Laying the Foundation for a Nuclear Reactions Program using Rare Ion Beams and Artificial Intelligence

Nicholas Scielzo | 22-ERD-020

Project Overview

Many of the processes that are central to our nuclear-security missions and to the creation of the elements in the cosmos are driven by a complex network of nuclear reactions. However, we have no data for most of the nuclei involved because they are radioactive, and the models used to simulate these processes are limited as they often fail to properly capture the relationships that exist between nuclear-reaction rates. As a result, LLNL and the broader NNSA community currently lack the tools needed to identify and measure the nuclear processes that have the greatest impact on our missions. During this project we took the first steps toward addressing this long-standing problem through an integrated effort that combines (1) building statistical models of nuclear data with deep-learning techniques and (2) demonstrating several key capabilities needed to establish a world-leading experimental program at the Facility for Rare Isotope Beams (FRIB). We showedd the promise of these approaches by demonstrating some of the radiochemistry and nuclear-physics techniques needed to harness the unprecedented radioisotope production at FRIB and taking advantage of the paradigm shift in data handling due to recent advances in machine learning and artificial intelligence. This effort established the multidisciplinary research teams in nuclear science, radiochemistry, high-performance computing and data science to address these pressing nuclear-data needs. The success of this project had a key role in laying the foundation for a larger Strategic Initiative LDRD project to address these challenges on a larger scale.

Mission Impact

The advances initiated here lay the groundwork to revolutionize our ability to address a wide range of nuclear-data needs that are central to stockpile stewardship and nuclear forensics. These national-security missions rely in part on interpreting the telltale radioactive signatures found in the debris of a nuclear detonation that arise from nuclear reactions involving the actinide fuel, the resulting fission products, and any radiochemical diagnostics or structural materials present. The development of new, world-leading capabilities that bring together nuclear physics, radiochemistry, and data science are needed to keep LLNL at the forefront of addressing the most pressing nuclear-data needs that underpin these security missions. This effort is essential to position LLNL to deliver the highest-impact nuclear-science program over the next decade and beyond.

Publications, Presentations, and Patents

Kmak, K.N. et al., 2022. "Development of Chemical Procedures for Isotope Harvesting: Separation of Trace Hafnium from Tungsten." Solvent Extraction and Ion Exchange 40 1-17 (2022). https://doi.org/10.1080/07366299. November 2022.2079502.

Bence, J.A. et al., 2022. "Solid-Phase Isotope Harvesting of 88Zr from a Radioactive Ion Beam Facility." Applied Radiation and Isotopes 189, 110414 (2022). https://doi.org/10.1016/j.apradiso.2022.110414.

Ratkiewicz, A. 2022. "Constraining Neutron-Induced Reactions Through the Surrogate Reaction Method." Talk at the Workshop for Applied Nuclear Data Activities (WANDA) Meeting. Virtual. February 2022.

Ratkiewicz, A. 2022. "Surrogate Reactions for Nuclear Astrophysics." Presentation at the International Nuclear Physics Conference (INPC). Cape Town South Africa, September 2022.