Technical Meeting on Tritium Breeding Blankets and Associated Neutronics

A First Approach to the Development of Formulations for Lithium Isotope Enrichment for Fusion by Liquid Extraction with Complexing Agents Using Machine Learning Techniques

Not scheduled

20m

IAEA Headquarters, Vienna, Austria

Poster Track I: Breeding blanket design and performance Topics I, II, III Posters

Speaker

Annalisa Congiu (Eni S.p.A.)

Description

One of the crucial aspects in the development and commercialization of D-T fusion is the feeding of Tritium, which requires mass-enriched lithium 6, a minority isotope in nature (about 7%).
Although chemical properties are directly related to the electronic configuration of the atom, the mass difference between the isotope nuclei confer slightly different behavior to molecules and ions containing different isotopes of the same element and termed isotopologues. This phenomenon is known as the isotope effect and has been observed in a multiplicity of ways that have inspired the proposal of many technologies for isotopic separation [1-5]. The realization of a process for lithium isotope enrichment is complex and energy intensive because of the high number of stages required to achieve the targets and estimated quantities needed for fusion reactors currently under consideration. Except for mercury amalgam exchange technology (COLEX process, developed for military purposes in the 1960s and therefore still classified) possible technologies have very low TRLs and little information is disclosed about their potential development.
One of the most studied technologies is liquid-liquid extraction with complexing agents on which a substantial literature exists since the 1980s [6-7]. Isotopic enrichment occurs by complexation and extraction from lithium salts dissolved in water. Rationalization of the many parameters that impact equilibria is complicated by the fragmented nature of the sources. This presentation illustrates the creation of a PoC of a Machine Learning model capable of supporting experimentation through design of experiment.
The activity performed concerns:
• development of an Extraction Process Database and its Analysis
• implementation and optimization of Machine Learning (ML) pipeline
• development of an explainability process to uncover the relations between variables and target values;
• development of a user interface

Development and critical points of this approach are discussed.

REFERENCES
[1] Giegerich T., Report on the assessment of a viable route for the separation of lithium isotopes, internal report TRANSversal Actions for Tritium, KIT, 2019
[2] Murali, A., Zhang, Z., Free, M. L., Sarswat, P. K., A Comprehensive Review of Selected Major Categories of Lithium Isotope Separation Techniques, Phys. Status Solidi A, 2100340 (2021) 1-21
[3] Ault, T, Brozek, K., Fan, L., Folsom M., Kim, J, Zeismer, J., Lithium Isotope Enrichment: Feasible Domestic Enrichment Alternatives, Report UCBTH-12-005, Department of Nuclear Engineering University of California, Berkeley, 2012
[4] Symons, E. A., Lithium Isotope Separation: A Review of Possible Techniques, Separation Science and Technology, 20 9-10, (1985) 633-651
[5] Badea, S. L., Niculescu, V. C., & Iordache, A. M., New Trends in Separation Techniques of Lithium Isotopes: A Review of Chemical Separation Methods. Materials, 16 3817 (2023) 1-15
[6] Nishizawa, K, Takano, T. Ikeda, I., Okohara M., Extractive Separation of Lithium Isotopes by Crown Ethers Separation Science and Technology, 23 4-5 (1988) 333-345
[7] Pei, H. Yan, F., Liu, H. He, B. Li, J., The selective complexation of crown ethers for lithium isotope separation: A critical review, Separation and Purification Technology 341 126857 (2024) 1-13