven to be printable without major defects are characterized and tested. The results of these experiments provide feedback to the other activities. During the presentation, we will discuss the advances in terms of benchmarking DFT and CALPHAD to calculate the properties of rare earth metals, alloys, and oxides; the development of processing flowsheets representative of current practices for extraction of La, Ce, and Y from Bastnaesite, and associated costs and environmental impacts; and the fabrication of alloy specimens of different compositions. Acknowledgement: We acknowledge support from Naval Nuclear Laboratory (NNL) award No. 60676.1.1043353 - Fluor NNL Processing PO153761